Ann Surg 2008 Yekebas

8/10/2019 Ann Surg 2008 Yekebas

1/10

ORIGINALARTICLES

En Bloc Vascular Resection for Locally Advanced PancreaticMalignancies Infiltrating Major Blood Vessels

Perioperative Outcome and Long-term Survival in 136 Patients

Emre F. Yekebas, MD,* Dean Bogoevski, MD,* Guellue Cataldegirmen, MD,* Christina Kunze, MD,*

Andreas Marx, MD, Yogesh K. Vashist, MD,* Paulus G. Schurr, MD,* Lena Liebl, MD,*

Sabrina Thieltges, MD,* Karim A. Gawad, MD,* Claus Schneider, MD,* and Jakob R. Izbicki, MD*

Background:To assess in-hospital complication rates and survival

duration after en bloc vascular resection (VR) for infiltration of

pancreatic malignancies in major vessels.Methods: Between 1994 and 2005, 585 patients underwent poten-

tially curative pancreatic resection without adjuvant chemotherapy.

Four hundred forty-nine patients (77%) underwent standard onco-

logic resection (VR), whereas 136 (23%) received VR (VR). For

calculation of in-hospital morbidity and mortality rates, all 136

patients who underwent VR were considered. In contrast, for sur-

vival analysis, only pancreatic adenocarcinoma patients (n 100)

were included.

Results:One hundred twenty-eight VR patients underwent portal

or superior mesenteric vein resection and 13 hepatic artery (HA) or

superior mesenteric artery (SMA) resection. In 5 patients, synchro-

nous VR addressing both the mesenterico-portal axis and either the

HA or SMA was performed. In-hospital morbidity and mortalityrates of VR patients (39.7%/4.0%) nearly equaled that of VR

patients (40.3%/3.7%). From the 100 patients with pancreatic ade-

nocarcinoma, histopathology confirmed true vascular invasion in

77 patients. Twenty-three patients had peritumoral inflammation,

mimicking tumor invasion. Median survival was 15 months (11.2

18.8) in patients with histopathologic proven vascular invasion and

16 months (14.017.9) in those without (P 0.86). Two-year

survival probabilities were 36% (without) versus 34% (with vascular

invasion;P 0.9). Among VR patients with histopathologically

evidenced vascular invasion, 19 survived longer than 30 months,

and 6 were still alive 5 years after surgery. Multivariate modeling

identified nodal involvement (N1) and poor grading (G3) as the only

predictors of decreased survival. Evidence of vascular invasion had

no adverse impact on survival.

Conclusion: Postoperative morbidity and mortality rates after en

bloc VR are comparable with standard pancreatectomy proce-

dures. Median survival of 15 months in patients with vascular invasion

is superior to that of patients who undergo palliative therapy and nearly

equals that of patients who are not in need for VR.

(Ann Surg2008;247: 300309)

Although early detection of pancreatic malignancies is theonly theoretical chance for cure,1 the majority of pancre-atic cancers is diagnosed at an advanced or even incurablestage. Narrowing or encasement of the celiac axis, superiormesenteric, or splenic vessels visualized either by state-of-the-art computed tomography (CT) scanning2 or intraopera-tively are considered as signs of locally advanced tumors,

albeit differential diagnosis of true vascular tumor infiltrationand peritumoral inflammation is often difficult.

Until the early 1990s, the presence of portomesentericinvasion was generally accepted as a contraindication forcurative surgery. However, institutional experiences gainedin several high-volume pancreatic centers, which evidencedthat advanced pancreatic carcinoma infiltrating majorperipancreatic vessels can be safely treated by right-sided,left-sided, subtotal or total pancreatectomy with en blocresection of infiltrated major vascular structures led to in-creasingly aggressive surgical approaches. This attitude wassupported by the reported prognosis of these patients thatseems to resemble that in patients without resection of major

vascular structures.37

Therefore, the apodictic assumption that involvementof mesenteric, portal, and splenic veins indicates irresectabil-ity is with regard to technical aspects and prognosis at leasthighly debatable, if not even no more sustainable. A morecontroversial area concerns an infiltration of the hepaticartery (HA), the superior mesenteric artery (SMA) and of theceliac trunk. In patients undergoing distal pancreatectomywith en bloc celiac axis resection for advanced cancer of thepancreatic body, a Japanese group has most recently reporteda R0 resection rate of 91% (21/23), a mortality rate of 0%,and an estimated 5-year survival probability of 42%.8 Insti-tutional experiences with en bloc vascular resection (VR) of

From the *Department of General, Visceral and Thoracic Surgery, andInstitute of Pathology, University Medical Centre Hamburg-Eppendorf,University of Hamburg, Hamburg, Germany.

Reprints: E. F. Yekebas, MD, Department of Surgery, University Hospi-tal Hamburg-Eppendorf, Martinistrasse 52, 20246 Hamburg, Ger-many. E-mail: [email protected].

Copyright 2008 by Lippincott Williams & WilkinsISSN: 0003-4932/08/24702-0300DOI: 10.1097/SLA.0b013e31815aab22

Annals of Surgery Volume 247, Number 2, February 2008300


2/10

even more than one vessel in selected patients, as reported inthe presented series, have not yet been published in thesurgical literature.

Matters of debate are as follows: (1) Is vascular infiltrationalways a sign of advanced (and therefore unresectable) cancer oris it a consequence of tumor location? (2) Are perioperative

morbidity and mortality rates comparable between patients withand without VR? (3) Does the widespread assumption thatoncologic overall prognosis is worse in patients with vascularinvolvement withstand an accurate analysis of patients?

Therefore, we conducted an analysis based on a cohort of585 curatively operated patients with histopathologically provenpancreatic, ampullary or distal bile duct malignancies who werespared from adjuvant chemotherapy. Among these patients, 136underwent extended surgical procedures with en bloc VR. Forcomparison of perioperative morbidity and mortality rates, theentire cohort of patients undergoing VR, irrespective of whetherhistopathology confirmed true vascular invasion was included.In contrast, the analysis of oncologic long-term outcome was

limited to patients with ductal adenocarcinoma.

