Ultrasound in Med. & Biol., Vol. 35, No. 10, pp. 1596–1600, 2009Copyright � 2009 World Federation for Ultrasound in Medicine & Biology

Printed in the USA. All rights reserved0301-5629/09/$–see front matter

asmedbio.2009.04.022

doi:10.1016/j.ultr

d Original Contribution

VASCULARITY INDEX OF LARYNGEAL CANCER DERIVED FROM 3-DULTRASOUND: A PREDICTING FACTOR FOR THE IN VIVO ASSESSMENT

OF CERVICAL LYMPH NODE STATUS

JUN ZHOU,* SHANG-YONG ZHU,y RUO-CHUAN LIU,y FENG LUO,z and DE-XI SHU**First College of Clinical Medical Science, China Three Gorges University; yFirst Affiliated Hospital of Guangxi Medical

University and zFirst Affiliated Hospital of Guangxi Traditional Chinese Medical College, Nanning, Guangxi, China

(Received 13 September 2008, revised 20 April 2009, in final form 22 April 2009)

AUltrasoGorgesE-mail

Abstract—To demonstrate whether a calculated vascularity index (VI) can predict metastases of cervical lymphnodes, the VI values of the primary tumors were obtained by using 3-D sonography in 87 subjects with laryngealcancer confirmed by laryngoscope and biopsy. N-staging of the subjects was determined by pathological nodalharvesting. The relationship between the VI and pathological N-staging was evaluated by correlation coefficient.To test the accuracy of the VI for predicting cervical lymph node involvement, a receiver operating characteristic(ROC) curve was constructed, and the best operating point was determined by Youden’s index. For comparison, 2-Dsonography was applied to detect metastatic cervical lymph nodes. The accuracy, sensitivity and specificity of theVI, 2-D sonography and a combination of the two methods for diagnosis of metastatic cervical lymph nodes werecompared. There was a positive linear correlation between the VI and pN-staging (r 5 0.740, p , 0.001). Thearea under the ROC curve for the VI was 0.919. The best operating point of the VI was 4.4565, which derivedhigher sensitivity than that of 2-D sonography (95% vs. 81%, p 5 0.031), but lower specificity (75% vs. 95%,p 5 0.012). The combination of the two methods yielded a higher accuracy (97% vs. 85% and 89%, p 5 0.002and 0.016), a higher sensitivity to 2-D sonography (95% vs. 81%, p 5 0.031) and a higher specificity to VI(98% vs. 75%, p 5 0.002). The VI of laryngeal cancer can be a useful factor for predicting metastases of cervicallymph nodes. (E-mail: zjsts8@163.com) � 2009 World Federation for Ultrasound in Medicine & Biology.

Key Words: Laryngeal cancer, Cervical lymph node, Metastasis, Three-dimensional sonography, Vascularity index.

INTRODUCTION AND LITERATURE

Some studies have demonstrated that sonography is a very

useful method for differential diagnosis of the cervical

lymph nodes (Takeuchi et al. 1999; Yonetsu et al. 2001;

Ahuja and Ying 2003). However, the ability of 2-D sonog-

raphy to detect retropharyngeal lymph nodes and those

smaller than 5.0 mm is still controversial (Don et al.

1995; van den Brekel et al. 1990). A more accurate

approach needs to be developed.

As well, tumor angiogenesis plays an important role

in the pathogenesis of tumor growth and metastases (Folk-

man 1992). Some studies based on sectioned slides have

indicated that tumor angiogenesis is a predictive marker

for nodal status in laryngeal cancer (Murray et al. 1997;

ddress correspondence to: Jun Zhou, Department of Diagnosticund, the First College of Clinical Medical Science, China ThreeUniversity, 183 Yi-Ling Road, Yichang, Hubei 443003, China.

: zjsts8@163.com

1596

Kupisz et al. 1999; Teknos et al. 2002). Furthermore,

a study has revealed the value of tumor vascularization

imaged with 2-D sonography for prediction of axillary

status in breast carcinoma (Santamarıa et al. 2005).

