R a d i o f R e q u e n c y a b l a t i o no f l i v e R t u m o R s

An experimental and clinical study

lars frich

The Department of Surgery and

The Interventional CentreRikshospitalet University Hospital

Thesis submitted for the degree of Dr. med.

October, 2006

Faculty of MedicineUniversity of Oslo

© Lars Frich, 2007

Series of dissertations submitted to theFaculty of Medicine, University of OsloNo. 476

ISBN 978-82-8072-693-3

All rights reserved. No part of this publication may be reproduced or transmitted, in any form or by any means, without permission.

Cover: Inger Sandved AnfinsenPrinted in Norway: AiT e-dit AS, Oslo, 2007.

Produced in co-operation with Unipub AS.The thesis is produced by Unipub AS merely in connection with thethesis defence. Kindly direct all inquiries regarding the thesis to the copyrightholder or the unit which grants the doctorate.

Unipub AS is owned byThe University Foundation for Student Life (SiO)

III

Surgery is a conservative art. It takes to novel methods reluctantly as an old dog to new tricks. It was slow to adopt the ligature; slow to adopt the principles of antisepsis; slow to adopt the fastidious technique and painstaking haemostasis

that have largely put a stop to operating by the clock. It has been equally slow to adopt the principles of electrosurgery which, from a technical standpoint, are likely to be no less revolutionizing.

— Ha rv ey Cus h i n g ,

IV

The present work was carried out at the Department of Surgery and the Interventional Centre, Rikshospitalet University Hospital during the years 2001-2006. The work was financially supported by the Norwegian Cancer Society, the University of Oslo, Rikshospitalet University Hospital, Skaugens foundation (Bergliot og Sigurd Skaugens fond til bekjempelse av kreft) and Henrik Homans foundation (Legatet til Henrik Homans minde).

I would like to express my gratitude to my primary supervisor Ivar Gladhaug, whose unique combination of clinical proficiency, scientific judgment, openness to pursue new concepts and gentle personality is highly appreciated. Co-supervisor Professor Atle Bjørnerud provided invalu-able guidance in the exciting world of magnetic resonance imaging.

My sincere thanks to Professor Erik Fosse at the Interventional Centre who provided excellent facilities for conducting experimental research in an inspiring multidisciplinary environment. I am in debt to former head of the Department of Surgery Anstein Bergan, present head of the Department of Surgery Karl-Erik Giercksky, and head of the Section for Gastrointestinal Surgery Øystein Mathisen for supporting the clinical study and for providing good working facilities.

As my predecessor research fellow, Tom Mala introduced me to the field of thermal ablation of liver tumors. His research and experience provided me with a foundation for which I am most grateful. I am also grateful to Bjørn Edwin whose early enthusiasm for thermal ablation and incorporation of these techniques into clinical practice was an important inspiration for con-ducting experimental research. I would like to thank the staff at the Interventional Centre whose professionalism and positive spirit have been a most valuable resource. Especially, I would like to thank Sumit Roy and Per Kristian Hol for inspiring scientific discussions and Terje Tillung for as-sistance in magnetic resonance imaging. It has been a pleasure to cooperate with Knut Brabrand, Gaute Hagen and Trond Mogens Aaløkken at the Department of Radiology in the treatment and follow-up of patients.

Ac k now l e d g e m e n ts

V

I have much appreciated discussions of electrosurgical principles with Professor Sverre Grimnes at the Department of Biomedical and Clinical Engineering. Sigrid Fossheim kindly provided lipo-somes. Laura Killingbergtrø and the staff at Section for Gastrointestinal Surgery provided unique care to our patients. Kristin Bjørnland and Solveig Pettersen provided expertise in zymography. I am grateful to Professor Ole Petter F. Clausen and the staff at the Department and Institute of Pathology for preparation and examination of histopathologic specimens. Dag Sørensen and the staff at the Center for Comparative Medicine provided excellent care of animals. My sincere thanks also goes to the patients who participated in the clinical study. Hopefully, you have ben-efited from our common research effort.

Finally, I would like to thank my family and my family-in-law. The ones not here are deeply missed. Special thanks to my children Ingeborg and Jens Andreas who have been a highly loved source of joy. I am immensely grateful for the support my wife Ellen provided during these years. This thesis would not have been possible without her encouragement, love and patience.

VI

con t e n ts

LIST OF PAPERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII

ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

RADIOFREQUENCy ABLATION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

AIMS OF THE STUDy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

SUMMARy OF PAPERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

FUTURE PERSPECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

APPENDIx I - THE BIOHEAT EQUATION . . . . . . . . . . . . . . . . . . . . . . . 45

APPENDIx II - A NOTE ON TERMINOLOGy . . . . . . . . . . . . . . . . . . . . . 47

APPENDIx III - A NOTE ON ANIMAL MODELS . . . . . . . . . . . . . . . . . . . . 49

REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

ERRATA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

VII

I Frich L, Mala T, Gladhaug IP. Hepatic radiofrequency ablation using perfusion electrodes in a pig model: Effect of the Pringle manoeuvre. Eur J Surg Oncol. 2006;32(5):527-32.

II Frich L, Bjørnerud A, Fossheim S, Tillung T, Gladhaug I. Experimental application of thermosensitive paramagnetic liposomes for monitoring magnetic resonance imaging guided thermal ablation. Magn Reson Med. 2004;52(6):1302-9.

III Frich L, Hol PK, Roy S, Mala T, Edwin B, Clausen OPF, Gladhaug IP. Experimental hepatic radiofrequency ablation using wet electrodes: electrode-to-vessel distance is a significant predictor for delayed portal vein thrombosis. Eur Radiol. 2006;16(9):1990-9.

IV Frich L, Bjørnland K, Pettersen S, Clausen OPF, Gladhaug IP. Increased activity of matrix metalloproteinase 2 and 9 after hepatic radiofrequency ablation. J Surg Res. 2006;135(2):297-304.

V Frich L, Hagen G, Brabrand K, Edwin B, Mathisen Ø, Aaløkken TM, Gladhaug IP. Radiofrequency ablation of colorectal liver metastases: evaluation of local tumor

recurrence by 3D volumetric analysis. Submitted.

l i s t of pA pe r s

VIII

2D 2-dimensional3D 3-dimensionalCEA carcinoembryonal antigenCRC colorectal cancerCT computed tomographyECM extracellular matrixHCC hepatocellular carcinomaMDCT multi-detector computed tomographyMMP matrix metalloproteinaseMR magnetic resonanceMRI magnetic resonance imagingPET positron emission tomographyRF radiofrequencyROI region of interestRPV right portal veinSD standard deviationSI signal intensityT1 longitudinal relaxationT2 transversal relaxationVOI volume of interest

A bbr e v i At ions

1

i n t roduc t ion

Colorectal cancer is the second most com-mon malignancy in Norway, following can-cer of the prostate in men and the breast in women. Average age-adjusted incidence rates in the period 2000-2004 for cancer of the colon were 27.0 per 100 000 person-years in men and 23.9 in women, with corresponding rates for cancer of the rectum 15.9 in men and 10.5 in women. Approximately 3 400 pa-tients are diagnosed with colorectal cancer in Norway per year, and more than 1 600 deaths per year are attributed to this entity. The 5-year survival for localized colorectal cancer is 80-90%. The corresponding 5-year survival in patients with metastatic disease is below 10% [1]. Synchronous metastases, i.e. those recog-nized at the time of diagnosis of the primary tumor, are found in approximately 15-25% of patients with colorectal cancer. Metachronous metastases, those detected after diagnosis of the primary tumor, are found in additional 25-50% [2-5].

Metastasis (from Greek methistanai, change of place) is the result of a series of inter-related steps in which cancer cells leave the original tumor and migrate to distant organs via the blood or the lymphatic system [6, 7].

Different primary malignant tumors have a propensity to spread to specific organs [7]. The patophysiology of organ specificity in tumor metastasis can be considered in terms of two mutually not exclusive hypotheses: the “seed-and-soil” hypothesis and the “mechani-cal” hypothesis. The “seed-and-soil” hypothe-sis was proposed by Paget in 1889 and focuses on the interaction between the cancer cells and the specific organ microenvironment [8]. The “mechanical” hypothesis was proposed by Ewing in 1928, and suggests that the greater the number of tumor cells delivered to a spe-cific site, the more likely it is that a metastasis will develop at that site. The “mechanical” hypothesis is therefore mainly concerned with the anatomy of possible dissemination routes from the organ containing the primary malignancy [7, 9, 10]. Primary tumors in or-gans drained by the portal circulation have a higher probability of metastasis in the liver without metastasis elsewhere than primary tumors in organs drained by the systemic venous circulation [11, 12]. In line with the “mechanical” hypothesis, this may be related to the fact that the liver is the first visceral or-gan that malignant cells encounter following release into the portal circulation [13]. These

2

data suggest that survival may be improved in selected patients if metastatic disease in the liver can be controlled.

The first documented liver resection was performed on a trauma patient in 1716 [14]. Successful elective liver resection for a mass lesion was performed in 1888 [15]. A total of 59 cases of partial hepatectomy had been described worldwide by 1899 [16]. Resection for metastatic tumors was first described in 1940. Cancer metastastic to the liver was for a long time considered incurable, and systematic attempts to treat liver metastasis by surgical resection were not made until the 1960s. In a retrospective study from 1963, an average survival of 35 months was found in 25 patients with liver metastasis from dif-ferent primaries treated by surgical resection [17]. A 1978 literature review of more than 400 patients treated with liver resection for metastastic cancer from various primary tu-mors concluded that cure could be achieved, especially in patients with primary colorectal tumors [18]. Liver resection was associated with a mortality of nearly 30% in early series [14]. Several technical advances have gradual-ly transformed liver resection into its present state as a safe routine surgical procedure [19]. Due to improvements in surgical technique and intra- and postoperative patient support, the current operative mortality rates are less than 5% with a 5-year survival of 30-50% [20-22]. Recent studies have shown improved survival compared with historical series, with 5-year survival of 58% [23-25]. With the reduced perioperative mortality, a more

aggressive approach has emerged towards re-section of colorectal liver metastases. Repeat liver resections in patients with isolated liver recurrence after the first hepatectomy have shown survival rates comparable to those achieved by first hepatectomies [26-28]. Curative hepatectomy has been performed in selected patients with initially unresectable colorectal liver metastases after downsizing by chemotherapy [29]. However, at present only 10-15% of patients with colorectal liver me-tastases are candidates for potential curative hepatic resection [5, 29].

Evaluating the possible benefit of a treatment requires knowledge of the natural course of the disease. Patients with colorectal liver metastases accepted for surgical treatment represent a subset with limited disease and favorable prognosis. Patient selection bias may therefore present problems in interpret-ing the results of interventions in patients with limited hepatic colorectal metastases. Most studies of the natural history of malig-nant liver metastases are retrospective studies conducted in the 1960s and 1970s, without the advantage of modern imaging modalities. Median survival for patients with untreated liver metastases from colorectal cancer rarely exceeds 9 months [30, 31]. In a retrospective study of 252 patients with biopsy proven he-patic metastases that were not resected, 5-year survival was less than 3% [32]. A prospective study of 484 patients conducted between 1980-90 found a median 3-year survival of 0.9% [33]. Improved short-term survival, but no 5-year survival was found in 62 patients

3

with potentially resectable metastases that were not resected [34, 35]. Although the effi-cacy of liver resection versus no treatment has not been assessed in prospective randomized studies, it is generally accepted that resection of colorectal liver metastases improves long-term survival [21, 36, 37].

Surgical treatment is by its nature directed towards localized disease. Precise knowledge of the extent and anatomical location of ma-lignant disease constitute a fundamental pre-requisite for rational surgical management. Accurate preoperative staging is dependent on the sensitivity of the imaging modality used. Although the sensitivity of sonography, computed tomography (CT) and magnetic resonance imaging (MRI) is high for hepatic metastases above 1-2 cm diameter, hepatic metastases below 1 cm diameter can only oc-casionally be detected by these imaging mo-dalities. Thus, colorectal cancer patients may have more widespread disease than antici-pated from preoperative staging [38]. Several classification systems are currently used for staging of colorectal cancer. Dukes’ classifica-tion for rectal cancer, proposed in 1932 and modified in 1958 [39, 40], is widely used as a valuable prognostic indicator in patients with colorectal cancer. According to this classifica-tion, colorectal cancer is staged into A, B or C based on the tumor involvement of the wall of the colon and rectum. Dukes’ classification has later been expanded to include stage D for patients with distant metastasis. The TNM classification of malignant tumors was intro-duced by the International Union Against

Cancer in 1958 [41]. The TNM system pro-vide anatomic information on primary tumor, lymph node involvement and the presence or absence of distant metastasis. These staging systems do not separate between patients with resectable or non-resectable hepatic metastases. It has therefore been argued that new staging systems should be developed for patients with advanced disease [42].

Treatment of cancer can be provided as pal-liative or curative therapy. Palliative therapy is provided in the context that symptoms are present that can possibly be ameliorated by the treatment, but that eradication of the cancer is not expected. Curative therapy im-plies that a chance, albeit sometimes small, exists for complete eradication of the cancer. It is generally accepted that surgical resec-tion for malignant liver disease should only be performed with a curative intent [31]. Criteria used for evaluating if a patient with malignant disease confined to the liver is eligible for hepatectomy can be categorized as technical criteria, such as tumor proxim-ity to major intrahepatic vessels precluding safe resection, biological criteria assumed to be associated with inferior prognosis such as the number, size or distribution of tumors, or general criteria related to the physical condi-tion of the patient. Disease in patients select-ed to hepatectomy was previously limited to one to three unilobar metastases, preferably presenting metachroneously and resectable with a margin of 1 cm [42]. The definition of resectability has evolved, and the current defi-nition of resectability with curative intent is

4

based on the ability of the surgeon to achieve clearance of all measurable disease from the liver, while leaving a healthy future remnant liver of at least 20% of the total liver volume [42, 43]. Extrahepatic malignant disease has been considered an absolute contraindication to hepatectomy for colorectal liver metastases [21, 44], but this dogma has recently been challenged [45, 46].

Methods for local ablationi of malignant tu-mors were developed in order to achieve local tumor control and possibly increased survival in patients that could not be offered curative hepatic resection. Ablation techniques can be categorized into chemical or thermal meth-ods according to the physical principle used for induction of tissue damage. Early attempts in using cold to treat liver tumors were crude and involved direct application of liquid ni-trogen to tumors or to metallic instruments in contact with tumors [47]. Modern cryoabla-tion devices use nitrogen or argon gas under high pressure, which is led into a cryoprobe whose tip is cooled down to subzero tempera-tures. However, the inability to visualize deep liver tumors limited the use of local ablation until intraoperative sonography became avail-able in the late 1980s [48]. Several minimally- invasive methods producing localized heat such as microwave coagulation therapy [49],

laser ablation [50] and radiofrequency ab-lation [51] are currently in clinical use for treatment for hepatic tumors. High-intensity focused ultrasound offers a non-invasive ap-proach to tissue destruction, but has not yet come into clinical use for treatment of malig-nant liver tumors [52]. Several recent reviews present current methods for local ablation of hepatic tumors [53-57].

The therapeutic approach to colorectal hepat-ic metastases has changed dramatically over the last three decades, from an evaluation of whether surgical resection was indicated or not, to a complex multidisciplinary field in-volving surgeons, radiologists and oncologists offering an increasing number of treatment modalities [54, 58, 59]. Current treatment of colorectal hepatic metastases includes strate-gies for achieving hepatic resection in previ-ously non-resectable patients such as neoadju-vantii chemotherapy, preoperative portal vein embolization to increase the future remnant liver [60], 2-stage resection approaches [61] and ex situ resection with liver autotrans-plantation [62]. Non-resection treatments currently used are tumor ablation [63, 64], locoregional or systemic adjuvant chemo-therapy [65], local radiation by microspheres [66], external beam radiotheraphy [67], im-munotherapy [68] and gene therapy [69].

i From Latin ablatio, removal or destruction. In the context of this thesis used to describe methods used for local tissue destruction by heat, cold or chemical substances.

ii From Latin adiuvans, to help. Adjuvant chemotherapy: chemotherapy given after tumor surgery to prevent cancer recurrence. Neoadjuvant chemotherapy: chemotherapy given prior to surgical removal of a tumor to shrink the tumor, with the intent of making the patient operable or the surgical procedure less extensive.

5

Radiofrequency ablationiii is an electrosurgical method utilizing the flow of electrical current through tissue to dissipate heat. An electrode is placed inside the tumor, and tissue in the vicinity of the electrode is heated, producing localized tissue destruction [70]. This modal-ity has recently been introduced in the treat-ment of patients with non-resectable hepatic metastases [51, 71].

This study was carried out to investigate vari-ous experimental aspects of hepatic radiofre-quency ablation, as well as the clinical applica-tion of this technique in a well-defined patient population with colorectal liver metastases.

iii The term radiofrequency refers to the frequency of the alternating current used in these systems (300 - 500 kHz), which is within the radiofrequency spectrum.

6

7

ELEctrosurgicaL principLEs

Electrosurgery can be defined as the therapeu-tical use of heat dissipated by high-frequency electrical current passing through biological tissue. Electrosurgery can be performed with either monopolar or bipolar techniques; monopolar techniques are more commonly used for tumor ablation. For monopolar elec-trosurgery an asymmetrical electrode system is utilized where a small active electrode is placed in the tissue to be treated and a large dispersive electrode is placed on the skin of the patient. With the bipolar technique, two identical electrodes are placed in the tissue close to each other, and the current passes through tissue locally between the electrodes.

Fundamental quantities when describing an electrical circuit is the electrical potential (volt, V), the flow of electric charge per time unit (ampere, A), and the resistance through which the current passes (ohm, Ω). Electrical current may be unidirectional (direct current, DC) or may change direction (alternating cur-rent, AC). Alternating current can be charac-terized by the frequency of cycles per second (hertz, Hz). The opposition to alternating

current is denominated impedance, and is determined not only by the resistance, but also by the reactance, which is the capacitive and inductive components of the electrical circuit. The reactance and hence the imped-ance is dependent on the current frequency. The impedance is represented by the symbol Z, and is measured in ohm (Ω).

The relationship between the electrical po-tential, current and resistance in a resistive direct current circuit is expressed by Ohm’s law, where V is the potential difference, I is the current, and R is the resistance:

(1)

Power in a purely resistive circuit is represent-ed by the symbol P (watt, W), and is deter-mined by the potential and current:

(2)

By combining Ohm’s law (1) with equation (2), the power can also be expressed as:

(3)

r A diof r equ e nc y A bl At ion

V = R ⋅ I

P = I ⋅ V

P = I2 ⋅ R

8

The relationship between energy, current, re-sistance and time can be determined by Joule’s law where Q is energy (joule, J), I is current (A), R is electrical resistance (Ω) and t is the length of time the current is permitted to flow (sec):

(4)

The central concept in electrosurgery is that current passing through tissue will dissipate electrical energy as heat. Equation (4) shows that energy generated in an electrical circuit is a function of the current to the second power. The current passing through the active elec-trode, the tissue and the dispersive electrode is the same per time unit. Because the resistance of the metal electrodes is negligible, virtually no heat will be dissipated in the electrodes. Both the active and the dispersive electrode are therefore cold during electrosurgery.

Biological tissue consists of cells containing ionic electrolytes, surrounded by cell mem-branes with capacitive properties. The cells are surrounded by extracellular electrolytes and organized into organs enclosed by membranes with different electrical properties than the in-tracellular and extracellular electrolytes. The impedance of tissues is much higher than the impedance of electrodes [72]. Hence, current passing through tissue will dissipate heat.

The current density ( J), defined as the electric current per cross-sectional area (A/m2), is constant throughout a cable if the cross sec-

tion of the cable is constant and the material of the cable is homogeneous. When current-carrying electrodes are brought into contact with tissue in a closed circuit, current spreads out in the tissue volume. An idealized model of a spherical electrode positioned in a homo-geneous medium illustrate that the current spread out from the electrode in the shape of a sphere:

(5) where J is the current density, I is the current passing through the electrode, r is the dis-tance from the electrode, a is the radius of the electrode.

In a monopolar system, the active electrode is small, leading to a high current density. In contrast, the dispersive electrode is large in order to permit the limited heat produced by the low current density to be dissipated by conduction [73]. Assuming adiabatic condi-tions, i.e. that no heat is transferred to or from the tissue (for instance by blood flow), and no phase difference between the current density and the electric field, the temperature rise in a tissue volume is a function of the current den-sity to the second power:

(6)

where ∆T is the temperature rise, J is the cur-rent density, t is the time, σ is the tissue con-ductivity, c is specific heat capacity and ρ is the density.

Q = I2 ⋅ R ⋅ t

J = I4 ⋅ π ⋅ r2 r ≥ a

∆T = J 2

σt

cρ

9

The power density Wv (watt/cubic metres) is an expression of the electrical energy per time unit that tissue is exposed to under ideal conditions [74]. The power density falls off extremely rapidly with distance from the elec-trode, as it is inversely proportional to the dis-tance from the electrode to the fourth power:

(7)

where σ is the tissue electrical conductivity, V the voltage, a the radius of the electrode, and r the distance from the electrode.

Although the relationship between the tem-perature rise in a tissue volume and the current density described by equation (6) is valid for an adiabatic model system, additional factors such as heat loss and metabolic tissue heating influence tissue temperature in vivo. The bio-heat equation (Appendix I) is a mathematical model that incorporates these factors [76]. The general concept of the bioheat equation can be simplified to [55]:

extent of coagulation = energy deposited ×

tissue interactions – heat loss

This model can be used to predict the extent of tissue coagulation produced by thermal energy under a variety of circumstances [77]. During thermal ablation in a given tissue vol-ume in vivo, increase of coagulation volume can be achieved by increasing energy deposi-tion or by reducing heat loss. During mono-polar electrosurgery, the electrode-to-tissue interface at the active electrode is exposed

to a high power density which may cause a rapid temperature rise. Consequently, tissue in proximity to the active electrode is suscep-tible to carbonization. Tissue carbonization leads to increased tissue impedance and an abrupt fall in the flow of electrical current, which limits tissue heating and thereby the extent of the coagulation [75]. The balance between obtaining a sufficient power density at a distance from the electrode while avoid-ing tissue carbonization at the electrode-to-tissue interface is a fundamental biophysical limitation of electrosurgery which was noted as early as 1928 [73].

Wv = σV 2 a2

r4

10

tissuE rEsponsE to hEat

The tissue response to heat is a function of temperature and exposure time. Tumors are believed to be more thermosensitive than normal tissue, possibly due to inability to dissipate heat effectively [78-80]. Cellular homeostasis is maintained with a mild eleva-tion of temperature to approximately 40˚C. When temperatures are increased to 42-45˚C cells become more susceptible to damage by other agents such as chemotherapy and ra-diationiv [81]. Below a tissue temperature of 45˚C thermal changes are largely reversible [82]. With a temperature of 50-52˚C, cellu-lar death occurs within 4-6 minutes [70, 80, 83]. At temperatures above 60˚C near instan-taneous protein coagulation and irreversible cellular damage occurs [55]. The liquid in the tissue evaporates at temperatures above 90˚C, resulting in desiccation if the tissue is heated slowly, or vaporization if the heat is delivered rapidly [84, 85]. Carbonization of the tissue occurs at temperatures above 105˚C.

The aim of radiofrequency ablation is to achieve and maintain a temperature of 50–100°C throughout the entire target volume. Thermally treated tissue is associated with different findings on postprocedural imaging (Appendix II). The term “coagulation” is used in this thesis to describe the localized tissue alterations caused by thermal ablation.

hEpatic radioFrEQuEncy abLation: a concEptuaL history

development of electrosurgery

The term electricity was introduced in 1600 by the natural philosopher and physician William Gilbert to describe the phenomenon that rubbing furs against amber (Greek elek-tron) would cause an attraction between the two materials. Gilbert found that a number of materials had similar abilities, but errone-ously believed that the attraction was caused by a subtle effluvium that was released after rubbing had pushed away the air between the two materials [86]. The first device generat-ing static electricity was invented in 1663 by Otto von Guericke by using a rotating sphere of sulphur on a shaft. By 1740 electrostatic machines were in widespread use in Europe. In 1745 the Leyden jar was invented, which made it possible to store and discharge static electricity [87].

In the early eighteenth century, dissected frogs were regularly used to explore the ef-fects of static electricity. While investigat-ing these effects the Italian physicist Luigi Galvani accidentally discovered in 1780 that muscle spasms were induced in frog legs after the sciatic nerve was brought into contact with two dissimilar metals. Galvani believed that the muscle contractions were caused by

iv The use of loco-regional or whole body hyperthermia with a temperature of 42-44˚C to sensitize tumor cells to the cytotoxic effects of ionizing radiation and chemotherapy should not be confused with thermal ablation, which require temperatures > 50˚C.

11

a disturbance of the equilibrium of electrical fluid carried to the muscles by the nerves. He coined the term animal electricity to describe what was believed to be a new form of elec-tricity [85, 86]. This discovery and subsequent experiments led to the concept of electrical current and the birth of bioelectricity. In 1799 Allesandro Volta constructed the first device that produced continuous electrical current by a chemical reaction, thereby showing that this assumed new form of electricity could be generated independently of animals. The de-vice produced by Volta, known as the voltaic pile, is the progenitor of modern batteries. The current produced by this device was known as Galvanic current, corresponding to the modern term direct current. Michael Faraday discovered electromagnetic induction in 1831 and thereby the principles necessary for construction of the dynamo, which was able to deliver alternating current.

Exposing living organisms to direct current or low-frequency alternating current is as-sociated with neuromuscular stimulation and an involuntary tetanic response. In 1891 Jacques-Arsène d’Arsonval discovered that by using frequencies above 10 kHzv, electrical current could be passed through tissue with-out producing other physiological effect than that of heat. Based on these findings, an ap-paratus was built that produced a spark across a gap, which could be used to generate super-ficial tissue destruction [88-90]. Carl Franz

Nagelschmidt developed the theory that creation of heat observed during treatment was caused by molecular agitation induced by high frequency electrical currents, and intro-duced the concept diathermy in 1897 (from Greek dia, through; therma, heat) [86, 90]. The first surgical use of electrical current is believed to be the treatment of a carcinoma-tous ulcer of the hand by the French physician Joseph Rivére in 1900 [91]. During the next decade, the use of electricity in treating lesions of the skin, oral cavity as well as coagulation of vascular tumors and hemorrhoids became widespread [85]. Pozzi introduced the term fulguration (from Latin fulgur, lightning) re-ferring to the superficial carbonization result-ing when an electrical spark was used to treat skin [92].

In 1909 Eugéne-Louis Doyen introduced the term electrocoagulation (from Latin coagulare, to curdle) to describe a method where the tis-sue was touched directly with the treatment electrode and a neutral electrode was attached to the patient. This arrangement produced deep tissue coagulation rather than carbon-ization on the tissue surface [93, 94]. In 1911 William Lawrence Clark used a high voltage spark gap oscillator to remove skin tumors. By microscopy of the treated tissue he observed that it shrunk from dehydration, and used the term desiccation (from Latin desiccare, to dry out) to describe this process [95]. In 1923 George A. Wyeth used electrosurgery for cut-

v It has later been shown that nerve and muscle exitation can be elicited at frequencies well above 10 kHz, but that increasingly higher current density is needed to elicit such a response with increasing frequency (LaCourse, J.R., et al., Effect of high-frequency current on nerve and muscle tissue. IEEE Trans Biomed Eng, 1985. 32(1): p. 82-6.)

12

ting tissues, utilizing an apparatus he termed an endotherm knife (from Greek endo, within; thermé, heat) [96]. In 1925 Ward discovered that a continuous sine wave from a vacuum tube oscillator was effective for cutting tis-sue, whereas coagulation was achieved by a damped sinusoid from a spark gap oscillator [97].

These discoveries inspired the physicist William T. Bovie to develop a device that could be used both for cutting and for ob-taining hemostasis by tissue coagulation (Fig 1). In 1926 the device, known as the Bovie unit, was used by the neurosurgeon Harvey Cushing for surgical treatment of patients with intracranial tumors that had previously been considered inoperable [73, 94]. The Bovie unit consisted of an alternating current

generator, interchangeable small active elec-trodes mounted on an insulated pistol-grip of bakelite, and a large dispersive electrode placed on the skin of the patient. By varying the modulation of the current used, the unit was capable of superficial dehydration, cut-ting or coagulation of a limited tissue volume [73]. Modern electrosurgical units are based on the electrophysiological principles of the Bovie unit.

development of hepatic radiofrequency ablation

Radiofrequency ablation was suggested for treatment of hepatic malignancies in 1990 [98, 99]. The same year an experimental study using a needle electrode with 10 mm non-insulated tip showed that it was possible to produce an ellipsoid coagulation in porcine liver with maximal diameter perpendicular to the electrode axis of 1.4 cm [99]. In 1992 percutanous ultrasound guided radiofre-quency ablation was performed in 10 pigs, with typical coagulation diameters of 1-2 cm [100]. The first publication presenting hepat-ic malignancies treated with radiofrequency ablation appeared in 1993, and presented 13 patients with hepatocellular carcinoma. The unit which was used achieved a coagulation volume of 1.8 mL [101], corresponding to a coagulation diameter of 1.5 cm. In a 1995 study using needle electrodes connected to a generator commonly used for cardiac abla-tions, various combinations of electrode tip exposure, gauge, duration of treatment and temperature were examined to assess the op-

Figure 1. The Bovie electrosurgical unit, manufactured from 1926 by the Liebel-Flarsheim Co, Cincinnati, OH.

13

timal settings for generating the largest pos-sible tissue coagulation volume. The authors concluded that coagulations larger than 1.6 cm in diameter could not be produced by any combination of factors with a needle elec-trode [102]. Numerous patient series were published in the second half of the 1990s [103-110]. The populations treated were of-ten heterogeneous with both primary hepatic tumors and metastases from various primary tumors. The majority of early reports are ob-servational studies with short follow-up and missing data on long-term patient survival.

Established oncological criteria suggest that hepatic malignancies should be eradicated radically including a 1-cm margin of appar-ently healthy tissue to eliminate microscopic

foci of malignant cells, and to compensate for the uncertainty in determining the exact tumor margin [111, 112]. In order to adhere to the principle of a 1-cm tumor-free margin, a tumor must be completely enveloped by a continuous coagulation that encompasses the tumor as well as a 1-cm margin on all sides. The diameter of a spherical coagula-tion would therefore need to be at least 2 cm larger than the largest diameter of the tumor (Fig 2) [113]. This implies that a coagulation volume several times the tumor volume must be created (Fig 3). The needle electrode was not capable of generating adequate coagula-tion volumes with one electrode insertion [55, 102]. Furthermore, the use of overlap-ping coagulations is not an effective method to increase coagulation volume (Fig 4) [113].

Figure 2. The figure shows an electrode placed in a tumor. In order to obtain a 1-cm tumor-free margin, the tumor with a diameter of 2 cm requires a tissue coagulation with a diameter of at least 4 cm.

Figure modified after Rhim, H., et al., Essential techniques for successful radio-frequency thermal ablation of malignant hepatic tumors. Radiographics, 2001. 21 Spec No: p. S17-35.

Figure 3. The figure shows the tumor volume (filled) and the volume of the corresponding 1-cm margin. The volume of a 2-cm spherical tumor is 4.2 mL, the volume of the surrounding 1-cm tumor free margin (excluding the tumor) is 29.3 mL. The required coagulation volume to treat a 2-cm spherical tumor is therefore 33.5 mL, approximately 8-fold the tumor volume. Present radiofrequency ablation systems produce coagulation volumes of approximately 40 mL.

14

Devices producing larger coagulations with one electrode insertion would therefore have to be developed to make radiofrequency ab-lation useful for treatment of most clinically relevant hepatic tumors (i.e. those measuring > 1-2 cm in diameter). Several strategies were proposed in the last half of the 1990s in or-der to increase the coagulation volume [114]. Although generators capable of adjusting power output according to tissue impedance and temperature were introduced, it was prin-cipally the rapid development in electrode design that enabled substantially larger tissue coagulations to be obtained.

Loop electrode.— The loop electrode consisted of a super-elastic metal with a distal loop with a radius of 1 cm, which was intro-duced into the tissue by a cannula. After the electrode had regained its loop shape, the electrode was rotated simultaneously with activation of the electrosurgical unit in cut mode. This produced a spherical coagulated rim of 1 mm which physically separated a non-coagulated spherical tissue volume with a diameter of 2 cm from the surrounding vital parenchyma [115, 116].

Bipolar electrode.—The use of bipo-lar electrodes, i.e. insertion of two identical electrodes in the target tissue, generates high current density at both electrodes and in tis-sue between the electrodes [117]. A bipolar system was found to produce larger coagula-tions than the needle electrode, but also more irregular coagulation geometry [118].

Cluster electrode.—The use of cluster electrodes, i.e. insertion of three or more elec-trodes that are activated simultaneously, was examined in 1995. Simultaneous application of energy to three electrodes placed 1.5 cm apart would lead to a coagulation diameter of 3 cm [119].

Internally cooled electrode.—The in-ternally cooled electrode was described in 1996 [120, 121]. This design consisted of a hollow electrode with two internal lumens extending to the electrode tip. One lumen delivered cooled saline to the electrode tip, and the other lumen returned the saline to a

2.0

cm

2.8 cm

4 cm

Figure 4. The figure illustrates the result of overlapping coagulations for a device generating a spherical tissue coagulation of 2 cm (thick line). The coagulation diameter can be increased by positioning four coagulations in the x-y plane (thin lines) and two additional coagulations along the z-axis (not shown). The positioning of six coagulations will produce a spherical coagulation with a diameter of 2.8 cm (dotted line). A coagulation with a diameter of 2 cm correspond to a volume of 4.2 mL, whereas a coagulation with a diameter of 2.8 cm correspond to a volume of 11.5 mL. Hence, increasing the number of coagulations by a factor of 6 only increase the spherical coagulation volume by a factor of 2.7.

Figure modified after Rhim, H. and G.D. Dodd, 3rd, Radiofrequency thermal ablation of liver tumors. J Clin Ultrasound, 1999. 27(5): p. 221-9.

15

collection unit outside the body. The perfus-ate did not come into contact with patient tissues. A thermocouple was embedded at the tip of the electrode to allow for continu-ous temperature measurement. By cooling the electrode tip, the risk of excessive heating and carbonization was reduced, enabling a higher current density and higher deposition of elec-trical energy in the tissue than was tolerated with conventional electrodes. Using this prin-ciple coagulation with a diameter of 2.5 cm could be made in porcine liver in vivo [120].

Perfusion electrode.— It was noted in 1990 that adding saline to the electrode-to-tissue interface would increase coagulation volume [98]. The explanation proposed for this finding was that installation of saline produced a more electrically uniform tis-sue, thereby limiting tissue carbonization. It was further hypothesized that a high local ion concentration at the electrode-to-tissue interface would increase the effective surface of the electrode and thereby the coagulation

volume. In 1997 the effect of continuous sa-line infusion during ablation was evaluated [105]. This study showed that radiofrequency ablation with a needle electrode produced co-agulations with a diameter of less than 1 cm in vivo, whereas continuous intraparenchymal infusion of 0.9% saline at 1 mL/min during ablation increased maximum coagulation di-ameter to 4 cm.

Multitined expandable electrode.—A device utilizing four expandable electrodes was presented in 1998 [122, 123]. The device consisted of an insulated shaft containing four retractable electrodes which could be de-ployed to a maximum diameter of 3 cm. Each electrode tip contained a thermocouple for temperature monitoring. Coagulations with diameters of 3.1 cm were created in an experi-mental study [122].

Commercially available hepatic radiofre-quency ablation systems have been developed based on the principle of the cluster electrode,

table 1. The four most commonly used systems for hepatic radiofrequency ablation.

Generator name Frequency (kHz) Power output (W) Electrode type

HiTT 1061 375 60 Perfusion

Cool-tip RF2 480 200 Internally cooled

RF 30003 480 200 Multitined expandable

1500x4 460 250 Multitined expandable

1 Integra LifeSciences, Tuttlingen, Germany, formerly Berchtold Tuttlingen, Germany 2 Valleylab, Boulder, CO, formerly Radionics, Burlington, MA3 Boston Scientific Corp., Boston, MA, formerly RadioTherapeutics Corp., Sunnyvale, CA4 RITA Medical Systems, Mountain View, CA

16

Figure 5. The figure shows the perfusion electrodes used in this study.

a. The diameter of the electrodes are 1.7 mm with a insulated shaft with a length of 15 or 20 cm incorporating a lumen for saline perfusion.

b. The terminal non-insulated 1.5 cm segment of the electrode has three pairs of two side holes, placed at an angle of 120˚ from each other.

c. Saline is distributed to the tissue around the electrode through the side holes.

the internally cooled electrode, the perfusion electrode, the multitined expandable electrode as well as combinations of these principles. Technical details of the four most commonly used systems for hepatic radiofrequency abla-tion are summarized in Table 1.

An impedance-controlled radiofrequency ablation system with perfusion electrodes (Elektrotom HiTT 106, Berchtold GmbH & Co, Germany) based on the principle of con-tinuous intraparenchymal infusion of saline during ablation was used in this study (Fig 5).

17

hEpatic radioFrEQuEncy abLation: cLinicaL chaLLEngEs

patient survival

Following the first report in 1996 of 6 pa-tients with colorectal liver metastases treated with radiofrequency ablation [104], larger series were published in 1999 and 2000 [110, 124, 125]. Follow-up was short and survival at standardized time intervals (i.e. 1, 3 and 5 years) was not presented. More recent patient series indicate that radiofrequency ablation of non-resectable colorectal liver metastases is associated with a 5-year survival of approxi-mately 30% [126-129].

Facilitated by downsizing as a result of neoad-juvant chemotherapy, and the ability of local ablation techniques to treat non-resectable tumors, patients with previously non-resect-able tumors are offered therapy with curative intent. The natural course of the disease in this novel and highly selected patient catego-ry is not known. Treatment is offered within a complex multimodal framework consisting of repeat surgical interventions, local abla-tion procedures and different adjuvant and neoadjuvant chemotherapy regimens. The interventions may be implemented in varying order and at non-standardized time points. No randomized studies comparing radio-frequency ablation with hepatic resection or with chemotherapy in patients with colorec-

tal hepatic metastases have been performed to date. Although preliminary data indicate that improved survival can be achieved in selected patients with non-resectable colorectal he-patic metastases treated with radiofrequency ablation, formal scientific proof is lacking.

Local tumor control

Complete eradication of malignant growth is considered mandatory to achieve increased survival in patients with colorectal liver me-tastases [35]. This assumption constitutes the oncological rationale for local ablation of non-resectable hepatic metastases. However, local tumor control, i.e. lack of tumor recur-rence at the site of treated tumors, is not a sufficient condition for increased patient sur-vival, as intrahepatic recurrence not related to the treatment site and extrahepatic recurrence is common in patients treated with R0-re-sectionvi [130] and radiofrequency ablation [24, 131] of colorectal liver metastases. Local tumor control should therefore be considered a surrogate endpoint, but is nevertheless a valuable indicator of the technical ability to eradicate a tumor.

