ATROPOS: Survival Analysis of ICI-associated Myocarditis

"ATROPOS (Assessing lethaliTy and Risk factOrs in immune checkPoint inhibitors-induced myOcardia survival analysis based on an international registry) is an observational, non-interventional analysis of the international REDCap registry of immune checkpoint inhibitor-associated myocarditis (ICI-M).

The study population consists of cancer patients exposed to immune checkpoint inhibitors and recorded in the registry with possible, probable, or definite ICI-M according to registry/adjudication definitions. The main objective is to evaluate associations between immunosuppressive treatment strategies and overall mortality, as a function of myocarditis severity, focusing on first-line glucocorticoids and the subsequent use of second-line immunosuppressants. The study will also assess whether these associations are modified by myocarditis severity. Secondary analyses will examine mortality at 30, 90 and 360 days, cause-specific death (e.g. myotoxicity, cancer-related) where available, and clinical risk factors related to oncology history, ICI exposure, symptoms, ECG, echocardiography, cardiac MRI, biomarkers including CK and troponin, other immune-related adverse events, heart failure, respiratory muscle failure, arrhythmias and conductive disorders, sepsis, and pharmacological treatments.

The statistical approach will include descriptive analyses and Cox proportional-hazards modelling. Time-dependent covariates will be considered for variables such as troponin, left ventricular ejection fraction, severity/grade of the myocarditis/myotoxicity, glucocorticoids and other immunosuppressants. Missing data will be described and handled through prespecified methods, with MICE and JointAI used to cross-check robustness of results. Because treatment strategies are not assigned by the protocol, all treatment-effect analyses will be interpreted as observational associations, adjusted for prespecified clinical confounders.

ATROPOS does not assign study treatment, change patient management, require extra visits, or involve a study drug/device. The central analysis dataset uses coded or pseudonymised data, with direct identifiers retained only by local sites when applicable under their own governance."