This Article Abstract Full Text (PDF) Publisher's Note Erratum (v25,p3790) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Maki, R. G. Articles by Hensley, M. L. Search for Related Content PubMed PubMed Citation Articles by Maki, R. G. Articles by Hensley, M. L.

Randomized Phase II Study of Gemcitabine and Docetaxel Compared With Gemcitabine Alone in Patients With Metastatic Soft Tissue Sarcomas: Results of Sarcoma Alliance for Research Through Collaboration Study 002

Robert G. Maki, J. Kyle Wathen, Shreyaskumar R. Patel, Dennis A. Priebat, Scott H. Okuno, Brian Samuels, Michael Fanucchi, David C. Harmon, Scott M. Schuetze, Denise Reinke, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker, Martee L. Hensley

From the Department of Medicine, Memorial Sloan-Kettering Cancer Center New York, NY; Department of Biostatistics & Applied Mathematics and Sarcoma Center, Department of Medicine, The University of Texas M.D. Anderson Cancer Center, Houston, TX; Washington Cancer Institute, Section of Hematology/Oncology, Washington, DC; Mayo Clinic, Rochester, MN; University of Illinois at Chicago, Oncology Specialists, Park Ridge, IL; Winship Cancer Institute, Emory University School of Medicine, Atlanta, GA; Partners' Health Care/Massachusetts General Hospital Cancer Center, Boston, MA; and the University of Michigan Comprehensive Cancer Center, Ann Arbor, MI

Address reprint requests to Robert G. Maki, MD, PhD, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, Box 223, New York, NY 10021; makir{at}mskcc.org

	ABSTRACT

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Purpose: Gemcitabine as a single agent and the combination of gemcitabine and docetaxel have activity in patients with metastatic soft tissue sarcoma. To determine if the addition of docetaxel to gemcitabine improved clinical outcome of patients with metastatic soft tissue sarcomas, we compared a fixed dose rate infusion of gemcitabine versus a lower dose of gemcitabine with docetaxel.

Patients and Methods: In this open-label phase II clinical trial, the primary end point was tumor response, defined as complete or partial response or stable disease lasting at least 24 weeks. A Bayesian adaptive randomization procedure was used to produce an imbalance in the randomization in favor of the superior treatment, accounting for treatment-subgroup interactions.

Results: One hundred nineteen of 122 randomly assigned patients had assessable outcomes. The adaptive randomization assigned 73 patients (60%) to gemcitabine-docetaxel and 49 patients (40%) to gemcitabine alone, indicating gemcitabine-docetaxel was superior. The objective Response Evaluation Criteria in Solid Tumors response rates were 16% (gemcitabine-docetaxel) and 8% (gemcitabine). Given the data, the posterior probabilities that gemcitabine-docetaxel was superior for progression-free and overall survival were 0.98 and 0.97, respectively. Median progression-free survival was 6.2 months for gemcitabine-docetaxel and 3.0 months for gemcitabine alone; median overall survival was 17.9 months for gemcitabine-docetaxel and 11.5 months for gemcitabine. The posterior probability that patients receiving gemcitabine-docetaxel had a shorter time to discontinuation for toxicity compared with gemcitabine alone was .999.

Conclusion: Gemcitabine-docetaxel yielded superior progression-free and overall survival to gemcitabine alone, but with increased toxicity. Adaptive randomization is an effective method to reduce the number of patients receiving inferior therapy.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Soft tissue sarcomas are rare, accounting for less than 1% of all cancers that occur in the United States each year.¹ The standard of care for most primary soft tissue sarcomas is surgery, with radiation also used for larger primary extremity tumors.² Despite good local control, 40% to 50% of patients will develop distant recurrence, which is nearly always fatal.

The most active chemotherapy agents for metastatic soft tissue sarcoma are doxorubicin and ifosfamide.² Gemcitabine and docetaxel each have modest activity in sarcomas alone.^3-9 Gemcitabine may have greater activity when given as a fixed dose rate infusion (10 mg/m²/min) compared with the recommended schedule (a 30-minute infusion).^4,10 The combination of fixed dose rate infusion gemcitabine and docetaxel has been shown to be effective against metastatic leiomyosarcoma (LMS)¹¹ and other soft tissue sarcomas.^12,13 However, it is unclear if the activity of the combination is due to the prolonged infusion of gemcitabine or synergy between the two drugs.

We therefore conducted a multicenter, open-label, phase II study of gemcitabine given via fixed dose rate infusion versus a lower dose of fixed dose rate infusion gemcitabine with docetaxel in patients with metastatic soft tissue sarcomas. The goal was to select the better of two treatment regimens, within each of the four prognostic subtypes, defined by LMS histology versus other, and prior pelvic irradiation versus none.

