PARPi after PARPi in epithelial ovarian cancer.

The objective of this study was to describe the treatment experience of patients with recurrent epithelial ovarian cancer who are retreated with an inhibitor of poly(ADP-ribose)-polymerase (PARPi). We conducted a multi-institutional, retrospective review of ovarian cancer patients who received ≥2 lines of therapy containing a PARPi. Demographic, clinical, and pathological data were analyzed with descriptive statistics. Twenty-two patients were identified. For initial PARPi (PARPi1), 12 patients (54.5%) received veliparib, 7 (31.8%) olaparib and 3 (13.6%) rucaparib resulting in 10 patients who had no evidence of disease at the completion of therapy (NED), 3 partial responses (PR), 4 stable disease (SD), and 3 progressive disease (PD). (All 10 CRs involved veliparib given in conjunction with cytotoxic chemotherapy). PARPi1 was used as maintenance in 2 patients. PARPi1 was discontinued because planned number of cycles was reached (n = 10), progression (n = 8), toxicity (n = 2), other (n = 2). For second PARPi (PARPi2), 10 patients (45.4%) received niraparib, 6 (27.3%) olaparib, and 6 (27.3%) rucaparib resulting in 3 PR, 13 SD, and 3 PD. PARPi2 was used as maintenance in 3 patients. The 3 patients who experienced a PR to PARPi2 had a BRCA mutation and were NED following PARPi1. PARPi2 was discontinued because of progression (n = 13), toxicity (n = 6), other (n = 2). One patient currently remains on PARPi2. Toxicity after PARPi1 was not associated with toxicity from PARPi2 (p > 0.05). With 3 approved PARPi for different indications including frontline and recurrence, the opportunity to reuse PARPi has increased. Characterizing those who should be re-challenged is an important initiative moving forward.

On the TRAIL to successful cancer therapy? Predicting and counteracting resistance against TRAIL-based therapeutics.

Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and agonistic antibodies against TRAIL death receptors (DR) kill tumor cells while causing virtually no damage to normal cells. Several novel drugs targeting TRAIL receptors are currently in clinical trials. However, TRAIL resistance is a common obstacle in TRAIL-based therapy and limits the efficiency of these drugs. In this review article we discuss different mechanisms of TRAIL resistance, and how they can be predicted and therapeutically circumvented. In addition, we provide a brief overview of all TRAIL-based clinical trials conducted so far. It is apparent that although the effects of TRAIL therapy are disappointingly modest overall, a small subset of patients responds very well to TRAIL. We argue that the true potential of targeting TRAIL DRs in cancer can only be reached when we find efficient ways to select for those patients that are most likely to benefit from the treatment. To achieve this, it is crucial to identify biomarkers that can help us predict TRAIL sensitivity.

Gab2 regulates the migratory behaviors and E-cadherin expression via activation of the PI3K pathway in ovarian cancer cells.

Ovarian cancer, the most deadly gynecologic malignancy, is often diagnosed late and at the advanced stage when the cancer cells have already migrated and invaded into other tissues and organs. Better understanding of the mechanism of metastasis in ovarian cancer cells is essential to the design of effective therapy. In this study, we investigated the function of scaffolding adaptor protein Gab2 in ovarian cancer cells. Gab2 is found to be overexpressed in a subset of ovarian tumors and cancer cell lines. Gab2 expression mainly regulates the migratory behaviors of ovarian cancer cells. Overexpression of Gab2 promotes the migration and invasion, and downregulates E-cadherin expression in ovarian cancer cells with low-Gab2 expression. Conversely, knockdown of Gab2 expression inhibits the migration and invasion, and promotes E-cadherin expression in ovarian cancer cells with high-Gab2 expression. By expressing Gab2 wild-type and Gab2 mutants that are defective in activation of the PI3K and Shp2-Erk pathways, we find that Gab2 inhibits E-cadherin expression and enhances the expression of Zeb1, a transcription factor involved in epithelial-to-mesenchymal transition (EMT), and cell migration and invasion through the activation of the PI3K pathway. Knockdown of Zeb1 expression blocks Gab2-induced suppression of E-cadherin expression and increase in cell invasion. LY294002 and GDC-0941, inhibitors of PI3K, or Rapamycin, an inhibitor of PI3K downstream target mTOR, can reverse the effects of Gab2 on migration and invasion. Overall, our studies reveal that Gab2 overexpression, via activation of the PI3K-Zeb1 pathway, promotes characteristics of EMT in ovarian cancer cells.

