- © 2008 by American Society of Clinical Oncology
Gemcitabine Plus Paclitaxel Versus Paclitaxel Monotherapy in Patients With Metastatic Breast Cancer and Prior Anthracycline Treatment
- Kathy S. Albain,
- Shona M. Nag,
- German Calderillo-Ruiz,
- Johann P. Jordaan,
- Antonio C. Llombart,
- Anna Pluzanska,
- Janusz Rolski,
- Allen S. Melemed,
- Jose M. Reyes-Vidal,
- Jagdev S. Sekhon,
- Lorinda Simms and
- Joyce O'Shaughnessy
- From the Loyola University Chicago Stritch School of Medicine, Cardinal Bernardin Cancer Center, Maywood, IL; Lilly Oncology and Department of Biostatistics, Eli Lilly & Company, Indianapolis, IN; Baylor-Sammons Cancer Center, TX Oncology, PA, US Oncology, Dallas, TX; Jehangir Hospital, Pune; Dayanand Medical College and Hospital, Ludhiana, Punjab, India; Instituto Nacional de Cancerologia, Mexico City, Mexico; Addington Hospital, Durban, South Africa; Instituto Valenciano de Oncologia, Valencia, Spain; Klinika Chemioterapii, Lodz, Poland; Centrum Onkologii, Krakow, Poland; and the Clinica Las Condes, Santiago, Chile
- Corresponding author: Kathy S. Albain, MD, Loyola University Chicago Stritch School of Medicine, Cardinal Bernardin Cancer Center, 2160 S First Ave, Maywood, IL 60153; e-mail: kalbain{at}lumc.edu
Abstract
Purpose The objective of this phase III global study was to compare the efficacy of gemcitabine plus paclitaxel (GT) versus paclitaxel in patients with advanced breast cancer. It was designed as a pivotal study for the approval of G for a breast cancer treatment indication.
Patients and Methods Patients who relapsed after adjuvant anthracyclines were randomly assigned to gemcitabine,1,250 mg/m2 days 1 and 8 plus paclitaxel, 175 mg/m2 on day 1; or, to paclitaxel at same dose on day 1 (both arms administered every 21 days, unblinded). The primary end point was overall survival (OS) and secondary end points were time to progression (TTP), response rate (RR), progression-free survival, response duration, and toxicity. This final OS analysis was planned at 380 deaths.
Results A total of 266 patients were randomly assigned to GT and 263 to paclitaxel. Median survival on GT was 18.6 months versus 15.8 months on paclitaxel (log-rank P = . 0489), with an adjusted Cox hazard ratio of 0.78 (95% CI, 0.64 to 0.96; P = .0187). The TTP was longer (6.14 v 3.98 months; log-rank P = .0002) and the RR was better (41.4% v 26.2%; P = .0002) on GT. There was more grade 3 to 4 neutropenia on GT and grade 2 to 4 fatigue and neuropathy were slightly more prevalent on GT.
Conclusion This phase III study documents a role for gemcitabine in advanced breast cancer after anthracycline-based adjuvant therapy. The results establish GT as a reasonable choice for women who require cytoreduction with manageable toxicities and validate ongoing testing of GT in the adjuvant setting.
INTRODUCTION
Breast cancer, the most frequent malignancy in women, is a global problem and leading cause of cancer mortality.1,2 Median survival from diagnosis of metastatic disease is 2 to 3 years, with 5% to 10% survival beyond 5 years.3,4 The goals of treatment are to prolong survival while controlling symptoms and minimizing toxicity. New strategies to achieve these goals worldwide are imperative.
