Objectives: Color Doppler ultrasonography (CDUS) has been proposed as an alternative to portal pressure gradient (PPG) measurement to detect transjugular intrahepatic portosystemic shunt (TIPS) dysfunction but with inconsistent results. This study aimed at developing and validating CDUS criteria to assess TIPS dysfunction.
Methods: A total of 117 consecutive follow-up simultaneous CDUS and hemodynamic evaluations in 34 patients with TIPS were analyzed. TIPS dysfunction was defined as a PPG >12 mmHg. A predictive model was obtained with logistic regression and was validated in an independent, prospective sample of 119 consecutive paired CDUS/hemodynamic evaluations in 55 patients.
Results: TIPS dysfunction was present in 57 of the 117 studies in the retrospective series. At multivariate analysis, mean maximum flow velocity at the portal vein (mVPmax) and direction of flow in the intrahepatic portal vein branches (FD) were the only independent predictors of TIPS dysfunction. The prediction rule for TIPS dysfunction derived from the model (mVPmax <28 cm/s when flow is hepatofugal or mVPmax <39 cm/s when flow is hepatopetal) had 90% sensitivity, 45% specificity, and negative likelihood ratio of 0.23. This prediction rule was validated both in patients with bare stents and in patients with polytetra fluoroethylene (PTFE)-covered stents, showing an overall 87% sensitivity, 57% specificity, and 0.23 negative likelihood ratio.
Conclusions: The combination of two CDUS parameters correlate with TIPS dysfunction with high sensitivity and low specificity but with a good negative likelihood ratio. TIPS catheterization can be safely avoided in half of the patients using this predictive rule.