Adenosarcoma in a patient with vaginal endometriosis.

BACKGROUND: Adenosarcoma in a patient with extraovarian endometriosis is a rare event and can be easily overlooked. CASE: A woman with a history of endometriosis underwent multiple resections of a vaginal mass and medical treatment for presumed recurrent endometriosis. Eventually, a vaginal adenosarcoma was diagnosed. CONCLUSION: The possibility of adenosarcoma should be considered if an enlarging mass occurs at the site of extraovarian endometriosis.

Gardnerella vaginalis isolated from patients with bacterial vaginosis and from patients with healthy vaginal ecosystems.

The differences in the phenotype and genotype of Gardnerella vaginalis isolates from patients with bacterial vaginosis (BV) and from patients without BV are unknown. In our study, 43 isolates of G. vaginalis were examined for biotype (hippurate hydrolysis, lipase, and beta-galactosidase activity), sensitivity to metronidazole, and genotype. Of the 117 women visiting the gynecology clinic at Rush-Presbyterian-St. Luke's Medical Center who were included in the study, 27.4% were found to have BV. G. vaginalis was found in samples from 87.5% of women with BV, from 34.0% of women with intermediate BV, and from 26.4% of women with healthy vaginal ecosystems. Among patients with G. vaginalis, biotypes 7 and 8 were isolated from 32% and 20% of patients, respectively. Biotype 5 was predominantly associated with a healthy vaginal ecosystem (P=.0004). Biotypes 5 and 7 were the most resistant to metronidazole. No specific phenotype or genotype of G. vaginalis causes BV.

Characteristics and survival of cervical cancer patients managed at adjacent urban public and university medical centers.

OBJECTIVE: The goal of this work was to compare characteristics and survival of cervical cancer patients at adjacent public and university hospitals to define the effects of poverty and ethnicity on disease. METHODS: A retrospective chart review was conducted of cervical cancer patients managed by gynecologic oncologists at two adjacent urban hospitals between 1992 and 1998. Continuous variables were compared by t test, categorical variables by chi(2), and survival by the Kaplan-Meier and log-rank methods. RESULTS: In all, 372 patients were identified, with 209 (56%) at the public hospital and 163 (44%) at the adjacent university hospital. Ethnic distribution differed between the two hospitals: 100 (52%) versus 46 (28%) African-American, 56 (29%) versus 13 (8%) Hispanic, 31 (16%) versus 96 (60%) Caucasian, and 5 (3%) versus 6 (4%) other (P < 0.001). In addition, public hospital patients presented with more advanced cancers (stages II--IV) than those managed at the university hospital, 96 (48%) versus 53 (34%) (P = 0.008), and squamous cancers were more common at the public hospital, 154 (89%) versus 120 (76%) (P = 0.03). However, with a median follow-up of 17 months, stage-adjusted survival did not differ between the two institutions. CONCLUSIONS: The higher proportions of advanced and squamous cervical cancers encountered at the public hospital likely reflect suboptimal screening. Equal access to gynecologic oncologists eliminated disparities in stage-adjusted survival. Efforts at earlier diagnosis should be directed at indigent, especially minority women.

Allelotype analysis of uterine leiomyoma: localization of a potential tumor suppressor gene to a 4-cM region of chromosome 7q.

