Prediction of prognosis and resectability using MR imaging, clinical, and histopathological findings in patients with perihilar cholangiocarcinoma.

PURPOSE

To predict poor overall survival (OS) and risk of residual tumor after surgery using MR imaging, clinical, and histopathological findings in perihilar cholangiocarcinoma.

METHODS

196 patients with perihilar cholangiocarcinoma who underwent preoperative MRI and curative-intent surgery were retrospectively included. MRI findings were assessed by two radiologists. Clinical characteristics and histopathological results such as serum carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA 19-9), T and N stage, and resection status, were also investigated. Cox regression analysis and the Kaplan-Meier method were used to identify prognostic factors for OS. We further analyzed the correlation between MRI features and residual tumors using logistic regression analysis.

RESULTS

The median OS was 25.0 ± 26.6 months. T stage (hazard ratio [HR] 6.26, p = 0.014), N stage (HR 1.86, p = 0.002), CA-19-9 >37 U/mL (HR 2.06, p < 0.001), enlarged LN on MRI (HR 1.69, p = 0.006), and residual tumor (HR 1.52, p = 0.034) were important predictors of poor survival. The 5-year OS of the complete resection group (n = 107) was significantly better than that of the residual tumor group (n = 89) (35.5% vs. 18.8%, p = 0.002). Additionally, peritumoral fat stranding (odds ratio[OR] 2.09, p = 0.027), Bismuth type III/IV (OR 1.95, p = 0.022), and common bile duct (CBD) involvement (OR 2.3, p = 0.008) on MRI were important predictors of residual tumors in univariate analyses. However, absence of peritumoral fat stranding was a significant independent predictor for complete resection (OR 1.99, p = 0.048) and showed the highest sensitivity, at 79.8%.

CONCLUSION

MR imaging, clinical and histopathological results are useful for predicting poor survival after surgery for perihilar cholangiocarcinoma. Furthermore, MRI findings, including peritumoral fat stranding, CBD involvement, and Bismuth type, are important for the prediction of residual tumors.