Wasting phenotype for differentiating spinal tuberculosis from spinal pyogenic infection

Spinal tuberculosis (STB) and pyogenic spinal infection (PSI) often present with overlapping clinical manifestations and imaging features, leading to delayed diagnosis and suboptimal outcomes. Identifying reliable laboratory-based markers may improve early differential diagnosis.

To investigate wasting phenotype-related clinical and laboratory indicators for differentiating STB from PSI and to establish a clinically applicable diagnostic model.

In this prospective study, 253 patients with confirmed spinal infections were enrolled, including 159 with STB (62.85%) and 94 with PSI (37.15%). Demographic, clinical, and routine laboratory data were collected. Eight candidate metabolic and inflammatory variables were assessed using univariate analyses and multivariable logistic regression. Model performance was evaluated by receiver operating characteristic (ROC) analysis, calibration testing, and Youden-derived optimal thresholds.

Three key variables-body temperature, high-density lipoprotein (HDL), and blood glucose-were independently associated with STB. In the adjusted model, each 1 °C increase in temperature reduced the odds of STB by approximately 69% (OR = 0.318; 95% CI: 0.138-0.731; P = 0.009), each 1 mmol/L increase in HDL increased the odds by 3.7-fold (OR = 3.692; 95% CI: 1.311-10.394; P = 0.011), and each 1 mmol/L increase in blood glucose reduced the odds by 24% (OR = 0.764; 95% CI: 0.622-0.938; P = 0.014). The model demonstrated moderate discrimination (AUC = 0.673, 95% CI: 0.604-0.740) but good calibration (P = 0.609). ROC-derived optimal thresholds were T ≤ 36.7 °C, HDL ≥ 0.89 mmol/L, and blood glucose ≤ 5.85 mmol/L, providing practical reference points for clinical application.

A composite wasting phenotype defined by lower body temperature, lower blood glucose, and elevated HDL significantly improves early differentiation of STB from PSI. While individual thresholds show limited standalone diagnostic value, the combined model provides a biologically plausible, interpretable, and clinically useful tool to aid decision-making in managing spinal infections.

Level 3 (According to the Oxford CEBM 2016 criteria for diagnostic studies).