Uterine leiomyoma is a benign smooth muscle tumor of the myometrium and is the most commonly encountered neoplasm in women of reproductive age. As for most benign tumors, the pathogenesis of leiomyoma remains obscure, especially at the molecular genetic level. The purpose of this study was to perform a genome-wide allelotype analysis to identify potential sites of tumor suppressor gene inactivation. Fifty-two cases of uterine leiomyoma were subjected to allelotype analysis by using matched pairs of tumor and blood DNA. Loss of heterozygosity (LOH) was assessed at 61 microsatellite markers distributed throughout the genome and representing all 41 chromosome arms. In general, LOH was very rare except on chromosome 7q, where LOH was observed in 34% of all informative tumors. Fine-deletion mapping with 25 microsatellite markers from the 7q22 region revealed a minimal deletion unit of approximately 4 cM, bounded by the markers D7S2453 proximally and D7S496 distally, that probably harbors a novel tumor suppressor gene involved in the etiology of this tumor.

Clinical characteristics of clear cell carcinoma of the ovary.

OBJECTIVE: The aim of this study is to evaluate the clinical characteristics of clear cell carcinoma of the ovary. METHODS: Between 1986 and 1996, 45 patients with clear cell carcinoma of the ovary were identified by scanning the medical records department and the tumor registry at our institution. RESULTS: Median age was 55 years (range 31-80 years). Tumors were 60% (27/45) stage I, 11% (5/45) stage II, 20% (9/45) stage III, and 9% (4/45) stage IV. All patients presented with a pelvic mass ranging in size from 2 x 3 to 20 x 30 cm and all except 1 had optimal cytoreduction. All patients received postoperative platinum-based chemotherapy, 47% (21/45) in combination with paclitaxel. One stage Ia patient refused therapy. Of the 6 stage III/IV patients with measurable residual tumor, 67% (4/6) partially responded to first line chemotherapy by CT scan or second look laparotomy. Recurrences occurred in 37% (10/27) stage I patients, including 18% (2/11) stage Ia, 33% (1/3) stage Ib, and 54% (7/13) stage Ic. Time to recurrence was 16 and 38 months for the two stage Ia patients and 35 months (median, range 18-56 months) for the stage Ic patients. Survival after recurrence was significantly related to disease-free interval after primary chemotherapy. With a median follow-up of 40 months (range 4-145 months), 93% (25/27) of stage I patients are alive, 20% (5/25) with disease, while 46% (6/13) of stage III/IV patients are alive. Median survival for the stage III/IV patients was 22 months (range 4-70 months). CONCLUSIONS: Clear cell tumors of ovary frequently present at early stages. However, these tumors have a propensity for recurrence even after primary chemotherapy in early stage tumors.

BRCA1, BRCA2, and hereditary nonpolyposis colorectal cancer gene mutations in an unselected ovarian cancer population: relationship to family history and implications for genetic testing.

OBJECTIVE: Our purpose was to determine the prevalence of BRCA1, BRCA2, and hereditary nonpolyposis colorectal cancer gene mutations in a large, unselected population of ovarian cancer patients and to evaluate the relationship between mutation status and a routinely obtained family history of cancer. STUDY DESIGN: One hundred sixteen consecutive ovarian cancer patients seen for routine clinical care were examined for BRCA1, BRCA2, hMSH2, and hMLH1 gene mutations with use of the polymerase chain reaction, single-strand conformation polymorphism analysis, and direct gene sequencing. Fisher's exact test was used to evaluate possible associations between BRCA1 and BRCA2 mutation status and specific familial characteristics. RESULTS: Among 116 unselected ovarian cancer patients we identified a total of 13 germline mutations in 12 patients: 10 in BRCA1, one each in hMSH2 and hMLH1, and a single BRCA2 mutation, which occurred in a patient also carrying a BRCA1 mutation. More than half the patients with BRCA1 mutations had family histories that would generally be considered unremarkable. Of 22 family history variables analyzed, only two (maternal family history of breast or ovarian cancer, p=0.037, and maternal family history of any cancer, p=0.020) conferred a significantly increased risk of carrying a BRCA1 mutation compared with ovarian cancer patients without such a history. However, the majority of ovarian cancer patients with these family histories and other suggestive histories tested negative for mutations. CONCLUSIONS: Approximately 10% of ovarian cancers occur in association with genetic mutations known to predispose to the disease. A routinely obtained family history is an unreliable way to identify patients who might harbor mutations. The majority of ovarian cancer patients with suggestive family histories test negative for known gene mutations, perhaps suggesting the existence of additional undiscovered genes predisposing to ovarian cancer.

