Prognostic indicators for invasive carcinoma of the vulva

GYNECOLOGIC

ONCOLOGY

42, 239-244

(1991)

Prognostic Indicators for Invasive Carcinoma of the Vulva FELIX N. RUTLEDGE,*,’ ‘Department

of Gynecology,

MICHELE FOLLEN MITCHELL,* MARK F. MUNSELL,~ E. NEELY ATKINSON,? SYLVIA BASS,* VIRGIE MCGUFFEE,* AND ELVIO SILVA$ TDepartment

of Biomathematics, and $Department of Pathology, Center, Houston, Texas 77030

the University

of Texas M.D.

Anderson

Cancer

Received January 3, 1991

Three hundred sixty-five patients with invasive squamouscell carcinoma of the vulva have been treated at M.D. Anderson CancerCenter between1944and 1990.We undertook a rigorous review of the medical records, and a Cox proportional hazards modelwasappliedto examinepredictorsof both failure to survive and recurrence. Significant predictors of both failure to survive and recurrenceincluded tumor size, clinical stage, therapy aim, pelvic or inguinal nodal metastases,and positive margins. We then undertook an analysisof StageI and II lesionstreated with a curative aim to seeif there was a difference in survival or in disease-free interval betweenthose patients treated with radical vulvectomy and those treated with radical wide local excision. There was no survival advantage from the radical vulvectomy procedure. We conclude that careful selectionmay allow us to choosesomepatients for lessradical procedures. 0 1991 Academic Press, Inc.

cedure. Prognostic indicators studied by other authors include tumor size; tumor grade; FIG0 stage; whether therapy aim is palliative or curative; inguinal, pelvic, or Cloquet’s nodal metastases; deep, vaginal, and lateral surgical margins; age; labial versus clitoral lesion location; lesion appearance; TNM stage; medical status; and lymphatic vascular space involvement. We have recently subjected our 365 cases of invasive carcinoma to a Cox proportional hazards analysis in a retrospective study to make the series comparable to other reports in the literature; however, we were not able to examine lymphatic vascular space involvement in this study. MATERIALS

AND METHODS

Three hundred sixty-five patients with invasive vulvar carcinoma have been followed at The University of Texas M.D. Anderson Cancer Center between 1944 and 1990. A retrospective review of the medical records was performed focusing on age; race; marital status; age at menopause; medical status; associated genital disease; clinical symptoms; duration of symptoms; gross appearance of the tumor; tumor size; site of lesion; histology; tumor grade; FIG0 stage; TNM stage; associated malignancies; aim of therapy; primary treatment; secondary treatment; tertiary treatment; surgically positive nodes; deep, vaginal, or lateral margins; length of hospital stay; early and late complications; and recurrence. Failure to survive and recurrence were estimated for all of the indicators of interest using the Kaplan-Meier estimate. Log-rank tests were performed on the survival and disease-free interval curve to determine whether there was a difference in the levels of the variables at the P = 0.05 level. Contingency tables were formed to examine the homogeneity of data for surgical procedure comparing radical wide local excision versus radical vulvectomy for the variables of interest [3].

INTRODUCTION

Within the past 10 years there has been intensified interest in prognostic indicators for carcinoma of the vulva. This interest has been associated with a move to improve the selection of patients for individualized treatment. After the treatment of carcinoma of the vulva by radical vulvectomy and lymphadenectomy was established as superior (1, 21, this treatment was followed rigidly for all patients who were physically suited. There is now an effort to modify this management for some Stage I and II lesions to lessen the undesirable postoperative leg edema and deformity created by the extensive resection. It is essential that these attempts at conservatism not sacrifice survival. Prognostic indicators have a role in determining’ who may be treated with a less radical pro’ To whom correspondence should be addressed at Department of Gynecology, The University of Texas M.D. Anderson Cancer Center, Box 67, 1515 Holcombe Blvd., Houston, TX 77030. 239

0090-8258/91$1.50 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form’ reserved.

