J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 25 NUMBER 5 FEBRUARY JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Phase III Trial of With or Without Paclitaxel in Advanced Uterine Carcinosarcoma: A Gynecologic Oncology Group Study Howard D. Homesley, Virginia Filiaci, Maurie Markman, Pincas Bitterman, Lynne Eaton, Larry C. Kilgore, Bradley J. Monk, and Frederick R. Ueland From the Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, Brody School of Medicine, Greenville, NC; Gynecologic Oncology Group Statistical and Data Center, Roswell Park Cancer Institute, Buffalo, NY; The University of Texas M.D. Anderson Cancer Center, Houston, TX; Department of Pathology, Rush University Medical Center, Chicago, IL; Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, Ohio State University/James Cancer Hospital, Columbus, OH; Department of Obstetrics and Gynecology, University of Alabama at Birmingham, Birmingham, AL; Division of Gynecologic Oncology, University of California-Irvine, Chao Family Comprehensive Cancer Center, Orange, CA; and the Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Kentucky, Lexington, KY. Submitted March 7, 2006; accepted November 21, Supported by National Cancer Institute grants to the Gynecologic Oncology Group (GOG) Administrative Office (Grant No. CA 27469) and the GOG Statistical and Data Center (Grant No. CA 37517). Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Address reprint requests to Denise Mackey, Gynecologic Oncology Group, Administrative Office, Four Penn Center, 1600 JFK Blvd, Suite 1020, Philadelphia, PA 19103; dmackey@gog.org by American Society of Clinical Oncology X/07/ /$20.00 DOI: /JCO A B S T R A C T Purpose To determine if paclitaxel added to ifosfamide as first-line treatment for advanced uterine carcinosarcoma (CS) improves overall survival (OS), progression-free survival (PFS), response, and toxicity. Patients and Methods Eligible patients had measurable stage III or IV, persistent, or recurrent uterine CS. Random assignment to treatment was between ifosfamide 2.0 g/m 2 intravenously (IV) daily for 3 days (arm 1) or ifosfamide 1.6 g/m 2 IV daily for 3 days plus paclitaxel 135 mg/m 2 by 3-hour infusion day 1 (arm 2). Mesna was administered similarly (both arms); filgrastim began on day 4 (arm 2). Cycles were repeated every 21 days up to eight cycles. Results Of 214 patients enrolled, 179 were eligible (arm 1, 91 patients; arm 2, 88 patients). Arm 2 patients experienced more frequent and severe sensory neuropathy (grade 1 to 4; 8% v 30%). The crude response rate was 29% (arm 1) and 45% (arm 2). The odds of response stratified by performance status were 2.21 greater in arm 2 (P.017). Median PFS and OS, respectively, for arm 1 compared with arm 2 were 3.6 v 5.8 months and 8.4 v 13.5 months, respectively. There was a 31% decrease in the hazard of death (hazard ratio [HR], 0.69; 95% CI, 0.49 to 0.97; P.03) and a 29% decrease in the hazard of progression (HR, 0.71; 95% CI, 0.51 to 0.97; P.03) relative to arm 1 when stratifying by performance status. Conclusion OS was significantly improved in arm 2, and toxicities were as expected and manageable. However, the need for active new agents persists, given that OS remains relatively poor in this disease. J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION Patients with previously untreated disseminated carcinosarcomas of the uterus respond less frequently to chemotherapy than those with uterine carcinomas 1 and have short survivals. Previous Gynecologic Oncology Group (GOG) studies investigating a number of single agents have reported the following response rates: etoposide (7%) 2 ; doxorubicin (10%) 3 ; cisplatin (18%) 4 ; ifosfamide (32%, 5 36% 6 ), and topotecan (10%). 7 Although the combination of ifosfamide and cisplatin in a prior GOG study produced superior response rates compared with ifosfamide alone, there was no survival advantage. 6 A recent GOG study of paclitaxel to treat this disease was noted to have moderate activity (18% response rate). 8 The current randomized trial of ifosfamide alone versus ifosfamide plus paclitaxel sought to compare response, progression-free survival (PFS), and overall survival (OS) in women with disseminated carcinosarcoma of the uterus. Because of previously reported toxicity (primarily neurotoxicity) when ifosfamide was administered as a 5-day regimen, 5,6 a 3-day schedule was used in this trial. The ifosfamide starting dose was 1.6 g/m 2 in the combination arm due to reported toxicity and possible treatment-related deaths in the previous GOG randomized trial. 6 There are relatively few other references to specific chemotherapy regimens (with or without radiation) for carcinosarcoma. In our study, ifosfamide alone was selected as the control arm to assess the impact on survival (the primary end point) of adding paclitaxel while maintaining a tolerable level of toxicity. 526