Deletion mapping of two potential chromosome 14 tumor suppressor gene loci in ovarian carcinoma.

Previous allelotyping studies of epithelial ovarian carcinoma suggest that loss of heterozygosity on chromosome 14q may be a common genetic alteration in this tumor type. The purpose of this study was to determine a precise frequency of chromosome 14q allelic loss in ovarian carcinomas and to define a minimal region(s) of deletion. Seventy-six ovarian carcinomas representative of the complete spectrum of grade, stage, and histological subtype were selected for PCR-based deletion mapping analysis using 15 highly polymorphic microsatellite markers spanning the length of this chromosome arm. Loss of heterozygosity was observed in 49% of the tumors studied, placing 14q among the most frequently affected chromosomal regions in ovarian cancer. Deletions were observed in all tumor grades and stages and in all histological subtypes except tumors of low malignant potential. Deletion of the entire chromosome arm was rare; the majority of tumors displayed partial losses, providing an informative basis for detailed deletion mapping. Two distinct minimal regions of deletion were delineated. One region was defined by markers D14S80 and D14S75 at 14q12-13, and the other region was defined by markers D14S65 and D14S267 at 14q32. These data implicate the involvement of two tumor suppressor genes on chromosome 14q in a substantial fraction of ovarian carcinomas.

Racial differences in survival from gynecologic cancer.

OBJECTIVE: To determine whether survival from gynecologic cancer is different between African-American and white patients at an inner-city hospital with both a large clinic and a private service. METHODS: We studied 538 patients (89 African American, 449 white) diagnosed with cervical, uterine, or ovarian cancer at a single institution from January 1, 1989 through December 31, 1993. Information was obtained on age, stage, site of disease, histology, and type of health insurance (public or commercial). Insurance coverage was used as a proxy for socioeconomic status. Overall survival was estimated by the method of Kaplan and Meier and compared by the log-rank test. Cox proportional hazard modeling was used to evaluate the effects of multiple factors on survival. RESULTS: African-American patients were significantly older and were more likely to have cervical cancer and public insurance than white patients. Overall survival was worse for African-American patients than for white patients (P < .05). However, stage for stage, there was no significant difference in survival between the groups. There was also no difference when patients were grouped by insurance status. African Americans had a significantly worse survival for cervical cancer than whites, and African-American patients older than 65 years had a worse survival than whites of similar age. On multivariate analysis, only stage and insurance coverage were significant predictors of survival. CONCLUSIONS: African-American patients with gynecologic cancer at our institution have worse overall survival than white patients. The survival difference seems to be due predominantly to differences in socioeconomic status and stage at diagnosis.

Adenovirus-mediated gene therapy of ovarian cancer in a mouse model.