The widespread inclusion of anthracyclines in the adjuvant setting and concerns regarding cardiotoxicity limit their use as first-line therapy in metastatic disease. During the past decade, other cytotoxic drugs with activity in advanced breast cancer were identified, including the taxanes (paclitaxel and docetaxel), gemcitabine, vinorelbine, and capecitabine. For example, paclitaxel yielded single-agent response rates of 21% to 49% after prior anthracycline therapy.5-12 Overall response rates of 14% to 42% were reported for gemcitabine, usually after both anthracyclines and taxanes.13-19
Whether to employ these active agents for metastatic breast cancer in sequence or as various combinations is debated for patients with prior anthracycline treatment. For example, the doublet of docetaxel and capecitabine was superior to monotherapy, but produced significant toxicity.20 Other doublets that improve outcomes with minimal toxicity are needed. Gemcitabine is an attractive drug to combine with a taxane, given distinct mechanisms of action, additive or synergistic activity in vitro, nonoverlapping toxicity, and lack of cardiotoxicity. Phase II studies of the gemcitabine plus paclitaxel doublet demonstrated consistently high response rates (40% to 71%) and manageable toxicity as first-line or salvage therapy in patients with advanced breast cancer.21-27 The optimal dose, schedule, and sequence of administration were determined in these studies and utilized in this trial.26,27
Based on these data, a randomized, phase III multicenter study was conducted to compare the overall survival (with other efficacy end points) and safety of gemcitabine plus paclitaxel (GT) versus paclitaxel in patients with advanced breast cancer. This reports the planned, final analysis for overall survival (OS), updating prior presentations of time to documented progression (TTP) and the interim OS analyses.28,29 Changes in pain, analgesic use, and quality of life (QOL) were also examined and are summarized in a separate publication.
PATIENTS AND METHODS
Eligibility Criteria
Eligible women had locally recurrent or metastatic breast carcinoma with the following criteria: minimum of one bidimensionally measurable lesion outside a prior radiation portal (≥ 1 × 1 cm2 by radiographs or ≥ 2 × 2 cm2 by physical examination); good Karnofsky performance status (KPS); adequate bone marrow reserve, liver and renal function; normal calcium levels; and an estimated life expectancy ≥ 12 weeks. The cancer was recurrent after one neoadjuvant/adjuvant anthracycline-based regimen (or nonanthracycline based if the use of anthracyclines was clinically contraindicated). Prior gemcitabine or taxane was not allowed, nor was any prior chemotherapy for metastatic disease. Patients with brain or bone-only metastases were excluded. Written informed consent was obtained from all patients.
Study Design
This was a phase III, multicenter, open-label study of GT versus paclitaxel. After registration, patients underwent standardized, prestudy evaluations. Randomization occurred after central confirmation of eligibility. A minimizing algorithm with probability factor 0.7530 was applied to balance the treatment arms for the strata KPS (70 to 80 v 90 to 100), prior anthracycline therapy in the adjuvant setting (yes v contraindicated), prior hormone therapy (yes v no), visceral metastases (yes v no), disease progression after prior adjuvant therapy (≤ 6 v > 6 months), and center.
In the GT arm, gemcitabine (Gemzar; Eli Lilly and Company, Indianapolis, IN) at 1,250 mg/m2 was given intravenously on days 1 and 8 over 30 to 60 minutes. In both arms, paclitaxel (Taxol; Bristol-Meyers Squibb, New York, NY) at 175 mg/m2 was administered intravenously on day 1 over 3 hours (before gemcitabine on the GT arm), with standard paclitaxel premedication. Treatment was repeated every 21 days. Prophylactic antiemetics (including corticosteroids) and growth factors for hematologic toxicity were permitted. Dose reduction criteria were mandated and toxicity reporting used the National Cancer Institute-Common Toxicity Criteria version 2.0. Treatment was continued until disease progression, intolerable toxicity, or patient withdrawal. Off-study chemotherapy was at the discretion of the investigator.
Study Evaluations
The prestudy evaluation (≤ 4 weeks before registration) included an ECG, chest x-ray, computed tomography, or magnetic resonance imaging scan of chest and abdomen, and bone scan. Blood chemistries and complete blood counts were measured within 2 weeks before registration and at every treatment cycle. Medical history, physical examination, KPS, and analgesic level documentation, and measurements of palpable or visual tumor lesions were carried out no more than 1 week before random assignment and before each cycle. Radiographic measurements were performed every 8 weeks ± 1 week using the baseline method(s). Toxicity was assessed after each cycle.