2 Paclitaxel for Uterine CS PATIENTS AND METHODS Patients Eligible patients had primary, histologically confirmed heterologous or homologous carcinosarcoma (mixed mesodermal tumors) of the uterus. On central review, primary cell type and extent of disease was confirmed by the GOG pathology committee. Patients were to have stage III or IV, persistent, or recurrent disease not amenable to curative intent by other means. All patients had measurable disease in two dimensions by palpation, x-ray, magnetic resonance imaging, computed tomography, or ultrasound with a minimum measurement of 1 cm by imaging. Patients were to have adequate bone marrow function, with an absolute neutrophil count of 1,500/ L, platelets 100,000 L, creatinine 2 mg %, or creatinine clearance 50 ml/min, with bilirubin 1.5 normal, AST 3 normal, and serum albumin 3 g/dl. Patients were required to have a GOG performance status of 0 to 2. At least 6 weeks must have passed since prior radiotherapy (RT) for the current malignancy and at least 3 months must have passed since RT to the site of current measurable disease. Patients who had received prior chemotherapy for uterine carcinosarcoma were ineligible, as were those with septicemia, severe infection, acute hepatitis, or GI bleeding, or a performance status of 3 to 4. Patients having other invasive malignancies (with the exception of nonmelanoma skin cancer), prior or current evidence of other cancer within 5 years, or any previous cancer therapy determined to be a contraindication for this protocol, were ineligible. Patients could not have a history of congestive heart failure, unstable angina, or myocardial infarction within the last 6 months. The institutional review boards of participating institutions approved the protocol before enrolling any patient, and all patients signed an informed consent consistent with all federal, state, and local requirements, before study entry. Chemotherapy Study drugs were available commercially. Patients randomly assigned to arm 1 (ifosfamide alone) received ifosfamide 2.0 g/m 2 intravenously (IV) daily for 3 days every 21 days for eight cycles. The starting dose was 1.2 g/m 2 for patients who had received prior RT. Mesna was administered either IV or orally. The IV administration consisted of 2 g IV during 12 hours beginning 15 minutes before the ifosfamide infusion; for oral administration, the total dose wastobe4ginthree divided doses of 1.33 g administered 1 hour before and 8 hours after the ifosfamide infusion for 3 days. For arm 2 (the combination regimen), ifosfamide was to be administered at a dose of 1.6 g/m 2 daily for 3 days (or was reduced to 1.2 g/m 2 for patients who had received prior radiation). Paclitaxel 135 mg/m 2 during 3 hours IV was administered on day 1. Mesna was to be administered as indicated for arm 1. Filgrastim 5 g/kg/d was begun on day 4 until the granulocyte count was 2,000/m 2. The suggested premedication for paclitaxel was dexamethasone 20 mg orally or IV 12 and 6 hours before paclitaxel, diphenhydramine 50 mg IV 30 minutes before paclitaxel, cimetidine 300 mg IV 30 minutes before paclitaxel, or ranitidine 50 mg IV 30 minutes before paclitaxel. Patients on both arms were to receive a maximum of eight cycles of protocol therapy unless disease progression or adverse effects prohibited additional treatment. Depending on toxicity, ifosfamide was to be reduced in 0.4 g/m 2 decrements and paclitaxel was to be reduced to 100 mg/m 2. If hematologic toxicity was grade 1, ifosfamide was to be escalated to a maximum of 2 g/m 2 and paclitaxel was to be escalated to a maximum of 200 mg/m 2. No subsequent chemotherapy was to be administered until the absolute neutrophil count was 1,500/ L and platelets were 100,000/ L. If therapy was delayed for myelosuppression, weekly recounts were obtained. For nonhematologic toxicity, doses were to be reduced by one level for microscopic hematuria. The dose was held until serum albumin 3 g %. If grade 2 ifosfamide-related neurologic symptoms (such as confusion) developed, ifosfamide was to be reduced one dose level. Study therapy was discontinued for the following reasons: treatment delay lasting 6 weeks or more due to myelosuppression; persistent low ( 3 g %) serum albumin unresolved after 6 weeks; recurring grade 2 ifosfamide-induced neurologic symptoms