Thus, using sonography to study vascularity in laryngeal

cancer and, further, to predict cervical lymph node status

becomes feasible.

However, Santamarıa et al. (2005) has also reported

that there is no significant relationship between the density

of microvascularization on stained sections and the

number of vessels detectable by 2-D sonography. Because

of a nonplanar meshwork of vessels, which causes the

presentation of avascular and poorly vascularized regions

in tumor (Less et al. 1991), both stained sections and 2-D

sonography cannot reveal the distribution of vessels in

tumor literally. Accordingly, 3-D sonography would

play a more suitable role for the in vivo assessment of

spatial tumor vascularization.

The purpose of this study was to evaluate primary

tumor vascularization by using 3-D sonography and to

Vascularity index of layrngeal cancer d J. ZHOU et al. 1597

assess the potential value of the vascularity index (VI) as

a factor for the prediction of cervical lymph node status in

patients with laryngeal cancer.

MATERIALS AND METHODS

Eighty-seven consecutive patients (7 women and

80 men) with laryngeal cancer diagnosed by direct or indi-

rect laryngoscope and biopsy were recruited between

August 2005 and June 2008. They ranged in age from

28–85 years (mean 6 SD, 58.6 6 8.4). A pilot study

showed that subjects with tumors that penetrated the

thyroid cartilage brought severe error to calculations that

computed tumor volume with the ellipsoid formula, and

these subjects were accompanied with high N-staging,

which could be easily detected by 2-D sonography. For

these reasons, similar patients had been excluded from

our study. Of the involved four sonographers with 12–

22 years of experience, one took charge of recruitment

and examination of 2-D and 3-D sonography and stored

the images on a disk (Hitachi Global Storage Technolo-

gies Ltd., Bangkok, Thailand); the others who were

blinded to the status of the cervical lymph nodes reviewed

the 3-D images on the disk.

After the sonographic examination, all subjects

underwent total laryngectomy and bilateral radical neck

dissection. The postoperative histopathology confirmed

the diagnoses and determined the presence or absence of

metastatic cervical lymph nodes. The pathological

N-staging (pN-staging) was defined according to the

2002 International Union Against Cancer TNM classifica-

tion (Sobin and Wittekind 2002). This study was approved

by the institutional review board where the work was per-

formed. Oral informed consent was obtained from all

patients.

Sonography was performed with a 10.0-MHz linear

array transducer connected to a Technos MPX scanner

(Esaote, Genoa, Italy). Subjects were asked to lie in the

supine position. The shoulders were supported with

a pillow to keep the neck extended. The transducer was

placed on the anterolateral aspect and in the middle of

the neck to evaluate neoplasm and larynx on longitudinal

and transverse sections. The scan region was from the

level of the hyoid bone to the level of the cricoid cartilage.

To investigate the infiltration extent of the tumor, we

systematically identified these specific structures in each

subject: base of tongue, hyoid bone, strap muscles, laryn-

geal cartilages, epiglottis, aryepiglottic folds, vocal cords,

ventricular bands, pre-epiglottic space, paraglottic spaces,

thyroid gland, neck vessels and the subglottic region.

Morphologic changes of the laryngeal structure with

lesions and fixity of the vocal fold were deemed invasion.

On the basis of previous reports (Gritzmann et al. 1989;

Erkan et al. 1993), hypoechoic margins were defined as

the extent to which infiltration occurred. The maximum

length of the tumor was measured in centimeters on the

longitudinal scan, whereas the maximum width and thick-

ness were measured in centimeters on the transverse

section that was perpendicular to the longitudinal scan.

Then the volume of the neoplasm was calculated in cubic

centimeters by using the ellipsoid formula.