240

RUTLEDGE

A subset of variables was then chosen for further investigation. Log-log plots of the Kaplan-Meier estimates of survival were obtained to examine the assumption of proportional hazards [4]. The proportional hazards model of Cox was used to identify prognostic factors [5]. No obvious departures from the assumption of proportional hazards were discovered for the variables of interest. The Cox proportional hazards model was fit with each of the variables of interest for both failure to survive and recurrence. The same analysis was then applied to Stage I and II cases treated with curative intent to look at radical wide local excision versus radical vulvectomy and their effects on both survival and recurrence. The result of the data are presented in Tables l-4. The age, length of follow-up, ethnicity, stage, and therapy are outlined in Table 1. Three hundred thirty-one (90.7%) of the patients were treated with curative intent, while thirty-four (9.3%) were treated with palliative intent only. Twenty of the patients had no surgery; the surgical treatment of the remaining three hundred fortyfive patients is detailed in Table 1. Radiation was used perioperatively in 67 patients (18.3%), and this was used as the single treatment modality in 20 patients (5.5%). Chemotherapy was used in 18 patients (4.9%). In the total 365 patients, recurrences were noted in 116 (31.8%), of which 51 (14%) had local recurrences, 31 (8.5%) had regional recurrences, and 34 (9.3%) had distant recurrences.

RESULTS When the univariate Cox proportional hazards model was used to examine failure to survive, significant predictors included tumor size, FIG0 stage, therapy aim (palliative versus curative), positive groin or pelvic nodes, lateral margins, and TNM stage. Nonsignificant predictors of failure to survive included age, grade, positive Cloquet’s node, positive deep and vaginal margins, site of lesion (labia versus clitoris and/or mons), gross tumor appearance, and medical status. These findings may be found in Table 2. Significant predictors of recurrence included size of tumor, FIG0 stage, therapy aim (palliative versus curative), groin nodes, pelvic nodes, deep surgical margin, and clinical TNM stage. Nonsignificant predictors of recurrence were grade, positive Cloquet’s node, lateral and vaginal surgical margins, age, site of lesion, gross tumor appearance, and medical status. These findings may be found in Table 3. The significant predictors for both failure to survive and recurrence included tumor size, clinical stage, therapy aim, positive pelvic or inguinal nodes, and positive margins. Positive lateral margins were significant for failure

ET AL.

TABLE 1 Invasive Vulvar CarcinomaStudy 1944-1985:Summary of Patient Characteristics(N = 365) Median, 64 years Age: Range, 17-97 years Median, 83 months Follow-up: Range, 11 days-428 months Ethnicity: 278 (76.2%) White, 60 (16.4%) Black, 27 (7.4%) Hispanic Number

Percentage

Stage I II III IV Unstaged

57 122 148 32 6

(15.6) (33.4) (40.6)

Total

365

100.0

Therapy Chemo only XRT only Surgery only Surgery + XRT Surgery + chemo XRT + chemo Surgery + XRT + chemo

1 21 262 61 20 0 0

(0.3) (5.7) (71.8) (16.7) (5.5) (0.0) (0.0)

Total

365

100.0

G3.8) (1.6)

5-year Survival (%) 93.8 88.9 64.3 15.6

0.0 32.8 81.4 58.0 78.2 0.0

5-Year survival for entire group: 74.2% Surgical treatments Simple vulvectomy 2 nodes Radical wide local excision + nodes Modified radical vulvectomy 2 nodes Radical vulvectomy 2 nodes Exenteration Total

25

(7.4)

52.7

26

(7.6)

71.2

20

(5.9)

86.8

261

(76.8)

79.5

8

(2.3) 100.0

37.5

(37.2) (10.6)

52.1 21.4

340

S-Year survival for entire group: 76.7% Positive nodes Inguinal Pelvic

103/277 141132

Note. XRT, radiotherapy; Chemo, chemotherapy.

to survive while positive deep margins were significant for recurrence. The results for Stage I and II patients who underwent surgical therapy with a curative intent were subjected to further analysis to see whether there was a difference when they underwent radical wide local excision versus radical vulvectomy. When we examined the outcomes of both failure to survive and recurrence, we found no statistically significant difference between the two therapies.

