# A predictive model for left ventricular reverse remodeling after pharmacological therapy in children with recent-onset dilated cardiomyopathy

## TL;DR

This study creates a model to predict heart improvement in children with a heart condition called dilated cardiomyopathy after drug treatment.

## Contribution

The study introduces a validated nomogram to predict left ventricular reverse remodeling in pediatric dilated cardiomyopathy patients.

## Key findings

- Age, left ventricular end-diastolic dimension Z-score, and QRS interval were significant predictors of LVRR.
- The nomogram showed high discrimination (C-index 0.903) and good calibration (Hosmer-Lemeshow P = 0.207).
- The model was clinically useful for threshold probabilities greater than 4%.

## Abstract

Pharmacological advances have improved pediatric dilated cardiomyopathy (DCM) prognosis, which manifests as left ventricular reverse remodeling (LVRR). However, significant inter-individual variability exists in therapeutic response. Identifying predictors is critical for individualizing management to inform device and transplant timing.

To develop a nomogram for predicting LVRR in pediatric DCM.

A retrospective analysis of 146 children hospitalized for DCM from January 2012 to June 2023. 55 exhibited LVRR. A nomogram predicting pediatric DCM-LVRR was developed using univariate analysis and logistic regression to select predictors. The nomogram was validated via bootstrapping and receiver operating characteristic curves for discrimination. Calibration was assessed with the Hosmer-Lemeshow test. Decision curve analysis evaluated performance and utility.

Age, left ventricular end-diastolic dimension Z-score, and QRS interval were associated with the occurrence of LVRR. Discrimination was high (C-index 0.903) and internally validated on bootstrapping with 1000 repetitions (Adjusted C-index 0.895). The Hosmer-Lemeshow test revealed no significant deviation between nomogram predictions and outcomes (χ2 =  10.883; P =  0.207). DCA revealed that the model was clinically useful at threshold probabilities > 4%.

We developed and internally validated a nomogram predicting LVRR for pediatric DCM patients, exhibiting high sensitivity, specificity and clinical utility.