	PATIENTS AND METHODS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Study Design
In the gemcitabine-only arm, gemcitabine was administered as a fixed dose rate of 10 mg/m²/min¹⁰ during a 120-minute intravenous infusion, at 1,200 mg/m² days 1 and 8, every 21 days. In the gemcitabine-docetaxel arm, the gemcitabine dose was a fixed dose rate 900 mg/m² intravenous infusion during 90 minutes days 1 and 8, with docetaxel 100 mg/m² intravenously during 60 minutes day 8, every 21 days. Gemcitabine and docetaxel were provided by the manufacturers and distributed by a third-party central pharmacy to participating sites. Filgrastim 5 µg/kg subcutaneously daily for 7 to 10 days, or pegfilgrastim 6 mg subcutaneously once, was administered to all patients starting on day 9 to 10 of each cycle. Up to two 25% dose reductions of each agent were permitted in subsequent cycles of therapy for patients experiencing febrile neutropenia (temperature > 38°C with neutrophil count < 1,000/µL), grade 2 neuropathy, grade 3 liver function test abnormalities, or other grade 3 to 4 nonhematologic toxicity. Patients with prior pelvic irradiation started therapy with 25% dose reductions. The clinicaltrials.gov identifier for this study was NCT00142571.

The study was performed at eight Sarcoma Alliance for Research through Collaboration sites in the United States. An institutional review board or ethics committee approved the study protocol and the informed consent form at each site. Each participant provided written informed consent. Patients were stratified at the time of enrollment according to histology (LMS versus other) and prior pelvic radiation. Given that we used an outcome adaptive randomization (AR) procedure based on the interim data, data were collected and analyzed continuously during the trial. Specifically, once a result was entered by a treating institution, those data were immediately incorporated into the randomization model. Data were collected from each participating institution via a secure Web site in the Biostatistics and Applied Mathematics department of M.D. Anderson Cancer Center (Houston, TX) and analyzed automatically each time a patient was randomly assigned. Response Evaluation Criteria in Solid Tumors (RECIST)¹⁴ response determinations were made by radiologists familiar with sarcomas at the treating institutions; these images were not reviewed centrally.

Eligibility Criteria
Patients were eligible if they met the following criteria: diagnosis of soft tissue sarcoma (excluding GI stromal tumor and Kaposi sarcoma); age older than 10 years; recurrent or progressive disease by examination or imaging studies; lack of clinical evidence that a second cancer, if present, was the disease requiring therapeutic intervention; zero to three prior chemotherapy regimens; disease measurable per RECIST; Eastern Cooperative Oncology Group (ECOG) performance status (PS) 2; peripheral neuropathy grade 1 by National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) version 3.0; at least 3 weeks since prior radiation or cytotoxic chemotherapy; neutrophil count 1,000/µL, hemoglobin 8.0 g/dL; platelet count 100,000/µL; total bilirubin institutional upper limit of normal; ALT and AST 5x the institutional upper limit of normal; alkaline phosphatase 2.5x the institutional upper limit of normal; serum creatinine 2.0 mg/dL, negative serum pregnancy test in women of child-bearing potential; use of effective contraception while on study; and ability to provide written informed consent. Patients were excluded if active or uncontrolled infection was present, if prior therapy with gemcitabine or docetaxel had been administered, if a known hypersensitivity to polysorbate 80 was present, if the patient was pregnant or lactating, or if uncontrolled CNS metastases were present.

Clinical examinations and laboratory testing were performed at a screening visit, at the time of the first dose of therapy, and at the start of each subsequent cycle of therapy for as many as eight cycles of therapy. At that point, patients could either continue therapy or stop at the investigator's discretion. A physical examination, ECOG PS, complete blood count, and biochemical profile were performed on day 1 of each cycle of therapy, and complete blood counts were continued weekly for at least the first six cycles of therapy. Tumors were measured or imaged again after every two cycles of therapy.

Statistical Methods
The primary end point of the study was tumor response, defined as complete or partial response within 24 weeks, or stable disease lasting at least 24 weeks. Bayesian AR was used to assign patients to the two treatment arms, based on the estimated probabilities of treatment success (S), defined as RECIST complete or partial response at the end of two, four, six, or eight 3-week cycles of therapy, or stable disease through all eight cycles, and treatment failure (F), defined as RECIST disease progression or death during any of the eight cycles.¹⁵ Patients' data for application of the AR procedure were recorded based on the first finding of RECIST response or failure. A RECIST confirmed response was not used by the AR procedure. Denoting the S and F probabilities of these events by P_S and P_F, respectively, the AR procedure was based on the weighted average P = .435(P_S) + 0.565(1 – P_F); the weights reflect the utilities elicited from the investigators that nonfailure was 30% more important than treatment success.