The mode we selected for 3-D sonography was

‘‘Free-hand.’’ Color Doppler settings were standardized

as follows: the pulse repetition frequency was 1.0–

1.3 KHz, the color gain was adjusted to a level without

background noise, the band-pass filter was set on ‘‘L’’

(low) and the gain of gray scale was turned down to

clearly identify the tumor’s border. Numerous points

were fastened on the tumor’s border and were linked

together, and then the gathered image was clipped with

the ‘‘Electronic-cutter’’ to delete the pixels of surrounding

structures and to define the tumor boundary. The ‘‘Elec-

tronic-cutter’’ is a program feature of the system. After

that, a 3-D spatial image was reconstructed by system soft-

ware to display the distribution of the blood vessel

network in the primary larynx tumor. For clear identifica-

tion of terminal vessel branches, the spatial image was

rotated on the x, y, z or any other axis, and the gray scale

was adjusted once more. The terminal vessel branches of

the tree-like framework within the tumor were counted

one by one, by consensus of the three sonographers, and

then divided by the volume of the tumor to obtain the

value of the VI.

For comparison, the cervical lymph nodes were de-

tected by using 2-D sonography. According to a previ-

ously reported distribution of neck node metastasis

(Shah 1990), we scanned the submental and submandib-

ular triangle, upper middle and lower third of the neck

and the posterior triangle of the neck. On the basis of

cervical lymph node differentiation studies (Tschammler

et al. 1998; Ahuja and Ying 2003), we used the four diag-

nostic rules of metastatic cervical lymph nodes: the ratio

of the short to long axis diameter $0.5, hypoechoic,

absence of echogenic hilus and distortions of the intrano-

dal angioarchitecture. The suspicious metastatic lymph

nodes found by 2-D sonography were marked with black

dots on the cervical skin and described with respect to the

subcutaneous depth. Black silk suture was threaded

through the node labeled by sonography as an intraopera-

tive symbol to compare with the pathologic findings.

The relationship between the VI and pN-staging was

evaluated by correlation coefficient r that was obtained

from the Spearman’s rho in nonparametric correlations.

To investigate the ability of the VI to predict lymph

node metastasis, we combined the subjects with involved

cervical lymph nodes and constructed a receiver operating

characteristic (ROC) curve. The best operating point was

determined by Youden’s index (Youden 1950; Hilden and

Table 1. Classified VI values corresponding to pN-staging

pN-staging

VI pN0 pN1 pN2

#5 34 3 1�10 10 9 9�15 0 5 6.15 0 2 8

VI 5 vascularity index; pN-staging 5 pathological N-staging.

1598 Ultrasound in Medicine and Biology Volume 35, Number 10, 2009

Glasziou 1996). The area under the ROC curve was calcu-

lated to assess the accuracy of the VI to predict metastases

of the cervical lymph nodes.

Two methods were combined by recruiting the VI as

an inclusion criterion for positive node and 2-D sonog-

raphy as an exclusion criterion for negative node. The accu-

racy, sensitivity and specificity of the VI, 2-D sonography

and the combination of the two methods in the diagnoses of

metastatic cervical lymph nodes were compared. A McNe-

mar’s test was applied to evaluate their differences.

SPSS statistical software (SPSS Inc., Chicago, IL)

was used for data analysis. A p-value of , 0.05 was

considered significant.

RESULTS

Postoperative pathology confirmed that each subject

had a single tumor in the larynx. All cases were squamous

cell carcinoma. The reconstructed 3-D sonography images

clearly presented the spatial distribution of the vessel

network in laryngeal cancer (Fig. 1).

The values of the VI were from 0–20.642; only one

case had a VI value .20. There was no case with pN3.

Therefore, subjects were classified in four grades on the

basis of the VI values and classified in three grades accord-

ing to the pN-staging (Table 1). The r value between the VI

and pN-staging was 0.740 (p , 0.001), which demon-

strated a noticeable positive linear correlation. A denser

vascular network and more terminal vessel branches

existed in tumors of the higher pN-staging cases (Fig. 2).

The area under the ROC curve for the VI was 0.919.

It denoted an excellent accuracy for diagnosing metastases

Fig. 1. 3-D sonogram of a 42-year-old man with pN1 cervicallymph nodes. The 3-D sonogram shows the tumor vessel

network (arrows) clearly.

of the cervical lymph nodes. An analysis of the coordi-

nates of the curve showed that the largest Youden’s index

was 0.719, at which the corresponding value of the VI was

4.4565. This was determined to be the best operating point

(Fig. 3). In referring to postoperative pathology, this coor-

dinate of the curve yielded an accuracy of 85%, a sensi-

tivity of 95% and a specificity of 75%.

