Abstracts Presented for the Twenty-Ninth Annual Meeting of the Society of Gynecologic Oncologists

94

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

effectiveness of RU486 or tamoxifen alone or in combination with topotecan in growth inhibition of leiomyosarcoma cells. Materials and methods. Effectiveness of RU486 or tamoxifen (Tam) and topotecan (Topo) was determined by the in vitro sulforhodamine B cytotoxicity assay. The cell line SK-LMS-1 was used in all experiments. Cells grown in 96-well dishes were treated with Topo for 2 h, followed by 96 h of incubation, while RU486 and Tam incubations were continuous for 4 days. Binding of the sulforhodamine B to fixed cells was quantitated by measuring the optical density at 540 nm. Results: Cellular responses to single agents were determined with Tam (2.5– 40 mM), RU486 (0.25–20 mg/ml), and Topo (0.5–32 mM). IC50 values of these agents were 15 mM (Tam), 2.5 mg/ml (RU486), and 2 mM (Topo). Combination treatments of Topo with RU486 or Tam were performed. We demonstrate that combinations of Topo with RU486 or Tam have synergistic effects on the leiomyosarcoma cells. At 10 – 0.25 mg/ml RU486, Topo activity was enhanced 16 to 80%, even at concentrations at which RU486 alone had no growth inhibitory effect. A similar enhancement was seen with Topo and Tam treatments (2.5–20 mM), where a 23 to 63% increase in Topo cytotoxicity was demonstrated over the single agent Topo alone. Combination of RU486 and Tam did not result in enhanced activity. Conclusions: Our data demonstrate the potential use of Topo, RU486, and Tam alone and in combination for treatment of leiomyosarcoma. Further studies are underway to test the feasibility of this approach. 87. Correlation of Integrin Expression and Cell Invasion in Human Endometrial Cancer Cells. Frank D. Yelian, J. C. Todt, Y. Eliraz, C. Carenno, J. J. Song, Gunter Deppe, and Michael P. Diamond. Wayne State University/ Hutzel Hospital, Detroit, Michigan. Poster Session Objective: To investigate endometrial cancer cell adhesion and invasion of the extracellular matrix and the possible role of integrins. Methods: Two human endometrial cancer cell lines, AN3 CA and HEC-1-A, were used in this study. Cell invasion through the basement membrane protein laminin was determined using a BIOCOAT invasion chamber. A modified solid-phase adhesion assay was used to compare cell adhesion to the extracellular matrix protein, fibronectin, vitronectin, and entactin. The expression of b3 integrins in the endometrial cancer cells was determined by flow cytometry and Western blot analysis using a specific monoclonal antibody. Results: AN3 CA cells were originally isolated from a metastatic lymph node, while HEC-1-A cells were from a grade 1A tumor located at an endometrial site. AN3 CA cells were significantly more invasive than HEC-1-A cells. We also found that AN3 CA cells were less adherent to fibronectin and vitronectin than HEC-1-A cells. AN3 CA cells were completely nonadherent to enactin, while HEC-1-A cells adhere well. Since avb3 integrin is one of the major receptors for these matrix proteins, we examined the expression of the b3 integrin subunit. Although both cell lines had detectable levels of b3, the amount of protein in AN3 CA cells was significantly lower than that in HEC-1-A cells. Most importantly, we found that the b3 integrin was completely absent on the surface of AN3 CA cells. Conclusion: This study demonstrates for the first time that deficient expression of b3 integrin in endometrial cancer cells is highly correlated with their reduced adhesion and enhanced invasive ability, suggesting that this integrin may serve as a negative marker for tumor invasiveness. 88. RU486 Inhibits Choriocarcinoma Cell Adhesion to the Basement Membrane Protein, Laminin. Frank D. Yelian, Adnan R. Munkarah, Robert T. Morris, Mei F. Xu, Richard E. Leach, Michael P. Diamond, and Gunter Deppe. Wayne State University/Hutzel Hospital, Detroit, Michigan. Poster Session Objective: Our previous study demonstrated that choriocarcinoma tissues express high levels of progesterone receptor. This study was designed to examine if the progesterone antagonist, RU486, will affect tumor adhesion to the basement membrane protein, the major step for tumor invasion. Methods: Both JEG-3, a choriocarcinoma cell line, and HTR-8/SVneo, a first-trimester trophoblast cell line, were used in this study. Indirect immunofluorescence staining was used to detect the progesterone receptor in these cells. Cell

adhesion assays were conducted following treatment with various concentrations of RU486 for 24 h. Cells were transferred onto a 96-well microtiter plate precoated with laminin. After 2 h of incubation at 37°C, adherent cells were fixed and stained with 2% Geimsa. Relative cell adhesion in each group was determined by measuring optical density (OD) using an automatic microplate reader at a wavelength of 630 nm. Results: We found that both JEG-3 and HTR-8/SVneo cells had detectable levels of progesterone receptors. Interestingly, JEG-3 cell adhesion to laminin was significantly inhibited by RU486 in a dose-dependent manner (P , 0.01). The ODs for various doses of RU486 (0, 0.1, 1, and 10 mg/ml) were 0.390 6 0.025, 0.285 6 0.04, 0.92 6 0.019, and 0.128 6 0.027, respectively. In contrast, the adhesion of normal trophoblast HTR-8/SVneo cells to laminin was not significantly affected by RU486 treatment (P . 0.05). Conclusion: This study demonstrates that RU486 inhibits choriocarcinoma cell adhesion to laminin which is the first step of invasion. This suggests that the progesterone antagonist may have a potential use as an antimetastasis agent for human choriocarcinoma. 89. Mapping Alterations of Chromosome 3p25-pter: Identification of Loci Distal to VHL Implicated in Epithelial Ovarian Cancer. Sophy Yeung,* Anne-Marie Mes-Masson,† Hafida Lounis,† Robert B. Seymour,* Diane Provencher,† and Patricia N. Tonin.* *Montreal General Hospital Research Institute, Montreal, PQ Canada; and †Centre Hospitelier Universityersite de Montreal, Montreal, PQ Canada. Basic Science Session Objective: Recently, we have described 3p alterations in epithelial ovarian cancer (EOC) suggesting the presence of more than one tumor suppressor gene (TSG) implicated in EOC. Loss of heterozygosity (LOH) was observed in 3p25-pter. We have extended the analysis of 3p25-pter to include new samples and additional markers for the purposes of mapping TSGs implicated in EOC. Methods: LOH was detected by PCR-gel electrophoresis assay using eight microsatellite repeat markers localized to 3p25-pter (including markers representing the VHL locus) in a total of 105 ovarian tumors representing 78 malignant, 18 low malignancy potential (LMP), and 9 benign tumors. Fiftyfour samples represent primary cultures derived from 41 malignant, 9 LMP, and 4 benign solid tumors or ascites. Forty-seven samples represent solid tumors. Four spontaneously transformed cell lines, TOV21G, TOV81D, TOV122D, and OV90, derived from solid malignant tumors (TOV) or malignant ovarian ascites (OV) also were analyzed. Results: For those cases which were informative, the analysis revealed LOH for at least one 3p marker in 25/78 malignant, 1/18 LMP, and 2/9 benign samples. Interstitial or LOH involving some of the 3p markers occurred in 19 cases. Only distal 3pter deletions occurred in cell line TOV112D. No association was observed with LOH and grade, stage, or histopathology of disease. Conclusions: LOH analysis suggests the presence of a TSG on 3p25-pter implicated in EOC. LOH was observed in benign, LMP, and malignant epithelial tumors, suggesting that these events are not associated with late-stage disease. LOH analysis excludes VHL as a candidate TSG in EOC. Interstitial deletions were observed in cultured cells and a cell line which will further enable the localization and characterization of putative 3p25-pter TSGs.

CERVICAL 90. The Effect of Route of Delivery on Regression of Cervical Dysplasia in the Postpartum Period. David Y. Ahdoot, M.D., Gaylord S. Rose, M.D., Kristi M. Van Nostrand, M.D., Nicole J. Nguyen, M.D., Debu S. Tewari, M.D., Vicki M. Darrow, M.D., Tom Kurosaki, M.D., and Philip J. DiSaia, M.D. University of California, Orange, California. Poster Presentation Objective: The objective of this study was to determine whether cervical dysplasia, diagnosed antepartum, resulted in differing postpartum regression rates, with respect to vaginal versus abdominal delivery. Methods: Four hundred twenty-six patients diagnosed with cervical dysplasia antenatally between 1990 and 1996 were identified through clinic/pathology logs and data were collected from inpatient and outpatient records. Complete demographic,

95

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS clinical, and cytology reports in the antepartum and postpartum period were available for 134 patients with the data separated into groups of high-grade and low-grade dysplasia. Regression was defined as normalization of or regression of the antepartum Pap smear results from high-grade dysplasia to low-grade dysplasia. Statistical analysis was performed utilizing the Fisher’s exact and Wilcoxon rank sums tests. Results: Of 108 total patients, 86 delivered vaginally and 22 delivered via cesarean section. Postpartum regression was noted in 62% of vaginal versus 32% in the patients delivered via cesarean section (P , 0.016). No differences between the two groups with respect to age, parity, or smoking history were noted. Of the 55 patients with high-grade dysplasia, 44 delivered vaginally and 11 delivered via cesarean section. Regression of high-grade CIN was noted in 59% of patients who delivered vaginally versus 0% in the ceserean section group (P , 0.0004). Conclusion: Since the reported spontaneous regression rates of cervical dysplasia ranges from 6 to 31 and 62% for high- and low-grade lesions, respectively, this study supports the clinical observation that spontaneous regression of cervical dysplasia occurs with increased frequency in the postpartum period. Most importantly, this study noted a significantly higher regression among patients with high grade lesions who delivered vaginally versus by cesarean section. 91. Adenocarcinoma in Situ of the Uterine Cervix: The Management and Outcome. Masoud Azodi and Joseph T. Chambers. Yale University, New Haven, Connecticut. Poster Session Objective: To review the evaluation and outcome of patients (pts) with adenocarcinoma (AC) in situ of cervix. Methods: A retrospective review of records between January 1988 and December 1996 identified 52 pts with AC in situ on colposcopic or cone biopsies (Bx), who had complete information; with median follow-up of 38 months (range 8 –98). Results: The mean age was 37 years (range 19 –72), mean parity was 1.5 (range 0 – 6). Fifty-two percent of patients had prior abnormal pap smears. The initiating pap was abnormal in 49 (94%). Cervical Bx’s showed 36 (69%) AC in situ and glandular dysplasia (dys), 2 (4%) squamous dys, 4 (8%) chronic inflammation, 2 (4%) no pathologic changes, 3 (6%) did not have Bx, and 5 (10%) were unknown. Subsequently 50 (96%) had cone Bx’s. The endocervical margins were positive (pos) in 27% of cold knife cones (CKC), 82% of LEEPs, and 50% of laser cones. The ectocervical margins were pos in 7% of CKC’s, 18% of LEEP’s, and 50% of laser cones. The cone histology showed 41 AC in situ/glandular dys, 3 AC in situ with superficial invasion, 5 invasive AC, and only 1 squamous dys. ECC’s above the cones were performed in 31 pts, and only 1(3%) was pos. The definitive treatment was hysterectomy in 37, repeat cone in 6, no additional therapy in 7, and unknown in 2. The definitive pathology showed residual disease in 56%. From 14 cones with negative margins, subsequent treatment showed residual disease in 6 (43%) of specimens (1 invasive AC, 1 AC in situ, 1 mild glandular dys, 3 glandular atypia). The pts with negative ECC’s above the cone had residual disease in 48% of specimens. Conclusion: Women with AC in situ of uterine cervix had residual disease in greater than 42% of cases with negative margins in cone biopsies and/or with negative ECC’s. LEEP cones had higher rate of pos endocervical margins (82%) compared to CKC (27%) and laser cone (50%). If maintaining reproductive capacity is desired, we would recommend CKC; however this does not guarantee absence of disease.

follows for our patients: CIN I (n 5 7), CIN II (n 5 8), CIN III (n 5 4). Methods: Cervical mucus was obtained on all patients prior to the LLETZ procedure. Sixteen patients were evaluable and of these 16 patients 10 had normal Pap smears after therapy; however, 6 patients had abnormal Paps (HGSIL, 1, LSGIL, 2, ASCUS favor dysplasia, 3). These 6 recurrent dysplasia patients were compared to the 10 normal patients for the following parameters: cervical mucus concentrations of IgG, IgA, TNF-a, IL-12, IL-10, IL-6. Results: (See table).

Cytokine (pg/ml) and Immunoglobulin (mg/ml) Concentrations in Cervical Mucus Norm (n 5 10)

CIN (n 5 6)

P value

IgG 85.2 IgA 27.1 TNF-a 25.5 IL-12 186.3 IL-10 109 IL-6 2436

(624 ) 128.4 (670) (66.5 ) 27.7 (68.9) (625 ) 89.2 (665.5) (682.9) 315 (6231) (618 ) 134 (655) (6769 ) 5306 (62899)

0.49 0.95 0.30 0.54 0.61 0.25

Conclusion: Immunosuppression is a known risk factor for CIN; however, these data indicate that a heightened immune response may also lead to recurrence of CIN. These data may explain why NSAIDs may be beneficial in chemoprevention trials. 93. Sutureless Technique for Cold-Knife Conization: A Report of 74 Consecutive Cases. Ivor Benjamin, Sally Y. Segel, and Stephen C. Rubin. University of Pennsylvania Cancer Center, Philadelphia, Pennsylvania. Objective: To assess the clinical utility and morbidity of cold-knife conization of the uterine cervix using a sutureless technique. Methods: We prospectively collected data on 74 consecutive patients who underwent cold-knife conization for cervical dysplasia at our institution between 1993 and 1997. Monopolar electrocautery and direct pressure with Monsel’s soaked cotton swabs were the only maneuvers utilized to achieve hemostasis. Age, indication for cone, surgical time, estimated blood loss (EBL), final pathology, margin status, complications, and postoperative cervical examination were recorded. Results: The mean age was 37 years (range 19 –73), EBL was 36 ml (range 10 –150 ml), and mean surgical time was 29 min (range 5– 63 min). Four of 74 (5.4%) had exocervical and 9 of 74 (12%) had endocervical margins positive for dysplasia. Final pathology revealed CIN III (25/74, 34%), CIN II (11/74, 15%), CIN I (2/74, 2.7%), CIS (7/74, 9.5%), and cancer (3/74, 4.1%), with the remaining 34% having no residual dysplasia. In no cases were the margins unevaluable. No patients required readmission in the perioperative period. One of 74 (1.4%) was seen in the emergency department for evaluation of vaginal bleeding that was controlled by direct pressure and Monsel’s solution. The postoperative examination (at 3– 4 weeks postcone) revealed good healing in all patients without evidence of stenosis or suture-related deformation of the cervix. To date, two patients have carried full-term pregnancies. Conclusions: Based upon our experience, cold-knife conization may be performed quickly and safely without the need for sutures to achieve hemostasis.

