Adjust cut-off values of immunohistochemistry models to predict risk of distant recurrence in invasive breast carcinoma patients.

Background: Multigene assays are recommended for hormone receptor-positive invasive breast carcinoma to determine the risk of recurrence, but they are highly expensive. We investigated the prognostic values of immunohistochemistry (IHC)-based prognostic models as an alternative to multigene assays.
Methods: The risk categories estimated by the IHC-based prognostic models were correlated to those estimated by the multigene assays in 71 cases and the follow-up results in 642 consecutive cases of HER2- luminal-type early breast cancer. Cut-off values of IHC-based models were adjusted based on survival outcome to reveal maximum Harrell C index or based on the maximum positive likelihood ratio correlated to multigene assay.
Results: All investigated IHC-based models could predict the risk of distant recurrence, but their cut-off values required adjustment. Using distant recurrence-free survival (DRFS) to refine the cut-off values could improve the prognostic values. Adjusting the cut-off values using the results of multigene assays, the positive predictive values of an estimate of low risk or low recurrence score (≤ 21) were higher than 90%. On average, 23% of cases got different results of risk assessment after adjustment. Although cut-off values adjusted by multigene assay were not identical to those refined by survival, the adjusted values (17.1 and 23.8) and the refined values (17.5 and 24.5) of the best model (Magee Eq. 1) were close. Among all the evaluated models, Magee equation 2 was the only one without Ki67, and its prognostic values were the lowest. Using 20% as cut-off for Ki67 as suggested by St. Gallen consensus, we could confidently define luminal A cancer.
Conclusion: It is necessary to adjust the cut-off values of IHC-based prognostic models to fit the purpose. If the estimated risk is clearly high or low, it may be reasonable to omit multigene assays when cost is a consideration.
(Copyright © 2016. Published by Elsevier Taiwan LLC.)