The accuracy, sensitivity and specificity of the VI,

2-D sonography and a combination of the two methods

are shown in Table 2 for comparison. There was no signif-

icant difference in accuracy between the VI and 2-D

sonography (p 5 0.629). The specificity for the VI was

lower than that of 2-D sonography, but the sensitivity

for the VI was higher compared with that of 2-D sonog-

raphy. Supernumerary six cases with lymph nodes metas-

tasis discovered by the VI were not detected by 2-D

sonography. In the pathologically negative neck group,

10 cases diagnosed node-negative by 2-D sonography

were taken as node-positive by the VI, whereas only one

Fig. 2. 3-D sonogram shows the vessel network and terminalvessel branches (arrows) clearly in a 54-year-old man withpN2 cervical lymph nodes. The vessel network is denser than

that in the pN1 case (Fig. 1).

Fig. 3. ROC curve of VI for predicting cervical lymph nodemetastases. The best operating point (solid dot on the curve)determined by Youden’s index corresponds to a sensitivity of

95% and a specificity of 75%.

Vascularity index of layrngeal cancer d J. ZHOU et al. 1599

case the VI took as node-negative was diagnosed node-

positive by 2-D sonography. Differences in the sensitivity

and specificity between the VI and 2-D sonography were

significant (p 5 0.031 and 0.012). The combination of the

two methods yielded a higher accuracy (97% vs. 85% and

89%, p 5 0.002 and 0.016), a higher sensitivity to 2-D

sonography (95% vs. 81%, p 5 0.031) and a higher spec-

ificity to VI (98% vs. 75%, p 5 0.002).

DISCUSSION

There are a series of tumor-related predicting factors

for laryngeal cancer. Among these, lymph node involve-

ment is widely accepted as the most important clinical

predicting factor (Lefebvre et al. 2002). Many studies

have indicated that sonography shows a high accuracy

in differentiating metastatic from benign nodes (Takeuchi

et al. 1999; Yonetsu et al. 2001; Ahuja and Ying 2003).

Table 2. Accuracies, sensitivities and specificities for VI,2-D sonography and combination of the two methods in

diagnosis of metastatic cervical lymph nodes

Diagnostic method Accuracy (%) Sensitivity (%) Specificity (%)

VI 85 (74/87) 95 (41/43) 75 (33/44)2-D sonography 89 (77/87) 81 (35/43) 95 (42/44)VI 1 2-D 97 (84/87) 95 (41/43) 98 (43/44)

Data in parentheses are the number of patients.VI 5 vascularity index; 2-D sonography 5 2-dimensional sonog-

raphy; VI 1 2-D 5 combination of vascularity index and 2-D sonog-raphy.

However, the limitation of using 2-D sonography alone

to determine whether metastasis exists in lymph nodes still

remains. Retropharyngeal lymph nodes and many meta-

static nodes ,5.0 mm are undetectable by sonography

(Don et al. 1995; van den Brekel et al. 1990). Even path-

ological examination may provide a false sense of security

because of occult metastatic disease and the limited ability

of the surgeon to pick appropriate lymph nodes (Finn et al.

2002). Thus, the purpose of this study was to find a param-

eter that was complementary or would surpass the current

assessment of lymph node status.

Murray et al. (1997) and Kupisz et al. (1999) inves-

tigated the sectioned slides of laryngeal cancer and re-

ported that the degree of angiogenesis may become

a reliable independent indicator that can be used to predict

nodal status, tumor recurrence and possibly survival. To

our knowledge, there are no available documents

describing the use of sonography to assess angiogenesis

or vascularity in laryngeal cancer. However, one study

concerning quantitative assessment by power Doppler

sonography on the angiogenesis of cervical cancer (Cheng

et al. 1999) and another study on breast carcinoma vascu-

larization evaluated by power Doppler sonography (San-

tamarıa et al. 2005) can be references to us. As stated in

the latter paper, tumor vascularization detected with 2-D

sonography contributes to predication of auxiliary node

status; nevertheless, there is no significant relationship

between area of microvascularization on stained sections

and the number of vessels detected by 2-D sonography.

