Evaluating Biomarkers as Predictors of Cancer Therapy Cardiotoxicity: All you Need is a Meta-Analysis?. Letter Regarding the Article ‘Troponins and Brain Natriuretic Peptides for the Prediction of Cardiotoxicity in Cancer Patients: A Meta-Analysis.’

We read with great interest the meta-analysis by Michel et al.1 who investigated troponins and B-type natriuretic peptides (NPs) as indicators of cardiac damage and predictors of systolic dysfunction in adult patients receiving anticancer drugs (anthracyclines, HER2 inhibitors or their combination, myeloablative therapy for stem cell transplantation, radiotherapy, or other chemotherapy regimens). Herein, we would like to highlight some possible issues with their analysis, which may inform future research on this topic. The literature search was done in November 2018. It did not include recent significant evidence such as the study reporting that the risk of cardiac dysfunction over 4 years is two-fold higher when high-sensitivity troponin T (hs-TnT) is >14 ng/L at the end of anthracycline treatment and that a doubling of N-terminal fragment of pro-B-type NP during therapy is associated with a 1.6-fold higher risk of cardiac dysfunction.2 Similarly, the Authors did not consider a novel and promising field of research, namely troponins as indicators of myocarditis by immune checkpoint inhibitors.3 Furthermore, the whole analysis revolved around a comparison between normal and elevated troponin levels, but we are not entirely convinced of the opportunity to pool together troponin T and I assays of different generations, with highly different upper reference values (e.g. 14 ng/L for hs-TnT and 80 ng/L for troponin I with the I-STAT assay).4 We may add that reducing plasma troponin values to a binary variable (elevated vs. normal) is likely to underexploit the information conveyed by a troponin dosage. Troponin increase from baseline to the end of treatment might further refine prognostic stratification, also because high-sensitivity assays have a low intra-individual coefficient of variation, which means that random fluctuations are small. Troponin levels are reliably associated with the severity of ongoing myocardial damage.5 Finally, reduced left ventricular ejection fraction is only a surrogate endpoint, and biomarkers should be preferably evaluated as predictors of more clinically robust outcome measures, such as heart failure hospitalization or cardiovascular mortality. In summary, Michel et al. should be congratulated for their effort to provide a synthetic assessment of the sparse and heterogeneous evidence on troponins and NPs as indicators and predictors of cardiotoxicity. Nonetheless, their analysis is burdened by some limitations that could be overcome only by pooling together individual patient data, or by designing prospective studies aiming to answer specific questions.