ASCO—It may soon be possible to determine which breast cancer patients will respond to adjuvant endocrine therapy, based on the level of cell-cycle response in the surgical specimen after neoadjuvant endocrine therapy, Matthew Ellis, MD, PhD, reported at the 2007 ASCO annual meeting (abstract 570). Dr. Ellis is director of the Breast Cancer Program and associate professor of medicine at Washington University, St. Louis.
P024 Study
The new approach was evaluated in a retrospective analysis of the P024 study, a double-blind randomized trial of neoadjuvant endocrine therapy that compared 16 weeks of letrozole(Drug information on letrozole) (Femara) 2.5 mg daily with tamoxifen(Drug information on tamoxifen) 20 mg daily.
In P024, letrozole proved to be more effective than tamoxifen in terms of tumor response and rates of breast conservation. Biomarker studies also indicated that letrozole was a more effective antiproliferative agent, producing greater declines in tumor cell Ki67 values.
"Ki67 measures proliferation. If endocrine therapy is maximally effective, the Ki67 should disappear. If Ki67 is present, by definition the tumor is relatively endocrine-therapy resistant," Dr. Ellis explained.
In the P024 trial, the rate of pathological complete responses (pCR) to endocrine therapy was very low. "We have therefore proposed that a cell-cycle complete response (CCCR)—in which Ki67 staining is ≤ 1% in the surgical specimen—may be an alternative to pCR because CCCR indicates that the endocrine therapy was maximally effective in causing cell-cycle arrest," he said.
Initial studies by Dr. Ellis' team had already shown that CCCR is more likely with letrozole than with tamoxifen, and that there is a strong association between the absence of a CCCR and the presence of HER2 gene amplification.
Survival rates in P024 patients were analyzed after a median of 5 years' follow-up on adjuvant tamoxifen. The letrozole-treated patients whose tumors exhibited a CCCR had superior relapse-free (P = .0077) and breast-cancer-specific survival rates (P = .0006), Dr. Ellis reported. These relationships were shown in both univariate and multivariate analyses of predictive factors.