despite dose reduction; development of grade 3 or 4 neurotoxicity; and disease progression. Outcome Parameters Patients without repeat assessments due to reasons related to their disease were considered inassessable for response and were included in the no response category. Response was defined as follows: complete response was the disappearance of all gross evidence of disease lasting for at least 4 weeks; partial response was a reduction of 50% or greater in the bidimensional product from measurement of each lesion sustained for at least 4 weeks; increasing disease was a 50% or greater increase in the product from any lesion or the appearance of any new lesion(s) within 8 weeks of entry onto the study. Stable disease was any condition not meeting any of the above criteria. PFS was the length of time a patient survived from study entry without tumor progression (defined as reappearance or increasing disease). Patients alive without tumor progression are censored at the date of last contact. OS was defined as the observed length of life from study entry to death as a result of any cause or, for living patients, the date of last contact. All adverse events believed to be related to protocol treatment were assessed according to the GOG Common Toxicity Criteria. Statistical Considerations The GOG Statistical and Data Center randomly assigned therapy to each patient. Random assignment was carried out with equal probability of assignment to each treatment regimen using a balanced block randomization to balance assigned treatment regimens within strata defined by institution and performance status (0 to 1 or 2). Within each parent institution, separate treatment blocks were maintained for patients with performance status 0 to 1 and performance status 2. The treatment assignment was concealed from the institution and the patient until telephone registration with verification of eligibility. The primary end point for comparison of the treatment regimens was OS; treatment-related toxicity, PFS, and clinical response were secondary end points. One hundred forty-six deaths would provide statistical power of 90% to detect a proportional decrease of 38.5% in the death hazard when testing at the level of.05 with a one-tail test. 9 Although a one-tail testing procedure was chosen for the purposes of study design and analyses, all significance levels provided in this report are based on two-tail tests. The P values reported for all analyses are unadjusted for multiple comparisons. This report includes 150 deaths, with a median follow-up of 20 months among living patients. An interim futility analysis 10 of OS with 81 reported deaths was reported to the data monitoring committee in June The intention-to-treat principle was applied in treatment group comparisons of OS, PFS, and clinical response after excluding ineligible patients. A log-rank test 11 stratified by performance status was used to test the independence of treatment with PFS and OS. The product-limit method 12 was used to estimate the cumulative probability of PFS and OS. The treatment effect on PFS and OS, stratified by performance status, was estimated using a Cox proportional hazards model. 13 Treatment hazard ratios with 95% CIs are reported. Explorations analyses using proportional hazards regression models of clinical, pathologic, and host characteristics were carried out to identify putative prognostic factors and for crude assessment of the heterogeneity of treatment effects. An exact test 14 stratified by performance status was used to test the independence of treatment with clinical response. The conditional maximum likelihood estimate of the common relative odds ratio is reported with exact 95% confidence bounds. 15 In addition, toxicity analyses included only patients who received assigned treatment. The Kruskal-Wallis test with correction for numerous ties was used to compare the frequency and severity of treatment-related toxicity between treatment arms. 16 RESULTS During the period from November 1997 to April 2004, 214 women were enrolled onto this phase III study, 35 of whom were deemed ineligible (17 randomly assigned to arm 1 and 18 randomly assigned to arm 2). The reasons for exclusion include wrong cell type (n 15 in each arm), wrong stage (n 1 in each arm), primary disease not 527