92. Heightened Immune Response in Cin May Lead to Recurrence. Maria C. Bell,* Edward J. Mayeaux,* Elba Turbat-Herrera,* Robert P. Edwards,† Eric C. Ragan,† and Peg A. Crowley-Nowick.† *LSUMC–Shreveport, Shreveport, Louisiana; and †Magee Women’s Hospital, Pittsburgh, Pennsylvania.

94. Comparison of Laparoscopic-Assisted Radical Vaginal Hysterectomy and Radical Abdominal Hysterectomy in the Treatment of Early Stage Cervical Carcinoma. Brendan S. Bolger, Alberto De Barros Lopes, Ofer D. Lavie, and John M. Monaghan. Queen Elizabeth Hospital, Gateshead, United Kingdom. Poster Session

Objective: Approximately 2,500,000 women are diagnosed with a precancerous cervical abnormality each year in the United States. Many precancerous lesions will spontaneously regress. This project assesses the local immune response in patients with CIN. Patients: Nineteen patients were identified in colposcopy clinic and had biopsy proven CIN. The ages ranged from 16 to 37 years of age with a mean age of 26.7 years. Pretherapy diagnoses were as

Objective: To compare the clinical, surgical, and histopathological features of patients treated following the introduction of laparoscopically assisted radical vaginal hysterectomy (LARVH) for early-stage cervical carcinoma compared with women treated by radical abdominal hysterectomy (RAH). Materials: The data were retrieved from the departmental data base on the first 17 cases of LARVH (April 1996 to June 1997) and compared with a historical

96

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

control group of 17 consecutive RAH (January 1994 to February 1995) in the treatment of tumors ,2 cm. Results: Six of 34 had stage 1A2 disease and 28/34 had stage 1B1. The stage, age, and weight were comparable between the two groups. Diagnosis was made on LEEP in 31/34. The mean duration of surgery was 196 and 110 min for LARVH and RAH, respectively (P , 0.001). Transfusion was required more frequently following RAH (7 vs 2, P 5 0.055). Cystotomy occurred in 1 patient at LARVH, with an uncomplicated primary repair. No significant surgical complication was seen following RAH. Pelvic hematoma and febrile morbidity were similar in each group (1 patient and 2 patients, respectively, in each group). The mean hospital stay was shorter following LARVH (4.3 and 8.5 days, P , 0.001). The mean number of lymph nodes was not significantly different between the groups (13 LARVH vs 15.5 RAH, P 5 0.12); there was nodal metastasis in 1 patient in each group. There was residual tumor in the hysterectomy in 7 LARVH and 4 RAH. Incomplete resection at the parametrial margins was seen in 2 patients following LARVH and no cases following RAH. Conclusion: This study identifies parametrial resection margins as a possible limiting factor of this procedure. There is a need to address the importance of appropriate case selection, the impact of ‘‘the learning curve,’’ and the implicit limitations of LARVH. 95. Socioeconomic Barriers to Cervical Cancer Screening in Orange County. Wendy R. Brewster,* Tom T Kurosaki,† Alberto Manetta,* and Hoda Anton-Culver.† *UCI Medical Center, Orange, California; and †University of California, Irvine, California. Objectives: To identify significant socioeconomic barriers to cervical cancer screening. Methods: Data were extracted from the 1989 –1991 County Health Surveys which included information on female genital tract screening. Variables analyzed included: history and frequency of Papanicolaou smears, race, age, household income, occupation, and medical insurance status. Results: One thousand four hundred fourteen adult female residents participated in the survey. Of the women surveyed, 79% received cervical cancer screening at least once every 3 years. By x2 analysis the factors that most strongly correlated with cervical cancer screening were level of education P , 0.0005; insurance status P , 0.0005; and annual household income greater than the lowest quartile ($30,000) P , 0.00005. In the lowest income quartile there was no difference in the frequency of Papanicolaou smears irrespective of insurance status or whether the subject was a single parent. The frequency of Papanicolaou smears was directly related to patient age. Younger respondents (18 –29 years old) were less likely to have received screening compared to older participants, 17% vs 4%, P , 0.005. Other positive factors included professional occupations P , 0.005; and being Caucasian P , 0.005. Using multivariate logistic regression analysis, the factors that remained as independent predictors for screening were household income P , 0.0001; professional occupations P , 0.006; and race P , 0.037. Insurance status was not an independent predictor. Conclusions: The importance of cervical cancer screening has been well established. Household income, occupation, and race are the significant socioeconomic barriers to its application. 96. Hormone Receptors Status Postradiation Therapy in Patients with Cervical Cancer. I. Cass, A. L. Fields, D. Breining, B. Schwartz, J. Jones, P. S. Anderson, C. D. Runowicz, and G. L. Goldberg. AECOM & Montefiore Medical Center, Bronx, New York. Poster Session Objective: Endometrial activity following radiation therapy (RT) in cervical cancer has been reported. Cervical cancer is not considered a hormonesensitive tumor and yet the use of hormone replacement therapy (HRT) in premenopausal women after RT is unclear despite its known benefits. Endometrial function after RT and the need for combined HRT remains controversial. We evaluated hormone receptor status in the uteri of women following standard pelvic RT for cervical cancer to examine the histologic effect of radiation in order to make recommendations for appropriate HRT in these women. Methods: We reviewed the histology and the presence of estrogen and progesterone receptors (ER, PR) in the uteri of 12 women with squamous cell carcinoma of the cervix stages IB–IVA treated with standard-dose pelvic RT

who subsequently had pelvic exenteration for recurrent disease from 1994 to 1997. Three postmenopausal women without a history of HRT use who had hysterectomies for benign conditions were used as controls. Results: Four patients were premenopausal, ages 20 – 45, and 8 patients were postmenopausal, ages 47–75 at the time of RT. Fewer endometrial glands and less stroma were present in women treated with radiation regardless of menopausal status prior to RT compared to controls. ER staining was identical in premenopausal and postmenopausal radiated patients and controls. There was a significant difference in PR staining between premenopausal radiated patients and controls (P 5 0.01). However, there was no significant difference in the ER/PR staining of premenopausal and postmenopausal radiated patients. Conclusions: The current study suggests that the presence of ER/PR is not significantly affected by standard pelvic RT used to treat patients with cervical cancer. A larger study to confirm these data appears warranted because these findings suggest that HRT in these women should probably include a progestational agent. 97. Recurrence Factors of High Grade SIL after Conization with Free Margins. Luis M. Chiva, Pablo Mazariegos, Luis Sr. Chiva, Santiago Lizarraga, Angel Hernandez, Francisco Saiz-Pardo, Luis M. Granados, Regina Cardenas, Nicolas L. Garcia. HGU Gregorio Maran˜on, Madrid, Spain. Poster Session Objective: To analyze possible factors of recurrence of high-grade SIL in patients treated with conization with clear margins. Methods: From 1990 to 1995, 167 patients with suspected high grade SIL (only CIN III) of the cervix underwent conization with free margin for dysplasia in the specimen. We review retrospectively the role of type of cone, glandular involvement in the cone specimen, HIV infection, histological findings of HPV, HPV typification, and endocervical curettage as predictors of recurrence of dysplasia. From 1990 to 1992, most conizations were performed with cold knife; in contrast, after 1993 the customary treatment has been LEEP. Results: Seventeen of 167 patients had relapse of SIL (CIN III, 6 cases; CIN II, 3 cases; CIN I, 4 cases; and HPV, 4 cases). Pap smear was abnormal as first finding of relapse in 49% of the cases. Four of 17 women were infected by HIV. At least one of the known high-risk subtypes of HPV was found originally in the specimen of conization in 36% of patients by DNA hybridization. Glandular involvement was found in 17% of the cases. Mean time of relapse was 19 months (range 2– 47 months). Patients who underwent a LEEP suffered a statistically significantly higher risk of relapse than those who were treated with cold knife conization. All other factors increased the percentage of recurrence even though they had no statistical significance. Age of HIV-positive patients was statistically lower and the finding of aggressive types of HPV (16, 18, 31, 33, 35, and 51) was higher in this group (P less than 0.05). Glandular involvement was correlated positively with positive endocervical curettage. Conclusion: Classic conization statistically diminished the risk of relapse of high-grade SIL compared with LEEP in our series. 98. HPV-Based Triage for Cervical Cancer Screening. Chris J. Colby, Kinney K. Walter, Manos M. Michelle, and Robert A. Hiatt. Kaiser Permanente, Oakland, California. Objectives: We studied an HPV-based triage system for ASCUS Pap smears. This approach used hybrid capture testing for high-risk HPV DNA, performed on ThinPrep, monolayer liquid cytology medium, collected at the time of the original Pap smear. Among the study objectives was estimating the early disease detection, the reduction of Pap smears and colposcopies, and the cost savings of such a clinical management approach. Methods: Based on data collected from a large prepaid health plan we used a Monte Carlo simulation model to evaluate the costs and benefits of HPV-based triage, using current standards of follow-up practice as the baseline for comparison. The model simulates the natural progression of disease and then superimposes both the current follow-up program and the proposed HPV-based triage program. Sensitivity analysis was used to determine if changes in any of the base assumptions significantly changed the results. Results: Under a set of conser-

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

97

vative assumptions, HPV-based triage for ASCUS Pap smears was shown to decrease follow-up Pap smears by over 50% and colposcopies by 10% with no resulting decrease in the number of HSILs missed. HPV-based triage reduced follow-up costs without compromising disease detection. Conclusions: HPVbased ASCUS triage is a useful and potentially cost-saving technology if HPV DNA testing is conducted on the liquid cytology medium collected at the time of the original Pap smear.

after cone biopsy for bleeding. Conclusion: Patients diagnosed with CIS should be counseled and tested for HIV infection because of a significant seropositive rate. Gynecologic follow-up is poor in this patient population; special efforts should be made to enhance compliance. Cone biopsy and hysterectomy appear to be equally safe and effective in the treatment of CIS. CIS is a poor prognostic indicator for death from any cause. In HIV patients, however, cervical cancer is an unlikely cause of these deaths.

99. Prognostic Factors for Pelvic Lymph Node Metastases and/or Recurrence in Early-Stage Adenocarcinoma of the Cervix. Allan L. Covens,* Joanne Kirby,* Patricia Shaw,* William Chapman,* and Edmee Franseen.† *University of Toronto, Toronto, Ontario, Canada; and †Toronto Sunnybrook Regional Cancer Centre, Toronto, Ontario, Canada.

101. A Retrospective Study of Radical Vaginal Hysterectomy with Pelvic Lymphadenectomy versus Abdominal Radical Hysterectomy for the Treatment of Early Cervical Cancer. Peter R. Dottino, Manolis T. Tsatsas, Ann Marie Beddoe, Daniel H. Tobias, and Carmel J. Cohen. Mt. Sinai Medical Center, New York, New York.

Objectives: To assess the prognostic significance of local tumor factors in predicting lymph node metastases and/or recurrence in early adenocarcinoma of the cervix. Methods: All patients with stage I adenocarcinoma of the cervix having tumor thickness ,10 mm, in whom pelvic lymph node dissection and radical hysterectomy or radical trachelectomy was performed at our center were selected from our prospective cervical cancer database (1986 to present). All pathology specimens including cone biopsy, LEEP, and hysterectomy were re-reviewed to assess histological subtype, depth, volume, grade, and presence of capillary lymphatic space involvement (CLS). Volume was calculated as ‘‘depth 3 length 3 1.5 max(depth, length).’’ These factors were then correlated with clinical data extracted from the database. Results: One hundred twelve patients were identified from the database; however 44 patients were excluded from this analysis due to pathology specimens not retrievable. Sixtyeight patients had complete pathological and clinical data. The mean age was 40 years, and the median follow-up was 40 months (range 8 –102). The median tumor depth and volume was 2.8 mm (range 0.3– 8.0), and 237 mm3 (range 0.1–7996), respectively. Pelvic lymph node metastases were identified in 2 patients with 3.6 and 4.9 mm of invasion, and tumor volumes of 1196 and 2624 mm3, respectively. Both specimens were CLS2. Recurrence occurred in 3 patients with 2.5, 3.7, and 5.0 mm of invasion and tumor volumes of 843, 671, and 3480 mm3, respectively. Two of these 3 specimens were CLS1. Conclusions: The incidence of positive pelvic lymph nodes and/or recurrence in our patient population was very low. Because all patients with metastatic disease or recurrence had tumor volumes .500 mm3, volume of disease rather than depth of invasion may be the single most important prognostic factor for the above events.

Objectives: To compare the technique of radical hysterectomy with laparoscopic pelvic lymphadenectomy to the traditional radical abdominal hysterectomy. Methods: Retrospective review of 50 consecutive cases of operable early-stage cervical cancer was conducted. The patient population was reviewed, noting patient demographics, estimated blood loss, transfusion requirements, lymph node yield, uterine weight, operative time, febrile morbidity, length of hospital stay, and recovery of bladder function. Results: The two groups of patients, radical abdominal hysterectomy (n 5 31) and radical vaginal hysterectomy (n 5 19) did not differ with respect to age, parity, weight, stage of disease, and anesthetic risk. There was a statistically significant reduction in blood loss, transfusion requirements, and length of hospital stay in the vaginal group. There was no difference in number of lymph nodes removed, and the mean operative time differed only by 5 min. There was also a trend toward reduction of febrile morbidity in the vaginal group. The laparoscopic lymphadenectomy time averaged 115 min. There was both a reduction in operative time and an increase in the number of lymph nodes collected as the surgeons gained familiarity with the procedure. Operative complications were distributed equally in the two groups. Conclusions: We observed a reduction of blood loss, length of hospital stay, and febrile morbidity in the vaginal group. This was accomplished without compromising the efficacy of the procedure and without an increase in complications. As the surgeons gain more familiarity with the procedure, there was an increase in lymph node yield and a reduction of operative time.

