Preoperative inflammatory indices derived from blood cell counts and their association with early postoperative pulmonary complications in patients with esophageal squamous cell carcinoma undergoing neoadjuvant chemoimmunotherapy: a retrospective coh

Postoperative pulmonary complications (PPCs) are important adverse events in patients with esophageal squamous cell carcinoma (ESCC) undergoing neoadjuvant chemoimmunotherapy followed by radical esophagectomy. Although preoperative inflammatory indices have been associated with postoperative outcomes, their perioperative relevance and incremental predictive value in this specific population remain uncertain. This study aimed to evaluate the associations of the preoperative systemic immune-inflammation index (SII), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) with early PPCs and to assess their incremental value beyond preoperative clinical factors. This retrospective cohort study included 330 patients with ESCC who underwent radical esophagectomy after neoadjuvant chemoimmunotherapy. Early PPCs occurring within 7 days after surgery were the primary outcome. Associations of preoperative SII, NLR, and PLR with early PPCs were evaluated using logistic regression. Model 1 adjusted for clinical stage, age, albumin, hypertension, tumor location, chronic lung disease, and recent smoking exposure, whereas Model 2 additionally adjusted for one-lung ventilation duration, total intraoperative infusion volume, and operative time. Incremental predictive value beyond the preoperative clinical reference model was assessed using changes in area under the receiver operating characteristic curve (AUC), DeLong tests, likelihood-ratio tests, reclassification measures, calibration, and bootstrap internal validation. Early PPCs occurred in 115 of 330 patients (34.85%). Patients with early PPCs had higher preoperative SII, NLR, and PLR levels than those without early PPCs. When analyzed as continuous variables, higher NLR, SII, and PLR remained associated with increased odds of early PPCs in Model 1 (NLR: OR 1.279, P = 0.004; SII per 100-unit increase: OR 1.129, P = 0.001; PLR per 100-unit increase: OR 1.797, P = 0.004), with similar associations in Model 2. Quartile-based associations were more consistent for PLR and SII than for NLR. The preoperative clinical reference model had an AUC of 0.564, which increased to 0.617, 0.625, and 0.623 after addition of NLR, PLR, and SII, respectively. The AUC increases were significant for PLR ( P = 0.027) and SII ( P = 0.038), whereas the comparison for NLR yielded P = 0.050. Bootstrap internal validation indicated some model optimism. Exploratory combined-marker analyses did not demonstrate clear superiority over individual markers. Higher preoperative NLR, SII, and PLR were consistently associated with increased odds of early PPCs in patients with ESCC undergoing esophagectomy after neoadjuvant chemoimmunotherapy. These routinely available inflammatory indices added clinically relevant information beyond conventional preoperative factors, with PLR and SII showing the most consistent associations across analytical approaches. Given their simplicity, accessibility, and low cost, NLR, PLR, and SII may serve as practical adjuncts for identifying patients at increased pulmonary risk and refining perioperative risk stratification.