PROGNOSTIC

INDICATORS

FOR INVASIVE

TABLE 2 Summary of Cox Proportional Hazards Model for Vulva Data: Predictors of Failure to Survive Relative hazard Tumor size (cm) (P value = 0.005) O-2 1.0000 3-4 1.1312 5-6 2.1833 >6 2.3683 Tumor grade (P value = 0.182) I 1 .oooo II 0.9556 III 1.4748 FIG0 stage (P value = 0.000) I 1.0000 II 2.9842 III 8.6452 IV 40.9031 Therapy aim (P value = 0.000) Curative 1.0000 Palliative 8.2821 Groin nodes (P value = 0.000) Bilaterally negative 1.0000 Unilaterally positive 6.8656 Bilaterally positive 20.3212 Pelvic nodes (P value = 0.000) Bilaterally negative 1.0000 Unilaterally positive 5.9230 Bilaterally positive 10.9852 Cloquet’s nodes (P value = 0.631) Negative 1 0000 Positive 1.4548 Deep surgical margins (P value = 0.931) Negative 1.0000 Invasive 1.0914 Vaginal surgical margins (P value = 0.906)” Negative 1.0000 Invasive 0.9063 Lateral surgical margins (P value = 0.010)’ Negative 1.0000 Invasive 5.2267 Age (P value = 0.100) <41 1.0000 4160 1.8266 61-80 2.5600 >80 3.0251 Site of lesion (P value = 0.359) Clitoris or mons 1.oooo Labia 2.3082 Gross tumor appearance (P value = 0.668) Nonulcerative 1.0000 Ulcerative 1.0905 Clinical stage (tumor) (P value = 0.000) Tl 1.oooo T2 2.3574 T3 4.2916 T4 35.1826 Clinical stage (nodes) (P value = 0.000) NO 1.0000 Nl 1.3209 N2 3.0702 N3 11.2343

95% Confidence interval

(0.6364, 2.0108) (1.2116, 3.9342) (1.2310. 4.5560) (0.5716, 1.5974) (0.8653, 2.5136) (0.8694, 10.2431) (2.6973, 27.7087) (12.2898, 136.1343)

VULVAL

CANCER

241

TABLE 2-Continued Relative hazard Clinical stage (metastasis) MO MIA or MlB Medical status (P value - 0.289) No illness Obesity Diabetes Hypertension Cardiovascular abnormalities

95% Confidence interval --

1.0000 14.5149

(5.6785, 37.1022)

1.0000 0.7925 0.2702 1.1532

(0.3355, 1.8723) (0.0374, 1.9509) (0.6884. 1.9318)

1.4235

(0.7582, 2.6725)

” Of the 133 patients with deep surgical margins, only 4 had invasive carcinoma. * Of the 327 patients with vaginal surgical margins, only 8 had invasive carcinoma. ’ Of the 331 patients with lateral surgical margins, only 6 had invasive carcinoma.

(5.1064, 13.4328) (3.5492, 13.2808) (10.0935, 40.9127) (2.9335, 11.9590) (2.4803, 48.6537) (0.3424, 6.1807) (0.1523, 7.8227) (0.2233, 3.6779) (1.8889, 14.4626) (0.6416, 5.2002) (0.9264, 7.0745) (0.9068, 10.0917) (0.3045, 17.4957) (0.7338, 1.6205) (1.0462, 5.3119) (1.9420, 9.4837) (10.0462, 123.2127) (0.7026, 2.4834) (1.8513, 5.0916) (6.3559, 19.8571)