Local tumor recurrence rates after radiofre-quency ablation are higher than after hepatic resection, with reported rates varying from less than 5% to above 50% [110, 125, 132-134]. A recent meta-analysis of 5227 liver tumors treated with radiofrequency ablation found that tumor diameter > 3 cm, perivas-

vi R0-resection implies that no microscopic residual tumor is present in the resections margin(s), and no residual macroscopic malignant disease is detected.

18

the treatment site. Temperature sensors are attached to or embedded in the ablation elec-trodes of most commercial radiofrequency ablation electrodes. However, thermal infor-mation is restricted to a predefined number of measuring points, and provide no informa-tion on tissue temperatures at a distance away from the electrode. Non-invasive temperature measurement is an attractive concept with the potential to visualize the three-dimensional temperature distribution during treatment [145].

Cooling caused by blood flow in large vessels passing through or near the treatment site is an important limitation of the extent of the coagulation in vivo [122, 146]. Several studies have examined the effect of temporary reduc-tion of intrahepatic perfusion by pharmaco-logical modulation [147, 148] or temporary mechanical occlusion of hepatic vessels [146, 149-152]. Interruption of hepatic inflow dur-ing ablation increases efficiency [153], and enables coagulations that are larger, with a more spherical geometry [149, 151], which would be beneficial in order to achieve com-plete treatment.

assessment of local tumor recurrence

Malignant tissue treated by radiofrequency ablation is left in situ, necessitating repeated imaging follow-up to determine if the tumor has been eradicated. Early detection of local tumor recurrence may facilitate early reinter-vention with potential benefits for patient survival. The use of fine needle aspiration cy-

cular tumor localization and multiple tumors were independent predictors for local tumor recurrence. Thus, the ability of radiofre-quency ablation to eradicate liver tumors is limited by a number of tumor-related factors. Furthermore, higher local recurrence rates were found for procedures using a percutane-ous approach [135].

In order to attain reliable tumor destruction by radiofrequency ablation, knowledge of the expected coagulation geometry and volume as well as the relationship of the coagulated volume to the position of the electrode is crucial. A systematic review of the literature from 1990 to 2003 found that adequate ex-perimental data were missing for most of the commercially available electrodes [136]. Reproducibility of coagulation geometry and volume is low under standardized treatment protocols in experimental studies [137, 138] and in clinical series [139]. Reliable intra-procedural monitoring of the extent of tissue coagulation is therefore essential to ensure ad-equate treatment of a tumor and to avoid dam-age to abutting structures. Transabdominal sonography provide accurate guiding of the ablation electrode, but is unreliable in assess-ment of the extent of tissue coagulation [100, 140]. The use of CT [141], MR [142, 143] and contrast enhanced sonography [144] have been evaluated for intraprocedural mon-itoring of radiofrequency ablation. Real-time monitoring of the temperature distribution during the procedure would allow the op-erator to optimize the treatment and possibly avoid damaging normal structures adjacent to

19

induced in the treated tumor and surround-ing hepatic parenchyma have not been well characterized due to the fact that the treated tumors and surrounding parenchyma are left in situ. It has been hypothesized that radiofre-quency ablation generates an antigen source that may induce antitumor immunity [167], which may be a beneficial effect. However, of special interest are recent experimental find-ings that radiofrequency ablation strongly promotes proliferation of residual neoplastic cells [168]. Examination of biological effects induced by radiofrequency ablation may be of importance for understanding mechanisms for post-ablation tumor recurrence, and should be further characterized.

tology is controversial due to the risk of needle tract seeding [154, 155]. Consequently, non-invasive methods should preferably be used for detection of persistent or residual malig-nant tissue at the coagulated site. However, it can be difficult to distinguish benign tissue alterations caused by radiofrequency ablation from residual tumor on post-ablation follow-up imaging [156-158]. Non-invasive meth-ods for assessment of local tumor recurrence should therefore be refined.

complications

Two large retrospective studies have assessed complications associated with hepatic radio-frequency ablation [159, 160]. The reported mortality in both studies was below 1%, with complications occurring in less than 10% of patients. Complications directly related to the application of thermal energy such as damage to bile ducts or intrahepatic vessels occurred in less than 1% of patients. In one of these studies of 3670 patients, 3 of 20 deaths was attributed to portal vein thrombosis and 1 death to bile duct stricture [159]. The risk of damage to intrahepatic vessels and bile ducts [150, 161-163], as well as the feasibil-ity of protecting bile ducts by intraductal cooling have been examined in recent studies [164-166].

biological alterations at the coagulated site

Although the gross and microscopic find-ings after hepatic radiofrequency ablation are well characterized [140], biological effects

20

21

This thesis examines the use of hepatic radiofrequency ablation in four experimental studies in non-tumor animal models and in one clinical study in patients with colorectal liver metastases. An impedance-controlled perfusion electrode radiofrequency ablation system was used in all studies. Specifically, the aims were:

1. Characterize coagulation geometry and coagulation volume generated in porcine liver during maintained and interrupted hepatic inflow.

2. Investigate the feasibility of a thermosensitive liposomal agent for thermal monitoring of MRI-guided hepatic radiofrequency ablation in rabbit liver.

3. Determine if electrode-to-vessel distance or the use of hepatic inflow occlusion are as-sociated with acute or delayed portal vein thrombosis in porcine liver.

4. Examine if hepatic radiofrequency ablation increases the expression of matrix metallo-proteinase-2 and matrix metalloproteinase-9 in the transition zone separating coagulated tissue from normal porcine hepatic tissue.

5. Determine if alterations in coagulation volume on post-ablation CT images acquired every third month facillitate early detection of local tumor recurrence in patients with colorectal hepatic metastases treated with radiofrequency ablation.

A i m s of t h e s t u dy

22

23

s u m m A ry of pA pe r s

I Frich L, Mala T, Gladhaug IP. Hepatic radiofrequency ablation using perfusion electrodes in a pig model: Effect of the Pringle manoeuvre. Eur J Surg Oncol. 2006;32:527-32.

Coagulation geometry, coagulation volume and delivered energy per coagulated tissue volume was assessed in a non-tumor porcine model. Mechanical occlusion of the portal vein and the hepatic artery (Pringle maneuver) during ablation was performed in 6 of 12 animals. One coagulation was made in each animal close to the portal vein. All animals were sacrificed 4 days after ablation. Use of hepatic inflow occlusion increased the ef-fective coagulation diameter by 1 cm, but was associated with elongated coagulations extending outside the expected treatment site. Displacement of the geometrical centre of the coagulation relative to the position of the electrode was observed. The energy deliv-ered during ablation could not be used to predict coagulation volume.

24

II. Frich L, Bjørnerud A, Fossheim S, Tillung T, Gladhaug I. Experimental application of thermosensitive paramagnetic liposomes for monitoring magnetic resonance imaging guided thermal ablation. Magn Reson Med. 2004;52:1302-9.

The utility of a thermosensitive liposomal agent to assess the extent of tissue coagulation during MRI-guided laser ablation and radiofrequency ablation in non-tumor rabbit liver was examined. One coagulation was made in each animal prior to administration of the agent, and two additional coagulations were made after administration of the agent. Tissue temperature during treatment was measured by a MRI-compatible temperature fiber. T1-weighted MR images were acquired prior to heating, during heating and after heating for each coagulation. Signal intensity (SI) of the coagulations generated prior to injection of the agent decreased 7% below baseline during heating, but increased 5% above baseline at normalization of tissue temperature. Both coagulations generated after administration of the agent showed increased SI during heating (5% vs. 19%), followed by a further increase after tissue temperature had normalized (14% vs. 26%). Increased SI was found for coagulations made 50-55 min after injection of the agent compared to those made 12-14 min after injection of the agent. Persistent signal enhancement on T1-weighted MR images was found in areas assumed to be exposed to a temperature above the phase transition temperature of 57˚C, indicating that thermosensitive liposomes may be used for monitoring of radiofrequency ablation.

25

III Frich L, Hol PK, Roy S, Mala T, Edwin B, Clausen OPF, Gladhaug IP. Experimental hepatic radiofrequency ablation using wet electrodes: electrode-to-vessel distance is a significant predictor for delayed portal vein thrombosis. Eur Radiol. 2006;16:1990-9.

Possible explanatory variables associated with acute and delayed portal vein thrombosis after hepatic radiofrequency ablation close to portal vessels were assessed in an experi-mental study. Coagulations were created within 1.5 cm of the right portal vein branch in 12 pigs with (n=6) or without (n=6) hepatic vascular occlusion (Pringle maneuver). Sham operations with Pringle maneuver were performed in four animals. Vessel patency was assessed by rotational portal venography prior to ablation, 10 min after ablation and 4 days after ablation. Distance between the ablation electrode and the right portal vein branch was measured from 3-dimensional reconstructions of the portal venograms. No animals developed acute portal vein thrombosis. Delayed portal vein thrombosis oc-curred in 2 of 6 animals in the Pringle group, and in 3 of 6 animals in the group where coagulations were made during physiological hepatic circulation. All five occurrences of delayed portal vein thrombosis occurred in animals with a distance between the ablation electrode and the right portal vein branch of 5 mm or less. Delayed portal vein throm-bosis may occur after hepatic radiofrequency ablation during physiologic hepatic circula-tion. An electrode-to-vessel distance of < 5 mm was a significant predictor for delayed portal vein thrombosis.

26

IV Frich L, Bjørnland K, Pettersen S, Clausen OPF, Gladhaug IP. Increased activity of matrix metalloproteinase 2 and 9 after hepatic radiofrequency ablation. J Surg Res. 2006;135(2):297-304.

Activity of MMP-2 and MMP-9 in plasma and in tissue lysates from the transition zone surrounding coagulated hepatic tissue was examined in a non-tumor porcine model. Hepatic vascular occlusion was performed in 6 of 12 animals. MMP activity was quanti-fied by gelatin zymography. Cellular localization of MMPs was determined by immuno-histochemistry. MMP-2 and MMP-9 activity in tissue lysates from the transition zone was significantly increased compared to normal hepatic parenchyma with ratios of 3.0 and 2.6, respectively. MMP-2 and MMP-9 activity in plasma 1 h after radiofrequency ablation was significantly increased compared to baseline levels, with ratios of 1.2 and 1.5, respectively, but had normalized 4 days after radiofrequency ablation. Hepatic inflow occlusion during ablation did not inflence MMP activity. Expression of MMP-2 and MMP-9 was localized to macrophages in the transition zone.

27

V Frich L, Hagen G, Brabrand K, Edwin B, Mathisen Ø, Aaløkken TM, Gladhaug IP. Radiofrequency ablation of colorectal liver metastases: evaluation of local tumor recur-

rence by 3D volumetric analysis. Submitted.

In this clinical study, 11 patients with colorectal liver metastases were treated with radio-frequency ablation. Multi-detector computed tomography (MDCT) was performed at 1 and 3 months after radiofrequency ablation, and then at intervals of 3 months until 24 months, and each sixth month thereafter. A total of 81 post-ablation scans were analyzed, with a median follow-up of 27 months (17-43 months). The theoretical treatment margin 1 month after radiofrequency ablation was assessed retrospectively. Post-ablation coagula-tion volume was retrospectively determined using a semi-automatic three-dimensional (3D) method. Local tumor recurrence was found in 7 patients, and was detected median 9 months (6-21 months) after radiofrequency ablation. A positive margin was not achieved in 3 patients. In patients without local tumor recurrence, coagulation volume was less than 30% of baseline value at 24 months. In contrast, coagulation volume increased markedly in 6 of 7 patients with local tumor recurrence. Local tumor recurrence was identified simultaneously by conventional morphological criteria and semi-automatic 3D volumetric analysis. Semi-automatic 3D volumetric analysis provides added diagnostic information compared with conventional morphological evaluation, and may increase specificity of non-invasive diagnosis of local tumor recurrence after radiofrequency ablation of colorectal cancer liver metastases.

28

29

ExpErimEntaL studiEs

study i

Hepatic radiofrequency ablation systems were not compared in experimental studies using standardized treatment protocols until 2003 [137, 138]. All evaluated systems showed substantial variability in coagulation volume and geometry. Furthermore, coagulation vol-umes were higher in ex vivo experiments than in vivo, leading to the conclusion that ex vivo findings did not predict the in vivo efficacy of these systems [137]. Although the systems were able to produce coagulations with trans-verse diameters in the range 2.5 to 4 cm, the coagulations were smaller than expected and deviated from the assumed spherical geom-etry. The perfusion electrode system used in the present study was the most efficient [153] (i.e. requiring the least amount of electrical en-ergy per coagulated tissue volume), produced among the largest volumes of coagulated tis-sue [138], but with larger variations in coagu-lation geometry and coagulation volume than systems based on other principles [137, 138].

Active heating of tissue during radiofrequen-cy ablation is counteracted by heat loss caused by circulating blood, as shown by the bioheat equation (Appendix I). Temporary interrup-tion of hepatic blood flow during ablation would therefore be expected to influence co-agulation geometry and coagulation volume. Interruption of hepatic inflow during abla-tion is associated with two different effects. The term heat-sink effect refers to cooling by adjacent blood vessels with diameter > 3 mm, which may lead to irregular coagulation ge-ometry in proximity to the vessels [122, 169]. Additionally, parenchymal perfusion on the capillary level is reduced. In line with the bio-heat equation, reduction of heat loss caused by parenchymal perfusion would be expected to lead to a general increase in the coagulation volume [146]. Interruption of hepatic inflow during ablation with multitined expandable electrodes and internally cooled electrodes produced coagulations that were larger, less el-liptic and less distorted in experimental stud-ies [149, 151, 170] and clinical series [152, 171]. In contrast, the impact of interruption of hepatic inflow on coagulation geometry

di sc us sion

30

and coagulation volume made with the per-fusion electrode system had not been inves-tigated. In study I we examined coagulation geometry and coagulation volume produced by the perfusion electrode system in a porcine model with and without hepatic inflow occlu-sion (Pringle maneuver). The portal vein and the hepatic artery was mechanically occluded during ablation in 6 of 12 animals, based on the assumption that reduction of heat loss would lead to a coagulation geometry more in line with an ideal sphere as suggested by equation (7).

Our results confirmed that the use of a stan-dardized ablation protocol during physiologi-cal hepatic circulation did not produce pre-dictable coagulation geometry or predictable coagulation volumes. As expected, reduction of heat loss by Pringle maneuver was associ-ated with increased coagulation volume, which may be considered a beneficial effect. However, the coagulation geometry was char-acterized by elongated coagulations extend-ing outside the expected treatment site, and only a fraction of the increased coagulation volume contributed to a spherical shape. The explanation for the elongated coagulations observed in our study may be related to the effect of saline infusion during ablation. With the settings used in this experimental study, saline was infused at a rate of 105 mL/hour, corresponding to 16 mL during a 9 min co-agulation, which was larger than the mean coagulation volume of 10 mL. A recent clini-cal study confirmed that saline was distrib-

uted outside the planned treatment volume with the perfusion electrodes used in this study [172]. Thermal damage was observed to abutting organs, which may be caused by direct thermal damage from heated saline. Additionally, uneven intraparenchymal dis-tribution of saline may alter both the local current density and tissue conductivity, which would influence the tissue temperature, as seen from equation (6). After determination of the largest circle that could be inscribed in each slice from a coagulation, the term effec-tive coagulation diameter was assigned to the diameter of the largest of these circles. This entity was used to calculate the effective co-agulation volume, which is the volume of the largest sphere that could be inscribed into the coagulation volume. Knowledge of the largest spherical coagulation volume produced is of clinical interest as it would allow the operator to adequately coagulate a tumor without con-sidering variation in coagulation geometry relative to the axis of the electrode. In this study, determination of geometrical proper-ties were based on photographs of tissue slices from the coagulation which had been cut in parallel sections. A limitation of this method is that it is dependent on the plane of cutting, and that the accuracy decreases with increas-ing slice thickness. Nevertheless, the clinically important geometrical concepts introduced in this study are robust, easily comprehensible and may be applied for evaluation of geo-metrical properties of future ablation systems. The terminology to describe coagulation ge-ometry is still under development. Of interest,

31

standardization of the methods and terminol-ogy for describing coagulation geometry has recently been proposed [173].

Our findings indicated displacement of the geometrical centre of the effective coagula-tion volume relative to the position of the electrode. Inability to predict the position of the coagulated area relative to the electrode constitutes a fundamental limitation in the use of local ablation techniques (Fig 6). The ablation electrode was positioned close to the right portal vein branch in all animals. It is reasonable to assume that displacement of the geometrical center of the coagulation relative to the electrode may have been caused by the heat-sink effect [146, 174]. If this is the case one would expect displacement to be less pronounced in the animals where he-patic inflow occlusion was performed during ablation. Displacement of the geometrical

centre of the coagulation relative to the elec-trode could not be quantified reliably by the methods used in our study because the exact position of the ablation electrode was not known at reoperation 4 days after the ablation procedure. Lastly, due to the dependence on electrically conducting saline in generating a large coagulation volume, displacement of the coagulation volume may also be a limitation of the perfusion electrode design.

study ii

As shown in study I, coagulation geometry and volume is not predictable under stan-dardized experimental ablation protocols. Furthermore, coagulations are smaller and more irregular than expected in clinical series of patients treated with hepatic radiofrequen-cy ablation [139, 175]. Intraoperative visual-ization of the thermally induced necrosis may be beneficial to achieve complete treatment of the index tumor and to avoid damage to organs adjacent to the planned treatment site. Assessment of the extent of tissue coagula-tion by sonography, CT or MRI during and immediately after ablation is unreliable [55, 140, 141, 175]. Non-invasive thermometry is an attractive concept with the potential to visualize the three-dimensional temperature distribution created during ablation. Several methods have the potential to measure tem-perature non-invasively, including MRI-based thermometry, ultrasound thermometry, mi-crowave-based thermometry and electrical impedance tomography [176, 177]. MRI-

Figure 6. The figure shows an electrode placed in a tumor and the position of the anticipated coagulation diameter (dotted line). The actual position of the coagulation is indicated by a circle. Due to displacement of the coagulation, a part of the tumor is incompletely treated.

32

based methods can provide non-invasive ther-mal mapping of relative temperature changes with acceptable accuracy, spatial resolution and temporal resolution [145]. However, radiofrequency ablation generates significant electromagnetic interference in the frequen-cies used for transmission and reception in MRI, precluding the use of several MRI-based methods [178]. Temperature-sensitive MRI contrast agents represent a novel approach for thermal imaging by indicating an absolute temperature threshold.

In study II we examined the feasibility of a thermosensitive liposomal agent with a membrane phase transition temperature of 57˚C and a size of 110 nm to assess the extent of tissue coagulation during intra-procedural MRI-guided laser ablation and radiofrequency ablation. The liposomal mem-brane can be tailormade to a predetermined transition temperature by modulating the composition of membrane phospholipids. The liposomes used in our study contained gadodiamide (GdDTPA-BMA), which is a MRI contrast agent that acts by reducing the longitudinal relaxation time (T1) of tissue wa-ter protons. At physiologic temperatures the paramagnetic compound is retained within the liposome and does not influence the T1-relaxation of the tissue water protons due to slow transmembrane water exchange condi-tions. At temperatures above the transition temperature the T1-relaxation efficacy of the liposomal agent increases markedly either due to fast transmembrane water exchange

conditions or leakage of gadodiamide into the tissue. T1-shortening of the tissue water protons will be shown as high signal intensity (SI) on standard T1-weighted MR images. A transition temperature of 57˚C of the liposo-mal membrane was chosen because protein denaturation and irreversible cellular damage would be expected to have occurred above this temperature threshold [55]. Increased SI on T1-weighted MR images would therefore indicate the extent of irreversible tissue dam-age. In study II coagulations made after injec-tion of the agent showed slightly increased SI during heating, followed by a nonreversible increase in SI after the tissue had regained physiological temperature. We speculated that this may have been caused by leakage of GdDTPA-BMA from the liposome interior and subsequent entrapment due to coagula-tion of vessels. This mechanism has also been suggested by a recent study in which the fea-sibility of using these liposomes as indicators for local drug delivery during hyperthermia has been examined [179]. The diameter of the coagulations as measured from photographs of macroscopic specimens and as determined from MR images were inconsistent. This may possibly be explained by the fact that irre-versible thermal damage may have occurred at a temperature below the phase transition temperature of the agent [80]. Nevertheless, a consistent pattern in the diameters was not observed. This finding might have been caused by the introduction of saline into the tissue during treatment, which would be expected to decrease SI on T1-weighted images and lead

33

to local inhomogeneities in the concentration of the liposomal agent or of GdDTPA-BMA released in the tissue. Although the mecha-nisms responsible for our observed results are not completely understood, consistent results were found that confirm the ability of the liposomal agent to differentiate between tis-sue exposed to thermal treatment and normal liver tissue. The persistent signal enhancement in areas exposed to a temperature above the threshold temperature may be of clinical value during MRI-guided radiofrequency ablation.

study iii

The presence of large vessels in the proxim-ity of tumors treated with radiofrequency ablation is associated with a high rate of in-complete tumor destruction and local tumor recurrence, assumed to be caused by cooling effects [135, 169, 180]. Temporary reduction of flow in the portal vein and hepatic artery during radiofrequency ablation may enhance coagulation necrosis close to hepatic ves-sels [169]. However, as maintained flow in large vessels serve to protect the vessels from thermal damage [181], vessels may be more vulnerable to thermal injury and subsequent thrombosis if radiofrequency ablation is per-formed during interruption of hepatic vascu-lar inflow. Although segmental portal vein thrombosis may be asymptomatic, patients with a deficit in functional hepatic reserve may be at risk of postoperative hepatic fail-ure after portal vein thrombosis. The safety of performing radiofrequency ablation in

proximity to large intrahepatic vessels has not been established. It has been suggested that radiofrequency ablation should only be per-formed for tumors at least 2 cm from major intrahepatic vessels [105]. By adhering to this principle, potentially operable patients could be excluded from treatment. We therefore considered it important to examine factors associated with risk of damage to intrahepatic vessels during radiofrequency ablation in an experimental model.

In study III we examined risk factors associ-ated with portal vein thrombosis after hepatic radiofrequency ablation, with special atten-tion to the effect of temporary hepatic inflow occlusion and the electrode-to-vessel distance. The presence or absence of portal vein throm-bosis was examined by performing rotational portal venography prior to radiofrequency ablation, 10 min after radiofrequency abla-tion and 4 days after radiofrequency ablation. Consequently, we could determine by an objective method in the living animal when portal vein thrombosis occurred and the ana-tomical location of a portal vein thrombosis. Acute portal vein thrombosis was defined as thrombosis on portal venograms acquired 10 min after completion of the ablation, whereas delayed portal vein thrombosis was defined as thrombosis on portal venograms acquired 4 days after ablation. Delayed portal vein thrombosis occurred in 2 of 6 animals in the group in which hepatic inflow occlusion was performed, and in 3 of 6 animals in the group where coagulations were made during physi-

34

ological hepatic circulation. Hence, the use of temporary hepatic inflow occlusion was not associated with portal vein thrombosis. We found that an electrode-to-vessel distance of less than 5 mm was a significant predictor for delayed portal vein thrombosis. Our findings conflict with those of a study which reported that portal vein thrombosis invariably oc-curred when hepatic radiofrequency ablation was performed during vascular inflow oc-clusion, and that no portal vein thrombosis was found during physiological hepatic cir-culation [163]. These apparently diverging results may be caused by differences in study design and the method used for detection of portal vein thrombosis. The electrode-to-vessel distance, which was quantified by 3D portal venography in our study, proved to be of significance for the occurrence of portal vein thrombosis. Due to variations in electrical power and electrode designs used in different ablation systems, we consider our findings to be valid only for the specific radio-frequency ablation system used in this study. Even though the electrode-to-vessel distance may be of importance in regard to portal vein thrombosis in humans, the pig is considered to be more susceptible to activation of the co-agulation system following endothelial dam-age than humans [182]. Consequently, the electrode-to-vessel distance needed to induce post-ablation thrombosis in humans may be shorter than the 5 mm distance found to be of significance in our experimental study. Our results indicate that portal vein thrombosis may occur without temporary occlusion of

hepatic vessels. Another finding of clinical importance is that normal portal venography 10 min after radiofrequency ablation does not exclude later occurrence of portal vein throm-bosis. This finding may be of importance for postoperative evaluation of patients with as-sumed high risk of post-ablation portal vein thrombosis.

study iV

Several established risk factors for local tu-mor recurrence after radiofrequency ablation of hepatic metastases exist, such as tumor di-ameter > 3 cm and tumor proximity to major vessels [135]. Local tumor recurrence after radiofrequency ablation can be attributed to incomplete treatment of the index tumor or growth of micrometastatic disease around the tumor [181]. Micrometastatic disease in the vicinity of the macroscopic tumor has been detected in 2% to 70% of patients who underwent liver resection for colorectal liver metastases [183, 184]. These contradic-tory results possibly illustrate methodological limitations. Therefore, the true rate of micro-metastatic disease in this patient category re-mains uncertain. Patients treated with hepatic radiofrequency ablation have higher local, in-trahepatic and distant recurrence rates, as well as lower overall survival than comparable re-sected patients [24]. A reasonable explanation for these findings is negative patient selection bias, with micrometastatic disease being pres-ent in a larger proportion of patients treated with radiofrequency ablation. An alternative

35

explanation is that thermal ablation may in-duce biological alteration in the tumor tissue or peritumoral hepatic parenchyma that fa-cilitate spread and invasion of malignant cells. However, biological effects induced by radio-frequency ablation in the treated tumor and surrounding hepatic parenchyma are not well characterized due to the fact that the treated tumors and surrounding parenchyma are left in situ. In a recent experimental tumor model in mice, hepatic radiofrequency ablation pro-moted intrahepatic growth of residual neo-plastic cells [168]. The mechanisms explain-ing these findings is not well understood, although increased expression of growth fac-tors have been found after thermal ablation in experimental tumor models [185].

Matrix metalloproteinases (MMPs) are a family of matrix-degrading endopeptidases involved in a variety of physiological and pathological events [186-188]. The gelatinases MMP-2 and MMP-9 are involved in colorec-tal cancer tumor cell invasion, metastasis and angiogenesis [188]. In study IV we examined the activity of MMP-2 and MMP-9 in the transition zone separating the coagulated tis-sue and normal hepatic parenchyma after he-patic radiofrequency ablation in a non-tumor porcine model. MMP-2 and MMP-9 activity in tissue lysates from the transition zone 4 days after radiofrequency ablation was three-fold increased compared to normal hepatic parenchyma. Increased activity of MMPs probably represents an inflammatory response caused by radiofrequency ablation. However,

up-regulation of MMP activity may provide a microenvironment that facilitates the risk of local tumor recurrence and distant metastasis by vascular invasion in the presence of viable malignant cells. This mechanism would be of particular importance if a tumor is incom-pletely treated by radiofrequency ablation, leaving viable malignant cells, or if microme-tastases are present in or near the transition zone. Furthermore, hepatic radiofrequency ablation is increasingly combined with he-patic resection, which induces hematogenous dissemination of colorectal cancer cells [189, 190].

The data generated in this non-tumor pre-liminary animal study does not allow us to draw conclusions concerning the possible role of MMPs in local tumor recurrence, invasion or metastasis after radiofrequency ablation of colorectal liver metastases. However, based on the assumption that increased expression of MMPs in peritumoral tissue is associated with increased risk of growth and metastasis of malignant cells, our results indicate that he-patic radiofrequency ablation should only be performed when complete eradication of the target tumor including an adequate margin is possible.

36

cLinicaL study

study V

Hepatic radiofrequency ablation has been used for treatment of patients with non-re-sectable liver metastases from colorectal can-cer at our institution since 2003. Prior to in-troduction of hepatic radiofrequency ablation in clinical practice, a non-randomized phase II study was designed and approved by the Regional committee for medical research eth-ics. Written informed consent was obtained from all patients prior to radiofrequency ab-lation. All patients in the study were system-atically followed by multi-detector computed tomography (MDCT) of the liver at 1 and 3 months after radiofrequency ablation, and then at intervals of 3 months until 24 months, and each sixth month thereafter.

Non-invasive diagnosis of local tumor recur-rence is desirable due to the risk of needle tract seeding and negative impact on patient survival if biopsy is performed in operable pa-tients [191, 192]. However, it can be difficult to distinguish between local tumor recur-rence and benign tissue alterations on post-ablation follow-up imaging [156-158]. In study V our hypotheses were that inadequate treatment margins as determined from scans 1 month after radiofrequency ablation may identify patients who would develop local tumor recurrence, and that increase in coagu-lation volume during follow-up may be used to detect local tumor recurrence earlier than conventional morphological criteria. The

geometrical concepts of effective coagulation diameter and effective coagulation volume de-veloped in study I were applied in the analysis of the treatment results.

Based on previous published results from other groups, occurrences of local tumor re-currence at the treated site were expected to be diagnosed during follow-up within the first 6 months. However, local recurrence had also been reported to appear up to 23 months after the procedure [125, 193, 194]. In order to compare groups with and without local tu-mor recurrence, we retrospectively identified 11 patients who had either developed local tumor recurrence or who had been followed for at least 24 months without local tumor recurrence, assuming that local tumor recur-rence would not develop in this last category.

Local tumor recurrence was found in 7 of 11 patients. Intrahepatic recurrence not re-lated to the radiofrequency ablation site was found in 6 patients, and extrahepatic tumor growth in 6 patients. Of the 7 patients with local tumor recurrence, 4 patients presented with inoperable intrahepatic or extrahepatic tumors at the time of diagnosis of local tumor recurrence. Fine needle aspiration cytology of the area treated with radiofrequency ablation was performed in 4 patients during follow-up. In 3 patients malignancy was verified, whereas no malignant cells were found in 1 patient who did not develop local tumor recurrence. Malignant growth is common in this patient category despite successful treatment of the index tumor, with inoperable intrahepatic

37

or extrahepatic malignant growth occurring in 3 of 4 patients where the index tumor had been successfully treated with radiofrequency ablation. These findings are comparable to a previous study [24]. The largest coagulation diameter 1 month post-ablation was larger than the pre-ablation tumor diameter in all patients. However, when considering the ef-fective coagulation diameter 1 month post ablation, a positive treatment margin had only been achieved in 8 patients. The 3 pa-tients with an effective coagulation diameter smaller or equal to the largest tumor diameter developed local tumor recurrence. A nega-tive treatment margin may be of clinical use to identify patients who later develop local tumor recurrence. However, the prognostic value of this parameter should be examined in prospective studies.

In the patients who did not develop local tu-mor recurrence, coagulation volume decreased during follow-up, with all values below 30% of baseline values at 24 months. In contrast, coagulation volume increased significantly in the patients who developed local tumor recur-rence. Our findings suggest that the combina-tion of semi-automatic 3D volumetric analysis and conventional morphological evaluation may increase the specificity of non-invasive diagnosis of local tumor recurrence after ra-diofrequency ablation of colorectal cancer liver metastases, thereby possibly avoiding un-necessary biopsies. Furthermore, systematic post-ablation follow-up is associated with de-tection of treatable malignant disease, with 3 of 11 patients re-treated with curative intent.

We found a high rate of local recurrence in this study, which can partly be explained by the fact that the selection criteria used in this study leads to an overrepresentation of pa-tients with local tumor progression. Hepatic radiofrequency ablation using expandable and internally cooled electrodes increase parenchymal pressure [195], and it has been speculated that this may be a mechanism for spread of malignant cells to surrounding tis-sue. One would expect that intraparenchymal infusion of saline by the ablation system used in this study would lead to further increase in parenchymal pressure. Furthermore, study I confirmed that the perfusion electrodes pro-duced irregular and non-predictable coagula-tions, which may be unfavorable in regard to achieving adequate treatment margins. This was in line with our clinical experience, and is one of the reasons we no longer use the perfu-sion electrode system in clinical treatment at our institution. Finally, a learning curve exists for new procedures. Therefore, as this patient series represents our initial experience with radiofrequency ablation, a lower rate of local tumor recurrence would be expected with in-creasing experience.

Study V is a small clinical study. Some authors regard small clinical studies as unscientific and unethical [196]. On the other hand, one could argue that in rapidly lethal diseases, one should look for large effects in small trials rather than small effects in large trials [197]. This non-randomized study with a median follow-up of 27 months does not allow us to determine long-term survival effects of

38

radiofrequency ablation of colorectal liver metastases. However, of the 11 patients in-cluded in the study, 9 patients are alive with a median follow-up of 28 months (20-43 months). This compares favorably to recent studies of patients with metastastic colorectal cancer treated with chemotherapy, in which a median survival of 20-22 months has been achieved [198, 199]. However, the patients in our study is a highly selected group, with liver-only involvement which may have bet-ter prognosis than patients included in che-motherapy studies. Patient selection bias can therefore not be excluded as an explanation for the apparent short-term benefit of radio-frequency ablation. In 7 of the 11 patients in study V, liver resection had been performed previously. Therefore, a time-effect is also present, with these 7 patients having a median survival of 12 months (2-77 months) after development of metastatic disease to the liver prior to entering our study.

39

New technology that offers the promise of improved patient care is attractive, and may be adopted despite lack of evidence of either efficacy or superiority over existing proce-dures [200]. When hepatic radiofrequency ablation was introduced in the early 1990s, it was viewed as a palliative minimally-invasive experimental method [135]. Within a decade after its introduction, the use of hepatic radio-frequency ablation is increasing despite lack of formal scientific evidence.

In order to investigate the therapeutic poten-tial of the concept of local ablation of liver metastases, local tumor recurrence at the site treated with radiofrequency ablation should be minimized. This involves development of radiofrequency ablation systems capable of generating predictable tissue coagulations, and improved intra-procedural monitoring in order to secure adequate treatment margins. The ability of radiofrequency ablation to eradicate liver tumors is limited by a number of tumor-related factors [135] which should be taken into consideration when determin-ing which patients are eligible for this treat-ment modality. Novel diagnostic modalities

such as positron emission tomography (PET) may improve staging of malignant disease, which would facilitate selection of patients with localized non-resectable disease who may benefit from local tumor ablation [201].

It is important to assess whether high local control rates are of clinical significance in re-gard to long-term survival, since intrahepatic tumor recurrence not related to the area treat-ed with radiofrequency ablation or extrahe-patic tumor recurrence significantly limits the value of local tumor control. Furthermore, an unresolved apparent inconsistency exists in post-treatment recurrence rates in patients with colorectal liver metastases treated with hepatic resection or radiofrequency ablation, with significantly higher recurrence rates ob-served in patients treated with radiofrequen-cy ablation [24]. If these differences cannot be attributed to biological differences in the treated tumors, differences in recurrence rates must be attributed to the treatment, and not the disease [202]. Further exploration of the biological effects induced by radiofrequency ablation may provide insights in the underly-ing mechanisms of these findings.

f u t u r e pe r s pe c t i v e s

40

Long-term patient survival is the primary clinical endpoint in evaluation of potential curative interventions in cancer patients. Even though recent non-randomized series indicate that radiofrequency ablation of non-resect-able colorectal liver metastses is associated with a 5-year survival of approximately 30% [126, 129], these patients belong to a highly selected group treated within a multimodal framework. Patient survival in non-random-ized series cannot reliably be attributed to any single intervention, as the outcome may reflect patient selection bias and the cumu-lated effect of a multimodal treatment regime [51]. Randomized studies should therefore be performed to determine the possible impact of radiofrequency ablation on long-term sur-vival. However, randomization of resectable patients to radiofrequency ablation is consid-ered unethical, as superior long-term survival in well-defined patient populations is estab-lished for surgical resection [134, 202, 203]. Randomization of non-resectable patients to chemotherapy or radiofrequency ablation may be ethically acceptable. In 2002 a multi-center randomized phase III study examining the possible benefit of hepatic radiofrequency ablation was initiated by the European orga-nization for research and treatment of cancer (EORTC). The objective of the study was to examine the effect of performing radiofre-quency ablation in addition to chemotherapy by randomizing 390 patients with non-resect-able colorectal liver metastases to chemothera-py vs. chemotherapy and radiofrequency abla-tion. However, in this multicenter study with

nearly 50 participating centers, recruitment of patients was insufficient, and the study was redesigned to a randomized phase II study, with planned inclusion of 152 patients [204]. The lack of sufficient recruitment may reflect ethical or practical concerns of the individual physicians in charge of patient inclusion, or that patients refuse to volunteer for random-ization. Nevertheless, this study will prob-ably be able to determine if radiofrequency ablation provide an added survival benefit to chemotherapy.

Despite expanding indications for resec-tion of colorectal liver metastases, several recent studies have shown improved survival compared with historical series, with 5-year survival of 58% [23-25]. These studies may establish a new gold standard for long-term survival after resection of colorectal hepatic metastases against which other potentially curative treatments must be compared. Long-term results after radiofrequency ablation are promising, but do not currently match those achieved by resection. Radiofrequency ablation has been promoted by some as an inexpensive minimal-invasive alternative to resection [205]. However, the short-term benefits of less invasiveness should not be considered a valid argument for performing a treatment which may be less effective than established treatment. Therefore, based on currently available studies, only non-resect-able tumors should be treated with radio-frequency ablation [202, 206]. Considering the complexity of the multi-modal treatment

41

available to patients with colorectal liver me-tastases, radiofrequency ablation should only be performed in patients in which resection is not considered possible after evaluation by a multidisciplinary team including oncologists, radiologists and surgeons experienced in he-patic surgery [202].

42

43

conc lusions

1. The perfusion electrode system produces irregular coagulations during physiological hepatic perfusion. Mechanical occlusion of the portal vein and the hepatic artery during ablation increases the effective coagulation volume, but is associated with elongated coagulations extending outside the expected treatment site.

2. A liposomal thermosensitive agent showed persistent signal enhancement on T1-weight-ed MR images of tissue exposed to a temperature above the phase transition temperature of 57˚C, and may be of use during MRI-guided radiofrequency ablation.

3. Delayed portal vein thrombosis may occur after radiofrequency ablation during physi-ological hepatic flow. An electrode-to-vessel distance of < 5 mm was a significant predic-tor for delayed portal vein thrombosis.

4. Radiofrequency ablation of normal porcine liver is associated with significantly increased activity of matrix metalloproteinase-2 and -9 in the transition zone separating tissue coagulation from normal hepatic parenchyma. The expression of matrix metal-loproteinases was localized to macrophages in the transition zone.

5. Semi-automatic 3D volumetric analysis of post-ablation CT images acquired every third month does not facilitate earlier detection of local tumor recurrence than conventional morphological evaluation, but may increase specificity of non-invasive diagnosis of local tumor recurrence after radiofrequency ablation of colorectal cancer liver metastases.

6. Local tumor progression was found in a high proportion of patients with colorectal liver metastases treated with radiofrequency ablation.

44

7. Intrahepatic and extrahepatic malignant growth is commonly found during follow-up in patients treated with radiofrequency ablation despite successful treatment of the index tumor.

8. Systematic post-ablation follow-up may identify patients with malignant disease that can be offered re-resection or re-ablation in curative intent.