100. Hysterectomy versus Cone Biopsy in HIV-Seropositive Patients with Carcinoma in Situ of the Cervix. Giuseppe Del Priore and Rita M. Roure. New York University Medical Center, New York, New York. Poster Session Objective: To evaluate the safety and efficacy of hysterectomy vs cone biopsy in HIV patients with carcinoma in situ (CIS) of the cervix. Methods: We performed a retrospective case– control study of all HIV(1) patients diagnosed with CIS of the cervix between 1989 and 1995, using a university hospital tumor registry and other computer databases. Results: There were 276 patients with CIS; 27 were HIV positive (9.8%). Nine were treated by hysterectomy, 16 were treated by cone biopsy, and 2 remained untreated. Overall, 56% of patients did not receive any follow-up Pap smear (44% of hysterectomy patients versus 64% of cone biopsy patients; P 5 0.32). CD4 count (hysterectomy, 233 6 171; cone biopsy, 186 6 172; P 5 0.64) and age (mean, 36 years) did not differ by treatment. According to Pap smear results, 60% (3/5) of hysterectomy patients were abnormal sometime after treatment compared to 67% (4/6) cone biopsy patients (P 5 0.9). Median time to recurrence was 12 months in hysterectomy patients versus 8 months in cone biopsy patients (P 5 0.49). Deaths occurred in 22% of hysterectomy patients, 31% of cone biopsy patients, and 50% of untreated patients, none due to cervical cancer. Median time from CIS to death was 27.5 months for hysterectomy patients and 11 months for cone biopsy patients (P 5 0.53). There were no complications in the hysterectomy group; however, 2 patients were readmitted

102. The Prognostic Significance of Lymphvascular Space Invasion (LVSI) and Adjuvant Radiotherapy in Stage IB Node-Negative Cervical Cancer. Alaa (Al) Elbendary, Thomas Herzog, Perry Grigsby, John Pfeifer, and David Mutch. Washington University, St. Louis, Missouri. Plenary Session Objective: To assess the prognostic significance of (LVSI) and the impact of adjuvant radiotherapy in node-negative stage IB cervical cancer. Methods: We retrospectively reviewed all 334 women surgically treated for stage IB disease at our institution between January 1986 and December 1995. Three hundred thirty-four patients underwent radical hysterectomy and lymphadenectomy. Information regarding LVSI, presence of lymph node metastasis, tumor size, tumor histology, depth of invasion, grade, positive surgical margins, parametrial disease, and adjuvant radiotherapy was extracted from the medical records. LVSI was correlated with tumor size, grade, and depth of invasion. The impact of surgicopathological risk factors on disease recurrence was examined using multivariate analysis. The influence of radiotherapy on the incidence and site of recurrence was examined. Results: Three hundred of 334 pts were eligible for this study. LVSI was identified in 126 pts, whereas 174 pts had no LVSI. In the group with LVSI, 51 pts (40%) had positive lymph nodes compared to 22 (12%) in the group without LVSI (P , 0.001). One hundred fifty-two patients with no nodal disease nor LVSI were compared to 75 pts with LVSI and no nodal disease. The two groups were balanced with respect to age and race. Patients with LVSI had a higher incidence of grade 2 or 3 tumors (92% vs 46% P 5 0.03), deep invasion (56% vs 18%, P 5 P 5 0.017). Compared to tumors with no LVSI, tumors with LVSI recurred more frequently (21% vs 7% P 5 0.001) with a tendency for distant recurrence (54%

98

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

vs 36%). Multivariate analysis confirmed LVSI and bulky lesions but not grade nor depth of invasion as independent predictors of recurrence. Adjuvant radiotherapy was prescribed more frequently in patients with LVSI (27% vs 11%, P 5 0.002). In patients with no LVSI, adjuvant radiotherapy did reduce the risk of local recurrence; however, there was no reduction in the overall incidence of recurrence. For patients with LVSI, no significant reduction in local or overall recurrence rates could be demonstrated with adjuvant radiotherapy. Conclusion: Our results confirm LVSI as a predictor for nodal metastasis as well as recurrence. Tumors with LVSI are more aggressive and carry a worse prognosis; however, adjuvant radiotherapy does not appear to reduce the risk of recurrence. 103. Pelvic Lymphatic Fluid BUN/CR: A Possible Screening Test for Intraperitoneal Urinary Leakage. James Fanning and Kelly J. Manahan. Medical College of Ohio, Toledo, Ohio. Poster Session Introduction: Frequently, postoperatively following radical hysterectomy with lymphadenectomy there is increased pelvic suction catheter output with resultant decreased urinary output. The major differential is increased pelvic peritoneal/lymphatic output vs urinary leakage from the bladder or ureters. Since there are no standard laboratory values for BUN/CR in pelvic lymphatic fluid, we wish to prospectively evaluate these levels to determine if standard values could be defined. If standard values are defined, evaluation of lymphatic fluid BUN/CR may prove to be a valuable screening test to differentiate excessive lymphatic output vs urinary leak. Methods: We prospectively evaluated 16 consecutive patients undergoing radical hysterectomy with lymphadenectomy. No patient had clinical evidence of intraperitoneal urinary leak. On POD 2 and 3, serum, urine, and pelvic suction catheter samples were evaluated for BUN/CR. Results were compared using Mann–Whitney U. Results: Median BUN was serum 11, lymphatic fluid 11, and urine 469. Median CR was serum .9, lymphatic fluid 1.0, and urine 138. Intrapatient serial levels of serum, lymphatic fluid, and urine BUN/CR from postoperative days 2 and 3 showed no significant difference (P 5 0.31). Lymphatic fluid BUN/CR was essentially identical to serum BUN/CR and statistically significantly lower than urine BUN/CR (P , 0.0001). Conclusions: There was strong intrapatient consistency of serial lymphatic BUN/CR. Lymphatic fluid and serum BUN/CR were essentially identical. We conclude that following radical hysterectomy with lymphadenectomy, pelvic lymphatic fluid BUN/CR standard values are the same as serum levels. Since urine BUN/CR are 40 to 100 times greater than pelvic lymphatic fluid BUN/CR, evaluation of pelvic fluid, BUN/CR may prove to be a valuable screening test to differentiate excessive lymphatic output vs urinary leak. 104. Prospective Trial of Aggressive Postoperative Bowel Stimulation Following Radical Hysterectomy in an Attempt to Decrease Hospital Stay. James Fanning and Sandia S. Yu. Medical College of Ohio, Toledo, Ohio. Introduction: In patients treated with radical hysterectomy, the major deterrent of early hospital discharge is slow return of bowel function. We performed a prospective trial of aggressive postoperative bowel stimulation in an attempt to hasten return of bowel function. Methods: We prospectively treated 16 consecutive patients undergoing radical hysterectomy with bowel stimulation consisting of milk of magnesia (30 cc tid) and bisacodyl (1 per rectum qd) starting on POD 1. We evaluated time to flatus, bowel movement, diet, discharge, and complications. Comparison was made to 40 similar patients from a previously reported prospective trial who were not treated with bowel stimulation. Results were compared using the Mann–Whitney U. Results: Median age was 46 years. Median weight was 79 kg. Median time to flatus was 3 days, BM 3 days, diet 3 days, and discharge 4 days. No patient developed ileus or bowel obstruction. Time to discharge in our previous trial without bowel stimulation was 8 days. This difference was statistically significant (P 5 ,0.0001). Conclusion: Aggressive postoperative bowel stimulation following radical hysterectomy resulted in early return of bowel function

and no morbidity. Compared to our previous trial without bowel stimulation, patients were discharged a median of 4 days earlier. 105. The Pre- and Postoperative Value of the Endocervical Curettage in the Detection of Cervical Intraepithelial Neoplasia and Invasive Cervical Cancer. Bruce A. Fine and Glen Feinstein. UTHSCSA, San Antonio, Texas. Poster Session Objective: To evaluate the pre- and postoperative value of endocervical curettage in the detection of cervical intraepithelial neoplasia and invasive cervical cancer. Methods: A retrospective study of patients undergoing cervical conization was performed to evaluate the correlation of the grade of the preoperative endocervical curettage (ECC) and the grade of dysplasia in the conization specimen as well as the incidence of invasion in the cone specimen with respect to use of preoperative ECC. In addition, an evaluation of the role of routine preoperative ECC in satisfactory and unsatisfactory colposcopy and of the need for routine ECC in the detection of postoperative residual dysplasia was carried out. Results: Two hundred ninety-seven of 391 (76%) patients had an ECC as part of preoperative assessment for cervical dysplasia on Papanicolaou smear. When evaluating the association of grade of preoperative ECC with respect to the grade of conization specimen, there was a weighted k of 0.135 (P 5 0.0003). There was perfect agreement in 36.7% of cases and agreement within one grade in 60.4% of cases. Of 17 patients with invasive disease on conization specimen who had a preoperative ECC, only patients with a positive ECC had invasion at conization. None of 113 patients with a negative preoperative ECC had invasive disease on their conization specimens. One hundred forty-eight of 184 patients with a negative ECC had colposcopic examination and 63 (76.8%) had satisfactory colposcopy. The proportion of satisfactory colposcopic examinations between positive and negative ECC is not significantly different (P 5 0.673). Follow-up of patients with positive margins of resection at the time of conization was performed with a Papanicolaou smear only in 20 patients and with a pap smear and ECC in 53 patients. In the latter group, only 2/45 (4%) had a positive ECC with negative pap smear. This study group was too small to show statistical significance (P 5 0.310). Conclusions: There is good agreement between the grade of dysplasia on the preoperative ECC and on subsequent conization specimen. All patients with invasive disease on conization had positive preoperative ECC and no patients with a negative preoperative ECC had invasive disease. Therefore, the preoperative ECC is a good predictor of conization findings. Colposcopic examination was not a good predictor of pathology in the endocervical canal. Routine ECC should be a part of the preoperative assessment of an abnormal Papanicolaou smear; however, routine postoperative ECC may be unnecessary in the evaluation for residual dysplasia in patients with positive margins at conization. 106. Treatment Outcomes of Cervical Cancer Patients Treated with High Dose Rate Brachytherapy. Ihn Han,* P. J. Chuba,* F. Shamsa,* Adnan R. Munkarah,† Robert T. Morrris,† Carl W. Christensen,† John M. Malone Jr.,† Gunter Deppe,† and Arthur T Porter.* *Wayne State University/Harper Hospital, Detroit, Michigan; and; †Wayne State University/Hutzel Hospital, Detroit, Michigan. Objective: Over the past decade, high-dose rate (HDR) brachytherapy combined with external beam radiation has become widely used for the treatment of cervical cancer. The goal of this study was to determine the long-term efficacy and toxicity of this treatment modality. Materials: We identified 131 patients with cervical cancer who were treated at our institution with external radiotherapy and HDR brachytherapy since 1987. From a retrospective review of the medical records, patients’ demographics, clinical information, and follow-up data were analyzed. Results: The mean age was 54 years. Patient distribution by FIGO clinical staging was 33 stage IB, 8 stage IIA, 49 stage IIB, 4 stage IIIA, 35 stage IIIB, and 2 stage IBA. In 122 patients brachytherapy was delivered using the HDR technique and consisted of 8 –12 outpatient applications (3.86 cGy per fraction, three fractions per week). In the remaining 9 patients, the brachytherapy was a combination of low-dose rate

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS cesium and HDR techniques. After a median follow-up of 47 months, 50% patients were alive with no evidence of disease, 2% were alive with disease; and 36% have died of cervical cancer. The sites of disease recurrence were pelvis (38%), distant (53%), and both local and distant (9%). Only two severe treatment complications (higher than grade 2 GOG) were encountered. Conclusion: HDR brachytherapy is a safe and effective alternative to low-dose rate brachytherapy for the definitive treatment of cervical cancer patients. 107. Prognostic Factors Influencing Survival of Cervical Cancer Patients Treated with Radiation Therapy. Ihn Han,* P. Chuba,* F. Shamsa,* Adnan R Munkarah,† Gunter Deppe,† and Arthur A Porter.* *Wayne State University/Harper Hospital, Detroit, Michigan; and †Wayne State University/Hutzel Hospital, Detroit, Michigan. Objective: The goal of our study was to identify prognostic factors influencing the survival of cervical cancer patients treated with conventional external beam radiation and low-dose rate (LDR) brachytherapy. Methods: We identified 216 cervical cancer patients treated with external beam radiation therapy and LDR brachytherapy at our institution in the time period 1980 – 1990. From a retrospective review of the medical records, patients’ demographics, clinical information, and follow-up data were analyzed. Results: Our study population had the following patients characteristics: 19% were under age 40, 45% were 41– 60, and 36% older than 60; 52% were Caucasian, 46% African–American, and 2% belonged to other races. The patient distribution by FIGO clinical staging was 20.8% stage IB, 34.3% stage II, 42.5% stage III, and 2.3% stage IVA. Histologic evaluation revealed squamous cell carcinoma in 91.2% of patients, adenocarcinoma in 5.5%, adenosquamous in 0.9%, and other 2.3%. The actuarial disease-free survival for all patients was 60% at 5 years and 55% at 10 years. After a median follow-up of 114 months, disease recurrence was diagnosed in 40% of patients. A stepwise multivariate analysis revealed that age, race, stage, and treatment time had an impact on survival. The worst prognostic factors were younger age (Conclusion: In addition to stage and overall radiation treatment times, our study shows that age and race are independent prognostic factors affecting the survival of cervical cancer patients treated with radiation therapy. 108. Lymph Node Evaluation in Squamous Cell Carcinoma of the Cervix: Touch Imprint Cytology versus Frozen Section Histology. Annette Hasenburg, Suzanne C. Ledet, Thuy Ardaman, Tallia Levy, and Dirk G. Kieback. Baylor College of Medicine, Houston, Texas. Plenary Session Objective: Pelvic lymph node (LN) involvement is the most important negative predictor of survival in early-stage cervical cancer and may exclude the patients with recurrence from exenteration . We perform extraperitoneal LN dissection with frozen section analysis prior to radical surgery. Nonsurgical treatment options are chosen in case of LN metastases. To reduce operating time, alternative diagnostic methods to assess LN involvement could be of value. Preparation and analysis of frozen sections takes 10 –15 min for each LN, whereas a touch imprint cytology requires 5 min or less. A prospective study was performed to determine the usefulness of cytologic imprint technique vs frozen section vs paraffin histology. Methods: Three hundred eighteen pelvic and paraaortic LN from 32 patients with stage I–IV cervical cancer were bisected and submitted for frozen section after touch imprints had been prepared. Results: Twenty-nine nodes (9.1%) revealed metastatic squamous cell carcinoma (SCC) by frozen section histology. Twenty-six of these 29 were diagnosed by touch imprint and confirmed histologically. Reasons for the 3 false negatives included inadequate preparation of the touch imprint and the presence of small metastases within the lymph node parenchyma deep to the bisected surface. Permanent histology always agreed with the frozen section result. Touch imprint evaluation of pelvic LN for metastatic squamous cell carcinoma at the time of intraoperative consultation had a sensitivity and specificity of 90 and 100%. Conclusions: Adequate imprint preparation is very important and multiple sections should be made through thick LN. With those precautious in mind, touch imprints may provide a sensitive, specific, and time-efficient method to diagnose LN metastases in SCC of the cervix.