The AR method allowed for possible treatment-subgroup interactions for the four subgroups: LMS, prior pelvic radiation (PPR); non-LMS, PPR; LMS, no PPR; or non-LMS, no PPR. Thus, the value of P was permitted to vary among the four subgroups depending on treatment-subgroup interactions.¹⁵ Toxicity was evaluated using NCI CTCAE version 3.0. Patients were assessed on an intention-to-treat basis with respect to overall survival. For progression-free survival (PFS), patients were observed from the first day of treatment until progression, toxicity, or completion of at least eight cycles (24 weeks) of therapy. Patients stopping treatment due to toxicity before the first radiologic evaluation on therapy were removed from study and deemed inassessable for response (n = 1, gemcitabine; n = 3, gemcitabine-docetaxel); these four patients were observed for overall survival (OS). Patients developing toxicity after at least one radiologic assessment were censored at the time of their toxicity, and their evaluations were used by the randomization model. Thus, stable disease at a re-evaluation was a positive development, in that treatment did not overtly fail.

Bayesian regression analyses of the ability of the covariates LMS, PPR, ECOG PS, and treatment to predict PFS and OS were conducted assuming a log-normal distribution for PFS or OS time.¹⁶ The log-normal distribution was chosen based on preliminary goodness-of-fit analyses considering several possible models, including the Weibull, exponential, and log logistic. Model selection was based on posterior model probabilities¹⁷ and the Bayes information criterion. Two log-normal regression models were fit, defined in terms of their linear predictors. The first model included only main effects, treatment + LMS + PPR + PS; the second model included these main effects plus treatment-covariate interactions. For each model, noninformative prior distributions were assumed on all parameters. All Bayesian computations were carried out in Winbugs V1.4¹⁸ and using the custom program that was the basis for implementing the AR. All other computations were carried out in S-Plus (version 3.3; Statistical Sciences, Seattle, WA).¹⁹

	RESULTS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
During the recruitment period from January 2003 to December 2005, 122 patients at eight sites were randomly assigned using the AR procedure to receive gemcitabine with or without docetaxel. Accrual was terminated on January 1, 2006. The research database was locked on April 1, 2006. The disposition and baseline characteristics of the two treatment groups are listed in Table 1 and the CONSORT diagram (Fig 1). The median number of patients per site was 15 (range, two to 30).

View larger version (32K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 1. CONSORT diagram.

Seventy-three patients (60%) were randomly assigned to gemcitabine-docetaxel and 49 patients (40%) were randomly assigned to gemcitabine alone by the AR procedure, indicating that gemcitabine-docetaxel had superior outcomes as defined, compared with gemcitabine alone. Given the final data, the posterior probability that the two-drug combination was superior to gemcitabine alone was .98 for PFS for all subgroups, and .97 for OS (Table 3).

The RECIST partial response rate for patients receiving gemcitabine-docetaxel (16%; 12 of 73) was greater than the partial response rate for gemcitabine alone (8%; four of 49; Table 3), and includes two unconfirmed RECIST partial responses (gemcitabine, malignant fibrous histiocytoma/high-grade undifferentiated pleomorphic sarcoma [MFH/HGUPS]; gemcitabine-docetaxel, uterine LMS). One of nine LMS patients receiving gemcitabine had a partial response (11%), compared with five of 29 (17%) who received gemcitabine-docetaxel. Six of 19 patients (32%) with MFH/HGUPS experienced partial responses (two of eight receiving gemcitabine, and four of 11 receiving gemcitabine-docetaxel, including one complete response). Best responses by treatment arm and histology are listed in Table 4.

View larger version (16K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 2. Kaplan-Meier curves for (A, B) progression-free survival (PFS) and (C, D) overall survival (OS) for patients with Eastern Cooperative Oncology Group (ECOG) performance status (A, C) 0 and (B, D) more than 0 in the gemcitabine and gemcitabine-docetaxel arms. (E) Cumulative probability of stopping therapy for toxicity as a function of time for each treatment arm.

Safety and Tolerability
The safety analysis is based on the 120 patients who received at least one dose of chemotherapy. Twenty-six percent of patients receiving gemcitabine and 46% patients receiving gemcitabine-docetaxel required at least one dose reduction. The mean dose intensities per cycle were 94% (gemcitabine) and 90% (gemcitabine-docetaxel). Toxicity by patient is listed in Table 5. The most common NCI CTCAE grade 3 to 4 toxicity was thrombocytopenia (46 of 120 assessable patients; 38%). Febrile neutropenia was observed in seven of 120 patients (6%). Grade 3 fatigue and/or grade 3 myalgias or muscle weakness were observed in 25% of patients receiving gemcitabine-docetaxel versus 10% of patients receiving gemcitabine only. Despite planned dose reductions, patients were removed from study more frequently on the combination arm (Fig 2E). When a Bayesian analysis with uninformative priors is to calculate the median time to removal for toxicity on each treatment arm, there is a .999 probability that the median time to removal for toxicity is shorter with gemcitabine-docetaxel than with gemcitabine alone.