45

A ppe n di x i - t h e bioh e At equAt ion

In 1948 Pennes developed a mathematical model for heat transfer in perfused tissue based on temperature measurements in the human forearm [76]. Pennes suggested that the rate of heat transfer between blood and tissue is proportional to the product of the volumetric perfusion rate and the difference between the arterial blood temperature and the local tissue temperature. The equation proposed by Pennes is known as the bioheat equation [55]:

where ρ density of tissue, blood (kg/m3)c specific heat of tissue, blood (W sec/kg ˚C)k thermal conductivitym perfusion (blood flow rate/unit mass tissue) (m3/kg sec)Qp power absorbed per unit volume tissueQm metabolic heating per unit volume of tissue.

For the sake of this thesis, the most important contribution of the bioheat equation is to offer a biophysiological explanation of the effects of reduction of intrahepatic blood flow during radiofrequency ablation.

ρtct∂T(r, t) /∂t = ∇(kt∇T) − cbρb mρt (T − Tb ) + Qp (r, t) + Qm (r, t)

46

47

A ppe n di x i i - A not e on t e r m i nol o g y

The terminology used for description of radiofrequency ablation systems, electrodes, and postoperative image findings is inconsistent. Incompatible sets of terminology have been proposed by different groups [181, 207]. The terminology used in this thesis adheres to the proposal of the International working group on image-guided tumor ablation [181]. Due to the general evolvement of the terminology and the fact that authors are encouraged to implement the terminology used by a specific scientific journal, the terminology used in the papers in this thesis is inconsistent.

1. Radiofrequency ablation system and electrodes

A single radiofrequency ablation system was used for all studies in this thesis. The current generator is denominated “impedance-controlled”, whereas the system including the electrodes is denominated as “saline-enhanced radiofrequency ablation system”. The preferred term for the electrodes used in this study is “perfusion electrodes”, which correspond to the term “saline-cooled” used in paper II and “wet electrode” used in paper III. In the literature, the terms “open perfused system” and “Berchtold system” are occasionally used to describe this system.

2. Thermally treated tissue

Postprocedural imaging may identify zones with altered signal intensity (at magnetic resonance imaging), echogenicity (at sonography), or attenuation (at computed to-mography) that can be attributed to the thermal treatment. When describing findings on postprocedural images, the zone in which the induced treatment effect is present is called “coagulation” or “ablation zone”. This corresponds to the terms “lesion” or “abla-tion” used in paper II.

48

3. Failure of local therapy

A distinction is made between tumor growth within or at the site treated with radio-frequency ablation, intrahepatic foci of malignancy not related to the site treated with radiofrequency ablation, and extrahepatic tumor growth. The preferred terms for tumor growth within or at the site previously treated with radiofrequency ablation are “local tumor progression” and “local tumor recurrence”.

49

A ppe n di x i i i - A not e on A n i m A l mode l s

All animals were supplied by the Department of Comparative Medicine, Rikshospitalet University Hospital. Two different non-tumor animal models were used in this study.

In paper I, III and IV we used a pig model. The porcine liver is not morphologically similar to the human liver. Nevertheless, the pig is widely used as a model for hepatic thermal ablation. Furthermore, this model has proved to be valuable to our group in previous experimental re-search. A total of 26 pigs were operated. Eight pigs were included in a pilot study to establish the model. Of the consecutive 13 pigs who were randomized to hepatic vascular occlusion, one pig was euthanized within 2 hours of the primary operation due to respiratory distress attributed to complicated tracheal intubation. Consequently, 12 animals were successfully randomized, operated and reoperated 4 days later. Data from these 12 animals are included in paper I, III and IV. Additionally, 5 animals constituted a control group where only hepatic vascular occlusion but no radiofrequency ablation was performed. One of the animals in the control group was euthanized 2 days postoperatively due to symptoms of paralytic ileus. Data from the remaining 4 animals in the control group were used in paper III.

In paper II we studied a liposomal paramagnetic agent which was only available in limited quantities. In order to obtain adequate plasma levels, the experiments were performed in a small-animal model. The rabbit was chosen because it is a well-established liver model in magnetic resonance imaging research. A total of 14 rabbits were operated. Five rabbits were included in a pilot study to establish the model. Of the remaining 9 animals, one animal with preoperative weight loss was euthanized during the ablation procedure due to signs of distress. In the remain-ing 8 animals the procedure was successfully completed.

50

51

r e f e r e nc e s

1. Cancer in Norway 2004. 2006. 111 p.2. Bengmark, S. and L. Hafstrom, The natural history of primary and secondary malignant

tumors of the liver. I. The prognosis for patients with hepatic metastases from colonic and rectal carcinoma by laparotomy. Cancer, 1969. 23(1): p. 198-202.

3. Zavadsky, K.E. and y.T. Lee, Liver metastases from colorectal carcinoma: incidence, re-sectability, and survival results. Am Surg, 1994. 60(12): p. 929-33.

4. Bird, N.C., D. Mangnall, and A.W. Majeed, Biology of colorectal liver metastases: A review. J Surg Oncol, 2006. 94(1): p. 68-80.

5. Manfredi, S., C. Lepage, C. Hatem, O. Coatmeur, J. Faivre, and A.M. Bouvier, Epidemiology and management of liver metastases from colorectal cancer. Ann Surg, 2006. 244(2): p. 254-9.

6. Boedefeld, W.M., 2nd, K.I. Bland, and M.J. Heslin, Recent insights into angiogenesis, apoptosis, invasion, and metastasis in colorectal carcinoma. Ann Surg Oncol, 2003. 10(8): p. 839-51.

7. Fidler, I.J., The pathogenesis of cancer metastasis: the ‘seed and soil’ hypothesis revisited. Nat Rev Cancer, 2003. 3(6): p. 453-8.

8. Paget, S., The distribution of secondary growths in cancer of the breast. Lancet, 1889. 1: p. 571-3.

9. Ewing, J., Neoplastic diseases: a treatise on tumors. 3rd ed. 1928, Philadelphia: Saunders. 1127 p.

10. Weiss, L., Metastasis of cancer: a conceptual history from antiquity to the 1990s. Cancer Metastasis Rev, 2000. 19(3-4): p. I-xI, 193-383.

11. Pickren, J.W., y. Tsukada, and W.W. Lane, Liver metastasis: analysis of autopsy data, in Liver metastasis, L. Weiss and H.A. Gilbert, Editors. 1982, G. K. Hall Medical Publishers: Boston. p. 2-18.

12. Hess, K.R., G.R. Varadhachary, S.H. Taylor, W. Wei, M.N. Raber, R. Lenzi, et al., Metastatic patterns in adenocarcinoma. Cancer, 2006. 106(7): p. 1624-33.

52

13. Radinsky, R. and I.J. Fidler, The biology of liver metastasis, in Liver metastases: biology, diagnosis and treatment, O.J. Garden, J.G. Geraghty, and D.M. Nagorney, Editors. 1998, Springer-Verlag: London. p. 19-35.

14. Fan, M.H. and A.E. Chang, Resection of liver tumors: technical aspects. Surg Oncol, 2002. 10(4): p. 139-52.

15. Langenbuch, C., Ein Fall von Resektion eines linksseitigen Schurlappens der Leber. Berliner Klinische Wochenschrift, 1888. 25: p. 37.

16. Keen, W., Report of a case of resection of the liver for removal of a neoplasm. Ann Surg, 1899. 30: p. 267-69.

17. Woodington, G.F. and J.M. Waugh, Results of resection of metastatic tumors of the liver. Am J Surg, 1963. 105: p. 24-9.

18. Foster, J.H., Survival after liver resection for secondary tumors. Am J Surg, 1978. 135(3): p. 389-94.

19. Frich, L., T. Mala, B. Edwin, I. Gladhaug, O. Mathisen, and A. Bergan, Malignant liver tumors. A review of current surgical treatment options. Experience from a Norwegian hepatobiliary center. Gastroenterologia Polska, 2003. 10(4): p. 349-56.

20. Ekberg, H., K.G. Tranberg, R. Andersson, B. Jeppsson, and S. Bengmark, Major liver resection: perioperative course and management. Surgery, 1986. 100(1): p. 1-8.

21. Scheele, J. and A. Altendorf-Hofmann, Resection of colorectal liver metastases. Langenbecks Arch Surg, 1999. 384(4): p. 313-27.

22. Jarnagin, W.R., M. Gonen, y. Fong, R.P. DeMatteo, L. Ben-Porat, S. Little, et al., Improvement in perioperative outcome after hepatic resection: analysis of 1,803 consecu-tive cases over the past decade. Ann Surg, 2002. 236(4): p. 397-406.

23. Choti, M.A., J.V. Sitzmann, M.F. Tiburi, W. Sumetchotimetha, R. Rangsin, R.D. Schulick, et al., Trends in long-term survival following liver resection for hepatic colorectal metasta-ses. Ann Surg, 2002. 235(6): p. 759-66.

24. Abdalla, E.K., J.N. Vauthey, L.M. Ellis, V. Ellis, R. Pollock, K.R. Broglio, et al., Recurrence and outcomes following hepatic resection, radiofrequency ablation, and combined resec-tion/ablation for colorectal liver metastases. Ann Surg, 2004. 239(6): p. 818-825.

25. Pawlik, T.M., C.R. Scoggins, D. Zorzi, E.K. Abdalla, A. Andres, C. Eng, et al., Effect of surgical margin status on survival and site of recurrence after hepatic resection for colorec-tal metastases. Ann Surg, 2005. 241(5): p. 715-22.

26. Nordlinger, B., J.C. Vaillant, M. Guiguet, P. Balladur, F. Paris, P. Bachellier, et al., Survival benefit of repeat liver resections for recurrent colorectal metastases: 143 cases. Association Francaise de Chirurgie. J Clin Oncol, 1994. 12(7): p. 1491-6.

27. Bismuth, H., R. Adam, F. Navarro, D. Castaing, L. Engerran, and A. Abascal, Re-resec-tion for colorectal liver metastasis. Surg Oncol Clin N Am, 1996. 5(2): p. 353-64.

53

28. yan, T.D., K.Q. Lian, D. Chang, and D.L. Morris, Management of intrahepatic recurrence after curative treatment of colorectal liver metastases. Br J Surg, 2006. 93(7): p. 854-9.

29. Bismuth, H., R. Adam, F. Levi, C. Farabos, F. Waechter, D. Castaing, et al., Resection of nonresectable liver metastases from colorectal cancer after neoadjuvant chemotherapy. Ann Surg, 1996. 224(4): p. 509-20.

30. Bengtsson, G., G. Carlsson, L. Hafstrom, and P.E. Jonsson, Natural history of patients with untreated liver metastases from colorectal cancer. Am J Surg, 1981. 141(5): p. 586-9.

31. Penna, C. and B. Nordlinger, Surgery of liver metastases from colorectal cancer: new promises. Br Med Bull, 2002. 64: p. 127-40.

32. Wagner, J.S., M.A. Adson, J.A. Van Heerden, M.H. Adson, and D.M. Ilstrup, The natural history of hepatic metastases from colorectal cancer. A comparison with resective treat-ment. Ann Surg, 1984. 199(5): p. 502-8.

33. Stangl, R., A. Altendorf-Hofmann, R.M. Charnley, and J. Scheele, Factors influencing the natural history of colorectal liver metastases. Lancet, 1994. 343(8910): p. 1405-10.

34. Wilson, S.M. and M.A. Adson, Surgical treatment of hepatic metastases from colorectal cancers. Arch Surg, 1976. 111(4): p. 330-4.

35. Scheele, J., R. Stangl, and A. Altendorf-Hofmann, Hepatic metastases from colorectal carcinoma: impact of surgical resection on the natural history. Br J Surg, 1990. 77(11): p. 1241-6.

36. Wei, A.C., P.D. Greig, D. Grant, B. Taylor, B. Langer, and S. Gallinger, Survival after hepatic resection for colorectal metastases: a 10-year experience. Ann Surg Oncol, 2006. 13(5): p. 668-76.

37. Simmonds, P.C., J.N. Primrose, J.L. Colquitt, O.J. Garden, G.J. Poston, and M. Rees, Surgical resection of hepatic metastases from colorectal cancer: a systematic review of published studies. Br J Cancer, 2006. 94(7): p. 982-99.

38. Wallace, J.R., K.K. Christians, F.A. Quiroz, W.D. Foley, H.A. Pitt, and E.J. Quebbeman, Ablation of liver metastasis: is preoperative imaging sufficiently accurate? J Gastrointest Surg, 2001. 5(1): p. 98-107.

39. Dukes, C.E., The classification of cancer of the rectum. J Pathol Bacteriol, 1932. 35: p. 323-32.

40. Dukes, C.E. and H.J. Bussey, The spread of rectal cancer and its effect on prognosis. Br J Cancer, 1958. 12(3): p. 309-20.

41. Sobin, L.H. and C. Wittekind, TNM classification of malignant tumours (UICC). 6th ed. 2002, New Jersey: John Wiley & Sons. 272 p.

42. Poston, G., R. Adam, and J.N. Vauthey, Downstaging or downsizing: time for a new stag-ing system in advanced colorectal cancer? J Clin Oncol, 2006. 24(18): p. 2702-6.

54

43. Garden, O.J., M. Rees, G.J. Poston, D. Mirza, M. Saunders, J. Ledermann, et al., Guidelines for resection of colorectal cancer liver metastases. Gut, 2006. 55 Suppl 3: p. iii1-8.

44. Roh, M.S., Expanding the indications for hepatic resection in patients with colorectal liver metastases. Ann Surg Oncol, 2004. 11(3): p. 238-9.

45. Elias, D., L. Sideris, M. Pocard, J.F. Ouellet, V. Boige, P. Lasser, et al., Results of R0 resec-tion for colorectal liver metastases associated with extrahepatic disease. Ann Surg Oncol, 2004. 11(3): p. 274-80.

46. Elias, D., G. Liberale, D. Vernerey, M. Pocard, M. Ducreux, V. Boige, et al., Hepatic and extrahepatic colorectal metastases: when resectable, their localization does not matter, but their total number has a prognostic effect. Ann Surg Oncol, 2005. 12(11): p. 900-9.

47. Fong, y., L.H. Blumgart, and A.M. Cohen, Surgical treatment of colorectal metastases to the liver. CA Cancer J Clin, 1995. 45(1): p. 50-62.

48. Gage, A.A., History of cryosurgery. Semin Surg Oncol, 1998. 14(2): p. 99-109.49. Izzo, F., Other thermal ablation techniques: microwave and interstitial laser ablation of

liver tumors. Ann Surg Oncol, 2003. 10(5): p. 491-7.50. Vogl, T.J., K. Eichler, R. Straub, K. Engelmann, S. Zangos, D. Woitaschek, et al., Laser-

induced thermotherapy of malignant liver tumors: general principals, equipment(s), procedure(s)--side effects, complications and results. Eur J Ultrasound, 2001. 13(2): p. 117-27.

51. Tanabe, K.K., S.A. Curley, G.D. Dodd, A.E. Siperstein, and S.N. Goldberg, Radiofrequency ablation: the experts weigh in. Cancer, 2004. 100(3): p. 641-50.

52. Kohrmann, K.U., M.S. Michel, J. Gaa, E. Marlinghaus, and P. Alken, High intensity fo-cused ultrasound as noninvasive therapy for multilocal renal cell carcinoma: case study and review of the literature. J Urol, 2002. 167(6): p. 2397-403.

53. Scudamore, C.H., E.J. Patterson, A.M. Shapiro, and A.K. Buczkowski, Liver tumor abla-tion techniques. J Invest Surg, 1997. 10(4): p. 157-64.

54. Dodd, G.D., 3rd, M.C. Soulen, R.A. Kane, T. Livraghi, W.R. Lees, y. yamashita, et al., Minimally invasive treatment of malignant hepatic tumors: at the threshold of a major breakthrough. Radiographics, 2000. 20(1): p. 9-27.

55. Goldberg, S.N., G.S. Gazelle, and P.R. Mueller, Thermal ablation therapy for focal ma-lignancy: a unified approach to underlying principles, techniques, and diagnostic imaging guidance. AJR Am J Roentgenol, 2000. 174(2): p. 323-31.

56. Lau, W.y., T.W. Leung, S.C. yu, and S.K. Ho, Percutaneous local ablative therapy for hepatocellular carcinoma: a review and look into the future. Ann Surg, 2003. 237(2): p. 171-9.

57. Erce, C. and R.W. Parks, Interstitial ablative techniques for hepatic tumours. Br J Surg, 2003. 90(3): p. 272-89.

55

58. Curley, S.A. and R. Vecchio, New trends in the surgical treatment of colorectal cancer liver metastases. Tumori, 1998. 84(3): p. 281-8.

59. Ruers, T. and R.P. Bleichrodt, Treatment of liver metastases, an update on the possibilities and results. Eur J Cancer, 2002. 38(7): p. 1023-33.

60. Jaeck, D., E. Oussoultzoglou, E. Rosso, M. Greget, J.C. Weber, and P. Bachellier, A two-stage hepatectomy procedure combined with portal vein embolization to achieve curative resection for initially unresectable multiple and bilobar colorectal liver metastases. Ann Surg, 2004. 240(6): p. 1037-49.

61. Adam, R., A. Laurent, D. Azoulay, D. Castaing, and H. Bismuth, Two-stage hepatectomy: A planned strategy to treat irresectable liver tumors. Ann Surg, 2000. 232(6): p. 777-85.

62. Gruttadauria, S., J.W. Marsh, D.L. Bartlett, B. Gridelli, and A. Marcos, Ex situ resection techniques and liver autotransplantation: last resource for otherwise unresectable malig-nancy. Dig Dis Sci, 2005. 50(10): p. 1829-35.

63. Sotsky, T.K. and T.S. Ravikumar, Cryotherapy in the treatment of liver metastases from colorectal cancer. Semin Oncol, 2002. 29(2): p. 183-91.

64. Vogl, T.J., R. Straub, K. Eichler, O. Sollner, and M.G. Mack, Colorectal carcinoma me-tastases in liver: laser-induced interstitial thermotherapy--local tumor control rate and survival data. Radiology, 2004. 230(2): p. 450-8.

65. Lewis, A.M. and R.C. Martin, The treatment of hepatic metastases in colorectal carci-noma. Am Surg, 2006. 72(6): p. 466-73.

66. Herba, M.J. and M.P. Thirlwell, Radioembolization for hepatic metastases. Semin Oncol, 2002. 29(2): p. 152-9.

67. Malik, U. and M. Mohiuddin, External-beam radiotherapy in the management of liver metastases. Semin Oncol, 2002. 29(2): p. 196-201.

68. Sato, T., Locoregional immuno(bio)therapy for liver metastases. Semin Oncol, 2002. 29(2): p. 160-7.

69. Havlik, R., L.R. Jiao, J. Nicholls, S.L. Jensen, and N.A. Habib, Gene therapy for liver metastases. Semin Oncol, 2002. 29(2): p. 202-8.

70. Goldberg, S.N., Radiofrequency tumor ablation: principles and techniques. Eur J Ultrasound, 2001. 13(2): p. 129-47.

71. Decadt, B. and A.K. Siriwardena, Radiofrequency ablation of liver tumours: systematic review. Lancet Oncol, 2004. 5(9): p. 550-60.

72. Grimnes, S. and O.G. Martinsen, Electrical properties of tissue, in Bioimpedance and bioelectricity basics, S. Grimnes and O.G. Martinsen, Editors. 2000, Academic Press: San Diego, California. p. 87-125.

73. Cushing, H.W., Electro-surgery as an aid to the removal of intracranial tumors. With a preliminary note on a new surgical-current generator by W. T. Bovie. Surg Gynecol Obstet, 1928. 47: p. 751-84.

56

74. Grimnes, S. and O.G. Martinsen, Geometrical analysis, in Bioimpedance and bioelectric-ity basics, S. Grimnes and O.G. Martinsen, Editors. 2000, Academic Press: San Diego, California. p. 127-152.

75. Organ, L.W., Electrophysiologic principles of radiofrequency lesion making. Appl Neurophysiol, 1976. 39(2): p. 69-76.

76. Pennes, H.H., Analysis of tissue and arterial blood temperatures in the resting human forearm. J Appl Physiol, 1948. 1: p. 93-122.

77. Liu, Z., S.M. Lobo, S. Humphries, C. Horkan, S.A. Solazzo, A.U. Hines-Peralta, et al., Radiofrequency tumor ablation: insight into improved efficacy using computer model-ing. AJR Am J Roentgenol, 2005. 184(4): p. 1347-52.

78. Cavaliere, R., E.C. Ciocatto, B.C. Giovanella, C. Heidelberger, R.O. Johnson, M. Margottini, et al., Selective heat sensitivity of cancer cells. Biochemical and clinical stud-ies. Cancer, 1967. 20(9): p. 1351-81.

79. LeVeen, H.H., S. Wapnick, V. Piccone, G. Falk, and N. Ahmed, Tumor eradication by radiofrequency therapy. Responses in 21 patients. Jama, 1976. 235(20): p. 2198-200.

80. Storm, F.K., W.H. Harrison, R.S. Elliott, and D.L. Morton, Hyperthermic therapy for human neoplasms: thermal death time. Cancer, 1980. 46(8): p. 1849-54.

81. Alexander, H.R., Jr., Hyperthermia and its modern use in cancer treatment. Cancer, 2003. 98(2): p. 219-21.

82. Lepock, J.R., How do cells respond to their thermal environment? Int J Hyperthermia, 2005. 21(8): p. 681-7.

83. Goldberg, S.N., G.S. Gazelle, E.F. Halpern, W.J. Rittman, P.R. Mueller, and D.I. Rosenthal, Radiofrequency tissue ablation: importance of local temperature along the electrode tip exposure in determining lesion shape and size. Acad Radiol, 1996. 3(3): p. 212-8.

84. O’Connor, J.L. and D.A. Bloom, William T. Bovie and electrosurgery. Surgery, 1996. 119(4): p. 390-6.

85. Massarweh, N.N., N. Cosgriff, and D.P. Slakey, Electrosurgery: history, principles, and current and future uses. J Am Coll Surg, 2006. 202(3): p. 520-30.

86. Bren, S.P.A., Historical introduction to EMF health effects. IEEE Eng Med Biol Mag, 1996. 15(4): p. 24-30.

87. Grimnes, S. and O.G. Martinsen, History of bioimpedance and bioelectricity, in Bioimpedance and bioelectricity basics, S. Grimnes and O.G. Martinsen, Editors. 2000, Academic Press: San Diego, California. p. 313-9.

88. d’Arsonval, A., Action physiologique des courants alternatifs. C R Seances Soc Biol Fil, 1891. 43: p. 283-286.

89. d’Arsonval, A., Production des courants de grande intensité leurs effets physiologiques. C R Seances Soc Biol Fil, 1893. 45: p. 122-24.

57

90. LaCourse, J.R., A.D. Rothwell, and S.M. Selikowitz. Development of electrosurgery. I: A historical perspective. in The 1993 IEEE Nineteenth Annual Northeast. 1993. p. 35-6.

91. Riviére, A.J., Action des courants de haute fréquence et des effleuves du résonateur Oudin sur certaines tumeurs malignes. J Méd Interne, 1900. 4: p. 776-7.

92. Pozzi, M., Remarques sur la fulguration. Bull Assoc, 1909. Franc Cancer(2): p. 64-9.93. Doyen, D., Sur la destruction des tumeurs cancéreuses accessibles par la méthode de la

voltaisation bipolaire et de l’électro-coagulation thermique. Arch Ekec Med, 1909. 17: p. 791-5.

94. Pollack, S.V., A. Carruthers, and R.C. Grekin, The history of electrosurgery. Dermatol Surg, 2000. 26(10): p. 904-8.

95. Clark, W.L., The desiccation treatment of congenital and new growths of the skin and mucous membrane. JAMA, 1914. 63: p. 925-9.

96. Wyeth, G.A., The endotherm. American Journal of Electrotherapeutics Radiology, 1924. 42: p. 5.

97. Pearce, J.A., Electrosurgery. 1986, New york: John Wiley & Sons.98. McGahan, J.P., P.D. Browning, J.M. Brock, and H. Tesluk, Hepatic ablation using radio-

frequency electrocautery. Invest Radiol, 1990. 25(3): p. 267-70.99. Rossi, S., F. Fornari, C. Pathies, and L. Buscarini, Thermal lesions induced by 480 KHz

localized current field in guinea pig and pig liver. Tumori, 1990. 76(1): p. 54-7.100. McGahan, J.P., J.M. Brock, H. Tesluk, W.Z. Gu, P. Schneider, and P.D. Browning, Hepatic

ablation with use of radio-frequency electrocautery in the animal model. J Vasc Interv Radiol, 1992. 3(2): p. 291-7.

101. Rossi, S., F. Fornari, and L. Buscarini, Percutaneous ultrasound-guided radiofrequency electrocautery for the treatment of small hepatocellular carcinoma. J Intervent Radiol, 1993. 8: p. 97-103.

102. Goldberg, S.N., G.S. Gazelle, S.L. Dawson, W.J. Rittman, P.R. Mueller, and D.I. Rosenthal, Tissue ablation with radiofrequency: effect of probe size, gauge, duration, and tempera-ture on lesion volume. Acad Radiol, 1995. 2(5): p. 399-404.

103. Rossi, S., M. Di Stasi, E. Buscarini, L. Cavanna, P. Quaretti, E. Squassante, et al., Percutaneous radiofrequency interstitial thermal ablation in the treatment of small hepa-tocellular carcinoma. Cancer J Sci Am, 1995. 1(1): p. 73.

104. Rossi, S., M. Di Stasi, E. Buscarini, P. Quaretti, F. Garbagnati, L. Squassante, et al., Percutaneous RF interstitial thermal ablation in the treatment of hepatic cancer. AJR Am J Roentgenol, 1996. 167(3): p. 759-68.

105. Livraghi, T., S.N. Goldberg, F. Monti, A. Bizzini, S. Lazzaroni, F. Meloni, et al., Saline-enhanced radio-frequency tissue ablation in the treatment of liver metastases. Radiology, 1997. 202(1): p. 205-10.

58

106. Solbiati, L., S.N. Goldberg, T. Ierace, T. Livraghi, F. Meloni, M. Dellanoce, et al., Hepatic metastases: percutaneous radio-frequency ablation with cooled- tip electrodes. Radiology, 1997. 205(2): p. 367-73.

107. Lencioni, R., O. Goletti, N. Armillotta, A. Paolicchi, M. Moretti, D. Cioni, et al., Radio-frequency thermal ablation of liver metastases with a cooled-tip electrode needle: results of a pilot clinical trial. Eur Radiol, 1998. 8(7): p. 1205-11.

108. Cuschieri, A., J. Bracken, and L. Boni, Initial experience with laparoscopic ultrasound-guided radiofrequency thermal ablation of hepatic tumours. Endoscopy, 1999. 31(4): p. 318-21.

109. Jiao, L.R., P.D. Hansen, R. Havlik, R.R. Mitry, M. Pignatelli, and N. Habib, Clinical short-term results of radiofrequency ablation in primary and secondary liver tumors. Am J Surg, 1999. 177(4): p. 303-6.

110. Curley, S.A., F. Izzo, P. Delrio, L.M. Ellis, J. Granchi, P. Vallone, et al., Radiofrequency ablation of unresectable primary and metastatic hepatic malignancies: results in 123 pa-tients. Ann Surg, 1999. 230(1): p. 1-8.

111. Cady, B., R.L. Jenkins, G.D. Steele, Jr., W.D. Lewis, M.D. Stone, W.V. McDermott, et al., Surgical margin in hepatic resection for colorectal metastasis: a critical and improvable determinant of outcome. Ann Surg, 1998. 227(4): p. 566-71.

112. Blumgart, L.H., Liver resection for benign disease and for liver and biliary tumors, in Surgery of the liver and biliary tract, L.H. Blumgart and y. Fong, Editors. 2000, W. B. Saunders: London. p. 1639-1713.

113. Dodd, G.D., 3rd, M.S. Frank, M. Aribandi, S. Chopra, and K.N. Chintapalli, Radiofrequency thermal ablation: computer analysis of the size of the thermal injury cre-ated by overlapping ablations. AJR Am J Roentgenol, 2001. 177(4): p. 777-82.

114. Gazelle, G.S., S.N. Goldberg, L. Solbiati, and T. Livraghi, Tumor ablation with radio-frequency energy. Radiology, 2000. 217(3): p. 633-46.

115. Lorentzen, T., The loop electrode: in vitro evaluation of a device for ultrasound-guided interstitial tissue ablation using radiofrequency electrosurgery. Acad Radiol, 1996. 3(3): p. 219-24.

116. Lorentzen, T., C.P. Nolsoe, S.T. Torp-Pedersen, N.E. Holm Christensen, S. Pedersen, T. Horn, et al., The loop electrode: a new device for US-guided interstitial tissue ablation using radiofrequency electrosurgery - an animal study. Minim Invasive Ther, 1996. 5(6): p. 511-516.

117. Greenwood, J.J., Two point coagulation. A new principle and instrument for applying coagulation current in neurosurgery. Am J Surg, 1940. 50: p. 267-79.

118. McGahan, J.P., W.Z. Gu, J.M. Brock, H. Tesluk, and C.D. Jones, Hepatic ablation using bipolar radiofrequency electrocautery. Acad Radiol, 1996. 3(5): p. 418-22.

59

119. Goldberg, S.N., G.S. Gazelle, S.L. Dawson, W.J. Rittman, P.R. Mueller, and D.I. Rosenthal, Tissue ablation with radiofrequency using multiprobe arrays. Acad Radiol, 1995. 2(8): p. 670-4.

120. Goldberg, S.N., G.S. Gazelle, L. Solbiati, W.J. Rittman, and P.R. Mueller, Radiofrequency tissue ablation: increased lesion diameter with a perfusion electrode. Acad Radiol, 1996. 3(8): p. 636-44.

121. Lorentzen, T., A cooled needle electrode for radiofrequency tissue ablation: thermody-namic aspects of improved performance compared with conventional needle design. Acad Radiol, 1996. 3(7): p. 556-63.

122. Patterson, E.J., C.H. Scudamore, D.A. Owen, A.G. Nagy, and A.K. Buczkowski, Radiofrequency ablation of porcine liver in vivo: effects of blood flow and treatment time on lesion size. Ann Surg, 1998. 227(4): p. 559-65.

123. Rossi, S., E. Buscarini, F. Garbagnati, M. Di Stasi, P. Quaretti, M. Rago, et al., Percutaneous treatment of small hepatic tumors by an expandable RF needle electrode. AJR Am J Roentgenol, 1998. 170(4): p. 1015-22.

124. de Baere, T., D. Elias, C. Dromain, M.G. Din, V. Kuoch, M. Ducreux, et al., Radiofrequency ablation of 100 hepatic metastases with a mean follow-up of more than 1 year. AJR Am J Roentgenol, 2000. 175(6): p. 1619-25.

125. Siperstein, A., A. Garland, K. Engle, S. Rogers, E. Berber, A. Foroutani, et al., Local recur-rence after laparoscopic radiofrequency thermal ablation of hepatic tumors. Ann Surg Oncol, 2000. 7(2): p. 106-13.

126. Gillams, A.R. and W.R. Lees, Radio-frequency ablation of colorectal liver metastases in 167 patients. Eur Radiol, 2004. 14(12): p. 2261-7.

127. Navarra, G., A. Ayav, J.C. Weber, S.L. Jensen, C. Smadga, J.P. Nicholls, et al., Short- and-long term results of intraoperative radiofrequency ablation of liver metastases. Int J Colorectal Dis, 2005. 20(6): p. 521-8.

128. Berber, E., R. Pelley, and A.E. Siperstein, Predictors of survival after radiofrequency ther-mal ablation of colorectal cancer metastases to the liver: a prospective study. J Clin Oncol, 2005. 23(7): p. 1358-64.

129. Machi, J., A.J. Oishi, K. Sumida, K. Sakamoto, N.L. Furumoto, R.H. Oishi, et al., Long-term outcome of radiofrequency ablation for unresectable liver metastases from colorectal cancer: evaluation of prognostic factors and effectiveness in first- and second-line man-agement. Cancer J, 2006. 12(4): p. 318-26.

130. Topal, B., L. Kaufman, R. Aerts, and F. Penninckx, Patterns of failure following curative resection of colorectal liver metastases. Eur J Surg Oncol, 2003. 29(3): p. 248-53.

131. Elias, D., O. Baton, L. Sideris, V. Boige, D. Malka, G. Liberale, et al., Hepatectomy plus intraoperative radiofrequency ablation and chemotherapy to treat technically unresect-able multiple colorectal liver metastases. J Surg Oncol, 2005. 90(1): p. 36-42.

60

132. Kuvshinoff, B.W. and D.M. Ota, Radiofrequency ablation of liver tumors: influence of technique and tumor size. Surgery, 2002. 132(4): p. 605-11.

133. van Duijnhoven, F.H., M.C. Jansen, J.M. Junggeburt, R. van Hillegersberg, A.M. Rijken, F. van Coevorden, et al., Factors influencing the local failure rate of radiofrequency abla-tion of colorectal liver metastases. Ann Surg Oncol, 2006. 13(5): p. 651-8.

134. Aloia, T.A., J.N. Vauthey, E.M. Loyer, D. Ribero, T.M. Pawlik, S.H. Wei, et al., Solitary colorectal liver metastasis: resection determines outcome. Arch Surg, 2006. 141(5): p. 460-6.

135. Mulier, S., y. Ni, J. Jamart, T. Ruers, G. Marchal, and L. Michel, Local recurrence after hepatic radiofrequency coagulation: multivariate meta-analysis and review of contribut-ing factors. Ann Surg, 2005. 242(2): p. 158-71.

136. Mulier, S., y. Ni, y. Miao, A. Rosiere, A. Khoury, G. Marchal, et al., Size and geometry of hepatic radiofrequency lesions. Eur J Surg Oncol, 2003. 29(10): p. 867-78.

137. Denys, A.L., T.D. Baere, V. Kuoch, B. Dupas, P. Chevallier, D.C. Madoff, et al., Radio-frequency tissue ablation of the liver: in vivo and ex vivo experiments with four different systems. Eur Radiol, 2003. 13(10): p. 2346-52.

138. Pereira, P.L., J. Trubenbach, M. Schenk, J. Subke, S. Kroeber, I. Schaefer, et al., Radiofrequency ablation: in vivo comparison of four commercially available devices in pig livers. Radiology, 2004. 232(2): p. 482-90.

139. Montgomery, R.S., A. Rahal, G.D. Dodd, 3rd, J.R. Leyendecker, and L.G. Hubbard, Radiofrequency ablation of hepatic tumors: variability of lesion size using a single abla-tion device. AJR Am J Roentgenol, 2004. 182(3): p. 657-661.

140. Raman, S.S., D.S. Lu, D.J. Vodopich, J. Sayre, and C. Lassman, Creation of radiofrequency lesions in a porcine model: correlation with sonography, CT, and histopathology. AJR Am J Roentgenol, 2000. 175(5): p. 1253-8.

141. Cha, C.H., F.T. Lee, Jr., J.M. Gurney, B.K. Markhardt, T.F. Warner, F. Kelcz, et al., CT versus sonography for monitoring radiofrequency ablation in a porcine liver. AJR Am J Roentgenol, 2000. 175(3): p. 705-11.

142. Steiner, P., R. Botnar, S.N. Goldberg, G.S. Gazelle, and J.F. Debatin, Monitoring of radio frequency tissue ablation in an interventional magnetic resonance environment. Preliminary ex vivo and in vivo results. Invest Radiol, 1997. 32(11): p. 671-8.

143. Boaz, T.L., J.S. Lewin, y.C. Chung, J.L. Duerk, M.E. Clampitt, and J.R. Haaga, MR monitoring of MR-guided radiofrequency thermal ablation of normal liver in an animal model. J Magn Reson Imaging, 1998. 8(1): p. 64-9.

144. Liu, J.B., B.B. Goldberg, D.A. Merton, N.M. Rawool, L. Needleman, and F. Forsberg, The role of contrast-enhanced sonography for radiofrequency ablation of liver tumors. J Ultrasound Med, 2001. 20(5): p. 517-23.

61

145. Frich, L., Non-invasive thermometry for monitoring hepatic radiofrequency ablation. Minim Invasive Ther Allied Technol, 2006. 15(1): p. 18-25.

146. Goldberg, S.N., P.F. Hahn, K.K. Tanabe, P.R. Mueller, W. Schima, C.A. Athanasoulis, et al., Percutaneous radiofrequency tissue ablation: does perfusion-mediated tissue cooling limit coagulation necrosis? J Vasc Interv Radiol, 1998. 9(1 Pt 1): p. 101-11.

147. Goldberg, S.N., P.F. Hahn, E.F. Halpern, R.M. Fogle, and G.S. Gazelle, Radio-frequency tissue ablation: effect of pharmacologic modulation of blood flow on coagulation diam-eter. Radiology, 1998. 209(3): p. 761-7.

148. Horkan, C., M. Ahmed, Z. Liu, G.S. Gazelle, S.A. Solazzo, J.B. Kruskal, et al., Radiofrequency ablation: Effect of pharmacologic modulation of hepatic and renal blood flow on coagulation diameter in a Vx2 tumor model. J Vasc Interv Radiol, 2004. 15(3): p. 269-74.

149. Chinn, S.B., F.T. Lee, Jr., G.D. Kennedy, C. Chinn, C.D. Johnson, T.C. Winter, 3rd, et al., Effect of vascular occlusion on radiofrequency ablation of the liver: results in a porcine model. AJR Am J Roentgenol, 2001. 176(3): p. 789-95.

150. Denys, A.L., T. De Baere, C. Mahe, J.C. Sabourin, A. Sa Cunha, S. Germain, et al., Radio-frequency tissue ablation of the liver: effects of vascular occlusion on lesion diameter and biliary and portal damages in a pig model. Eur Radiol, 2001. 11(10): p. 2102-8.

151. Chang, C.K., M.P. Hendy, J.M. Smith, M.H. Recht, and R.E. Welling, Radiofrequency ablation of the porcine liver with complete hepatic vascular occlusion. Ann Surg Oncol, 2002. 9(6): p. 594-8.

152. de Baere, T., B. Bessoud, C. Dromain, M. Ducreux, V. Boige, N. Lassau, et al., Percutaneous radiofrequency ablation of hepatic tumors during temporary venous occlusion. AJR Am J Roentgenol, 2002. 178(1): p. 53-59.

153. Brieger, J., P.L. Pereira, J. Trubenbach, M. Schenk, S.M. Krober, D. Schmidt, et al., In vivo efficiency of four commercial monopolar radiofrequency ablation systems: a comparative experimental study in pig liver. Invest Radiol, 2003. 38(10): p. 609-16.

154. Metcalfe, M.S., F.H. Bridgewater, E.J. Mullin, and G.J. Maddern, Useless and dangerous--fine needle aspiration of hepatic colorectal metastases. Bmj, 2004. 328(7438): p. 507-8.

