Jing Feng, Juan Zhang
This internally validated dynamic perioperative nomogram showed preliminary discrimination and calibration for estimating major PPC risk after RC within the study cohort. Because intraoperative net fluid balance is required, the model is intended for risk updating near the end of surgery rather than purely preoperative assessment. It may provide a preliminary basis for planning postoperative surveillance and respiratory care, but external validation in larger independent cohorts is required before clinical implementation.
BACKGROUND: Radical cystectomy (RC) is associated with a considerable risk of major postoperative pulmonary complications (PPCs). Generic surgical risk models may not adequately account for the procedure-specific characteristics of RC. This study aimed to develop and internally validate a dynamic perioperative nomogram that combines preoperative characteristics with an intraoperative variable to estimate the individualized risk of major PPCs following RC for bladder cancer.
METHODS: A retrospective cohort study was conducted on 456 patients who underwent RC between January 2020 and June 2025. The least absolute shrinkage and selection operator (LASSO) regression and multivariable logistic regression analyses were utilized to screen independent perioperative predictors. These robust predictors were subsequently incorporated to construct a visual nomogram. Model performance was comprehensively evaluated in terms of discrimination (Harrell's C-index), calibration, and clinical utility via decision curve analysis (DCA). Internal validation was performed using 1,000 bootstrap resamples to assess optimism-corrected performance.
RESULTS: Among the 456 included patients, 32 (7.0%) developed major PPCs within 30 days postoperatively. Six independent predictors were identified and integrated into the nomogram: age, smoking history, American Society of Anesthesiologists (ASA) physical status, preoperative serum albumin level, forced expiratory volume in one second percentage of predicted value (FEV1%pred), and intraoperative net fluid balance. The nomogram demonstrated excellent discrimination with an apparent C-index of 0.824 [95% confidence interval (CI): 0.758-0.890] and an optimism-corrected C-index of 0.811. The bootstrap-corrected calibration curve showed high concordance between predicted and observed probabilities (Hosmer-Lemeshow P=0.60). DCA suggested a positive net benefit relative to the treat-all and treat-none strategies across threshold probabilities of 4% to 65% within the study cohort.
CONCLUSIONS: This internally validated dynamic perioperative nomogram showed preliminary discrimination and calibration for estimating major PPC risk after RC within the study cohort. Because intraoperative net fluid balance is required, the model is intended for risk updating near the end of surgery rather than purely preoperative assessment. It may provide a preliminary basis for planning postoperative surveillance and respiratory care, but external validation in larger independent cohorts is required before clinical implementation.