99

109. Liposuction-Assisted Nerve-Sparing Extended Radical Hysterectomy: Oncologic Rationale, Surgical Anatomy, and Feasibility Study. Michael Hoeckel,* Fritz Casper,* Hendryk Pilch,* Paul G. Knapstein,* Claus P. Heussel,† and Moritz A Konerding.‡ *Department of Ob/Gyn, Mainz, Germany; †Department of Radiology, Mainz, Germany; and ‡Department of Anatomy, Mainz, Germany. Poster Session Objective: To improve the therapeutic index of radical hysterectomy by extending the resection of parametrial tissue without further impairing pelvic autonomic nerve functions. Methods: We studied the sectional anatomy of normal parametrial tissue, residual parametrial tissue remaining in situ after conventional type III radical hysterectomy, and the location of pelvic wall recurrences with high-resolution MRI. Next, we investigated in fresh female human cadavers the topographic relationship of the pelvic autonomic nerve system to the parametrial tissues. Finally, we performed a clinical feasibility study of the liposuction-assisted nerve-sparing extended radical hysterectomy. Results: MRI demonstrated that perispinous adipose tissue (PSAT) is retained in situ after conventional type III radical hysterectomy and is a frequent site of tumor relapse. Liposuction-assisted dissection studies in five cadavers revealed that PSAT contains the pelvic plexus and the pelvic splanchnic nerves. Based on these findings we developed the liposuction-assisted nerve-sparing extended radical hysterectomy. The new surgical procedure has been applied in eight consecutive patients with cervical/vaginal cancer. No intra- and postoperative complications occurred. Postoperative MRI assured that PSAT was cleared in all cases. A metastatic lymph node was identified in the PSAT removed by liposuction in one patient. Suprapubic cystostomies could be removed after a median period of 10 days (range 7–30 days) postoperatively. Conclusion: The liposuction-assisted nerve-sparing extended radical hysterectomy is feasible. The removal of PSAT in addition to the en bloc parametrectomy in anatomically defined planes is possible without further compromising the bladder function as evaluated preliminarily. 110. Effects of b-Carotene on Cervical Intraepithelial Neoplasia (CIN): A Phase II Trial. Kristin A. Keefe,* Sharon Wilczynski,† Westley Lagerberg,* Julia Chapman,‡ Cheryl Brewer,* Sandhya Upasani,* Michael L Berman.* *UCI Medical Center, Orange, California; †City of Hope Medical Center, Duarte, California; and ‡University of Kansas Medical Center, Kansas City, Kansas. Poster Session Objective: Studies suggest that b-carotene (C) may prevent cervical cancer. We are evaluating the effect of C taken over a 2-year period on CIN 2 and 3 lesions. HPV typing by PCR was carried out to determine if lesion regression was related to HPV type. This report summarizes our 6-month findings. Methods: One hundred eleven women were randomized to receive C (30 mg daily) or placebo, with evaluations every 3 months by cytology, colposcopy, and cervicography. Colposcopically directed biopsies were performed on all women before therapy (tx) and at 6 months to evaluate treatment response. Women with persistent CIN 3 or whose lesions progressed to CIN 3 were removed from the study, while tx continued for 2 years on all others. Response was defined as an improvement in CIN by 2 grades. Regression of CIN was analyzed using a Mantel–Haenszel x2 test stratified by CIN. Fisher’s exact test was used to analyze the effect of HPV types and degree of CIN response to tx. Results: The mean age of the 111 studied was 29.8 years. Eight patients whose biopsies were only CIN 1 were excluded leaving 103 for analysis. Ten patients withdrew prior to 6 months and were considered nonresponders. The overall response rate was 47%. Response was similar in both treatment arms when stratifying for CIN status (P 5 0.84). Data on 93 women who had HPV typing at enrollment and at 6 months revealed that 81 (87%) were HPV positive with 68% possessing high-risk types (16, 18), 22% intermediate risk (31, 33, 35, 39, 52, 56), and 10% low risk (6, 11, 53) or unknown. No significant difference in subtypes was found across tx arms (P 5 0.51) or response rates (P 5 0.75). A significant relationship was found between HPV and CIN grade (P 5 0.05) with high-risk types associated with CIN 3. Conclusion: b-Carotene does not enhance regression of CIN at 6 months. HPV type was associated with CIN grade but not response to treatment.

100

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

111. Weekly Cisplatin Chemotherapy during Irradiation Improves Survival and Reduces Relapses for Patients with Bulky Stage Ib Cervical Cancer Treated with Irradiation and Adjuvant Hysterectomy: Results of a Randomized GOG Trial. Henry M. Keys,* Brian N. Bundy,† Frederick B. Stehman,‡ Laila I. Muderspach,§ Weldon E. Chafe,¶ Charles L. Suggs,i Joan L Walker,** and Deborah Gersell.†† *Albany Medical College, Albany, New York; †Gynecologic Oncology Group, Buffalo, New York; ‡Indiana University Medical Center, Indianapolis, Indiana; §University of Southern California, Los Angeles, California; ¶Eastern Virginia Medical School, Norfolk, Virginia; iUniversity of Mississippi, Jackson, Mississippi; **University of Oklahoma, Oklahoma City, Oklahoma; and ††Washington University School of Medicine, St. Louis, Missouri. Plenary Session Objective: To evaluate the role of concomitant weekly cisplatin chemotherapy with irradiation followed by adjuvant hysterectomy in patients with bulky stage IB cervical cancer. Methods: Between February 24, 1992 and April 7, 1997, 374 patients with bulky (4 cm or larger) cervical cancers were entered onto protocol and randomly assigned to be treated with irradiation (external and intracavitary) and adjuvant hysterectomy with (Rt 1 Cis) or without (Rt only) weekly cisplatin during irradiation. Forty-three percent were classified as having ‘‘barrel-shaped’’ cervices, while 57% were nonbarrel bulky cancers. Squamous histology accounted for 80% of cases. The radiotherapy was the same for both regimens. Cisplatin was to be delivered once weekly during the external radiation course at 40 mg/m2 (max dose 70 mg/week) for 6 weeks. Hysterectomy was due 3– 6 weeks after Rt. Results: Residual cancer in the hysterectomy specimen was significantly different between the Rt only and Rt 1 Cis groups (57 and 47%, respectively). Both survival and recurrence-free interval (RFI) were statistically significantly better in the Rt 1 Cis arm (P , 0.05 for both). The risk of relapse for Rt 1 Cis was reduced to about half that seen in the control arm (RR 5 0.53), while the risk of death had a similar reduction (RR 5 0.52). Survival at 24 months was 89% vs 79%, and RFI was 81% vs 69%. All 52 patient deaths are due to cancer recurrence. Grade 3 and 4 hematologic and GI toxicities were more frequent in the Rt 1 Cis group, while other toxicities were equivalent in both treatment groups. Conclusion: Weekly Cisplatin chemotherapy during external irradiation improved RFI and S for patients treated with combined irradiation and adjuvant extrafascial hysterectomy for bulky stage IB cervical cancer. 112. Incidence and Mortality Rates from Invasive Adenocarcinoma of the Cervix in Young Women in an HMO Population: Can More Intensive Cytologic Screening Be Justified for Oral Contraceptive Users? Walter Kinney,* Hai-Yen Sung,* Kathy Kearney,* Marie Miller,* George Sawaya,† and Robert Hiatt. *TPMG, Oakland, California; and †UCSF, San Francisco, California. Objective: To evaluate all cases of invasive adenocarcinoma of the cervix in a large HMO, and estimate the burden of incident and fatal disease in younger women, thereby providing context for the assertion that oral contraceptive (OC) users, who may be at higher risk of adenocarcinoma of the cervix, should have Pap smears more frequently than current practice. Methods: Medical records of all members of our prepaid health plan developing invasive adenocarcinoma of the cervix during the period from 1988 to 1994 were reviewed and details of patient interaction with the medical care system, diagnosis, treatment, and outcome were abstracted. Results: In the period 1988 –1994, comprising approximately 8,497,000 woman/years of membership, there were 144 cases of invasive adenocarcinoma of the cervix. Of these, 4/144 (2.8%) occurred in women under the age of 30, 35/144(24.3%) occurred in women ages 30 –39, and the remaining 105/144 (72.9%) occurred in women 40 and older. Cases were evenly distributed over the study period and no trend toward an increased number of cases in young women was noted. At a median follow-up of approximately 4 years from diagnosis, 31/144 patients (21.5%) were dead of disease. There were 0 deaths under the age of 30, 1/31 deaths (3%) ages 30 –39, and 30/31 deaths (97%) occurred in women ages 40 and older. During this same time period (1988 –1994) approximately 125,000

women per year filled prescriptions for oral contraceptives. Conclusions: Invasive adenocarcinoma of the cervix is rare in younger women in this population. Given this low incidence, it seems unlikely that more frequent cytologic screening of current OC users would be cost effective for prevention of invasive adenocarcinoma of the cervix. 113. The ASCUS Pap Smear as a Harbinger of Invasive Cervical Cancer (ICC). Walter Kinney,* Hai-Yen Sung,* Kathy Kearney,* Marie Miller,* George Sawaya,† and Robert Hiatt.* *TPMG, Oakland, California; and †UCSF, San Francisco, California. Objective: To assess the risk of ICC within 6 months following an ASCUS pap. Methods: Medical records of all members of our prepaid health plan developing invasive cervical cancer during the period from 1988 to 1994 were reviewed and details of patient interaction with the medical care system including clinic visits, symptoms, Pap smear results, diagnostic procedures, tumor histology and FIGO stage, treatment, and outcome were abstracted. Because the Bethesda system of cytologic interpretation was not uniformly utilized during this time period, those smears whose result would have been categorized as ASCUS under the Bethesda system (‘‘inflammation,’’ ‘‘cervicitis,’’ ‘‘squamous metaplasia,’’ ‘‘reactive changes,’’ and ‘‘atypia, not otherwise specified’’) are included in the ASCUS category. Smears with any other abnormal result were excluded. Results: Five hundred twenty-five of 649 patients diagnosed with ICC had one or more smears in the 6 months preceding the diagnosis. One hundred nine of 525 (21%) had an ASCUS smear in the 6 months preceding the diagnosis of ICC. An ASCUS smear was the first abnormal in the 6 months prior to diagnosis in 18/525 (3.4%). If one assumes a 4% ASCUS rate, there were an estimated 126,000 ASCUS smears in KPNCR during the years 1988 –1994. Risk of diagnosis of ICC within 6 months of an ASCUS smear is estimated at .0865% or 1/1156. Conclusion: Risk of diagnosis of ICC within 6 months of an ASCUS smear is small but real. Proposed follow-up or triage algorithms for women with ASCUS smears should take this risk into account. 114. Cytologic Antecedents of Fatal Invasive Cervical Cancer (ICC): Relative Contribution of ‘‘False Negative’’ Smears. Walter Kinney,* Hai-Yen Sung,* Kathy Kearney,* Marie Miller,* George Sawaya,† and Robert Hiatt. *TPMG, Oakland, California; and †UCSF, San Francisco, California. Objective: To assess the contribution of ‘‘false-negative’’ Pap smears to death from cervical cancer in an HMO population. Methods: Medical records of all women developing ICC during the period from 1988 to 1994 who were members of our prepaid health plan continuously for the 3 years prior to diagnosis were reviewed, and diagnosis, treatment, and outcome were abstracted. Assessment of screening history was limited to the period from 36 to 3 months prior to diagnosis to avoid including smears which were part of the evaluation that led to the recognition of ICC. Results: During the 7-year study period there were 8,497,000 woman-years of membership and 3,146,000 Pap smears were performed; 455 women meeting the criteria of 3 years of continuous membership were diagnosed with ICC. At a median follow-up of 4 years from diagnosis, 103/455 patients (22.6%) diagnosed with ICC were dead of their malignancy. In the period from 36 to 3 months preceding the diagnosis of fatal invasive cervical cancer, 65/103 patients (63.1%) dying of ICC did not have a Pap smear within our health plan, 11/103 (10.7%) had at least one abnormal Pap mandating follow-up care, and 27/103 (26.2%) had only normal Paps or minimally abnormal Paps which would not have engendered follow-up (‘‘metaplasia’’, for example). Among these 27 patients will be those with interval cancers (occurring after the last Pap) and those with false-negative smears. The average age at death for the fatal cases was 58 years old. Conclusions: The relatively small contribution of false-negative smears to death from ICC suggests that expensive new technologies intended to reduce false-negative rates will have to provide benefits in addition to reduction in mortality (such as contribution to ASCUS triage, reduction in medicolegal risk) to be cost effective.

