Incremental prognostic value of clinical, functional and imaging variables in acute myocarditis: a hierarchical risk modelling approach

Background

Risk stratification in acute myocarditis remains challenging due to the heterogeneity of clinical presentation and disease severity. Individual prognostic markers—including electrocardiographic abnormalities, biomarkers of myocardial injury, left ventricular systolic function, and cardiac magnetic resonance (CMR) findings—have been associated with outcomes; however, their relative and incremental contributions within a unified risk framework are not well established.

Purpose

To assess the incremental prognostic value of clinical, laboratory, echocardiographic, and CMR variables using a hierarchical modelling strategy for the prediction of cardiovascular-related hospitalization in patients with acute myocarditis.

Methods

In a retrospective cohort of patients with clinically suspected acute myocarditis, four hierarchical logistic regression models were sequentially constructed to predict cardiovascular-related hospitalization during follow-up. Model 1 included ECG status (normal vs altered). Model 2 added peak troponin levels. Model 3 further incorporated left ventricular ejection fraction (LVEF). Model 4 additionally included CMR late gadolinium enhancement (LGE), dichotomized according to median extent. Model discrimination was assessed using receiver operating characteristic (ROC) curve analysis and area under the curve (AUC).

Results

A total of 138 patients with acute myocarditis were included (81.2% male; median age 28 years [IQR 22–38]). On admission, 51.4% of patients exhibited ST-segment elevation on ECG, while 39.1% had a normal ECG. Median peak troponin was 6,803 ng/L (IQR 1,653–14,325). Median LVEF was 58% (IQR 55–61). CMR demonstrated myocardial oedema in 71.3% and LGE in 86.1% of patients, with a median of 3 involved segments. During follow-up, cardiovascular-related hospitalization occurred in 13.8% of patients.

Discriminative performance improved progressively across hierarchical models: Model 1 (ECG only) AUC 0.55; Model 2 (ECG + troponin) AUC 0.55; Model 3 (ECG + troponin + LVEF) AUC 0.59; and Model 4 (ECG + troponin + LVEF + CMR LGE) AUC 0.64. The addition of imaging-based markers, particularly CMR LGE, resulted in the highest discriminatory performance, although overall predictive accuracy remained moderate.

Conclusions

In patients with acute myocarditis, a hierarchical risk modelling approach demonstrates incremental prognostic value with the stepwise addition of functional and tissue-characterization variables. While ECG findings and biomarkers alone show limited discrimination, incorporation of left ventricular systolic function and CMR late gadolinium enhancement improves risk stratification for cardiovascular-related hospitalization. These findings support the role of multimodality assessment in the early prognostic evaluation of myocarditis and provide a framework for future, larger-scale risk prediction studies.For image description, please refer to the figure legend and surrounding text.For image description, please refer to the figure legend and surrounding text.