2023. Neoplastic vs. Bland Portal Vein Thrombosis in Hepatocellular Carcinoma: Quantitative Analysis Based on MR Diffusion-Weighted Imaging
Authors* Denotes Presenting Author
Maimoona Siddique *;
Pakistan Kidney and Liver Institute & Research Centre
LINCS Diagnostics & IR Services
Magnetic resonance (MR) imaging is an emanating potential modality for detecting portal venous thrombosis (PVT). We investigated to establish specific criteria to diagnose PVT in patients with hepatocellular carcinoma (HCC), based on MR diffusion-weighted imaging (DWI). Our aim was to distinguish the bland from neoplastic PVT by doing quantitative analysis between attenuation values on Triphasic Computed Tomography (TP-CT) and Apparent Diffusion Coefficient (ADC) values on MR DWI.
Materials and Methods:
Age and sex-matched, HCC cases with PVT, classified based on Doppler ultrasound and TP-CT imaging-based criteria of PV expansion and venous contrast filling defect, were selected prospectively. The cases underwent unenhanced MR imaging, with a 1.5-T magnet, between April 2019 and October 2019. Quantitative parameters including ADC values for regions of interest drawn over thrombus, background liver parenchyma, and dominant HCC lesion, as well as the ratios of the thrombus to liver and tumor (T:L ADC and T: T ADC, respectively) values, were recorded. Receiver operator curve analysis (ROC) was used to define cut off ADC values to differentiate malignant from benign PVT.
Of the total 47 cases with PVT [32 males, 15 females; mean age 48 years±10.2], the mean Alpha-Fetoprotein level was 2173 ±946 ng/ml. On TP-CT, neovascularity and early arterial enhancement of PVT were depicted in 36/47 cases with malignant PVT and non-depicted in 11/47 cases with benign PVT. PVTs were complete in 40 patients and partial in seven, were in a major portal vein (n=38) or segmental portal vein (n =9). In all cases, the venous contrast filling defects on TP-CT showed correlative signal involving the entire width of the portal vein lumen or its segmental which approximated (with T1 weighting) and exceeded (with T2 weighting) the intensity of the hepatic parenchyma while the hepatic veins showed a complete flow void. There was slightly lower sensitivity for detecting segmental PVT compared with that of major PVT in the malignant PVT cases, possibly related to partial volume effect. The mean ADC value for malignant PVT was 1.14 ± 0.17 × 10-3 mm2 s-1, 1.48 ± 0.28 × 10-3 mm2 s-1 for the liver parenchyma, and 1.08 ± 0.23 × 10-3 mm2 s-1 for HCC lesions; henceforth, T:L ADC was 0.80 ± 0.61 and T: T ADC was 1.05±0.74. For bland PVT, mean ADC value was 1.78 ± 0.17 × 10-3 mm2 s-1, and 1.56 ± 0.42 × 10-3 mm2 s-1 for liver parenchyma, and 1.04 ± 0.38 × 10-3 mm2 s-1 for HCCs; T: L ADC ratio was 1.14 ± 0.40 and T: T ADC was 1.71 ± 0.45. There was a statistically significant [P value < 0.05] difference between ADC and ADC ratio values of the benign versus malignant PVT. ROC curve revealed cutoff value (?1), with sensitivity (100%), and specificity (84%).
MR DWI has an excellent diagnostic yield in the differentiation of malignant from benign PVT in HCC. However, the sensitivity is limited in cases with segmental PVT, particularly in cases with curative intent. Non-contrast MR can serve as a better substitute for equivocal cases on dynamic CT imaging, or those with a history of iodinated contrast allergy.