OBJECTIVE: Our purpose was to test the feasibility of adenovirus-mediated gene therapy of ovarian cancer. STUDY DESIGN: Ovarian cancer cell lines were exposed to an adenovirus vector expressing a reporter gene (lacZ) and to the same vector bearing the herpes simplex virus thymidine kinase gene (Ad.RSVtk) followed by ganciclovir. lacZ expression and growth inhibition were quantitated. Immunodeficient mice were injected intraperitoneally and subcutaneously with human ovarian cancer cells and treated with Ad.RSVtk and ganciclovir. Statistical analyses included one-way analysis of variance and t tests. RESULTS: Staining for lacZ demonstrated viral transduction in vitro. After exposure to Ad.RSVtk all cell lines showed significant (p < 0.0001, analysis of variance) cytotoxicity to ganciclovir. Human ovarian tumor cells established subcutaneously or intraperitoneally in immunodeficient mice responded to therapy with Ad.RSVtk followed by ganciclovir. Treated mice had a 10- to 20-fold lower subcutaneous tumor burden than did control mice. Additionally, no intraperitoneal tumors were observed in treated mice. CONCLUSIONS: Ovarian cancer cells are readily transduced with recombinant adenovirus and become sensitive to ganciclovir after transduction with Ad.RSVtk. These data support the development of this method for human clinical trials.

Clinical and pathological features of ovarian cancer in women with germ-line mutations of BRCA1.

BACKGROUND: We tested the hypothesis that ovarian cancers associated with germ-line mutations of BRCA1 have distinct clinical and pathological features as compared with sporadic ovarian cancers. METHODS: We reviewed clinical and pathological data on patients with primary epithelial ovarian cancer found to have germ-line mutations of BRCA1. Survival among patients with advanced-stage cancer and such mutations was compared with that in control patients matched stage, grade, and histologic subtype of the tumors. A combination of single-strand conformation and sequencing analyses was used to examine the 22 coding exons and intronic splice-donor and splice-acceptor regions of BRCA1 for mutations in pathological specimens. Alternatively, some patients were known to be obligate carriers of the mutant BRCA1 gene because of their parental relationships with documented mutant-gene carriers. RESULTS: We identified 53 patients with germ-line mutations of BRCA1. The average age at diagnosis was 48 years (range, 28 to 78). Histologic examination in 43 of the 53 patients showed serous adenocarcinoma. Thirty-seven tumors were of grade 3, 11 were of grade 2, 2 were of grade 1, and 3 were of low malignant potential. In 38 patients, the tumors were of stage III; 9 patients (including those with tumors of low malignant potential) had stage I disease, 5 had stage IV, and 1 had stage II. As of June 1996, with a median follow-up among survivors of 71 months from diagnosis, 20 patients had died of ovarian cancer, 27 had no evidence of the disease, 4 were alive with the disease, and 2 had died of other diseases. Actuarial median survival for the 43 patients with and advanced-stage disease was 77 months, as compared with 29 months for the matched controls (P<0.001). CONCLUSIONS: As compared with sporadic ovarian cancers, cancers associated with BRCA1 mutation appear to have a significantly more favorable clinical course.

Mutations of the BRCA2 gene in ovarian carcinomas.

Inherited mutations in the recently discovered BRCA2 gene are believed to be responsible for a significant fraction of early-onset hereditary breast cancers. Unlike BRCA1, however, which confers a high risk to both breast and ovarian cancer, the incidence of ovarian cancer appears to be much lower In BRCA2-linked families, causing uncertainty as to the relevance of BRCA2 to hereditary ovarian cancer. Numerous allelotype studies indicate that allelic deletions Including the BRCA2 locus on chromosome 13q are common in ovarian cancers in general, suggesting that somatic mutations of this gene may be involved in sporadic ovarian tumorigenesis. The purpose of this study was to test the hypothesis that germline or somatic mutations of BRCA2 are associated with hereditary and/or sporadic ovarian cancers, respectively. The entire 10.2-kb coding region of BRCA2 was screened for mutations in 130 consecutive ovarian tumors, the only selection criterion being a pathological diagnosis of epithelial ovarian carcinoma. Loss of heterozygosity at markers flanking BRCA2 was observed in 56% of the tumors. Four germline mutations and two somatic mutations were identified; two of the germline mutations are recurrent, having been previously described. Remarkably, the patients with germline mutations were late-onset cases with no medical or family histories suggestive of hereditary cancer. These data suggest that mutations of BRCA2 are rare in sporadic ovarian cancers, and that the proportion of ovarian cancers resulting from hereditary predisposition may be higher than previously suspected based on estimates derived from studies of highly penetrant genetic loci.