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS 115. A Randomized Trial of Chemotherapy versus Radiation Therapy versus Observation after Radical Hysterectomy in Stage IB–IIB Cervical Cancer Patients with Pelvic Lymph Node Metastases and/or Vascular Invasion. Manfred Lahousen, Josef Haas, Hellmuth Pickel, and Raimund Winter. University of Graz, Graz, Germany. Objective: Adjuvant treatment modalities after hysterectomy have long been used in an attempt to eradicate microscopic tumor residuals in patients at high risk for recurrence. However, it has not been clearly demonstrated that adjuvant radiation or in combination with chemotherapy or adjuvant chemotherapy alone improves the outcome. Methods: To evaluate the effect of an adjuvant treatment the Austrian Gynecologic Oncology Group conducted between 1989 and 1995 a clinical trial in 76 patients with stage Ib–IIb cervical cancer treated with radical hysterectomy. Patients with pelvic lymph node metastases and/or vascular invasion received either adjuvant chemotherapy consisting of carboplatin (400 mg/m2) and bleomycin (30 mg) or standardized external radiation therapy to the pelvis or no further treatment. Results: After 5 years (median 3.7 years) no statistically significant differences in survival time have been identified between the three randomization treatment arms. Conclusions: The data suggest that the most important treatment factor in patients with cervical cancer and lymph node and/or vascular invasion is radical abdominal hysterectomy and systematic lymphadenectomy. Adjuvant therapy seemed to affect neither survival 116. Efficacy of Extrafascial Hysterectomy after Radiation Therapy for Stage Ib2 Cervix Carcinoma. Frank T. Lake, James B. Hall, and Robert V. Higgins. Carolinas Medical Center, Charlotte, North Carolina. Objective: To evaluate the efficacy of preoperative radiation therapy and extrafascial hysterectomy in women with stage IB2 cervix cancer. Methods: The study population consisted of all patients diagnosed with stage IB2 cervix cancer treated at one institution with preoperative radiation therapy (RT) followed by an extrafascial hysterectomy (EH). Pretreatment surgical staging was not performed. Preoperative therapy was designed to deliver a total radiation dose of 7500 cGy to point A and 5500 cGy to point B. External radiation (ER) was given at a rate of 180 –200 cGy/day. Two weeks after completing ER, all patients received a cesium intracavitary implant. EH was performed 4 – 6 weeks after completion of RT. Univariate and multiple analysis were used to test the independent variables related to disease recurrence. A P value of #0.05 was considered statistically significant. Results: Fiftythree women were enrolled in the study. The average age was 41.1 years (22.7–77.9), 64% were Caucasian, 30% were African–American, and 6% were Asian. Fourteen of 53 women (26.4%) developed recurrence; mean time to recurrence was 25 months (1.2–34). Patient’s age, gravidity, parity, race, smoking history, weight, cervical size, and histologic cell type were not predictive of recurrent cancer. There was no statistically significant difference between the two groups in the number of days to complete RT. The presence of tumor in the hysterectomy specimen strongly predicted recurrence of disease (P 5 0.001). Evaluation of the patients after treatment, using the National Cancer Institute toxicity criteria, indicated that no patient experienced more than a grade 2 toxicity. Conclusion: EH after preoperative RT is a well-tolerated and effective method to treat stage IB2 cervix cancer. Presence of tumor in the hysterectomy specimen is highly predictive of disease recurrence. 117. Cardiac Metastasis from Carcinoma of the Cervix, Report of Two Cases. Julio Fernando Lemus,* Gazi Abdulhay,† Craig Sobolewski,† and Victor R. Risch.† *University of Miami, Miami, Florida; and †Lehigh Valley Hospital, Allentown, Pennsylvania. Background: The presence of cardiac metastasis from cervical carcinoma is extremely rare. The diagnosis is made almost exclusively postmortem. There are few cases of premortem diagnosis, and it is believed that when cardiac metastasis is found, the prognosis is extremely poor. Due to the rarity of this condition it is very difficult to standardize care for these patients. Considering

101

the evidence provided by the cases in this report, it is possible that aggressive therapy may lengthen patient survival and quality of life. Case: We present two cases of cervical carcinoma with metastases to the heart. Both patients presented with symptomatology of cardiac tamponade. Both patients had invasion of the myocardium where the prognosis is even worse. We took an aggressive therapeutic approach to our patients and had excellent results in one. Our report includes the longest survival reported for a patient to date with premortem diagnosis of intramyocardial metastasis from cervical carcinoma. Conclusion: We concluded that the prognosis for cardiac metastasis from cervical carcinoma is extremely poor. The stage of the disease at initial presentation does not predict the future development of cardiac metastasis. Taking an aggressive therapeutic approach, including thoracentesis, chemotherapy, and radiation therapy to the heart, survival and quality of life can be improved. 118. A Phase II Trial of CPT-11 in Chemotherapy-Naive Patients with Recurrent Squamous Carcinoma of the Cervix (SCC): A Gynecologic Oncology Group (GOG) Study. Katherine Y. Look,* John A. Blessing,† Charles Levenback,‡ Matthew F. Kohler,§ Weldon E. Chafe,¶ and Lynda D. Roman.i *Indiana University Medical Center, Indianapolis, Indiana; †Roswell Park Cancer Institute, Buffalo, New York; ‡M. D. Anderson Cancer Center, Houston, Texas; §UC-Irvine, Orange, California; ¶Eastern Virginia Medical School, Norfolk, Virginia; and iWomen & Childrens, Los Angeles, California. Poster Session Objective: To report the response rate and associated toxicity of weekly CPT-11 in previously untreated patients (pts) with recurrent SCC. Methods: From October 1994 to May 1996, the GOG conducted a phase II trial in pts with recurrent SCC using weekly intravenous CPT-11 at 125 mg/m2 times 4 weeks with a 2-week rest to be repeated until disease progression or unacceptable toxicity. Eligibility criteria were a GOG performance status of 0 –2, adequate bone marrow reserve, adequate liver function, serum creatinine ,2 mg% and a signed informed consent. None of the patients had received prior chemotherapy other than radiation sensitizers. Standard GOG toxicity and response criteria were used. Results: Fifty-four patients were entered into the trial. Three patients were ineligible because of wrong cell type [n 5 2] or inadequate pathology material [n 5 1]. Two were inevaluable for toxicity because of inadequate trial of drug. Another 4 pts were inevaluable for response because of insufficient assessment of response. Thus, 49 pts are evaluable for toxicity and 45 are evaluable for response. Prior radiotherapy had been given to 41 pts. The median age of patients was 45 years (range 29 –71). The median number of weekly doses delivered was 7 (range 1– 46). The overall response rate was 13.9% (6/45). The incidence of grade 4 neutropenia and anemia was 6.1 and 4.1%, respectively. Nineteen patients (38.8%) developed gastrointestinal toxicity (GI) including 8 with grade 3 and 11 with grade 4 severity. There was 1 pt death from GI toxicity. There was 1 complete response of 8.8 month duration outside the pelvis and 5 partial responses of which 3 were extrapelvic and 2 pelvic. Conclusion: This schedule of CPT-11 exhibits activity with moderate toxicity. A schedule of cisplatin combined with weekly CPT-11 should be studied in this patient population. 119. Modified Radical Hysterectomy in the Treatment of Early Squamous Cervical Cancer. Javier F. Magrina, M. Goodrich, T. Lidner, A. Weaver, J. Cornella, and K. Podratz. Mayo Clinic–Scottsdale, Scottsdale, Arizona. Objective: To evaluate the results of modified radical hysterectomy in the treatment of early cervical cancer. Material and Methods: Retrospective chart review of 56 patients with Stage I (IA, 35; IB, 21) squamous cervical carcinoma treated by modified radical hysterectomy and followed for a minimum of 5 years (mean 12 years, range 5.1–29) was conducted. All pathology slides were reviewed for tumor size, grade, depth of penetration, and lymphvascular invasion. Results: The mean depth of invasion was 0.5 cm (range 0.1–2.5 cm) and the mean tumor size was 1.1 cm (range 0.1–7 cm). There were only three (5.4%) patients with positive nodes. None of the patients with tumors #2 cm

102

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

in size had positive nodes, compared to an incidence of 33.3% for the group with tumors .2 cm in size. Two patients developed a recurrence, for a 5-year recurrence rate of 3.6%. There were 10 deaths during the entire follow-up period but only two related to cervical cancer. The disease-free and overall 5-year survival rates were 96.4 and 94.6%, respectively. The disease-free survival rate among the 47 patients with tumors #2 cm was 100% compared to 88.9% for the 9 patients with tumors .2 cm. Univariate analysis revealed stage, lymph node status, and tumor size as statistically significant prognostic factors for overall survival. Tumor grade, lymphvascular permeation and depth of invasion (1–3 mm vs .3 mm) were not statistically significant for overall survival. Conclusions: Modified radical hysterectomy is effective surgical therapy for patients with squamous cervical carcinoma #2 cm in size. The number of patients with tumors .2 cm is too small (n 5 9) to establish any valid conclusions. 120. Cone Biopsy Does Not Affect the Incidence of Complications from Radical Hysterectomy nor Adversely Influences Survival in Patients with Stage Ib Cervical Carcinoma. V. K. Malviya,* C. T. Figueras,* A. A. Dharmani,† and G. Deppe.† *Providence Hospital, Southfield, Michigan; and †Wayne State University, Detroit, Michigan. Objective: To determine if cone biopsy (bx) as a diagnostic procedure for invasive cervical cancer (CA Cx) (i) resulted in an increased incidence of complications from subsequent radical hysterectomy or (ii) influenced the overall survival of patients (pts) with stage 1B CA Cx. Methods: The medical records of 142 pts with stage 1B Ca Cx who underwent radical hysterectomy were retrospectively reviewed. Data were analyzed using x2 analysis and BMDP statistical software version 7.0. P , 0.05 was accepted as statistically significant. Results: Forty-eight patients underwent a cone bx to confirm the diagnosis of invasive cancer, whereas 79 did not have a cone bx. Fifteen patients were excluded from the analysis due to insufficient data. There were no differences between groups in age, race, history of smoking, or associated medical disease (P . 0.05). Thirty-two percent of patients who had undergone conization had deep invasion on final histopathology compared with 67% who did not have cone bx (P , 0.05). In addition, 19% of patients in the former group and 42% in the latter group had poorly differentiated cancer (P . 0.05). There were no difference between groups in the incidence of intraoperative or postoperative complications, EBL (1176 6 166 ml vs 1271 6 163 ml), and the mean overall survival (72.2 months vs 59.2 months) (P . 0.05). The time interval between the cone bx and surgical procedure ranged from 7 to 92 days (mean 29.5 6 2.6). Conclusions: Our preliminary results indicate that cone bx did not increase the incidence of surgical complications from radical hysterectomy. Although our data also suggest that cone bx did not influence overall survival, these data must be interpreted with caution since patients in the cone bx group had a lower incidence of deep invasion and poorly differentiated disease. 121. Effects of Cigarette Smoking and Race upon Recurrence Rate and Survival in Patients with Stage Ib Cancer of the Cervix. V. K. Malviya,* C. T. Figueras,* A. A. Dharmani,* and G. Deppe.† *Providence Hospital, Southfield, Michigan; and †Wayne State University, Detroit, Michigan. Poster Session Objective: Previous investigators have demonstrated an increased risk of cervical cancer (CA Cx) in cigarette smokers. In addition, survival patterns in patients (pts) with CA Cx have indicated poorer survival among black patients compared to their white counterparts. The specific aim of this study was to determine if a history of cigarette smoking or the race of the pt altered recurrence rates and progression-free interval (PFI), or increased the likelihood of postoperative complications in patients with Stage IB CA Cx undergoing radical hysterectomy. Methods: The medical records of 210 pts with Stage IB CA Cx who underwent a radical hysterectomy with or without lymphadenectomy were retrospectively reviewed. Patients with prior radiation therapy (RT) or chemotherapy were excluded. Overall, survival duration and PFI were computed. Data were analyzed using analysis of variance, unpaired Student’s t test, and x2 analysis as appropriate.

The generalized savage test statistic was used to test the difference between survival functions. P , 0.05 was accepted as statistically significant. Results: Thirty-nine patients were excluded from the study because of insufficient data (22), prior chemotherapy and/or RT (14), and race other than black or white (5). Ninety-five patients were smokers and 74 were nonsmokers. Sixty-five patients were black and 140 where white. There were no differences between groups with regard to age, parity, associated medical conditions, depth of invasion, or the incidence of angiolympatic invasion (P 5 NS). No significant differences in overall survival or postoperative complications among smokers and nonsmokers and between black and white patients were found (P . 0.05). Conclusion: Our data suggest that cigarette smoking or race does not adversely influence the survival or the recurrence rates in patients with Stage IB CA Cx undergoing radical hysterectomy. 122. Accuracy of the Bethesda System for Cervicovaginal Cytology: Correlation with Repeat Cytology and Colposcopic Biopsy. L. Stewart Massad,*,† Yvonne Collins,* and Helen E. Cejtin.* *Cook County Hospital, Chicago, Illinois; and †Rush Medical College, Chicago, Illinois. Poster Session Objective: To determine the reproducibility and colposcopic biopsy correlates of the Bethesda System (TBS) for cervicovaginal cytologic diagnosis. Methods: Data on referral cytology, cytology at colposcopy, and colposcopic biopsy were retrieved for 1226 women evaluated in the resident clinics of Cook County Hospital (n 5 775, July 1, 1996 to June 30, 1997) and Rush– Presbyterian–St. Luke’s Medical Center (n 5 451, October 1, 1992 to June 30, 1997). Cytology results were classified under TBS. Women with unknown referral cytology were excluded, and women with inadequate colposcopy or no or unknown biopsy result were excluded from biopsy comparisons. Results: Of 962 women with both available, referral and colposcopy cytology correlated exactly in 311 (32%) and within one grade in 716 (74%). Of 283 women referred for ASCUS, biopsy was normal or HPV in 172 (61%), CIN1 in 77 (27%), and CIN2-3 in 34 (12%). Of 352 referred for LSIL, biopsy was normal or HPV in 175 (50%), CIN1 in 122 (35%), and CIN 2–3 in 55 (16%). Of 128 referred for HSIL, biopsy was normal or HPV in 38 (30%), CIN 1 in 27 (21%), and CIN2-3 in 61 (48%), with cancer in 2. Of 153 women with ASCUS at the time of colposcopy, biopsy was normal or HPV in 76 (50%), CIN1 in 51 (33%), and CIN2-3 in 25 (16%), with cancer in 1. Of 155 with LSIL at colposcopy, biopsy was normal or HPV in 72 (46%), CIN1 in 47 (30%), and CIN2–3 in 36 (23%). Of 82 with HSIL at colposcopy, biopsy was normal or HPV in 22 (27%), CIN1 in 15 (18%), and CIN2–3 in 45 (55%). Conclusions: Under TBS, cytology results are often reproducible within one grade, but exact reproducibility is uncommon. CIN2–3 is infrequent among women with ASCUS or LSIL but is present in half of those with HSIL. 123. Fluorescence Spectroscopy: A Diagnostic Tool for Cervical Intraepithelial Neoplasia (CIN) Michele Follen Mitchell,* Shannon Abikhaled,† Anant Agrawal,‡ Nirmala Ramanujam,‡ Scott Cantor,* and Rebecca Richards-Kortum.‡ *University of Texas M. D. Anderson Cancer Center, Houston, Texas; †UT Health Science Center, Houston, Texas; and ‡UT MDACC Austin, Austin, Texas. Poster Session Objectives: The development of a cost-effective fluorescence system to reduce the number of patient visits by enabling diagnosis and treatment in a single visit. Methods: A portable fluorometer was developed and utilized to acquire fluorescence spectra from 381 cervical sites in 95 patients at 337, 380, and 460 nm excitation prior to colposcopy. A multivariate statistical algorithm was used to extract clinically useful information from tissue spectra acquired in vivo. The algorithm was compared to sensitivity and specificity of colposcopy. For cost analysis, we constructed a simple decision-analytic model for the diagnosis and management of CIN in a referral population. We compared ‘‘see and treat’’ LEEP based on spectroscopy and colposcopy used together at one visit. We then analyzed the costs and number of cases that were appropriately and inappropriately treated, missed, and not treated in a hypothetical cohort of 100 patients seen in a referral clinic. Results: A diagnostic algorithm that differentiates high-grade SIL