PATIENTS AND METHODS

Patient Characteristics, Inclusion CriteriaA total of 585 consecutive patients with pancreatic,

ampullary and distal common bile duct (CBD) cancer who

underwent potentially curative resection at our institutionfrom April 1994 to July 2005 were analyzed. Patients thatreceived postoperative adjuvant chemotherapy under con-trolled study conditions (until 2004) or in the routine clinicalsetting (after 2004) were excluded. The ethic committee ofthe chamber of physicians in Hamburg approved this pro-

spective study. Informed consent was obtained from allpatients before including them in a prospective database.

The age of patients ranged from 32 to 90 years, with amedian of 61 years. There was a slight predominance of malepatients (Table 1). Patients with macroscopically incompleteresection (R2-status) were excluded from analysis. Depend-ing on the location of tumors, patients were subjected toclassic pancreatico-duodenectomy (c-PD), pylorus-preserv-ing pancreatico-duodenectomy (pp-PD), distal spleno-pan-createctomy (DSP), subtotal pancreatectomy (st-P), and totalpancreatico-duodenectomy (t-PD). Guide-mark stitches ofthe resected vessels after the removal of the specimen enabledto examine whether or not intraoperatively suspected vascular

tumor invasion could be confirmed by histopathology and, inturn, to calculate the rate of cases in which peritumoralinflammatory changes mimicked vascular involvement. Tu-mor stage and grade were classified according to the sixthedition of the tumor-node-metastasis classification of theInternational Union against Cancer.9

TABLE 1. Characteristics of the Study Population (n 585)

Total Patients(n 585)

Without VR(n 449)

With VR(n 136) P

Gender (M/F) 349/236 264/185 85/51 0.441

Age 61 (3290) 61 (3282) 62 (3390) 0.949Type of operation

c-PD (Whipple) 277 (47%) 215 (48%) 62 (46%)

pp-PD (Whipple) 174 (30%) 144 (32%) 30 (22%)

DSP 55 (9%) 45 (10%) 10 (7%)

St-P 41 (7%) 23 (5%) 18 (13%)

t-PD 38 (6%) 22 (5%) 16 (12%)

Resection margins 0.963

R0 502 (85.8%) 383 (85.3%) 119 (87.5%)

R1 83 (14.2%) 66 (14.7%) 17 (12.5%)*

Operation time 360 (220520) 350 (220520) 360 (250500) 0.23

Surgical morbidity 0.719

Hemorrhage 31 (5.3%) 25 (5.6%) 6 (4.4%)

Pancreatic fistula 45 (7.7%) 36 (8.0%) 9 (6.6%)

Bile leak 29 (5.0%) 20 (4.5%) 9 (6.6%)

PV thrombosis 5 (0.8%) 3 (0.7%) 2 (1.5%)

Arterial thrombosis 2 (0.3%) 1 (0.2%) 1 (0.7%)

Sepsis 26 (4.4%) 19 (4.2%) 7 (5.1%)

Medical complications 95 (16.2%) 74 (16.5%) 21 (15.4%)

No complications 352 (60.2%) 271 (60.4%) 81 (59.6%)

Periop. mortality 23 (3.9%) 18 (4.0%) 5 (3.7%) 0.895

*In only one VRpatient, histopathology showed microscopic tumor invasion extending to the resected vessel margins,whereas in all other patients, R1 status involved the retroperitoneal, eg SMA resection margin.

Arterial thrombosis involved the main hepatic artery in one patient without VR. In a second patient, thrombosis of avenous graft replacing the SMA occurred.

pp indicates pylorus-preserving; DSP, distal spleno-pancreatectomy; st-P, subtotal pancreatectomy; t-PD, total pancre-atico-duodenectomy; PV, portal vein.

Annals of Surgery Volume 247, Number 2, February 2008 En Bloc Vascular Resection in Pancreatic Surgery

2008 Lippincott Williams & Wilkins 301


3/10

Preoperative evaluation included abdominal ultra-sonography (US) and contrast computed tomography (CT).Helical CT with arterial, pancreatic, and hepatic phases wasroutinely done. Endoscopic retrograde cholangiopancreatog-raphy, and endoscopic ultrasound were in individual patientsperformed. Distant metastases, celiac trunk infiltration, or

thrombosis of the mesenterico-portal axis with evident cav-ernomatous transformation evidenced either by preoperativeimaging or intraoperatively were considered contraindicationsfor curative surgery. In contrast, narrowing of the mesenterico-portal venous axis without cavernomatous thrombosis of theportal vein and encasement of the hepatic and mesenteric arter-ies were accepted an indication for explorative laparotomy.

Surgical Technique

Type of Pancreatic Resection and ResectabilityCriteria

After abdominal exploration, patients with distant me-tastases, peritoneal dissemination, arterial infiltration of theceliac trunk, mesenteric root encasement, in either case his-topathologically proven by frozen sections, underwent pallia-tive procedures. All other patients were considered poten-tially eligible for resection. c-PD and pp-PD represented thestandard procedures for masses located right to the mesen-terico-portal axis, whereas in those left to the SMV, DSP wasthe procedure of choice. Pancreatic body malignancies wereusually either treated by st-P (extended left pancreatectomy)or by TP. Bile duct and pancreatic transsection margins wereroutinely checked by frozen section. Standard lymphadenec-tomy as described elsewhere was performed.10,11 Irrespectiveof the type of resection (c-PD, pp-PD, DSP, st-P, TP),mobilization of the specimen was performed before VR,

hereby resulting in en bloc resection of the specimen includ-ing the involved vessel as the last step of the operation.Technical and general eligibility criteria for en bloc resectionwere as follows: (1) presumable achievement of R0-status;(2) tumor infiltration proximal to the peripheral branching ofthe SMV and SMA; (3) tumor respecting the celiac trunk; (4)no evidence for hypercoagulopathy syndromes.