155. Jones, O.M., M. Rees, T.G. John, S. Bygrave, and G. Plant, Biopsy of potentially oper-able hepatic colorectal metastases is not useless but dangerous. Bmj, 2004. 329(7473): p. 1045-6.

156. Berber, E., A. Foroutani, A.M. Garland, S.J. Rogers, K.L. Engle, T.L. Ryan, et al., Use of CT Hounsfield unit density to identify ablated tumor after laparoscopic radiofrequency ablation of hepatic tumors. Surg Endosc, 2000. 14(9): p. 799-804.

157. Chopra, S., G.D. Dodd, 3rd, K.N. Chintapalli, J.R. Leyendecker, O.I. Karahan, and H. Rhim, Tumor recurrence after radiofrequency thermal ablation of hepatic tumors:

62

spectrum of findings on dual-phase contrast-enhanced CT. AJR Am J Roentgenol, 2001. 177(2): p. 381-7.

158. Dromain, C., T. de Baere, D. Elias, V. Kuoch, M. Ducreux, V. Boige, et al., Hepatic tumors treated with percutaneous radio-frequency ablation: CT and MR imaging follow-up. Radiology, 2002. 223(1): p. 255-62.

159. Mulier, S., P. Mulier, y. Ni, y. Miao, B. Dupas, G. Marchal, et al., Complications of radio-frequency coagulation of liver tumours. Br J Surg, 2002. 89(10): p. 1206-1222.

160. Livraghi, T., L. Solbiati, M.F. Meloni, G.S. Gazelle, E.F. Halpern, and S.N. Goldberg, Treatment of focal liver tumors with percutaneous radio-frequency ablation: complica-tions encountered in a multicenter study. Radiology, 2003. 226(2): p. 441-51.

161. Shen, P., S. Fleming, C. Westcott, and V. Challa, Laparoscopic radiofrequency ablation of the liver in proximity to major vasculature: effect of the Pringle maneuver. J Surg Oncol, 2003. 83(1): p. 36-41.

162. Marchal, F., D. Elias, P. Rauch, A. Leroux, J. Stines, J.L. Verhaeghe, et al., Biliary lesions during radiofrequency ablation in liver. Study on the pig. Eur Surg Res, 2004. 36(2): p. 88-94.

163. Ng, K.K., C.M. Lam, R.T. Poon, T.W. Shek, S.T. Fan, and J. Wong, Delayed portal vein thrombosis after experimental radiofrequency ablation near the main portal vein. Br J Surg, 2004. 91(5): p. 632-9.

164. Dominique, E., A. El Otmany, A. Goharin, D. Attalah, and T. de Baere, Intraductal cool-ing of the main bile ducts during intraoperative radiofrequency ablation. J Surg Oncol, 2001. 76(4): p. 297-300.

165. Elias, D., L. Sideris, M. Pocard, C. Dromain, and T. De Baere, Intraductal cooling of the main bile ducts during radiofrequency ablation prevents biliary stenosis. J Am Coll Surg, 2004. 198(5): p. 717-21.

166. Stippel, D.L., C. Bangard, H.U. Kasper, J.H. Fischer, A.H. Holscher, and A. Gossmann, Experimental bile duct protection by intraductal cooling during radiofrequency ablation. Br J Surg, 2005. 92(7): p. 849-55.

167. den Brok, M.H., R.P. Sutmuller, R. van der Voort, E.J. Bennink, C.G. Figdor, T.J. Ruers, et al., In situ tumor ablation creates an antigen source for the generation of antitumor immunity. Cancer Res, 2004. 64(11): p. 4024-9.

168. von Breitenbuch, P., G. Kohl, M. Guba, E. Geissler, K.W. Jauch, and M. Steinbauer, Thermoablation of colorectal liver metastases promotes proliferation of residual intrahe-patic neoplastic cells. Surgery, 2005. 138(5): p. 882-7.

169. Lu, D.S., S.S. Raman, P. Limanond, D. Aziz, J. Economou, R. Busuttil, et al., Influence of large peritumoral vessels on outcome of radiofrequency ablation of liver tumors. J Vasc Interv Radiol, 2003. 14(10): p. 1267-74.

63

170. Scott, D.J., J.B. Fleming, L.M. Watumull, G. Lindberg, S.T. Tesfay, and D.B. Jones, The ef-fect of hepatic inflow occlusion on laparoscopic radiofrequency ablation using simulated tumors. Surg Endosc, 2002. 16(9): p. 1286-91.

171. yamasaki, T., F. Kurokawa, H. Shirahashi, N. Kusano, K. Hironaka, and K. Okita, Percutaneous radiofrequency ablation therapy for patients with hepatocellular carcinoma during occlusion of hepatic blood flow. Comparison with standard percutaneous radio-frequency ablation therapy. Cancer, 2002. 95(11): p. 2353-60.

172. Gillams, A.R. and W.R. Lees, CT mapping of the distribution of saline during radiofre-quency ablation with perfusion electrodes. Cardiovasc Intervent Radiol, 2005. 28(4): p. 476-80.

173. Mulier, S., y. Ni, L. Frich, F. Burdio, A. Denys, J.F. DeWispelaere, et al., Experimental and clinical radiofrequency ablation in solid organs: proposal for standardized description of coagulation size and geometry. Ann Surg Oncol, 2007. In press.

174. Lu, D.S., S.S. Raman, D.J. Vodopich, M. Wang, J. Sayre, and C. Lassman, Effect of vessel size on creation of hepatic radiofrequency lesions in pigs: assessment of the “heat sink” effect. AJR Am J Roentgenol, 2002. 178(1): p. 47-51.

175. Stippel, D.L., H.G. Brochhagen, M. Arenja, J. Hunkemoller, A.H. Holscher, and K.T. Beckurts, Variability of size and shape of necrosis induced by radiofrequency ablation in human livers: a volumetric evaluation. Ann Surg Oncol, 2004. 11(4): p. 420-5.

176. Miyakawa, M. and J. Bolomey, eds. Non-invasive thermometry of the human body. 1996, CRC Press: Boca Raton. 257 p.

177. van Rhoon, G.C. and P. Wust, Introduction: Non-invasive thermometry for thermo-therapy. Int J Hyperthermia, 2005. 21(6): p. 489-95.

178. Vigen, K.K., J. Jarrard, V. Rieke, J. Frisoli, B.L. Daniel, and K. Butts Pauly, In vivo porcine liver radiofrequency ablation with simultaneous MR temperature imaging. J Magn Reson Imaging, 2006. 23(4): p. 578-84.

179. Salomir, R., J. Palussiere, S.L. Fossheim, A. Rogstad, U.N. Wiggen, N. Grenier, et al., Local delivery of magnetic resonance (MR) contrast agent in kidney using thermosensi-tive liposomes and MR imaging-guided local hyperthermia: a feasibility study in vivo. J Magn Reson Imaging, 2005. 22(4): p. 534-40.

180. Elias, D., O. Baton, L. Sideris, T. Matsuhisa, M. Pocard, and P. Lasser, Local recurrences after intraoperative radiofrequency ablation of liver metastases: a comparative study with anatomic and wedge resections. Ann Surg Oncol, 2004. 11(5): p. 500-5.

181. Goldberg, S.N., C.J. Grassi, J.F. Cardella, J.W. Charboneau, G.D. Dodd, 3rd, D.E. Dupuy, et al., Image-guided tumor ablation: standardization of terminology and reporting crite-ria. Radiology, 2005. 235(3): p. 728-39.

182. Swindle, M., Surgery, anesthesia, and experimental techniques in swine. 1998, Ames: Iowa State University Press. xIII, 329 p.

64

183. Kokudo, N., y. Miki, S. Sugai, A. yanagisawa, y. Kato, y. Sakamoto, et al., Genetic and histological assessment of surgical margins in resected liver metastases from colorectal carcinoma: minimum surgical margins for successful resection. Arch Surg, 2002. 137(7): p. 833-40.

184. yokoyama, N., y. Shirai, y. Ajioka, S. Nagakura, T. Suda, and K. Hatakeyama, Immunohistochemically detected hepatic micrometastases predict a high risk of intra-hepatic recurrence after resection of colorectal carcinoma liver metastases. Cancer, 2002. 94(6): p. 1642-7.

185. Nikfarjam, M., V. Muralidharan, and C. Christophi, Altered growth patterns of colorectal liver metastases after thermal ablation. Surgery, 2006. 139(1): p. 73-81.

186. Nelson, A.R., B. Fingleton, M.L. Rothenberg, and L.M. Matrisian, Matrix metalloprotein-ases: biologic activity and clinical implications. J Clin Oncol, 2000. 18(5): p. 1135-49.

187. Parks, W.C., C.L. Wilson, and y.S. Lopez-Boado, Matrix metalloproteinases as modula-tors of inflammation and innate immunity. Nat Rev Immunol, 2004. 4(8): p. 617-29.

188. Wagenaar-Miller, R.A., L. Gorden, and L.M. Matrisian, Matrix metalloproteinases in colorectal cancer: is it worth talking about? Cancer Metastasis Rev, 2004. 23(1-2): p. 119-35.

189. Weitz, J., M. Koch, P. Kienle, A. Schrodel, F. Willeke, A. Benner, et al., Detection of he-matogenic tumor cell dissemination in patients undergoing resection of liver metastases of colorectal cancer. Ann Surg, 2000. 232(1): p. 66-72.

190. Topal, B., J.L. Aerts, T. Roskams, S. Fieuws, J. Van Pelt, P. Vandekerckhove, et al., Cancer cell dissemination during curative surgery for colorectal liver metastases. Eur J Surg Oncol, 2005. 31(5): p. 506-11.

191. Rodgers, M.S., R. Collinson, S. Desai, R.S. Stubbs, and J.L. McCall, Risk of dissemination with biopsy of colorectal liver metastases. Dis Colon Rectum, 2003. 46(4): p. 454-8.

192. Jones, O.M., M. Rees, T.G. John, S. Bygrave, and G. Plant, Biopsy of resectable colorectal liver metastases causes tumour dissemination and adversely affects survival after liver resection. Br J Surg, 2005. 92(9): p. 1165-8.

193. Solbiati, L., T. Livraghi, S.N. Goldberg, T. Ierace, F. Meloni, M. Dellanoce, et al., Percutaneous radio-frequency ablation of hepatic metastases from colorectal cancer: long-term results in 117 patients. Radiology, 2001. 221(1): p. 159-66.

194. Adam, R., E.J. Hagopian, M. Linhares, J. Krissat, E. Savier, D. Azoulay, et al., A com-parison of percutaneous cryosurgery and percutaneous radiofrequency for unresectable hepatic malignancies. Arch Surg, 2002. 137(12): p. 1332-9.

195. Kotoh, K., S. Morizono, M. Kohjima, M. Enjoji, H. Sakai, and M. Nakamuta, Evaluation of liver parenchymal pressure and portal endothelium damage during radio frequency ablation in an in vivo porcine model. Liver Int, 2005. 25(6): p. 1217-23.

65

196. Halpern, S.D., J.H. Karlawish, and J.A. Berlin, The continuing unethical conduct of un-derpowered clinical trials. Jama, 2002. 288(3): p. 358-62.

197. Horrobin, D.F., Are large clinical trials in rapidly lethal diseases usually unethical? Lancet, 2003. 361(9358): p. 695-7.

198. Giacchetti, S., G. Bjarnason, C. Garufi, D. Genet, S. Iacobelli, M. Tampellini, et al., Phase III trial comparing 4-day chronomodulated therapy versus 2-day conventional delivery of fluorouracil, leucovorin, and oxaliplatin as first-line chemotherapy of metastatic colorectal cancer: the European Organisation for Research and Treatment of Cancer Chronotherapy Group. J Clin Oncol, 2006. 24(22): p. 3562-9.

199. Souglakos, J., N. Androulakis, K. Syrigos, A. Polyzos, N. Ziras, A. Athanasiadis, et al., FOLFOxIRI (folinic acid, 5-fluorouracil, oxaliplatin and irinotecan) vs FOLFIRI (fo-linic acid, 5-fluorouracil and irinotecan) as first-line treatment in metastatic colorectal cancer (MCC): a multicentre randomised phase III trial from the Hellenic Oncology Research Group (HORG). Br J Cancer, 2006. 94(6): p. 798-805.

200. Wilson, C.B., Adoption of new surgical technology. Bmj, 2006. 332(7533): p. 112-4.201. Erturk, S.M., T. Ichikawa, H. Fujii, S. yasuda, and P.R. Ros, PET imaging for evaluation

of metastatic colorectal cancer of the liver. Eur J Radiol, 2006. 58(2): p. 229-35.202. Abdalla, E.K. and J.N. Vauthey, Colorectal metastases: resect or ablate? Ann Surg Oncol,

2006. 13(5): p. 602-3.203. Primrose, J.N., Treatment of colorectal metastases: surgery, cryotherapy, or radiofrequency

ablation. Gut, 2002. 50(1): p. 1-5.204. Ruers, T., W. Bechstein, and J. Lederman, CLOCC trial (chemotherapy + local ablation

versus chemotherapy). Randomized phase II study of local treatment of liver metastases by radiofrequency combined with chemotherapy versus chemotherapy alone in patients with unresectable colorectal liver metastases. EORTC protocol 40004. Version 6.0. October 13, 2005, EORTC. 130 p.

205. Oshowo, A., A. Gillams, E. Harrison, W.R. Lees, and I. Taylor, Comparison of resec-tion and radiofrequency ablation for treatment of solitary colorectal liver metastases. Br J Surg, 2003. 90(10): p. 1240-3.

206. McKay, A., E. Dixon, and M. Taylor, Current role of radiofrequency ablation for the treat-ment of colorectal liver metastases. Br J Surg, 2006. 93(10): p. 1192-1201.

207. Mulier, S., y. Miao, P. Mulier, B. Dupas, P. Pereira, T. de Baere, et al., Electrodes and mul-tiple electrode systems for radiofrequency ablation: a proposal for updated terminology. Eur Radiol, 2005. 15(4): p. 798-808.

66

67

paper iii, page 1995

The correct heading of the second column of Table 2 is:

Control group (n=6)

e r r AtA

I

Hepatic radiofrequency ablation using perfusion electrodes in a pig model:

Effect of the Pringle manoeuvre

L. Fricha,b,*, T. Malab, I.P. Gladhaugb,c

aThe Interventional Centre, Rikshospitalet University Hospital, N-0027 Oslo, NorwaybDepartment of Surgery, Rikshospitalet University Hospital, N-0027 Oslo, Norway

cDepartment of Surgery, Faculty Division Rikshospitalet, University of Oslo, N-0027 Oslo, Norway

Accepted 27 February 2006

Available online 3 April 2006

Abstract

Aim: To assess the influence of the Pringle manoeuvre on volume and geometry of coagulations close to the portal vein using an impedance-

controlled radiofrequency ablation system with perfusion electrodes.

Methods: Twelve pigs were randomly assigned to a control group (nZ6) and a group where the Pringle manoeuvre was applied during

ablation (nZ6). One coagulation was made in each animal close to the portal vein. All animals were sacrificed 4 days after ablation, and the

livers were removed for gross and histopathologic analysis.

Results: Effective coagulation volume in the Pringle group (10.8G5.0 cm3) was significantly increased (pZ0.03) compared to the control

group (4.1G4.1 cm3). The efficacy ratio, defined as the effective coagulation volume divided by the coagulation volume, was not

significantly different in the Pringle group (0.47G0.27) compared to the control group (0.33G0.22). The geometrical centre of the effective

coagulation volume did not correspond to the position of the ablation electrode. Thermal damage of the gallbladder was found in three

animals, all belonging to the Pringle group.

Conclusions: The Pringle manoeuvre was associated with increased effective coagulation volume, but did not significantly influence the

predictability of coagulation volume or geometry.

q 2006 Elsevier Ltd. All rights reserved.

Keywords: Radiofrequency; Liver; Neoplasm; Animal experimentation

Introduction

Radiofrequency ablation (RF ablation) is used for in situ

destruction of malignant liver tumours in patients who are

not candidates for hepatic resection.1 RF ablation should be

subject to established oncological criteria for surgical

treatment of malignant liver tumours including complete

tumour clearance and treatment margins of 0.5–1 cm.2,3

Coagulation geometry is as important as coagulation

volume in order to achieve adequate treatment margins.3–5

Interruption of hepatic inflow during ablation enable RF

coagulations that are larger, less elliptic and less distorted in

experimental studies using multitined expandable electro-

des.4,6 In contrast, effects of the Pringle manoeuvre on

coagulation volume and geometry made with an impedance-

controlled RF ablation system using perfusion electrodes

have not been investigated.

Perfusion mediated tissue cooling is believed to be

particularly prominent in the perivascular regions.7 It is

therefore of special interest to characterize volume and

geometry of coagulations made in proximity to large

intrahepatic vessels. The aim of this study was to assess

the impact of the Pringle manoeuvre on the volume and

geometry of coagulations made with an impedance-

controlled perfusion electrode system close to the right

portal vein branch (RPV) in normal porcine liver.

Materials and methods

Study design

Twelve domestic pigs with a mean weight of 30.3 kg

(SDZ3.4 kg) were randomly assigned to a control group

EJSO 32 (2006) 527–532 www.ejso.com

0748-7983/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ejso.2006.02.021

* Corresponding author. Address: Department of Surgery, Rikshospitalet

University Hospital, N-0027 Oslo, Norway. Tel.:C47 23071825; fax:C47

23072267.

E-mail address: lars.frich@labmed.uio.no (L. Frich).

(nZ6) and a group where the Pringle manoeuvre was

applied during ablation (nZ6). One coagulation was made

in each animal. Coagulation volume and geometry was

assessed from specimens retrieved 4 days after ablation.

RF ablation protocol

An impedance-controlled RF ablation system (Elek-

trotom 106 HiTT, Berchtold GmbH and Co., Germany) was

used. The system incorporates a generator that delivers

alternating current at 375 kHz. The cumulative energy

output is displayed in real-time. The 1.7 mm diameter

perfusion electrode (EZ 707-15-04, Berchtold, Germany)

has a 15 cm long shaft incorporating a lumen for saline

perfusion, with a terminal non-insulated segment of 15 mm.

RF energy was applied for 1 min with an effect of 30 W,

immediately followed by 8 min with 50 W.

Operative procedure

Surgery was performed during general anaesthesia

induced by intravenous pentobarbital 150–400 mg and

morphine 10–30 mg and maintained by inhaled isoflurane

1–1.2% in oxygen. The analgesic effect was supplemented

with morphine infusion at 10 mg/h. A 134 cm2 dispersive

plate (EZ 344-02, Berchtold, Germany) was placed over the

right hip. Laparatomy was performed and a silicone vessel

loop (Sterion, MN, USA) was passed twice loosely around

the hepatoduodenal ligament.

The ablation electrode was positioned 2 cm into the liver

parenchyma in the anterior–posterior plane within 1.5 cm of

RPV. In the Pringle group, the vessel loop was tightened

around the hepatoduodenal ligament 1 min prior to appli-

cation of RF energy and released immediately after ablation

had been completed. The needle track was coagulated by

applying 25 W during withdrawal of the electrode. The

animals were euthanised 4 days after ablation and the livers

were removed for gross and histopathological analysis.

Assessment of coagulation volume and geometry

All coagulationswere excisedwithwidemargins and cut at

4–6 mm intervals transverse to RPV. All fresh tissue slices

were photographedwith a L-shapedABFOno. 2 photomacro-

graphic scale (Lightning Powder Company, Jacksonville, FL,

USA) (Fig. 1(a)). The images were imported into the scientific

image analysis program ImageJ version 1.33 (National

institute of mental health, Bethesda, MA, USA). The border

of normal liver tissue vs tissue with macroscopically visible

changes was traced on both sides of each slice, and the

perimeter and area of the coagulation was measured.

Maximum coagulation diameter was determined by measur-

ing the Feret’s diameter, which is the longest distance between

any two points along the perimeter of the coagulation.

Effective coagulation diameter was defined as the diameter

of the largest circle that could be fit into the coagulation

(Fig. 1(b)). The coagulation volume of each slice was

calculated by multiplying the slice thickness with the mean

of the coagulation area from both sides of the slice.

Coagulation volume was determined by addition of the

corresponding slice coagulation volumes. Delivered energy

per coagulation volume was calculated for all coagulations.

Effective coagulation volumewas defined as the volumeof the

largest sphere that could be inscribed into the coagulation

volume. The efficacy ratio was defined as the effective

coagulation volume divided by the total coagulation volume.

An image navigation and display software system8was used to

create 3D surface renderings of the coagulation volume and

the inscribed effective coagulation volume on the basis of the

electronically traced perimeters of the coagulation slices.

Histopathological examination

Tissue slices were fixed in 4% formaldehyde for at least

10 days, then dehydrated and embedded in paraffin.

Figure 1. (a) Macrophotograph of tissue coagulation made during Pringle

manoeuvre. The right portal vein branch is seen in the periphery of the

coagulated area. The assumed position of the ablation electrode is indicated

by an arrowhead. (b) Geometrical evaluation of the coagulation in (a). The

thick line denotes the border of normal liver tissue vs tissue with

macroscopically changes consistent with coagulation. Dmax denotes

maximum coagulation diameter and Deff effective coagulation diameter.

The geometrical centre of the inscribed circle does not correspond to the

assumed position of the ablation electrode (black dot).

L. Frich et al. / EJSO 32 (2006) 527–532528

A 4–5 mm thick section was cut from each slice, stained

with haematoxylin and eosin and examined by light

microscopy by an experienced pathologist.

Statistical analysis

The animals were allocated to the two groups by block

randomization, using a block size of two. SPSS for Mac

OSX version 11.0.2 (SPSS, Chicago, IL, USA) was used for

statistical analyses. Data are presented as meanGSD unless

otherwise indicated. Independent samples t-test was used

for comparison of the two groups. A p-value !0.05 was

considered statistical significant.

Results

Coagulation characteristics and delivered energy per

coagulation volume in the two groups are summarized in

Table 1.

Coagulation volume

Coagulation volume was increased three-fold, although

not significantly, in the Pringle group compared to the

control group (Fig. 2(a)). Effective coagulation volume

was significantly increased in the Pringle group

(Fig. 2(b)). Delivered energy per coagulation volume

was six-fold lower, although not significantly, in the

Pringle group.

Coagulation geometry

Maximal coagulation diameter was increased, although

not significantly, in the Pringle group. Effective coagulation

diameter was significantly increased in the Pringle group.

The efficacy ratio was not significantly different in the two

groups. The geometrical centre of the effective coagulation

volume did not correspond to the position of the ablation

electrode (Fig. 1(b)).

Complications

Thermal damage to abutting organs was present on

re-operation in three animals, all belonging to the Pringle

group. In two animals, gallbladder perforation was caused

by severely deformed coagulations with maximum lengths

of 11.3 and 13.7 cm extending to the gallbladder (Fig. 3).

Additionally, perforation of the gastric wall was found in

one of these animals. Perforation of the gallbladder was also

found in a third animal where the coagulation had not

extended to the gallbladder.

Table 1

Summary of coagulation characteristics and delivered energy per coagulation volume in the two experimental groups

Control group (nZ6),

meanGSD (range)

Pringle group (nZ6),

meanGSD (range)

pa

Maximal coagulation diameter (cm) 4.1G1.5 (1.6–6.3) 7.6G4.2 (3.4–13.3) 0.10

Effective coagulation diameter (cm) 1.7G0.8 (0.6–2.7) 2.7G0.4 (2.1–3.2) 0.03

Coagulation volume (cm3) 9.9G7.6 (0.9–21.9) 31.2G25.5 (11.6–77.2) 0.08

Effective coagulation volume (cm3) 4.1G4.1 (0.1–10.6) 10.8G5.0 (4.9–17.9) 0.03

Efficacy ratio 0.33G0.22 (0.1–0.7) 0.47G0.27 (0.2–0.9) 0.36

Delivered energy per coagulation volume

(J/cm3)

7450G11,004 (1143–29,623) 1224G768 (315–2135) 0.20

a Independent samples t-test with unequal variances assumed.

Figure 2. Coagulation volume (a) and effective coagulation volume (b) in the control group (nZ6) and the Pringle group (nZ6). Boxes represent the

interquartile range, horizontal bars within the boxes indicate median values. Mean values are indicated by a plus sign (C).

L. Frich et al. / EJSO 32 (2006) 527–532 529

Histopathology

A sharp demarcation between coagulated tissue and

normal hepatic parenchyma was found at gross examin-

ation, corresponding to the 1–2 mm border between

coagulated and non-coagulated tissue at histopathologic

examination (Fig. 4). The hepatic microstructure was

preserved in parts of the coagulated area. Cytoplasm of

coagulated hepatocytes was more eosinophilic, and showed

pycnotic nuclei compared to viable hepatocytes. In other

coagulated areas, the hepatocytes had undergone karyolysis

and were denucleated.

Discussion

Coagulation volumes in our control group were smaller

than in previous experimental studies using perfusion

electrodes,5,9 but in conformity with clinical results.10

These diverging findings can be attributed to the use of

different ablation protocols and methods used for volume

estimation. Furthermore, as the coagulations in our study

were made close to a portal vein, perfusion mediated tissue

cooling possibly limited the coagulation volume. These

studies nevertheless support our findings of unpredictable

ablation volumes when no hepatic vascular occlusion is

used. Use of the Pringle manoeuvre was associated with

three-fold increase in coagulation volume, which is

comparable to previous studies with multitined expandable

electrodes.4 The effective coagulation volume was nearly

three-fold increased in the Pringle group, demonstrating a

beneficial effect of the Pringle manoeuvre. The

maximal coagulation diameter is of importance in order to

avoid thermal damage to abutting organs, but is oncologi-

cally irrelevant.3 In contrast, knowledge of the effective

coagulation diameter is important to avoid

incomplete coagulation of the index tumour.11 Although

the effective coagulation diameter was increased by 1 cm

in the Pringle group, the Pringle manoeuvre was associated

with elongated coagulations extending outside the expected

treatment site.

The isoperimetric ratio has been used as an indicator of

coagulation circularity, but is not appropriate for character-

ization of irregular non-spherical volumes as different ratios

can be acquired depending on the orientation of the

measurement planes. In this study, we, therefore, introduced

the concept of the efficacy ratio which directly relates to the

ability to produce a spherical coagulation volume (Fig. 5).

A low efficacy ratio indicates that a large proportion of the

coagulation occurs in locations not relevant for the

oncological goal of the treatment. The efficacy ratio was

below 0.5 in both groups, but tended to increase in the

Pringle group. Our observation of a displacement of the

geometrical centre of the effective coagulation volume

relative to the position of the electrode may be caused by the

proximity of the electrode to large portal vessels. This

finding may be a of clinical importance and should be

further assessed.

Direct heat transfer from the treatment site can cause

burns of adjacent viscera, which is a well known

complication of hepatic RF ablation regardless of ablation

system used.12,13 However, when using perfusion electro-

des, diffusion of heated saline may cause thermal damage to

structures not in direct contact with the treatment site.10,14 In

Figure 3. Thermal injury to abutting organ. The coagulation has extended to

the gallbladder (G) and caused a full wall damage. The main portal vein

trunk (P) has been incised. The arrowhead indicates the assumed position of

the ablation electrode.

Figure 4. Haematoxylin and eosin stained section. A sharp demarcation

(arrowheads) is seen between the coagulation (R) and normal hepatic

parenchyma (N).

L. Frich et al. / EJSO 32 (2006) 527–532530

the present study, thermal damage to abutting structures was

observed in three animals. Gallbladder perforation

was observed in one of these animals even though

the coagulation did not involve the gallbladder, suggesting

that this perforation may be caused by diffusion of heated

saline.

The effect of saline infusion on coagulation volume and

geometry has not been completely elucidated, but is

believed to be three-fold; a liquid electrode effect where

the effective surface of the ablation electrode is increased by

the electrical conductive qualities of saline, reduction of

charring in the proximity of the electrode due to cooling

effects, and extended thermal damage by diffusion of heated

saline into the tissue.15 A marked divergence from a

spherical geometry for coagulations produced with per-

fusion electrodes during physiological liver perfusion has

been reported previously,5,9 and may be caused by diffusion

of saline along vessels, tissue planes or surfaces, generating

low resistance electrical paths that could lead to heating

outside the expected target area. In our study, use of the

Pringle manoeuvre resulted in unpredictable coagulation

volumes and geometry as well as thermal damage to the

gallbladder and the gastric wall. A possible explanation for

this finding is that reduced tissue capillary level

microperfusion due to hepatic inflow occlusion may

increase the risk of low resistance electrical paths or heated

saline to cause thermal damage.

Our findings are valid only for the specific RF ablation

system and electrodes used in the study. The lack of

statistical significance for several of the examined variables

may be explained by the limited statistical power of the

study. In contrast to the human liver, the deep fissures and

the thin lobes in the porcine liver may influence the volume

and geometry of the coagulations. Extrapolation of our

findings to clinical settings should, therefore, be done with

caution.

In conclusion, the reproducibility of the coagulation

volumes and geometry including the efficiency ratio was

low in the control group. The Pringle manoeuvre was

associated with increased effective coagulation volume, but

did not significantly increase the predictability of coagu-

lation volume, coagulation geometry or the efficacy ratio.

Furthermore, the Pringle manoeuvre was associated with

coagulations extending outside the expected treatment site

causing thermal damage to adjacent organs.

Acknowledgements

We thank the staff at the Interventional Centre at

Rikshospitalet for providing excellent assistance. We also

thank the Department of Comparative Medicine at

Rikshospitalet for support in animal care, and Ole Petter

F. Clausen, Department of Pathology, Rikshospitalet

University Hospital for examination of the histopatological

specimens. This work was supported by the Norwegian

Cancer Society, grant no. D02042/001.

References

1. Ni Y, Mulier S, Miao Y, Michel L, Marchal G. A review of the general

aspects of radiofrequency ablation. Abdom Imaging 2005;30(4):

381–400.

2. Blumgart LH. Liver resection for benign disease and for liver and

biliary tumors. In: Blumgart LH, Fong Y, editors. Surgery of the liver

and biliary tract. 3rd ed. London: W.B. Saunders; 2000. p. 1639–713.

3. Mulier S, Ni Y, Miao Y, Rosiere A, Khoury A, Marchal G, et al. Size

and geometry of hepatic radiofrequency lesions. Eur J Surg Oncol

2003;29(10):867–78.

4. Chinn SB, Lee Jr FT, Kennedy GD, Chinn C, Johnson CD, Winter

III TC, et al. Effect of vascular occlusion on radiofrequency ablation of

the liver: results in a porcine model. AJR Am J Roentgenol 2001;

176(3):789–95.

5. Pereira PL, Trubenbach J, Schenk M, Subke J, Kroeber S,

Schaefer I, et al. Radiofrequency ablation: in vivo comparison of

four commercially available devices in pig livers. Radiology 2004;

232(2):482–90.

6. Chang CK, Hendy MP, Smith JM, Recht MH, Welling RE. Radio-

frequency ablation of the porcine liver with complete hepatic vascular

occlusion. Ann Surg Oncol 2002;9(6):594–8.

7. Lu DS, Raman SS, Limanond P, Aziz D, Economou J, Busuttil R,

et al. Influence of large peritumoral vessels on outcome of

radiofrequency ablation of liver tumors. J Vasc Interv Radiol

2003;14(10):1267–74.

Figure 5. 3D surface renderings of coagulation volume (blue) and the

corresponding effective coagulation volume (red). Pringle manoeuvre was

used for both coagulations. The volumes are shown with correct

proportions relative to each other. (a) Coagulation volume is 42.9 cm3,

effective coagulation volume is 15.6 cm3. The efficacy ratio is 0.36. (b)

Coagulation volume is 12.8 cm3, effective coagulation volume is 11.5 cm3.

The efficacy ratio is 0.90.

L. Frich et al. / EJSO 32 (2006) 527–532 531

8. Rosset A, Spadola L, Ratib O. OsiriX: an open-source software for

navigating in multidimensional DICOM images. J Digit Imaging 2004;

17(3):205–16.

9. Denys AL, Baere TD, Kuoch V, Dupas B, Chevallier P, Madoff DC,

et al. Radio-frequency tissue ablation of the liver: in vivo and ex vivo

experiments with four different systems. Eur Radiol 2003;13(10):

2346–52.

10. Kettenbach J, Kostler W, Rucklinger E, Gustorff B, Hupfl M, Wolf F,

et al. Percutaneous saline-enhanced radiofrequency ablation of

unresectable hepatic tumors: initial experience in 26 patients.

AJR Am J Roentgenol 2003;180(6):1537–45.

11. Patterson EJ, Scudamore CH, Owen DA, Nagy AG, Buczkowski AK.

Radiofrequency ablation of porcine liver in vivo: effects of blood flow

and treatment time on lesion size. Ann Surg 1998;227(4):559–65.

12. Mulier S, Mulier P, Ni Y, Miao Y, Dupas B, Marchal G, et al.

Complications of radiofrequency coagulation of liver tumours. Br

J Surg 2002;89(10):1206–22.

13. Livraghi T, Solbiati L, Meloni MF, Gazelle GS, Halpern EF,

Goldberg SN. Treatment of focal liver tumors with percutaneous

radio-frequency ablation: complications encountered in a multicenter

study. Radiology 2003;226(2):441–51.

14. Gillams AR, Lees WR. CT mapping of the distribution of saline during

radiofrequency ablation with perfusion electrodes. Cardiovasc Inter-

vent Radiol 2005;28(4):476–80.

15. Livraghi T, Goldberg SN, Monti F, Bizzini A, Lazzaroni S,

Meloni F, et al. Saline-enhanced radio-frequency tissue ablation

in the treatment of liver metastases. Radiology 1997;202(1):

205–10.

L. Frich et al. / EJSO 32 (2006) 527–532532

II

Experimental Application of ThermosensitiveParamagnetic Liposomes for Monitoring MagneticResonance Imaging Guided Thermal Ablation

Lars Frich,1,2* Atle Bjørnerud,3 Sigrid Fossheim,4 Terje Tillung,1 and Ivar Gladhaug2

The use of a liposomal paramagnetic agent with a T1-relaxivitythat increases markedly at temperatures above the phase transi-tion temperature (Tm) of the liposomal membrane was evaluatedduring magnetic resonance imaging (MRI) guided hyperthermiaablation. A neodymium-yttrium aluminum garnet (Nd-YAG) laserunit and a radiofrequency ablation system were used for tissueablation in eight rabbit livers in vivo. One ablation was made ineach animal prior to administration of the liposomal agent. Lipo-somes with a Tm of 57°C containing gadodiamide (GdDTPA-BMA)were injected iv, and two additional ablations were performed.T1-weighted scans were performed in heated tissue, after tissuetemperature had normalized, and 15–20 min after normalization oftissue temperature. Increase in signal intensity (SI) for ablationsprior to injection of the agent was 13.0% (SD 5.7) for the lasergroup and 9.1% (SD 7.9) for the radiofrequency group. Signalintensity after administration of the agent unrelated to heatingwasnot statistically significant (SI 1.4%, P 0.35). For ablationsmade after injection of the agent, a significant increase was foundin the laser (SI 34.5%, SD 11.9) and radiofrequency group(SI 21.6%, SD 22.7). The persistent signal enhancementfound in areas exposed to a temperature above the thresholdtemperature above Tm allows thermal monitoring of MRI guidedthermal ablation. Magn Reson Med 52:1302–1309, 2004. © 2004Wiley-Liss, Inc.

Key words: thermometry; contrast agent; liver; radiofrequencyablation

Hyperthermia ablation is used for treatment of benign andmalignant tumors (1–4). The tissue response to heating isrelated to the absolute temperature induced in the tissueand exposure time. Irreversible tissue destruction occurswhen tissue is exposed to temperatures in excess of 50–55°C (5,6). Hyperthermia ablation has been widely usedfor treating malignant liver tumors not amenable to con-ventional surgical resection. Heating modalities in clinicaluse include laser, microwave, radiofrequency, and focusedultrasound (1,7–9). The exact size and shape of the result-ing ablation is difficult to predict because of heterogeneousdistribution of the thermal energy due to local disparity intissue composition, tissue anatomy, and vascularization

(10,11). In clinical series, local tumor progression afterablation therapy has been reported to be as high as 30–60%, indicating the presence of viable malignant cells atthe treated site (8,12,13). Improved intraprocedural moni-toring may reduce the local tumor progression rates asso-ciated with thermal ablation and prove beneficial in regardto long-term patient survival.

Multiplanar imaging capabilities and excellent soft tis-sue contrast make magnetic resonance imaging (MRI) ide-ally suited for peroperative visualization, positioning ofenergy applicators, and monitoring of thermal ablationprocedures. Additionally, noninvasive, three-dimensionalmapping of relative temperature changes is feasible withMRI, based on the longitudinal relaxation time (T1), thediffusion coefficient (D), or proton resonance frequency(PRF) of tissue water. These methods rely on image sub-traction and are sensitive to tissue motion (14). Thermalimaging of the liver therefore presents certain challengesas the liver is displaced during respiration due to move-ment of the diaphragm. A substantial body of researchexists on MRI-based methods for temperature mappingduring laser ablation (15–17). These methods can be ap-plied to thermal monitoring of other MRI-compatible localthermal ablation therapies such as focused ultrasound(18). Thermal monitoring of radiofrequency ablation posescertain challenges. MRI-based methods for thermal moni-toring cannot be used during radiofrequency ablation, asthe application of radiofrequency energy leads to electro-magnetic noise that severely deteriorates image quality.Simple switching circuits (19) and filtering of the RF signalby modification of the hardware (20) have been developedthat allow image acquisition in near real-time or during RFablation. At present such solutions are not incorporated incommercially available systems used for RF ablation. Con-sequently, reliable MRI-based thermometry of radiofre-quency ablation requires development of new methods forthermal imaging.

Thermosensitive MRI contrast agents represent a novelapproach for thermal imaging that offers a robust, motion-insensitive method for indication of an absolute tempera-ture threshold on standard imaging hardware using con-ventional MRI sequences. Mechanisms of action includetemperature-induced transition from a diamagnetic to aparamagnetic state using bistable molecular complexes(21) and temperature-dependent relaxivity by incorpora-tion of conventional MRI contrast agents within thermo-sensitive liposomes (22,23). In the present study we usedliposomes with a phase transition temperature (Tm) of57°C containing gadodiamide (GdDTPA-BMA) duringthermal ablation in vivo. The agent is designed to be “MRdormant” at physiologic temperatures and “MR activated”

1The Interventional Centre, Rikshospitalet University Hospital, 0027 Oslo,Norway.2Department of Surgery, Rikshospitalet University Hospital, 0027 Oslo, Nor-way.3Department of Radiology, Rikshospitalet University Hospital, 0027 Oslo,Norway.4Amersham Health AS, Part of GE Healthcare, Postbox 4220 Nydalen, 0401Oslo, Norway.Grant sponsor: Norwegian Cancer Society; Grant number: D02042/001.*Correspondence to: Lars Frich, Department of Surgery, Rikshospitalet Uni-versity Hospital, N-0027 Oslo, Norway. E-mail: lars.frich@labmed.uio.noReceived 2 February 2004; revised 28 July 2004; accepted 28 July 2004.DOI 10.1002/mrm.20289Published online in Wiley InterScience (www.interscience.wiley.com).