	DISCUSSION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

The reason LMS and MFH/HGUPS respond better to this combination is unknown. Both LMS and MFH/HGUPS are tumors with aneuploid karyotypes, unlike the approximately 25% to 30% of translocation-associated sarcomas. LMS and MFH/HGUPS are different qualitatively from well-differentiated and dedifferentiated liposarcomas, which are also aneuploid, but have ring and giant chromosomes bearing amplification of chromosome 12q and genes CDK4 and HDM2.²⁵ Interestingly, patients with pleomorphic liposarcoma also responded to this combination; pleomorphic liposarcomas appear more like MFH/HGUPS than other sarcomas by gene expression array analysis.²⁶ These data support the idea that as high-grade sarcoma genotypes evolve, they lose features consistent with their primary lineage, reaching a more undifferentiated state. This also accounts for data that many MFH/HGUPS have features of other more differentiated sarcoma subtypes.²⁷

Although hematologic toxicity was similar in both treatment arms, more than 40% of patients receiving gemcitabine-docetaxel discontinued treatment for a variety of nonhematologic toxicities within 6 months of therapy, despite dose reductions. Constitutional symptoms such as myalgias and fatigue were the most significant cumulative adverse effects of gemcitabine-docetaxel, suggesting that the dose and schedule used in this study are too high for long-term use. Nonetheless, the relative ease of administration and toxicity profile of gemcitabine-docetaxel compare favorably with that of doxorubicin-ifosfamide, another commonly used combination in metastatic soft tissue sarcomas.

Bayesian AR was first proposed in 1933.²⁸ AR was first used in a study of extracorporeal membrane oxygenation for respiratory failure.²⁹ Bayesian study designs have been used in clinical trials involving anesthesia,³⁰ stroke,^31,32 and medical devices.³³ The US Food and Drug Administration recently issued draft guidance for use of Bayesian statistical designs in medical device clinical trials, reflecting increased acceptance of these designs in the medical and regulatory communities.³⁴ Notably, although Bayesian study designs have been used in numerous phase I and standard phase II trials by oncologists,^35-39 AR has been used infrequently. Although standard so-called frequentist phase III study designs examine long-term trends of repeated random events, Bayesian designs use an approach of assigning a prior belief of an event, and observing how that prior belief is modified by the data, yielding a posterior probability. Thus, rather than using traditional P values for comparing treatment arms, Bayesian methods use posterior probabilities and credible intervals to quantify the treatment effect magnitude, which provide an intuitive way to think about outcomes of a clinical study such as this one.

We enrolled 73 patients on the superior treatment and 49 on the inferior treatment. Thus, 24 more patients (20%) enrolled on study received superior therapy or avoided inferior therapy than would have been the case with conventional 1:1 randomization. This trial highlights AR as clinically and ethically attractive for comparative trials of new systemic agents for metastatic cancer, given that data can be fed quickly back into a randomization model in real time to treat potentially fewer patients with inferior therapy in comparison to standard frequentist clinical trial designs. In any case, due to the likelihood principle,¹⁶ which states that all of the information for making statistical inferences is contained in the data actually observed, use of AR to conduct the trial does not invalidate its results in comparison to traditional randomized study designs.

The use of Bayesian analysis of a standard clinical trial design was highlighted recently in an editorial commenting on a phase III randomized study of salmeterol and fluticasone in 6,112 patients with chronic obstructive pulmonary disease.⁴⁰ Despite the study size, the hazard ratio for death for patients receiving both agents versus placebo was only of borderline significance (P = .052). The editorial concluded, "Believe it or not, we still need more data, from even larger trials."⁴¹ However, a Bayesian interpretation of the clinical trial was clear: "On further weighing these results, however, I think the treatment with long-acting beta agonists was a winner and that with inhaled corticosteroids was a clear loser."⁴¹ However, Bayesian trial designs are not a panacea. Our study raises a note of caution to investigators interested in AR models. Despite involving centers familiar with sarcoma clinical trial conduct, clerical errors caused randomization misassignments when the randomization model was most sensitive to such errors.

We conclude that the combination of gemcitabine-docetaxel is superior to a higher dose of gemcitabine, given the data from this study, and conclude that the synergy of gemcitabine-docetaxel accounts for the bulk of the combination arm's activity, rather than the fixed dose rate infusion of gemcitabine. Given that RECIST response rates on both treatments were low, but a number of patients had prolonged stable disease, our data also lend support to the idea of stable disease as an important clinical end point for patients with metastatic soft tissue sarcomas.⁴²

	AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment: N/A Leadership: N/A Consultant: N/A Stock: N/A Honoraria: N/A Research Funds: Robert G. Maki, Eli Lilly & Co, Sanofi; Shreyaskumar R. Patel, Eli Lilly & Co, Sanofi; Michael Fanucchi, Sanofi; Laurence H. Baker, Eli Lilly & Co, Sanofi Testimony: N/A Other: N/A