OS was measured from randomization to the date of death from any cause, with censoring at last visit date. The TTP was measured from randomization to the date of documented disease progression, with censoring at last visit date (if alive) or date of death. Progression-free survival (PFS) was from randomization to the documented date of progression or date of death from any cause, whichever was earlier, with censoring at the last visit date. Patients still alive but without progression were censored at their last visit date. Response was assessed according to WHO criteria. A complete response (CR) or partial response (PR) was confirmed at least 4 weeks after first observation of the response. Duration of response (investigator-assessed CR or PR) was measured from randomization to the date of progression, with censoring at last visit date (if alive) or date of death.
A panel of imaging experts evaluated all radiology tests to independently assess end points dependent on response and progression. The TTP, PFS, and duration of response end points were assessed in two ways: by utilizing only the investigator assessments of progression, or by using the earliest of either investigator-assessed or independent panel-assessed progression. Because these two methods yielded essentially the same results, only the investigator-assessed results are presented in this report.
Statistical Methods
All randomly assigned patients were analyzed for efficacy on the basis of treatment arm allocation using the intent-to-treat principle and all who received at least one dose of chemotherapy were evaluated for safety. The primary analyses of distribution of time-to-event end points were estimated using the Kaplan-Meier method (unadjusted) and compared by log-rank test. Exploratory analyses (prespecified before data lock) were conducted to adjust for multiple significant covariates via stepwise Cox regression models on OS and TTP. Thirteen baseline prognostic variables were tested univariately and those significant at the .05 level were included in a Cox model. Response rates were compared using an unadjusted normal approximation for the difference of two binomial proportions (two-sided z test). Two-sided CIs for all estimated variables were constructed (95% level.)
The protocol was designed to detect a hazard ratio (HR) of 0.75 (ie, a 33% improvement) of GT over paclitaxel for both coprimary end points OS and TTP, based on literature reports for paclitaxel,5-12 with a resulting sample size of approximately 526. The initial significance level for OS was 0.03, requiring approximately 440 deaths to achieve 80% power; for TTP, it was 0.028 (75% power). However, based on discussions with the US Food and Drug Administration at the time of submission for approval on the TTP end point, but prior to any survival analysis, OS was made the sole primary end point, to be tested at a two-sided .05 significance level (HR, 0.75 with a censor rate < 30%). Consequently, to preserve the protocol-specified 80% power to detect a 33% improvement in OS, the final OS analysis was to be performed at a minimum of 377 deaths (this report).
A planned interim analysis that evaluated only toxicity, response, and TTP was reported previously.28 An interim OS analysis, conducted at the request of the US Food and Drug Administration during final consideration for drug approval on the TTP end point, was done at the nominal significance level of .0001 and was previously reported.29 This left a significance level of .049983 for the final OS analysis (this report), based on the overall α of .05 for OS.
RESULTS
A total of 599 women were registered from 19 countries with 98 participating centers for the prestudy evaluation (Fig 1). Of these, 529 were eligible and randomly assigned from August 1999, to April 2002, 266 to GT and 263 to paclitaxel (the intent-to-treat population for efficacy analyses). Of the 70 patients not randomly assigned, 52 did not meet eligibility criteria, five withdrew, two physicians declined, one died before random assignment, and 10 were withdrawn from one site due to questions of data integrity. A total of 261 patients on GT and 260 patients on paclitaxel received drug treatment and were evaluated for safety.
CONSORT flow diagram. Disposition of study participants. (*) Patients included in the final analyses of survival and secondary efficacy variables per intention-to-treat principle. (†) Patients treated and therefore included in the safety analysis.(‡) Included: protocol violation, entry criteria not met, death unrelated to disease, patient moved, unable to contact patient.
Patient Characteristics and Treatment Delivered
Baseline characteristics and demographics were balanced between arms (Table 1). Median age was 53 years (range, 26 to 83 years). Metastatic disease was present in 97% of patients on each arm. Visceral disease occurred in 73% of patients, and 42% had three or more metastatic sites. Approximately one half of the patients received prior hormone therapy and 96% received prior anthracyclines.