Magnetic Resonance in Medicine 52:1302–1309 (2004)

© 2004 Wiley-Liss, Inc. 1302

when reaching the predefined Tm of the liposome mem-brane. At physiologic temperatures, the T1-relaxation en-hancement of the liposomal agent is low due to limitedwater exchange. As the temperature approaches Tm, thetransmembrane water exchange kinetics is faster and theT1-relaxivity increases rapidly and markedly before level-ing off as Tm is exceeded (Fig. 1). Tissue with a tempera-ture above Tm would therefore be expected to appear withincreased signal intensity (SI) on T1-w images. An impor-tant secondary mechanism in vivo is leakage of GdDTPA-BMA from the liposomes into the surrounding tissue attemperatures above Tm. Following thermal coagulation ofparenchyma and vessels, it is assumed that the agent can-not migrate and therefore will interact with the tissuewhere it was released. Consequently, a persistent signalintensity enhancement on T1-w images may verify that atemperature of Tm or above was achieved.

The purpose of this study was to quantify relative signalintensity alterations associated with the use of paramag-netic thermosensitive liposomes during and after laser andradiofrequency ablation of rabbit liver in an interventionalMRI system.

MATERIALS AND METHODS

Liposomes

Liposomal gadodiamide (GdDTPA-BMA), whose membraneconsisted of 90% (w/w) distearoylphosphatidylcholine(DSPC) and 10% distearoylphosphatidylglycerol (DSPG),was prepared by the thin film hydration method (24). Theliposomes were subsequently sized down by membrane ex-trusion and untrapped GdDTPA-BMA was removed by dial-ysis. Key physicochemical liposomal properties, as mea-sured by standard methods described elsewhere (22,24),were effective Gd concentration 27 mM, extraliposomal pH7.2, and osmolality 355 mosmol/kg. Intensity weighted lipo-some size was 110 nm with a polydispersity index of 0.16,indicating a highly monomodal size distribution. The transi-tion temperature (Tm) of the liposomal membrane was 57°Cwith a SD of the Tm measurements of 1%. The liposomalagent was injected iv at a dose of 1.1 mL/kg, corresponding to30 mol Gd/kg. The thermosensitive liposomal agent was

supplied by Amersham Health AS. Figure 1 shows the invitro temperature dependence of the T1-relaxivity for liposo-mal gadodiamide (GdDTPA-BMA) used in this study. For therelaxometric measurements, the liposomal dispersion wasdiluted with pH-adjusted isosmotic glucose solution to aneffective Gd concentration of 1 mM and dispended in NMRtubes. Each tube was heated to a predetermined temperaturewithin a temperature range of 37–57°C in a thermostatedheating block at a heating rate of 10°C/min. When the desiredtemperature was reached, each sample was inserted into a0.47-T relaxometer (Minispec PC-120b, Bruker GmbH, Ger-many) for immediate relaxometric measurements.

Laser and Radiofrequency Equipment

The laser apparatus consisted of a neodymium-yttriumaluminum garnet (Nd-YAG) laser unit with a wavelengthof 1064 nm (Dornier mediLas Fibertom 4100, Dornier Med-Tech GmbH, Germany). The laser light was distributedthrough 1.9-mm-diameter flexible MRI-compatible lightguides designed for interstitial tissue ablation (DornierMedTech GmbH, Germany) with a 20-mm diffuser at thetip of the fiber.

A saline-enhanced, impedance-controlled radiofre-quency system (Elektrotom HiTT 106, Berchtold GmbH &Co, Germany) was used for radiofrequency ablation. The375-kHz radiofrequency generator delivered a nonmodu-lated high-frequency alternating current to a maximum of1.2 amp. Maximum power output of the generator could beset between 5 and 60 W. The actual power output wasadjusted automatically in accordance with the impedanceof the patient circuit. A syringe pump (Pilot C, FreseniusVial, France) was connected by a RS-232 connection to theRF generator. Output of a saline solution of 154 mmol(0.9%) NaCl per liter water was controlled by an algorithmbuilt into the RF generator. The saline-cooled MRI-com-patible RF electrodes had a diameter of 1.7 mm and a15-cm shaft. The saline solution was distributed to thetissue in the immediate vicinity of the tip of the electrodethrough three groups of two side holes each in the 15-mmexposed tip, placed at 120° angles from each other.

MRI-Compatible Fiber-Optic Thermometry System

Tissue temperature during the thermal treatment was mea-sured by a MRI-compatible fiber-optic thermometry sys-tem consisting of a flexible fiber-optic probe and an opto-electronic signal-processing unit (Multitemp 1601, Op-tomed, Norway). The outer diameter of the temperatureprobe was 1.5 mm. The probe had 16 measuring pointswith a sensor interval of 3 mm. All 16 sensors could beread out simultaneously in real-time with a temporal res-olution of 1 sec and an accuracy of 0.1°C. The systemwas calibrated to a temperature range of 0–150°C. Thetemperature monitoring system was connected by a RS-232 interface to a laptop with specially designed software(Multitemp). Temperature readings for all 16 sensors weresampled each second.

Animals and Operative Procedure

The experimental protocol was approved by the Depart-ment of Comparative Medicine at Rikshospitalet Univer-

FIG. 1. In vitro temperature dependence of the T1-relaxivity forliposomal and nonliposomal gadodiamide (GdDTPA-BMA). The ver-tical dashed line indicates the phase transition temperature of theliposomal membrane at 57°C.

Thermosensitive Paramagnetic Liposomes 1303

sity Hospital under the surveillance of the Norwegian An-imal Research Authority. Eight chinchilla-bastard rabbitswith a mean weight of 3.2 kg (2.7–4.0 kg) were treated inaccordance with current legislation governing animal care.After premedication with a subcutaneous injection of 0.1mL/kg fentanyl 0.315 mg/mL fluanisone 10 mg/mL(Hypnorm, Janssen), an iv cannula (Neoflon, Becton–Dick-inson) was inserted in the dorsal ear vein and 0.1 mL/kgfentanyl 0.315 mg/mL fluanisone 10 mg/mL (Hypnorm,Janssen) was administered iv. All rabbits were shaved inthe upper abdominal midline using a hair clipper (OsterGolden A5). The four rabbits to undergo radiofrequencyablation were shaved in an area of 10 20 cm on the lowerback, and a 134-cm2 dispersive plate (EZ 344–02, Berch-told, Germany) was applied. The rabbits were intubatedwith a 3- or 3.5-mm tracheal tube. All animals were anes-thetized by Isoflurane 1–1.2%, 50% oxygen, using aminute volume of 1.2 liter and respiratory frequency of30/min. A midline incision was made and the rabbits wereplaced inside the MRI head coil. To eliminate motionartifacts during scanning, the right medial, left medial, andleft lateral liver lobes were exteriorized and placed on ahorizontal 5-mm polycarbonate sheet fixated to the headcoil. Care was taken not to compromise the circulation ofthe liver.

Four animals were treated with laser ablation and fouranimals with radiofrequency ablation. The laser fiber orthe radiofrequency electrode was inserted into one of theliver lobes axially. The temperature fiber was insertedparallel in close proximity to the laser fiber or the RFelectrode, with the tip of the temperature sensor extending15 mm in front of the energy applicator. Each treatmentsession consisted of 9 min active heating. For the animalstreated with laser ablation, a maximum power setting of 20W and duration of 3 min was applied three times succes-sively for each ablation. Automated stepwise reduction ofthe laser effect (20/15/10 W each for 30 sec and 7 W for 90sec) was used. For the radiofrequency system a powersetting of 20 W and a continuous treatment time of 9 minwas used. One ablation was created in each of the threeexteriorized liver lobes of each animal. The first ablation ineach animal, denoted lesion A, was made prior to injectionof the agent. Tissue temperature during active heating andpassive cooling was monitored by real-time readout fromthe temperature probe in the liver. When all sensors on thetemperature probe were below 37°C, the laser fiber/RFelectrode and the temperature probe were removed andinserted into one of the two remaining liver lobes. Theliposomal agent was injected in a dorsal ear vein with aninfusion speed of 1 mL/min, and the vein was flushed with10 mL saline solution of 154 mmol NaCl. The secondablation, denoted lesion B, was made 12–14 min afteradministration of the agent, using the same energy modal-ity and protocol as the first ablation. The tissue tempera-ture was allowed to normalize, and the third ablation,denoted lesion C, was made accordingly in the remaininglobe 50–55 min after injection of the agent. The A, B, andC lesions were randomly assigned to each of the threeexteriorized liver lobes. After the MRI scans had beencompleted, the animals were killed by injecting100 mg/kgpentobarbital iv.

MRI Protocol

All procedures were performed on a 0.5-T vertically openwhole-body MRI system (Signa SP2, GE Medical Systems)using a transmit and receive head coil. An axial T1-weighted fast spin echo (FSE) sequence was used for im-aging. The imaging parameters were repetition time (TR) 300 msec, echo time (TE) 19 msec, flip angle (FA) 90°,slice thickness 4 mm, spacing 4.5 mm, 256 256 acqui-sition matrix, field of view 180 180 mm, and four signalaverages. Total duration of the sequence was 2 min, 16 sec.A scan was performed after each repositioning of the laserfiber or RF electrode, but before energy was applied. Forthe four animals operated with laser, one scan was per-formed every 3 min during heating for each of the abla-tions. The mean signal intensity of the three scans duringlaser heating was used for further analysis. Due to theelectromagnetic noise generated by the radiofrequency cir-cuit while operating, scanning in heated tissue duringtreatment was not possible for the four animals treatedwith radiofrequency ablation. A scan was therefore per-formed within 10 sec after termination of the radiofre-quency treatment. The liver tissue was allowed to reachbaseline temperature and a postablation scan at baselinetemperature was performed for all animals. This scan pro-tocol was repeated for each of the three ablations. Scansperformed prior to heating, in heated tissue, and aftercooling of the tissue were denoted t0, t1, and t2, respec-tively. The scan at t0 for B and C lesions was made15–20 min after cooling of the tissue of the previous abla-tions. Signal intensity of A and B lesions 15–20 min aftercooling of the tissue, denoted t3, could therefore be mea-sured from the t0 scans of B and C lesions, respectively. Noscan was done 15–20 min after normalization of the tissuetemperature for C lesions. Hence, signal intensity at t3 istherefore not available for C lesions, and statistical analy-sis is performed only for the values measured at t1 and t2for all lesion types. To evaluate signal intensity changesdue to the agent unrelated to heating, signal intensity ofthe reference regions of interest (ROI)s was measured pre-vious to administration of the agent and 3 min after injec-tion of the agent, but prior to the generation of the Blesions.

Analysis of MR Images

The signal intensity of the ablated area versus the un-treated tissue was measured from uncompressed DICOMimages by image analysis software designed for quantita-tive image analysis (DIMview, Rikshospitalet UniversityHospital, Oslo, Norway) running on a Windows XP com-puter system. Ablations as seen on T1-w images consistedof a dark central area at the needle track with diameters of1–5 mm surrounded by a brighter rim. Each ablation wassegmented manually using a polygon tool, avoiding thecentral darker part. The mean value of the signal intensityof this ROI (SIlesion) was used for further analysis. Tocompensate for a potential general variation of signal in-tensity in the liver during the procedure, each signal in-tensity measurement was normalized using two indepen-dent regions of interest (SIref1 and SIref2) from liver tissuenot exposed to the local thermotherapy on the same image

1304 Frich et al.

as the ablations. The subtraction of the mean signal inten-sity of these two reference ROIs from the signal intensity ofthe ablation was used as an expression of the relativealteration of signal intensity between the ablation and thesurrounding liver tissue.

SI% SIlesion SIref1 SIref2

2 100

The largest diameter of each ablation was measuredfrom the MR images by an investigator blinded to theresults of the corresponding measurements on the macro-scopic specimens.

Analysis of Liver Tissue

The livers were harvested immediately after the animalswere killed. To retain the orientation of the ablated area, amarker was inserted into the tracks from the laser fiber orRF electrode, and each ablation was excised with widemargins. Using a custom-built slicer, the ablated tissuewas cut in 4-mm slices axial to the inserted marker. Allslices were photographed with a 2.1-megapixel digitalcamera (Canon, Tokyo, Japan) under standardized light-ning conditions with a L-shape ABFO No. 2 photomacro-graphic scale (Lightning Powder Company, Jacksonville,FL). The images were imported into the scientific imageanalysis program ImageJ version 1.3 (National Institute ofMental Health, Bethesda, MD) running on an AppleMacintosh OSX computer system. Each image was cali-brated by measuring the number of pixels per unit lengthof the photomacrographic scale. Image resolution was typ-ically 10 pixel/mm. Slices from each ablation were man-ually segmented using the border of normal liver tissueversus tissue with macroscopically visible changes, andthe largest diameter of the ablation was determined.

Statistical Analysis

SPSS for Mac OSX version 11.0.2 (SPSS, Chicago, IL) wasused for statistical analysis. Analysis of the normalizedsignal intensity of the ablations was performed based on avariable component model, with animal as random com-ponent and lesion type and relative time as fixed compo-nents. This statistical model accounts for dependency inobservations from the same animal. Post hoc tests werebased on Scheffe’s method. Paired samples t test was usedfor analysis of significance of the observed signal intensityalterations of the agent unrelated to heating.

RESULTS

In one of the rabbits in the laser treatment group only 2ablations were made due to the anatomy of the liver. In theremaining seven animals 1 ablation was made in each ofthe three exteriorized liver lobes. Hence, 11 laser ablationswere made in four animals and 12 radiofrequency abla-tions were made in four animals. At the time of removal ofthe liver, macroscopic lesions were identifiable for allattempted ablations. Energy deposited during each radio-frequency ablation was typically 10,500 J. Examples oftemperature distribution during laser and radiofrequency

ablation are shown in Fig. 2a and b. During laser ablation,temperatures in close proximity to the laser applicator tipexceeded the 150°C calibration limit of the temperaturesensor for several of the ablations. With the radiofrequencysystem, the highest registered temperature was 104°C. Atemperature above 57°C was registered by at least one ofthe temperature sensors in the first 3 min of passive cool-ing following the treatment. During scanning in this ther-mal imaging window, increased signal intensity of theareas with a temperature above the transition temperatureof the agent could be expected. Tissue temperature re-turned to baseline within 10 min following termination ofthe energy application.

Signal Intensity prior to Administration of the Agent

For the animals in the laser group, decreased SI was seenfor the A lesions for scans at t1 during heating (SI 3.5%, SD 6.4) followed by a signal increase (SI 12.2%, SD 2.3) at t2 when the tissue had returned tobaseline temperature. On scan t3 15–20 min after the tis-sue had reached baseline temperature, no further discern-able increase in SI (SI 13.0%, SD 5.7) was seen (Fig.

FIG. 2. a: Temperature distribution in rabbit liver during and afterlaser ablation. Laser energy was applied in three repeated cycles,each of 3 min. The distance between each temperature sensor was3 mm. The horizontal dashed line indicates the transition tempera-ture of 57°C for the MRI agent used. Temperature has reached 37°Cin all sensors 15 min after the ablation. b: Temperature distributionin rabbit liver during and after radiofrequency ablation. The distancebetween each temperature sensor was 3 mm. The horizontaldashed line indicates the transition temperature of 57°C for the MRIagent used. Temperature reached 37°C in all sensors 15 min afterthe ablation.

Thermosensitive Paramagnetic Liposomes 1305

3a). A similar pattern was seen for the A lesions in theanimals in the radiofrequency group, where scans at t1showed decreased SI (SI -6.9%, SD 8), followed by asignal increase at t2 (SI 4.7%, SD 10.5) and t3 (SI 9.1%, SD 7.9) (Fig. 3b).

Signal Intensity after Administration of the Agent

The change in signal intensity prior to and 3 min after ivinjection of the contrast agent unrelated to heating was notstatistically significant (SI 1.4%, SD 4.5, P 0.35).For animals in the laser group a slight increase in signalintensity was measured for B lesions at t1 (SI 2.3%,SD 6.1), followed by an increase at t2 (SI 18.8%,SD 20.0) and a further increase at t3 (SI 34.5%, SD 11.9). For C lesions an increase during ablation at t1 (SI 16.4%, SD 7.1) followed by a further increase at t2 (SI 31.5%, SD 8.9) was found (Fig. 3a). Test for compar-isons of SI measurements found significance at the 0.05level between A lesions and C lesions with a P value lessthan 0.05 and a confidence interval (CI) of 6.95–22.05. Thedifference between B lesions and C lesions was also sig-nificant (P 0.049, CI 0.03–15.13). Difference betweenA lesions and B lesions did not reach significance whenperforming an analysis including only SI measurements att1 and t2 (P 0.053, CI -0.07–13.91). The results for theanimals in the radiofrequency group (Fig. 3b) were com-parable to those in the laser treatment group. The B lesionsshowed a minor signal increase at t1 (SI 5.4%, SD 9.5) followed by a further increase at t2 (SI 13.7%, SD 9.7) and at t3 (SI 21.6%, SD 22.7). For C lesions anincrease at t1 (SI 19.3%, SD 12.7) was found, fol-lowed by a further increase at t2 (SI 26.3%, SD 10.9).In the radiofrequency group, test for comparisons of meansof SI found a significance between A lesions and C lesionswith a P value less than 0.05 and a CI of 7.67–27.66. Thedifference between B lesions and C lesions was also sig-nificant (P 0.048, CI 0.09–20.08). The difference be-tween A lesions and B lesions did not reach significancewhen performing an analysis including only SI measure-ments at t1 and t2 (P 0.165, CI -2.41–17.58). TypicalMR images of an ablation made after administration of theagent are shown in Fig. 4. Figure 5 shows the correspond-ing macroscopic specimen.

FIG. 3. a: Mean relative signal intensity on T1-weighted imagesduring and after laser ablation of four rabbit livers. A lesions weremade prior to injection of the agent. B lesions were made 12–14 minafter injection of the agent and C lesions 50–55 min after injection ofthe agent. Relative time is indicated on the horizontal axis; duringheating (t1), after the tissue temperature is normalized (t2), and15–20 min after normalization of tissue temperature (t3). The valueof t1 is the average of three scans made during application of laserenergy. Intensity measurement at t3 is not available for C lesions. b:Mean relative signal intensity on T1-weighted images during andafter radiofrequency ablation of four rabbit livers. A lesions weremade prior to injection of the agent. B lesions were made 12–14 minafter injection of the agent and C lesions 50–55 min after injection ofthe agent. Relative time is indicated on the horizontal axis; duringheating (t1), after the tissue temperature is normalized (t2), and15–20 min after normalization of tissue temperature (t3). Intensitymeasurement at t3 is not available for C lesions.

FIG. 4. T1-weighted images of radiofrequency ablation (lesion B) in rabbit liver after injection of liposomal contrast. Prior to heating (t0),during heating (t1), after normalization of tissue temperature (t2), and 15–20 min after normalization of tissue temperature (t3). Note theincreasing signal intensity in the periphery of the thermal lesion in the right liver lobe (white arrows). The radiofrequency electrode has beenrepositioned in the left liver lobe at t3 (white arrowhead).

1306 Frich et al.

Diameter of Ablations on MRI and MacroscopicSpecimens

The largest diameter of the ablations perpendicular to theprobe track was measured from the photographs of themacroscopic specimens by one of the investigators (LF).The mean diameter of the 11 laser ablations was 16.3 mm(SD 1.2; 14–18 mm). The corresponding figure for the 12radiofrequency ablations was 24.3 mm (SD 3.8; 18–32 mm). Another investigator (AB) blinded for the resultsof the analysis of the macroscopic specimens measured thelargest diameter for all ablations from the MR images at t2.For the 11 laser ablations the mean and SD of the differ-ence between the diameter as measured from the photo-graphs and the MR images were 1.7 1.3 mm. For 3 of the12 radiofrequency ablations, a reliable measurement of thelargest diameter could not be established from the MRimages. For the 9 radiofrequency ablations where a diam-eter could be determined, the corresponding figures were0.3 3.4 mm. Figure 6 shows a scatterplot of the diam-eter of the ablations as measured from the MR images andthe photographs of the specimens.

DISCUSSION

Several active and passive methods for noninvasive ther-mometry within the human body exist, utilizing transmis-sion or reflection of microwaves, ultrasound, or MRI (25).In the present study, we examined the feasibility of theliposome encapsulated paramagnetic agent gadodiamide(GdDTPA-BMA) as a thermosensitive probe during MRI-guided laser ablation and radiofrequency ablation of rabbitliver in vivo.

Local application of heat in tissues leads to alternationsin metabolic, macromolecular, extracellular, and vascularparameters (26). Irreversible cellular damage is a functionof temperature and exposure time and is dependent on thetype of tissue. Denaturation of proteins occurs after expo-sure for temperature above 50–55°C and leads to macro-scopically visible tissue changes in liver tissue. Tissuebecomes desiccated when reaching 90–100°C and carbon-ized when temperatures exceed 150°C (5,27,28). Revers-ible and irreversible structural and biologic tissue changesrelated to ablation (26) as well as the influence of liposo-

mal GdDTPA-BMA contributed to the alterations in signalintensity on T1-w images observed in the present study.Additionally, a direct thermal T1-effect was present forscans in heated tissue. The correlation between tissueT1-relaxation time and temperature is complex and non-linear if one considers the entire temperature range inves-tigated in the present study (37 to 150°C) (26,29). In thetemperature range of 37–45°C, a linear increase in T1 withtemperature is expected (14,30). At higher temperatures, co-agulation and tissue denaturation occur, leading to nonlinearand irreversible T1 changes, and the T1-temperature correla-tion becomes complex and tissue dependent (29,31).

Alterations of T1 related to the ablation and a directthermal T1-effect contributed to the observed signal inten-sity for A lesions made prior to injection of the liposomalagent. A lesions therefore provided a reference that facili-tated quantification of T1-changes that could be attributedto the T1-effect of the liposomal agent for subsequent B andC lesions. Signal intensity of A lesions decreased to 4–7%below baseline in both the laser and the radiofrequencygroups on scans at t1 performed in heated tissue. On scansat t2 after normalization of the tissue temperature, SIincreased 5–12% above baseline. These findings are inaccordance with previous results from laser ablation ofrabbit liver (32). No further increase in SI was seen for Alesions at t3, 15–20 min after normalization of the tissuetemperature. The slope between t1 and t2 can possibly beexplained by the direct thermal T1-effect causing a netincrease in T1 for scans acquired in heated tissue at t1. Inscans at normalized temperature at t2, the direct thermalT1-effect is no longer present and the observed 5–12%increase in signal intensity at t2 can be attributed to struc-tural and biologic tissue changes related to the ablation.Intravenous injection of the liposomal agent in tissue withnormalized temperature unrelated to heating did not causea general increase in signal intensity, confirming that theT1-relaxation enhancement of the agent was negligible at atemperature well below the phase transition temperatureof the liposomal membrane (Fig. 1). For B lesions a meanincrease in SI of 2–5% compared to baseline was found at

FIG. 6. Largest diameter of 11 laser ablations and 9 radiofrequencyablations measured from macroscopic specimens and T1-weightedMR images.

FIG. 5. Macrophotograph of the ablated liver tissue depicted inFig. 4.

Thermosensitive Paramagnetic Liposomes 1307

t1. The signal intensity for B lesions increased to 14–19%at t2 and further to 22–35% at t3. The nonreversible con-trast enhancement after cooling could be explained by aleakage of GdDTPA-BMA from the liposome interior andsubsequent entrapment due to coagulation of vessels. Theincreased signal intensity stabilized and was present onscans performed more than 60 min after termination of thethermal treatment. Signal intensity for C lesions was sig-nificantly increased to 16–19% at t1 with a further in-crease to 26–32% at t2. The magnitude of the normalizedsignal intensity enhancements for B and C lesions was com-parable to that found in a recent study of focused ultrasoundablation of rabbit liver using the same liposomal agent (23).The slope between t1 and t2 observed for B and C lesions wascomparable to the slope between t1 and t2 observed for Alesions, suggesting that the considerably lower SI seen duringscanning in heated tissue at t1 compared to t2 for B and Clesions was primarily caused by two counteracting processestaking place during heating. The alterations of tissue param-eters induced by the thermal treatment and the direct ther-mal T1-effect caused a net increase of tissue T1 as seen on t1scans of A lesions, whereas the activated liposomal agentdecreased T1 of the tissue.

Increased signal intensity by a constant offset was ob-served in C lesions compared to B lesions and in B lesionscompared to A lesions in both the laser group and the radio-frequency group. Conventional liposomes are rapidly opso-nized in blood after iv injection and taken up by the mono-nuclear phagocyte system (MPS) with the liver as a majororgan. For a liposome size of 110 nm as used in this study,Kuppfer cells are responsible for the liposome uptake inliver, but a minor uptake by the hepatocytes is also to beexpected. This intracellular uptake increases with timepostinjection and reaches a plateau, with the latter depen-dent on factors such as liposome size and coating (33). Bio-distribution studies of the current liposome formulation inrabbits using the same dosage as used in the present studyshow a rapid blood clearance with only 35% of the injecteddose remaining in the blood 5 min postinjection. The liveruptake at 3 h postinjection was about 40%, a typical figurefor the liposome size and membrane investigated (unpub-lished data). The offset between B lesions made 12–14 minafter administration of the agent and C lesions made 50–55 min after administration of the agent may therefore beexplained by a time-dependent hepatic intracellular uptakeof the liposomal agent, where the intracellular concentrationof GdDTPA-BMA was higher at the time of creation of Clesions compared to B lesions.

An increase in signal intensity was observed for B lesionsat t3 compared to t2 in both in the laser and the radiofre-quency group (Fig. 3a and b). A corresponding slope betweent2 and t3 was not present for A lesions. The reason for thisdelayed enhancement is not established. However, delayedfocal signal increase can be seen on T1-w images in infarctedmyocardial tissue following injection of extracellular gado-linium agents (34). This finding is believed to be caused byloss of integrity of the cell membranes in the infarcted tissue,with a subsequent local increase of the distribution volumeof the agent. A hypothesis could be made that a similar localincrease in the distribution volume occurs in liver tissueexposed to thermal ablation, giving rise to a time-dependentaccumulation of GdDTPA-BMA released from the liposome

interior into the surrounding tissue at tissue temperaturesabove Tm.

The correlation between the largest diameters of the abla-tions as measured from the MR images and the macroscopicspecimens was not comparable to that found in a studyexamining the correlation between GdDTPA-enhanced MRIand histomorphologic findings of laser ablation in rabbit liver(35). Several factors might contribute to this inconsistency.The contrast agent used in this study enhances the ablationsite, and not the surrounding normal liver tissue, as is thecause for conventional MRI liver contrast agents. Irreversibletissue damage is dependent on both the temperature and theexposure time. The temperature needed for irreversible tis-sue damage with an exposure time of 9 minutes as used inthis study might be lower than the 57°C temperature thresh-old of the liposomal agent (5). Hence, irreversible tissuedamage could occur in areas exposed to a temperature belowthe transition temperature of the liposomal agent. Interest-ingly, the diameters measured from the laser specimens werelarger than those measured from the MR images for all abla-tions except for one, suggesting that enhancement is onlypresent in limited areas within the thermal lesion. For theradiofrequency ablations, which were larger and more irreg-ular than the laser ablations, the correlation was poorer.During radiofrequency ablation, saline was introduced at theablation site. Saline is associated with increasing T1-relax-ation times and could cause a decrease in signal intensity onT1-w images. Additionally, the presence of saline might leadto local inhomogeneities in the concentration of the liposo-mal agent. The lack of correlation between images and spec-imens could also partly be explained by methodological lim-itations. For example, even if care was taken to retain theorientation of the ablated area, slicing of the liver specimenswere not necessarily identical to the plane of the axial T1-wimages.

This experimental study was performed in a small ani-mal model subject to several limitations. Findings in rabbitliver tissue are not necessarily transferable to those in ahuman liver. Furthermore, all treatments were conductedin normal liver parenchyma. Vascular divergence betweenintratumoral and peritumoral blood flow and surroundingnormal liver (36) could lead to significant inhomogeneousdistribution of the contrast agent of possible importance.In our study, the temperature-sensitive liposomal agentwas not compared to other established MRI contrast-en-hanced methods. Compared to our A lesions, conspicuitycan probably be improved by conventional MRI contrastagents (37) or by using other imaging modalities such ascontrast-enhanced ultrasound (38). However, consistentresults were found that confirm the ability of the liposomalagent to differentiate between tissue exposed to the ther-mal treatment and normal liver tissue.

In conclusion, a significant nonreversible increase insignal intensity was found for B and C lesions made afterinjection of the liposomal agent compared to control Alesions in both the laser ablation group and the radiofre-quency group. The persistent signal enhancement in areasexposed to a temperature above the threshold temperaturemay be of clinical value during MRI-guided thermal abla-tion, especially for local heating modalities, such as radio-frequency ablation, where other MRI-based thermometrymethods are not feasible.

1308 Frich et al.

ACKNOWLEDGMENTS

The authors thank Unni Nordby Wiggen and Astrid Rogs-tad, Amersham Health AS, for the preparation and char-acterization of the liposomes. Thore Egeland, Rikshospita-let University Hospital, provided statistical assistance. Wethank the staff at the Interventional Centre at Rikshospita-let for providing excellent assistance. We also thank theDepartment of Comparative Medicine at Rikshospitalet forsupport in animal care.

REFERENCES

1. Vogl TJ, Straub R, Eichler K, Sollner O, Mack MG. Colorectal carcinomametastases in liver: laser-induced interstitial thermotherapy—local tu-mor control rate and survival data. Radiology 2004;230:450–458.

2. Farrell MA, Charboneau WJ, DiMarco DS, Chow GK, Zincke H, CallstromMR, Lewis BD, Lee RA, Reading CC. Imaging-guided radiofrequency ab-lation of solid renal tumors. Am J Roentgenol 2003;180:1509–1513.

3. Marangio A, Prati U, Luinetti O, Brunetti E, Filice C. Radiofrequencyablation of colorectal splenic metastasis. Am J Roentgenol 2002;178:1481–1482.

4. Nishida T, Inoue K, Kawata Y, Izumi N, Nishiyama N, Kinoshita H,Matsuoka T, Toyoshima M. Percutaneous radiofrequency ablation oflung neoplasms: a minimally invasive strategy for inoperable patients.J Am Coll Surg 2002;195:426–430.

5. Storm FK, Harrison WH, Elliott RS, Morton DL. Hyperthermic therapyfor human neoplasms: thermal death time. Cancer 1980;46:1849–1854.

6. Goldberg SN, Gazelle GS, Mueller PR. Thermal ablation therapy forfocal malignancy: a unified approach to underlying principles, tech-niques, and diagnostic imaging guidance. Am J Roentgenol 2000;174:323–331.

7. Morikawa S, Inubushi T, Kurumi Y, Naka S, Sato K, Tani T, YamamotoI, Fujimura M. MR-guided microwave thermocoagulation therapy ofliver tumors: initial clinical experiences using a 0.5 T open MR system.J Magn Reson Imaging 2002;16:576–583.

8. Bleicher RJ, Allegra DP, Nora DT, Wood TF, Foshag LJ, Bilchik AJ.Radiofrequency ablation in 447 complex unresectable liver tumors:lessons learned. Ann Surg Oncol 2003;10:52–58.

9. Kohrmann KU, Michel MS, Gaa J, Marlinghaus E, Alken P. High inten-sity focused ultrasound as noninvasive therapy for multilocal renal cellcarcinoma: case study and review of the literature. J Urol 2002;167:2397–2403.

10. Patterson EJ, Scudamore CH, Owen DA, Nagy AG, Buczkowski AK.Radiofrequency ablation of porcine liver in vivo: effects of blood flowand treatment time on lesion size. Ann Surg 1998;227:559–565.

11. Aschoff AJ, Merkle EM, Wong V, Zhang Q, Mendez MM, Duerk JL,Lewin JS. How does alteration of hepatic blood flow affect liver perfu-sion and radiofrequency-induced thermal lesion size in rabbit liver? JMagn Reson Imaging 2001;13:57–63.

12. Solbiati L, Livraghi T, Goldberg SN, Ierace T, Meloni F, Dellanoce M,Cova L, Halpern EF, Gazelle GS. Percutaneous radio-frequency ablationof hepatic metastases from colorectal cancer: long-term results in 117patients. Radiology 2001;221:159–166.

13. Kettenbach J, Kostler W, Rucklinger E, Gustorff B, Hupfl M, Wolf F,Peer K, Weigner M, Lammer J, Muller W, Goldberg SN. Percutaneoussaline-enhanced radiofrequency ablation of unresectable hepatictumors: initial experience in 26 patients. Am J Roentgenol 2003;180:1537–1545.

14. Quesson B, de Zwart JA, Moonen CT. Magnetic resonance temperatureimaging for guidance of thermotherapy. J Magn Reson Imaging 2000;12:525–533.

15. Jolesz FA, Bleier AR, Jakab P, Ruenzel PW, Huttl K, Jako GJ. MRimaging of laser-tissue interactions. Radiology 1988;168:249–253.

16. Gould SWT, Vaughan NV, Gedroyc W, Lamb G, Goldin R, Darzi A.Monitoring of interstitial laser thermotherapy with heat-sensitive co-lour subtraction magnetic resonance imaging: calibration with absolutetissue temperature and correlation with predicted lesion size. LasersMed Sci 1999;14:250–256.

17. Germain D, Chevallier P, Laurent A, Savart M, Wassef M, Saint-JalmesH. MR monitoring of laser-induced lesions of the liver in vivo in a low-

field open magnet: temperature mapping and lesion size prediction. JMagn Reson Imaging 2001;13:42–49.

18. Palussiere J, Salomir R, Le Bail B, Fawaz R, Quesson B, Grenier N,Moonen CT. Feasibility of MR-guided focused ultrasound with real-time temperature mapping and continuous sonication for ablation ofVX2 carcinoma in rabbit thigh. Magn Reson Med 2003;49:89–98.

19. Zhang Q, Chung YC, Lewin JS, Duerk JL. A method for simultaneous RFablation and MRI. J Magn Reson Imaging 1998;8:110–114.

20. Oshiro T, Sinha U, Lu D, Sinha S. Reduction of electronic noise fromradiofrequency generator during radiofrequency ablation in interven-tional MRI. J Comput Assist Tomogr 2002;26:308–316.

21. Bartholet A, Goudemant J, Laurent S, Kahn O, Vander Elst L, Muller R.Spin transaction molecular materials: intelligent contrast agents formagnetic resonance thermometry. In: Proceedings of the 8th AnnualMeeting of ISMRM, 2000. p 380.

22. Fossheim SL, Il’yasov KA, Hennig J, Bjornerud A. Thermosensitiveparamagnetic liposomes for temperature control during MR imaging-guided hyperthermia: in vitro feasibility studies. Acad Radiol 2000;7:1107–1115.

23. McDannold N, Fossheim SL, Rasmussen H, Martin H, Vykhodtseva N,Hynynen K. Heat-activated liposomal MR contrast agent: initial in vivoresults in rabbit liver and kidney. Radiology 2004;230:743–752.

24. Fossheim SL, Fahlvik AK, Klaveness J, Muller RN. Paramagnetic liposomesas MRI contrast agents: influence of liposomal physicochemical properties onthe in vitro relaxivity. Magn Reson Imaging 1999;17:83–89.

25. Miyakawa M, Bolomey J, editors. Non-invasive thermometry of thehuman body. Boca Raton, FL: CRC Press; 1996. 257 p.

26. Graham SJ, Stanisz GJ, Kecojevic A, Bronskill MJ, Henkelman RM.Analysis of changes in MR properties of tissues after heat treatment.Magn Reson Med 1999;42:1061–1071.

27. Martinez AA, Meshorer A, Meyer JL, Hahn GM, Fajardo LF, Prionas SD.Thermal sensitivity and thermotolerance in normal porcine tissues.Cancer Res 1983;43:2072–2075.

28. Heisterkamp J, van Hillegersberg R, IJzermans JNM. Critical tempera-ture and heating time for coagulation damage: implications for inter-stitial laser coagulation (ILC) of tumors. Lasers Surg Med 1999;25:257–262.

29. Graham SJ, Bronskill MJ, Henkelman RM. Time and temperature de-pendence of MR parameters during thermal coagulation of ex vivorabbit muscle. Magn Reson Med 1998;39:198–203.

30. Cline HE, Hynynen K, Hardy CJ, Watkins RD, Schenck JF, Jolesz FA.MR temperature mapping of focused ultrasound surgery. Magn ResonMed 1994;31:628–636.

31. Peller M, Reinl HM, Weigel A, Meininger M, Issels RD, Reiser M. T1relaxation time at 0.2 Tesla for monitoring regional hyperthermia:Feasibility study in muscle and adipose tissue. Magn Reson Med 2002;47:1194–1201.

32. Morrison PR, Jolesz FA, Charous D, Mulkern RV, Hushek SG, MargolisR, Fried MP. MRI of laser-induced interstitial thermal injury in an invivo animal liver model with histologic correlation. J Magn ResonImaging 1998;8:57–63.

33. Fossheim SL, Colet JM, Mansson S, Fahlvik AK, Muller RN, KlavenessJ. Paramagnetic liposomes as magnetic resonance imaging contrastagents. Assessment of contrast efficacy in various liver models. InvestRadiol 1998;33:810–821.

34. Pereira RS, Prato FS, Wisenberg G, Sykes J, Yvorchuk KJ. The use ofGd-DTPA as a marker of myocardial viability in reperfused acute myo-cardial infarction. Int J Cardiovasc Imaging 2001;17:395–404.

35. Germer C, Isbert CM, Albrecht D, Ritz JP, Schilling A, Roggan A, WolfKJ, Muller G, Buhr H. Laser-induced thermotherapy for the treatment ofliver metastasis. Correlation of gadolinium-DTPA-enhanced MRI withhistomorphologic findings to determine criteria for follow-up monitor-ing. Surg Endosc 1998;12:1317–1325.

36. Srivastava DN, Mahajan A, Berry M, Sharma MP. Colour Doppler flowimaging of focal hepatic lesions. Australas Radiol 2000;44:285–289.

37. Aschoff AJ, Rafie N, Jesberger JA, Duerk JL, Lewin JS. Thermal lesionconspicuity following interstitial radiofrequency thermal tumor abla-tion in humans: A comparison of STIR, turbo spin-echo T2-weighted,and contrast-enhanced T1-weighted MR images at 0.2 T. J Magn ResonImaging 2000;12:584–589.

38. Liu JB, Goldberg BB, Merton DA, Rawool NM, Needleman L, ForsbergF. The role of contrast-enhanced sonography for radiofrequency abla-tion of liver tumors. J Ultrasound Med 2001;20:517–523.

