Guiding treatment decisions in early breast cancer: A model-based comparison of the OncotypeDX and MammaPrint tests

Abstract

Introduction

Genomic tests may improve chemotherapy allocation in early-stage breast cancer beyond traditional clinical factors. We compared the clinical usefulness of two multi-genomic tests, MammaPrint and OncotypeDX, in guiding adjuvant chemotherapy decisions.

Methods

The MINDACT and TAILORx trials provided prospective validation for the MammaPrint and OncotypeDX tests, respectively. We generated two synthetic cohorts to evaluate both tests in both trial contexts. Chemotherapy was assumed to be indicated if it was expected to reduce the risk of an event by at least 5 %, defining events as 10-year distant metastases or breast cancer–related death. We compared treatment decision making informed by clinical risk information alone versus clinical information plus either the MammaPrint (dichotomous score: high/low risk) or OncotypeDX test (dichotomous and continuous scores). These strategies were evaluated using Net Benefit: a weighted difference of preventing events through treatment and the number of treatments given.

Results

Treatment decision-making informed by clinical information alone would result in a positive balance between preventing events and treatments given in both synthetic cohorts (4.8 net benefit in MINDACT, 3.0 in TAILORx per 1000 patients). Incorporating OncotypeDX into risk assessment improved treatment allocation more than MammaPrint (+2.6 vs + 1.6 in MINDACT, +1.3 vs + 1.0 in TAILORx contexts), with substantial uncertainty. Using the dichotomized OncotypeDX score limited its clinical usefulness compared to using its underlying continuous score.

Discussion

Both MammaPrint and OncotypeDX tests improve identifying candidates for chemotherapy among women with early breast cancer, with broadly equivalent clinical usefulness. The tests should be implemented into existing risk algorithms to maximize their clinical usefulness.