Healthcare Professionals   /   OncoLink Scientific Meetings Coverage   /   Scientific Meetings   /   OncoLink at the San Antonio Breast Cancer Symposium 2006   /   Highlights from SABCS 2006

Benefit from exemestane (EXE) as extended adjuvant therapy after 5 years of tamoxifen (TAM): intent-to-treat analysis of NSABP B-33

Reviewer: John P. Plastaras, MD, PhD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: December 16, 2006

Presenter: E. Mamounas
Affiliation: NSABP Operations & Biostatistical Centers, Pittsburgh, PA

Background

• For hormone receptor positive breast cancers, 5 years of adjuvant tamoxifen had been the standard. Aromatase inhibitors (AIs) were tested in three different settings:
  1. AI x 5 yr vs. Tamoxifen x 5 yr
  2. Sequential Tam/AI (2-3 yr each) x 5 yr vs. Tam x 5 yr
  3. Extended AI vs. placebo after Tam for 5 yr.
• The NSABP intended to test the 3 rd approach with a trial that was started in May 2001. It was designed to test the benefit of adding exemestane (EXE) therapy sequentially after 5 year of tamoxifen.
• The ATAC trial reported results in December 2001 showing that the AI was superior to Tam in terms of disease free survival. In 10/03, the NCIC MA.17 trial reported that extended adjuvant letrozole after 5 years of tamoxifen was superior to placebo. At that time, accrual in this NSABP B-33 trial was halted, unblinded and patients were allowed to crossover to EXE.
• The intention to treat analysis is presented here.

Methods

• 1598 post-menopausal women with hormone receptor positive breast cancer, clinical stage T1-3 N0-1 who were disease free after 5 years of tamoxifen
• Randomized, double-blind, multicenter, phase III, two-arms:
  • Tam x 5 yr, followed by placebo x 5 yr
  • Tam x 5 yr, followed by EXE (25 mg po daily) x 5 yr
• The original design was for 2 years of EXE, but the protocol was amended in 2002 to extend the duration to 5 years
• The trial was originally powered to detect a 21% reduction in the hazard rate with a power of 80% with an accrual goal of 3000 patients. After the NCIC MA.17 trial was reported, only 1598 patients had been randomized.
• Endpoints:
  • Primary: disease free survival
• Secondary: overall survival, relapse free survival, bone mineral density, blood lipid profile, and quality of life

Results

• Patients: Arms were well-balanced. 80% were ER+/PR+, 12% were ER+/PR-, and 3% were ER-/PR+
• After accrual was closed and unblended, women on both arms were offered EXE, free of charge
  • 560 of 783 (72%) on EXE continued on EXE
  • 344 of 779 (44%) on placebo started taking EXE
  • Most switched between 6-12 months
  • It is unknown whether the remaining women started taking an alternate AI, such as anastrazole.
• Disease free survival was better in EXE arm, but did not reach significance:
  • EXE 91% vs. Placebo 89%, p=0.07 (RR 0.68)
• Distant disease free survival favored but EXE, but did not reach significance:
  • EXE 94% vs. Placebo 93% (RR 0.69, p=0.13)
• Relapse free survival favored EXE:
  • EXE 96% vs. Placebo 94% (RR 0.50, p=0.03)
• Overall survival favored Placebo, but was not significant
  • 16 deaths on EXE vs. 13 deaths on Placebo (RR 1.2, p=0.63)
• Toxicity favored placebo:
  • There was 1% grade 4 toxicity in each arm
  • Grade 3: EXE 9% vs. 6%, p=0.03
    • The most common were arthralgia, fatigue, and bone pain
  • Fractures were not significantly different: EXE 28 vs. Placebo 20, p=0.33
• Quality of Life: 454 participants. There were no differences between the arms.

Author's Conclusions

• Despite early closure and crossover, there was a borderline significant benefit in disease free survival and a significant benefit in relapse free survival with the addition of 5 years of EXE after 5 years of tamoxifen
• The toxicity of EXE was acceptable.

Clinical/Scientific Implications

• The optimal choice of AI and sequence with tamoxifen is not known, but AI's are currently standard treatment in hormone positive, post menopausal breast cancer
• Unfortunately, after cross over, use of other AIs was not recorded, which would allow an analysis based on treatment received in addition to this intention to treat analysis.
• Thus conclusions cannot be definitively drawn from this study