Thermosensitive Paramagnetic Liposomes 1309

III

Eur Radiol (2006) 16: 1990–1999DOI 10.1007/s00330-006-0177-6 EXPERIMENTAL

Lars FrichPer Kristian HolSumit RoyTom MalaBjørn EdwinOle Petter F. ClausenIvar P. Gladhaug

Received: 8 December 2005Revised: 19 January 2006Accepted: 20 January 2006Published online: 16 March 2006# Springer-Verlag 2006

Experimental hepatic radiofrequency ablationusing wet electrodes: electrode-to-vesseldistance is a significant predictor for delayedportal vein thrombosis

Abstract The aim of this study was toexamine possible explanatoryvariables associated with acute anddelayed portal vein thrombosis afterhepatic radiofrequency (RF) ablationusing wet electrodes. Coagulationswere created within 1.5 cm of the rightportal vein (RPV) branch in 12 pigswith (n=6) or without (n=6) Pringlemanoeuvre. Sham operations withPringle manoeuvre were performed infour animals. Rotational portalvenography was performed prior toablation, 10 min after ablation and 4days after ablation. Vessel diametersand vessel patency was determinedfrom the portal venograms. Distancebetween the ablation electrode andRPV was measured from 3-dimen-sional reconstructions of the portalvenograms. The portal veins wereexamined by microscopy. Delayedportal vein thrombosis was found intwo of six animals in the Pringle groupand three of six animals in the control

group 4 days after ablation (P=1.0,Fisher’s exact test). All five occur-rences of delayed portal vein throm-bosis were found in the six animalswith a distance between the ablationelectrode and RPV of 5 mm or less(P=0.030), indicating that theelectrode-to-vessel distance may be anindependent explanatory factor fordelayed portal vein thrombosis afterRF ablation with wet electrodes.

Keywords Radiofrequency (RF)ablation . Portal vein thrombosis .Complication . Experimentalanimal study

Introduction

Hepatic resection may be contraindicated in patients withtumours located close to central hepatic vessels, if resectionof these vessels is associated with unacceptable risk ofhepatic insufficiency due to extensive loss of functional liverparenchyma. For cirrhotic patients with a deficit in functionalhepatic reserve this risk is greatly increased [1]. Radio-frequency (RF) ablation minimises loss of functional liverparenchyma and is a treatment option for patients with non-resectable malignant liver tumours [2–5]. Treatment of peri-vascular tumours not amenable to resection is a potentiallyimportant clinical application for hepatic RF ablation.

Large vessels in the proximity of tumours treated withRF ablation are a predictor of incomplete tumour destruc-tion and, consequently, local tumour progression [6–8].Ablation of perivascular tumours, i.e. tumours contiguouswith a vessel with a diameter of 3 mm or more, is restrictedby the heat-sink effect, where the region to be coagulated iscooled by blood flow in large vessels [9, 10]. Occlusion ofhepatic vascular inflow, by clamping of the portal vein andthe hepatic artery (Pringle manoeuvre), can be expected todecrease the heat-sink effect and reduce perfusion of thehepatic parenchyma. Use of the Pringle manoeuvre isassociated with increased energy efficiency [11] andincreased ablation volume [9, 12]. Additionally, hepatic

L. Frich . P. K. Hol . S. Roy . B. EdwinThe Interventional Centre,Rikshospitalet University Hospital,0027 Oslo, Norway

L. Frich (*) . T. Mala .B. Edwin . I. P. GladhaugDepartment of Surgery,Rikshospitalet University Hospital,0027 Oslo, Norwaye-mail: lars.frich@labmed.uio.noTel.: +47-230-71825Fax: +47-230-72267

O. P. F. ClausenDepartment and Institute of Pathology,Rikshospitalet University Hospital,0027 Oslo, Norway

I. P. GladhaugDepartment of Surgery,Faculty Division Rikshospitalet,University of Oslo,0027 Oslo, Norway

vascular occlusion enhances coagulation necrosis close tohepatic vessels [6] and may be advantageous for ablation ofperivascular tumours [8]. However, vascular inflow occlu-sion may make hepatic vessels more vulnerable to thermalinjury and subsequent thrombosis [13]. Portal vein throm-bosis related to RF ablation may lead to greater loss offunctional liver parenchyma than anticipated, increasingthe risk of postoperative hepatic insufficiency.Risk factors associated with occurrence of portal vein

thrombosis following RF ablation have not been fullyestablished. The aim of this experimental study was toexamine possible explanatory variables associated withacute and delayed portal vein thrombosis after RF ablationin proximity to the right portal vein (RPV) branch inporcine liver. Specifically, the impact of performing hepaticvascular occlusion, and the distance between the ablationelectrode and RPV was investigated.

Materials and methods

Study design

The protocol was approved by the institutional Departmentof Comparative Medicine. Sixteen domestic pigs with amean weight of 30.7 kg (SD=3.1) were included in thestudy. Twelve animals were randomly allocated to twoexperimental groups. In one group Pringle manoeuvre wasapplied during the ablation (n=6). In the control group, RFablation was performed without Pringle manoeuvre (n=6).Four animals were assigned to a sham group where Pringlemanoeuvre, but no RF ablation, was performed. Patency ofportal vessels in all groups was determined by rotationalportal venography prior to ablation, 10 min after ablationand 4 days after ablation. The portal veins were examinedby microscopy after euthanasia.

Radiofrequency equipment

An impedance-controlled RF ablation system (ElektrotomHiTT 106, Berchtold GmbH & Co, Germany) was used forcreating coagulations. The system incorporated a generatorthat delivered alternating current at 375 kHz to a maximumof 1.2 A. Power output of the generator could be setbetween 5 W and 60 W. The actual power output wasadjusted by the generator in accordance with the imped-ance of the patient circuit. Tissue impedance and cumula-tive deployed energy were displayed in real-time. Theactive radio-opaque electrode was 1.7 mm in diameter,with a 15 cm long shaft incorporating a lumen for perfusionof saline solution. The terminal non-insulated segment ofthe electrode was 15 mm in length. A syringe pump (PilotC, Fresenius Vial, France) was used for infusion of isotonicsaline solution (0.9% NaCl) through the electrode. Thesaline solution was distributed to the tissue via three pairs

of two side holes in the electrode tip, placed at a 120° anglefrom each other. The pump was connected by a RS-232serial port to the RF generator. The infusion rate of salinesolution was determined by the pre-selected power outputof the generator and the circuit impedance. Under a powersetting of 50 W, saline solution was infused at a rate of105 ml/h, which corresponded to a volume of 16 ml during9 min ablation. If an upper impedance threshold of 900 Ωwas reached, five-times the value of the nominal flow wasinfused for 1.2 s. Radiofrequency energy was applied for1 min with an effect of 30 W, immediately followed by8 min with 50 W. Maximum cumulated electrical energygenerated during one treatment cycle with this protocolwas 25,800 J. A new electrode was used for each animal.

Angiography system

A floor mounted C-arm based angiography system(Angiostar O.R., Siemens, Erlangen, Germany) was usedfor acquisition of rotational venographic images of theportal veins. During each portal venography, 60 mliodixanol 320 mg I/ml (Visipaque, Amersham Health,Oslo) was injected into the splenic vein by an injector at5 ml/s. Starting 3 s after the injection commenced, a seriesof 160 portal venograms was obtained while the C-armrotated through 198° over 8 s. The set of image data wastransferred to a post-processing workstation (Leonardo,Siemens), and a 3-dimensional (3D) portal venogram wascreated using a volume-rendering algorithm.

Operating procedure

After the animals had undergone premedication withintramuscular injection of ketamine 15–20 mg/kg (Ketalar,Pfizer), atropine sulphate 1 mg (Atropin, NycomedPharma) and azaperone 2–4 mg/kg (Stresnil, Janssen-Cilag), an intravenous cannula was placed in an ear vein,and the animal was transported to a combined angiographyand operating theatre. Ampicillin 250 mg was administeredintramuscularly prior to surgery and after the surgicalprocedure had been completed. General anaesthesia wasinduced with intravenous administration of pentobarbital150–400 mg and morphine 10–30 mg. After endotrachealintubation, anaesthesia was maintained with isoflurane 1–1.2% in oxygen. The analgesic effect was supplementedwith morphine infusion at 10 mg/h. A thermodilutionSwan–Ganz catheter 7.5 F (Edwards Lifesciences, USA)was introduced via the right external jugular vein andadvanced to the pulmonary artery for monitoring of vitalsigns. A 134 cm2 dispersive plate (EZ 344-02, Berchtold,Germany) was placed over the right hip. A midline incisionwith a transverse right subcostal extension was made. Thehepatoduodenal ligament, containing the hepatic artery, theportal vein and the bile duct, was identified, and a silicone

1991

vessel loop (Sterion, MN) was passed twice around allstructures. To minimise the risk of unintended damage toportal vessels, we did not use surgical diathermy whenexposing the hepatoduodenal ligament. The spleen wasmobilised, and a 5 F 100 mm long introducer (Radifocus,Terumo, Leuven, Belgium) was introduced into the distal2–4 cm of the splenic vein and advanced 5 cm in anantegrade direction for administration of intraportal con-trast medium. A dilute heparin solution 2.5 IU/ml heparin(Hepaflex, Baxter, Lessines, Belgium) was infused throughthe introducer at 60 ml/h to maintain catheter patency.The porcine intrahepatic portal vein shows a consistent

pattern of ramification into four segments with the RPVcoming off at an almost right angle to the portal arch [14].The extrahepatic portal vein with the bifurcation to theRPV was identified in all animals. The ablation electrodewas positioned in the anterior–posterior plane under directvisual guidance based on the point of entry of the RPV intothe hepatic parenchyma. The electrode was inserted 2 cminto the liver parenchyma between the portal arch andwithin 1.5 cm of RPV, perpendicular to the vessel (Fig. 1).The distance between the active tip of the ablationelectrode and RPV was determined by portal venography.The electrode was repositioned if the active tip was notwithin 1.5 cm of the RPV. When the position of theelectrode was found to be acceptable, rotational portalvenography was performed. In animals randomly allocatedto hepatic inflow occlusion, the silicone vessel loop wastightened around the hepatoduodenal ligament 1 min priorto application of RF energy and released immediately afterablation had been completed for a total duration of 10 min.In the sham group hepatic inflow was occluded for 10 minin all animals, without insertion of an ablation electrodeinto the liver. Portal flow was measured in the sham groupby a transmit time flowmeter (VeriQ, Medi-Stim ASA,Norway) using a 12 mm flow probe (Cardiac Output,Medi-Stim ASA).The acute effect of the ablation on the portal veins was

assessed by portal venography 10 min after completion ofthe ablation, while the RF electrode was still positioned inthe hepatic parenchyma. The electrode was removed aftercompletion of the portal venography. To prevent bleedingfrom the needle track, we activated the electrode at 25 Wduring its withdrawal. The introducer in the splenic veinwas removed, and the distal 2–4 cm of the splenic vein wasligated. The Swan–Ganz catheter was removed, and theright external jugular vein was ligated. Postoperatively, theanimals were kept under observation until full recoveryfrom anaesthesia and were allowed free access to food andwater. Postoperative pain was managed with intramuscularinjection of buprenorphine hydrochloride 0.01 mg/kg(Temgesic, Schering-Plough, N.J., USA) twice daily for 3days. The animals underwent repeat laparotomy 4 daysafter ablation, and the delayed effect of the ablation wasevaluated by rotational portal venography as describedpreviously. The animals were subsequently euthanised with

intravenous injection of pentobarbital 500 mg and potas-sium chloride 2–3 mmol/kg. The liver was excisedimmediately after asystole had occurred.

Interpretation of portal venograms

Diameter of the portal arch and RPV was measured at theperpendicular from the tip the ablation electrode to thevessel on native portal venograms by image acquisitionsoftware (Polytron T.O.P., Siemens). The 3D reconstruc-tions of the portal venograms allowed multidirectional viewof the volume data set consisting of the radio-opaqueelectrode and contrast medium-filled portal veins. The dis-tance between the non-insulated 15 mm tip of the ablationelectrode and the portal arch and RPV was determined byrotation of the 3D reconstruction to the projection where thelargest perpendicular distance between the tip of the elec-trode and the corresponding portal vein segment could bemeasured. If areas relevant to the examination were hiddenby other structures, the volume data could be trimmed and avolume of interest generated by post-editing of the volumedata set (Fig. 2). To correct for possible displacement of theelectrode relative to the vessels during ablation, we mea-sured the electrode-to-vessel distance on portal venogramsobtained both prior to and 10 min after ablation. The meanof these measurements was considered to be the electrode-to-vessel distance, and was used in the statistical analysis.Measured vessel diameters and electrode-to-vessel distanceswere rounded off to the nearest millimetre.

Fig. 1 Anteroposterior projection of the porcine extrahepatic andintrahepatic portal vein and the splenic vein. The radiofrequencyablation electrode (a) was positioned 2 cm into the liver parenchymabetween the portal arch (b) and within 1.5 cm of the right portal veinbranch (c). Vessel diameter was measured along a perpendicularfrom the electrode tip to the vessel wall (dashed lines). Contrastagent was distributed to the portal vein (d) by injection through anintroducer placed in the splenic vein (e)

1992

Occurrence of portal vein thrombosis was determined byindependent examination of the native 2-dimensional (2D)portal venograms and the 3D reconstructions by a surgeonand two experienced radiologists unaware of the group towhich the particular animal had been randomly allocated.Agreement between the examiners was reached byconsensus. Owing to the pseudostenosis phenomenonobserved at volume-rendered 3D angiography [15], native2D images were used to confirm abnormal findings on thereconstructed 3D images. The splenic vein, the extrahe-patic portal vein, the portal arch and the RPV wereclassified as normal or abnormal. Abnormal vessels werecategorised as either having a filling defect or as beingoccluded. An occluded portal vessel with lack of peripheralfilling on post-ablation portal venograms was considered tobe a portal vein thrombosis. Acute portal vein thrombosiswas defined as thrombosis on portal venograms acquired10 min after completion of the ablation, whereas delayedportal vein thrombosis was defined as thrombosis on portalvenograms acquired 4 days after ablation.

Histopathological examination

The portal arch was incised longitudinally, and thepresence of intraluminal thrombi and macroscopic endo-thelial damage was noted. The opening of the RPV wasidentified. All coagulations were excised with widemargins and cut in 4–6 mm slices transverse to the RPV,fixed in 4% formaldehyde for at least 10 days, dehydrated,and embedded in paraffin. One 4–5 μm-thick section wascut from each paraffin block and stained with haematoxylinand eosin for microscopic examination by an experiencedpathologist blind to the findings on the portal venograms.The presence of viable hepatocytes between the coagulatedarea and the portal vessel wall, the degree and extent ofdamage to the portal vessel wall itself, and the presence ofthrombotic material adherent to the intimal surface of thevessel was noted.

Statistical analysis

SPSS for Mac OSX version 11.0.4 (SPSS, Chicago, Ill.,USA) was used for statistical analyses. A P value less than0.05 was considered statistically significant. Data are

Fig. 2 Three-dimensionalreconstructions of the porcineextrahepatic and intrahepaticportal vein before and afterradiofrequency ablation. Orien-tation of the volume is indicatedby the square on the lower rightof each image. a After posi-tioning of the radiofrequencyelectrode but before ablation.The electrode (white arrow) ispositioned in proximity of theright portal vein branch.b Reoperation 4 days afterablation. Note lack of contrast inthe right portal vein branch(white arrowhead). c, d Volumedata have been trimmed toelucidate the spatial relationshipbetween the electrode, the rightportal vein branch and the portalarch

1993

presented as mean ± SD unless otherwise indicated. Asample size of six animals in each experimental group wasneeded to demonstrate a significant difference between thetwo groups with proportions of portal thrombosis of 0.05and 0.95 using a two-group continuity corrected χ2 testwith 80% power at the 0.05 two-sided significance level.The animals were allocated to the two experimental groupsby block randomisation, using a block size of two.Occurrence of delayed portal vein thrombosis in the twoexperimental groups was analysed by Fisher’s exact test.Independent samples t-test assuming unequal varianceswas used for comparison of the two experimental groupswith respect to energy delivered during the ablation,diameter of the portal arch, diameter of RPV, distancebetween the electrode and the portal arch and distancebetween the electrode and RPV. Subsequently, the animalswere regrouped on the basis of presence of delayed portalvein thrombosis on post-ablation portal venograms, and thesame variables were compared using independent samplest-tests. The distance between the RPV and the ablationelectrode was recoded into a binary variable with a cut-offvalue of 5 mm. Fisher’s exact test with Bonferronicorrection for multiple tests was used for examination ofthe relationship between Pringle manoeuvre and therecoded electrode-to-vessel distance as explanatory vari-ables, and delayed portal vein thrombosis as outcomevariable.

Results

One ablation was completed in each animal in the twoexperimental groups. Reversible systemic hypotensionwith a reduction of mean arterial pressure of 10 mmHgor more during the ablation was observed in five animals inthe Pringle group and two in the control group. Increase intissue impedance was frequently observed during ablationin the animals in the Pringle group.Portal venograms acquired prior to RF ablation were

considered normal in all animals. A filling defect in thevicinity of the electrode tip was seen 10 min after RFablation in one animal in each experimental group. Neitherof these two animals was considered to have an acute portalvein thrombosis. The remaining ten portal venogramsacquired 10 min after ablation were normal. Delayed portalvein thrombosis was found 4 days after ablation in two ofsix animals in the Pringle group and three of six animals inthe control group (Table 1). Delayed portal vein thrombosiswas present in the two animals with abnormal portalvenograms 10 min after ablation, and in three animals withnormal portal venograms 10 min after ablation. Four of theoccurrences of portal vein thrombosis were located in theRPV (Fig. 2). A fifth portal vein thrombosis was found inone animal in the Pringle group, where the portal arch witha diameter of 13 mm was found to be occluded. Grossexamination of the liver of this animal revealed thermal

damage to the luminal surface at the opening of the RPV,with a thrombus extending into the portal arch. Theendothelium of the portal arch was macroscopicallynormal. Diameters of the occluded RPV branches rangedfrom 4–9 mm (5.6±1.9 mm). The splenic vein and theextrahepatic portal vein were normal in all animals in thetwo experimental groups.No occurrences of thrombosis or filling defects were

found in the splenic veins, extrahepatic portal veins orintrahepatic portal veins in the sham group 10 min afterPringle or at follow-up 4 days later. Portal flow measure-ments in the sham-group animals during Pringle ma-noeuvre confirmed cessation of portal flow. Ligation of thedistal 2–4 cm of the splenic vein after removal of theintroducer used for administration of intraportal contrastagent did not significantly decrease portal flow comparedto baseline values (884±60 ml/min vs 934±146 ml/min,P=0.455).

Histopathology

A sharp demarcation between the coagulated area andnormal hepatic parenchyma was found at gross examina-tion, corresponding to the border between viable and non-viable tissue found at histopathological examination(Fig. 3a,b). The hepatic microstructure was preserved inparts of the coagulated area. The cytoplasm of coagulatedhepatocytes was more eosinophilic and showed pyknoticnuclei compared to viable hepatocytes. In other coagulatedareas, the hepatocytes had undergone karyolysis and weredenucleated. Fibroblast proliferation with a variable degreeof inflammation was prominent between the coagulationand the portal vein in all animals. The portal vein could notbe identified in one animal in the Pringle group withdelayed portal vein thrombosis due to extensive tissuenecrosis. In the remaining four animals with delayed portalvein thrombosis, thrombosis was confirmed by microsco-py. In addition, subtotal thrombotic occlusion of the RPVwas found in one animal in the Pringle group. Thecorresponding portal venogram was interpreted as normal,and this animal was therefore not considered to have aportal vein thrombosis in the statistical analysis.

Table 1 Acute and delayed portal vein thrombosis in the twoexperimental groups

Parameter Pringlegroup(n=6)

Controlgroup(n=6)

P*

Acute portal vein thrombosisa 0/6 0/6 –Delayed portal vein thrombosisb 2/6 3/6 1.0

*Fisher’s exact testaTen minutes after ablationbFour days after ablation

1994

Viable hepatocytes were present between the vessel walland the coagulated area in one of five animals in the Pringlegroup and in four of six in the control group. Damage to theendothelial lining was occasionally present without thermaldamage to the adjacent hepatocytes. Necrosis of all layersof the portal vein wall covering 20–100% of the circum-ference was present in all five pigs in the Pringle group inwhom the portal vein could be assessed. In the controlgroup, necrosis of the vessel wall was seen in four of sixanimals, covering 20–80% of the circumference. In oneanimal in the control group with portal vein thrombosis,extensive inflammation of the portal vein was found. In theremaining three animals where the portal vein could beassessed, necrosis of all layers of the portal vein waspresent.

Factors associated with portal vein thrombosis

Mean energy delivered during ablation was lower in thePringle group than in the control group (22,560±2,520 J vs25,670±360 J, P=0.046). No significant differencesregarding preoperative vessel diameter or electrode-to-vessel distances were found in the two experimental groups(Table 2). Hepatic vascular occlusion was not associatedwith occurrence of portal vein thrombosis (P=1.0, Fisher’sexact test). The largest diameter of a thrombosed RPV withPringle manoeuvre was 9 mm, whereas the correspondingvessel diameter in the control group was 5 mm. The longestdistance between the ablation electrode and a thrombosed

Table 2 Deployed energy, preoperative vessel diametersand electrode-to-vessel distances in the two experimental groups

Parameter Pringlegroup(n=6)

Controlgroupocclusion(n=6)

Pa

Mean ± SD Mean ± SD

Energy 22,560 ± 2,520 J 25,270 ± 361 J 0.046Diameter of portalarch

11.3 ± 1.5 mm 11.2 ± 1.6 mm 0.856

Diameter of rightportal vein branch

5.8 ± 1.6 mm 5.7 ± 1.8 mm 0.867

Distance of electrodeto portal arch

10.0 ± 4.2 mm 17.2 ± 7.4 mm 0.074

Distance of electrodeto right portalvein branch

6.0 ± 2.2 mm 6.8 ± 5.6 mm 0.745

aIndependent samples t-test with unequal variances assumed

Fig. 3 Haematoxylin and eosin stained sections of hepatic radio-frequency ablation. A sharp demarcation between the coagulatedareas (RF) and normal hepatic parenchyma is present (arrowheads).Note thrombus in portal veins in both specimens (arrows) a Pringlemanoeuvre during the ablation. The coagulation merges with theadjacent vessel. Necrosis of the vessel wall was present in 20–30%

of the circumference. b No hepatic vascular occlusion duringablation. The coagulated area has fragmented. Note that the contourof the coagulation has been altered by the adjacent portal vein,illustrating the heat sink effect. Extensive inflammation of the vesselwall was present

Fig. 4 Relationship between use of Pringle manoeuvre, appliedelectrical energy, distance between the ablation electrode and rightportal vein branch and delayed portal vein thrombosis in the twoexperimental groups (n=12). Occurrence of delayed portal veinthrombosis is indicated by a filled symbol. PVT delayed portal veinthrombosis, NOR no portal vein thrombosis, P+ Pringle manoeuvre,P− no hepatic vascular occlusion

1995

RPV with Pringle manoeuvre was 5 mm. The correspond-ing distance in the control group was 4 mm. The shortestdistance between the electrode and the RPV withoutoccurrence of portal vein thrombosis was 3 mm and wasfound in an animal with a RPV diameter of 5 mm in thePringle group. This animal received lowest cumulatedenergy of all animals (Fig. 4).The mean distance between the ablation electrode and

the RPV was significantly shorter in the five animals thatdeveloped delayed portal thrombosis (3.6±1.5 mm) than inthe seven animals with normal portal venograms (8.4±4.2 mm), with a P value of 0.023. No other significantdifferences were found in these two groups (Table 3).Delayed portal vein thrombosis occurred in five of sixanimals, with a distance between the electrode and the RPVof 5 mm or less, while no occurrences of portal veinthrombosis were found in the six animals where thisdistance was 6 mm or more (Fig. 4). A distance betweenthe electrode and the RPVof 5 mm or less was a significantpredictor for delayed portal vein thrombosis (P=0.030,Fisher’s exact test).

Discussion

Although a non-cirrhotic liver may tolerate a resection ofup to 80%, reduction of liver parenchyma by even 50%may be associated with postoperative hepatic insufficiency[1]. The hepatic artery normally supplies 25% of the totalblood flow to the liver but may provide up to 30–50% ofthe normal oxygen requirement of the hepatic parenchyma,because of the higher oxygen content in the arterial bloodthan in the partly deoxygenated portal blood. Increasedoxygen extraction from the hepatic arterial blood and

increased arterial hepatic blood flow can minimise theeffect of a significant reduction in portal flow [16]. Owingto these compensatory mechanisms the clinical implicationof segmental portal vein thrombosis after RF ablation couldbe questioned. Although segmental portal vein thrombosismay be asymptomatic [13], transient liver failure has beenreported after portal vein embolisation in patients withreduced hepatic reserve [17]. Liver failure caused by portalvein thrombosis was the most frequent cause of death in arecent study of 312 patients with hepatic tumours treated byradiofrequency ablation [18]. Patients with a deficit infunctional hepatic reserve may be at risk of postoperativehepatic failure after portal vein thrombosis. Of specialclinical concern are patients in whom an extensive hepaticresection has been performed and concomitant RF ablationin the remnant liver is planned [18]. Determination of riskfactors associated with occurrence of segmental portal veinthrombosis related to RF ablation may, therefore, be ofvalue for selecting patients suitable for RF ablation.Use of the Pringle manoeuvre was not associated with

increased risk of delayed portal vein thrombosis in thisexperimental randomised study. Histopathological exami-nation confirmed necrosis of the portal vein wall in all pigsin the Pringle group where the integrity of the portal veinwall could be assessed. Viable hepatocytes between thevessel wall and the coagulated area were present in onlyone of five animals in the Pringle group, compared withfour of six in the control group. These findings suggest thathepatic vascular occlusion during RF ablation enhancescellular devitalisation close to portal vessels and confirmthe increased risk of thermal damage to vessels duringhepatic inflow occlusion found in previous experimentalstudies [19, 20]. Nevertheless, this did not translate into ahigher incidence of acute or delayed portal vein thrombosisin this study. Even though the distribution of all variablesunder consideration, except energy, was similar in the twoexperimental groups, as expected from randomisation(Table 2), these figures are derived from a small data setof five occurrences of delayed portal vein thrombosis in 12animals and should be interpreted with caution. Hence, ourfinding that the Pringle manoeuvre was not associated withacute or delayed portal vein thrombosis may be explainedby the limited sample size.Less energy was delivered during ablation in the Pringle

group than in the control group. This finding can beexplained by the fact that the impedance-controlledradiofrequency system employed in this study reducedpower output if tissue impedance increased during abla-tion. The reduced or absent hepatic blood flow caused bythe Pringle manoeuvre can be expected to limit theperfusion-mediated cooling of the electrode-to-tissue in-terface, leading to rapid desiccation and subsequentimpedance rise. As the ablation protocol used in thisstudy consisted of a predefined time interval of 9 min and afixed power setting of 50 W, reduced power output resultedin lower accumulated energy output. The importance of the

Table 3 Deployed energy, preoperative vessel diameters andelectrode-to-vessel distances, using delayed portal vein thrombosisas the categorising variable

Parameter Delayed portalvein thrombosis(n=5)

No delayed portalvein thrombosis(n=7)

Pa

Mean ± SD Mean ± SD

Energy 24,313 ± 1,384 J 23,630 ± 2751 J 0.586Diameter of portalarch

11.6 ± 1.8 mm 11.0 ± 1.3 mm 0.547

Diameter of rightportal vein branch

5.6 ± 1.9 mm 5.9 ± 1.5 mm 0.810

Distance of electrodeto portal arch

10.2 ± 6.7 mm 16.0 ± 6.3 mm 0.167

Distance of electrodeto right portal veinbranch

3.6 ± 1.5 mm 8.4 ± 4.2 mm 0.023

aIndependent samples t-test with unequal variances assumed

1996

lower mean energy deposited in the Pringle group isunclear, as no statistical significant difference was foundwhen we compared deployed cumulated energy in theanimals with and without delayed portal vein thrombosis(Table 3).A recent study by Ng et. al. reported that use of the

Pringle manoeuvre during RF ablation invariably resultedin delayed portal vein thrombosis, whereas no such riskwas present for ablation without hepatic vascular occlusion[20]. In our study, delayed portal vein thrombosis occurredonly in two of six animals in the Pringle group. Perhaps ofmore importance was that three of six animals in the controlgroup developed delayed portal vein thrombosis. Theseapparently diverging results may have several explana-tions. In the study by Ng et. al., the distance between thevessel and the electrode was stated to be 5 mm in allanimals. Our results suggest that the distance betweenelectrode and vessel may be of significant importance fordevelopment of delayed portal thrombosis. The RF abla-tion systems used in the two studies were different. In thestudy by Ng et. al., an internally cooled electrode with anon-insulated tip of 30 mm was used, whereas we used awet ablation electrode design with a non-insulated tip of15 mm. The generator and generator setting used by Ng et.al. was not specified, but the electrodes used in the studyare usually connected to a generator that can deliver up to200 W, compared to the 50 W used in our study. Theduration of ablation (12 min vs 9 min) also varied. In thestudy by Ng et. al. the presence of delayed portal veinthrombosis was investigated 1 week after ablation. Histo-pathological examination revealed complete necrosis of theportal vein wall in the majority of the animals 4 days afterthe ablation in our study. We cannot exclude that theproportion of portal vein thrombosis could have beenhigher if our animals had been examined at 7 days afterablation compared to 4 days after ablation. Other dissimilaraspects are the methods used for assessment of portal veinthrombosis. The study by Ng et al. measured portal flow bymeans of intraoperative Doppler ultrasonography, whilerepeat portal venography was used for determination ofportal vein thrombosis in our study. The use of portalvenography was based on the assumption that venographywould facilitate objective detection of segmental portalvein thrombosis that might go unnoticed when alterationsof flow velocity at the portal trunk were measured.The approach used in our study involved exposing portal

endothelium injured by ablation to a non-ionic contrastagent and ligation of the distal 2–4 cm of the splenic vein.Both the use of intraportal contrast medium and ligation ofthe distal splenic vein could be potential influential factorsfor portal vein thrombosis, especially in the group withoutPringle manoeuvre. Even though the non-ionic contrastagent used in this study is associated with a low incidenceof thrombus-related events [21], the safety profile of the

agent has not been evaluated for porcine portal endothe-lium damaged by thermal ablation. We cannot, therefore,completely rule out the possibility that use of this contrastagent in the presence of portal endothelium damaged bythermal ablation could lead to formation of thrombi thatmay have influenced the results. Significantly reducedportal vein flow caused by decreased inflow from thesplenic vein could be a contributing factor for intrahepaticportal thrombosis. However, flow measurements in thesham group showed that ligation of the distal splenic veinwas associated with a statistically not significant reductionof portal flow of 5%. It is, therefore, not likely that reducedportal flow due to distal splenic vein ligation was aconfounding factor for the development of portal throm-bosis. The distal splenic vein is a possible source of embolito intrahepatic portal veins. However, filling defects werenot observed in the splenic veins or extrahepatic portalveins on any portal venogram. Furthermore, thrombosisdeveloped in the RPV, which comes off at almost a rightangle to the portal arch and is not a probable site ofthrombus formation caused by distant emboli.The absence of acute portal vein thrombosis following

RF ablation in proximity to portal veins has beenestablished by previous studies [20, 22]. Only two of thefive animals with delayed portal vein thrombosis 4 daysafter ablation had abnormal findings on the portalvenograms 10 min after ablation. Hence, a normal portalvenography obtained 10 min after ablation did not excludelater occurrence of thrombosis, whereas an abnormal portalvenography 10 min after ablation was associated withdelayed portal vein thrombosis. Thrombus formationrelated to RF ablation is believed to be initiated byendothelial cell injury with interruption of endothelialcontinuity, leading to platelet adhesion and activation aswell as thrombin production [23]. In the present study,histopathology confirmed that portal vein thrombosis wasinvariably accompanied by transmural necrosis of theportal wall or extensive inflammation of the portalendothelium. This suggests that acute endothelial damagecaused by the ablation represents a risk for subsequentthrombosis of the vessel.Our finding of an electrode-to-vessel distance of 5 mm

or less as a significant predictor of portal thrombosis isbiologically plausible. The relationship between factorsthat influence tissue temperature after deposition of energyis described by the bioheat equation [24]. According tothis, an increase in tissue temperature is dependent on theenergy deposited in the tissue, modified by local tissueinteractions, minus the heat loss [25]. In accordance withthe bioheat equation, our findings suggest that the criticaldistance from the electrode at which portal vessels are atrisk of thrombosis may be increased if perfusion-mediatedheat loss is minimised due to hepatic vascular occlusion.The tissue power density Wv (watts/cubic metres) is an

1997

expression of the electrical energy per time unit that tissueis exposed to under ideal conditions:

Wv ¼ δV 2 a2

r4

where δ is the tissue conductivity, V the voltage, a theradius of the electrode, and r the distance from theelectrode. Thus, the power density can be expected to falloff extremely rapidly with increasing distance from theelectrode, as it is inversely proportional to the distancefrom the electrode to the fourth power [26]. The RFablation system used in this study infuses saline solution tothe tissue through the active electrode during ablation.Although the effect of saline infusion has not beencompletely elucidated, the conductive properties of salinesolution are believed to increase the effective radius of theelectrode [27] and thereby increase power density at agiven distance from the electrode. The magnitude of thiseffect has not been quantified for the RF ablation systemused in this study.From an electrical point of view, the liver is inhomoge-

neous, with large variations in tissue conductivity. Conse-quently, the flow pattern of the electrical current around theelectrode is determined by the tissue morphology andcomposition. The temperature distribution is also related tonon-electrical tissue factors, such as the amount of heat lossby conduction. The complexity of the association betweenthese factors and the resulting thermal tissue damage isdemonstrated by the fact that size and shape of thecoagulated volume cannot be predicted, even understandardised ablation protocols [28]. Although the relation-ship between these factors and the resulting tissue temper-ature is not a trivial one, it seems reasonable to assume that,when the electrode-to-vessel distance is below a criticallimit, the vessel wall will be exposed to temperatures highenough to cause endothelial damage. Factors related to thespecific RF ablation equipment used, such as the generatoroutput, size and shape of the active electrode and thepresence of internal or external cooling of the electrode,influence the energy efficacy [11] and the extent of thecoagulation [29]. The RF ablation system used in this studyhas been shown to produce irregular coagulations whencompared with other RF ablation systems [29, 30].Therefore, our finding of an electrode-to-vessel distanceof 5 mm or less as being associated with delayed portalvein thrombosis is only valid for the specific RF ablationsystem used in this study.Several limitations of this study must be addressed. First,

findings after experimental RF ablation in normal porcine

liver are not necessarily applicable to clinical settings. Thepig has a coagulation pathway similar to that of humans,but it is considered to be more susceptible to activation ofthe coagulation system following endothelial damage [31].Hence, thermal ablation and endothelial damage causingthrombosis in pigs would not necessarily lead to portalthrombosis in humans. Secondly, for the prevention ofclotting in the introducer in the splenic vein, heparin at150 IU/h was administered through the introducer betweenthe contrast examinations during the 1–2 h procedure. Theaccumulated heparin dose used in our study was less than5% of the 100–150 IU/kg per hour considered a normaldose for surgical procedures in the pig [31]. We consider itunlikely that administration of heparin at 150 IU/h into thesplenic vein would have influenced the occurrence ofportal vein thrombosis or its natural course. Reversiblesystemic hypotension with a reduction of mean arterialpressure of 10 mmHg or more during ablation wasobserved in both experimental groups during ablation.Consequently, reduced hepatic blood flow may haveoccurred even in the control group. This study wasdesigned to detect thrombosis in the portal veins. Con-current thrombosis of segmental hepatic arteries would beexpected to increase the loss of hepatic function but wasnot assessed in this study. Finally, the experiments werecarried out in an animal model with an adequate hepaticreserve, which is unsuitable for demonstration of alteredhepatic function after segmental portal vein thrombosis.In conclusion, delayed portal vein thrombosis associated

with hepatic RF ablation close to portal veins occurred inboth the Pringle and the control group. Use of the Pringlemanoeuvre during RF ablation was not associated with ahigher risk of delayed portal thrombosis. On the otherhand, our findings indicate that the Pringle manoeuvrerendered portal veins more susceptible to thermal damage.A distance between the ablation electrode and the RPV of5 mm or less was significantly associated with develop-ment of delayed portal vein thrombosis after RF ablationwith wet electrodes. Further research is warranted toexamine the role of the electrode-to-vessel distance in thedevelopment of acute or delayed portal vein thrombosis forother hepatic RF ablation systems, and to assess alterationsof hepatic function associated with segmental portal veinthrombosis.

Acknowledgements We thank the staff at the Interventional Centrefor providing excellent assistance, and the Department of Compara-tive Medicine for support in animal care. Marijke Veenstra and MarteOlstad at the Institute of Biostatistics provided statistical assistance.This work was supported by the Norwegian Cancer Society, grant no.D02042/001.

