Comparison of multiparameter tests in the UK OPTIMA-Prelim trial

Introduction All published adjuvant chemotherapy trials in breast cancer have made the assumption that the proportional benefits of chemotherapy apply uniformly across molecular subgroups. However, it can be argued that chemotherapy effectiveness for luminal A breast cancer is low in comparison to other subtypes irrespective of tumour stage. A logical extension of this argument is that novel multiparametric tests that use biological features of breast cancers to assess risk may also inform chemotherapy benefit in luminal cancers. The OPTIMA trial is a multi-centre, partially blinded, randomised clinical trial with a non-inferiority endpoint, and an adaptive design, to compare standard treatment (chemotherapy followed by endocrine therapy) with multi-parameter test-guided treatment allocation to either chemotherapy followed by endocrine therapy or endocrine therapy alone. OPTIMA-prelim aimed to compare the predicted risk stratification, sub-type classification and cost effectiveness of different multiparameter tests performed on the same patient population. Methods Over 20 months of recruitment 285 patients were randomised to OPTIMA-prelim. Tissue was collected centrally, ER and HER2 status confirmed and samples provided for testing with Oncotype DX™, Prosigna™ (PAM50), Mammaprint™, Mammatyper™, IHC4-AQUA and IHC4 using conventional biomarkers. Sub-type classification was provided by Blueprint™, Mammatyper™ and Prosigna™. Each test was performed at central diagnostic laboratories (OncotypeDx, Mammaprint/Blueprint, Mammatyper) or in a central laboratory (Prosigna™/IHC4) strictly according to GLP practices. Results Samples from 181 patients randomised by January 2014 were tested and data analysed for this study. Patients were categorised as low/intermediate or high risk using predetermined cut-offs for each test. Oncotype DX predicted a proportion of low-risk tumours (79%; 95% CI 73-85%) similar to that predicted as either low or intermediate risk using Prosigna ROR_P (71%; 95% CI 64-78%) and IHC4 (69%; 95% CI 62-76%), whilst MammaPrint identified the fewest low-risk tumours (59%; 95% CI 52-66%). Strikingly, a comparison between tests showed modest agreement between tests when dichotomising results between high vs low/intermediate risk. Disagreement between different tests, in assigning individual tumours to risk categories, is not uncommon; for the four tests [Oncotype DX, MammaPrint, Prosigna ROR_P (low/int) and IHC4 (low/int)], only 71 (39%) tumours were classified as low/intermediate risk for all four tests and only 17 (9%) tumours were high risk for all four tests, 93 (52%) tumours were assigned to different risk categories by different tests. Similarly all three subtypes tests (Blueprint/Prosigna/Mammatyper) each assigned 59% of tumors to luminal A subtype but only 70% of these cases were classified as luminal A by all three assays. Conclusion Existing evidence on the comparative prognostic information provided by different tests suggests current multiparameter tests provide broadly equivalent risk information for the population of women with luminal breast cancers. However, for the individual patient, tests may provide differing risk categorisation or indeed subtype information