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS from non-high-grade was developed with a sensitivity and specificity of 79 6 2 and 78 6 2%, respectively. ‘‘See and treat’’ LEEP based on spectroscopy was the least expensive but least effective, costing $160,479 to accurately detect 31.5 cases of cervical precancer in 100 patients. The most expensive strategy was colposcopically directed biopsy at $311,808 to find 45.8 cases, but when both tests were used in a ‘‘see and treat’’ modality, more cases were found (46) at a cheaper cost ($285,133). The incremental cost-effectiveness of the joint strategy compared to the spectroscopy-only strategy was $8596 per case of cervical precancer detected. Conclusions: Fluorescence spectroscopy should be considered an important innovation in the diagnosis of CIN as demonstrated by its efficacy and economic advantage. 124. High Dose Rate Brachytherapy Is Equivalent to Low Dose Rate Brachytherapy in the Treatment of Stage Ib and II Cervical Cancer. Daniel G. Petereit, Julian C. Schink, David M. Potter, Elizabeth Grosen, Ellen M. Hartenbach, Jann Sarkaria, and Dolores A. Buchler, University of Wisconsin, Madison, Wisconsin. Poster Session Objective: To compare the efficacy and complications of high dose rate (HDR) versus low dose rate (LDR) brachytherapy in the treatment of patients with stage IB and II cervical cancer. Methods: One hundred twenty-three stage IB/II cervical cancer patients were treated with HDR brachytherapy between August 1989 and November 1996. These patients were compared to 141 stage IB/II patients treated with LDR brachytherapy between April 1977 and July 1988. Five HDR brachytherapy fractions were designed to deliver an LDR equivalent dose. All statistical endpoints were calculated by actuarial analysis using the log rank test. Any toxicity #RTOG grade 3 was analyzed. Results: (See table).

Three-Year Clinical Endpoints

Stage (patient number)

Survival (%)

RFS (%)

Pelvic control (%)

Complications (%)

Distant metastases (%)

86 82

85 81

85 91

16 9

8 7

65 58

69 61

80 78

14 9

17 25

103

lated with histologic and cytologic findings. Using histopathologic diagnosis as the reference standard, the sensitivity and positive predictive value (PPV) of Pap smear and high-risk (HR) HPV were calculated. The k test was used to correlate colposcopic and histopathologic findings. Results: Pap smears at the time of consultation demonstrated 25 pts with normal results, 39 with LGSIL, 30 with HGSIL, 1 SCC, and 5 ASCUS. Seventy-eight patients underwent cervical biopsy. Colposcopic findings significantly correlated significantly with histopathologic findings. Forty-four percent of patients tested positive for HPV DNA: 40 pts with HR HPV, 3 pts with low-risk (LR) HPV, and 1 pt with both HR and LR HPV. Sixteen patients (41%) with LGSIL on Pap smear tested positive for HR HPV; 37% of pts in this group required cervical conization because cervical biopsies demonstrated moderate/severe dysplasia. The diagnosis of moderate/severe dysplasia significantly correlated with the presence of HR HPV [OR 78.9 (8.31–389.30)]. There was no significant correlation between the HPV DNA signal strengths and the histologic grade of dysplasia. The sensitivity and the PPV of Pap smear alone in identifying moderate/severe dysplasia was 62 and 96%, respectively. The combination of HGSIL Pap smears and HR HPV increased the sensitivity but not the PPV for the detection of moderate/severe dysplasia to 77.7 and 95%, respectively (P 5 NS). Conclusion: The use of hybrid capture DNA testing does not significantly improve the sensitivity or PPV of the diagnosis of HGSIL when compared cytologically indicated, colposcopically directed cervical biopsies. 126. Preliminary Results of a Randomized Trial of Mitomycin C as an Adjunct to Radical Radiotherapy of Locally Advanced Cervix Cancer. Kenneth B. Roberts,* Nelson Urdaneta,† Raul Vera,‡ Andres Vera,§ Enrique Gutierrez,† Yadelis Aguilar,† Sara Ott,† Sara Rockwell,* Alan C. Sartorelli,* Diana B. Fischer,* James J. Fischer,* *Yale University School of Medicine, New Haven, Connecticut; †Central University of Venezuela, Caracas, Venezuela; ‡La Floresta Medical Institute, Caracas, Venezuela; and §Domingo-Luciani Hospital, Caracas, Venezuela. Poster Session

125. The Clinical Value of Digene Hybrid Capture DNA Testing in a ReferralBased Population with Abnormal Cervical Cytology. Fernando F. Recio, B. I. S. Srivastava, C. Wong, R. E. Hempling, G. H. Eltabbakh, N. Natarajan, and M. S. Piver. Roswell Park Cancer Institute, Buffalo, New York.

Objective: To determine the efficacy of mitomycin C (mito) with radiotherapy (RT) for the treatment of cervix cancer. Methods: A randomized study of mito combined with RT vs RT alone was initiated in 1990. As of April 1996, 160 patients (pts) had been enrolled. After excluding 10 pts for protocol violations, 150 pts having a mean follow-up of 35 months were analyzed as of May 1997. Intravenous mito, 15 mg/m2, was given on the first and sixth week of RT. Results: The 75 pts in the RT with mito group and 75 pts in the RT alone group have a comparable distribution by age and stage (mean age 47 years; stage IB 3%, IIA 11%, IIB 47%, IIIA 1%, IIIB 36%, IVA 3%). The 3-year actuarial survival rates for RT with mito and RT alone were 77 and 62%, respectively (P 5 0.16). The 3-year actuarial local recurrence-free survivals for patients receiving RT with mito and RT alone were 74 and 62%, respectively (P 5 NS). Complete response was seen in 66 pts (88%) of the RT with mito group and in 69 pts (92%) of the RT alone group (P 5 NS). Of those with an initial complete response, the 3-year actuarial local recurrence-free survivals for patients receiving RT with mito and RT alone were 84 and 68%, respectively (P 5 0.05). There were no treatment-related deaths. Mild hematologic toxicity was seen only in the group treated with mito, with 8 pts having a nadir white cell count , 2000/ml (but none below 1000/ml) and with 1 patient having a nadir platelet count below 25,000/l. No excess in nonhematologic toxicity has been observed thus far with combined mito and RT. Conclusion: In this open Phase III trial of mito as an adjunct to RT for cervix cancer there were minimal hematologic effects and no increase in acute radiation reactions. Thus far there are nonsignificant trends in favor of those receiving mito for survival and local control.

Objective: The purpose of this study was to evaluate the value of hybrid capture DNA testing in a referral-based patient population with abnormal cervical cytology. Methods: One hundred consecutive patients who were referred for evaluation of abnormal cervical cytology (ASCUS, LGSIL, HGSIL, or SCC) were evaluated by repeat Pap smear, hybrid capture HPV DNA analysis, and colposcopy. Colposcopic findings were recorded, and if appropriate, cervical biopsies were performed. Hybrid capture results were corre-

127. The Chemopreventive Agent Sulindac Modulates Expression of TGF-a and EGF-Receptor in the Uterine Cervix. Phillip Y. Roland,* Ronald D. Alvarez,† William E. Grizzle,† Russel B. Myers,† Mack Barnes,† Larry C. Kilgore,† and Edward E. Partridge.† *Northwestern University Medical School, Chicago, Illinois; and †University of Alabama at Birmingham, Birmingham, Alabama. Poster Session

Stage IB HDR (59 LDR (76) Stage II HDR (64) LDR (65)

Note. RFS, relapse-free survival. No statistically significant differences were identified between the 2 groups at 3 years at all clinical endpoints measured. Conclusions. HDR brachytherapy results in comparable survival, RFS, and pelvic control when compared to LDR brachytherapy for patients with stage IB and II cervical cancer. The slightly higher complications (P 5 ns) in the HDR patients were identified 3 years ago which resulted in a decrease of the HDR dose per fraction from 9 –10 to 5– 6.5 Gy/Fx. Since the literature supports the equivalency of both forms of brachytherapy, the HDR approach is more practical with elimination of both inpatient hospitalization and radiation exposure to personnel.

104

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

Objective: To conduct a prospective clinical trial evaluating the chemopreventive potential of sulindac in cervical dysplasia, as determined by changes in histology and molecular marker expression. Methods: Patients with an abnormal Pap test and colposcopic evidence of a dysplastic lesion were selected for study. Oral sulindac 150 mg. twice a day was administered for 8 weeks. Biopsies of each lesion were obtained before and after sulindac to assess for changes in histology and expression of the biomarkers TGF-a and EGFreceptor (EGF-R). Biomarker expression of dysplasia and normal tissue was characterized by an immunostaining score of basal, lower-, and upper-half cervical epithelium. The Wilcoxon signed-ranks test was used for comparison of scores before and after treatment. Results: A total of 23 cases were available for study. The majority of patients (82.3%) demonstrated no change in lesion histology before and after treatment. TGF-a was weakly detected in all tissues, with reduced expression in upper-half epithelial cells of normal and atypia lesions after sulindac administration. EGF-R cytoplasmic and membrane staining was detected in all tissues, with increased cytoplasmic expression detected in the lower-half epithelium of normal and atypia lesions after treatment. Conclusions: Sulindac is a NSAID which has demonstrated chemopreventive activity in epithelial neoplasia. There was no evidence of histologic regression of cervical dysplasia in the current study. However, alterations in biomarker expression suggest that sulindac has a direct effect upon the uterine cervix. In addition, TGF-a and EGF-R may be useful biomarkers in future chemoprevention trials targeting low-grade cervical lesions.

nectomy, WBC $3000/ml, platelet count $100,000/ml, serum creatinine #2 mg%, and adequate hepatic function. Radiation techniques and doses were defined. Chemotherapy doses (per m2) were as follows: Regimen I, hydroxyurea orally 3 g twice weekly; Regimen II, hydroxyurea orally 2 g twice weekly, 5-FU 1 g/day as a 96-h infusion days 1 and 29 1 cisplatin 50 mg days 1 and 29; and Regimen III, weekly cisplatin 40 mg/week. Results: From April 1992 to April 1997, 575 patients were enrolled. Fifty patients were ineligible, 44 of which were surgical ineligibilities. Four hundred eighty-seven patients are included in this analysis. Both platinum regimens, weekly cisplatin and HFC, have improved progression-free intervals when compared to hydroxyurea, P 5 0.004 and P 5 0.002, respectively, reducing the risk of progression by 43 and 44%, respectively. The percentage of patients recurrence free at 24 months was 70% for weekly cisplatin, 67% for HFC, and 50% for hydroxyurea. Grade 3 or 4 leukopenia and grade 4 GI toxicity was increased with HFC compared to weekly cisplatin or hydroxyurea (P 5 0.00001, P 5 0.02, respectively). Further follow-up is needed for survival analysis. Conclusions: Weekly cisplatin is more effective than hydroxyurea and more tolerable than HFC as a concomitant chemoradiation regimen for locally advanced cervical cancer.

128. The Chemopreventive Agent Sulindac Modulates Prostaglandin E2 (PGE2) Levels in Cervical and Vaginal Secretions. Phillip Y. Roland,* Ruiwen Zhang,† Y. Li,† Mack Barnes,† Larry C. Kilgore,† Edward E. Partridge,† and Ronald D. Alvarez.† *Northwestern University Medical School, Chicago, Illinois; and †University of Alabama at Birmingham, Birmingham, Alabama.

