Gene Assays Predict Recurrence in Node-Negative Breast Ca

April 1, 2006

Two studies presented at the 28th Annual San Antonio Breast Cancer Symposium suggest researchers are getting closer to developing patient-specific prognostic assays for breast cancer recurrence.

SAN ANTONIO—Two studies presented at the 28th Annual San Antonio Breast Cancer Symposium suggest researchers are getting closer to developing patient-specific prognostic assays for breast cancer recurrence. One study assessed the reliability of a 21-gene recurrence score assay, Oncotype DX (Genomic Health, Inc., Redwood City, California), and the other assessed a 76-gene prognostic signature.

Lead investigator Terry P. Mamounas, MD, MPH, chairman of the breast committee at the National Surgical Adjuvant Breast and Bowel Project (NSABP) and medical director of the Aultman Cancer Center, Canton, Ohio, reported results using the 21-gene recurrence score assay to predict locoregional recurrence (abstract 29). Study participants were women with early node-negative, ER-positive breast cancer who had been treated with tamoxifen.

The NSABP researchers had previously validated the recurrence score assay, which is based on 16 cancer genes and 5 reference genes, in predicting distant recurrence among tamoxifen-treated breast cancer patients. In the current study of locoregional recurrence, they evaluated the assay in three groups of patients: tamoxifen-treated patients from the NSABP B-14 and B-20 trials (n = 895), placebo patients from the NSABP B-14 trial (n = 355), and patients from NSABP B-20 treated with both tamoxifen and chemotherapy (n = 424).

The 10-year Kaplan-Meier estimate of the locoregional failure rate was calculated with second primary cancers, distant metastasis, and deaths before locoregional failure censored. The 10-year locoregional recurrence rate was 4.3% in tamoxifen-treated patients with a low recurrence score, 7.2% in those with an intermediate score, and 15.8% in women with a high score. In the placebo group, the recurrence rate was 10.8%, 18.4%, and 20%, respectively. The rate of locoregional recurrence was generally low in women who had received tamoxifen and chemotherapy: 1.6%, 2.7%, and 7.8%, respectively.

In a multivariate analysis, recurrence score was shown to be a significant independent predictor of locoregional failure among tamoxifen-treated patients, along with age and type of surgical procedure.

''In an exploratory subset analysis, we estimated the 10-year rate of locoregional failure according to recurrence score, patient age, and type of surgical procedure," Dr. Mamounas said. "This analysis should be interpreted with caution, however, given the small number of patients and the small number of events in most subgroups and the resulting wide confidence intervals." Among the mastectomy patients, a significant association was found between recurrence score and locoregional failure rate. In women younger than age 50, the rate was 1.5% for a low score, 7.6% for an intermediate score, and 23.8% or higher for a high score. Among women aged 50 or older, the recurrence rates were 2.6%, 3.8%, and 12.8%, respectively. In the group treated with lumpectomy and radiation, younger women had locoregional failure rates of 12.5%, 27.7%, and 26.5%, respectively; for women 50 or older, the rates were 3.6%, 3.7%, and 4.8%, respectively.

In a multicenter validation trial (abstract 28), John A. Foekens, PhD, of Erasmus Medical Center, Rotterdam, reported findings of high sensitivity with his group's 76-gene prognostic signature assay that has been developed in collaboration with Veridex LLC, a Johnson & Johnson Company. The group's aim has been to develop a prognostic profile that could be used for all node-negative breast cancer patients, irrespective of age, menopausal status, tumor size and grade, and hormone-receptor status.

The 76-gene prognostic signature assay was initially assessed in a single-center study (Lancet 365:671-679, 2005) involving 286 node-negative women, none of whom had received any systemic therapy. In this study, it was validated in a test set of 171 node-negative patients, using a clinical endpoint of distant metastasis-free survival and based on a median follow-up of 101 months. The genetic signature showed a 93% sensitivity and 48% specificity in identifying poor-prognosis patients. Among patients with an unfavorable genetic signature, the hazard ratio (HR) for distant metastasis was 5.7; for overall survival, 8.6.

In the group's more recent four-center validation study of samples from 180 systemic therapy-naive patients (164 ER-positive and 16 ER-negative), with a median follow-up of 100 months, the signature had a sensitivity of 90% and a specificity of 50%. "Among patients with a good profile, 96% were metastasis free at 5 years vs 74% in the poor-profile group," Dr. Foekens said. "At 10 years, 94% were metastasis free vs 65% in the poor-profile group" (HR for distant metastasis 7.4; for overall survival 5.4).

Among traditional prognostic indicators, only tumor size was significant. The HR for tumors larger than 2 cm vs smaller tumors was 2.08. Among postmenopausal patients, the HR for those with a poor prognostic signature was 9.84.

Dr. Foekens also reported preliminary findings from an ongoing study using the 76-gene assay to evaluate node-negative, ER-positive women taking adjuvant tamoxifen. In an interim analysis of the outcomes of 79 patients, the HR for a poor-prognosis signature was 5.8.