Technique of VRWhen tumors infiltrated the mesenterico-portal axis, the

vessel was freed from its surrounding tissue proximally anddistally to ensure sufficient vascular control. For SMV le-sions, the proximal vascular clamp was placed distal to the

venous confluence. In case of tumors invading the lateralaspect of the lower PV, proximal clamps were usually placedat an angle to the spleno-portal confluence without sacrificingthe splenic vein. For tumor infiltration involving the mid orhigh PV, the splenic vein was consistently divided to achievemobility. Whether tangential resection of the lateral SMV/PVor segmental sleeve resection was performed depended ontumor location. Tangential resection in case of attachment ofthe tumor to the right-sided SMV/PV was usually recon-structed by simple venous suture. Rarely, when more thanone-third of the lateral wall had to be resected, autologousvenous patches (internal jugular, saphenous, inferior mesen-teric vein) were performed to avoid venous narrowing. Vas-

cular infiltration of more than half of the venous circumfer-ence was consistently treated by segmental sleeve resection.The preferred reconstruction technique after segmental resec-tion was primary end-to-end anastomosis. Tension-free anas-tomosis was facilitated by mobilization of the mesentericroot. Occasionally, to achieve additional mobility of the

proximal stump, the hepatic ligaments were transsected. Onlywhen despite both maneuvers tension-free anastomosis weretechnically not feasible, venous continuity was restored byinterposition of autologous veins (see above). The mostrecently published technique of using the left renal vein as anautologous conduit to restore the mesenterico-portal continu-ity12 that, in contrast to harvesting the internal jugular orsaphenous veins, protects patients from additional cervical oringuinal surgery was not applied in any of the reportedpatients. When the venous confluence had to be resected, thedecision on whether or not the splenic vein stump wasreinserted was based on the presence or absence of adequatecollateralization of the splenic hilum via the short gastric

veins (Fig. 1).In 5 among 13 patients undergoing arterial sleeveresection, combined VR procedures addressing both the mes-enterico-portal axis and the hepatic or mesenteric artery wereperformed (Table 2). Such extended procedures were re-stricted to patients in good preoperative, clinical condition(ASA III). When tension-free reconstruction of the arterialcontinuity by end-to-end anastomosis was not feasible, anautologous vein interposition using the internal jugular, sa-phenous, or inferior mesenteric vein was performed. Periop-erative, systemic anticoagulation intending partial thrombo-plastin time of 60 to 70 s was only performed in patients whounderwent arterial resection. In all other patients, conven-

tional, prophylactic heparinization by low molecular weightheparin was performed.

In-Hospital ParametersThe following parameters were routinely assessed, on-

line included in a prospective data base, and analyzed in thisstudy: perioperative morbidity, especially surgical complica-tions, eg, occurrence of postpancreatectomy hemorrhage,pancreatic, and biliary fistula, thrombosis of the PV, SMV,SMA, and HA in patients undergoing VR, and of sepsis/intra-abdominal abscess formation. Perioperative mortality wasdefined as in-hospital mortality and deaths within the firstmonth after discharge of patients.

Follow-UpFollow-up (median: 14 months; range 3139 months)

was either performed by interviewing the patients generalpractitioners, in our institution on an outpatient basis, or,when no other information could be provided, from theregional Cancer Registry. These evaluations included regu-larly scheduled physical examinations, imaging tests andstudies of tumor markers (carcinoembryonic antigen and CA19-9). Events considered were death, local recurrence, anddistant metastasis. When no events were recorded, the pa-tients were censored at the last contact with them.

Yekebas et al Annals of Surgery Volume 247, Number 2, February 2008

2008 Lippincott Williams & Wilkins302


4/10

Statistical AnalysisAssociations between categoric and continuous vari-

ables at surgery were assessed by Fisher exact test andWilcoxon tests, where appropriate. The Kaplan-Meier

method was used to estimate survival probability 24 and 60months after surgery. Differences between patient groupswith respect to their survival were assessed by log-rank tests,considering differences to be statistically significant at a Pvalue of 0.05. Multivariate modeling including variableswith aPvalue 0.05 in univariate log-rank test was fit to thesurvival data using Cox proportional hazards methods. In thiscase, significance statements refer toPvalues of 2-tailed testswith a Pvalue 0.05. Those 23 pancreatic adenocarcinomapatients undergoing VR in whom histopathology onlyshowed tumor-mimicking involvement of vessels were, re-garding their perioperative course, pooled with patients whohad histologically proven true vascular tumor invasion (Table

1), whereas for long-term survival analysis, VR patientswithout histopathologically evidenced vascular infiltrationwere shifted in the subset of VRpatients without vascularinvolvement. To avoid potential bias because of different

histopathologic diagnosis (ductal adenocarcinoma, ampul-lary, ie, duodenal and papillary carcinoma, bile duct car-cinoma), comparison of survival between VR and VRpatients only considered patients with true pancreatic duc-tal adenocarcinoma.

RESULTS

Procedures and Perioperative PatientCharacteristics

Table 1 shows patient characteristics, surgical proce-dures, and perioperative outcome in the entire study cohortirrespective of the tumor entity (ductal, ampullary, distal

FIGURE 1.Reconstruction after enbloc resection of the portal vein orsuperior mesenteric vein. Sixty-fivepatients underwent tangential VR (A).

Whether portal blood flow was re-stored by simple suture (n 57) orvenous patch depended on the ex-tent of vessel wall resection. Segmen-tal vessel resection was performed in63 patients with different sites of

anastomosis. In the majority of thesepatients, the splenic vein could bemaintained (B,C), whereas in only 12,it had to be killed to achieve sufficientmobility (D,E). Five patients in whoma tension-free anastomosis could notbe performed required venous graftinterposition (F).




5/10

CBD carcinoma) and of whether or not true vascular invasionwas evidenced histopathologically. c-PD was performed in277 patients, pp-PD in 174 patients, DSP in 55 patients, st-Pin 41 patients, and t-PD in 38 patients (Table 1). Overall,these procedures were performed in 449 patients without

vascular resection (VR), whereas 136 patients underwenten bloc VR. Resection margins were tumor-free (R0) in 85%of VR patients and in 87.5% of VR patients (P 0.963),resulting, in the entire study cohort, in R0-status in 86%, andin R1-status in 14% of patients. With respect to operationtimes and median volume of intraoperatively transfusedblood units that accounted for a median of 3 units in bothgroups, no significant differences between VR and VRpatients were detected. Postoperative hemorrhage and pan-creatic fistula represented the most frequent surgical compli-cations, overall accounting for 5.3% and 7.7%, respectively.Portal vein thrombosis occurring in 3 VR patients and in 2VR patients (0.7% and 1.5%, respectively) was lethal in 2

patients (1 VR, 1 VR). In 2 patients, arterial thrombosiscomplicated the postoperative course: one patient developingHA thrombosis not related to VR after TP for intraductalpapillary mucinous neoplasia (IPMN) with invasive carci-noma had, as would be expected, long-lasting biliary com-plications, but survived. In the second patient who underwentsynchronous en bloc Whipple procedure with SMA and SMVresection, thrombosis of the venous graft replacing the SMAdue to pancreatic fistula with subsequent intraabdominal infec-tion was lethal (Table 2). In the whole cohort of 585 patients, 22(4%) died perioperatively, ie, during their hospital stay or within30 days after discharge. Overall, morbidity and mortality rates ofVR and VR patients were comparable.

Type of VR and ReconstructionAmong the 136 patients who underwent VR, 128 were

subjected to a resection of the mesenterico-portal axis. In 13patients a resection of the HA or SMA was performed with anoverlap of 5 patients undergoing synchronous VR of the PV

or SMV. Distribution of portal venous reconstruction tech-niques, depending on the tumor location, is shown in Figure1. Tangential resection of either the SMV or the PV wasperformed in 65 patients among which in 57, simple venoussuture was sufficient to restore portal blood flow, whereas 8underwent a venous patch (Fig. 1A). Sixty-three patientswere subjected to segmental mesenterico-portal sleeve resec-tion. In the majority of the latter (n 58), the mesenterico-portal axis was reconstructed by end-to-end anastomosis (Fig.1BE). Only 5 patients required venous graft interposition(Fig. 1F). Overall, in 10 from 17 patients that underwent atranssection of the splenic vein, a reinsertion of the venousstump was performed, hereby creating a neo-confluence

(Fig. 1D, E).In 13 patients, tumor invasion involved the HA (n 10)or the SMA (n 3), respectively (Table 2). Among the 5patients undergoing simultaneous VR of the mesenterico-portalaxis, 3 were subjected to t-PD and 2 to c-PD. In 11 patients,end-to-end anastomosis was technically feasible, whereas 2 withVR of the SMA required autologous vein interposition. Furtheranalysis of this subset showed that all patients surviving longerthan 2 years had been staged as pN0.

HistopathologyIn 482 of 585 patients (82%), histopathology confirmed

pancreatic ductal adenocarcinoma, whereas the remainder

TABLE 2. En Bloc Arterial Resection With or Without Mesenterico-Portal Resection

Patient No.

Resected

Vessels

Arterial

Reconstruction

Pancreatic

Resection

PV/SMV-

Histology

HA/SMA-

Histology

N

Status

Perioperative

Outcome

Survival

(months)

Synchronous VR(PV/SMVHA/SMA)

1 HASMV E/E c-PD Positive Positive pN1 Survived 18

2 HAPV E/E t-PD Positive Positive pN0 Survived 25


4 SMASMV Venous graft c-PD Positive Negative pN0 Died


Arterial resection(HA/SMA)

6 HA E/E t-PD Positive pN1 Survived 9

7 HA E/E t-PD Negative pN0 Survived Alive (39 mo)

8 HA E/E st-P Positive pN0 Survived Alive (32 mo)



11 HA E/E c-PD Positive pN1 Survived 15

12 SMA Venous graft t-PD Positive pN0 Survived Alive (29 mo)

13 SMA E/E st-P Positive pN1 Survived 10

Actual survival of patients with pancreatic adenocarcinoma who had true vascular invasion in main arterial branches confirmed by histopathology. Five among 13 patientsundergoing resection of the HA or SMA had simultaneous VR of the SMV or PV (Patient No. 1-5). All patients who survived longer than 2 years (no. 2, no. 7, no. 8, no. 12) werestaged as pN0. Regarding the latter, histopathology confirmed in 3 patients tumor invasion in the resected artery.

PV indicates portal vein; SMV, superior mesenteric vein; HA, hepatic artery; SMA, superior mesenteric artery; E/E, end-to-end anastomosis; c-PD, classical pancreatico-duodenectomy; t-PD, total pancreatico-duodenectomy; st-P, subtotal pancreatectomy.




6/10

had ampullary (duodenal, papillary; other, eg, NET) (n 54)or distal bile duct carcinoma (n 49). To avoid statisticalbias caused by the different biologic characteristics of amp-

ullary and distal CBD carcinoma, as compared with pancre-atic carcinoma, only patients with true pancreatic adenocar-cinoma were subjected to survival analysis (Fig. 2).

Among the 482 patients with ductal adenocarcinoma,100 (21%) underwent VR. The analysis of the VR subsetrevealed histopathologic evidence of vascular invasion in 77patients. In contrast, 23 patients had only tumor-mimickinglesions without histopathologic proven vascular involvement.For survival analysis, these 23 VR patients without vascu-lar invasion in histopathology were pooled with VR pa-tients. This resulted in 405 patients without and 77 patientswith vascular invasion representing the data base for furthersurvival analysis (Table 3).

Comparison of tumor characteristics showed slight dif-ferences of borderline significance regarding the primarytumor stage (T-stage). The higher rate of negative resectionmargins in patients with histopathologically evidenced vas-cular invasion (90%) as compared with those without (82%)did not reach statistical significance (P 0.1). In the entiresubset of VR patients, only one (distal CBD carcinoma)was identified in whom microscopic involvement (R1) con-cerned the resection margin of the resected vessel. In all otherVR patients with histopathologically evidenced R1 status,tumors extended to the retroperitoneal resection margin,whereas en bloc resected vessels were found to be tumor-free.Only minimal differences regarding nodal status and tumor

grading between patients with and without histopathologicvascular invasion were assessed (Table 3).

SurvivalMedian follow-up times were 13 months (range, 4123)

in VR patients with histopathologically evidenced vascularinvasion, 15 months (range, 4112) in VR patients withoutvascular invasion, and 14 months (range, 4139) in VRpatients, respectively. Overall, median survival of patients withhistopathologically confirmed vascular invasion was 15 months(95% confidence interval CI; 11.218.8 months), whereas thatof patients without vascular involvement was 16 months (95%CI; 14.017.9 months, P 0.856). In those 23 VR patientswithout histologically proven infiltration, median survival was23 months (95% CI, 1432).

Overall 2- and 5-year survival probabilities accountedfor 35.9% and 17% for patients without, and for 33.7% and14.6% for patients with vascular invasion (P 0.9). In thesubanalysis of the VR group, 2-, and 5-year survival prob-abilities of patients in whom histopathology did not confirmtrue vascular invasion were 41.1% and 24.2%, respectively.

In the entire cohort of 77 patients with vascular invasion,19 survived more than 30 months. Six patients were still alive 5years after surgery and one patient survived more than 10 years.In the arterial en bloc resection group (n 13) involving eitherthe HA or SMA, 8 patients survived more than 1 year and 4 evenmore than 2 years. Five from these patients were subjected tosimultaneous VR of the SMV or PV (Table 2).

FIGURE 2. Overall survival of patients with pancreatic ductal adenocarcinoma grouped according to the VR and histologicallyproven infiltration in the blood vessels. No survival benefit was detected in VR patients compared with VR patients irre-spective of whether histopathology showed true infiltration or tumor-mimicking, inflammatory pseudo-infiltration (P 0.948andP 0.279). Of the patients with VR and histologically proven infiltration 33.7% survived more than 30 months comparedwith 30.8% for patients without VR.




7/10

Further stratification of the entire cohort of 482 patientswith ductal adenocarcinoma according to nodal involvementshowed, irrespective from vascular invasion status, signifi-cantly better median survival of 24 months (95% CI, 18.429.6) in pN0 patients compared with a median survival of 11months (95% CI, 9.612.4) in pN1 patients (P 0.0001).Additional stratification of pN0 patients according to vascular

invasion revealed that the slightly better median overallsurvival of 24 months (95% CI, 16.831.2) in patients with-out vascular involvement did not reach statistical significance(P 0.12) when compared with those with histopathologi-cally proven vascular infiltration (median survival 13 months;95% CI, 5.220.8).

The overall survival did neither differ between differentVR techniques (primary suture vs. end-to-end vs. patch vs.graft interposition, data not shown, P 0.45) nor betweenmesenterico-portal versus arterial resection (data not shown,P 0.41).

Multivariate Analysis

Multivariate modeling selecting 7 variables using astepwise regression model identified only lymph node in-volvement (pN vs. pN1) and histologic grading (G1/2 vs. G3)to be independent predictive factors for survival. Histologi-cally proven vascular infiltration, as would be expected fromunivariate survival analysis, was not found to have an adverseindependent influence on long-term outcome of patients.Also, the resection margin status did not independently affectsurvival (Table 4).

DISCUSSIONThe considerable improvement of postoperative mor-

bidity and mortality rates after major pancreatic resection for

pancreatic carcinoma in the past 2 decades has decreased thethreshold in operating upon patients with surgically challeng-ing, locally advanced tumors. The impetus of extended vas-cular en bloc resection in case of preoperatively assumed orintraoperatively assessed tumor invasion in adjacent vessels,as long as distant metastases are absent, is to achieve a

potentially curative resection. This concerns in particular theSMV and PV, whereas the role of arterial en bloc resection ofthe HA and the SMA or even of the celiac trunk itself is stillhighly controversial.

The presented series is based on an analysis of patientswho were spared from adjuvant chemotherapy. Until 2004,we restricted the use of adjuvant chemotherapy to patientswho were enrolled in controlled studies, eg, the ESPAC-IIand ESPAC-III trials, whereas most of curatively resectedpatients did not receive additional therapy. The rationale forour reluctance towards the routine use of adjuvant chemo-therapy under clinical settings was that adequately powered,randomized studies evidencing the beneficial impact of adju-

TABLE 3. Histopathology in Pancreatic Ductal Carcinoma (n 482)

Without Vascular Infiltration in Histopathology With Vascular Infiltration

in Histopathology En blocVascular Resection P(1) P(2)Standard procedures En bloc Vascular Resection Total

No. 382 23 405 77

T stage 0.095 0.057

T1 53 (14) 2 (9) 55 (14) 5 (6)

T2 122 (32) 7 (30) 129 (32) 23 (30)

T3 199 (52) 12 (52) 211 (52) 44 (57)

T4 8 (2) 2 (9) 10 (2) 5 (6)

N stage 0.283 0.231

N0 60 (21) 1 (13) 61 (15) 8 (10)

N1 322 (79) 22 (87) 344 (85) 69 (90)

Grading 0.421 0.404

G1 53 (14) 3 (13) 56 (14) 10 (13)

G2 210 (55) 12 (52) 222 (55) 37 (48)

G3 119 (31) 8 (35) 127 (31) 30 (39)

R status 0.10 0.10

R0 313 (82) 21 (91) 334 (82) 69 (90)

R1 69 (18) 2 (9) 71 (18) 8 (10)

Values in the parentheses are given in percentage.P (1): Patients without histopathologically evidenced vascular invasion (standard pancreatic resections en bloc VR with negative vascular histopathology, n 405) versus

en bloc VR with histopathologically confirmed vascular involvement (n 77).P (2): Standard pancreatic resections (n 382) versus en bloc VR with histopathologically confirmed vascular involvement (n 77).

TABLE 4. Multivariate Analysis of Variables PotentiallyPredictive of Survival After Major Pancreatic Resection

Relative Risk 95.0% CI P

Sex 0.730 0.4301.237 0.242

Age 1.001 0.9911.012 0.798

Vascular invasion 1.063 0.8601.315 0.569

pT 1 & 2 0.768 0.4471.319 0.338

pN0 2.320 1.7943.000 0.0001

R0 0.886 0.6081.292 0.531

G1/2 2.464 1.9393.132 0.0001

95% CI indicates 95% confidence interval.




8/10

vant chemotherapy were not only scarce but also inconsistent.A Japanese randomized controlled trial reported even a tendencyto worse 5-year survival probability in patients who receivedadjuvant chemotherapy compared with control patients that onlyunderwent curative surgery (11.5% vs. 18.0%, not significant).13

After 2004, when the ESPAC-I14 and CONKO-00115 trials

evidenced statistically robust survival benefits of adjuvantchemotherapy, our institutional attitude substantially changed.Although one may argue that these studies, especially theESPAC-I trial, were afflicted with some weaknesses regard-ing the study design, the use of adjuvant chemotherapy hastherefore become our institutional standard in the clinicalsetting after primary curative resection of pancreatic cancer.

Opponents of extended en bloc VR raise 2 majorcounter-arguments against its use. The first argument arguingthat morbidity is substantially elevated,7,1619 has been rebut-ted by several surgical series that evidenced comparablein-hospital morbidity and mortality rates after VR and onco-logic standard procedures.37,10,17,20 Nonetheless, the belief

in the usefulness of VR is still controversial. This is reflectedby considerable differences between experienced US centerswith respect to the rates of VR that range from 3%20 to 38%.4

In the present series, even mean operative time andintraoperative blood loss, generally reported to be elevated incase of en bloc resection,4,20,21,22 did not substantially differin patients with VR compared with those without. Also,vascular reconstruction was not associated with increasedprevalence of specific vascular complications, such as hem-orrhage and thrombosis. Overall, the analysis of in-hospitalmorbidity and mortality rates of VR patients based on atotal of 136 patients with pancreatic adenocarcinoma (n 100) and with malignancies of nonpancreatic origin (ampul-

lary/distal CBD cancer, other; n 36) were nearly identicalcompared with VR patients.The second counter-argument directed against VR is

related to the putative limited survival benefit once the tumorinvades major vessels. This attitude ignores several aspects.First, definitive assessment of tumor adherence is hardlypossible by preoperative imaging. Even intraoperatively, it isfrequently misjudged to what degree vascular involvement iscaused by peritumoral inflammatory changes or by trueinvasion. Therefore, in the individual patient, decision-mak-ing on whether or not to perform VR is often based oninstitutional or even individual experiences and discretionrather than on histopathologic evidence. In the present study,

a considerable rate of almost one-fourth (23/100) of patientswho underwent VR for pancreatic cancer was found to havetumor-mimicking lesions without proven vascular involve-ment in histopathology. Prima facie, this rate of intraopera-tive misjudgement seems to be high. On close inspection,however, it is even rather lower than that reported in the mostprevious series according to which the rate of histopathologicconfirmation of intraoperatively suspected vascular infiltra-tion ranges from 26% to 85%.4,6,7,16,20,2330

Furthermore, the notion that histopathologically evi-denced vascular tumor invasion represents per se an adversefactor, as argued by some authors,16 has been challenged oreven disproved by several studies that showed comparable

long-term survival in these patients compared with thosewithout vascular invasion.6,7 The most conclusive explana-tion for this is that the crucial factor, which determineslong-term outcome of patients with pancreatic cancer is thepresence or absence of early tumor cell dissemination todistant organs undetectable by routine imaging techniques at

the time of potentially curative surgery. In support of this, onemay argue that all attempts aiming at achieving better overalloutcome by increasing the extent of peripancreatic lymphad-enectomy eventually failed. In this context, it was the impetusof several randomized trials in the past decade to decrease therisk of local recurrence by extended lymphadenectomy pro-cedures supposed to be beneficial as a result of its undoubt-edly superior lymphatic clearance.20,3134 Better local con-trol, so was the assumption, would result in improved long-term survival. However, although these trials reported asignificantly increased lymph node yield by extended lymph-adenectomy, overall survival of patients who underwent ex-tended lymphadenectomy was nearly identical compared with

patients subjected to standard lymphadenectomy.In the light of these considerations, it was a key finding

of the presented series that the median survival of 15 monthsof patients with true pancreatic adenocarcinoma who hadhistopathologically proven vascular invasion (n 77) nearlyequaled that of 382 patients without VR. Interestingly, al-though not reaching statistical significance, VR patients inwhom histopathology showed only tumor-mimicking vascu-lar involvement (n 23) had the longest median survival of23 months. In this context, it is important to stress that survivalanalysis only addressed patients with pancreatic carcinoma.Patients with malignancies of nonpancreatic origin and thosewho underwent adjuvant chemotherapy were excluded.

It has been recently suggested that the likelihood of R1resection margin status is increased in large tumors thatrequire VR35; a finding, which is inconsistent with the datareported herein. Several studies report a poor survival asso-ciated with VR due to a high incidence of positive resectionmargins.10,16,21,29,36 Although not significant, we observed aneven lower R1 rate of 10% (true infiltration) and 9%(infiltration-mimicking vascular involvement) after VR thanthat after standard resection (18%). One may hypothesize thatfactors such as large tumor size, R1 resection margin statusand the need for VR, frequently lacking to have significant,independent importance in multivariate modeling, are asso-ciated with a higher likelihood of N1 status that is currently

regarded the most important independent variable for predic-tion of overall prognosis. In the present study, the only factor,which had, apart from nodal metastasis, an independentinfluence on survival was tumor grading (G1/2 vs. G3). Allother variables, including tumor size, resection margin statusand, in particular, histopathologic vascular invasion, had nostatistical significance when introduced in multivariate re-gression analysis.

That R1 status had even in univariate analysis noindependent adverse influence on survival directs attention tothe issue of histopathologic staging accuracy. It is an issue ofincreasing concern that the assessment of a R1 status isclosely dependant on the technique of histopathologic exam-




9/10

ination. The R1 rates of approximately 10% to 20% in thepresented series, depending on whether or not VR was per-formed, were similar to those reported in the majority ofstudies published in the surgical literature. In contrast, somerecent studies report considerably higher R1 rates of 80% andover by implementing more sophisticated procedures of axial

slicing techniques of Whipple specimens.37

This raises thequestion whether disparities in margin status reporting derivefrom different histopathologic examination techniques ratherthan from true differences in tumor staging, hereby resultingin understaging of manyif not even mostpatients withpancreatic malignancies due to the lack of standardizedguidelines for the histopathologic and reporting of specimens.

In summary, to argue that locally advanced tumors withvascular invasion may have to some extent a worse prognosisthan early tumor stages is, in the individual patient, nothelpful in decision-making. First, the final histopathologicexamination will confirm surgically suspected true vascularinvasion in the majority (77% in this series; partly far below

this rate in other studies), but not all patients with a consid-erable rate of tumor-mimicking lesions. Second, even in thosepatients in whom clinical or intraoperative suspicion is con-firmed by histopathology, overall outcome seems to be de-pendent from other variables, such as nodal involvement andtumor grading rather than from vascular tumor invasion that,in the present study, had no an independent adverse impact onsurvival. This may even be the case in selected patients inwhom VR of the PV/SMV is combined with a resection ofeither the HA or the SMA. In the presented series, suchcombined procedures were performed in 5 patients. Apartfrom one patient who died perioperatively, actual survival inthese patients rangedin the absence of any adjuvant treat-mentfrom 11 to 25 months, which is far longer than that to

be expected when therapy in these patients would have beenrestricted to palliative measures.

In conclusion, our data suggest that major pancreaticsurgery can be safely combined with en bloc VR in case ofsuspected or evidenced vascular invasion. With adequateinstitutional experience, in-hospital morbidity and mortalityrates are comparable with that of standard procedures. Theneed for vascular reconstruction does not adversely impactlong-term survival. When potentially curative resection isachieved, 2-, and 5-year survival probabilities of 35% and15% of patients with histopathologically evidenced vascularinvasion nearly equal that of patients without vascular in-volvement. Median survival after en bloc VR is far longer

than the reported survival of patients in whom therapy isrestricted to palliative surgical procedures and chemother-apy. Further trials are needed addressing whether neoadju-vant treatment options, especially radiotherapy, may down-size locally advanced pancreatic tumors suspicious ofvascular invasion.

REFERENCES1. Ariyama J, Suyama M, Ogawa K, et al. The detection and prognosis of

small pancreatic carcinoma. Int J Pancreatol. 1990;7:3747.2. Bluemke DA, Fishman EK. CT and MR evaluation of pancreatic cancer.

Surg Oncol Clin N Am. 1998;7:103124.3. Yoshimi F, Asato Y, Tanaka R, et al. Reconstruction of the portal vein

and the splenic vein in pancreaticoduodenectomy for pancreatic cancer.Hepatogastroenterology. 2003;50:856860.

4. Tseng JF, Raut CP, Lee JE, et al. Pancreaticoduodenectomy withvascular resection: margin status and survival duration. J GastrointestSurg. 2004;8:935949; discussion 949950.

5. Koniaris LG, Staveley-OCarroll KF, Zeh HJ, et al. Pancreaticoduodenectomyin the presence of superior mesenteric venous obstruction. J GastrointestSurg. 2005;9:915921.

6. Harrison LE, Klimstra DS, Brennan MF. Isolated portal vein involve-ment in pancreatic adenocarcinoma. A contraindication for resection?Ann Surg. 1996;224:342347; discussion 347349.

7. Leach SD, Lee JE, Charnsangavej C, et al. Survival following pancre-aticoduodenectomy with resection of the superior mesenteric-portal veinconfluence for adenocarcinoma of the pancreatic head. Br J Surg.1998;85:611617.

8. Hirano S, Kondo S, Hara T, et al. Distal pancreatectomy with en blocceliac axis resection (DP-CAR) for locally advanced pancreatic bodycancer: long-term results. Ann Surg. 2007;246:4651.

9. Sobin LH, Wittekind Ch.TNM Classification of Malignant Tumours. 6thed. New York: Wiley; 2002.

10. van Geenen RC, ten Kate FJ, de Wit LT, et al. Segmental resection andwedge excision of the portal or superior mesenteric vein during pancre-atoduodenectomy.Surgery. 2001;129:158 163.

11. Yeo CJ, Cameron JL, Lillemoe KD, et al. Pancreaticoduodenectomywith or without distal gastrectomy and extended retroperitoneal lymph-adenectomy for periampullary adenocarcinoma, Part 2: Randomizedcontrolled trial evaluating survival, morbidity, and mortality. Ann Surg.2002;236:355366; discussion 366368.

12. Smoot RL, Christein JD, Farnell MB. An innovative option for venousreconstruction after pancreaticoduodenectomy: the left renal vein.J Gastrointest Surg. 2007;11:425 431.

13. Takada T, Amano H, Yasuda H, et al. Is postoperative adjuvant chemo-therapy useful for gallbladder carcinoma? A phase III multicenterprospective randomized controlled trial in patients with resected pan-creaticobiliary carcinoma. Cancer. 2002;95:16851695.

14. Neoptolemos JP, Stocken DD, Friess H, et al. A randomized trial ofchemoradiotherapy and chemotherapy after resection of pancreatic can-cer. N Engl J Med. 2004;350:12001210.

15. Oettle H, Post S, Neuhaus P, et al. Adjuvant chemotherapy withgemcitabine vs. observation in patients undergoing curative-intent re-section of pancreatic cancer: a randomized controlled trial. JAMA.2007;297:267277.

16. Allema JH, Reinders ME, van Gulik TM, et al. Portal vein resection inpatients undergoing pancreatoduodenectomy for carcinoma of the pan-creatic head. Br J Surg. 1994;81:16421646.

17. Fuhrman GM, Leach SD, Staley CA, et al. Rationale for en bloc veinresection in the treatment of pancreatic adenocarcinoma adherent to thesuperior mesenteric-portal vein confluence. Pancreatic Tumor StudyGroup.Ann Surg. 1996;223:154 162.

18. Harrison LE, Brennan MF. Portal vein resection for pancreatic adeno-carcinoma.Surg Oncol Clin N Am. 1998;7:165181.

19. Sindelar WF. Clinical experience with regional pancreatectomy foradenocarcinoma of the pancreas. Arch Surg. 1989;124:127132.

20. Bachellier P, Nakano H, Oussoultzoglou PD, et al. Is pancreaticoduo-denectomy with mesentericoportal venous resection safe and worth-while?Am J Surg. 2001;182:120 129.

21. Capussotti L, Massucco P, Ribero D, et al. Extended lymphadenectomyand vein resection for pancreatic head cancer: outcomes and implica-tions for therapy. Arch Surg. 2003;138:13161322.

22. Howard TJ, Villanustre N, Moore SA, et al. Efficacy of venous recon-struction in patients with adenocarcinoma of the pancreatic head.J Gastrointest Surg. 2003;7:10891095.

23. Baulieux J, Adham M, Oussoultzoglou E, et al. Is pancreatectomy withresection of the retro-pancreatic vessels for cancer justified?.Chirurgie.1998;123:438444.

24. Tashiro S, Uchino R, Hiraoka T, et al. Surgical indication and signifi-cance of portal vein resection in biliary and pancreatic cancer. Surgery.1991;109:481487.

25. Ishikawa O, Ohigashi H, Imaoka S, et al. Preoperative indications forextended pancreatectomy for locally advanced pancreas cancer involv-ing the portal vein. Ann Surg. 1992;215:231236.




10/10

26. Launois B, Franci J, Bardaxoglou E, et al. Total pancreatectomy for

ductal adenocarcinoma of the pancreas with special reference to resec-

tion of the portal vein and multicentric cancer. World J Surg. 1993;17:

122126; discussion 126127.

27. Takahashi S, Ogata Y, Tsuzuki T. Combined resection of the pancreas

and portal vein for pancreatic cancer. Br J Surg. 1994;81:11901193.

28. Nakao A, Harada A, Nonami T, et al. Clinical significance of portal

invasion by pancreatic head carcinoma. Surgery. 1995;117:5055.29. Roder JD, Stein HJ, Siewert JR. Carcinoma of the periampullary region:

who benefits from portal vein resection? Am J Surg. 1996;171:170174;

discussion 174175.

30. Poon RT, Fan ST, Lo CM, et al. Pancreaticoduodenectomy with en bloc

portal vein resection for pancreatic carcinoma with suspected portal vein

involvement.World J Surg. 2004;28:602608.

31. Pedrazzoli S, DiCarlo V, Dionigi R, et al. Standard versus extended

lymphadenectomy associated with pancreatoduodenectomy in the sur-

gical treatment of adenocarcinoma of the head of the pancreas: a

multicenter, prospective, randomized study. Lymphadenectomy Study

Group.Ann Surg. 1998;228:508 517.

32. Nguyen TC, Sohn TA, Cameron JL, et al. Standard vs. radical pancre-aticoduodenectomy for periampullary adenocarcinoma: a prospective,randomized trial evaluating quality of life in pancreaticoduodenectomysurvivors. J Gastrointest Surg. 2003;7:19; discussion 911.

33. Riall TS, Cameron JL, Lillemoe KD, et al. Pancreaticoduodenectomy withor without distal gastrectomy and extended retroperitoneal lymphadenec-tomy for periampullary adenocarcinoma, Part 3: Update on 5-year survival.J Gastrointest Surg. 2005;9:11911204; discussion 12041206.

34. Farnell MB, Pearson RK, Sarr MG, et al. A prospective randomized trialcomparing standard pancreatoduodenectomy with pancreatoduodenec-tomy with extended lymphadenectomy in resectable pancreatic headadenocarcinoma.Surgery. 2005;138:618628; discussion 628630.

35. Raut CP, Tseng J, Sun C, et al. Impact of resection status on pattern offailure and survival after pancreatoduodenectomy for pancreatic adeno-carcinoma.Ann Surg. 2007;246:5260.

36. Launois B, Stasik C, Bardaxoglou E, et al. Who benefits from portal veinresection during pancreaticoduodenectomy for pancreatic cancer?WorldJ Surg. 1999;23:926929.

37. Verbeke CS, Leitch D, Menon KV, et al. Redefining the R1 resection inpancreatic cancer.Br J Surg. 2006;93:12321237.



Ann Surg 2008 Yekebas

Documents

Transcript of Ann Surg 2008 Yekebas