Patient Demographics, Disease Characteristics, and Prior Therapy
Patients on the GT arm received more treatment (mean, 6.4 cycles) than those on the paclitaxel arm (mean, 5.7 cycles). Of the planned doses of gemcitabine, 8.4% were reduced and 6.9% were omitted. For paclitaxel on the GT arm, 5.2% of planned doses were reduced and 0.8% omitted, versus 2.0% and 0.1%, respectively, on the monotherapy arm. Relative dose intensities on the GT arm were 83.8% for gemcitabine and 92.8% for paclitaxel and on the monotherapy arm was 96.2% for paclitaxel. Treatment-related discontinuation was low (GT, 6.1%, 16 of 261 patients; paclitaxel, 3.5%, nine of 260 patients). Most drug-related withdrawals on both arms were due to neuropathy, whereas most dose reductions or omissions were due to neutropenia.
Efficacy
At the time of the final OS analysis (377 deaths), 84 (31.6%) of 266 patients on the GT arm and 68 (25.9%) of 263 patients on the paclitaxel arm were censored. The median OS was 18.6 months on GT versus 15.8 months on paclitaxel (log-rank P = .0489; Fig 2A). The HR was 0.82 (95% CI, 0.67 to 1.00; P = .0493) in favor of GT. At each time point of 12, 18, 24, and 30 months, there was a significantly higher survival probability for GT than for paclitaxel (P ≤ .0457). The survival curves separated at 6 months and maintained a clear separation over time until approximately 37 months (Fig 2A).
(A) Kaplan-Meier curve from analysis of the primary end point overall survival (n = 377 deaths). Median survival for the gemcitabine plus paclitaxel arm was18.6 months and for the paclitaxel arm was 15.8 months (log-rank P = .0489). (B) Kaplan-Meier curve for secondary end point of time to progression, favoring the combination of gemcitabine plus paclitaxel (median, 6.14 months) over paclitaxel alone (median, 3.98 months; log-rank P = .0002). GT, gemcitabine and paclitaxel; T, paclitaxel.
Several baseline covariates showed significant (P < .05) individual prognostic value on OS (univariate analyses, data not shown). These included KPS, number of tumor sites, prior surgery, time from diagnosis to randomization, disease progression during adjuvant chemotherapy, and hormone receptor status. When adjusted for the significant factors in the final Cox multivariate model for OS (Table 2), the benefit of GT versus paclitaxel became more apparent, with an adjusted HR of 0.78 (95% CI, 0.64 to 0.96; P = .0187). Other independent predictors of outcome were time from diagnosis to randomization, number of tumor sites, ER status, and KPS.
Cox Multivariate Model for Overall Survival
The median TTP was longer on GT than paclitaxel (6.14 v 3.98 months; log-rank P = .0002), with an unadjusted HR of 0.70 (95% CI, 0.59 to 0.85; Table 3 and Figure 2B), with 39 patients (14.7%) censored on GT and 26 (9.9%) censored on paclitaxel. There was also a significant PFS advantage for GT (5.9 v 3.9 months; unadjusted HR 0.73; 95% CI, 0.61 to 0.87; P = .0005), with 20 patients (7.5%) censored on GT and 16 (6.1%) on paclitaxel. Investigator response rates (Table 3) were GT of 41.4% and paclitaxel, of 26.2% (χ2 P = .0002; by independent radiology review, 43.1% and 26.9%; P = .0007). In patients with nonvisceral disease only at baseline, response rates were 56.9% (GT) and 38.0% (paclitaxel; P = .0235), and for those with visceral disease (with or without nonvisceral sites) were 35.6% (GT) and 21.9% (paclitaxel; P = .0030). The onset of response occurred at or before cycle 3 in more than 75% of patients on both arms and once a response was documented, the duration of response did not differ by arm (Table 3).
Analyses of Secondary Efficacy End Points
Details regarding poststudy chemotherapy are summarized in Table 4. Approximately 55% of patients received additional chemotherapy after completion of study drugs, ranging from one or two additional lines of therapy (36.5% GT; 39.5% paclitaxel) to three or more lines (16.5% GT arm; 17.1% paclitaxel arm). The types of additional chemotherapy were very similar in the two arms (Table 4), with the exception that more patients on the paclitaxel arm received off-study gemcitabine than on the GT arm (15.6% v 4.1%).
Summary of Post-Study Chemotherapy
Toxicity
Table 5 depicts the worst drug-related National Cancer Institute-Common Toxicity Criteria grade 2, 3, or 4 toxicities observed per patient by arm. Hematologic toxicity was more commonly observed on GT, especially neutropenia (GT, 47.9% grade 3 or 4; paclitaxel, 11.5%). The majority of the treatment-related dose delays on the GT arm (72 dose delays) were due to hematologic toxicities (data not shown), whereas those on the paclitaxel arm (41 delays) were due to a variety of toxicities (below). Febrile neutropenia occurred in 5.0% of patients on GT and 1.2% on paclitaxel. Twenty-eight patients on GT and 10 patients on paclitaxel received ≥ 1 transfusion(s), mostly of packed RBCs.
Drug-Related Common Toxicity Criteria Grade 2, 3, or 4 Toxicities
Overall, the incidence of grade 3 or 4 nonhematologic drug-related toxicities on the two arms was low (Table 5). Grade 3 or 4 fatigue was more common on the GT arm, generally lasted for one cycle (both arms) and was not associated with anemia. Grade 2, 3, or 4 sensory neuropathy occurred at similar frequency (24.1% GT, 21.6% paclitaxel); 8.8% on GT had motor neuropathy, versus 3.1% on paclitaxel. None of the grade 3 or 4 neuropathies resulted in hospitalization. Incidences of grade 3 transaminase elevations were slightly higher on the GT arm, whereas myalgia did not appear to differ. Five GT treated patients (1.9%) had grade 3 or 4 dyspnea, but all had active disease in the lungs or pleura. Two patients (one per arm) died from possible drug-related toxicities (non-neutropenic septicemia and severe asthenia).
DISCUSSION
The final results of this pivotal phase III study of GT versus paclitaxel prove that gemcitabine added to paclitaxel is effective therapy for women with advanced breast cancer who previously received anthracyclines. The US Food and Drug Administration approved a breast cancer indication for gemcitabine based on the highly significant prolongation of TTP28 and interim promising OS,29 with similar decisions from regulatory bodies in other countries. This final report confirms the OS benefit to GT, a 22% improvement in OS, along with a 43% prolongation in TTP, and both end points are statistically significant. The favorable impact of GT is quite durable out to 3 years of follow-up. The significant increase in objective responses from GT, plus the benefit observed in patients with poorer-prognosis visceral disease31 also support its first-line use when reduction in tumor burden is needed. The only difference between arms was the gemcitabine, with the dose and schedule of paclitaxel kept constant in both arms. Thus, gemcitabine is an important addition to the first-line armamentarium in the global treatment of metastatic breast cancer. In many parts of the world, very advanced, symptomatic presentations are quite common,1 so the GT doublet can be applied with greater success than the limited options2 previously available for many women.
The GT doublet is also a well-tolerated choice for women with advanced disease after adjuvant anthracycline therapy or for whom the cardiotoxic effects of anthracyclines preclude its use. There is increased neutropenia from GT compared with paclitaxel, as well as a slight increase in severe neuropathy and fatigue. Despite this increase in toxicity, the analysis of the global QOL end point from the QOL companion study favored the doublet.32
This trial was not designed to show that GT is superior to sequential monotherapy with paclitaxel followed by gemcitabine in the same doses and schedule. Another phase III trial showed quite similar degrees of superiority for OS, TTP, and RR of an antimetabolite/taxane doublet versus taxane alone in an anthracycline-pretreated population, but was not a crossover design either.20 The Eastern Cooperative Oncology Group conducted the only prospectively designed, phase III crossover study in metastatic disease (no prior anthracyclines) and found that while an anthracycline plus taxane doublet improved RR and TTP over each single agent with crossover to the other agent, there was no OS advantage to the doublet.31 It is possible that GT might yield equivalent OS to paclitaxel crossed over to gemcitabine at the same doses and schedule.
