Predicting postoperative prognosis of pancreatic cancer using a computed tomography-based radio-clinical model: exploring biologic functions.

UNLABELLED

Computed tomography (CT) imaging has the potential to assist in predicting the prognosis and treatment strategies for pancreatic cancer (PC). This study aimed to develop and validate a radio-clinical model based on preoperative multiphase CT assessments to predict the overall survival (OS) of PC and identify differentially expressed genes associated with OS.

METHODS

Patients with PC who had undergone radical pancreatectomy (R0 resection) were divided into development and external validation sets. Independent predictors of OS were identified using Cox regression analyses and included in the nomogram, which was externally validated. The area under the curve was used to measure the model's accuracy in estimating OS probability. RNA sequencing data from The Cancer Genome Atlas were used for gene expression analysis.

RESULTS

In the development and external validation sets, survival was estimated respectively for 132 and 27 patients. Multivariate Cox regression analysis identified 5 independent OS predictors: age (P = .049), sex (P = .001), bilirubin level (P = .005), tumor size (P = .020), and venous invasion (P = .041). These variables were incorporated into the nomogram. Patients were divided into high- and low-risk groups for OS and survival curves showed that all patients in the low-risk group had better OS than that of those in the high-risk group (P < .001). Differentially expressed genes in patients with a poor prognosis were involved in neuroactive ligand-receptor interaction.

CONCLUSION

The radio-clinical model may be clinically useful for successfully predicting PC prognosis.