Serum nutritional-metabolic recovery during tuberculosis treatment across BMI-defined undernutrition strata: A longitudinal proteomic reanalysis

Background Tuberculosis remains the leading cause of death from a single infectious agent, and undernutrition is a major modifiable risk factor. Serum proteomics can capture systemic host recovery during therapy, but whether baseline BMI-defined undernutrition alters this trajectory is unclear. Methods We reanalysed processed longitudinal tandem-mass-tag serum proteomics from 30 adults with pulmonary tuberculosis and durable cure sampled at baseline and weeks 8, 17, 26 and 52. Thirteen participants had baseline BMI <18.5 kg/m2 and 17 had BMI >=18.5 kg/m2. Protein-level analyses evaluated baseline differences and week-specific within-participant change; module and recovery-axis models evaluated BMI-by-time interactions, with random-intercept mixed models added as a repeated-measures sensitivity analysis. Three complementary treatment-responsive modules were integrated into an exploratory, nonredundant 36-protein nutritional-metabolic summary axis. Results The reconstructed dataset contained 1,923 quantified features across 150 participant-visits. Inflammatory/acute-phase and metal-handling/nutritional-immunity module scores declined during follow-up, whereas lipid/apolipoprotein-transport, nutritional-carrier and endocrine-immunometabolic scores increased. In the primary week-specific paired-change analyses, no protein-level between-BMI difference in longitudinal change survived false-discovery-rate correction; no curated-module BMI-by-time interaction survived correction. The 36-protein summary axis increased in both BMI strata from week 17 onward. In a random-intercept model without interaction, the BMI <18.5 kg/m2 stratum had a lower average axis score across visits than the BMI >=18.5 kg/m2 comparison stratum, although the confidence interval included zero (difference -1.34, 95% CI -2.76 to 0.08; p = .064). Categorical and continuous BMI-by-time interaction tests did not detect differential axis trajectories. Conclusions Tuberculosis treatment was accompanied by broad longitudinal remodelling of circulating inflammatory and nutritional-metabolic protein systems. An exploratory 36-protein summary axis increased in both baseline BMI strata, while formal interaction tests did not detect BMI-dependent trajectories and remained too imprecise to establish equivalence. BMI should be regarded as an anthropometric proxy rather than a comprehensive measure of nutritional state. The defined axis provides a framework for prospective validation alongside direct measures of body composition, micronutrients, treatment exposure and clinical outcomes.