3 Homesley et al documented (n 1), and required pre-entry tests not done (n 2). Among the 30 patients deemed ineligible due to wrong cell type, the slides representing 22 individuals were found by central pathology review to have no sarcomatous component and the slides representing eight individuals were considered to be purely sarcoma on central review. Thus, 179 (91 in arm 1 and 88 in arm 2) were eligible and constitute the basis of this report. Patient characteristics are listed in Table 1. There is an imbalance between treatment arms with respect to site of disease, with measurable disease confined to the pelvis in 23% of patients in arm 1 compared with 37% of patients in arm 2. For both arms, the median age is Characteristic Table 1. Patient Characteristics (n 179) (n 91) Paclitaxel Filgrastim No. % No. % Age at study entry, years Ethnicity Hispanic/Latino Non-Hispanic/non-Latino Unknown Race Other/unknown Black White GOG performance status Cell type MMT homologous MMT heterologous MMT, not otherwise specified Prior RT Missing No Yes Sites of measurable disease Pelvic Extrapelvic Pelvic and extrapelvic Stage III IV Recurrent/persistent Recurrence-free interval Primary or persistent months months months months months Abbreviations: GOG, Gynecologic Oncology Group; MMT, malignant mesenchymal tumor; RT, radiation therapy. 64 years and the median body-surface area is 1.75 at study entry. A history of prior radiation therapy was reported in 38% of patients in arm 1 and in 30% of patients in arm 2. Approximately 52% of patients in this study had recurrent disease; of those, the median time to recurrence (before enrolling onto this study) was 8 months (range, 1 to 121 months). Four patients (arm 1, n 3; arm 2, n 1) never received any protocol chemotherapy. The median total dose of ifosfamide received was 14.6 g/m 2 (arm 1) and 25.0 g/m 2 (arm 2); the total doses received were 1,784 g/m 2 (arm 1) and 2,170 g/m 2 (arm 2). Table 2 illustrates that 34% of eligible patients completed all eight cycles; however, the percentage varied by treatment arm. Patients in arm 1 tended to discontinue protocol therapy early due to progression (54%) compared with those in arm 2 (33%). Toxicity is summarized for all patients who received any study drug in Table 3. One patient in arm 2, for whom toxicity data was not reported, was not assessable. Arm 2 produced more frequent alopecia (grade 1 to 2; 40% v 58%) and more frequent and severe sensory neuropathy compared with arm 1 (grade 1 to 4; 8% v 30%), although detailed information regarding duration of this effect was not collected. In addition, arm 2 produced more frequent thrombocytopenia (11% v 46%), but it was seldom severe. There was very little difference between the arms with respect to neutropenia. Response rates (complete response partial response) stratified by performance status are listed in Table 4. The crude proportion of responding patients was 0.29 in arm 1 and 0.45 in arm 2. After stratifying by performance status, the odds of response in patients in arm 2 is 2.21 times that of patients in arm 1 (95% CI; 1.12 to 4.43; P.02). Thus, ifosfamide plus paclitaxel is superior to ifosfamide alone with respect to probability of response. Response did not vary significantly with site of disease or prior radiation. There is a statistically significant difference in PFS when comparing the two regimens (Fig 1). The crude median PFS for arm 1 was Outcome Table 2. Cycles Completed (n 179) (n 91) Paclitaxel Filgrastim No. % No. % No. of cycles completed Reason for discontinuation of study treatment Completed eight cycles Disease progression Patient refusal Toxicity Death Other disease JOURNAL OF CLINICAL ONCOLOGY

4 Paclitaxel for Uterine CS Adverse Event Table 3. Adverse Events (n 174) Adverse Event Grade, Adverse Event Grade, Paclitaxel Filgrastim (n 86) Leukopenia Thrombocytopenia Neutropenia Anemia Alopecia Cutaneous Cardiac Fever Fatigue Nausea and vomiting Other GI Creatinine Hematuria Other genitourinary Hepatic Peripheral neuropathy CNS Pain Pulmonary Infection/sepsis Metabolic Use of filgrastim began on day 4 of cycle 1 treatment. P months compared with 5.8 months for arm 2. The risk of progression or death relative to arm 1 stratified by performance status is 0.71 (95% CI, 0.51 to 0.97; P.03). This relative hazard estimate results in an increased percentage (from 50% to 61%) of patients surviving at 3.6 months. There is also a statistically significant difference in OS between the two regimens (Fig 2). The crude median OS for arms 1 and 2, respectively, was 8.4 and 13.5 months. The risk of death relative to arm 1 stratified by performance status is 0.69 (95% CI, 0.49 to 0.97; P.03). This relative hazard estimate translates as an increased percentage, from 50% to 62%, of patients surviving at 8.4 months. The majority of deaths were attributed to uterine carcinosarcoma. There was one death in arm 1 in which protocol