Objective: Previous studies by the GOG have demonstrated no activity with bolus etoposide in squamous cell carcinoma of the cervix. Prolonged oral etoposide, which exploits the schedule dependency of this agent, has demonstrated activity in non-small cell carcinoma of the lung and has been studied in combination therapy with cisplatin. To evaluate prolonged oral etoposide in previously treated squamous cell carcinoma of the cervix, the current phase II trial was conducted. Methods: Eligibility included squamous cell cancer of the cervix, measurable disease, 1 prior chemotherapy regimen which did not include etoposide, WBC $3000/ml, platelet count $100,000/ml, serum creatinine #2 mg%, and adequate hepatic function. The starting dose was 50 mg/m2/day (40 mg/m2/day for prior radiotherapy) for 21 days, every 28 days. Based on toxicity a dose escalation to a maximum dose of 60 mg/m2/day was prescribed. Results: Twenty-five patients were entered on this study; 24 are evaluable for toxicity. A median of 2 courses were given (range 1– 8). All but one had received radiation therapy and 20 had received prior chemotherapy. Oral etoposide was not well tolerated with grade 4 neutropenia occurring in 33.3% and grade 4 thrombocytopenia occurring in 15%. Eight patients received only 1 course of therapy. Of the remaining patients, 6 required dose reductions to 30 mg/m2/day. Only 3 patients were able to be dose escalated to 50 mg/m2/day. Seventeen patients are evaluable for response, of which 16 had received prior radiotherapy and 15 prior chemotherapy. Two partial responses (11.8%) were observed. Both of these patients had disease in nonirradiated sites and one was chemotherapy naive. Conclusion: Prior radiation therapy significantly limited our ability to deliver prolonged oral etoposide. At the dose tolerated this regimen is not significantly active in squamous cell carcinoma of the cervix and no benefit to the chronic oral schedule is evident.

Objective: To determine the pharmacokinetics of oral sulindac, a potential chemopreventive agent in dysplasia of the uterine cervix. Methods: Patients with abnormal Pap tests were administered sulindac 150 mg. orally twice each day for 8 weeks. Cervical–vaginal lavage (CVL) and blood plasma specimens were collected at baseline, 4 weeks, and 8 weeks. Concentrations of sulindac and its metabolites s-sulfone and s-sulfide were determined in CVL and plasma by HPLC, and levels of PGE2 were determined by ELISA. Results: A total of 21 patients were enrolled in the study. Mean plasma concentrations of sulindac and each metabolite increased in a time-dependent manner during the study (sulindac, 0.06 to 9.78 mg/ml; s-sulfone, 0.26 to 5.44 mg/ml and s-sulfide 0 to 2.21 mg/ml). Sulindac alone was detected in CVL specimens. Interestingly, both mean plasma and CVL PGE2 levels decreased at 4 weeks (201 to 131 and 1934 to 964 pg/ml), followed by an increase at 8 weeks (plasma, 355 pg/ml; and CVL, 1406 pg/ml). Conclusions: Sulindac has demonstrated chemopreventive activity in a number of tumor systems. Proposed mechanisms include the modulation of prostaglandin synthesis, inhibiting cell proliferation, and inducing apoptosis. These results suggest that oral sulindac has a direct effect on PGE2 production in cervical and vaginal secretions. These data have been used to develop a prospective clinical trial evaluating the chemopreventive efficacy of sulindac in cervical dysplasia. 129. Significant Preliminary Results of a Phase III Randomized Study of Concomitant Chemoradiation with Hydroxyurea vs Hydroxyurea, 5-FU Infusion, and Bolus Cisplatin (HFC) vs Weekly Cisplatin in Advanced Cervical Cancer, a GOG Study. Peter G. Rose, Brian Bundy, Tate Thigpen, Gunter Deppe, Mitchell Maiman, and Daniel Clarke-Pearson. University Hospitals of Cleveland, Cleveland, Ohio. Plenary Session Objective: The GOG performed a randomized study to compare the efficacy of hydroxyurea vs HFC vs weekly cisplatin as concomitant chemoradiation in advanced cancer of the cervix. Methods: Patients with primary, untreated, invasive squamous, adenosquamous, or adenocarcinoma of the cervix, Stages II-B, III, and IV-A, with negative paraaortic nodes were entered on this protocol. Eligibility included a retroperitoneal bilateral paraaortic lymphade-

130. Prolonged Oral Etoposide in Recurrent or Advanced Squamous Cell Carcinoma of the Cervix: A Gynecologic Oncology Group (GOG) Study. Peter G. Rose, John A. Blessing, Wesley C. Fowler, and Steve Waggoner. University Hospitals of Cleveland, Cleveland, Ohio. Poster Session

131. Pregnancies after Radical Trachelectomy (RT) for the Treatment of Early-Stage Cervical Cancer. Michel Roy and Marie Plante, M.D. Laval University, Quebec, Quebec, Canada. Poster Presentation Objective: To evaluate the role of a fertility-preserving surgery in the treatment of early-stage cervical cancer. Methods. Retrospective review of our first 20 patients treated by a laparoscopic pelvic lymphadenectomy followed by a vaginal RT from October 1991 to December 1996 was conducted. Results: Patients’ median age was 33 (22– 42) and 11 were nulliparous. Thirteen cancers were stage IB, 1 was 1A1, 4 were IA2, and 2 were IIA. The majority (13/20) were squamous. Eight lesions were diagnosed by conization, 4 by LEEP, and 8 by biopsies. Two tumors were .2 cm in size and only one had vascular space invasion. Margins were negative in all cases on final specimen. The median number of pelvic nodes

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS excised was 30 (11–71). The median OR time was 295 min (155– 455), blood loss 250 cc (50 –1200), and hospital stay 5 days (4 – 8). There were 3 intraoperative complications: one was attributed to the trachelectomy (uncontrollable pelvic bleeding noted by laparoscopy after the RT) and two were secondary to the lymphadenectomy (injury to the external iliac artery). Postoperative complications included vulvar edema and low back pain in two patients and prolonged bladder dysfunction in one. The median follow-up time is 24 months (8 – 68). One patient recurred in the left parametrium 18 months after RT. Five patients have attempted pregnancy so far: three have delivered by cesarean section at 34, 38, and 25 (premature labor) weeks of gestation; 2 patients are currently 12 and 20 weeks pregnant. One patient has been trying for 6 months. Conclusion: Based on our preliminary data, RT appears to be a safe procedure in well-selected patients with early-stage cervical cancer. Successful pregnancies are definitely possible after this procedure. This new surgical technique warrants further careful evaluation. 132. Treatment and Prognosis of Stage IA1–IIA Cervical Adenocarcinoma. John O. Schorge,* Ellice S. Lieberman,* Annekathryn Goodman,† Christopher P. Crum,* and Ellen E. Sheets.* *Brigham & Women’s Hospital, Boston, Massachusetts; and †Massachusetts General Hospital, Boston, Massachusetts. Objective: To review the treatment, surgical–pathological findings, and prognosis of patients with early-stage cervical adenocarcinoma. Methods: Two hundred forty women with stage IA1–IIA cervical adenocarcinoma were treated from 1982 to 1997. A retrospective chart review was performed. Median follow-up was 43 months. Results: All stage IA1/IA2 patients, 147 stage IB1 patients (94%), and 24 stage IB2/IIA patients (49%) had primary surgical treatment. The table

Clinical Variables by Stage of Cervical Adenocarcinoa Stage Varible No. No. No. No.

patients (1)LN/LND (%) CMT (%) Rec/PDz (%)

IA1/IA2

IB1

IB2/IIA

34 0/27 (0) 2 (6) 0 (0)

157 14/141 (10) 28 (18) 23 (15)

49 8/24 (33) 33 (67) 20 (41)

Note. (1)LN/LND, (1)lymph nodes/lymph node dissections; CMT, combined modality therapy; Rec/PDz, recurrences or persistent disease. shows the clinical variables by stage of disease. A majority of IB2/IIA patients received combined modality therapy (CMT) whether surgery (77%) or radiotherapy (59%) was the primary treatment. Conclusions: Patients with stage IA1/IA2 cervical adenocarcinoma have an excellent prognosis. A minority of stage IB1 patients and a majority of stage IB2/IIA patients receive CMT. Stage IA13 IIA cervical adenocarcinoma represents a disease spectrum similar to that of squamous cell carcinoma. 133. Treatment of Selected Patients with Stage IB Carcinoma of the Cervix after Radical Hysterectomy and Pelvic Lympadenectomy: Pelvic Radiation Therapy (Rt) versus No Further Therapy (NFT) (A GOG Study). Alexander Sedlis,* Brian N. Bundy,† Marvin Rotman,* Samuel Lentz,‡ Laila I. Muderspach,§ and Richard Zaino.¶ *Suny Health Science Center at Brooklyn, Brooklyn, New York; †GOG Statistical Office, Buffalo, New York; ‡Bowman Grey School of Medicine, Winston-Salem, North Carolina; §University of Southern California, Los Angeles, California; and ¶M. S. Hershey Medical Center, Hershey, Pennsylvania. Objectives: Large tumor size, deep stromal penetration (SP), and capillary space involvement (CLS) increase the risk of cancer recurrence and death in stage

105

IB cervical cancer patients (pts) treated by radical hysterectomy. The GOG conducted a randomized trial from 1988 to 1995 to evaluate Rt in reducing recurrence and mortality. Methods: Two hundred seventy-seven patients with at least two risk factors—reater than one-third SP, CLS, and large tumor size (definition depending on CLS and SP)—were randomized to Rt or NFT. Tumor size by palpation was $4 cm in 59% of pts. The deepest SP was at the middle third in 35% and at the deep third in 65% of pts; 70% of pts had CLS. There were no significant differences in the distribution of risk factors, race, performance status, and cell type between Rt and NFT. Of the 137 pts assigned to Rt, 85% received radiation within 90% of the 50.4 Gy dose; 9 pts refused radiation therapy. One hundred forty patients were assigned to NFT. Results: Twenty-one (15%) Rt and 38 (27%) NFT pts had a recurrence, 18 and 27 were vaginal/pelvic; respectively. Eighteen (13%) Rt pts died, 15 from cancer, whereas in the NFT, 28 (20%) died, 23 from cancer. There was a significant 44% reduction of recurrence (i.e., 0.56 relative risk) with recurrence-free rates of 88% for Rt and 79% for NFT at 2 years. Survival analysis awaits further follow-up. Severe (GOG grade 3/4) urologic adverse effects occurred in 4 (2.9%) Rt and 2 (1.4%) NFT pts, hematologic in 3 (2.2%) and 1 (0.7%), gastrointestinal in 4 (2.9%) and 0, neurologic in 1 (0.7%) and 0, respectively. One patient died of secondary treatment-related causes. Conclusions: Rt patients had less risk of recurrence at the cost of 2.9% severe GI and GU toxicity. 134. Prospective Evaluation of the Clinical Tumor Burden Score for FIGO Stage III Cervix Cancer. Eileen M. Segreti, Brian D. Kavanagh, Derrick T. Koo, Douglas W. Arthur, Randal J. West, Rupert K. Schmidt-Ullrich, and Dean R. Goplerud. Virginia Commonwealth University, Richmond, Virginia. Objective: To evaluate prospectively the clinical tumor burden score (TBS), a numerical index of pelvic disease extent in stage III uterine cervix cancer, in patients treated with radiotherapy (RT). Methods: A prior analysis of FIGO stage III cervix cancer patients determined the weighted prognostic contribution of components of presenting disease extent as follows: upper 1/3 vagina, 1; middle 1/3 vagina, 2; lower 1/3 vagina, 3; unilateral sidewall, 2; bilateral sidewall, 3; hydronephrosis, 2; bilateral parametria, 2. The sum for an individual patient equals the TBS, shown in a historical control group (HC) to predict clinical outcome, with worse prognosis for TBS .4. For a contemporary cohort (CC), actuarial local control (LC), distant metastasis (DM), and overall survival (OS) were analyzed with respect to TBS, initial hemoglobin (Hb), age, total dose of RT, and maximum measurable dimension of the tumor (MMD). Results: From 1989 to 1997, 51 consecutive patients with stage III cervix cancer received RT with curative intent. In this CC median pt A dose was 88 Gy, higher than for HC patients (median, 75 Gy). The strongest predictor for LC in the CC group was TBS. For CC patients 4-year LC was 78% for TBS #4 vs 58% for TBS .4 (P 5 0.19 by log rank), comparing favorably with 4-year LC of 66 and 43% for HC patients in same subgroups. The strongest predictor of DM for the CC was MMD; the 4-year freedom from DM was 80% for MMD ,6 cm vs 53% for MMD $6 cm (P 5 0.02). No other parameters predicted LC or DM, and none predicted OS. Conclusions: In the CC higher doses of RT improved LC, especially for patients with adverse prognostic features. TBS remained the strongest predictor of LC, while MMD emerged as a predictor of DM. TBS and MMD should be used to facilitate comparisons between institutional experiences and to stratify patients entering investigational treatment protocols. 135. Radical Abdominal Trachelectomy—A New Surgical Technique for the Management of Cervical Carcinoma. J. Richard Smith,* Deborah Cm. Boyle,* Giuseppe Del Priore,† Laszlo Ungar,‡ David J Corless,* Andrew D. Lawson,* and David Noakes,§ *Chelsea & Westminster Hospital, London United Kingdom; †Bellevue Hospital, New York, New York; ‡St. Stephen’s Hospital, Budapest Hungary; and §Royal Veterinary College, Hatfield, Hertfordshire, United Kingdom. Objective: To develop a curative procedure which allows for the preservation of the uterine body and hence fertility. Methods: Feasibility studies were