The patients were further compared by an examination of tumor size, tumor grade, clinical stage, and presence of positive nodes. There were no statistically significant differences between the two groups. What was statistically significant was that more radical wide local excisions have been performed since 1980. These findings are outlined in Table 4. The data were then subjected to a multivariate Cox proportional hazards model. When survival was examined, the most significant predictors of failure to survive were positive groin nodes and positive pelvic nodes. When recurrence was examined, the most significant predictors were the same. Since these multivariate models were limited to the patients on whom node samples were obtained, there may be bias in this analysis; therefore? we consider the univariate analyses to be of more importance than the multivariate analyses. DISCUSSION The Cox proportional hazards model allows one to examine failure time data and identify predictors of statistical significance. Hopefully, the information obtained in this study will help to support modifications of our present treatment to better fit individual patient requirements, allowing us to do less radical surgery on patients who are at a lower risk of recurrence. Tumor size had a significant impact on survival with those patients having lesions over 6 cm, being 2.4 times less likely to survive and 2.1 times more likely to recur than those whose tumors were O-2 cm in size. The presence of positive nodes was a very important predictor. Those patients who had bilaterally positive inguinal nodes were 20.3 times less likely to survive and were 7.4 times more likely to recur than those

RUTLEDGE

242

ET AL.

TABLE 3 Summary of Cox Proportional Hazards Model for Vulva Data: Predictorsof Recurrence Relative hazard Tumor size (cm) (P value = 0.009) O-2 1.0000 3-4 1.0963 5-6 1.9847 >6 2.0793 Tumor grade (P value = 0.202) I 1.0000 II 0.7853 III 1.1845 FIG0 stage (P value = 0.000) I 1.0000 II 1.9372 III 4.2226 IV 12.0024 Therapy aim (P value = 0.000) Curative 1.0000 Palliative 4.1624 Groin nodes (P value = 0.000) Bilaterally negative 1.0000 Unilaterally positive 2.5205 Bilaterally positive 7.3636 Pelvic nodes (P value = 0.000) Bilaterally negative 1.0000 Unilaterally positive 4.2484 Bilaterally positive 12.8353 Cloquet’s nodes (P value = 0.9305) Negative l.Oooo Positive 0.9406 Deep surgical margins (P value = OX1438)~ Negative 1.0000 Invasive 4.3174 Vaginal surgical margins (P value = 0.878) Negative 1.0000 Invasive 1.0986 Lateral surgical margins (P value = 0.090)d Negative l.OQOO Invasive 3.2678 Age (P value = 0.573) <41 l.OOQO 41-60 0.9131 61-80 0.9657 180 1.5446 Site of lesion (P value = 0.128) Clitoris or mans 1.0000 Labia 3.5553 Gross appearance (P value = 0.869) Other 1.0000 Ulcerative 0.9704 Clinical stage (tumor) (P value = 0.000) Tl 1.0000 T2 1.9348 T3 3.4069 T4 4.2068 Clinical stage (nodes) (P value = 0.000) NO 1.0000 Nl 1.1527 NZ 2.0261 N3 6.0573

95% Confidence interval

(0.6603, 1.8200) (1.1640, 3.3840) (1.1311, 3.8223) (0.4976, 1.2393) (0.7309, 1.9196) (0.8822, 4.2540) (2.0174, 8.8382) (5.1662, 27.8848) (2.4315, 7.1254) (1.5471, 4.1061) (4.2510, 12.7551) (2.1055, 8.5724) (2.9634, 55.5940) (0.2259, 3.9163) (1.3735, 13.5713) (0.3493, 3.4557) (1.0314, 10.3530) (0.4557, 1.8294) (0.4937, 1.8890) (0.6414, 3.7199)

TABLE 3-Continued Relative hazard Clinical stage (metastasis) (P value = 0.006) MO 1.0000 MlA or MlB 8.6496 Medical status (P value = 0.190) No illness 1.0000 Obesity 1.0991 Diabetes 0.4811 Hypertension 1.2850 Cardiovascular 1.7854

