Preoperative lactate dehydrogenase-to-albumin ratio as a nutrition-inflammation index associated with postoperative pneumonia after McKeown esophagectomy: a retrospective cohort study.

BACKGROUND: Postoperative pneumonia is a frequent and clinically consequential complication after esophagectomy. The lactate dehydrogenase (LDH)-to-albumin ratio (LAR) integrates LDH-associated metabolic-inflammatory burden and albumin-associated nutritional-inflammatory reserve, but its association with postoperative pneumonia after McKeown esophagectomy remains unclear. This study evaluated whether preoperative LAR is associated with postoperative pneumonia in a clinically homogeneous, surgery-first cohort. METHODS: We retrospectively analyzed 467 consecutive patients with esophageal cancer who underwent curative-intent McKeown esophagectomy without neoadjuvant therapy between January 2022 and December 2024. LAR was calculated from LDH and albumin levels measured within 7 days before surgery and was analyzed primarily as a continuous variable. ROC-derived binary LAR was retained as a supplementary sensitivity analysis. The primary endpoint was postoperative pneumonia within 30 days. Discrimination was assessed using ROC analysis and DeLong's test. Associations were evaluated using staged multivariable logistic regression analyses, restricted cubic spline analyses, decision curve analyses, subgroup and timing analyses, and LDH-albumin joint-effect analyses. RESULTS: Postoperative pneumonia occurred in 147 of 467 patients (31.5%). The median onset time was postoperative day (POD) 6 [interquartile range (IQR), 4-9], and 98 events (66.7%) occurred within the first postoperative week. LAR showed good discrimination for postoperative pneumonia (AUC = 0.803), with an optimal Youden index cutoff of 5.855. Continuous LAR remained independently associated with postoperative pneumonia after full adjustment [Model 3 odds ratio (OR) per one-unit increase, 2.568; 95% CI, 2.085-3.163; p < 0.001], and estimates were stable after additional adjustment for surgical approach and conversion to open surgery. Restricted cubic spline analysis showed a significant overall association and non-linearity (P-overall < 0.001; P-non-linearity = 0.004). Pneumonia incidence increased across RCS-derived LAR strata from 7.9 to 25.9 and 59.8%. Binary LAR, subgroup analyses, early- and later-onset pneumonia analyses, and LDH-albumin joint-effect analyses supported the robustness and biological plausibility of the association. CONCLUSION: Preoperative LAR was independently associated with postoperative pneumonia after McKeown esophagectomy when analyzed as a continuous nutrition-inflammation index. LAR may support perioperative risk stratification and early postoperative surveillance, but the internally derived threshold requires external validation before its clinical generalization.