Abstract / Summary
Background: Kawasaki disease (KD) is an acute childhood vasculitis in which coronary artery lesions (CALs) determine long-term cardiovascular risk. We evaluated a focused pre-IVIG biomarker profile in relation to baseline Z-score-defined coronary involvement.
Objective: To examine whether interleukin-17A (IL-17A), interleukin-8 (IL-8), and platelet count (PLT) were associated with baseline CAL and to assess the robustness and internal discrimination of this exploratory profile.
Methods: This single-center retrospective exploratory cohort study screened 242 children with suspected KD treated in 2023. After 43 exclusions, 199 children were analyzed (non-CAL, n = 166; CAL, n = 33). CAL was defined using pretreatment segment-level coronary Z-scores recorded in the institutional echocardiographic database. To limit model complexity, the primary Firth-penalized logistic model was restricted to three predictors (PLT, IL-8, and IL-17A; 11 events per predictor). A nine-covariate clinically informed model and illness-timing/incomplete-KD models were treated as sensitivity analyses. Internal validation used 1000-resample bootstrap optimism correction and 100 repeated stratified fivefold cross-validation with complete refitting.
Results: Compared with the non-CAL group, children with CAL had higher PLT, IL-6, IL-8, and IL-17A and lower hematocrit (HCT), among other group-level differences. In the primary Firth model, PLT (OR, 3.081 per 100 × 109/L; 95% CI, 1.582-6.002), IL-8 (OR, 1.167 per 10 pg/mL; 95% CI, 1.076-1.266), and IL-17A (OR, 6.555 per 10 pg/mL; 95% CI, 3.054-14.071) were associated with baseline CAL. The apparent AUC was 0.975, the optimism-corrected AUC was 0.972, and the repeated-cross-validation mean AUC was 0.966 (SD, 0.006).
Conclusions: Higher pre-IVIG IL-17A, IL-8, and PLT were associated with baseline Z-score-defined CAL. These internally validated but single-center observational findings neither establish mechanism nor justify biomarker-guided imaging or treatment. Prospective multicenter external validation is required before clinical use.