	AUTHOR CONTRIBUTIONS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Conception and design: Robert G. Maki, J. Kyle Wathen, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker, Martee L. Hensley

Administrative support: Denise Reinke

Provision of study materials or patients: Robert G. Maki, Shreyaskumar R. Patel, Dennis A. Priebat, Scott H. Okuno, Brian Samuels, Michael Fanucchi, David C. Harmon, Scott M. Schuetze, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker

Collection and assembly of data: Robert G. Maki, J. Kyle Wathen, Shreyaskumar R. Patel, Dennis A. Priebat, Scott H. Okuno, Brian Samuels, Michael Fanucchi, David C. Harmon, Scott M. Schuetze, Denise Reinke, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker, Martee L. Hensley

Data analysis and interpretation: Robert G. Maki, J. Kyle Wathen, Shreyaskumar R. Patel, Dennis A. Priebat, Michael Fanucchi, David C. Harmon, Scott M. Schuetze, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker, Martee L. Hensley

Manuscript writing: Robert G. Maki, J. Kyle Wathen, Shreyaskumar R. Patel, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker, Martee L. Hensley

Final approval of manuscript: Robert G. Maki, J. Kyle Wathen, Shreyaskumar R. Patel, Dennis A. Priebat, Scott H. Okuno, Brian Samuels, Michael Fanucchi, David C. Harmon, Scott M. Schuetze, Denise Reinke, Peter F. Thall, Robert S. Benjamin, Laurence H. Baker, Martee L. Hensley

	Appendix

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Reference for clinical trial design. The clinical trial design and modeling of different potential outcomes are described by Thall and Wathen (Thall PF, Wathen JK: Stat Med 24:1947-1964, 2005), the abstract for which may be found at http://www3.interscience.wiley.com/cgi-bin/abstract/110433019/.

Clinical outcomes based on Eastern Cooperative Oncology Group performance status. Table A3 provides the number of patients and events for the two subgroups stratified by Eastern Cooperative Oncology Group performance status (ECOG PS). A greater proportion of patients were enrolled to the gemcitabine-docetaxel arm regardless of ECOG PS, indicating that gemcitabine-docetaxel was the superior therapy for outcome as defined for both subsets of patients.

PFS. Regarding the model derived for progression-free survival (PFS) and overall survival (OS), a positive parameter value for a given covariate corresponds to longer time to the outcome. Denoting the coefficient for a given covariate by ß, we denote the Bayesian posterior probability Pr(ß > 0 | data). To interpret the results, values of Pr(ß_j > 0 | data) closer to either 0 or 1 correspond to a stronger effect of the predictor, and values close to .5 correspond to no effect.

Tables A1, A2, and Figure A1 show details of the PFS estimates based on a fitted Bayesian log-normal model containing prior pelvic radiation (PPR), leiomyosarcoma (LMS), and ECOG PS covariates. Denoting the median PFS for gemcitabine-docetaxel by M_G+D and gemcitabine by M_G, Pr(M_G+D > M_G  data) = .98, given that there is not a treatment by covariate interaction, and this value is the same for all groups. The Bayesian 95% posterior credible interval for each subgroup is listed in Table A2 as well. Of the three covariates, ECOG PS had the greatest impact on PFS. Patients with an ECOG PS 0 had longer PFS than patients with PS greater than 0 (Table A1). Table A2 lists the median PFS, with 95% Bayesian posterior credible intervals for each of the eight subgroups determined by the covariates. Kaplan-Meier PFS estimates for gemcitabine-docetaxel and gemcitabine alone are shown in Figure A1.

OS. Tables A5 and A4 and Figure A2 summarize the fitted Bayesian log-normal model for OS for each of the four subgroups. A positive parameter value for a given covariate corresponds to longer OS. As above, denoting the coefficient for a given covariate by ß, we denote the Bayesian posterior probability Pr(ß> 0 | data) by Pr more than 0. To interpret the results, values of Pr(ß_j > 0 | data) closer to either 0 or 1 correspond to a stronger effect of the predictor, and values close to .5 correspond to no effect.

According to this model, the posterior probability that the median OS with gemcitabine-docetaxel is greater than with gemcitabine alone is Pr(ß_GD > 0 | data) = .97. This probability is the same for all subgroups because there was no treatment-covariate interaction (Tables A5 and A4). Of the three covariates, LMS histology had the greatest impact on OS. Patients with LMS had longer OS than patients with other histologies. Patients with ECOG PS more than 0 had inferior survival to those with PS = 0 (Tables A5 and A4). Patients who had PPR had inferior OS compared to patients who did not have PPR. Table A4 lists the median OS with 95% Bayesian posterior credible intervals for each of the eight subgroups determined by the covariates. Figure A2 demonstrates the posterior median OS for each treatment for four different patient subgroups (LMS or other histology, PPR or not). Table A4 and Figure A2 indicate that gemcitabine-docetaxel had a substantive advantage over gemcitabine in terms of OS. The posterior probability distribution curves for OS associated with selected subgroups are indicated in Figures A3 (A and B).

