[Brain conventional MR and DWI in cirrhotic patients:correlated with portal hypertension]

To explore the relationship among brain routine magnetic resonance imaging (MRI), diffusion weight imaging (DWI), and parameters of portal hypertension of cirrhosis patients.

Two hundred and seventy-five patients with cirrhosis underwent routine MRI exams were analyzed retrospectively. Signal intensities of putamen and globus pallidus (GP) were measured on T1WI to compute globus pallidus index (GPI). In apparent diffusion coefficient (ADC) images, regions of interest (ROIs) were placed on the white matter of parietal lobe, frontal lobe, occipital lobe, caudate head, putamen, GP, and thalamus to get ADC values of each location. In abdomen CT images, liver areas of all slices were drawn to multiply slice thickness to get liver volumes, and maximum portal venous caliber and portosystemic shunting collateral caliber were recorded. Independent-samples t test was performed to explore the significance of GPI between GP hyperintensity group and GP non-hyperintensity group, as well as between cirrhosis groups underwent 1.5T and 3.0T MR exams. One-way ANOVA test was performed to evaluate differences among Child-Pugh classes. The difference between baseline and follow-up MR exams were compared with paired-samples t test. Pearson correlation test was performed to explore the relationships among GPI, ADC values, parameters of portal hypertension, and venous blood ammonia levels. Spearman rank correlation test was performed to explore the relationships among GPI, ADC values, and Child-Pugh classes.

(1) The GPI in GP hyperintensity group was significantly higher than GP non-hyperintensity group (0.128 ± 0.031 vs 0.057 ± 0.021, t = 22.244, P < 0.01). The ADC values and venous blood ammonia levels showed no significant difference between two groups. GP hyperintensity group showed larger portal venous and collateral calibers than GP non-hyperintensity group ( (18 ± 4)mm vs (17 ± 3) mm, (10 ± 6) mm vs (6 ± 4) mm, t was 2.816 and 5.048 respectively; P was 0.005 and <0.01 respectively). GP hyperintensity group showed less liver volumes than the other group ( (1 113 ± 445)ml vs (14 443 ± 698)ml, t = -4.280, P < 0.01). Child-Pugh classes showed no significant difference among each other (all P > 0.05). (2) The GPI of 1.5T and 3.0T exams showed significant difference between each other (t = -5.15, P < 0.01). Short-term follow-up patients showed no significant difference of GPI between two exams (t = -0.271, P = 0.788). (3) GPI showed negative correlation with liver volumes of cirrhosis patients (r = -0.240;P < 0.01), and GPI positively correlated with portal venous calibers and portosystemic shunting collateral calibers (r = 0.296, 0.355; both P < 0.01). ADC values of parietal lobe, frontal lobe, caudate head, GP, and occipital lobe positively correlated with GPI (r = 0.309, 0.156, 0.174, 0.155, 0.185; all P < 0.05). ADC values of parietal lobe positively correlated with portal venous calibers (r = 0.187, P = 0.025), ADC values of other locations showed no significant correlation with parameters of portal hypertension (all P > 0.05). GPI and thalamus ADC values showed positive correlation with Child-Pugh classes (r = 0.195 and 0.217; both P < 0.05), ADC values of other locations showed no significant correlation with Child-Pugh classes (all P > 0.05).

The GPI of cirrhosis patients was correlated with severity of portal hypertension, and showed significant correlation with ADC values, while ADC values had no relationship with portal hypertension. Besides, venous blood ammonia values and Child-Pugh classes have little effect on GPI and ADC values.