1998

References

1. Blumgart LH (2000) Liver resection forbenign disease and for liver and biliarytumors. In: Blumgart LH, Fong Y (eds)Surgery of the liver and biliary tract,3rd edn. Saunders, London,pp 1639–1713

2. Rossi S, Fornari F, Buscarini L (1993)Percutaneous ultrasound-guidedradiofrequency electrocautery for thetreatment of small hepatocellular car-cinoma. J Intervent Radiol 8:97–103

3. Curley SA, Izzo F, Delrio P, Ellis LM,Granchi J, Vallone P, et al (1999)Radiofrequency ablation of unresect-able primary and metastatic hepaticmalignancies: results in 123 patients.Ann Surg 230:1–8

4. Bleicher RJ, Allegra DP, Nora DT,Wood TF, Foshag LJ, Bilchik AJ(2003) Radiofrequency ablation in 447complex unresectable liver tumors:lessons learned. Ann Surg Oncol10:52–58

5. Abdalla EK, Vauthey JN, Ellis LM, EllisV, Pollock R, Broglio KR, et al (2004)Recurrence and outcomes following he-patic resection, radiofrequency ablation,and combined resection/ablation for co-lorectal liver metastases. Ann Surg239:818–825

6. Lu DS, Raman SS, Limanond P,Aziz D, Economou J, Busuttil R, et al(2003) Influence of large peritumoralvessels on outcome of radiofrequencyablation of liver tumors. J Vasc IntervRadiol 14:1267–1274

7. Elias D, Baton O, Sideris L, MatsuhisaT, Pocard M, Lasser P (2004) Localrecurrences after intraoperative radio-frequency ablation of liver metastases:a comparative study with anatomic andwedge resections. Ann Surg Oncol11:500–505

8. Mulier S, Ni Y, Jamart J, Ruers T,Marchal G, Michel L (2005) Localrecurrence after hepatic radiofrequencycoagulation: multivariate meta-analysisand review of contributing factors. AnnSurg 242:158–171

9. Goldberg SN, Hahn PF, Tanabe KK,Mueller PR, Schima W, AthanasoulisCA, et al (1998) Percutaneousradiofrequency tissue ablation: doesperfusion-mediated tissue cooling limitcoagulation necrosis? J Vasc IntervRadiol 9:101–111

10. Lu DS, Raman SS, Vodopich DJ, WangM, Sayre J, Lassman C (2002) Effect ofvessel size on creation of hepaticradiofrequency lesions in pigs:assessment of the heat sink effect.Am J Roentgenol 78:47–51

11. Brieger J, Pereira PL, Trubenbach J,Schenk M, Krober SM, Schmidt D,et al (2003) In vivo efficiency of fourcommercial monopolar radiofrequencyablation systems: a comparativeexperimental study in pig liver. InvestRadiol 38:609–616

12. Chinn SB, Lee FT Jr, Kennedy GD,Chinn C, Johnson CD, Winter TC 3rd,et al (2001) Effect of vascular occlusionon radiofrequency ablation of the liver:results in a porcine model. Am JRoentgenol 176:789–795

13. Mulier S, Mulier P, Ni Y, Miao Y,Dupas B, Marchal G, et al (2002)Complications of radiofrequencycoagulation of liver tumours. Br J Surg89:1206–1222

14. Dondelinger RF, Ghysels MP, BrisboisD, Donkers E, Snaps FR, Saunders J,et al (1998) Relevant radiologicalanatomy of the pig as a training modelin interventional radiology. Eur Radiol8:1254–1273

15. Hirai T, Korogi Y, Ono K, Yamura M,Uemura S, Yamashita Y (2004)Pseudostenosis phenomenon atvolume-rendered three-dimensionaldigital angiography of intracranialarteries: frequency, location, and effecton image evaluation. Radiology232:882–887

16. Mathie RT, Wheatley AM (2000) Liverblood flow: physiology, measurementand clinical relevance. In: BlumgartLH, Fong Y (eds) Surgery of the liverand biliary tract, 3rd edn. Saunders,London, pp 85–107

17. Di Stefano DR, de Baere T, Denys A,Hakime A, Gorin G, Gillet, M et al(2005) Preoperative percutaneousportal vein embolization: evaluation ofadverse events in 188 patients.Radiology 234:625–630

18. de Baere T, Risse O, Kuoch V, DromainC, Sengel C, Smayra T, et al (2003)Adverse events during radiofrequencytreatment of 582 hepatic tumors. Am JRoentgenol 181:695–700

19. Denys AL, De Baere T, Mahe C,Sabourin JC, Sa Cunha A, Germain S,et al (2001) Radio-frequency tissueablation of the liver: effects of vascularocclusion on lesion diameter and biliaryand portal damages in a pig model. EurRadiol 11:2102–2108

20. Ng KK, Lam CM, Poon RT, Shek TW,Fan ST, Wong J (2004) Delayedportal vein thrombosis after experi-mental radiofrequency ablation nearthe main portal vein. Br J Surg91:632–639

21. Davidson CJ, Laskey WK, HermillerJB, Harrison JK, Matthai W Jr,Vlietstra RE, et al (2000) Randomizedtrial of contrast media utilization inhigh-risk PTCA: the COURT trial.Circulation 101:2172–2177

22. Shen P, Fleming S, Westcott C, ChallaV (2003) Laparoscopic radiofrequencyablation of the liver in proximity tomajor vasculature: effect of the Pringlemaneuver. J Surg Oncol 83:36–41

23. Khairy P, Chauvet P, Lehmann J,Lambert J, Macle L, Tanguay JF, et al(2003) Lower incidence of thrombusformation with cryoenergy versusradiofrequency catheter ablation.Circulation 107:2045–2050

24. Pennes HH (1948) Analysis of tissueand arterial blood temperatures in theresting human forearm. J Appl Physiol1:93–122

25. Goldberg SN, Gazelle GS, Mueller PR(2000) Thermal ablation therapy forfocal malignancy: a unified approach tounderlying principles, techniques, anddiagnostic imaging guidance. Am JRoentgenol 174:323–331

26. Grimnes S, Martinsen OG (2000)Geometrical analysis. In: Grimnes S,Martinsen OG (eds) Bioimpedance andbioelectricity basics. Academic Press,San Diego, California, pp 127–152

27. Livraghi T, Goldberg SN, Monti F,Bizzini A, Lazzaroni S, Meloni F, et al(1997) Saline-enhanced radio-frequen-cy tissue ablation in the treatment ofliver metastases. Radiology202:205–210

28. Montgomery RS, Rahal A, Dodd GD3rd, Leyendecker JR, Hubbard LG(2004) Radiofrequency ablation ofhepatic tumors: variability of lesionsize using a single ablation device. AmJ Roentgenol 182:657–661

29. Pereira PL, Trubenbach J, Schenk M,Subke J, Kroeber S, Schaefer I, et al(2004) Radiofrequency ablation: invivo comparison of four commerciallyavailable devices in pig livers.Radiology 232:482–490

30. Denys AL, Baere TD, Kuoch V, DupasB, Chevallier P, Madoff DC, et al(2003) Radio-frequency tissue ablationof the liver: in vivo and ex vivoexperiments with four differentsystems. Eur Radiol 13:2346–2352

31. Swindle M (1998) Surgery, anesthesia,and experimental techniques in swine.Iowa State University Press, Ames

1999

IV

Increased Activity of Matrix Metalloproteinase 2 and 9 After HepaticRadiofrequency Ablation

Lars Frich, M.D.,*,†,1 Kristin Bjørnland, M.D., Ph.D.,†,‡,¶ Solveig Pettersen, B.S.,‡Ole Petter F. Clausen, M.D., Ph.D.,§ and Ivar P. Gladhaug, M.D., Ph.D.†,¶

*The Interventional Centre, †Department of Surgery, ‡Institute for Surgical Research, §Department and Institute of Pathology,Rikshospitalet University Hospital; and ¶Department of Surgery, Faculty Division Rikshospitalet, University of Oslo, Oslo, Norway

Submitted for publication December 22, 2005

Background. Radiofrequency (RF) ablation of he-patic metastases from colorectal cancer (CRC) is asso-ciated with a high rate of local and intrahepatic tumorrecurrence. Matrix metalloproteinases (MMPs) playan important role in inflammation, tissue repair andtumor cell invasion andmetastasis. MMP-2 andMMP-9are associated with increased risk of recurrence anddecreased survival in patients with colorectal cancer.The primary aim of the study was to determine if he-patic RF ablation increased MMP-2 and MMP-9 activ-ity in the transition zone surrounding the coagulatedhepatic tissue.

Materials and methods. Twelve pigs were random-ized to hepatic RF ablation with (n 6) or without(n 6) hepatic vascular occlusion (Pringle maneuver).Four days after ablation tissue specimens were col-lected from the transition zone surrounding coagu-lated hepatic tissue, and from normal hepatic pa-renchyma. MMP activity was quantified by gelatinzymography. Cellular localization of MMPs was de-termined by immunohistochemistry using antibodiesagainst MMP-2, MMP-9, and the macrophage markerCD68.

Results.MMP-2 andMMP-9 activity was increased inthe transition zone compared to normal hepatic pa-renchyma, with ratios of 3.0 (P 0.005) and 2.6 (P 0.001), respectively. Pringle maneuver did not influ-ence MMP activity. MMP-2 and MMP-9 expression waslocalized to macrophages in the transition zone.

Conclusions. Hepatic RF ablation is associated withincreased expression of MMP-2 and MMP-9 in macro-phages in the transition zone surrounding the coagu-lated hepatic parenchyma. These findings may contrib-

ute to the understanding of possible mechanisms for thehigh recurrence rates observed in patients after RF ab-lation of CRC hepatic metastases. © 2006 Elsevier Inc. All rights

reserved.

Key Words:matrixmetalloproteinases; radiofrequencyablation; liver; neoplasm; animal experimentation.

INTRODUCTION

Hepatic resection is the gold standard for treatmentof liver metastases from colorectal cancer (CRC), with5-year survival rates of approximately 30% in se-lected patients [1, 2]. The majority of patients withhepatic metastases from CRC are not candidates forpotential curative hepatic resection because of bilobartumor location, underlying hepatic disease or extrahe-patic tumor growth. Radiofrequency (RF) ablation isincreasingly used for in situ tumor destruction in pa-tients with non-resectable metastases limited to theliver [3]. Focal tissue necrosis is achieved by applyingalternating electrical current within a frequency rangeof 400 to 500 kHz to an active electrode positioned inthe target tumor. At temperatures above 55°C, proteindenaturation and cell death occur. Heating of tissue isassociated with increased tissue impedance and char-ring which limit the coagulation necrosis to 3 to 5 cmdiameter [4]. Furthermore, in vivo the extent of tissuecoagulation is restricted by the heat-sink effect, wherethe region to be coagulated is cooled by blood flow inlarge vessels [5, 6]. Increased coagulation volume canbe achieved by reduction of hepatic perfusion duringablation by temporary mechanical occlusion of the por-tal vein and the hepatic artery (Pringle maneuver)[7–9].Patients with CRC liver metastases treated with

hepatic RF ablation have higher local, intrahepatic,

1 To whom correspondence and reprint requests should be ad-dressed at Department of Surgery, Rikshospitalet University Hospi-tal, N-0027 Oslo, Norway. E-mail: lars.frich@labmed.uio.no.

Journal of Surgical Research 135, 297–304 (2006)doi:10.1016/j.jss.2006.05.010

297 0022-4804/06 $32.00© 2006 Elsevier Inc. All rights reserved.

and distant recurrence rates, as well as lower overallsurvival than comparable resected patients [10]. Aplausible explanation for these findings is negativepatient selection bias, with micrometastatic diseasebeing present in a larger proportion of patients treatedwith RF ablation [11]. An alternative explanation thatwarrants further investigation is that thermal ablationof malignant tumors may induce biological alterationsin the tumor tissue or peritumoral hepatic parenchymathat facilitate spread and invasion of malignant cells[12, 13].Matrix metalloproteinases (MMPs) are a family of

zinc-dependent matrix-degrading endopeptidases in-volved in a variety of physiological and pathologicalevents [14–16]. MMPs are associated with inflamma-tion and tissue repair and play an important role in thenormal physiological turnover of the extracellular ma-trix (ECM). Increased expression of MMPs is necessaryfor tumor cell invasion, metastasis and angiogenesis.The gelatinases MMP-2 and MMP-9 degrade type IVcollagen, which is a major component of the basementmembrane. Disruption of the basement membrane al-lows tumors to spread locally and distally. MMP-2 andMMP-9 contribute to CRC progression in experimentalmodels [14], are overexpressed in patients with CRChepatic metastasis [17–19], and are associated withincreased risk of tumor recurrence and decreased sur-vival in patients with CRC [16]. Increased MMP activ-ity may provide a physiological mechanism of tissueremodeling that facilitates growth and spread of ma-lignant cells.Our study hypothesis was that hepatic RF ablation

increases expression of MMPs and cytokines in thetransition zone separating coagulated tissue from nor-mal hepatic parenchyma. The primary aim of the studywas to determine activity of MMP-2 and MMP-9, andexpression of the cytokines tumor necrosis factor-(TNF-), interleukin-10 (IL-10) and prostaglandin E2

(PGE2) in hepatic tissue lysates from the transitionzone. Additionally, we wanted to examine if use of thePringle maneuver during RF ablation was associatedwith alterations in MMP activity or cytokine levels.Tissue was harvested 4 days after ablation to allowexamination of MMP activity and cytokine levels dur-ing an established inflammatory process with a welldemarcated transition zone.

MATERIALS AND METHODS

Experimental Protocol

The protocol was approved by the institutional Department ofComparative Medicine. Twelve domestic pigs with a mean weight of30.3 kg (SD 3.4 kg) were randomized to two experimental groups.In the first group RF ablation was performed without occlusion ofhepatic vessels (n 6). In the second group Pringle maneuver wasperformed during RF ablation (n 6).Premedication consisted of intramuscular injection of ketamine

15 to 20 mg/kg (Ketalar; Pfizer, Cambridge, MA), atropine sulfate 1mg (Atropin; Nycomed Pharma, Rockslide, Sweden) and azaperone 2to 4 mg/kg (Stresnil, Janssen-Cilag). Ampicillin 250 mg was admin-istered intramuscularly before and after the surgical procedure. Sur-gery was performed during general anesthesia induced by intrave-nous pentobarbital 150 to 400 mg and morphine 10 to 30 mg andmaintained by inhaled isoflurane 1 to 1.2% in oxygen. The analgesiceffect was supplemented with morphine infusion at 10 mg/h. A 7.5 Fpulmonary artery catheter (CCOmbo 774HF75; Edwards Life-sciences, Irvine, CA) was introduced via the right external jugularvein and advanced to the pulmonary artery for monitoring of vitalsigns. A 134 cm2 dispersive plate (EZ 344-02; Berchtold GmbH& Co.,Tuttlingen, Germany) was placed over the right hip. Laparatomywas performed and a silicone vessel loop (Sterion, Minneapolis, MN)was passed twice loosely around the hepatoduodenal ligament con-taining the hepatic artery, the portal vein and the bile duct.An impedance-controlled RF ablation system (Elektrotom 106

HiTT, Berchtold GmbH & Co., Germany) was used for ablation. Thesystem incorporates a generator that delivers alternating current at375 kHz to a maximum of 1.2 ampere. The active electrode is 1.7 mmin diameter, with a 15 cm long shaft incorporating a lumen for salineperfusion, and a terminal non-insulated segment of 15 mm. A sy-ringe pump (Pilot C; Fresenius Vial, Grenoble, France) was used forinfusion of saline through the active electrode. The saline was dis-tributed to tissue in the vicinity of the electrode tip through threepairs of two side holes. The electrode was positioned 2 cm into theliver parenchyma under visual guidance. RF energy was applied for1 min with an effect of 30 Watts, immediately followed by 8 min with50 Watts. In animals randomized to the Pringle group, the siliconevessel loop was tightened around the hepatoduodenal ligament 1min before application of RF energy and released immediately afterablation had been completed. After ablation had been completed thepulmonary artery catheter was removed, and the right externaljugular vein was ligated. Postoperatively, the animals were keptunder surveillance until full recovery from anesthesia and wereallowed free access to food and water. Pain was managed withintramuscular buprenorphine hydrochloride 0.01 mg/kg (Temgesic;Schering-Plough, Berkeley Heights, NJ) twice daily for 3 days. Fourdays after ablation, the animals were sacrificed by intravenous in-jection of pentobarbital 500 mg and potassium chloride 2 to 3 mmol/kg, and the livers were removed.

Blood Samples

Systemic MMP activity and cytokine expression was determinedin blood samples collected at three time points as follows: baseline(i.e., 1 h after laparotomy, but before RF ablation), 1 h after RFablation and 4 days after RF ablation. Blood samples were drawninto 5 mL citrate tubes, stored on ice during the operative procedure,and centrifuged at 1000 g for 10 min. The plasma fraction wasseparated and stored at 70°C until analyzed. Protein content ofeach sample was measured with a Bio-Rad BCA assay.

Tissue Samples

Immediately after removal of the liver, fresh tissue cubes of 4 4 4 mm were excised from the transition zone, incorporating theedge of the coagulated area and the surrounding hepatic paren-chyma. Corresponding samples were excised from normal hepaticparenchyma at least 5 cm from macroscopic signs of tissue coagula-tion. The samples were immediately frozen in liquid nitrogen. Pro-teins were extracted using 10 L lysis buffer per 1 mg tissue andstored at 70°C until analyzed. Protein content was measured witha Bio-Rad BCA assay. As RF ablation denaturizes protein, measuredprotein content of the tissue lysates would not necessarily reflect theprotein content before ablation. The amount of tissue lysates used inthe gelatin zymography was therefore determined by dry weight ofeach tissue sample before homogenization.

298 JOURNAL OF SURGICAL RESEARCH: VOL. 135, NO. 2, OCTOBER 2006

Gelatin Zymography

MMP-2 and MMP-9 enzyme activity was analyzed using gelatinzymography. Equal amounts of plasma or liver tissue lysates wereapplied to a non-reducing 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel containing 0.1% gelatin (Bloom 300; Sigma-Aldrich, St. Louis, MO), and run as previously described with minormodifications [20]. The gelatin zymograms were calibrated with hu-man MMP-2 and MMP-9 standard (CC073; Chemicon, Temecula,CA) and activated with 0.2% Brij 35 (Sigma-Aldrich). Gels werestained with Coomassie brilliant blue R-250 (Bio-Rad, Hercules, CA)and destained before being scanned. MMP enzyme activity was de-tected as clear bands against a staining background of undegradedsubstrate. The gels were photographed under standardized lightningconditions, and clear bands representing gelatinase activity wasquantified by densitometry using TotalLab software (Nonlinear Dy-namics, Newcastle upon Tyne, UK). Latent and active forms ofMMP-2 and MMP-9 were pooled in the statistical analysis.

Enzyme-Linked Immunosorbent Assays

Cytokine expression in plasma and tissue lysates was determinedby commercially available enzyme-linked immunosorbent assays(ELISAs) performed according to the instructions from the manufac-turer (R&D Systems, Minneapolis, MN) for the following markers:TNF- (PTA00), IL-10 (P1000) and PGE2 (DE0100). Values belowthe detection limit of an assay were treated as missing in the statis-tical analysis.

Immunohistochemical Analysis

Primary antibodies used in the study were a mouse anti-human-MMP-2 monoclonal antibody (MAB3308; Chemicon, Temecula, CA)and a rabbit anti-human MMP-9 polyclonal antibody (AB13458,Chemicon), both diluted at 1/150. Western immunoblotting using a7.5% polyacrydamide gel with a dual color protein standard (Preci-sion Plus Dual Color, Bio-Rad) and antibodies against MMP-2 andMMP-9 was performed to verify that the MMP antibodies used in thestudy recognized porcine MMP-2 and MMP-9, with protein bandsstaining at approximately 72 kDa and 92 kDa, respectively. A mono-clonal mouse anti-human CD68 antibody (M0876; DakoCytomation,Glostrup, Denmark) diluted at 1/100 was used as a macrophagemarker.Coagulations were excised from fresh liver tissue and cut in 4 to

6 mm thick slices. Each slice contained the central coagulated areaand the surrounding transition zone between coagulated and normalhepatic parenchyma. The slices were fixed in 4% formaldehyde for atleast 10 days, then dehydrated and embedded in paraffin. One 3 to5 m thick section was cut from each paraffin block and stained withhematoxylin and eosin. Additionally, 3 to 5 m thick sections weremounted on SuperFrost Plus adhesion slides (Menzel-Gläser, Braun-schweig, Germany). The sections were dried for 50 min at 56°Cfollowed by 37°C overnight and deparaffinized, hydrated throughalcohols, rinsed in water and pre-treated with dual endogenous en-zyme block for 10 min (EnVision Dual Link System-HRP (DAB);Dako, Glostrup, Denmark). Thereafter the sections were rinsed inwater and incubated in a wash solution (APK Wash Solution; Ven-tana Medical Systems, Tucson, AZ) and stained using an automatedimmunohistochemical stainer (NexES, Ventana Medical Systems)using iView DAB Detection Kit (Ventana Medical Systems) accord-ing to the manufacturer’s protocol. The sections were rinsed in run-ning tap water for 10 min and counterstained with Lillie’s hematox-ylin for 7 s, rinsed in lukewarm water for 15 min, dehydratedthrough alcohols and xylene and mounted in Eukitt Mounting Me-dium (Electron Microscopy Sciences, Hatfield, PA). Sections wereexamined by light microscopy by an experienced pathologist.

Statistical Analysis

SPSS for Mac OSX version 11.0.4 (SPSS, Chicago, IL) was used forthe statistical analyses. A P value of less than 0.05 was consideredstatistical significant. Data are presented as mean SD unlessotherwise indicated. Plasma values of MMPs and cytokines 1 h afterlaparotomy, but before RF ablation were normalized to a value of 1in each animal, and relative alterations in the corresponding plasmasamples collected 1 h after RF ablation were analyzed by a onesample t test, using a test value of 1.0. MMP activity and cytokineexpression in lysates from normal hepatic parenchyma were normal-ized to a value of 1 in each animal, and relative alterations in thecorresponding values from the transition zone were examined by aone sample t test, using a test value of 1.0. Data from all animalswere pooled in the statistical analysis. Differences in the two groupswere analyzed by a two-group t test. Sample size of the study wasestimated by nQuery Advisor 4 (Statistical Solutions, Saugus, MA).With a sample size of 6 in each experimental group, a single group ttest with a two-sided significance level of 0.05 would have 80% powerto detect a difference between a null hypothesis mean of 1 and analternative mean of 1.75, assuming a standard deviation of 0.5.Concerning differences between the two experimental groups, a sam-ple size of 6 in each group would have 80% power to detect adifference in means of 0.9 between the two groups assuming a com-mon standard deviation of 0.5 when using a two group t test with a0.05 two-sided significance level. The animals were allocated to thetwo experimental groups by block randomization, using a block sizeof two.

RESULTS

Blood samples from one animal in the non-Pringlegroups were missing. Hence, plasma was available foranalysis in five of the six animals in the non-Pringlegroup and six of the six animals in the Pringle group.

MMP Activity in Liver Tissue Lysates and Plasma

Five separate bands of gelatinolytic activity weredetected on the zymograms. A band with an approxi-mate molecular weight of 135 kDa represented theheterodimer of proMMP-9 and neutrophil gelatinase-associated lipocalin [21]. The next four bands repre-sented proMMP-9 (92kDa), active MMP-9 (82 kDa),proMMP-2 (72 kDa), and active MMP-2 (62 kDa)(Fig. 1). The active form of MMP-2 was observed in thenormal hepatic parenchyma and the transition zone in10 animals, whereas the active form of MMP-9 wasobserved in 6 animals, and only in specimens from thetransition zone.MMP-2 and MMP-9 activity in tissue lysates from

the transition zone was significantly increased com-pared to normal hepatic parenchyma with ratios of 3.0and 2.6, respectively (Table 1). MMP-2 and MMP-9activity in plasma 1 h after RF ablation was signifi-cantly increased compared with baseline levels withratios of 1.2 and 1.5, respectively (Table 2). MMP-2 andMMP-9 activity in plasma was normalized 4 days afterRF ablation.

299FRICH ET AL.: INCREASED MMP-2 AND MMP-9 AFTER RADIOFREQUENCY ABLATION

Cytokine Expression in Liver Tissue Lysates and Plasma

PGE2 levels in tissue lysates from the transition zonewas significantly increased compared to normal he-patic parenchyma with a ratio of 1.9. Correspondingtissue levels of TNF- and IL-10 were not increased,with ratios of 1.1 and 1.4, respectively (Table 1). Levelsof TNF- and PGE2 were not increased in plasma 1 hafter RF ablation compared with baseline levels, withratios of 1.0 and 0.8, respectively (Table 2). PGE2 wasonly detected in three of 11 available plasma samples.IL-10 levels were below the detection limit in allplasma samples. Levels of TNF- and PGE2 in plasmawere normalized 4 days after RF ablation.

Immunohistochemical Analysis

A sharp demarcation between coagulated tissue andthe surrounding hepatic parenchyma was present at

gross examination of the fresh tissue specimens in bothexperimental groups. This demarcation correspondedto a 1 to 3 mm transition zone between the coagulatedtissue and surrounding hepatic parenchyma seen onhematoxylin and eosin stained sections (Fig. 2). Thetransition zone consists of an inner necrotic zonewith dead hepatocytes, a rim of hemorrhagic mate-rial and thrombosed vessels, and an outer rim withhemorrhage and viable hepatocytes, as describedpreviously [22]. Immunohistochemical analysis us-ing antibodies directed against MMP-2 and MMP-9in sections from the transition zone showed cellsstrongly positive of MMP-2 and MMP-9 surroundedby proliferating fibroblasts and inflammatory cells.Positive staining for MMP-2 or MMP-9 antibodies wasnot detected in normal hepatic parenchyma (Fig. 3).

FIG. 1. Gelatin zymography showing MMP activity in three representative animals. Hepatic tissue was obtained 4 days after radiofre-quency ablation. Normal porcine hepatic parenchyma (N), transition zone separating coagulated and normal hepatic parenchyma (T). Thepositions of a 135-kDa complex of ProMMP-9 and neutrophil gelatinase-associated lipocalin, proMMP-9, active MMP-9, proMMP-2, and activeMMP-2 are indicated.

TABLE 1

Lysates from Tissue Samples Obtained 4 Days AfterRadiofrequency Ablation. Activity of Matrix Metallo-proteinase 2 and 9, and Levels of Tumor NecrosisFactor-, Interleukin-10 and Prostaglandin E2 in Ly-sates from the Transition Zone Relative to Values inNormal Hepatic Parenchyma

n Ratio P

MMP-2 12 3.0 2.0 (0.96–8.57) 0.005MMP-9 12 2.6 1.1 (1.38–5.52) 0.001TNF- 8* 1.1 0.3 (0.81–1.70) 0.234PGE2 12 1.9 1.2 (0.32–3.81) 0.025IL-10 12 1.4 1.0 (0.27–3.91) 0.230

Note. MMP matrix metalloproteinase; TNF- tumor necrosisfactor-; IL-10 interleukin-10; PGE2 prostaglandin E2. Valuesare mean standard deviation (min, max).* TNF- was below the detection limit in 4 animals.

TABLE 2

Plasma Samples Obtained 1 Hour After Radiofre-quency Ablation. Activity of Matrix Metalloproteinase2 and 9, and Levels of Tumor Necrosis Factor-,Interleukin-10 and Prostaglandin E2 Relative to Base-line Values 1 hour After Laparotomy

n2 Ratio P

MMP-2 11 1.2 0.2 (0.84–1.67) 0.036MMP-9 11 1.5 0.6 (0.84–2.45) 0.013TNF- 11 1.0 0.2 (0.68–1.18) 0.830PGE2 3 0.8 0.2 (0.66–0.93) 0.107IL-10 — — —

Note. MMP matrix metalloproteinase; TNF- tumor necrosisfactor-; IL-10 interleukin-10; PGE2 prostaglandin E2. Valuesare mean standard deviation (min, max).PGE2 was not detected in 8 animals.IL-10 was not detected in any animal.* Plasma was not available for analysis for one of the animals in

the non-Pringle group.

300 JOURNAL OF SURGICAL RESEARCH: VOL. 135, NO. 2, OCTOBER 2006

Sections from the transition zone contained cells stain-ing positive for the macrophage marker CD68, whereascells in the normal hepatic parenchyma did not stainpositive for the macrophage marker (Fig. 3).

Effect of Pringle Maneuver

Absolute MMP activity and cytokine values mea-sured in lysates from normal hepatic tissue were notsignificantly different in the non-Pringle group and thePringle group: MMP-2 (P 0.542), MMP-9 (P 0.592),

TNF- (P 0.951), IL-10 (P 0.751), and PGE2 (P 0.282). No differences were found in tissue lysates fromthe transition zone compared to normal hepatic paren-chyma: MMP-2 (P 0.531), MMP-9 (P 0.775), TNF-(P 0.107), PGE2 (P 0.612), and IL-10 (P 0.787).Thus, the Pringle maneuver did not affect the absolutevalues measured in normal hepatic tissue lysates orthe ratio in tissue lysates from the transition zonecompared to normal parenchyma. No qualitative dif-ferences between the two experimental groups wereseen on immunohistochemical analysis of sections fromthe transition zone. No significant differences werefound in plasma values for MMPs or cytokines in thenon-Pringle group and the Pringle group (data notshown).

DISCUSSION

RF ablation of malignant liver tumors is associatedwith local recurrence rates ranging from 2% to 60%[23]. Tumor size larger than 3 to 4 cm, perivasculartumor localization and multiple tumors are establishedrisk factors for local recurrence after hepatic RF abla-tion [23, 24]. Although local tumor recurrence rates com-parable to those found after hepatic anatomical andwedge resection has been achieved [25], patients treatedwith RF ablation have higher intrahepatic recurrencerates and inferior survival rate compared with resectedpatients. This difference cannot fully be explained byselection bias or local tumor recurrence at the site of the

FIG. 2. Hematoxylin and eosin stained section from porcine liver4 days after radiofrequency ablation. Normal hepatic parenchyma(N) and coagulated parenchyma (C) is separated by a sharply demar-cated transition zone (T).

FIG. 3. Immunohistochemical analysis of sections from hepatic parenchyma harvested 4 days after radiofrequency ablation. Originalmagnification 200 or 400 as indicated far left. Sections were incubated with antibodies directed against MMP-2 (not shown), MMP-9 andthe macrophage marker CD68. Large cells strongly positive for MMP-9 (arrow) are seen in the transition zone separating coagulated andnormal hepatic parenchyma (first column). Normal hepatic parenchyma showing the portal triad with hepatocytes and some inflammatorycells. No cells stained positive for MMP-9 (second column). Macrophages strongly positive for CD68 (arrow) are seen in the transition zone(third column). Normal hepatic parenchyma showing the portal triad with hepatocytes and some inflammatory cells. No cells stained positivefor CD68 (fourth column). f, fibroblast proliferation and inflammatory cells; h, hepatocytes; p, portal triad; m, macrophages.

301FRICH ET AL.: INCREASED MMP-2 AND MMP-9 AFTER RADIOFREQUENCY ABLATION

coagulation [10]. The biological effects induced by RFablation in the index tumor and surrounding hepaticparenchyma are not well characterized because of thefact that the treated tumors and surrounding paren-chyma are left in situ. Although cellular necrosis oc-curs when tissue is exposed to temperatures above 50to 55°C [26], cells in the periphery of a coagulated areaare exposed to sub-lethal temperatures associated witha wide range of cellular responses including proteindenaturation, inactivation of protein synthesis, cell cy-cle progression, and DNA repair processes [27]. Exper-imental animal data suggest that hepatic RF ablationpromotes intrahepatic growth of residual neoplasticcells [12]. Although the biological mechanisms leadingto induction of neoplastic growth after thermal RFablation remain unresolved, experimental studies haveshown that hepatic thermal ablation increase the ex-pression of heat shock proteins [28] and growth factors[13] adjacent to the treated site.In the present study MMP-2 and MMP-9 activity

was significantly increased in tissue lysates from thetransition zone 4 days after RF ablation. IncreasedMMP-2 and MMP-9 activity at the tumor-stroma in-terface in human tumor tissue sections are predomi-nantly expressed by reactive stromal cells and rep-resent a tumor-induced host response that can beadopted by malignant cells to promote tumor growthand invasion [29, 30]. The cellular origin of the in-creased MMP activity found in tissue lysates by gelatinzymography was examined in tissue sections from thetransition zone by immunohistochemical staining withhuman MMP-2 and MMP-9 antibodies as well as themacrophage marker CD68. The results of the CD68staining and the typical macrophage morphologyclearly show that the cells staining for MMP antibodiesin the transition zone were macrophages. Even thoughMMP-2 and MMP-9 activity was detected in normalhepatic tissue by zymography, no MMP-2 or MMP-9staining was found in normal hepatic tissue by immu-nohistochemical analysis of tissue sections. Westernblotting verified that the anti-human MMP-2 andMMP-9 antibodies used in the study recognized porcineMMP-2 and MMP-9. Therefore, we believe that thelack of staining for MMP-2 and MMP-9 in normalhepatic tissue was because of the very low MMP levelsin normal tissue, which most probably were below thedetection limit for immunohistochemistry.Increased plasma MMP levels were found 1 h after

RF ablation, possibly indicating that RF ablation in-duced acute MMP release from cells other than tissuemacrophages at the treatment site. However, our ap-proach of serial blood samples followed by analysis ofhepatic tissue lysates after euthanasia 4 days after RFablation does not allow us to draw conclusions regard-ing the temporal relationship between systemic andlocal tissue MMP activity.

Pringle maneuver during RF ablation is generallyassociated with increased coagulation volume [8],which is considered beneficial in treatment of largetumors to avoid viable residual tumor cells. To ourknowledge, no randomized studies have evaluated theimpact of hepatic vascular occlusion during RF abla-tion on local tumor recurrence. In a recent meta-analysis of local recurrence after hepatic RF ablation ofliver tumors, univariate analysis found slightly lowerlocal recurrence rates in patients in which hepatic vas-cular occlusion was used (9.3% versus 12.8%, P 0.038). In a multivariate analysis of the same data,hepatic vascular occlusion was not significant [23]. Al-though RF ablation with Pringle maneuver has notbeen shown to have higher recurrence rates comparedto RF ablation without Pringle, experimental data sug-gest that hepatic ischemia and reperfusion (I/R) in thepresence of hepatic micrometastases is a strong stim-ulus of tumor growth [31]. This is of importance ashepatic RF ablation is increasingly combined with he-patic resection, which has been shown to induce hema-togenous and intrahepatic dissemination of CRC cells[32]. Increased activity of MMP-9 in serum 1 weekafter I/R has been reported [33]. We therefore consid-ered it of interest to examine if Pringle influenced thetumor microenvironment by altering the expression ofMMPs and cytokines. Our data suggest that Pringlemaneuver for 10 min does not have an impact on ab-solute values of MMP-2 or MMP-9 activity or cytokineexpression in normal hepatic parenchyma. These ap-parently diverging findings may be explained by theshort duration of hepatic vascular occlusion used in ourstudy.The activity of MMPs is regulated at several levels

including gene transcription of a number of growthfactors, tumor promoters, oncogenes, and hormones[34]. Additionally, MMPs are regulated by proenzymeactivation and reversible inhibition of activated en-zymes by tissue inhibitors of MMP [15, 35]. IncreasedMMP expression can be detected in repair or remodel-ing processes as well as in diseased or inflamed tissuesand may act on pro-inflammatory cytokines, chemo-kines and other proteins to regulate inflammation [15].TNF- mediates many aspects of the acute inflamma-tory process, and is one of the earliest indicators ofsubsequent host responses, with peak plasma values1 h after intravenous injection of endotoxin [36]. In ourstudy, plasma levels of the pro-inflammatory mediatorTNF- or the anti-inflammatory mediators IL-10 andPGE2 was not present in plasma 1 h after hepatic RFablation, confirming findings in previous experimental[37] and clinical studies [38, 39]. Cytokine levels wereonly examined in blood samples drawn 1 h after abla-tion and 4 days postoperatively. We cannot exclude theexistence of a cytokine response in between these timepoints. In fact, a recent experimental study examining

302 JOURNAL OF SURGICAL RESEARCH: VOL. 135, NO. 2, OCTOBER 2006

the systemic inflammatory response after hepatic RFablation, cryotherapy or resection in porcine liver re-ported increased TNF- levels 6 h after RF ablation[39]. These diverging findings may therefore be influ-enced by the time of blood sample collection relative tothe procedure, but could also be related to differentextent of the coagulation induced in these studies. Inour study increased MMP activity in tissue lysates wasfound without a concurrent increase in tissue TNF-levels. Increased levels of PGE2 were found in tissuelysates from the transition zone, in accordance with theroles of prostaglandins in inflammatory reactions aftertissue damage [40].MMP activity is up-regulated in many forms of cancer,

including CRC [16, 30]. In a recent clinical study of 32patients with CRC liver metastasis treated by hepaticresection, both active and latent forms of MMP-2 andMMP-9 were increased in tissues containing metastatictumor compared to normal liver tissue, with median ra-tios of 17.6 for proMMP-2 and 2.9 for proMMP-9. Ofimportance, MMP expression was higher in the meta-static tumor itself than in the transitional tissue contain-ing the invasive edge and immediate surrounding tumorand normal tissue [18]. Our results show that RF abla-tion of normal hepatic parenchyma is associated with athree-fold increased activity of MMP-2 and MMP-9.Therefore, if tumor tissue is incompletely coagulated,viable cancer cells are exposed to a microenvironmentthat may increase the risk of tumor progression anddistant metastasis. On the other hand, one could hy-pothesize that in situ tumor destruction might providea beneficial antigen source for induction of antitumorimmunity [41]. However, based on the assumption thatincreased expression of MMPs in peritumoral tissue isassociated with increased risk of tumor cell growth, ourresults indicate that RF ablation should only be at-tempted when complete eradication of the target tumorincluding a safety margin is possible.This study was performed in an animal model sub-

ject to several limitations. All coagulations were per-formed in normal hepatic porcine parenchyma. MMPactivity in tissue was assessed at one time point 4 daysafter ablation. Our study design does not allow us todraw conclusions regarding the prognostic role of theincreased activity of MMP-2 and MMP-9 activity onclinical outcome data such as local tumor recurrence orpatient survival. Furthermore, all coagulations weremade with a RF ablation system using perfusion ofsaline into the tissue during ablation that may affectour results. Therefore, further research should be con-ducted to assess the expression of MMPs after hepaticRF ablation using other RF ablation systems.In conclusion, hepatic RF ablation was associated with

a three-fold increased activity of MMP-2 and MMP-9 inthe transition zone separating coagulated tissue fromnormal hepatic parenchyma. The use of Pringle ma-

neuver for 10 min during ablation did not influenceMMP activity. The clinical significance of the combina-tion of residual tumor cells and induction of a micro-environment that facilitates growth and of tumors isnot fully explored and should be assessed in animaltumor models with extended observation time.

ACKNOWLEDGMENTS

We thank the staff at the Interventional Centre for providingexcellent assistance, the Department of Comparative Medicine forsupport in animal care and Aasa R. Schjølberg at the Institute ofPathology for assistance with the immunohistochemical analysis.LF was supported by the Norwegian Cancer Society, grant no.

D02042/001. KB was supported by a postdoctoral fellowship fromUniversity of Oslo, Norway.

REFERENCES

1. Nordlinger B, Guiguet M, Vaillant JC, et al. Surgical resection ofcolorectal carcinoma metastases to the liver. A prognostic scoringsystem to improve case selection, based on 1568 patients. Associ-ation Francaise de Chirurgie. Cancer 1996;77:1254.

2. Fong Y, Fortner J, Sun RL, Brennan MF, Blumgart LH. Clin-ical score for predicting recurrence after hepatic resection formetastatic colorectal cancer: Analysis of 1001 consecutive cases.Ann Surg 1999;230:309.

3. Berber E, Pelley R, Siperstein AE. Predictors of survival afterradiofrequency thermal ablation of colorectal cancer metastasesto the liver: A prospective study. J Clin Oncol 2005;23:1358.

4. Pereira PL, Trubenbach J, Schenk M, et al. Radiofrequencyablation: in vivo comparison of four commercially availabledevices in pig livers. Radiology 2004;232:482.

5. Goldberg SN, Hahn PF, Tanabe KK, et al. Percutaneous radiofre-quency tissue ablation: Does perfusion-mediated tissue coolinglimit coagulation necrosis? J Vasc Interv Radiol 1998;9(Pt 1):101.

6. Lu DS, Raman SS, Vodopich DJ, Wang M, Sayre J, Lassman C.Effect of vessel size on creation of hepatic radiofrequency le-sions in pigs: Assessment of the “heat sink” effect. AJR Am JRoentgenol 2002;178:47.

7. Chinn SB, Lee FT Jr, Kennedy GD, et al. Effect of vascularocclusion on radiofrequency ablation of the liver: Results in aporcine model. AJR Am J Roentgenol 2001;176:789.

8. Chang CK, Hendy MP, Smith JM, Recht MH, Welling RE.Radiofrequency ablation of the porcine liver with complete he-patic vascular occlusion. Ann Surg Oncol 2002;9:594.