(0.5357, (0.1165, (0.7938, (1.0167,

2.2551) 1.9872) 2.0801) 3.1354) 10 with invasive invasive invasive

with bilaterally negative nodes. Similarly, those patients with bilaterally positive pelvic nodes were 10.9 times less likely to survive and 12.8 times more likely to recur than those with bilaterally negative nodes. Curiously, clitoral location of the lesion did not make patients less likely to survive nor more likely to have a recurrence. As expected, the TNM staging system significantly predicted both survival and recurrence. Of great interest, the surgical procedure used did not seem to influence survival or recurrence, lending support to the idea that less radical surgery can be used on certain select patients without compromising survival or recurrence. Our 5-year survival rate is comparable to that of other large series in the literature (see Table 5) [6-111. Andersson and co-workers assessed 408 patients with primary vulvar cancer, focusing on preoperative variables to TABLE 4 Summaryof Cox Proportional Hazards Model for Vulva Data: StagesI and 11Carcinomaof Vulva Treated with Curative Intent Relative hazard

(0.6759, 1.3934)

(0.6853, 1.9388) (1.2988, 3.1607) (3.4722, 10.5671)

(2.6773, 27.9443)

’ Of the 134 patients with Cloquet’s nodes, there are only positive nodes. b Of the 333 patients with deep surgical margins, only 4 had carcinoma. ’ Of the 327 patients with vaginal surgical margins, only 8 had carcinoma. d Of the 331 patients with lateral surgical margins, only 6 had carcinoma. ’ Of 365 patients, there are only 5 with MlA or MlB.

(0.4791, 26.3819)

(1.0011, 3.7396) (1.7832, 6.5091) (0.5374, 32.9322)

95% Confidence interval -

95% Confidence interval

Outcome is failure to survive (P value = 0.409) Surgery Simple” Radical

1.0000 2.1424

(0.2829J6.227)

Outcome is recurrence (P value = 0.651) Surgery Simple” Radical

1.0000 1.3031

(0.3964, 4.2830)

’ Simple refers to radical wide local excision and radical hemivulvectomy with or without node sampling.

PROGNOSTIC

INDICATORS

FOR INVASIVE

TABLE 5 Resultsof Studieson Survival from Invasive Vulvar Carcinoma .5-Year survival rate (%I

Author

Year

No. of patients

Gosling et al. [6] de Valera [7] Way Ill Inverson ef al. [9] Podratz et al. [lo] Cavanagh et al.

1961 1967 1978 1980 1983

133 125 367 424 224

39 66 48 67 75

1990 1990

415 365

67 74

[Ill Rutledge et al.” ’ This study.

see whether they could determine their effect on survival [12]. Age, clinical node involvement, secondary tumor spread, tumor size, and menses were found to be the best individual predictors of survival time. The authors divided their patients into four groups: group 1 had impalpable nodes and no secondary tumor spread; group 2 had palpable nodes and no secondary tumor spread; group 3 had impalpable nodes and secondary tumor spread; and group 4 had palpable nodes and secondary tumor spread. As they examined these four groups in various age categories, they found each time that group 2 did worse than group 1, group 3 worse than groups 1 and 2, and group 4 the worst in all categories. Podratz and colleagues [lo] examined 224 patients with invasive squamous cell carcinoma of the vulva over a 20year period at the Mayo Clinic. There was an overall survival rate of 75% compared with 89% for age-matched controls. The most significant predictors of failure to survive were lesion size of more than 3 cm, regional nodal metastases, and histologic differentiation exceeding grade 2. Andersson and Nyboe [13] reported a series of 137 patients. They found significant predictors of failure to survive to be clitoral location of the lesion, tumor size greater than 4 cm, obesity, and age. When they examined a multivariate model, the combination of groin nodes, clitoral lesion location, and advanced grade and age were the best predictors of failure to survive. They stated that the combination of clinical and histopathologic variables had a predictive power significantly better than that of clinical variables alone. Boyce and colleagues (141 reported on 84 patients treated by vulvectomy and evaluated in relation to inguinal node status and survival. Significant predictors of inguinal node status were tumor diameter, depth of invasion, clinical node status, vascular lymphatic space involvement, and pattern of invasion. When evaluated in

