Overall, 49% of all patients had a low-risk EndoPredict score. After multivariate adjustment, these patients were significantly more likely to be free of distant metastases in the first 5 years of follow-up (hazard ratio, 1.20; P less than .001) and thereafter (HR, 1.28; P = .001). Fully 96.3% of patients with a low-risk score were metastasis free between 5 and 10 years.
"The EndoPredict is clearly an independent prognostic parameter both in the years 0 to 5 and after 5 years," Dr. Dubsky maintained.
When the C-index for discrimination was calculated, EndoPredict significantly improved on prediction of distant metastases after 5 years of follow-up when combined with the Adjuvant Online score (P less than .001) and other factors.
However, the best prediction was achieved with the predefined EndoPredict clinical score, which combined the EndoPredict score, nodal status, and tumor size, and achieved a C-index of nearly 0.8.
Fully 64% of patients still at risk for distant metastases after 5 years of follow-up had a low EndoPredict clinical score. These patients were dramatically more likely to remain free of distant metastases thereafter (HR, 5.11; P less than .001). In absolute terms, 98.2% were free of distant metastasis at 10 years.
"Risks and side effects of extended therapy should be weighed against this outcome," Dr. Dubsky recommended.
In a final analysis teasing apart the role of various genes included in the EndoPredict score, the proliferation genes added independent negative prognostic information for early recurrence, whereas the genes associated with ER signaling added independent positive prognostic information for late recurrence.
Dr. Dubsky disclosed that he is an adviser to Sividon, Agendia, Genomic Health, and AstraZeneca; receives grant support from Sividon and Agendia; and is on the speakers bureau for Sividon.