9. de Baere T, Bessoud B, Dromain C, et al. Percutaneous radio-frequency ablation of hepatic tumors during temporary venousocclusion. AJR Am J Roentgenol 2002;178:53.

10. Abdalla EK, Vauthey JN, Ellis LM, et al. Recurrence and out-comes following hepatic resection, radiofrequency ablation, andcombined resection/ablation for colorectal liver metastases.Ann Surg 2004;239:818.

11. Yokoyama N, Shirai Y, Ajioka Y, Nagakura S, Suda T,Hatakeyama K. Immunohistochemically detected hepatic mi-crometastases predict a high risk of intrahepatic recurrenceafter resection of colorectal carcinoma liver metastases. Cancer2002;94:1642.

12. von Breitenbuch P, Kohl G, Guba M, Geissler E, Jauch KW,Steinbauer M. Thermoablation of colorectal liver metastasespromotes proliferation of residual intrahepatic neoplastic cells.Surgery 2005;138:882.

303FRICH ET AL.: INCREASED MMP-2 AND MMP-9 AFTER RADIOFREQUENCY ABLATION

13. Nikfarjam M, Muralidharan V, Christophi C. Altered growthpatterns of colorectal liver metastases after thermal ablation.Surgery 2006;139:73.

14. Nelson AR, Fingleton B, Rothenberg ML, Matrisian LM. Matrixmetalloproteinases: Biologic activity and clinical implications.J Clin Oncol 2000;18:1135.

15. Parks WC, Wilson CL, Lopez-Boado YS. Matrix metalloprotein-ases as modulators of inflammation and innate immunity. NatRev Immunol 2004;4:617.

16. Wagenaar-Miller RA, Gorden L, Matrisian LM. Matrix metal-loproteinases in colorectal cancer: Is it worth talking about?Cancer Metastasis Rev 2004;23:119.

17. Zeng ZS, Guillem JG. Unique activation of matrix metal-loproteinase-9 within human liver metastasis from colorectalcancer. Br J Cancer 1998;78:349.

18. Waas ET, Wobbes T, Lomme RM, DeGroot J, Ruers T, HendriksT. Matrix metalloproteinase 2 and 9 activity in patients withcolorectal cancer liver metastasis. Br J Surg 2003;90:1556.

19. Li BH, Zhao P, Liu SZ, Yu YM, Han M, Wen JK. Matrixmetalloproteinase-2 and tissue inhibitor ofmetallo-proteinase-2 incolorectal carcinoma invasion and metastasis. World J Gastroen-terol 2005;11:3046.

20. Kleiner DE, Stetler-Stevenson WG. Quantitative zymography:Detection of picogram quantities of gelatinases. Anal Biochem1994;218:325.

21. Kjeldsen L, Johnsen AH, Sengelov H, Borregaard N. Isolationand primary structure of NGAL, a novel protein associated withhuman neutrophil gelatinase. J Biol Chem 1993;268:10425.

22. Raman SS, Lu DS, Vodopich DJ, Sayre J, Lassman C. Creationof radiofrequency lesions in a porcine model: Correlation withsonography, CT, and histopathology. AJR Am J Roentgenol2000;175:1253.

23. Mulier S, Ni Y, Jamart J, Ruers T, Marchal G, Michel L. Localrecurrence after hepatic radiofrequency coagulation: Multivar-iate meta-analysis and review of contributing factors. Ann Surg2005;242:158.

24. Bleicher RJ, Allegra DP, Nora DT, Wood TF, Foshag LJ, BilchikAJ. Radiofrequency ablation in 447 complex unresectable livertumors: Lessons learned. Ann Surg Oncol 2003;10:52.

25. Elias D, Baton O, Sideris L, Matsuhisa T, Pocard M, Lasser P.Local recurrences after intraoperative radiofrequency ablationof liver metastases: A comparative study with anatomic andwedge resections. Ann Surg Oncol 2004;11:500.

26. Storm FK, Harrison WH, Elliott RS, Morton DL. Hyperthermictherapy for human neoplasms: Thermal death time. Cancer1980;46:1849.

27. Lepock JR. How do cells respond to their thermal environment?Int J Hyperthermia 2005;21:681.

28. Rai R, Richardson C, Flecknell P, Robertson H, Burt A, ManasDM. Study of apoptosis and heat shock protein (HSP) expres-sion in hepatocytes following radiofrequency ablation (RFA).J Surg Res 2005;129:147.

29. Mc Donnell S, Chaudhry V, Mansilla-Soto J, Zeng ZS, Shu WP,Guillem JG. Metastatic and non-metastatic colorectal cancer(CRC) cells induce host metalloproteinase production in vivo.Clin Exp Metastasis 1999;17:341.

30. Coussens LM, Fingleton B, Matrisian LM. Matrix metallopro-teinase inhibitors and cancer: Trials and tribulations. Science2002;295:2387.

31. van der Bilt JD, Kranenburg O, Nijkamp MW, et al. Ischemia/reperfusion accelerates the outgrowth of hepatic micrometastasesin a highly standardized murine model. Hepatology 2005;42:165.

32. Topal B, Aerts JL, Roskams T, et al. Cancer cell disseminationduring curative surgery for colorectal liver metastases. EurJ Surg Oncol 2005;31:506.

33. Kuyvenhoven JP, Molenaar IQ, Verspaget HW, et al. PlasmaMMP-2 and MMP-9 and their inhibitors TIMP-1 and TIMP-2during human orthotopic liver transplantation. The effect of apro-tinin and the relation to ischemia/reperfusion injury. ThrombHaemost 2004;91:506.

34. Overall CM, Kleifeld O. Tumour microenvironment—opinion:Validating matrix metalloproteinases as drug targets and anti-targets for cancer therapy. Nat Rev Cancer 2006;6:227.

35. Knittel T, Mehde M, Grundmann A, Saile B, Scharf JG,Ramadori G. Expression of matrix metalloproteinases andtheir inhibitors during hepatic tissue repair in the rat. His-tochem Cell Biol 2000;113:443.

36. Lin E, Calvano SE, Lowry SF. Inflammatory cytokines and cellresponse in surgery. Surgery 2000;127:117.

37. Chapman WC, Debelak JP, Wright Pinson C, et al. Hepaticcryoablation, but not radiofrequency ablation, results in lunginflammation. Ann Surg 2000;231:752.

38. Schell SR, Wessels FJ, Abouhamze A, Moldawer LL, CopelandEM 3rd. Pro- and antiinflammatory cytokine production afterradiofrequency ablation of unresectable hepatic tumors. J AmColl Surg 2002;195:774.

39. Ng KK, Lam CM, Poon RT, et al. Comparison of systemicresponses of radiofrequency ablation, cryotherapy, and surgicalresection in a porcine liver model. Ann Surg Oncol 2004;11:650.

40. Williams CS, MannM, DuBois RN. The role of cyclooxygenases ininflammation, cancer, and development. Oncogene 1999;18:7908.

41. den Brok MH, Sutmuller RP, van der Voort R, et al. In situtumor ablation creates an antigen source for the generation ofantitumor immunity. Cancer Res 2004;64:4024.

304 JOURNAL OF SURGICAL RESEARCH: VOL. 135, NO. 2, OCTOBER 2006

V

1

Radiofrequency ablation of colorectal liver metastases: evaluation of local tumor recurrence by 3D volumetric analysis

Lars Frich1, 2

Gaute Hagen 3

Knut Brabrand 3

Bjørn Edwin 1, 2

Øystein Mathisen 1

Trond Mogens Aaløkken 3

Ivar P. Gladhaug 1, 4

1. Department of Surgery, Rikshospitalet University Hospital, 0027 Oslo, Norway2. The Interventional Centre, Rikshospitalet University Hospital, 0027 Oslo, Norway3. Department of Radiology, Rikshospitalet University Hospital, 0027 Oslo, Norway4. Department of Surgery, Faculty Division Rikshospitalet, University of Oslo,

0027 Oslo, Norway

Word count Abstract 193 wordsManuscript including references 4548 words

Running title Volumetric evaluation of local tumor recurrence

Subject category Radiology, gastrointestinal, oncology

Corresponding author Lars FrichDepartment of SurgeryRikshospitalet University HospitalN-0027 OsloNORWAYTel. + 47 23071825Fax. + 47 23072267E-mail: lars.frich@labmed.uio.no

This work was supported by the Norwegian Cancer Society, grant no. D02042/001.

2

Abstract

Background: The purpose of this study was to characterize the relationship between local tumor

recurrence after radiofrequency (RF) ablation of colorectal liver metastases and post-ablation

alterations in coagulation volume.

Materials and methods: 11 patients with hepatic metastases from colorectal cancer were treated

with RF ablation. All patients were followed for at least 24 months or until local tumor recurrence.

Multi-detector computed tomography (MDCT) was performed at 1 and 3 months after RF ablation,

and then at intervals of 3 months until 24 months and each sixth month thereafter. Median follow-

up was 27 months (17-43 months). Local tumor recurrence was found in 7 patients. Coagulation

volume was measured from 81 follow-up MDCT scans using a semi-automatic three-dimensional

(3D) method.

Results: Patients with local tumor recurrence had a median increase in coagulation volume of 50%

(range 0-378%) compared to images acquired 3 months previously. Local tumor recurrence was

criteria.

Conclusions: Semi-automatic 3D volumetric analysis provides added diagnostic information

diagnosis of local tumor recurrence after RF ablation of colorectal liver metastases.

Key words: Radiofrequency ablation, liver, computed tomography

3

Introduction

Metastatic disease to the liver is a frequent event in patients with colorectal cancer [1]. Surgical

resection is a potentially curative treatment of liver metastases from colorectal cancer, with 5-

year survival rates of 30-50 per cent in selected patients [2-4]. The majority of patients with

colorectal liver metastases are not candidates for hepatic resection. Radiofrequency (RF) ablation is

increasingly used for in-situ tumor destruction in patients with non-resectable metastases limited to

ablation, the reported rate of local tumor recurrence varied between 2% and 60%. Follow-up was

short in several studies, probably contributing to an underestimation of the true rate of local tumor

but have been reported to appear up to 23 months after the procedure [7-9].

for patient survival. Malignant tissue treated by RF ablation is left in situ, necessitating repeated

follow-up imaging or biopsy to determine if the tumor has been eradicated. Furthermore, biopsy

of potentially operable colorectal liver metastases is controversial due to the risk of needle tract

seeding and negative impact on patient survival [10, 11]. Consequently, non-invasive methods

should preferably be used for detection of persistent or residual malignant tissue at the site of

alterations on follow-up imaging [12-14]. Analysis of coagulation geometry and coagulation

volume on early post-ablation scans may identify patients with inadequate treatment margins

relative to the index tumor, which would be expected to have a high risk of local tumor recurrence.

Various morphologic patterns are associated with local tumor recurrence on post-ablation computed

tomography (CT) [13]. An enhancing peripheral tissue rim surrounding the non-enhancing

coagulation may represent local tumor recurrence, but may also represent benign periablational

to thermally damaged cells [15, 16]. Additionally, local tumor recurrence may be diagnosed based

4

to quantify small alterations in coagulation diameter reliably due to inter-scan displacement and

deformation of the liver. Increase in coagulation volume may be a more sensitive indicator of

growth than increase in coagulation diameter; for a spherical volume, an increase in diameter of

20% corresponds to a volume increase of 73%. However, the relationship between local tumor

recurrence after RF ablation of hepatic colorectal metastases and alterations in coagulation volume

is not well characterized.

In this clinical pilot study, patients with colorectal liver metastases treated with RF ablation were

followed for at least 24 months or until local tumor recurrence. To assess theoretical treatment

margins, we determined the largest tumor diameter on pre-ablation images, and the effective

coagulation diameter 1 month after RF ablation. Post-ablation coagulation volumes were measured

on repeat follow-up scans using a semi-automatic three-dimensional (3D) technique. Our study

hypotheses were: (1) inadequate treatment margins as determined from scans 1 month after

radiofrequency ablation may identify patients who would develop local tumor recurrence, and (2)

increase in coagulation volume during follow-up may be used to detect local tumor recurrence

earlier than conventional morphological criteria.

5

Materials and methods

Study population

This study was approved by the Regional committee for medical research ethics. From 2003 to

2006 a total of 23 patients underwent RF ablation for non-resectable malignant liver tumors at

our institution. Written informed consent was obtained from all patients prior to the procedure.

following criteria: (a) non-resectable liver metastasis from colorectal cancer; (b) complete clearance

of all liver tumors possible either by RF ablation alone or combined with surgical resection; (c)

diameter of tumor treated with radiofrequency ablation less than 4 cm; (d) no extrahepatic disease;

(e) follow-up > 24 months, or to local tumor recurrence. The reasons for exclusion were non-

colorectal cancer liver metastasis (n=1), tumor diameter > 4 cm (n=2), extrahepatic tumor growth

present at time of treatment (n=1), and follow-up less than 24 months without occurrence of local

tumor recurrence (n=8).

All patients were treated with a curative intent. Patient age was 60.3 ± 8.0 yr (49-80 yr). The

primary tumor location was the colon in 8 patients and the rectum in 3 patients. All patients had

undergone curative resection of the primary colorectal tumor. Liver resection of colorectal liver

metastases had previously been performed in 7 of 11 patients. The median interval between

operation of the primary tumor (n=11) and RF ablation was 17 months (3-101 months). The median

interval between liver resection (n=7) and RF ablation was 12 months (2-77 months). The patients

included in this study had 1-4 liver tumors, with a median of 1 tumor. One tumor was treated by RF

Operative procedure

Patients were treated with a RF ablation system with perfusion electrodes (Elektrotom HiTT 106,

Berchtold GmbH & Co, Germany). This system incorporates a generator that delivers alternating

current at 375 kHz to a maximum of 1.2 ampere, and a syringe pump (Pilot C, Fresenius Vial,

France) for infusion of isotonic saline through the electrode tip during ablation. All procedures

6

were performed under general anesthesia. The approach used for RF ablation was percutaneous

(n=4), laparoscopic (n=4) and by laparotomy (n=3). Intraprocedural image guidance was contrast-

enhanced ultrasonography (n=6), intraoperative ultrasonography (n=4) and magnetic resonance

imaging (MRI) (n=1). RF ablation and hepatic resection was performed during the same procedure

in 4 patients, one of these resections was performed as a laparoscopic procedure. The median

number of electrode activations in each procedure was 2 (2-4). Hepatic vascular occlusion was not

used during RF ablation in any patients. CT or contrast-enhanced ultrasonography was performed in

all patients prior to discharge to assess the completeness of the treatment.

Follow-up

All patients were followed with a standardized imaging protocol. Sixteen or 64-channel multi-

detector computed tomography (MDCT) (GE LightSpeed Pro 16 or GE LightSpeed VCT; GE

Healthcare, Milwaukee, WI, USA), was performed using 150-200 mL iodixanol 320 mgI/mL

(Visipaque, GE Healthcare, USA). Images were acquired in portal venous phase. Scans were

performed at 1 and 3 months after RF ablation, and then every 3 months until 24 months, and every

sixth month thereafter. A total of 81 post-ablation scans from 1 month after ablation and onwards

were reviewed, with a follow-up of 27 months (17-43 months). CT of the chest was performed

every sixth month. Carcinoembryonic antigen (CEA) with a cut-off level of 5 μg/L was used a

marker of malignancy, and was measured at all time points. All patients were followed for at least

24 months or until local tumor recurrence.

Assessment of local tumor recurrence

morphological criteria on follow-up CT examinations; new lesion(s) in the periphery of the

coagulated area; at least 20% growth of the largest diameter of the coagulated area; persistent

attenuation in the periphery of the coagulated area different from the coagulated area and from

normal hepatic parenchyma. During follow-up a strict distinction was kept between local tumor

recurrence, i.e. tumor growth within or at the ablation margin, intrahepatic tumors not related to the

areas treated with RF ablation, and extrahepatic tumor growth.

7

Treatment margin

The largest coagulation diameter and the effective coagulation diameter were retrospectively

program ImageJ version 1.37 (National institute of mental health, Bethesda, MA, USA). A region

attributed to RF ablation on all slices where tissue coagulation was present. Measurement

of geometric properties of each ROI was performed using automatic algorithms. The largest

The largest pre-ablation tumor diameter was subtracted from the effective coagulation diameter, and

this difference was divided by 2 to achieve the theoretical treatment margin surrounding the tumor

on all sides.

Coagulation volume

All images were retrospectively reviewed by an experienced radiologist unaware of the course of

the disease in each patient, who had not been involved in the RF ablation procedures or prospective

interpretation of follow-up scans. Images of each patient were reviewed in chronological order.

Tumors on preopoerative images contiguous to a vessel with diameter of 3 mm or more were

Workstation 4.2; GE Healthcare, USA). The coagulation volumes were calculated by the software

package Volume Viewer (GE Healthcare, USA). Using a seeding based semi-automatic technique,

representing tissue coagulation and normal hepatic parenchyma. The extent of the VOI was

controlled on coronal and axial source images. The volume of the selected VOI was calculated by

the software package.

Statistical analysis

SPSS for Mac OSX version 11.0.4 (SPSS, Chicago, Illinois, USA) was used for the statistical

8

mean ± SD (range), or median (range). All post-ablation coagulation volumes were normalized to

the 1-month baseline value for each patient. Independent samples t-test was used for comparison

between groups. Wilcoxon signed rank test was used for repeated measurements within each group.

9

Results

Local, intrahepatic and extrahepatic tumor recurrence

Diameter of the tumors treated with RF ablation was 2.2 ± 0.8 cm (1.0-3.5 cm), with a tumor

volume of 7.1 ± 6.6 mL (0.5-22.5 mL) (Table 1). Treatment of the index tumor was initially

considered complete in all patients. Local tumor recurrence was found during follow-up in 7

patients, and was detected median 9 months (6-21 months) after RF ablation. Tumor diameter

in the patients who developed local tumor recurrence was 2.2 ± 0.3 cm (1.0-3.5 cm) compared

with 2.1 ± 0.4 cm (1.2-3.1 cm) in the patients without local tumor recurrence. This difference was

malignant cells were found in 1 patient who did not develop local tumor recurrence. Local tumor

recurrence was found in 5 of 7 perivascular tumors. Of the 7 patients with local tumor recurrence, 4

patients presented with inoperable intrahepatic (n=1) or extrahepatic (n=3) malignant growth at the

time of diagnosis of local tumor recurrence. The 3 patients with local tumor recurrence only were

re-treated with resection (n=1) or repeat RF ablation (n=2). Considering all 11 patients, intrahepatic

recurrence not related to the RF ablation site was found in 6 patients with a median of 14.5 months

(4-24 months) after RF ablation, and extrahepatic tumor growth was found in 6 patients with a

median of 11.5 months (4-22 months) after RF ablation, with the following distribution; lungs

(n=4), lymph nodes (n=1), bone (n=1). Two patients who had developed extrahepatic disease died

17 and 20 months after RF ablation. The remaining 9 patients, including the 3 patients who were re-

treated, were alive with a median follow-up of 28 months (20-43 months).

Treatment margin

Largest coagulation diameter 1 month after RF ablation was 4.8 ± 1.4 cm (2.9-7.5 cm), and was

larger than the corresponding pre-ablation tumor diameter in all patients. Effective coagulation

diameter was 2.9 ± 0.8 cm (1.6-4.2 cm). The effective coagulation diameter was larger than the

corresponding tumor diameter in 8 patients (Table 1), with a positive margin of 0.6 ± 0.3 cm (0.3-

1.0 cm). The 3 patients in which the effective coagulation diameter was smaller or equal to the

tumor diameter developed local tumor recurrence. In the 7 patients with local tumor recurrence

10

the margin was 0.3 ± 0.6 cm (-0.4-1.0 cm), compared to 0.6 ± 0.3 cm (0.3-0.9 cm) in the 4 patients

Coagulation volume

Baseline coagulation volume 1 month after RF ablation was 29.7 ± 16.0 mL (13-70 mL), and was

larger than the corresponding tumor volume in all patients. At 3 months, coagulation volume had

decreased ´in all patients, and was 54.1 ± 16.9% (38-88%) of the baseline value. In the patients

without local tumor recurrence (n=4), subsequent measurements showed generally decreasing

coagulation volumes, with all values below 50% of baseline values at 12 months and below 30%

of baseline values at 24 months. One patient had a volume increase of 3% between two successive

measurements, but decreasing volume on the following measurement (Figure 2a). Patients who

85 ± 131% (0-378%) from the examination 3 months prior to diagnosis of local tumor recurrence

until diagnosis of local tumor recurrence (p=0.028). The median volume increase in these patients

was 50% (Figure 2b). One patient had local tumor recurrence without an associated alteration in

coagulation volume (Figure 3). Coagulation volume measurements 3 months after diagnosis of local

tumor recurrence were available in 5 patients, showing further increase in coagulation volume in 4

patients (Table 2). Of importance, 4 of the 7 patients with local tumor recurrence had a coagulation

volume less than the baseline volume at time of diagnosis.

11

Discussion

Several methods have been proposed for early detection of local tumor recurrence after RF ablation

geometry [17]. In the present study we retrospectively examined if inadequate treatment margins

assessed 1 month post-ablation could predict local tumor recurrence, and if increase in coagulation

volume on post-ablation MDCT images could be used to detect local tumor recurrence earlier than

conventional morphological criteria.

Established oncological criteria suggest that hepatic malignancies should be eradicated radically

including a 1-cm margin of apparently healthy tissue to eliminate microscopic foci of malignant

cells, and to compensate for the uncertainty in determining the exact tumor margin [18, 19].

Considering the effective coagulation diameters produced in this study, we did not coagulate

a coagulation diameter at least 2 cm larger than the largest tumor diameter) in any patients.

Theoretically, a margin of at least 0.5 cm could have been achieved in 5 patients, whereas 3

patients had a positive margin less than 0.5 cm. An effective coagulation diameter smaller or equal

to the pre-ablation tumor diameter was found in 3 patients, and was invariably associated with

to be examined in larger prospective studies. Even though one would expect a large margin to

developed local tumor recurrence (Table 1). The calculated margins surrounding the tumors may

have been overestimated if the tumor diameter was larger than anticipated from pre-ablation images.

Furthermore, even if an adequate coagulation volume was achieved, suboptimal positioning of the

ablation electrode or displacement of the geometrical center of the coagulated volume relative to the

ablation electrode may result in inadequate treatment [20].

malignant and non-malignant tissue. Coagulation volume decreased from baseline at 1 month to

12

3 months in all patients. Consequently, decreasing coagulation volume from 1 to 3 months after

RF ablation cannot be used as a predictor for the absence of later development of local tumor

recurrence. A marked increase in coagulation volume when local tumor recurrence was diagnosed

was seen in 6 of 7 patients. The lack of increase in coagulation volume in one patient was probably

due to the occurrence of nodules at a small distance from the coagulated area which were not

This indicates that semi-automatic volumetric analysis should only be used as a supplement to

conventional morphological criteria. With a follow-up interval of 3 months, increase in coagulation

volume was not present until local tumor recurrence was detected based on conventional criteria

(Figure 2b).

The natural course of adequately treated colorectal metastases is a decrease in coagulation volume,

as illustrated by the patients who did not develop local tumor recurrence. In the 6 patients where

coagulation volume was increased at the time of local tumor recurrence, the increase in volume

was at least 19% compared to the previous examination 3 months earlier, with a median increase

determine with certainty whether local tumor recurrence was caused by incomplete coagulation of

a tumor that continued to grow, or if a new tumor grew at or near the site of the adequately treated

tumor [16]. Untreated colorectal liver metastases have a tumor volume doubling time of 3-5 months

[21]. Viable malignant tissue after RF ablation would in most cases be expected to constitute a small

fraction of the coagulated volume. It is therefore somewhat surprising that a median coagulation

volume increase of 50% in 3 months was observed in our patients with local tumor recurrence.

However, recent experimental research indicate that RF ablation induces biological alterations in

the tissue of possible importance for tumor invasion and metastasis [22], and that thermal ablation

promotes growth of residual neoplastic cells, possibly through stimulation by growth factors [23,

24]. Therefore, one could speculate that the growth rate of viable residual malignant cells exposed

for thermal ablation may be higher than for untreated malignant tissue.

Three patients with local tumor recurrence were biopsied. In the remaining 4 patients with local

tumor recurrence, biopsy was not obtained due to simultaneous intrahepatic or extraheptic tumor

13

growth, which precluded potential curative surgical resection or local ablation. In these 4 patients

retrospect, operable patients with suspected local tumor recurrence by morphological criteria and

a concurrent marked increase in coagulation volume could be considered for repeated treatment

without performing biopsy, which delays potentially curative treatment and may lead to tumor

diagnosis of local tumor recurrence after RF ablation of colorectal cancer liver metastases, and

thereby possibly avoid unnecessary biopsies.

In this study we found local tumor recurrence in 7 of 11 patients, corresponding to a rate of 64%.

Reported local tumor recurrence rates varies from less than 20% [25], to recently reported rates

of 47% and 58% [26, 27]. Tumor proximity to major vessels is a well-established risk factor of

local tumor recurrence [6]. A high proportion of the tumors treated in this study were perivascular,

produces irregular coagulations [20, 28, 29], which may contribute to the high rate of local tumor

recurrence. Furthermore, a learning curve exists for new procedures, including RF ablation [6, 30].

Therefore, as this patient series represents our initial experience with RF ablation, a lower rate of

local tumor recurrence would be expected with increasing experience.

This retrospective pilot study has some important limitations. It presents data from only 11 patients.

This sample size is too small to allow conclusions regarding the prognostic value of the treatment

effective coagulation diameter were made from 2D axial images. Local tumor recurrence was

adequate to detect local tumor recurrence, the number of patients with malignant residual tumor

the perfusion electrode RF ablation system used in this study. Dual-modality positron emission

tomography (PET)/CT may have higher sensitivity to detect viable malignant tumor after RF

14

ablation compared to CT [31, 32], however this image modality was not evaluated in this study.

In conclusion, by acquiring follow-up MDCT images each third month after hepatic RF ablation,

criteria and semi-automatic 3D volumetric analysis. In patients with local tumor recurrence an

increase in median coagulation volume by 50% was seen, whereas in patients without local tumor

occurrence decreasing coagulation volume during the follow-up period was observed, with all

coagulation volumes less than 30% of baseline at 24 months. Semi-automatic 3D volumetric

analysis provides added diagnostic information compared with conventional morphological

RF ablation of colorectal cancer liver metastases.

15

References

1. Hess, K.R., G.R. Varadhachary, S.H. Taylor, W. Wei, M.N. Raber, R. Lenzi, et al. (2006)

Metastatic patterns in adenocarcinoma. Cancer 106:1624-33.

2. Scheele, J. and A. Altendorf-Hofmann (1999) Resection of colorectal liver metastases.

Langenbecks Arch Surg 384:313-27.

3. Wei, A.C., P.D. Greig, D. Grant, B. Taylor, B. Langer, and S. Gallinger (2006) Survival after

hepatic resection for colorectal metastases: a 10-year experience. Ann Surg Oncol 13:668-

76.

4. Simmonds, P.C., J.N. Primrose, J.L. Colquitt, O.J. Garden, G.J. Poston, and M. Rees (2006)

Surgical resection of hepatic metastases from colorectal cancer: a systematic review of

published studies. Br J Cancer 94:982-99.

5. Berber, E., R. Pelley, and A.E. Siperstein (2005) Predictors of survival after radiofrequency

thermal ablation of colorectal cancer metastases to the liver: a prospective study. J Clin

Oncol 23:1358-64.

6. Mulier, S., Y. Ni, J. Jamart, T. Ruers, G. Marchal, and L. Michel (2005) Local recurrence

after hepatic radiofrequency coagulation: multivariate meta-analysis and review of

contributing factors. Ann Surg 242:158-71.

7. Siperstein, A., A. Garland, K. Engle, S. Rogers, E. Berber, A. Foroutani, et al. (2000) Local

recurrence after laparoscopic radiofrequency thermal ablation of hepatic tumors. Ann Surg

Oncol 7:106-13.

8. Solbiati, L., T. Livraghi, S.N. Goldberg, T. Ierace, F. Meloni, M. Dellanoce, et al. (2001)

Percutaneous radio-frequency ablation of hepatic metastases from colorectal cancer: long-

term results in 117 patients. Radiology 221:159-66.

9. Adam, R., E.J. Hagopian, M. Linhares, J. Krissat, E. Savier, D. Azoulay, et al. (2002) A

comparison of percutaneous cryosurgery and percutaneous radiofrequency for unresectable

hepatic malignancies. Arch Surg 137:1332-9.

10. Rodgers, M.S., R. Collinson, S. Desai, R.S. Stubbs, and J.L. McCall (2003) Risk of

dissemination with biopsy of colorectal liver metastases. Dis Colon Rectum 46:454-8.

11. Jones, O.M., M. Rees, T.G. John, S. Bygrave, and G. Plant (2005) Biopsy of resectable

16

colorectal liver metastases causes tumour dissemination and adversely affects survival after

liver resection. Br J Surg 92:1165-8.

12. Berber, E., A. Foroutani, A.M. Garland, S.J. Rogers, K.L. Engle, T.L. Ryan, et al. (2000) Use

ablation of hepatic tumors. Surg Endosc 14:799-804.

13. Chopra, S., G.D. Dodd, 3rd, K.N. Chintapalli, J.R. Leyendecker, O.I. Karahan, and H. Rhim

(2001) Tumor recurrence after radiofrequency thermal ablation of hepatic tumors: spectrum

14. Dromain, C., T. de Baere, D. Elias, V. Kuoch, M. Ducreux, V. Boige, et al. (2002) Hepatic

tumors treated with percutaneous radio-frequency ablation: CT and MR imaging follow-up.

Radiology 223:255-62.

15. Goldberg, S.N., G.S. Gazelle, and P.R. Mueller (2000) Thermal ablation therapy for focal

guidance. AJR Am J Roentgenol 174:323-31.

16. Goldberg, S.N., C.J. Grassi, J.F. Cardella, J.W. Charboneau, G.D. Dodd, 3rd, D.E. Dupuy,

et al. (2005) Image-guided tumor ablation: standardization of terminology and reporting

criteria. Radiology 235:728-39.

17. Bricault, I., R. Kikinis, P.R. Morrison, E. Vansonnenberg, K. Tuncali, and S.G. Silverman

(2006) Liver metastases: 3D shape-based analysis of CT scans for detection of local

recurrence after radiofrequency ablation. Radiology 241:243-50.

18. Cady, B., R.L. Jenkins, G.D. Steele, Jr., W.D. Lewis, M.D. Stone, W.V. McDermott, et

al. (1998) Surgical margin in hepatic resection for colorectal metastasis: a critical and

improvable determinant of outcome. Ann Surg 227:566-71.

19. Blumgart, L.H. (2000) Liver resection for benign disease and for liver and biliary tumors.

In: L.H. Blumgart and Y. Fong (eds) Surgery of the liver and biliary tract. W. B. Saunders,

London, pp 1639-1713

20. Frich, L., T. Mala, and I.P. Gladhaug (2006) Hepatic radiofrequency ablation using perfusion

electrodes in a pig model: Effect of the Pringle manoeuvre. Eur J Surg Oncol 32:527-532.

21. Finlay, I.G., D. Meek, F. Brunton, and C.S. McArdle (1988) Growth rate of hepatic

metastases in colorectal carcinoma. Br J Surg 75:641-4.

17

22. Frich, L., K. Bjornland, S. Pettersen, O.P. Clausen, and I.P. Gladhaug (2006) Increased

activity of matrix metalloproteinase 2 and 9 after hepatic radiofrequency ablation. J Surg

Res 135:297-304.

23. von Breitenbuch, P., G. Kohl, M. Guba, E. Geissler, K.W. Jauch, and M. Steinbauer (2005)

Thermoablation of colorectal liver metastases promotes proliferation of residual intrahepatic

neoplastic cells. Surgery 138:882-7.

24. Nikfarjam, M., V. Muralidharan, and C. Christophi (2006) Altered growth patterns of

colorectal liver metastases after thermal ablation. Surgery 139:73-81.

25. Bleicher, R.J., D.P. Allegra, D.T. Nora, T.F. Wood, L.J. Foshag, and A.J. Bilchik (2003)

Radiofrequency ablation in 447 complex unresectable liver tumors: lessons learned. Ann

Surg Oncol 10:52-8.

26. van Duijnhoven, F.H., M.C. Jansen, J.M. Junggeburt, R. van Hillegersberg, A.M. Rijken,

ablation of colorectal liver metastases. Ann Surg Oncol 13:651-8.

Percutaneous saline-enhanced radiofrequency ablation of unresectable hepatic tumors: initial

experience in 26 patients. AJR Am J Roentgenol 180:1537-45.

28. Denys, A.L., T.D. Baere, V. Kuoch, B. Dupas, P. Chevallier, D.C. Madoff, et al. (2003)

Radio-frequency tissue ablation of the liver: in vivo and ex vivo experiments with four

different systems. Eur Radiol 13:2346-52.

29. Pereira, P.L., J. Trubenbach, M. Schenk, J. Subke, S. Kroeber, I. Schaefer, et al. (2004)

Radiofrequency ablation: in vivo comparison of four commercially available devices in pig

livers. Radiology 232:482-90.

30. Poon, R.T., K.K. Ng, C.M. Lam, V. Ai, J. Yuen, S.T. Fan, et al. (2004) Learning curve for

radiofrequency ablation of liver tumors: prospective analysis of initial 100 patients in a

tertiary institution. Ann Surg 239:441-9.

31. Langenhoff, B.S., W.J. Oyen, G.J. Jager, S.P. Strijk, T. Wobbes, F.H. Corstens, et al. (2002)

recurrence after local ablative therapy for liver metastases: a prospective study. J Clin Oncol

20:4453-8.

18

32. Veit, P., G. Antoch, H. Stergar, A. Bockisch, M. Forsting, and H. Kuehl (2006) Detection of

residual tumor after radiofrequency ablation of liver metastasis with dual-modality PET/CT:

initial results. Eur Radiol 16:80-7.

19

Figure 1

a. 55-year old man with liver metastases from colorectal cancer. Portal venous phase multi-detector

computed tomography acquired 1 month after resection of segment II/III and radiofrequency

coagulation.

b. Geometric analysis of the coagulation in (a). The thick line denotes the border of normal liver

tissue vs. alterations attributed to coagulation. Dmax denotes maximum axial coagulation diameter

and Deff effective coagulation diameter. On the shown slice, Dmax is 4.6 cm, whereas Deff is 3.4 cm.

Figure 2

Coagulation volumes 1-24 months after radiofrequency ablation of colorectal liver metastases.

Coagulation volumes for each patient have been normalized to the corresponding coagulation

volume 1 month after radiofrequency ablation.

a. Patients without local tumor progression (n=4). All values were below 50% of baseline values at

12 months and below 30% of baseline values at 24 months.

b. Patients with local tumor progression (n=7). The time of diagnosis of local tumor progression is

marked by a circle. The patient marked by as asterisk had a volume increase to above 400% when

local tumor progression was diagnosed at 6 months.

Figure 3

55-year-old man with liver metastases from colorectal cancer in which local tumor progression

was not detected by semi-automated 3D volumetric analysis. Portal venous phase multi-detector

computed tomography. The preoperative CEA was 64 μg/L. Resection of segment II and III, local

resection in segment VI and radiofrequency ablation of a tumor in segment VIII was performed

in the same procedure. 1 month after radiofrequency ablation coagulation volume was 26 mL,

and CEA was 3 μg/L. a. 9 months after radiofrequency ablation. The coagulation volume was

20

hyperemia is seen.

b. 12 months after radiofrequency ablation. The coagulation volume was calculated to 7 mL. CEA

was 16 μg/L. A heterogeneous low-attenuation nodule representing local tumor progression is

(arrow).

21

Table 1

Patient characteristics.

Table 2

Post-ablation coagulation volumes in patients with local tumor progression (n=7).

22

Figure 1

a. 55-year old man with liver metastases from colorectal cancer. Portal venous phase multi-detector computed tomography acquired 1month after resection of segment II/III and radiofrequency ablation of a tumor with a diameter of 3.1 cm in segment V shows a well-

b. Geometric analysis of the coagulation in (a). The thick line denotes the border of normal liver tissue vs. alterations attributed to coagula-tion. Dmax denotes maximum axial coagula-tion diameter and Deff effective coagulation diameter. On the shown slice, Dmax is 4.6 cm, whereas Deff is 3.4 cm.

23

a. Patients without local tumor progression (n=4). All values were below 50% of base-line values at 12 months and below 30% of baseline values at 24 months.

b. Patients with local tumor progression (n=7). The time of diagnosis of local tumor progression is marked by a circle. The patient marked by as asterisk had a volume increase to above 400% when local tumor progression was diagnosed at 6 months.

Figure 2

Coagulation volumes 1-24 months after radiofrequency ablation of colorectal liver metastases. Co-agulation volumes for each patient have been normalized to the corresponding coagulation volume 1 month after radiofrequency ablation.

24

a. 9 months after radiofrequency ablation. The coagulation volume was calculated to 7 mL.

-ation coagulation without peripheral hyper-emia is seen.

b. 12 months after radiofrequency ablation. The coagulation volume was calculated to 7 mL. CEA was 16 μg/L. A heterogeneous low-attenuation nodule representing local tumor progression is seen at the margin of the coagulated area located dorsal and medial to

Figure 3

55-year-old man with liver metastases from colorectal cancer in which local tumor progression was not detected by semi-automated 3D volumetric analysis. Portal venous phase multi-detector com-puted tomography. The preoperative CEA was 64 μg/L. Resection of segment II and III, local resec-tion in segment VI and radiofrequency ablation of a tumor in segment VIII was performed in the same procedure. 1 month after radiofrequency ablation coagulation volume was 26 mL, and CEA was 3 μg/L.

25

Table 1

Patient characteristics

Patientno.

Tumor diameter(cm)

Largest coagulationdiameter(cm)

Effective coagulationdiameter(cm)

Theoreticaltreatmentmargin (cm)

Localtumorrecurrence

Time to local tumorrecurrence(months)

1 2.3 6.7 4.2 1.0 Yes 212 2.5 4.8 3.0 0.3 Yes 63 1.2 2.9 2.1 0.5 No -4 1.6 3.9 2.7 0.6 No -5 1.0 4.2 2.8 0.9 Yes 126 1.8 3.9 2.5 0.4 Yes 67 3.5 4.3 2.7 -0.4 Yes 128 2.5 7.5 4.2 0.9 No -9 3.1 4.6 3.6 0.3 No -

10 2.3 5.9 1.6 -0.3 Yes 9 11 2.1 4.2 2.1 0.0 Yes 9

26

Table 2

Post-ablation coagulation volumes in patients with local tumor progression (n=7).

Post-ablation coagulation volume(mL)

Patient no. Baselinevolume at 1 month

3 months prior to diagnosis

Atdiagnosis

3 months afterdiagnosis

1 43 19 27 802 34 30 47 545 26 7 7 66 22 10 93 -7 22 11 26 -

10 13 2 10 41 11 22 9 14 17