Lisette van Dam

Magnetic resonance thrombus imaging in portal vein thrombosis 6 105 venous thrombosis of the portal vein and at other locations. For instance, portal vein imaging is different from imaging of veins in the extremities and brain, due to the presence of ascites, gas and bowel movements which can hamper abdominal vein imaging. Furthermore, PVT occurs in different clinical circumstances (i.e. portal hypertension, cirrhosis) that are less relevant to typical VTE. 24 There are only a few case-reports available on non-contrast-enhanced MRI for the diagnosis of PVT, using different techniques and without results on the diagnostic accuracy of MRI for PVT. 25,26 In a study from Zirinsky et al, T1- and T2 weighted MR images of 14 patients with acute and chronic PVT on CT or ultrasonography and of 8 patients with portal hypertension but without evidence of PVT were evaluated. With MRI (sub-)acute ( < 5 weeks old) thrombi appeared hyperintense relative to liver and muscle on both T1- and T2 weighted sequences and older thrombi (2- 18 months old) appeared hypertense in some patients, but only on T2-weighted images. Therefore, it was suggested that chronic thrombosis may lose their relative hyperintensity on T1-weighted images and thus may be used for detecting and classification of PVT. 25 In a case-series from Haddad et al, thrombosis on non-contrast-enhanced T1-weighted images appeared different depending on the age of the thrombus, with a high signal intensity in subacute splanchnic vein thrombosis ( < 6 weeks) and low signal intensity in more chronic ( > 2 months old) thrombosis. 26 There are some limitations of the current study. First, this study includes only three patients and therefore, the validity of the results must be evaluated in a large cohort before we can proceed to an outcome study. Such an accuracy study is currently ongoing. Furthermore, the evaluation of MR-NTCI images by the expert panel was not blinded for other imaging studies including ultrasound and CT images. Additionally, due to the small number of patients the interobserver agreement could not be established. In conclusion, we were able to identify two MR-NTCI sequences, 3D T1 TFE and 3D T1 Dixon FFE, that were able to diagnose and differentiate acute from chronic PVT. With our previous experience based on imaging of DVT in the lower and upper extremities 9,11,19 , we believe that the image quality is sufficient to be of clinical value and initiate the clinical part of the Rhea study (NTR 7061) to establish the diagnostic accuracy of MR-NCTI for the differentiation of acute from chronic PVT. Furthermore, the interobserver agreement of these sequences for PVT imaging will be assessed.