VULVAL

CANCER

243

combination, only the clinical status of the inguinal nodes, depth of invasion, and pattern of invasion were predictive of inguinal node status. Survival was predicted by tumor diameter, inguinal node status, depth of invasion, pattern of invasion, and lymphatic vascular space involvement. When evaluated in combination with other parameters, the diameter of the lesion was the most important predictor, although depth of invasion and vascular lymphatic space involvement did add additional information. Shimm and colleagues [15] analyzed the records of 98 patients undergoing surgery for squamous cell carcinoma of the vulva between 1960 and 1982. They examined age, tumor size, FIG0 stage, surgically determined TNM stages, tumor differentiation, lymphatic vascular space involvement, extent of procedure, and adjuvant radiation to determine effects on local control, freedom from distant metastases, and survival using single and multivariate analyses. They found local control to be significantly related to FIG0 stage. Freedom from distant metastases was significantly related to surgical N-stage, tumor size, and surgical T-stage. Survival was significantly related to surgical end-stage, tumor size, surgical T-stage, age, and lymphatic vascular space involvement. Metastatic involvement of inguinal nodes was significantly correlated with tumor size and differentiation. Kurzl and Messerer [16] studied 124 patients with squamous cell carcinoma of the vulva treated between 1971 and 1980, examining predictors of recurrence. They found prediction of survival to be significantly affected by age, dissociated tumor growth, lymphatic spread, tumor thickness, and ulceration. Heaps and colleagues [17] reported on 13.5 patients treated between 1957 and 1985. They focused on local recurrence and found stage, margin status, depth of invasion, tumor thickness, growth pattern, lymphatic vascular space involvement, amount of keratin, and mitotic activity to be significant predictors of recurrence. Tumor size was significant at the P = 0.0505 level, grade was significant at the P = 0.18 level, and number of nucleoli was significant at the P = 0.0573 level. They concluded that a l-cm tumor-free surgical margin of the vulva results in a high rate of local control, whereas a margin of less than 8 mm is associated with a 50% chance of recurrence. We recently reported on 32 patients with clinically negative inguinal nodes and a depth of invasion greater than 1 mm who were treated with radical wide local excision and unilateral superficial inguinal lymph node dissection if the lesion was well localized or with bilateral superficial inguinal dissection if the lesion was midline [18]. The 5year life-table survival for the entire group was 84%. Univariate analysis of potential prognostic variables showed no significant recurrence or survival differences for age, symptom, duration, FIG0 stage, tumor grade, tumor location, depth of invasion, or margin status.

244

RUTLEDGE TABLE

6

Analysesof Vulvar CarcinomaPredictorsof Survival

Predictor of survival Age Size Grade FIG0 stage Clinical node status Surgical node status Margins Lymphatic vascular involvement Pattern

Number of studies examining the predictof

Number of studies finding the predictor significant over the number of studies

6 7 6 7 4 6 3

316’ 6/l’

3 2

216 317 214’ 6/6b I/3 3/3b

112

’ Seven authors, 1440 patients (including this study). b 250% studies found the predictor significant.

Seven studies, including ours, have examined predictors of survival using advanced statistical methods. Among the seven reports, there is some disagreement about which variables are important. Table 6 lists the various predictors and focuses on how many of the authors found them significant in their studies. A total of 1440 patients is included in this series. Less than complete vulvectomy and limited node dissection are already being used for selected Stage I and II patients. These ventures into more conservative resection require careful assessment of each cancer, both preand postoperatively. Simultaneous statistical analyses of these factors help us evaluate which patients are candidates for conservative therapy. There seems to be good agreement among all the authors that tumor size, surgical node status, and lymphatic vascular space involvement are important. It is therefore unfortunate that we are unable to look at lymphatic vascular space involvement in the series reported here. Thus, for the present, there appears to be no significant difference in survival or disease-free interval for patients treated with simple or radical procedures. Follow-up on these patients in the future will determine whether this is the case. REFERENCES 1. Way, S. Carcinoma of the vulva, in Progress in gynecology (J. V. Meigs and S. H. Sturgis, Eds.), Grune 81 Stratton, New York, pp. 489-506 (1957).

