Background: Previous reports identifying discordance between multiparameter tests at the individual patient level have been largely attributed to methodological shortcomings of multiple in silico studies. Comparisons between tests, when performed using actual diagnostic assays, have been predicted to demonstrate high degrees of concordance. OPTIMA prelim compared predicted risk stratification and subtype classification of different multiparameter tests performed directly on the same population.
Methods: Three hundred thirteen women with early breast cancer were randomized to standard (chemotherapy and endocrine therapy) or test-directed (chemotherapy if Oncotype DX recurrence score >25) treatment. Risk stratification was also determined with Prosigna (PAM50), MammaPrint, MammaTyper, NexCourse Breast (IHC4-AQUA), and conventional IHC4 (IHC4). Subtype classification was provided by Blueprint, MammaTyper, and Prosigna.
Results: Oncotype DX predicted a higher proportion of tumors as low risk (82.1%, 95% confidence interval [CI] = 77.8% to 86.4%) than were predicted low/intermediate risk using Prosigna (65.5%, 95% CI = 60.1% to 70.9%), IHC4 (72.0%, 95% CI = 66.5% to 77.5%), MammaPrint (61.4%, 95% CI = 55.9% to 66.9%), or NexCourse Breast (61.6%, 95% CI = 55.8% to 67.4%). Strikingly, the five tests showed only modest agreement when dichotomizing results between high vs low/intermediate risk. Only 119 (39.4%) tumors were classified uniformly as either low/intermediate risk or high risk, and 183 (60.6%) were assigned to different risk categories by different tests, although 94 (31.1%) showed agreement between four of five tests. All three subtype tests assigned 59.5% to 62.4% of tumors to luminal A subtype, but only 121 (40.1%) were classified as luminal A by all three tests and only 58 (19.2%) were uniformly assigned as nonluminal A. Discordant subtyping was observed in 123 (40.7%) tumors.
Conclusions: Existing evidence on the comparative prognostic information provided by different tests suggests that current multiparameter tests provide broadly equivalent risk information for the population of women with estrogen receptor (ER)–positive breast cancers. However, for the individual patient, tests may provide differing risk categorization and subtype information.