Diagnostic performance of ultrasound, computed tomography and magnetic resonance imaging in portal vein thrombosis: A systematic review with meta-analysis

Background: Ultrasonography (US), computed tomography (CT), and magnetic resonance imaging (MRI) are commonly used for the diagnosis of portal vein thrombosis (PVT), however, their accuracy compared to invasive tests is not clearly known. Methods: We systematically searched MEDLINE, Embase, Cochrane Central, and Scopus databases for studies assessing the sensitivity and specificity of US, CT, and MRI for PVT diagnosis. Inclusion criteria required at least 10 participants and comparison with invasive gold standards (angiography or surgery/pathology). The quality assessment was performed using the QUADAS-2 tool. Pooled estimates of sensitivity and specificity were calculated through mixed-effects logistic regression model and reported with 95 % confidence intervals (CI). Based on the obtained results, we computed Bayesian models simulating the work-up of PVT in case of different clinical pre-test probabilities of PVT. Results: Out of 8526 identified citations, 26 studies with 1499 patients were included in the meta-analysis. The risk of bias was low in 3 studies. Pooled sensitivities and specificities were 0.84 (95 %CI 0.69-0.92) and 0.96 (95 %CI 0.92-0.98) for US, 0.81 (95 %CI 0.66-0.90) and 0.96 (95 %CI 0.88-0.99) for CT, and 0.81 (95 %CI 0.65-0.90) and 0.98 (95 %CI 0.96-0.99) for MRI. In patients with a pre-test probability of PVT of 30 %, a positive US yields a post-test probability of PVT of 90 % and a second positive imaging result leads to a post-test probability of 99 %. Conclusions: US, CT, and MRI demonstrate comparable accuracy for PVT diagnosis, with a high specificity across modalities. A stepwise diagnostic approach integrating clinical probability and imaging results could optimize diagnostic confidence.