Covariates in the survival models. For each patient, three covariates are recorded: disease histology, PPR, and ECOG PS. Disease histology is denoted by [LMS] = 1 if a patient's histology was LMS and [LMS] = 0 otherwise. We denote [PPR] = 1 if the patient received PPR and [PPR] = 0 if not, and [PS] = 0 if a patient's ECOG PS was 0 and [PS] = 1 otherwise. We denote treatment by [GD] where [GD] = 1 if the patient was treated with gemcitabine-docetaxel and [GD] = 0 if the patient was treated with gemcitabine alone. Thus, the model assumed gemcitabine as the baseline treatment, and ß₁ more than 0 corresponds to longer PFS or OS with gemcitabine-docetaxel, ß₂ more than 0 corresponds to longer PFS or OS with PPR, and so on.

For covariate modeling, we denoted a patient's PFS or OS by x. For the purposes of the model, x log-normal(µ, ^–1), where denotes the precision, and where µ = ß₀ + ß₁[GD] + ß₂ [PPR] + ß₃ [LMS] + ß₄ [PS]. Assumed priors included ß₀, ß₁, ß₂, ß₃, ß₄ normal(0, 1000), exponential(mean = 100).

View larger version (13K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig A1. Kaplan-Meier progression-free survival (PFS) estimates for gemcitabine-docetaxel and gemcitabine alone for the four subgroups examined prospectively in this study. (A) Prior pelvic radiation (PPR), leiomyosarcoma (LMS) subgroup; (B) no PPR, LMS subgroup; (C) PPR, non-LMS subgroup; (D) no PPR, non-LMS subgroup. Note the difference in the abscissa for the different figures.

View larger version (13K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig A2. Kaplan-Meier overall survival (OS) estimates for gemcitabine-docetaxel and gemcitabine alone for the four subgroups examined prospectively in this study. (A) Prior pelvic radiation (PPR), leiomyosarcoma (LMS) subgroup; (B) no PPR, LMS subgroup; (C) PPR, non-LMS subgroup; (D) no PPR, non-LMS subgroup. Note the difference in the abscissa for the different figures.

View larger version (12K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig A3. Posterior probability distributions for median overall survival (OS) for selected subgroups. (A) Nonleiomyosarcoma (non-LMS), no prior pelvic radiation (PPR), Eastern Cooperative Oncology Group performance status (PS) = 0 patient subgroup; (B) non-LMS, no PPR, PS more than 0 subgroup. G, gemcitabine arm; G + D, gemcitabine-docetaxel arm.

	NOTES

 
Supported by the Kristen Ann Carr Fund, Eli Lilly & Co, and Sanofi-aventis; and in part by a National Cancer Institute program project Grant No. P01-CA47179, the Shuman Fund for GIST Research, and spin4survival.org (R.G.M.).

Presented in part at the 42nd Annual Meeting of the American Society of Clinical Oncology, June 2-6, 2006, Atlanta, GA.

R.G.M. and J.K.W. contributed equally to this manuscript.

Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.

	REFERENCES

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
1. Jemal A, Siegel R, Ward E, et al: Cancer statistics, 2006. CA Cancer J Clin 56:106-130, 2006[Abstract/Free Full Text]

2. Brennan MF, Singer S, Maki RG, et al: Soft tissue sarcoma, in DeVita VT Jr, Hellman S, Rosenberg SA (eds): Cancer: Principles & Practice of Oncology (ed 7). Philadelphia, PA, Lippincott Williams & Wilkins, 2005, pp 1581-1637

3. Merimsky O, Meller I, Flusser G, et al: Gemcitabine in soft tissue or bone sarcoma resistant to standard chemotherapy: A phase II study. Cancer Chemother Pharmacol 45:177-181, 2000[CrossRef][Medline]

4. Patel SR, Gandhi V, Jenkins J, et al: Phase II clinical investigation of gemcitabine in advanced soft tissue sarcomas and window evaluation of dose rate on gemcitabine triphosphate accumulation. J Clin Oncol 19:3483-3489, 2001[Abstract/Free Full Text]

5. Svancárová L, Blay JY, Judson IR, et al: Gemcitabine in advanced adult soft-tissue sarcomas: A phase II study of the EORTC Soft Tissue and Bone Sarcoma Group. Eur J Cancer 38:556-559, 2002[CrossRef][Medline]

6. Okuno S, Ryan LM, Edmonson JH, et al: Phase II trial of gemcitabine in patients with advanced sarcomas (E1797): A trial of the Eastern Cooperative Oncology Group. Cancer 97:1969-1973, 2003[CrossRef][Medline]

