Age-dependent immune heterogeneity and the limited diagnostic utility of peripheral blood biomarkers in pediatric tuberculosis: a real-world cohort study with multidimensional model evaluation

Background Pediatric tuberculosis (TB) remains a diagnostic challenge due to atypical presentations and low pathogen detection rates. Overlapping clinical features with Mycoplasma pneumoniae (MP) infection further complicate differential diagnosis. Immune-inflammatory parameters are age-dependent, yet how age modifies these profiles in children with TB and MP co-infection remains poorly understood. Objective To investigate age-dependent heterogeneity in immune-inflammatory profiles among children with tuberculosis (TB) and to evaluate the additive diagnostic value of peripheral blood biomarkers beyond age alone, while assessing the impact of Mycoplasma pneumoniae co-infection on TB-related immune parameters. Methods This retrospective observational study enrolled 280 children aged 0–18 years (186 TB cases, 94 non-TB controls) admitted between January 2024 and December 2025. Peripheral lymphocyte subsets, routine blood parameters, and systemic inflammatory indices (NLR, PLR, SII, MLR, SIRI) were compared between groups. Logistic regression identified independent correlates. ROC analysis, calibration curves, decision curve analysis (DCA), net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were used to evaluate model performance. Age-stratified analysis (three strata: <5, 5–12, ≥12 years) was performed using ANCOVA with age as a covariate. Age × Group interaction analyses and 1:1 age-matched sensitivity analyses were conducted to formally test effect modification and exclude age confounding. Sensitivity analyses comparing bacteriologically confirmed versus clinically diagnosed cases and MP-only controls versus TB-MP co-infected cases were also performed. Results Age was the sole variable independently associated with pediatric TB (adjusted OR = 1.118, 95% CI: 1.046–1.194, P = 0.001). TB patients exhibited significantly lower CD8 + T cell, NK cell, and lymphocyte counts and higher PLR compared with controls, but these differences were highly age-dependent. Age-stratified analysis revealed striking heterogeneity: immune-inflammatory differences were highly significant in adolescents ≥12 years (CD8 + T cells and NK cells: adjusted P < 0.009 and P < 0.008, respectively) but virtually absent in children <5 years (all P > 0.05). Formal Age × Group interaction analyses confirmed significant interactions for CD8 + T cells (P < 0.001, partial η 2 = 0.089), NK cells (P < 0.001, partial η 2 = 0.196), and lymphocyte count (P < 0.001, partial η 2 = 0.404). Age-matched sensitivity analysis further confirmed that CD8 + T cell and NK cell differences in the ≥12 years stratum remained significant after matching (CD8 + T: P = 0.004; NK: P < 0.001), whereas CD4 + T cell differences were abolished (P = 0.113). The age-only model produced an AUC of 0.686 (95% CI: 0.618–0.755). The age + CD8 + T cell combined model showed statistically significant improvement (AUC difference = 0.041, 95% CI: 0.010–0.071; DeLong P = 0.009; likelihood-ratio test P < 0.001), with good calibration (Hosmer–Lemeshow P = 0.228; Brier score = 0.187; McFadden R 2 = 0.127), positive net benefit across threshold probabilities of approximately 0.40–0.78 in DCA, and significant reclassification improvement (continuous NRI = 0.505, 95% CI: 0.258–0.755, P < 0.001; IDI = 0.038, 95% CI: 0.011–0.068, P = 0.003; event NRI + = 0.419). None of the other age-combined models (age + NLR, SII, PLR, or MP) showed significant improvement over age alone (all DeLong P > 0.05). Within the TB cohort, MP co-infection was not observed to significantly alter immune-inflammatory profiles after age adjustment (all P > 0.05, partial η 2 < 0.01). Sensitivity analyses confirmed immunological consistency between bacteriologically confirmed and clinically diagnosed cases (all P > 0.05, partial η 2 ≤ 0.017). Conclusion Age is the key variable associated with pediatric TB and demonstrates marked age-dependent immune-inflammatory heterogeneity. Peripheral blood immune biomarkers added modest diagnostic information beyond age alone, with statistically significant yet clinically modest improvements. The incremental value of CD8 + T cell count was most evident in intermediate-risk decision scenarios (threshold probabilities 0.40–0.78) and in adolescents ≥12 years, rather than as a universal screening tool. MP co-infection was not associated with significant alterations in TB-related immune-inflammatory profiles in this cohort. These findings emphasize age stratification as a core principle in pediatric TB evaluation and suggest caution regarding over-reliance on routine lymphocyte subset testing for diagnostic purposes, particularly in younger children where clinical and radiological assessment remains paramount.