treatment may have been contributory. This patient was found to have a Table 4. Response by Performance Status (n 179) Response (n 91) Paclitaxel Filgrastim No. % No. % GOG performance status 0 14of of of of of4 0 4of13 31 Overall response rate 26 of of Abbreviation: GOG, Gynecologic Oncology Group. bowel obstruction and sepsis, followed by death, after one cycle of protocol therapy. In exploratory analyses, after adjusting OS and PFS for treatment regimen and performance status, no other prognostic factors were significant at the P.05 level. Factors tested included time to initial recurrence, pelvic metastasis, lung metastasis, sarcomatous elements (homologous v heterologous v unspecified), prior RT, race (white v black v other), site of disease (pelvic v extrapelvic v pelvic extrapelvic), and age at study entry. The log of the treatment hazard ratio estimates for OS and PFS did not differ from the primary analysis estimates by more than 7% after adjusting for prior radiation and for extrapelvic and combination Proportion Surviving Progression-Free Treatment Group Alive, PF Dead Total paclitaxel Months on Study Fig 1. Progression-free (PF) survival by two randomized treatment groups

5 Homesley et al Proportion Surviving Treatment Group + paclitaxel Months on Study pelvic and extrapelvic sites of disease. Similarly, the estimates of the log odds of response to treatment did not differ by more than 4% after adjusting for radiation treatment and for sites of disease. DISCUSSION Alive Dead Total Fig 2. Overall survival by two randomized treatment groups. To our knowledge this is the first randomized prospective trial in uterine carcinosarcoma that clearly demonstrates a superior OS for combination chemotherapy compared with single-agent treatment. This conclusion is not altered when the analysis of the treatment effect is adjusted for prior radiation therapy and for sites of disease. In an earlier GOG trial, there was no evidence of a survival advantage for the combination of 5-day ifosfamide and cisplatin compared with 5-day ifosfamide alone. 6 In our trial, the toxicity of 5-day ifosfamide, especially neurotoxicity, was virtually eliminated by the 3-day ifosfamide regimen used in both arms. In this trial, in which ifosfamide alone was inferior to ifosfamide and paclitaxel, it has to be emphasized that ifosfamide was administered in the 3-day schedule, albeit with a slightly higher dose of 2 g/m 2 /d for 3 days in the single-agent arm versus 1.6 g/m 2 /d in the combination arm. The higher ifosfamide dose as a single agent was chosen to balance the presumed more toxic combined arm of ifosfamide plus paclitaxel. It is unrealistic to contemplate that a randomized trial of 3- v 5-day ifosfamide will be conducted in patients with this rare tumor. Although the results of the present study speak for themselves, rigorous assessment of the role of ifosfamide in relation to platinum, taxanes, and various combinations is a challenge without a ready solution. Because carcinosarcomas have, by definition, an epithelial component as well as a sarcomatous component, the question arises whether regimens should be designed based on the chemotherapeutic experience with uterine adenocarcinoma plus agents directed at the sarcomatous element. Paclitaxel has been determined to be highly active as both first- and second-line therapy in endometrial adenocarcinoma. 17,18 In addition, a recent GOG trial has established the triple combination of doxorubicin, cisplatin, and paclitaxel as the standard that now is being compared to carboplatin and paclitaxel in endometrial cancer. 19 However, doxorubicin does not seem to be highly active in carcinosarcoma. 3 In addition, the ability to combine multiple cytotoxic agents in effective doses is limited. Clearly, such a differential drug effect with respect to the two cellular elements of this cancer is speculative. Quality-of-life assessments were not made as this was not a component of all GOG trials at the initiation of this study. However, the reported toxicity of the combination arm, in general, was tolerable and minimal. Future trials, as appropriate, will have a quality-of-life assessment tool. The current GOG phase II chemotherapy trial uses docetaxel plus gemcitabine in patients with one prior chemotherapy regimen. If the results are promising, it is likely that the next phase III trial will be a comparison of 3-day ifosfamide and paclitaxel compared with docetaxel and gemcitabine. As an alternative, comparison with carboplatin and paclitaxel might be considered; that regimen has reached first-stage accrual in a current phase II trial in the patient population described in this report, and is being used in the current phase III endometrial adenocarcinoma trial. More active agents certainly are needed for this disease, and it seems premature to suggest the use of the present regimen as adjuvant therapy outside a clinical trial. Nevertheless, for chemotherapy of advanced uterine carcinosarcoma, the results of this trial indicate that 3-day ifosfamide plus paclitaxel should be used for comparison to other promising chemotherapy regimens. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The authors indicated no potential conflicts of interest. AUTHOR CONTRIBUTIONS Conception and design: Howard D. Homesley, Virginia Filiaci, Maurie Markman, Pincas Bitterman Provision of study materials or patients: Howard D. Homesley, Maurie Markman, Lynne Eaton, Larry C. Kilgore, Bradley J. Monk, Frederick R. Ueland Collection and assembly of data: Virginia Filiaci Data analysis and interpretation: Howard D. Homesley, Virginia Filiaci, Maurie Markman Manuscript writing: Howard D. Homesley, Virginia Filiaci Final approval of manuscript: Howard D. Homesley, Virginia Filiaci, Maurie Markman, Pincas Bitterman, Lynne Eaton, Larry C. Kilgore, Bradley J. Monk, Frederick R. Ueland REFERENCES 1. Thigpen JT, Brady MF, Homesley HD, et al: Phase III trial of doxorubicin with or without cisplatin in advanced endometrial carcinoma: A Gynecologic Oncology Group study. J Clin Oncol 22: , Slayton R, Blessing J, DiSaia PJ, et al: Phase II trial of etoposide in the management of advanced or recurrent mixed mesodermal sarcomas of the uterus: A Gynecologic Oncology Group study. Cancer Treat Rep 71: , Omura GA, Major FJ, Blessing JA, et al: A randomized study of adriamycin with and without dimethyltriazenoimidazole carboxamide in advanced uterine sarcomas. Cancer 52: , Thigpen J, Blessing JA, Orr J, et al: Phase II trial of cisplatin in the treatment of patients with advanced or recurrent mixed mesodermal sarcomas of the uterus: A Gynecologic Oncology Group study. Cancer Treat Rep 70: , JOURNAL OF CLINICAL ONCOLOGY

6 Paclitaxel for Uterine CS 5. Sutton GP, Blessing JA, Rosensstein N, et al: Phase II trial of ifosfamide and mesna in mixed mesodermal tumors of the uterus: A Gynecologic Oncology Group study. Am J Obstet Gynecol 161: , Sutton GP, Brunetto VL, Kilgore L, et al: A phase III trial of ifosfamide with or without cisplatin in carcinosarcoma of the uterus: A Gynecologic Oncology Group study. Gynecol Oncol 79: , Miller DS, Blessing JA, Schilder J, et al: Phase II evaluation of topotecan in carcinosarcoma of the uterus: A Gynecologic Oncology Group study. Gynecol Oncol 98: , Curtin JP, Blessing JA, Soper JT, et al: Paclitaxel in the treatment of carcino-sarcoma of the uterus: A Gynecologic Oncology Group study. Gynecol Oncol 83: , Rubinstein LV, Gail MH, Santner TJ: Planning and duration of a comparative clinical trial with loss to follow-up and a period of continued observation. J Chronic Dis 34: , Wieand S, Schroeder G, O Fallon JR: Stopping when the experimental regimen does not appear to help. Stat Med 13: , Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemo Rep 50: , Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53: , Cox DR: Regression model and life tables (with discussion). J R Statistical Soc B 34: , Cox DR: The Analysis of Binary Data. London, United Kingdom, Methuen & Co Ltd, 1970, see Note following References ; reprinted by Chapman and Hall Ltd, London, UK, Mehta CR, Patel NR, Gray R: Computing an exact confidence interval for the common odds ratio in several 2x2 contingency tables. J Am Stat Assoc 80: , Kruskal WH, Wallis WA: Use of ranks in one-criterion variance analysis. J Am Stat Assoc 47: , Ball HG, Blessing JA, Lentz SS, et al: A phase II trial of paclitaxel in patients with advanced or recurrent adenocarcinoma of the endometrium: A Gynecologic Oncology Group study. Gynecol Oncol 62: , Lincoln S, Blessing JA, Lee RB, et al: Activity of paclitaxel as second-line chemotherapy in endometrial carcinoma: A Gynecologic Oncology Group study. Gynecol Oncol 88: , Fleming GF, Filiaci VL, Bentley RC, et al: Phase III randomized trial of doxorubicin cisplatin versus doxorubicin 24-h paclitaxel filgrastim in endometrial carcinoma: A Gynecologic Oncology Group study. Ann Oncol 15: , 2004 Acknowledgment We thank Barbara Saczynski for dedicated management of the clinical data for GOG 161, and Caron Modeas for expert editorial assistance. Appendix The Appendix is included in the full-text version of this article, available online at It is not included in the PDF version (via Adobe Reader )