Design, conduct, and analyses of Breast International Group (BIG) 1-98: a randomized, double-blind, phase-III study comparing letrozole and tamoxifen as adjuvant endocrine therapy for postmenopausal women with receptor-positive, early breast cancer

Abstract

Background: Aromatase inhibitors provide superior disease control when compared with tamoxifen as adjuvant therapy for postmenopausal women with endocrine-responsive early breast cancer.

Purpose: To present the design, history, and analytic challenges of the Breast International Group (BIG) 1-98 trial: an international, multicenter, randomized, double-blind, phase-III study comparing the aromatase inhibitor letrozole with tamoxifen in this clinical setting.

Methods: From 1998-2003, BIG 1-98 enrolled 8028 women to receive monotherapy with either tamoxifen or letrozole for 5 years, or sequential therapy of 2 years of one agent followed by 3 years of the other. Randomization to one of four treatment groups permitted two complementary analyses to be conducted several years apart. The first, reported in 2005, provided a head-to-head comparison of letrozole versus tamoxifen. Statistical power was increased by an enriched design, which included patients who were assigned sequential treatments until the time of the treatment switch. The second, reported in late 2008, used a conditional landmark approach to test the hypothesis that switching endocrine agents at approximately 2 years from randomization for patients who are disease-free is superior to continuing with the original agent.

Results: The 2005 analysis showed the superiority of letrozole compared with tamoxifen. The patients who were assigned tamoxifen alone were unblinded and offered the opportunity to switch to letrozole. Results from other trials increased the clinical relevance about whether or not to start treatment with letrozole or tamoxifen, and analysis plans were expanded to evaluate sequential versus single-agent strategies from randomization.

Limitations: Due to the unblinding of patients assigned tamoxifen alone, analysis of updated data will require ascertainment of the influence of selective crossover from tamoxifen to letrozole.

Conclusions: BIG 1-98 is an example of an enriched design, involving complementary analyses addressing different questions several years apart, and subject to evolving analytic plans influenced by new data that emerge over time.

Figures

Figure 1

BIG 1-98 trial design. The design accommodated its genesis as a two-arm trial that evolved into a four-arm trial enabling both head-to-head comparison of letrozole vs tamoxifen as well as assessment of the role of sequential endocrine therapies.

Figure 2

Comparison of treatments for BIG 1-98 PCA cohorts with respect to disease-free survival (DFS). Hazard ratios (HR) and 95% confidence intervals are estimated from a Cox proportional hazards model. The locations of the squares indicate the estimated hazard ratio of a DFS event for letrozole vs tamoxifen, and the size of the squares is proportional to the statistical information available for each comparison. The solid vertical line indicates the overall treatment effect estimate of HR = 0.81, and the dashed vertical line shows the location of the null hypothesis of no treatment difference (HR = 1.0). FU = Follow-up

Figure 3

Comparison of treatments for BIG 1-98 monotherapy cohorts with respect to DFS. Hazard ratios and 95% confidence intervals are estimated from a Cox proportional hazards model. The locations of the squares indicate the estimated hazard ratio of a DFS event for letrozole vs tamoxifen, and the size of the squares is proportional to the statistical information available for each comparison. The solid vertical line indicates the overall treatment effect estimate of HR = 0.82, and the dashed vertical line shows the location of the null hypothesis of no treatment difference

Figure 4

DFS according to central and local classification hormone receptor status. Numbers of patients and events, and estimates of 3-year DFS and standard errors (SE) are summarized. [Reprinted with permission of the Journal of Clinical Oncology 25(25), 2007: 3846–3852].