Galectin-9, FABP4 and pentraxin-3 are associated with myocardial dysfunction and discriminate chemotherapy-related cardiac dysfunction in anthracycline-treated breast cancer: a single-centre retrospective study

Figures

• 

  Figure 1. Cohort flow (STROBE). Of 286 women screened, 67 were excluded for the reasons shown, leaving 219 in the analytic cohort, stratified by on-treatment cancer therapy-related cardiac dysfunction (CTRCD) into cardiotoxicity (n = 66) and non-cardiotoxicity (n = 153) groups. CTRCD was defined per the 2014 ASE/EACVI consensus and 2022 ESC cardio-oncology guideline; the component breakdown is in Supplementary Table 1.

• 

  Figure 2. Serum biomarkers after the third chemotherapy cycle, by group. Bars show group mean and whiskers the standard deviation for galectin-9, FABP4 and pentraxin-3, all higher in the cardiotoxicity group (***P < 0.001; independent-samples t-test).

• 

  Figure 3. Echocardiographic indices from baseline to post-chemotherapy, by group. Left ventricular ejection fraction (LVEF), absolute global longitudinal strain (|GLS|) and myocardial composite index (MCI) were comparable at baseline (all P > 0.40) and diverged after chemotherapy; significance refers to the between-group difference in change from baseline. Points are group means.

• 

  Figure 4. Correlation heat-map. Pearson correlation coefficients between each serum biomarker and LVEF, |GLS| and MCI; all correlations are inverse and significant (P < 0.001). Adjusted partial correlations are in Supplementary Table 5.

• 

  Figure 5. Discrimination of CTRCD and internal validation. Left, ROC curves for each biomarker and the combined model, with operating points (circles) at the reported sensitivity/specificity; curves are binormal fits to the reported AUCs. Right, apparent AUC (95% CI) and bootstrap optimism-corrected AUC. The combined model significantly outperformed each single marker (DeLong P < 0.05).

• 

  Figure 6. Decision-curve analysis for on-treatment CTRCD. Across threshold probabilities of roughly 10–50%, adding the three-biomarker panel to the clinical model yielded higher net benefit than the clinical model, treat-all or treat-none strategies.