Nonlinear association between body mass index and in-hospital pulmonary complications after esophagectomy following neoadjuvant therapy for esophageal squamous cell carcinoma

Abstract Background Pulmonary complications remain common after esophagectomy. Risk assessment is particularly relevant after neoadjuvant therapy (NAT), which can worsen nutritional status and muscle reserve. Body mass index (BMI) is readily available, but its association with pulmonary complication risk may be nonlinear. Methods We performed a single-center retrospective cohort study at Shanxi Province Cancer Hospital. Among 364 neoadjuvant-treated patients with esophageal squamous cell carcinoma who underwent curative-intent esophagectomy, the primary endpoint was in-hospital pulmonary complications (pneumonia, pleural effusion requiring intervention, or pneumothorax). BMI was modeled as a continuous exposure using multivariable logistic regression and restricted cubic splines. Models were adjusted for demographic and clinical covariates, neoadjuvant regimen, and surgical route and approach. Modified Poisson regression with robust variance-estimated adjusted risk ratios were performed. Sensitivity analyses were additionally adjusted for anastomotic leak. Results Pulmonary complications occurred in 134/364 patients (36.8%). Patients with pulmonary complications had lower BMIs (22.43 ± 3.22 vs. 23.27 ± 2.91; P = 0.013) and more frequent anastomotic leak [48 (35.8%) vs. 45 (19.6%); P < 0.001]. In the adjusted linear BMI model, BMI was not significantly different [OR = 0.93 (0.86–1.01); P = 0.085]. In contrast, spline modeling revealed a significant overall association and nonlinearity [ P overall = 0.006; P nonlinear = 0.006], with a higher risk associated with a low BMI [18.5 vs. 22.8: OR = 2.879 (1.504–5.511)]. Estimates at higher BMIs were imprecise [28 vs. 22.8: OR 1.387 (0.751–2.559)]. Restricted cubic spline-based modified Poisson analysis revealed a directionally consistent nonlinear pattern, with excess risk concentrated at lower BMIs [BMI 18.5 vs. 22.8 kg/m 2 : RR = 1.689 (95% CI 1.260–2.262)]. The spline association persisted after adjustment for anastomotic leak ( P overall = 0.004; P nonlinear = 0.006). Conclusions Preoperative BMI was nonlinearly associated with in-hospital pulmonary complications after neoadjuvant-treated esophagectomy, with the excess risk concentrated at low BMIs and greater uncertainty at BMI extremes. A possible increase in risk at higher BMIs requires validation in larger cohorts. In the neoadjuvant era, these findings support the interpretation of BMI as a nonuniform risk marker and provide hypothesis-generating evidence for perioperative optimization at the postneoadjuvant and preoperative time points, as well as for nutritional and prehabilitation strategies during neoadjuvant treatment to mitigate pulmonary risk.