7. Verweij J, Catimel G, Sulkes A, et al: Phase II studies of docetaxel in the treatment of various solid tumours: EORTC Early Clinical Trials Group and the EORTC Soft Tissue and Bone Sarcoma Group. Eur J Cancer 31A:S21-S24, 1995 (suppl 4)

8. Edmonson JH, Ebbert LP, Nascimento AG, et al: Phase II study of docetaxel in advanced soft tissue sarcomas. Am J Clin Oncol 19:574-576, 1996[CrossRef][Medline]

9. Köstler WJ, Brodowicz T, Attems Y, et al: Docetaxel as rescue medication in anthracycline- and ifosfamide-resistant locally advanced or metastatic soft tissue sarcoma: Results of a phase II trial. Ann Oncol 12:1281-1288, 2001[Abstract/Free Full Text]

10. Tempero M, Plunkett W, Ruiz Van Haperen V, et al: Randomized phase II comparison of dose-intense gemcitabine: Thirty-minute infusion and fixed dose rate infusion in patients with pancreatic adenocarcinoma. J Clin Oncol 21:3402-3408, 2003[Abstract/Free Full Text]

11. Hensley ML, Maki R, Venkatraman E, et al: Gemcitabine and docetaxel in patients with unresectable leiomyosarcoma: Results of a phase II trial. J Clin Oncol 20:2824-2831, 2002[Abstract/Free Full Text]

12. Leu KM, Ostruszka LJ, Shewach D, et al: Laboratory and clinical evidence of synergistic cytotoxicity of sequential treatment with gemcitabine followed by docetaxel in the treatment of sarcoma. J Clin Oncol 22:1706-1712, 2004[Abstract/Free Full Text]

13. Bay JO, Ray-Coquard I, Fayette J, et al: Docetaxel and gemcitabine combination in 133 advanced soft-tissue sarcomas: A retrospective analysis. Int J Cancer 119:706-711, 2006[CrossRef][Medline]

14. Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumors: European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:205-216, 2000[Abstract/Free Full Text]

15. Thall PF, Wathen JK: Covariate-adjusted adaptive randomization in a sarcoma trial with multi-stage treatments. Stat Med 24:1947-1964, 2005[CrossRef][Medline]

16. Gelman A, Carlin JB, Stern HS, et al: Bayesian Data Analysis. New York, NY, Chapman & Hall, 1995

17. Raftery AE: Approximate Bayes factors and accounting for model uncertainty in generalised linear models. Biometrika 83:251-266, 1996[Abstract/Free Full Text]

18. Spiegelhalter D, Thomas A, Best N, et al: Bayesian inference using Gibbs Sampling Manual (BUGS 0.5). Cambridge, United Kingdom, MRC Biostatistics Unit, 1995

19. Venables WN, Ripley BD: Modern Applied Statistics with S-Plus (ed 3). New York, NY, Springer, 1999

20. Antman K, Crowley J, Balcerzak SP, et al: An intergroup phase III randomized study of doxorubicin and dacarbazine with or without ifosfamide and mesna in advanced soft tissue and bone sarcomas. J Clin Oncol 11:1276-1285, 1993[Abstract/Free Full Text]

21. Van Glabbeke M, van Oosterom AT, Oosterhuis JW, et al: Prognostic factors for the outcome of chemotherapy in advanced soft tissue sarcoma: An analysis of 2,185 patients treated with anthracycline-containing first-line regimens—A European Organization for Research and Treatment of Cancer Soft Tissue and Bone Sarcoma Group Study, J Clin Oncol 17:150-157, 1999[Abstract/Free Full Text]

22. Edmonson JH, Ryan LM, Blum RH, et al: Randomized comparison of doxorubicin alone versus ifosfamide plus doxorubicin or mitomycin, doxorubicin, and cisplatin against advanced soft tissue sarcomas. J Clin Oncol 11:1269-1275, 1993[Abstract/Free Full Text]

23. Santoro A, Tursz T, Mouridsen H, et al: Doxorubicin versus CYVADIC versus doxorubicin plus ifosfamide in first-line treatment of advanced soft tissue sarcomas: A randomized study of the European Organization for Research and Treatment of Cancer Soft Tissue and Bone Sarcoma Group. J Clin Oncol 13:1537-1545, 1995[Medline]

24. Borden EC, Amato DA, Rosenbaum C, et al: Randomized comparison of three adriamycin regimens for metastatic soft tissue sarcomas. J Clin Oncol 5:840-850, 1987[Abstract/Free Full Text]

25. Fletcher CDM, van den Berg E, Molenaar WM: Pleomorphic malignant fibrous histiocytoma/undifferentiated high grade pleomorphic sarcoma, in Kleihues P, Sobin LH (eds): World Health Organization Classification of Tumors. Lyon, France, IARC Press, 2002, pp 120-122