106

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS

conducted to assess the possibility of excising the cervix, parametrium, and upper vagina in addition to a pelvic lymphadenectomy to achieve total clearance of an early-stage tumor. To fully dissect the ureters via the abdominal route, the uterine arteries must be divided. Initially, feasibility of performing the dissection was assessed on a woman undergoing a radical hysterectomy. The procedure was then performed in its entirety on a young woman with an attempt to anastomose the divided uterine arteries in the belief that four of the original six arteries are required for uterine viability. Excision of the cervix and division of the uterine arteries leaves the uterus supplied only by the ovarian arteries. In this case the uterus remained viable, supplied in this way leading to questions regarding the necessity of anastomosing the uterine arteries. Results: We have demonstrated that the dissection is possible and, by frozen section histology at the time of surgery on the upper cut surface of the cervix and lymph nodes followed by formal histology, that complete tumor excision is possible. Conclusions: Further studies are required to assess the necessity and best method of anastomosing uterine arteries and observe performance of the anastomosis during pregnancy. Animal studies are in progress to address these issues. 136. Laparoscopic Radical Hysterectomy (Type III) with Aortic and Pelvic Lymphadenectomy: Surgical Morbidity and Short-Term Follow-up. Nicola M. Spirtos, Samuel C. Ballon, Gerardo M. Perez, Jeffrey L. Stern, and John B. Schlaerth. Women’s Cancer Center, Palo Alto, California. Poster Session Objective: The purpose of this study was to identify and quantify surgical morbidity and provide short-term follow-up results in patients consented to undergo complete laparoscopic radical hysterectomy (Type III) and aortic and pelvic lymphadenectomy. Materials and Methods: Forty-six consecutive patients were consented to undergo this surgical procedure using argon beam coagulation (ABC) and endoscopic staplers. All had a Quetelet index of ,35. The average age was 38.5 years (range 26 –58 years). Thirty-six patients had squamous cell carcinomas and 10 had adenocarcinomas of the cervix. Results: All but three surgical procedures were completed laparascopically. The average operative time was 225 min (range 180 to 430 min). The average blood loss was 250 ml (range 50 –700 ml). No patient was transfused. One patient with three previous cesarian sections suffered an intraoperative cystotomy repaired laparoscopically. One patient underwent laparotomy due to equipment complications. Two other patients underwent laparotomy to control bleeding sites. The average lymph node count was 32 (range 19 – 65). Six patients had positive lymph nodes. All surgical margins were macroscopically negative but two had microscopically positive and/or ‘‘close’’ surgical margins. One patient developed a ureterovaginal fistula postoperatively requiring reoperation. Follow-up has been provided on a q 3-month basis. There has been one documented recurrence. Conclusions: Laparoscopic radical hysterectomy (Type III) can be successfully completed in patients with early-stage cervical cancer with acceptable morbidity. Short-term follow-up validates the adequacy of this procedure at this time. 137. Missed Opportunities for Cervical Cancer Screening of HMO Members Developing Invasive Cervical Cancer (ICC). Hai-Yen Sung,* Kathy Kearney,* Marie Miller,* George Sawaya,† Robert Hiatt,‡ and Walter Kinney.‡ *TPMG, Oakland, California; †UCSF, San Francisco, California; and ‡TPMG, Sacramento, California. Plenary Session Objective: To identify potential missed opportunities for cervical cancer screening in members of a prepaid health plan who had no Pap smear performed in the 3 years prior to the diagnosis of ICC, thereby informing the debate about whether the best use for resources aimed at increasing screening rates is efforts at ‘‘outreach’’ (mail, telephone, etc.) or ’’inreach’’ (obtaining Pap specimens on patients presenting to primary care providers for other reasons). Methods: Medical records of all 649 members of our prepaid health plan developing invasive cervical cancer during the period from 1988 to 1994 were reviewed and details of patient interaction with the medical care system including clinic visits, symptoms, Pap smear results, diagnostic procedures,

tumor histology and FIGO stage, treatment, and outcome were abstracted. Assessment of screening history and history of clinic visits was limited to the period from 36 to 6 months prior to diagnosis to avoid including smears or visits which were part of the evaluation that lead to the recognition of ICC. Results: Of women developing ICC, 390/649 (60%) had not had a smear in the period from 36 to 6 months preceding the diagnosis. Of these 390 ‘‘nonscreeners,’’ 241 had been insured by our HMO for at least 30 of the 36 months prior to diagnosis. Of these women, 200/241 (83%) had had contact with the medical system as clinic outpatients during the specified period. Considering only internal medicine clinic visits, 162/241 (67%) had been seen at least once, and 95/241 (39%) had been seen three or more times. Of these 241 women, 18/241 (7%) had been seen one or more times in an Ob/Gyn clinic. Conclusions: Most women in this population developing ICC without benefit of recent Pap smear screening were seen in outpatient clinics in the 3 years preceding their diagnosis. 138. Use of Pap Smear Screening among Women Developing Invasive Cervical Cancer (ICC) While HMO Members. Hai-Yen Sung,* Kathy Kearney,* Marie Miller,* George Sawaya,† Robert Hiatt,‡ and Walter Kinney.‡ *TPMG, Oakland, California; †UCSF, San Francisco, California; and ‡TPMG, Sacramento, California. Objective: To assess the cytologic antecedents of invasive cervical cancer in an HMO population, thereby guiding the expenditure of resources for cervical cancer prevention. Methods: Medical records of all members of our prepaid health plan developing invasive cervical cancer during the period from 1988 to 1994 were reviewed and details of patient interaction with the medical care system including clinic visits, symptoms, Pap smear results, diagnostic procedures, tumor histology and FIGO stage, treatment, and outcome were abstracted. Assessment of screening history was limited to the period from 36 to 3 months prior to diagnosis to avoid including smears or visits which were part of the evaluation that lead to the recognition of ICC. Results: During the years 1988 –1994, 455 women were diagnosed with ICC who had 3 years of continuous membership prior to diagnosis. In the period from 36 to 3 months prior to the diagnosis of ICC 213/455 (46.8%) patients did not have a pap smear within the health plan, 92/455 (20%) had at least one abnormal Pap mandating follow-up, and 150/455 (33%) had one or more Paps interpreted as normal or abnormal not requiring follow-up (‘‘metaplasia,’’ for example), and no abnormal Paps. Conclusions: Despite universal access to medical care, in an HMO population with a high overall screening rate, underscreening remains the most important contributor to ICC. This serves to emphasize the importance of patient and provider education if the incidence of cervical cancer is to be reduced in this population. 139. Cytokine Production In Response To Human Papillomavirus (HPV) Type 16 E7 Peptides in Patients with Squamous Intraepithelial Lesions (SIL): Correlation with Follow-up Cervical Cytology and HPV Status. Patrick Timmins, J. Ketz, G. Ho, Y. Wang, L. Cipriani, C. Chang, P. Anderson, S. Romney, R. Burk, and A. Kadish. AECOM & Montefiore Medical Center, Bronx, New York. Poster Session Objective: We hypothesize that characterizing lymphoproliferative (LP) responses by cytokine production will enable us to identify a subset of patients (pts) with HPV infection and SIL who will undergo spontaneous regression. Methods: In this study peripheral blood mononuclear cells from 36 untreated pts with biopsy-proven SIL and HPV infection detected by PCR were cultured with overlapping peptides from the HPV 16 E7-transforming protein for 3, 7, 14, and 21 days. Production of Interleukin-4 (IL-4) in culture supernatants was measured by ELISA and cytokine production was correlated with follow-up Pap and HPV status. Results: Among patients with positive LP to E7 peptides 9/10 (90%) with high IL-4 production and 7/7 (100%) with low IL-4 production had follow-up Pap negative for SIL, whereas only 11/19 (58%) pts with absent IL-4 production had a negative follow-up Pap (P 5 0.039). Of patients with high IL-4 production 7/9 (78%) were negative for HPV, while only 3/5 (ic nodes who have survived with no evidence of disease is 59 months (48 – 69),

SOCIETY OF GYNECOLOGIC ONCOLOGISTS—ABSTRACTS whereas the 5 patients with recurrence all died within 17 months of surgery. Conclusions: The results suggest that TH2 responses to E7 peptides, characterized by IL-4 production, are associated with regression of SIL as detected by Pap smear, and these immune responses are associated with disappearance of HPV infection, including high-risk types. If larger studies confirm these preliminary results, cytokine production may provide an intermediate biomarker for differentiating those patients who will develop persistent disease from those who will spontaneously regress. [Supported in part by an ACOG/Searle research award (P.F.T.).] 140. Digital Fluoroscopy Reduces Procedure Time during Gynecologic Brachytherapy and Facilitates Radiation Dose Calculation. Randal J. West, Daniel C. Han, Eileen M. Segreti, and Brian D. Kavanagh. Virginia Commonwealth University, Richmond, Virginia. Objective: Intraoperative radiographs are widely used to verify proper device positioning in gynecologic brachytherapy, but procedure time can be prolonged during film processing, especially when devices must be repositioned. To evaluate the use of digital fluoroscopy for rapid device verification, digital radiographic images (DRIs) were analyzed with respect to image acquisition time and radiation dosimetric accuracy. Methods: In 25 tandem and ovoid (T&O) cases, DRIs were recorded with a commercially available digital fluoroscope. Total time to complete a set of AP and lateral images was measured. The DRIs were copied in a bitmap format and printed on ordinary paper. In 10 cases DRIs were used for radiation dose calculations, which were compared to values generated from film radiographs. Results: Average time to obtain a set of DRIs was 3.1 min, comparing favorably with the 8.5 min needed for film transport and processing. Storing images in bitmap format allows on-screen demarcation of anatomic structures and source positions. Reference point dose calculations based on DRIs were accurate to within 63% of values from film radiographs, the same variability expected from serial films of the same implant. DRIs and corresponding isodose distributions can be printed on transparencies that can be placed on external beam simulation films to aid design of customized blocks for parametrial boost irradiation. Conclusion: Digital fluoroscopy allows near real-time device placement verification and accurate dosimetric calculation for gynecologic brachytherapy. The electronic format offers convenient storage and image processing. Because DRIs are obtained faster than film radiographs, there can be reduced anesthesia time and savings in total operating room charges, especially for cases requiring multiple device adjustments. 141. Long-Term Results in Cervical Cancer Patients with Positive Paraaortic Nodes. Raimund Winter, Manfred Lahousen, and Josef Haas. University of Graz, Graz, Austria. Objective: The effect of systematic paraaortic lymphadenectomy in patients with cervical cancer is unknown. Methods: Between 1985 and July 1992 systematic pelvic and paraaortic lymphadenectomy was performed in 79 patients with invasive cervical cancer. The indications for the procedure were FIGO stage and IB disease and positive pelvic lymph nodes, stage IIB disease, or a preoperative tumor volume of .30 ccm as measured by MRI. Patients with lymph node involvement adjuvantly received carboplatin (400 mg/m2) and bleomycin (30 mg). Results: Forty-four of 79 patients (56%) had positive nodes and 12 (15%) had positive paraaortic nodes. The 5-year survival rate was 78% for patients with negative nodes and 63% for patients with positive pelvic but negative paraaortic nodes. Of the 12 patients with positive paraaortic nodes, 1 died postoperatively of a pulmonary artery embolism and 1 died intercurrently without evidence of disease at 20 months. Of the remaining 10 patients, 5 (50%) are free of recurrence. The average follow-up time of the 5 women with positive paraaortic nodes who have survived with no evidence of disease is 59 months (48 – 69), whereas the 5 patients with recurrence all died within 17 months of surgery. Conclusions: The results suggest that systematic pelvic and paraaortic lymphadenectomy and adjuvant chemotherapy have a therapeutic effect in patients with advanced cervical cancer and positive paraaortic lymph nodes.

107

OVARIAN 142. Laparoscopic Complications in Gynecologic Surgery for Benign or Malignant Disease. Nadeem Abu-Rustum, Richard R. Barakat, and John P. Curtin. Memorial Sloan–Kettering Cancer Center, New York, New York. Poster Session Objective: To describe the intraoperative and postoperative complications associated with laparoscopic gynecologic surgery in patients with benign or malignant disease. Methods: We retrospectively reviewed all laparoscopic procedures performed by the Gynecology Service during a 6-year period (July 1991 to April 1997), and identified all intraoperative and postoperative complications related to laparoscopy. Results: Seven hundred twenty-two laparoscopic procedures were performed, with 410 patients (56.8%) having malignant disease and 312 patients (43.2%) having benign conditions. The three most common procedures performed were bilateral salpingo-oophorectomy (271 patients), laparoscopically assisted vaginal hysterectomy (154 patients), and second-look laparoscopy (108 patients). Forty-nine patients (6.8%) had laparoscopy-related complications: postoperative abdominal-wall hematoma, the most frequent complication, occurred in 8t patients (1.1%); wound infection occurred in 7 patients (1%); and 6 (0.8%) required incidental enterotomy. Conversion to laparotomy occurred in 110 of the 722 patients (15.2%), with only 6 patients (0.8%) requiring conversion to correct intraoperative laparoscopic complications, and 6 patients (0.8%) requiring reoperation to correct postoperative complications. There was no significant difference between the complication rate in patients with malignant disease compared to those with benign conditions: 34 of 410 (8.3%) and 15 of 312 (4.8%), respectively (P 5 0.09). The ‘‘open laparoscopy’’ technique was associated with enterotomy in 3 patients (0.4%). There was one postoperative mortality (0.1%) in a patient who had intestinal injury. Conclusion: Despite the more aggressive gynecologic laparoscopic procedures, the major intraoperative and postoperative complication rate remained steady at an acceptable rate of 6.8%, with only 1.7% of patients requiring conversion to laparotomy or reoperation to correct major complications. 143. Comparison of Paclitaxel/Cisplatinol to Cyclophosphamide/Cisplatinol in Incompletely Resected Papillary Serous Peritoneal Cancer. James K. Aikins, James Paulson, Gerald Arnold, and Thomas F. Rocereto. Robert Wood Johnson Medical School, Camden, New Jersey. Objective: To assess the response rates in women with primary papillary serous tumors of the peritoneum treated with cyclophosphamide and cisplatinol as compared to paclitaxel and cisplatinol. Methods: We retrospectively reviewed patients with primary peritoneal cancer treated at our institution from January 1990 to January 1997. Eligible patients included women with residual disease more than 1 cm treated with cyclophosphamide/cisplatinol (750 and 75 mg/m2) or Taxol/cisplatinol (135 and 75 mg/m2). Survival analysis was performed using Kaplan–Meier method and comparisons were based on log-rank test. Results: Twenty-eight patients with primary peritoneal cancer comparable to Stage 3 ovarian cancer (omental or retroperitoneal lymph node metastasis) were identified that met all criteria. Primary reductive surgery was performed in all of the 28 patients. Bilateral salpingo-oophorectomy (BSO), omentectomy with hysterectomy in 16, and without hysterectomy in 12. Bowel resection occurred in 2/16 hysterectomized patients. Twenty-two of 28 of these patients were available for review. The median age of this group was 71 (41– 81 years). Histological review showed 14 Grade 3 and 8 Grade 2 papillary serous tumors. Ten of the patients received Taxol/cisplatinol with 80% response, while 12 received cisplatinol/cyclophosphamide with 67% response (P 5 0.48). The surgical response was slightly higher among Taxol/cisplatinol group, but was not statistically significant. The mean disease-free interval was 20 months for cyclophosphamide/cisplatinol and 12.5 months for Taxol/cisplatinol (P 5 0.27). Conclusion: Incorporating paclitaxel into the platinumbased therapy for peritoneal cancer had no significant statistical difference for median disease-free interval and overall survival. The complete surgically