All-trans-retinoic acid inhibits the proliferation of cell lines derived from human cervical neoplasia.

The purpose of this study was to determine whether all-trans-retinoic acid (RA) can inhibit the growth of cervical neoplastic cells by inducing differentiation or by increasing the secretion of transforming growth factor-beta (TGF-beta). Normal and HPV DNA-positive cervical cells (2 cell lines derived from cervical intraepithelial neoplasia (CIN), 2 HPV DNA-transfected cell lines, and 6 cervical carcinoma cell lines) were treated with RA (1 to 1000 nM), and both total viable cell count and [3H]thymidine incorporation were used to evaluate proliferation. In vitro differentiation was evaluated in organotypic (collagen gel raft) cultures with hematoxylin/eosin staining, and using specific immunostaining for fillagrin and cytokeratin 10. TGF-beta 1 and TGF-beta 2 secretion were measured with specific SELISAs. One-way analysis of variance and t tests were performed. RA causes a dose-dependent (P<0.05) growth arrest of comparable magnitude in normal ectocervical cells, in HPV DNA-transfected cell lines, in CIN-derived cell lines, and in four of six carcinoma cell lines. Endocervical cells and two carcinoma cell lines are unaffected. In vitro differentiation is decreased in CIN cells and is unchanged in carcinomas treated with RA as compared to control. Secretion of either TGF-beta 1 or TGF-beta 2 is significantly increased (P<0.05) in response to RA, both in RA-sensitive and in RA-resistant cells. RA induces growth inhibition in cervical neoplastic cell lines, including cervical carcinoma cells. This does not appear to be the result of increased differentiation or of increased TGF-beta secretion.

A bleomycin/ifosfamide/cisplatin regimen exhibits poor activity against persistent or recurrent squamous gynecologic cancers.

Persistent or recurrent squamous malignancies of the female genital tract are usually incurable by conventional therapy, and results of single agent chemotherapy have been disappointing. We undertook this study to confirm a previously reported response rate of 69%, using a regimen of bleomycin 30U, ifosfamide 5g/m2 with mesna 6g/m2, and cisplatin 50 mg/m2 (BIP) for recurrent cervical cancer. This regimen was used to treat persistent or recurrent squamous cancers in women with cervical cancer (n = 11), vaginal cancer (n = 1) and vulvar cancer (n = 1). Results were reviewed retrospectively and toxicities graded according to the criteria of the Gynecologic Oncology Group. No complete responses were seen. One patient had a partial response (10%, 95% confidence interval 0-28%). Five patients (50%), exhibiting stable disease during therapy with BIP, progressed after cessation of therapy. Of 9 women with symptoms after one cycle. Significant toxicities included neutropenic fever (3 grade 3, 3 grade 4), emesis (1 grade 3), confusion (2 grade 4), vaginal bleeding (2 grade 3), and renal failure (1 grade 3). Eight patients were transfused with a total of 28 units of red cells. After 23 months of follow-up, all patients were dead of disease. Mean survival was 10 months. Toxicity associated with this regimen can be significant, and results appear no better than those reported with single agent therapy.

Mutation analysis of the BRCA1 gene in ovarian cancers.