26. Baird K, Davis S, Antonescu CR, et al: Gene expression profiling of human sarcomas: Insights into sarcoma biology. Cancer Res 65:9226-9235, 2005[Abstract/Free Full Text]

27. Fletcher CD, Gustafson P, Rydholm A, et al: Clinicopathologic re-evaluation of 100 malignant fibrous histiocytomas: Prognostic relevance of subclassification. J Clin Oncol 19:3045-3050, 2001[Abstract/Free Full Text]

28. Thompson WR: On the likelihood that one unknown probability exceeds another in view of the evidence of two samples. Biometrika 25:285-294, 1933[Free Full Text]

29. Bartlett RH, Roloff DW, Cornell RG, et al: Extracorporeal circulation in neonatal respiratory failure: A prospective randomized study. Pediatrics 76:479-487, 1985[Abstract/Free Full Text]

30. Dougherty TB, Porche VH, Thall PF: Maximum tolerated dose of nalmefene in patients receiving epidural fentanyl and dilute bupivacaine for postoperative analgesia. Anesthesiology 92:1010-1016, 2000[CrossRef][Medline]

31. Grieve AP, Krams M: ASTIN: A Bayesian adaptive dose-response trial in acute stroke. Clin Trials 2:340-351, 2005; discussion 352-358, 364-378[Abstract/Free Full Text]

32. Krams M, Lees KR, Hacke W, et al: Acute Stroke Therapy by Inhibition of Neutrophils (ASTIN): an adaptive dose-response study of UK-279,276 in acute ischemic stroke. Stroke 34:2543-2548, 2003[Abstract/Free Full Text]

33. Brophy JM, Belisle P, Joseph L: Evidence for use of coronary stents: A hierarchical Bayesian meta-analysis. Ann Intern Med 138:777-786, 2003[Abstract/Free Full Text]

34. US Food and Drug Administration.Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials: Draft Guidance for Industry and FDA Staff. Bethesda, MD, US Food and Drug Administration, 2006

35. Muler JH, McGinn CJ, Normolle D, et al: Phase I trial using a time-to-event continual reassessment strategy for dose escalation of cisplatin combined with gemcitabine and radiation therapy in pancreatic cancer. J Clin Oncol 22:238-243, 2004[Abstract/Free Full Text]

36. Pisters PW, Patel SR, Prieto VG, et al: Phase I trial of preoperative doxorubicin-based concurrent chemoradiation and surgical resection for localized extremity and body wall soft tissue sarcomas. J Clin Oncol 22:3375-3380, 2004[Abstract/Free Full Text]

37. Thall PF, Cook JD: Dose-finding based on efficacy-toxicity trade-offs. Biometrics 60:684-693, 2004[CrossRef][Medline]

38. Thall PF, Simon RM, Estey EH: Bayesian sequential monitoring designs for single-arm clinical trials with multiple outcomes. Stat Med 14:357-379, 1995[Medline]

39. Giles FJ, Kantarjian HM, Cortes JE, et al: Adaptive randomized study of idarubicin and cytarabine versus troxacitabine and cytarabine versus troxacitabine and idarubicin in untreated patients 50 years or older with adverse karyotype acute myeloid leukemia. J Clin Oncol 21:1722-1727, 2003[Abstract/Free Full Text]

40. Calverley PM, Anderson JA, Celli B, et al: Salmeterol and fluticasone propionate and survival in chronic obstructive pulmonary disease. N Engl J Med 356:775-789, 2007[Abstract/Free Full Text]

41. Rabe KF: Treating COPD: the TORCH trial, P values, and the Dodo. N Engl J Med 356:851-854, 2007[Free Full Text]

42. Van Glabbeke M, Verweij J, Judson I, et al: Progression-free rate as the principal end-point for phase II trials in soft-tissue sarcomas. Eur J Cancer 38:543-549, 2002[CrossRef][Medline]

Submitted December 20, 2006; accepted April 9, 2007.

This article has been cited by other articles:

				S. L. Spunt, S. X. Skapek, and C. M. Coffin Pediatric Nonrhabdomyosarcoma Soft Tissue Sarcomas Oncologist, June 1, 2008; 13(6): 668 - 678. [Abstract] [Full Text] [PDF]

				I. Ray-Coquard, A. Le Cesne, J. S. Whelan, P. Schoffski, B. N. Bui, J. Verweij, S. Marreaud, M. van Glabbeke, P. Hogendoorn, and J.-Y. Blay A Phase II Study of Gefitinib for Patients with Advanced HER-1 Expressing Synovial Sarcoma Refractory to Doxorubicin-Containing Regimens Oncologist, April 1, 2008; 13(4): 467 - 473. [Abstract] [Full Text] [PDF]

				R. G. Maki Gemcitabine and Docetaxel in Metastatic Sarcoma: Past, Present, and Future Oncologist, August 1, 2007; 12(8): 999 - 1006. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Publisher's Note Erratum (v25,p3790) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted