Presentation on theme: "Abstract 1000 Phase III Trial of Gemcitabine plus Docetaxel (GD) Compared to Capecitabine plus Docetaxel (CD) with Planned Crossover to the Alternate Single."— Presentation transcript:
1Abstract 1000Phase III Trial of Gemcitabine plus Docetaxel (GD) Compared to Capecitabine plus Docetaxel (CD) with Planned Crossover to the Alternate Single Agent in Metastatic Breast Cancer (MBC)Andrew D. Seidman, Adam Brufsky, Rafat H. Ansari, James R. Rubinsak, Richard S. Stein, Lee S. Schwartzberg, John F. Stewart, Luping Zhao, John F. Gill, D. Fritz TaiGood morning. On behalf of my co-investigators, I am pleased to present the results of this study.
2DISCLOSURES This work was funded by Lilly USA, LLC. Dr. Seidman has been paid by Lilly and Sanofi-Aventis as a consultant and speaker.Dr. Brufsky has been paid by Lilly as a consultant and speaker.Drs. Ansari, Rubinsak, Stein, Schwartzberg, and Stewart have no disclosures.Drs. Zhao, Gill, and Tai are employees and shareholders of Lilly USA, LLC.I have been remunerated by Eli Lilly Company and Sanofi-Aventis for consulting and speaking services; other disclosures are as noted.
3BACKGROUNDDocetaxel (D) + capecitabine (C) improves response rate (RR) and time to progressive disease (TTP) over docetaxel as first-line treatment of MBC after prior anthracycline.1Paclitaxel + gemcitabine (G) improved RR and TTP over paclitaxel as first-line treatment of MBC after prior adjuvant anthracycline therapy.2GD and CD yielded nearly identical efficacy in anthracycline- pretreated MBC, but GD resulted in improved TTF due to toxicity-related treatment discontinuation with CD.3In patients with prior anthracycline exposure, the addition of capecitabine to docetaxel and the addition of gemcitabine to paclitaxel has resulted in improved response rates and TTP compared to the taxane monotherapy arm. As recently reported, the doublet of docetaxel plus gemcitabine yielded nearly identical RR and PFS as docetaxel plus capecitabine, with superior TTF for GD due to greater toxicity-related treatment discontinuation with CD.O’Shaughnessy J, et al. JCO 2002Albain KS, et al. JCO 2008Chan S, et al. JCO 2009
4RATIONALEBoth combination and single-agent chemotherapy have a role in the management of MBC.Few large, randomized, Phase III trials have examined pre-planned crossover chemotherapy strategies in MBC.Prior Phase III trials of CD have been complicated by treatment-limiting palmar-plantar erythrodysesthesia (PPE), resulting in frequent capecitabine dose reduction outside the clinical trial setting.This study compared safety and efficacy of GD and CD in patients with MBC, where the alternate, crossover monotherapy (GDC or CDG) was predetermined.Despite much debate on the issue, both poly- and mono-chemotherapy play a role in the management of MBC. Few randomized phase III trials have examined pre-planned crossover chemotherapy strategies in MBC. Prior phase III trials of docetaxel plus capecitabine have been notable for treatment-limiting PPE resulting in frequent capecitabine dose reduction outside the clinical trial setting. This study compared the safety and efficacy of GD and CD among patients with MBC where the alternate crossover monotherapy was predetermined.4
5STUDY OBJECTIVES Primary Secondary TTP Toxicity Overall response rate (ORR)Progression-free survival (PFS)Overall survival (OS)The primary objective of this trial was time to disease progression. Secondary objectives were to compare toxicity, response rate, PFS and OS.5
6STUDY DESIGN AND TREATMENT Patient Stratification FactorsFirst- versus second-line treatmentPrior anthracycline therapyVisceral dominant diseaseECOG PSMeasurable versus non-measurable (evaluable) only diseaseRANDOMIZEGD Induction:G 1000 mg/m2IV on Days 1, D 75 mg/m2 IV on Day 1every 21 days until disease progression (PD)CD Induction:D 75 mg/m2IV on Day 1 +C 1000 mg/m2PO BID on Days 1-14every 21 days until PDG Crossover:IV on Days 1,8 every 21days until PDC Crossover:Off studyFollow-up(1:1)After stratification for line of therapy, prior anthracycline exposure, visceral dominant disease, performance status, and measurable vs. evaluable-only disease, patients were randomized 1:1 to GD induction to be followed by capecitabine upon PD or prohibitive toxicity or CD induction to be followed by gemcitabine monotherapy upon PD or prohibitive toxicity. The docetaxel-gemcitabine regimen has docetaxel at 75 mg/m2 every 3 weeks, with gemcitabine 1000 mg/m2 days 1 and 8 every 3 weeks. The docetaxel-capecitabine arm employs a 20% lower dose of capecitabine than in both the O’Shaughnessy and Chan trials, given the high incidence of prohibitive hand-foot syndrome in those trials. At crossover, the monotherapy regimens were identical to the single agent non-taxane components of the induction doublets. Treatment was to progression or prohibitive toxicity.6
7MAJOR ELIGIBILITY CRITERIA Prior taxane therapy for MBC.Prior gemcitabine or capecitabine therapy.Concurrent trastuzumab therapy.CNS metastases.InclusionExclusionLocally advanced or metastatic breast cancer.No more than one prior course of chemotherapy for MBC.Measurable and non-measurable (evaluable) disease.Adequate renal, hepatic, and bone marrow function.ECOG PS ≤ 1Eligible patients had locally advanced or metastatic breast cancer, no more than one prior chemotherapy regimen for metastatic disease, measurable or evaluable disease, adequate end-organ function and performance status. Patients were excluded if they had prior taxane for MBC, prior gemcitabine or capecitabine, concurrent trastuzumab or central nervous system metastases.7
8STATISTICAL CONSIDERATIONS Using Freedman’s method, it was estimated that 442 patients (221 per arm) would be needed to obtain the 385 progression events required to observe a 2-month difference in TTP between GD and CD treatment arms with an 80% statistical power.Time‑to‑event analyses were estimated from date of randomization using the Kaplan-Meier method and results for each arm were compared by log-rank test.All 2‑sided statistical comparisons between the treatment arms were judged relative to a significance level of α=0.05.Data-lock for this final analysis was on 17 March Data management was performed by query analyses and data reviews.An exploratory, post-hoc analysis of the sum of induction and crossover TTP was performed.The sample size was driven by the comparison in TTP between the GD and CD arms. Using Freedman’s method, it was estimated that 442 patients would be needed to obtain the 385 progression events required to observe a 2-month difference in TTP between treatment arms with 80% statistical power. Time to event analyses were estimated from the date of randomization using the Kaplan-Meier method, and results were compared by log-rank test. 2-sided statistical comparisons between treatment arms were judged relative to an alpha of Data lock for this final analysis was on March 17th, 10 weeks ago. Finally, an exploratory, post-hoc analysis of the sum of induction and crossover TTP was performed.8
9Patients randomly assigned PATIENT DISPOSITIONPatients randomly assigned(N = 475)Accrual: 15 February 2002 –23 December 2008GD Arm(n = 239)ITTN = 475CD Arm(n = 236)Withdrew (n = 3)Withdrew (n = 9)SafetyN = 463GD Arm(n = 236)CD Arm(n = 227)Discontinued (n = 236)PD (n = 98)Adverse event (n = 43)Investigator decision (n = 40)Patient request (n = 38)Other (n = 17)Discontinued (n = 227)PD (n = 83)Adverse event (n = 67)Investigator decision (n = 31)Patient request (n = 27)Other (n = 19)CrossoverN = 158GDC(n = 77)CDG(n = 81)In a trial such as this where two lines of therapy are planned, a careful accounting of patient disposition is critical.475 patients comprised the intent-to-treat population, and 463 were evaluable for safety. 43 patients (point) exited the trial due to an adverse event during first-line therapy with GD, while 67 patients (point) stopped protocol therapy due to an adverse event on the CD arm. Other reasons for patient removal from the trial were fairly balanced. A total of 158 patients (point), or one-third of the ITT population, proceeded to the crossover monotherapy. The majority of patients receiving monotherapy discontinued it due to disease progression (point), in contrast to the induction phase patients above.Withdrew (n = 1)Withdrew (n = 1)Discontinued (n = 76)PD (n = 50)Adverse event (n = 8)Investigator decision (n = 7)Patient request (n = 6)Other (n = 5)Discontinued (n = 80)PD (n = 53)Adverse event (n = 8)Investigator decision (n = 7)Patient request (n = 9)Other (n = 3)PD = Progressive Disease; ITT = Intent-to-Treat9
10PHASE III MBC TRIALS Pre-Planned Sequential Lines of Chemotherapy StudyFirst-LineTherapySecond-Line% CrossoverSecond-Line/First-LineSledge et al. JCO, 2003DoxorubicinPaclitaxel129/245 (52.6%)128/242 (52.9%)Paridaens et al. JCO 200077/165 (46.7%)91/166 (54.8%)Joensuu et al. JCO 1988CEFEpirubicinMVMitomycin C88/150 (58.7%)74/153 (50.3%)Seidman et al. ASCO 2009GDCDCapecitabineGemcitabine77/237 (32.5%)79/226 (35.0%)Several prior phase III trials have attempted to deliver two consecutive sequential lines of chemotherapy for metastatic breast cancer. In the ECOG 1193 trial, and the EORTC trial studying sequential doxorubicin and paclitaxel, approximately 50% of the study population actually crossed over to the designated second-line therapy. In the Joensuu trial, after combination chemotherapy, about one-half of patients went on to receive the preplanned subsequent single agent regimen. In the current trial, only about one-third of patients crossed over from the induction taxane-based combination to the designated preplanned monotherapy.CEF = cyclophosphamide, epirubicin, 5-fluorouracilMV = mitomycin C, vinblastineGD = gemcitabine + docetaxelCD = capecitabine + docetaxel10
11PATIENT CHARACTERISTICS Baseline ParameterGDN=239CDN=236Age, median (range)57 (27-81)54 (27-82)Age group, n (%)≤65 years>65 years188 (78.7)51 (21.3)193 (81.8)43 (18.2)ECOG performance status, n (%)1175 (73.2)60 (25.1)176 (74.6)55 (23.3)ER status, n (%)PositiveNegative135 (56.5)89 (37.2)134 (56.8)80 (33.9)PR status, n (%)102 (42.7)114 (47.7)115 (48.7)102 (43.2)Visceral dominant disease, n (%)YesNo158 (66.1)79 (33.1)161 (68.2)73 (30.9)The median ages for patients on the GD and CD arms were 57 and 54, respectively. Approximately three quarters of patients had an ECOG PS of 0. Treatment arms were well-balanced for hormone receptor status, and approximately 2/3 of patients in each arm had visceral-dominant disease.ECOG = Eastern Cooperative Oncology Group;ER = estrogen receptor; PR = progesterone receptor11
12PATIENT CHARACTERISTICS Prior Therapy ParameterGDN=239CDN=236Prior chemotherapy for MBC, n (%)Yes26 (10.9)25 (10.6)No211 (88.3)208 (88.1)Prior anthracycline therapy, n (%)136 (56.9)133 (56.4)99 (41.4)99 (41.9)Prior hormonal therapy, n (%)135 (57.2)101 (42.3)100 (42.4)Prior adjuvant taxane therapy, n (%)46 (19.2)47 (19.9)31 (13.0)Unknown162 (67.8)164 (69.5)The large majority of patients treated were to receive first and second-line chemotherapy on this trial; about 11% (point) were intended to receive protocol therapy as second- and third-line chemotherapy. More than half of all patients had prior anthracycline (point), and the arms were well-balanced for prior antiestrogen therapy and prior taxane exposure.12
13DRUG ADMINISTRATION * Total number of patients with dose adjustment. ParameterGDN=237CDN=226GDCN=77CDGN=79Total cycles received, n17681668390344Mean cycles received, n22.214.171.124.4Median cycles received, n63Dose adjustments, docetaxel, n (%) *117 (49.4)106 (46.9)-Dose adjustments, gemcitabine, n (%)211 (89.0)39 (49.4)Dose adjustments, capecitabine, n (%)190 (84.1)40 (51.9)Relative dose intensity, docetaxel, %91.091.7Relative dose intensity, gemcitabine, %72.587.5Relative dose intensity, capecitabine, %77.492.8Patients received a median of 6 cycles of combination chemotherapy and 3 cycles of subsequent monotherapy. Docetaxel dose reduction or delay occurred in less than half of patients on both induction arms (point). Dose adjustments for gemcitabine and capecitabine were also made in over 80% of patients in the combination arms (point), and in approximately half of patients during subsequent monotherapy with these agents (point). Despite this, the relative dose intensity for docetaxel was over 90% (point). The RDI for the non-taxane components in the combination phase was approximately 75% (point), and approximately 90% during the crossover monotherapy phase (point).* Total number of patients with dose adjustment.13
14DRUG-RELATED TOXICITY (Induction) Hematologic Adverse event, n (%)*GDN=237CDN=226PGrade 3-4 neutropenia181 (76.4)69 (30.5)< .001Grade 3-4 leukopenia68 (28.7)17 (7.5)Grade 3-4 thrombocytopenia19 (8.0)0 (0.0)Grade 3-4 anemia10 (4.2)8 (3.5).812Grade 3-4 febrile neutropenia17 (7.2)14 (6.2).713Grade ¾ neutropenia, leukopenia, and thrombocytopenia were statistically more common with GD than CD. There was no difference in the incidence of grade ¾ anemia nor in the incidence of febrile neutropenia. Prophylactic hematopoietic growth factor support was not allowed.* Adverse events were defined using NCI-CTC, version 2.0.Prophylactic use of hematopoietic growth factors was not allowed14
15DRUG-RELATED TOXICITY (Induction) Nonhematologic Adverse event, n (%)*GDN=237CDN=226PGrade 3-4 fatigue25 (10.5)12 (5.3).041Grade 3-4 ALT/AST increased6 (2.5)0 (0.0).031Grade 3-4 nausea/vomiting8 (3.4)18 (8.0).042Grade 3-4 hand-foot syndrome3 (1.3)57 (25.2)< .001Grade 3-4 mucositis10 (4.4).049Grade 2-4 diarrhea54 (22.8)50 (22.1).912Grades ¾ non-hematologic toxicities are shown here. Fatigue and elevation of hepatic transaminase was more common with GD. Nausea and vomiting, hand-foot syndrome, and mucositis were more common with CD.* Adverse events were defined using NCI-CTC, version 2.0.ALT = alanine aminotransferase; AST = aspartate aminotansferase15
16EFFICACY Tumor Response (Measurable Disease) Best Overall ResponseGDCDP*Induction patients, n207191Induction response, n (%)72 (34.8)78 (40.8).216GDCCDGCrossover patients, n7270Crossover response, n (%)11 (15.3)5 (7.1).184A total of 398 patients with measurable disease were evaluable for response to induction therapy, and 142 for the subsequent monotherapy phase. 34.8% of patients responded to GD, and 40.8% to CD, a non-significant difference. 15.3% patients who crossed over after GD responded to capecitabine, whereas 7.1% of patients who crossed over from CD to gemcitabine responded to this single agent, a non-significant difference.* Fisher’s exact test16
17Patients censored, n (%) OVERALL SURVIVAL1.00.80.60.40.20.0OS, monthsGD (N=239)CD (N=236)PPatients censored, n (%)75 (31.4)72 (30.5)Median, (95% CI)23.0 (18.8, 25.7)23.3 (18.6, 25.5).785Survival ProbabilityIntent-to-treat populationThe intent-to-treat survival curves for patients in this trial are virtually superimposable, and was just under 2 years.Time to Death (Months)17
18TIME TO PROGRESSIVE DISEASE Induction Phase 1.00.80.60.40.20.0TTP, monthsGD (N=239)CD (N=236)PPatients censored, n (%)68 (28.5)83 (35.2)Median, (95% CI)9.3 (7.7, 10.8)8.9 (7.4, 11.1).385Despite over-accrual, censoring resulted in only324 TTP events, compared to 385 planned events.Progression-Free ProbabilityIntent-to-treat populationSimilarly, there was no significant difference in median time to progression between the GD and CD induction arms; 9.3 months for GD and 8.9 months for CD. Despite over-accrual, censoring resulted in 324 TTP events compared to the 385 planned events.Time to Progressive Disease (Months)18
19TIME TO PROGRESSIVE DISEASE Crossover Phase 1.00.80.60.40.20.0TTP, monthsGDC (n=77)CDG (n=81)PPatients censored, n (%)13 (16.9)10 (12.3)Median, (95% CI)4.5 (2.1, 7.8)2.3 (2.0, 3.8).145Progression-Free ProbabilityIntent-to-treat populationThe median TTP on capecitabine montherapy for those 77 patients who did indeed crossover after GD was 4.5 months. The median TTP on gemcitabine montherapy for the 81 patients who crossed over from CD was 2.3 months, a non-significant difference.Time to Progressive Disease (Months)19
20TTP INDUCTION-CROSSOVER SUM Definition FurtherPDCrossoverBaselineScanInductionPDPatient 1TTP1DiscontinuedPatient 2TTP2TTP Induction-Crossover Sum is a sub-set analysis of only those patients whoreceived single-agent crossover therapy (C or G).Induction TTP (TTP 1) was estimated for all patients from time of randomizationto first PD.Crossover TTP (TTP2) was estimated for all crossover patients from the time offirst single-agent dose to further PD.TTP Induction-Crossover Sum was calculated as TTP1 + TTP2.In an exploratory analysis, we examined the Summation Time to Progression for those patients who received single agent crossover therapy. Induction TTP, or TTP1, was estimated for all patients from time of randomization to first progression. Crossover TTP, or TTP2, was from the time of first single agent dose to further disease progression. The induction-crossover sum was calculated as TTP1 + TTP2.
21TIME TO PROGRESSIVE DISEASE Summary ParameterGDGDCCDCDGPInduction patients, n239236Induction, median, months(95% CI)9.3(7.7, 10.8)8.9(7.4, 11.1).385Crossover patients, n7680Crossover, median, months4.5(2.1, 7.8)2.3(2.0, 3.8).145Induction-crossover sum* patients, nInduction-crossover sum,* median, months14.3(10.4, 17.8)9.2(7.1, 11.6).093Recognizing that only one-third of the population crossed over from induction phase to crossover monotherapy, we nevertheless analyzed the sum of TTP for crossover patients in the two arms. (POINT) The median induction-crossover summation TTP was 14.3 months for the GD to C patients, and 9.2 months for the CD to G patients, with a p-value of* Exploratory, post-hoc analysis21
22TIME TO PROGRESSIVE DISEASE Induction-Crossover Sum 1.00.80.60.40.20.0TTP, monthsGDC (N=76)CDG (N=80)PPatients censored, n (%)13 (17.1)10 (12.5)Median, (95% CI)14.3 (10.4, 17.8)9.2 (7.1, 11.6).093Exploratory, post-hoc analysisProgression-Free ProbabilityTreated PatientsThe Kaplan-Meier time to progressive disease curves from protocol initiation through two lines of chemotherapy are shown here for those patients who indeed crossed over. Recognizing that crossover patients might not be balanced for prognostic features, we examined this possibility (next slide)….Time to Progressive Disease (Months)22
23PATIENT CHARACTERISTICS Crossover Population ParameterGDCN=77CDGN=81Age, median (range)56 (27-74)54 (27-82)Age group, n (%)≤65 years>65 years65 (84.4)12 (15.6)68 (84.0)13 (16.1)ECOG performance status, n (%)161 (79.2)14 (18.2)65 (80.2)15 (18.5)ER status, n (%)PositiveNegative40 (51.9)31 (40.3)40 (49.4) 29 (35.8)PR status, n (%)33 (42.9)35 (45.5)34 (42.0)36 (44.4)Visceral dominant disease, n (%)YesNo55 (71.4)22 (28.6)55 (67.9)26 (32.1)While this is an exploratory analysis, it is noteworthy that the patients who crossed over were well-balanced for baseline demographic features such as age, performance status (which was also balanced at crossover), hormone receptor status and the presence of visceral dominant disease.ECOG = Eastern Cooperative Oncology Group;ER = estrogen receptor; PR = progesterone receptor23
24PATIENT CHARACTERISTICS Prior Therapy: Crossover Population ParameterGDN=77CDN=81Prior chemotherapy for MBC, n (%)Yes4 (5.2)7 (8.6)No73 (94.8)74 (91.4)Prior anthracycline therapy, n (%)39 (50.6)50 (61.7)37 (48.1)30 (37.0)Prior hormonal therapy, n (%)42 (54.5)42 (51.9)34 (44.2)38 (46.9)Prior adjuvant taxane therapy, n (%)16 (20.8)14 (17.3)7 (9.1)10 (12.4)Unknown54 (70.1)57 (70.4)Crossover patients were also balanced for line of therapy, and prior exposure to anthracycline, antiestrogen, and taxane. (While we unfortunately failed to capture prior adjuvant taxane use for a significant proportion of the study population, it is noteworthy that prior taxane exposure did not seem to bias against benefit from paclitaxel and bevacizumab in ECOG 2100).24
25PATIENT CHARACTERISTICS Baseline All PatientsCrossover PatientsParameterGDN=239CDN=236GDCN=77CDGN=81ECOG performance status, n (%)1175 (73.2)60 (25.1)176 (74.6)55 (23.3)61 (79.2)14 (18.2)65 (80.2)15 (18.5)Prior chemotherapy for MBC, n (%)YesNo26 (10.9)211 (88.3)25 (10.6)208 (88.1)4 (5.2)73 (94.8)7 (8.6)74 (91.4)Visceral dominant disease, n (%)158 (66.1)79 (33.1)161 (68.2)73 (30.9)55 (71.4)22 (28.6)55 (67.9)26 (32.1)Finally, the one-third of patients crossing over in this study had similar baseline performance status, line of therapy and likelihood of visceral-dominant disease as compared to the overall study population.ECOG = Eastern Cooperative Oncology Group25
26SUMMARYORR, TTP, and OS were not significantly different comparing GD and CD.More Grade 3-4 fatigue, hepatotoxicity, neutropenia, and thrombocytopenia with GD; more Grade 3-4 PPE, gastrointestinal toxicity, and mucositis with CD (despite the lower capecitabine dose).More patients in the CD arm discontinued therapy due to toxicity (28.4% versus 18.0%, p = .009).In an exploratory analysis, the TTP sum from induction through crossover was 5.1 months greater for the GDC sequence compared to CDG, but did not reach statistical significance (p = .093).In summary (read) -
27CONCLUSIONSGD and CD had similar efficacy with toxicity profiles consistent with prior clinical experience.In an exploratory, post-hoc analysis of patients who crossed over to the pre-specified second-line therapy, C crossover from GD trended toward greater clinical benefit compared to G crossover from CD.This suggests that the GDC sequence may be preferable.This trial, like others that have attempted to prescribe 2 consecutive lines of therapy, had a modest rate of crossover.GD and CD had similar efficacy with toxicity profiles consistent with prior clinical experience. This trial, like others that have attempted to prescribe 2 consecutive lines of therapy, had a modest rate of crossover. This illustrates the practical issues inherent in such a trial design, an observation that has implications for the design of future clinical trials.In an exploratory, post-hoc analysis of patients who crossed over to the prespecified second-line therapy, C crossover from GD trended toward greater clinical benefit compared to G crossover from CD, suggesting that this sequence may be preferable.
28ACKNOWLEDGEMENTSWe are indebted to the patients who participated in this trial.We thank the supporting institutions and all the health care professionals who provided care and collected data.We thank Melissa Humbert and Jane Bromund for their diligence in trial management and data acquisition.We thank the patients who participated in this trial, the supporting institutions and health care professionals in The U.S., Taiwan, Mexico, Argentina, Australia, South Korea and Brazil who provided care and collected data, and Melissa Humbert and Jane Bromund for their diligence in trial management and data acquisition.THANK YOU VERY MUCH.
33DURATION FROM PD TO CROSSOVER ParameterGDN=74CDN=77Median duration from PD to crossover, days14105 patients who did not reach PD but received crossover therapywere excluded from this analysis; 2 patients in GDC and 3 patientsin CDG.2 patients did not receive crossover therapy and were excluded fromthis analysis.33
34ECOG PERFORMANCE STATUS Patients at Baseline and at Crossover ParameterGDCN=77CDGN=81ECOG performance status at baseline, n (%)161 (79.2)14 (18.2)65 (80.2)15 (18.5)ECOG performance status at crossover, n (%)63 (84.0)12 (16.0)63 (78.8)17 (21.3)ECOG = Eastern Cooperative Oncology Group34