Germline mutations of the BRCA1 tumor suppressor gene on chromosome 17q are involved in a significant fraction of hereditary breast and ovarian cancers. Allelic deletions that include the BRCA1 locus are common in breast and ovarian cancers, implying that somatic mutations of this gene may play an important role in the more common sporadic forms of these tumors as well. The recent cloning of BRCA1 allows direct testing of this hypothesis. A combination of single strand conformation and sequencing analyses was used to examine the 22 coding exons and intronic splice donor and acceptor regions of BRCA1 for mutations in 115 unselected cases of epithelial ovarian carcinoma. Seven mutations were identified, all of which were present in the germlines of patients with remarkable family or medical histories of breast and/or ovarian cancer. Eighty-nine of these tumors were examined for loss of heterozygosity in the BRCA1 region of chromosome 17q, and 67% of the tumors studied exhibited allelic deletions that included this region. These data are consistent with the hypothesis that BRCA1 mutations are involved in the etiology of hereditary ovarian carcinomas but occur rarely in sporadic tumors, and that the frequent allelic loss on chromosome 17q in this cancer type reflects the involvement of an additional tumor suppressor gene(s).

Expression and mutation analysis of the p53 gene in uterine papillary serous carcinoma.

BACKGROUND: The status of p53 protein expression was determined by immunohistochemistry and correlated with genetic analysis and clinical outcome in patients with uterine papillary serous carcinoma (UPSC). METHODS: Twenty-two patients with UPSC were identified and immunohistochemical staining of p53 protein was performed. Staining was analyzed by quantitating nuclear reactivity in 500 randomly counted cells per specimen. DNA analysis was performed on the tumors using single-strand conformation polymorphism (SSCP) analysis of exons 4-10 of the p53 gene, followed by DNA sequencing of all variants. Clinical data and patient status were ascertained from chart reviews. RESULTS: Sixteen of the 22 (73%) tumors were scored as p53-overexpressing as determined by immunohistochemical analysis. Patients whose tumors overexpressed p53 had a statistically significant shorter survival than those whose tumors did not (P < 0.022). DNA analysis of the 22 tumors revealed five with mutations of the p53 gene. Only three of these mutations were observed in tumors that overexpressed p53. CONCLUSIONS: A relatively large percentage of UPSC tumors exhibited high p53 immunoreactivity. Overexpression is correlated with poor prognosis. Positive immunohistochemistry for p53 protein in UPSC is not necessarily indicative of a genetic mutation.

Prognostic indicators of survival in advanced endometrial cancer.

Advanced endometrial cancer represents 14% of all stages but 54% of all deaths attributed to endometrial cancer. From 1973 to 1990, the charts of 137 patients with endometrial cancer (Stage III and IV) treated by the section of Gynecologic Oncology at Rush Medical College were retrospectively reviewed. The log rank method was used for univariate analysis and Cox proportional hazards regression was used for multivariate analysis. The patients were stratified as follows: Stage III, 92 (67.2%), Stage IV, 45 (32.8%); Grade 1, 15 (10.9%), Grade 2, 47 (34.3%), Grade 3, 67 (48.9%); adenocarcinoma, 93 (67.9%), adenosquamous, 18 (13.1%), adenoacanthoma, 2 (1.5%), clear cell, 1 (0.7%), papillary serous, 23 (16.8%). Using univariate analysis, median survival was 1.71 years for Stage III versus 0.68 years for Stage IV. Median survival based on treatment was as follows: radiotherapy (RT) only (n = 16), 0.89 years, surgery only (n = 36), 0.75 years, preoperative RT+surgery (n = 7), 2.5 years, surgery+postoperative RT (n = 56), 2.63 years, and other treatments (hormonal only n = 12, chemotherapy only n = 1, and no treatment n = 9), 0.6 years. Patients with vaginal extension survived a median of 0.82 years, versus 2.49 years without this factor (P = 0.002). Patients with clinically apparent parametrial involvement survived a median of 0.70 years versus 2.65 years without this factor (P = 0.0003). Multivariate analysis was possible via a surgical database (n = 99). Age > 60 (P = 0.01), parametrial involvement (P = 0.04), and abdominal metastases (P = 0.003) were significant prognostic indicators. Papillary or clear cell histology, advanced grade, and mode of treatment were not significant. Patients with abdominal metastases or parametrial extension of tumor have a significant decrease in mean survival.