Accordingly, the nonplanar meshwork of vessels in tumor

(Less et al. 1991) may not be thoroughly revealed by

stained sections and 2-D sonography. Therefore, our study

does not focus on stained sections and vessels detected

by 2-D sonography, but on the vessels imaged with 3-D

sonography to evaluate the spatial vascularity of a tumor

and subsequently predict the metastasis of nodes.

The r value and the area under the ROC curve con-

cerning the VI and pN-staging indicate that VI may

become a promising predicting indicator for cervical

lymph node status. The results of our study show that,

for detecting metastatic nodes in subjects with laryngeal

cancer, the VI did not have many advantages over 2-D

sonography in accuracy and specificity. However, the

sensitivity of the VI was higher than that of 2-D sonog-

raphy, which will reduce recurrence and salvage surgery.

Moreover, a combination of the VI and 2-D sonography

yielded satisfactory accuracy, sensitivity and specificity,

and remedied the defects of the two methods that occur

when each are used individually.

A limitation of sonography in this study is the severe

calcification of the laryngeal cartilage that hinders the

detection of the neoplasm in the larynx and the reconstruc-

tion of the 3-D sonogram. In our series, however, no

subject failed to obtain a 3-D sonogram because of the

1600 Ultrasound in Medicine and Biology Volume 35, Number 10, 2009

calcification. By changing the angle of the ultrasound

beam and using the remaining uncalcified cartilage as an

acoustic window, the tumors of all subjects in this study

were visualized. A study by Arens and Glanz (1999) intro-

duced the concept that endoscopic high-frequency sono-

graphy may become a useful diagnostic tool. The

limitations of our study may be remedied by the use of

endoscopic high-frequency sonography. Another limita-

tion of the study is related to artifacts of 3-D sonography.

To diminish the artifact to an acceptable degree, subjects

were asked to hold their breath while the image was being

obtained. Slowly and smoothly gliding the 2-D probe or

acquiring the 3-D data with a volumetric probe could

also decrease the motion artifacts. It was because of

conspicuous artifacts of power Doppler that color Doppler

flow imaging was recruited in this study, although the

latter presented vascular discontinuities in the color image

because of low velocities and disadvantageous Doppler

angles. In addition, the process of counting the terminal

vessel branches was time consuming, but appendant dedi-

cated analysis software is not required. It was important to

adjust the ‘‘scale’’ properly, because an excessive ‘‘scale’’

may lead the observer to misconstrue some arteriolar walls

distorted by the pulse as terminal branches.

The role of the VI in the clinical management of

laryngeal cancer has not been revealed thoroughly. On

the one hand, lymph node metastasis is regarded as the

single greatest predictor of recurrence in patients with

laryngeal cancer (Murray et al. 1997), but on the other,

some metastatic nodes cannot be revealed by current

methods (Finn et al. 2002). Therefore, further study will

be carried out to investigate the interval between laryngeal

excisions and recurrence, and subsequently to evaluate the

ability of the VI to predict recurrence.

SUMMARY

In conclusion, the combination of the VI and 2-D

sonography improves accuracy, sensitivity and specificity

compared with 2-D sonography or the VI alone. The VI

obtained by 3-D sonography may become a useful predict-

ing factor for the metastases of cervical lymph nodes in

patients with laryngeal cancer.

REFERENCES

Ahuja A, Ying M. Sonographic evaluation of cervical lymphadenopathy:Is power Doppler sonography routinely indicated? Ultrasound MedBiol 2003;29:353–359.

Arens C, Glanz H. Endoscopic high-frequency ultrasound of the larynx.Eur Arch Otorhinolaryngol 1999;256:316–322.

Cheng WF, Lee CN, Chu JS, Chen CA, Chen TM, Shau WY, Hsieh CY,Hsieh FJ. Vascularity index as a novel parameter for the in vivoassessment of angiogenesis in patients with cervical carcinoma.Cancer 1999;85:651–657.

