Abstract
BACKGROUND: Sepsis may precipitate acute respiratory distress syndrome (ARDS), a life-threatening pulmonary complication characterized by high mortality. The glucose-to-lymphocyte ratio (GLR) is a novel composite biomarker has demonstrated prognostic significance across multiple diseases. However, its association with short-term mortality in patients with sepsis-induced ARDS has not yet been clearly established. METHODS: In this multicenter retrospective cohort analysis, data from 2,485 individuals in the MIMIC-IV database were used to construct the primary derivation cohort, while an additional set of 298 patients from the Affiliated Hospital of Xuzhou Medical University served as an independent cohort for external validation. All-cause mortality at 28 days was defined as the primary outcome of the study. Least absolute shrinkage and selection operator (LASSO)-Boruta selected predictors. We evaluated the relationship between the GLR and mortality by employing multivariable Cox proportional hazards modeling, complemented by restricted cubic spline (RCS) and subgroup analyses. Receiver operating characteristic curves (ROC) and decision curve analysis (DCA) quantified incremental value of GLR over the Acute Physiology Score III (APS III) and the Simplified Acute Physiology Score II (SAPS II). RESULTS: In the derivation cohort, higher GLR values were independently linked to an increased risk of 28-day mortality (adjusted HR 1.13 per unit increase, 95% CI: 1.053-1.211, P < 0.001). The finding was replicated in the external validation cohort. Short-term mortality increased linearly with rising GLR levels, as shown by RCS analysis. Subgroup analyses identified significant interactions: the prognostic value of GLR was significantly attenuated or lost in patients with high illness severity scores (SAPS II > 40, APS III > 53) or severe liver disease, while it remained robust in patients with lower severity scores and without severe liver disease. Incorporating GLR into APS III and SAPS II models improved their AUC values (APS III: 0.694 vs. 0.708; SAPS II: 0.678 vs. 0.696). CONCLUSIONS: An elevated GLR at admission independently predicts 28-day mortality in patients with sepsis-induced ARDS. This marker is especially useful for early risk stratification among patients without advanced liver dysfunction or severe physiological abnormalities.