There are other caveats regarding this trial design. It does not allow assessment of GT after adjuvant taxanes or the activity of gemcitabine alone in the first-line setting. This study showed superiority of GT versus paclitaxel and validated other reports of single-agent paclitaxel in similar doses given every 3 weeks over 3 hours.5,6,8,10,11 However, unlike when this study was designed, currently the weight of evidence suggests that either weekly paclitaxel, dose-dense paclitaxel, or other taxane yields results superior to the control arm employed in this study. Thus, no information can be provided regarding the superiority of the GT doublet to other paclitaxel schedules or other taxanes in this setting. There may be more effective gemcitabine plus taxane doublets than the GT dose and schedule used in this trial33 and other studies are in progress to directly compare various antimetabolite/taxane doublet doses, schedules, and sequences. Finally, the potential interactions of estrogen receptor status, human epidermal growth factor receptor 2 status, and benefit from GT (or not), along with defining the optimal multigene predictors of treatment benefit from GT deserve future investigation.
Therefore, the final results of this pivotal, phase III study demonstrated the effectiveness of gemcitabine for patients with advanced breast cancer and prior anthracycline-based adjuvant therapy, as well as the clinical utility of the GT combination. The significant TTP and OS advantage to GT along with a manageable toxicity profile make it an attractive treatment option when combination chemotherapy is chosen as the optimal approach. The GT regimen is currently being evaluated in several phase III adjuvant trials because the addition of gemcitabine to paclitaxel appears to favorably alter the natural history of advanced breast cancer.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a ‘U’ are those for which no compensation was received; those relationships marked with a ‘C’ were compensated. 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 or Leadership Position: Allen S. Melemed, Eli Lilly & Company (C); Lorinda Simms, Eli Lilly & Company (C) Consultant or Advisory Role: Kathy S. Albain, Eli Lilly & Company (U); Antonio C. Llombart, Eli Lilly & Company (C); Joyce O'Shaughnessy, Eli Lilly & Company (C) Stock Ownership: Allen S. Melemed, Eli Lilly & Company; Lorinda Simms, Eli Lilly & Company Honoraria: Kathy S. Albain, Eli Lilly & Company; Joyce O'Shaughnessy, Eli Lilly & Company Research Funding: Jose M. Reyes-Vidal, Eli Lilly & Company; Jagdev S. Sekhon, Eli Lilly & Company Expert Testimony: None Other Remuneration: None
AUTHOR CONTRIBUTIONS
Conception and design: Kathy S. Albain, Allen S. Melemed, Joyce O'Shaughnessy
Financial support: Allen S. Melemed
Provision of study materials or patients: Kathy S. Albain, Shona M. Nag, German Calderillo-Ruiz, Johann P. Jordaan, Antonio C. Llombart, Anna Pluzanska, Janusz Rolski, Allen S. Melemed, Jose M. Reyes-Vidal, Jagdev S. Sekhon, Joyce O'Shaughnessy
Collection and assembly of data: Shona M. Nag, Antonio C. Llombart, Allen S. Melemed, Jagdev S. Sekhon, Lorinda Simms
Data analysis and interpretation: Kathy S. Albain, Antonio C. Llombart, Allen S. Melemed, Lorinda Simms, Joyce O'Shaughnessy
Manuscript writing: Kathy S. Albain, Allen S. Melemed, Lorinda Simms, Joyce O'Shaughnessy
Final approval of manuscript: Kathy S. Albain, Shona M. Nag, German Calderillo-Ruiz, Johann P. Jordaan, Antonio C. Llombart, Anna Pluzanska, Janusz Rolski, Allen S. Melemed, Jose M. Reyes-Vidal, Jagdev S. Sekhon, Lorinda Simms, Joyce O'Shaughnessy
Acknowledgments
We thank all the patients, investigators, and institutions involved in this study. We also acknowledge the editorial assistance of Chris Carucio, Tina Rutschman, Mary Dugan Wood, Mary Alice Miller, and Pete Fairfield; Yun Ding for data analysis; and Andrea Stow for data management.
Footnotes
-
Supported by Eli Lilly and Company (drug supply, data management and analysis, site audits, funding for independent radiology review, institutional review board fees and clinical research associate support).
Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 31-June 3, 2003, and the 40th Annual Meeting of the American Society of Clinical Oncology, New Orleans, LA, June 5-8, 2004.
Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received March 30, 2007.
- Accepted April 8, 2008.