Don DM, Anzai Y, Lufkin RB, Fu YS, Calcaterra TC. Evaluation ofcervical lymph node metastases in squamous cell carcinoma of thehead and neck. Laryngoscope 1995;105:669–674.

Erkan N, Tolu I, Aslan T, Guney E. Ultrasonography in laryngealcancers. J Laryngol Otol 1993;107:65–68.

Finn S, Toner M, Timon C. The node-negative neck: Accuracy of clinicalintraoperative lymph node assessment for metastatic disease in headand neck cancer. Laryngoscope 2002;112:630–633.

Folkman J. The role of angiogenesis in tumor growth. Semin Cancer Biol1992;3:65–71.

Gritzmann N, Traxler M, Grasl M, Pavelka R. Advanced laryngealcancer: Sonographic assessment. Radiology 1989;171:171–175.

Hilden J, Glasziou P. Regret graphs, diagnostic uncertainty and You-den’s Index. Stat Med 1996;15:969–986.

Kupisz K, Chibowski D, Klatka J, Klonowski S, Stepulak A. Tumorangiogenesis in patients with laryngeal cancer. Eur Arch Otorhinolar-yngol 1999;256:303–305.

Lefebvre JL, Lartigau E, Kara A, Sarini J. Oral cavity, pharynx, andlarynx cancer. In: Sobin LH, Wittekind Ch, (eds). TNM Classifi-cation of Malignant Tumors. ed 6. New York: Wiley-Liss; 2002. p.151–159.

Less JR, Skalak TC, Sevick EM, Jain RK. Microvascular architecture ina mammary carcinoma: Branching patterns and vessel dimensions.Cancer Res 1991;51:265–273.

Murray JD, Carlson GW, McLaughlin K, Pennington M, Lynn M,DeRose PB, Williams JK, Cohen C. Tumor angiogenesis as a prog-nostic factor in laryngeal cancer. Am J Surg 1997;174:523–526.

Santamarıa G, Velasco M, Farre X, Vanrell JA, Cardesa A,Fernandez PL. Power Doppler sonography of invasive breast carci-noma: Does tumor vascularization contribute to prediction of axillarystatus? Radiology 2005;234:374–380.

Shah JP. Patterns of cervical lymph node metastasis from squamouscarcinomas of the upper aerodigestive tract. Am J Surg 1990;160:405–409.

Sobin LH, Wittekind Ch. Head and neck tumors. In: Sobin LH,Wittekind Ch, (eds). TNM Classification of Malignant Tumors. ed6. New York: Wiley-Liss; 2002. p. 36–42.

Takeuchi Y, Suzuki H, Omura K, Shigehara T, Yamashita T,Okumura K, Shimada F. Differential diagnosis of cervical lymphnodes in head and neck cancer by ultrasonography. Auris NasusLarynx 1999;26:331–336.

Teknos TN, Cox C, Barrios MA, Chepeha DB, Bradford CR, Fisher SG,Wolf GT. Tumor angiogenesis as a predictive marker for organpreservation in patients with advanced laryngeal carcinoma. Laryn-goscope 2002;112:844–851.

Tschammler A, Ott G, Schang T, Seelbach-Goebel B, Schwager K,Hahn D. Lymphadenopathy: Differentiation of benign from malig-nant disease–Color Doppler US assessment of intranodal angioarch-itecture. Radiology 1998;208:117–123.

van den Brekel MW, Stel HV, Castelijns JA, Nauta JJ, van der Waal I,Valk J, Meyer CJ, Snow GB. Cervical lymph node metastasis:Assessment of radiologic criteria. Radiology 1990;177:379–384.

Yonetsu K, Sumi M, Izumi M, Ohki M, Eida S, Nakamura T. Contribu-tion of Doppler sonography blood flow information to the diagnosisof metastatic cervical nodes in patients with head and neck cancer:Assessment in relation to anatomic levels of the neck. AJNR Am JNeuroradiol 2001;22:163–169.

Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32–35.