Evaluation of Risk Prediction Models to Identify Cancer Therapeutics Related Cardiac Dysfunction in Women with HER2+ Breast Cancer.

ABSTRACT

Cancer-therapeutics-related cardiac dysfunction (CTRCD) is an important concern in women receiving trastuzumab therapy for HER2+ breast cancer. However, the ability to assess CTRCD risk remains limited. In this retrospective cohort study, we apply three published risk prediction models (Ezaz et al., NSABP-31 cardiac risk scores (CRS), and HFA-ICOS trastuzumab proforma) to 629 women (mean age 52.4 ± 10.9 years) with Stage I-III HER2+ breast cancer treated with trastuzumab ± anthracyclines to assess their performance to identify CTRCD during or immediately post treatment. Using these models, patients were classified into CTRCD risk categories according to the pre-treatment characteristics. With NSABP-31 CRS and HFA-ICOS proformas, patients in the highest risk category had a 1.7-to-2.4-fold higher relative risk of CTRCD than the low-risk category (p = 0.010 and 0.005, respectively). However, with all three risk models, those in the low-risk category had a high absolute risk of CTRCD (15.5-25.5%). The discrimination of the models for CTRCD (AUC 0.51-0.60) and their calibration was limited. NSAP-31 CRS and HFA-ICOS proformas can identify relative differences in CTRCD risk between patients, but when considering absolute risk, they are only able to identify the highest risk patients. There remains an ongoing need for accurate CTRCD risk prediction models in women with HER2+ breast cancer.