ET AL. 2. Taussig, F. Cancer of the vulva: An analysis of 155 cases, Am. J. Obstet. Gynecol. 40, 764779 (1940). 3. SAS Institute, Inc. SAS user’s guide: Statistics, Cary, NC, Version 5 (1985). 4. Becker, R. A., Chambers, J. M., and Wilks, A. R. The new S language: A programming environment for data analysis and graphics, Wadsworth & Brooks/Cole Advanced Books & Software, Pa-

cific Grove, CA (1988). 5. Kalfleisch, J. D., and Prentice, R. L. The statistical analysis of failure time data, Wiley, New York (1980). 6. Gosling, J. R. G., Abell, M. R., Drolette, B. M., and Loughrin, T. D. Infiltrative squamous cell (epidermoid) carcinoma of vulva, Cancer 14, 330 (1961). I. de Valera, E. Radical vulvectomy, Guest lecture presented at the Thirty-fifth Annual Meeting of the Central Association of Obstetricians and Gynecologists, Detroit, MI (September 14-16, 1967). 8. Way, S. The surgery of vulvar carcinoma: An appraisal, Clin. Obstet. Gynecol. 5, 623 (1978). 9. Iversen, T., Aalders, J. G., Christensen, A., and Kolstad, P. Squamous cell carcinoma of the vulva: A review of 424 patients, Gynecol. Oncol. 9, 271 (1980). 10. Podratz, K. C., Symmonds, R. E., Taylor, W. F., and Williams, T. J. Carcinoma of the vulva: Analysis of treatment and survival, Obstet. Gynecol. 61, 63-74 (1983). 11. Cavanagh, D., Fiorica, J. V., Hoffman, M. S., Roberts, W. S., Bryson, C. P., LaPolla, J. P., and Barton, D. P. J. Invasive carcinoma of the vulva: Changing trends in surgical management, Am. J. Obstet. Gynecol. 163, 1007 (1990). 12. Andersson, J. A., Philips, P. R., Senthilselvan, A., Guthrie, D., and Way, S. A. Preoperative prognosis for cancer of the vulva, Obstet. Gynecol. 58, 364 (1981). 13. Andersson, B., and Nyboe, J. Value of prognostic parameters in squamous cell carcinoma of the vulva, Gynecol. Oncol. 22, 341 (1985). 14. Boyce, J., Fruchter, R. G., Kasambilides, E., Nicastri, A. D., Sedlis, A., and Remy, J. C. Prognostic factors in carcinoma of the vulva, Gynecol. Oncol. 20, 364 (1985). 15. Shimm, D. S., Fuller, A. F., Orlow, E. L., Dosoretz, D. E., and Aristizabal, S. A. Prognostic variables in the treatment of squamous cell carcinoma of the vulva, Gynecol. Oncol. 24, 343 (1986). 16. Kurzl, R., and Messerer, D. Prognostic factors in squamous cell carcinoma of the vulva: A multivariate analysis, Gynecol. Oncol. 32, 143 (1989). 17. Heaps, J. M., Fu, Y. S., Montz, F. J., Hacker, N. F., and Berek, S. S. Surgical-pathologic variables predictive of local recurrence in squamous cell carcinoma of the vulva, Gynecol. Oncol. 38, 309 (1990). 18. Burke, T. W., Stringer, C. A., Gershenson, D. M., Edwards, C. L., Morris, M., and Wharton, J. T. Radical wide excision and selective inguinal node dissection for squamous cell carcinoma of the vulva, Gynecol. Oncol. 38, 328 (1990).