Development and validation of a nomogram for predicting cerebrospinal fluid leak after endoscopic craniopharyngioma resection.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
OBJECTIVE: To identify risk factors for cerebrospinal fluid (CSF) leak after extended endoscopic endonasal surgery for craniopharyngiomas and develop a predictive model for predicting postoperative CSF leak. METHODS: Six hundred and sixty cases of craniopharyngioma (training cohort: n = 462; validation cohort: n = 198) were retrospectively reviewed between October 2018 and May 2024, and relevant risk factors were identified. A nomogram was built using a stepwise logistic regression method based on the Akaike information criterion. The performance of the nomogram was evaluated using area under the curve (AUC), calibration curve, and decision curve analysis. RESULTS: The overall rate of postoperative CSF leak was 4.5%. Higher prognostic nutritional index (PNI) level (OR 0.819, 95% confidence interval [CI] 0.735-0.912; p < 0.001) and larger dural defect (OR 6.789, 95% CI 3.112-14.807; p < 0.001) were identified as independent predictors for postoperative CSF leak in multivariable logistic regression analysis. The AUCs of the nomogram were 0.870 (95% CI, 0.782-0.957; p < 0.001) and 0.842 (95% CI, 0.722-0.963; p < 0.001) in the training and validation sets, respectively. Calibration curves in the training and validation cohorts showed satisfactory agreement between predictive and actual outcomes (p = 0.608 and p = 0.564, respectively). Decision curve analysis further confirmed the clinical usefulness of the nomogram. CONCLUSIONS: Higher PNI levels may help reduce the risk of postoperative CSF leak, while a larger dural defect size was demonstrated as an independent risk factor. We developed and validated a nomogram for predicting CSF leak after endoscopic craniopharyngioma resection, which showed strong predictive performance and could assist clinicians in formulating personalized treatment strategies.