August 26, 2026Cardio-Oncology0 citationsOpen Access
Assessment of the HFA-ICOS risk tool for predicting cancer therapy–related cardiac dysfunction: a prospective cohort study
Key Result
The HFA-ICOS risk tool showed poor discrimination for predicting time to cancer therapy-related cardiac dysfunction in women with early-stage breast cancer, with a Harrell C-index of 0.43.
Key Points
- To determine the incidence of cancer therapy–related cardiac dysfunction (CTRCD) and evaluate the predictive performance of the HFA-ICOS risk stratification tool in early-stage breast cancer patients.
- Prospective cohort study enrolling 186 women with early-stage breast cancer receiving doxorubicin and/or trastuzumab, with 177 patients analyzed over 12 months of serial cardiac monitoring.
- Stratified baseline CTRCD risk into low (67%), moderate (28%), and high (5%) tiers using HFA-ICOS proformas and evaluated time-to-event outcomes using Cox models and Harrell C-index.
- CTRCD occurred in 53 patients (28.6%), yielding an overall incidence rate of 0.266 events per person-year (95% CI 0.200–0.348), with group-specific rates of 0.326 (low), 0.145 (moderate), and 0.248 (high).
- Pairwise restricted mean survival time comparisons showed no significant differences after Holm adjustment (all adjusted p ≥ 0.124), and the HFA-ICOS tool demonstrated poor discrimination for CTRCD onset (Harrell C-index 0.43, 95% CI 0.37–0.49).
Study Design
Type
Cohort (n=177)
Multicenter
Yes
Structured PICO
P
Population
177 women with early-stage breast cancer receiving doxorubicin and/or trastuzumab, followed for 12 months to assess the incidence of cancer therapy-related cardiac dysfunction.
E
Exposure
Classification into moderate or high risk categories using the Heart Failure Association-International Cardio-Oncology Society (HFA-ICOS) risk stratification tool prior to chemotherapy.
C
Comparator
Classification into the low risk category using the HFA-ICOS risk stratification tool.
O
Outcome
Time to the first occurrence of cancer therapy-related cardiac dysfunction (CTRCD) of any severity following chemotherapy initiation over 12 months follow-up.surrogate
The HFA-ICOS risk stratification tool demonstrated poor discrimination for predicting predominantly asymptomatic, imaging-detected cancer therapy-related cardiac dysfunction at 12 months in breast cancer patients.
Main Result
Effect estimate: Harrell C-index 0.43 (95% CI 0.37-0.49)
Limitations
- Symptomatic CTRCD could not be confidently adjudicated by the study team
- Natriuretic peptides (BNP/NT-proBNP) measurements were unavailable and not incorporated
- Serial changes in troponin concentrations were not incorporated into the working definition of CTRCD
- Very high-risk patients were excluded from the study
Abstract
Anthracyclines and HER2-targeted breast cancer treatments can cause cancer therapy–related cardiac dysfunction (CTRCD). The Heart Failure Association–International Cardio-Oncology Society (HFA-ICOS) risk tool assesses this risk, but its accuracy remains uncertain. We aimed to (1) determine CTRCD incidence using the 2021 CTRCD definition in breast-cancer patients receiving doxorubicin and/or trastuzumab; (2) evaluate the HFA-ICOS risk proformas, and (3) assess the individual risk factors incorporated in the proformas. In this prospective study, 186 women with early-stage breast cancer were enrolled to undergo baseline and serial cardiac evaluations over 12 months. After exclusions, 177 patients were included for risk stratification using the Heart Failure Association–International Cardio-Oncology Society (HFA-ICOS) score. Baseline risk distribution was 67% low ( n = 116), 28% moderate ( n = 52), and 5% high ( n = 9). During follow-up, 53 patients (28.6%) developed cancer therapy–related cardiac dysfunction (CTRCD), corresponding to an overall incidence rate of 0.266 events per person-year (95% CI 0.200–0.348). Incidence rates were 0.326, 0.145, and 0.248 in the low-, moderate-, and high-risk groups, respectively; pairwise 365-day restricted mean survival time (RMST) comparisons did not remain statistically significant after Holm adjustment (all adjusted p > = 0.124). In Cox models adjusted for treatment type, CTRCD risk did not differ significantly between the moderate- and high-risk groups compared with the low-risk group. The ordinal HFA-ICOS category showed poor discrimination for time to CTRCD (Harrell C-index 0.43, 95% CI 0.37–0.49). The HFA-ICOS proforma showed limited ability to stratify CTRCD risk in this cohort. These findings highlight limitations in current guideline-based risk models and the need for improved stratification strategies.