CT-based nomogram for predicting survival after R0 resection in patients with gallbladder cancer: a retrospective multicenter analysis.

Objectives: To establish a prognostic nomogram for patients undergoing R0 resection for gallbladder cancer based on preoperative CT.
Methods: A total of 151 patients (64 males, 87 females; mean age, 73.26 years) with gallbladder cancer who underwent CT and surgery with margin-negative resection were retrospectively collected at two tertiary institutions. The demographic and radiologic parameters were analyzed using univariate and multivariate Cox regression analyses to identify independent prognostic factors. The final CT-based nomogram was constructed to predict prognosis after curative resection of gallbladder cancer. Calibration curves for the survival probabilities were obtained for internal validation.
Results: Mass-forming type (hazard ratio [HR], 28.80), bile duct invasion (HR, 4.76), duodenal invasion (HR, 6.32), colon invasion (HR, 4.37), gallstones (HR, 0.09), and cholecystitis (HR, 2.56) were significant independent predictors for recurrence-free survival (p < .05). Mass-forming type (HR, 8.16, p < .001), bile duct invasion (HR, 2.92, p = .013), duodenal invasion (HR, 3.72, p = .012), and regional lymph node metastasis (HR, 2.07, p = .043) were independent predictors of poor cancer-specific survival (CSS) and were used to construct the nomogram. The nomogram showed a good predictive ability for the probabilities of survival on the calibration curves, and the concordance index of the model in predicting CSS was .768.
Conclusion: Preoperative CT findings could predict the prognosis of gallbladder cancer, and the CT-based nomogram accurately predicted CSS in patients with gallbladder cancer after attempted curative resection.
Key Points: • Among the preoperative imaging features, mass-forming type, bile duct invasion, duodenal invasion, and regional lymph node metastasis were independent predictors of poor cancer-specific survival. • The nomogram constructed using preoperative CT findings showed a good predictive ability for the survival on calibration curves, and the concordance index of the model in predicting cancer-specific survival was 0.768.