Nomograms to predict long-term survival for patients with gallbladder carcinoma after resection

Abstract

Background: Surgical resection remains the primary treatment option for gallbladder carcinoma (GBC). However, there is a pressing demand for prognostic tools that can refine patients' treatment choices and tailor personalized therapies accordingly.

Aims: The nomograms were constructed using the data of a training cohort (n = 378) of GBC patients at Eastern Hepatobiliary Surgery Hospital (EHBH) between 2008 and 2018. The model's performance was validated in GBC patients (n = 108) at Guangzhou Centre from 2007 to 2018.

Methods and results: The 5-year overall survival (OS) rate in the training cohort was 24.4%. Multivariate analyses were performed using preoperative and postoperative data to identify independent predictors of OS. These predictors were then incorporated into preoperative and postoperative nomograms, respectively. The C-index of the preoperative nomogram was 0.661 (95% CI, 0.627 to 0.694) for OS prediction and correctly delineated four subgroups (5-year OS rates: 48.1%, 19.0%, 15.6%, and 8.1%, p < 0.001). The C-index of the postoperative nomogram was 0.778 (95%CI, 0.756 -0.800). Furthermore, this nomogram was superior to the 8th TNM system in both C-index and the net benefit on decision curve analysis. The results were externally validated.

Conclusion: The two nomograms showed an optimally prognostic prediction in GBC patients after curative-intent resection.

Keywords: TNM system; decision curve analysis; gallbladder carcinoma; nomogram; prognosis; resection.

FIGURE 1

Nomogram for predicting the OS of gallbladder carcinoma patients after resection. The individual patient's value is initially identified on the variable axis and a vertical line is drawn from the value to identify the number of points for the variable value. The sum of these is found on the Total Points axis and a downward line from this value indicates the position of the predicted 3 or 5‐year survival rate. (A) The pre‐operative nomogram. (B) The post‐operative nomogram.

FIGURE 2

Calibration curves for OS prediction by the pre‐operative nomogram. (A and B): 3‐ and 5‐ year postoperative OS in the training cohort. (C and D) E: 3‐ and 5‐ year postoperative OS in the external validation cohort. The x axis represents the nomogram prediction; the y axis indicates the actual OS.

FIGURE 3

Calibration curves for OS prediction by the post‐operative nomogram. (A and B): 3‐ and 5‐ year postoperative OS in the training cohort. (C and D): 3‐ and 5‐ year postoperative OS in the external validation cohort. The x‐axis represents the nomogram prediction; the y axis indicates the actual OS.