Enhancing decision-making about adjuvant chemotherapy in early breast cancer following EndoPredict testing

Objective: Chemotherapy side-effects can be substantial. There is increasing recognition that some oestrogen receptor positive (ER +ve), human epidermal growth factor receptor 2 negative (HER2 -ve) patients with breast cancer derive no benefit from chemotherapy and experience only iatrogenic harm. Gene expression profiling tests help refine recurrence risk and likely chemotherapy benefit. EndoPredict (R) is one such test, which classifies risks of distant recurrence as low or high in patients treated with surgery and adjuvant endocrine therapy alone. We compared treatment decisions pre-test and post-test results, patients' anxiety, decisional conflict, and oncologists' confidence about the decisions made. Methods: Fourteen oncologists in 7 UK hospitals saw 149 pts judged to have equivocal indications for chemotherapy. Provisional treatment decisions were recorded then reconsidered when EPClin results were available. Pre-test and post-test results, patients completed State/Trait Anxiety Inventories (STAI), and the decisional conflict scale. Oncologists also recorded basic clinical details, their agreement with, and confidence about treatment decisions. Results: Sixty-seven percent patients initially prescribed endocrine alone with high risk result upgraded to endocrine+chemotherapy (E+C); 83% prescribed E+C and had low risk scores, downgraded to E. None of 46 patients initially favouring E alone, who were low risk, changed decisions. Oncologists' confidence about decisions was significantly increased following the results (P = 0.002). Patients with downgraded treatment decisions had significantly lower anxiety scores (P = 0.045); those upgraded had increased scores (P = 0.001). Overall decisional conflict and uncertainty fell significantly post-test (P < 0.022). Conclusions: EndoPredict scores increased oncologists' and patients' decision-making confidence, generally improving the matching of risk with therapy decisions.