Computed tomography-based score model/nomogram for predicting technical and midterm outcomes in transjugular intrahepatic portosystemic shunt treatment for symptomatic portal cavernoma.

Background: Transjugular intrahepatic portosystemic shunt (TIPS) may be technically difficult in patients with cavernous transformation of the portal vein (CTPV). Computed tomography (CT) is widely used for assessing the situation of the portal vein and its tributaries before TIPS, and an ultrasound-based Yerdel grading system has been developed, which is deemed useful for liver transplantation. Therefore, we hypothesized that a CT-based CTPV scoring system could be useful for predicting technical and midterm outcomes in TIPS treatment for symptomatic portal cavernoma.

Aim: To investigate the clinical significance of a CT-based score model/nomogram for predicting technical success and midterm outcome in TIPS treatment for symptomatic CTPV.

Methods: Patients with symptomatic CTPV who had undergone TIPS from January 2010 to June 2017 were retrospectively analysed. The CTPV was graded with a score of 1-4 based on contrast-CT imaging findings of the diseased vessel. Outcome measures were technical success rate, stent patency rate, and midterm survival. Cohen's kappa statistic, the Kaplan-Meier and log-rank tests, and uni- and multivariable analyses were performed. A nomogram was constructed and verified by calibration and decision curve analysis.

Results: A total of 76 patients (45 men and 31 women; mean age, 52.3 ± 14.7 years) were enrolled in the study. The inter-reader agreement () of the CTPV score was 0.81. TIPS was successfully placed in 78% of patients (59/76). The independent predictor of technical success was CTPV score (odds ratio [OR] = 5.56, 95% confidence interval [CI]: 3.55-9.67, = 0.002). The independent predictors of primary TIPS patency were CTPV score and splenectomy (OR = 9.22, 95%CI: 4.78-13.45, = 0.009; OR = 4.67, 95%CI: 2.59-7.44, = 0.017). The survival rates differed significantly between the TIPS success and failure groups. The clinical nomogram was made up of patient age, model for end-stage liver disease score, and CTPV score. The calibration curves and decision curve analysis verified the usefulness of the CTPV score-based nomogram for clinical practice.

Conclusion: TIPS should be considered a safe and feasible therapy for patients with symptomatic CTPV. Furthermore, the CT-based score model/nomogram might aid interventional radiologists in therapeutic decision-making.