External validation of the HFA-ICOS risk stratification tool for prediction of cardiovascular complications after cancer treatment

Abstract Background The 2022 ESC Guidelines on Cardio-oncology recommends using the HFA-ICOS (Heart Failure Association and the International Cardio-Oncology Society) four-category risk stratification tool to predict risk of heart failure (HF) and other cardiovascular disease (CVD) complications during and after cancer therapy. However, the actual risk associated with assignment to these categories is unknown. Purpose We aimed to assess the prognostic performance the HFA-ICOS risk stratification tool in a large cancer subpopulation of a general health survey linked to Norwegian health registries. Methods Participants in a large-scale population study diagnosed with cancer between 2006 and 2012 were included (n=2290) and followed for CVD endpoints until Sept 2023. CVD risk factors were measured at baseline between 2006 and 2008. The Norwegian Cancer Registry provided detailed information about cancer characteristics and therapy. We retrieved CVD endpoints from the local hospitals’ registries and the national causes of death registry. Kaplan Meier survival curves were used to estimate the risk of the combined endpoint of HF hospitalization, myocardial infarction (MI), stroke and CVD death, as well as HF hospitalization alone and non-CVD death. Harrel’s C-statistic was used to validate the discriminative abilities. Results Mean (SD) age was 65.8 (12.4) years, 48% were female, cancer types and cardiovascular risk factors are shown in Table 1. During a median (interquartile range) follow-up of 6.9 (1.4 to 12.2) years, there were 460 first CVD events (HF hospitalizations 87 (19%), MIs 161 (35%), strokes 130 (28%), CVD deaths 82 (18%)) and 1003 (44%) non-CVD deaths. When applying the HFA-ICOS tool, 649 (28%) were assigned to the low-risk category, 912 (40%) to medium, 661 (29%) to high-risk and 68 (3%) to the very high-risk category. Between the HFA-ICOS risk categories, there were clear differences in the observed risk of CVD events, HF hospitalizations and non-CVD mortality (Figure 1). CVD incidence was highest in the highest HFA-ICOS risk categories (7% in the low risk group, 21% in the medium risk group, 31% in the high risk group and 37% in the very high-risk group (p-value <0.001)). C-statistics for predicting CVD events was 0.696 (95% CI, 0.674 to 0.718). When only assessing patients with HF hospitalizations as (first) outcome (n=116), the risk of HF was higher in the high and very high-risk groups (0% in the low risk group, 4% in the medium, 9% in high and 21% in the very high risk group (C-statistics 0.794 (95% CI, 0.762 to 0.827)). Conclusion The HFA-ICOS baseline risk assessment was useful to identify patients at risk of both clinical overt heart failure and other major CV complications and non-CV death in a population with a variety of cancers. These results can help communicate absolute risks to cancer patients, which may be helpful in the shared decision-making for cardioprotective treatment and intensified follow-up.