Clinical predictors of postoperative cerebrospinal fluid leakage following endoscopic endonasal surgery for pituitary neuroendocrine tumors: a retrospective cohort study.

OBJECTIVE: Postoperative cerebrospinal fluid leakage (CSFL) remains a clinically relevant complication following endoscopic endonasal surgery (EES) for pituitary neuroendocrine tumors (PitNETs). This study aimed to identify independent clinical predictors of postoperative CSFL and to facilitate perioperative risk stratification. METHODS: We retrospectively analyzed 302 consecutive patients who underwent EES for PitNETs at a single tertiary center. Preoperative anatomical features, tumor characteristics, pathological subtypes, and intraoperative variables were collected. Given the limited number of postoperative CSFL events, we used penalized logistic regression (Firth correction) with a prespecified parsimonious set of predictors to reduce small-sample bias. RESULTS: Penalized multivariable analysis (Firth correction) identified fibrous tumor texture (OR 5.13), strong sellar barrier thickness (> 1 mm; OR 0.09), and intraoperative CSFL grade (Kelly grade 1–2: OR 3.69; Kelly grade 3: OR 34.30) as significant independent predictors. ACTH-secreting pathology demonstrated a trend towards association but did not reach statistical significance (OR 4.76; 95% CI 0.89–25.36; P = 0.068). CONCLUSIONS: Postoperative CSFL following EES for PitNETs is associated with a distinct set of clinical and anatomical risk factors. Recognition of these predictors may support individualized reconstruction and perioperative decision-making within a risk-stratified intraoperative decision algorithm; the proposed algorithm is exploratory and requires external validation.