Ischemia-modified albumin in children with clinically suspected acute myocarditis: diagnostic performance and incremental value beyond conventional biomarkers

Figures

• 

  Figure 1. Study design and participant grouping of the single-center case–control study. This schematic summarizes the prospective single-center case–control workflow, including prespecified inclusion/exclusion criteria, allocation into clinically suspected acute myocarditis (n = 47) and age-/sex-matched healthy controls (n = 47), laboratory measurements (IMA and conventional biomarkers), and the planned statistical analyses (group comparisons, ROC analysis, incremental discrimination modeling, clinical severity analyses, and correlation analysis). Numbers represent the final analyzed sample.

• 

  Figure 2. Violin plot of IMA (clinically suspected acute myocarditis vs. Control). Violin plots summarize the distribution of ischemia-modified albumin (IMA, ABSU) in each group; overlaid points represent individual participants (jittered to reduce overlap). The central tendency shown corresponds to the median distributional location; between-group comparison was performed using the Mann–Whitney U-test (Table 1).

• 

  Figure 3. ROC curve for IMA. Receiver operating characteristic (ROC) curve showing the ability of IMA to discriminate clinically suspected acute myocarditis cases from controls. The AUC with 95% confidence interval was obtained using bootstrap resampling, and the optimal cut-off was selected using the Youden index. The corresponding sensitivity and specificity at the Youden-optimal cut-off are reported in Table 4.

• 

  Figure 4. Pairwise differences in AUC between conventional biomarkers and IMA. Points represent pairwise bootstrap ΔAUC estimates (comparator minus IMA), and horizontal bars indicate 95% confidence intervals. Positive values indicate better discrimination than IMA.

• 

  Figure 5. ROC curves for incremental models (base vs. Base + IMA). ROC curves were constructed from predicted probabilities of logistic regression models. The base model included log10-transformed CK-MB, NT-proBNP, and CRP; the extended model added IMA. Model AUCs and ΔAUC (with uncertainty and p-value) were evaluated by bootstrap resampling and are reported in the Results.

• 

  Figure 6. Spearman correlation heatmap in clinically suspected acute myocarditis cases (r and p values). Heatmap displays Spearman's ρ for pairwise associations among biomarkers within the clinically suspected acute myocarditis group; each cell reports ρ along with the corresponding p-value.

• 

  Figure 7. Bubble plot summarizing AUC and strength of group differences (−log10 p). The x-axis indicates AUC from ROC analysis (Table 4), and the y-axis indicates −log10(p) from between-group comparisons (Table 1). Bubble size is proportional to AUC. This plot provides a visual summary of diagnostic discrimination alongside statistical evidence of group separation and does not replace Tables 1 and 4.