| Objective:To explore the feasiblity of diffusion weighted imaging based on I VIM mode in the diagnosis of non-alcoholic fatty liver disease(NAFLD) and determination of its severity, using histopathologic results of hepatic steatosis as the gold standard.Materials and methods:1.The establishment of animal models and experimental preparation35 healthy femal SD rats,180-200g in weight, were provided with normal drinking water and food. They were randomly divided into five groups:control group(n=7), fed.with chow diet; and model group(n=28),these rats were fed with high fat high cholesterol diet to establish the model of nonalcoholic fatty liver disease for 4 weeks,12weeks,16weeks after adaptation for 1 week. The diet of the experimental group included 87.5% basic food+10% lard+2% cholesterol+0.5% sodium cholate.The animals were anesthetized by intraperitoneal injection of 10% chloral hydrate(provoded by Sun Yat-sen University) at a dosage of 0.3mL/100g and then placed in the animal coil in a prone position, with liver placed in the center of the coil. The space between the rats and the coil was stuffed, in order to minized the motion of respiration.Our study was approved and supervised by the local Animal Research Committee.2. Magnetic resonance imaging (MRI) scan methodPhilips Archieve 1.5 T (Philips Healthcare, Best, Netherlands) superconducting MRI scanner with its corresponding software and animal coil were used in this study. Conventional MRI sequences included localization,axial T1WI, axial T2WI, axial T2WI fat supression, coronal T2WI fat supression (with SPIR in the abdomen region) and Axial DWI.Axial DWI images acquired by using a free-breathing single-shot spin echo sequence with diffusion gradients applied in three orthogonal directions with the following parameters:TR 1005 ms, TE 72 ms, excitation frequency (NEX) 4, matrix 64 x 66, field of view (FOV) 80-80 mm, slice thickness/gap,9.0/1.5mm) with weighting factors of b 0,10,20,40,80,130,200,400,800s/mm2.3. Image analysis and data processingThe bi-exponential model from an IVIM sequence was expressed by the equation Sb/SO=(1-f)exp(-bD)+ fexp[-b(D*+D)].The mono-exponential model was expressed by the equation Sb= SO exp (-bADC). Diffusion-weighted imaging data were transferred to a computer equipped with a manufacturer-supplied Software (PRIDE DWI Tool, version 1.5, Philips Healthcare) for IVIM and mono-exponential analysis. ROI measurements of IVIM parametric maps were obtained with Image J software (National institute of Health, Bethesda, MD). ROI was placed on the left lobe of each one at the slice of the largest area by two radiologists specializing in abdominal imaging who were blinded to the rats’groups. The parameters were measured for 3 times and the averages were finally calculated as the representative values. Vessels, biliary ducts and areas with severe artifacts by visual inspection were excluded from ROI in order to focus upon viable tissue in deriving IVIM parameters and ADC values. ADC was measured at the same slice with D map, and ROI was manually drawn on the same area as possible.4. Laboratory parametersBlood samples were obtained from hepatic central vein of all rats immediately after data of diffusion weighted imaging were obtained and the rats were sacrificed with excessive dosage of anesthetic. The serum level of alanine transaminase, aspartate transaminase, total bilirubin, triglycerides,total cholesterol, and C-reactive protein were measured.5. Histologic analysisAll tissue specimens were obtained from the left liver lobe, as near the areas of ROI as possible. They were stained with hematoxylin-eosin and Masson Each animal was given an NAFLD activity score, which was the unweighted sum of steatosis (score 0-3), lobular inflammation (score 0-3), and hepatocellular ballooning (score 0-2).The degree of fibrosis was evaluated by using the METAVIR scoring system, which was defiened as F0= no fibrosis, F1= portal fibrosis without septa, F2= portal fibrosis and a few septa, F3= numerous septa without cirrhosis,and F4= cirrhosis. The pathological results were reviewed with NIKON microscope by two independent pathologists who were blinded to the rats’ diet. Based on the NAFLD activity score, each subject was categorized into one of the NAFLD severity groups: normal (NAFLD activity score= 0),NAFL (NAFLD activity score= 1,2),borderline (NAFLD activity score= 3,4), or NASH (NAFLD activity score≥5).6. Statistical analysisThe intraclass correction coefficient(ICC) was calculated to derive the data varability for the 2 different observers. D values were compared with the ADC values with paired sample t test. Pretreatment ADC,D, f,D*values were compared between normal group, NAFL group, borderline group,and NASH group with one-way ANOVA. The homogeneity of variance for multiple comparisons was performed with SNK statistical method, and homogeneity of variance was performed with Dunnett’s T3 method. Measurement data were expressed as x±s, and the level of test a=0.05.SPSS statistical software 19.0 was used for above statistical analysis.ResultsIn the model group, the success rate was 100%. Among 25 rats which served as experimental animals (three of them died of unknown causes), steatosis was successfully observed. A total of 32 rats underwent MRI routine scan and multiple b value DWI scan.1.Pathological resultsThe liver in normal group rats were bright red, with sharp edge, but the liver in experimental rats manifested as tightened liver capsule, with blunt edge, and yellow, greasy, with visible fat particles.Under the optical microscope, the liver cells in normal group were radially arranged around the central vein, as normal liver lobules, with no infiltration of inflammatory cells. The experimental groups demonstrated different degrees of steatosis, with macrovesicular lipid droplet vacuoles orvesicular lipid droplet vacuoles,and hepatocellular ballooning and lobular inflammationliver in some subjects. Cell cords were arranged in disorder, with narrowed liver sinus.All seven animals in the control group were assigned an NAFLD activity score of 0,which is a normal liver histologic result. There were seven animals in the group with normal livers, five in the NAFL group, nine in the borderline group, and eleven in the NASH group. At histologic analysis, no fibrosis was observed in all the groups.2.Laboratory findingsThe results of serum biochemical tests showed that there was obvious abnormality of liver function and lipid disorders in the experimental group.3.The repeatability of parameters between different observersThe ICC of D value between two observers was 0.901 (95% CI:0.934～0.973). The ICC of F value between two observers was 0.78 (95% CI:0.675～0.785). The ICC of D* value between two observers was 0.77 (95% CI:0.821 ～0.877). The ICC of ADC values between two observers was 0.874 (95% CI:0.896～0.931). The repeatability of D, ADC between two observers was good, the best of which was D value, followed by the ADC, f, D*.4. The comparison of D value and ADC value32 rats were measured, andthe D value was(0.11±0.02)×10-3mm2/s, and the ADC value was (0.12±0.02)×10-3 mm2/s. ADC value was higher than D value (t=0.576, P=0.001).5. Comparison of ADC and IVIM Parameters in different NAFLD Severity Groups Pretreatment D value of normal group was (1.30±0.16) × 10-3mm2/s. Pretreatment f value of normal group was (17.06±6.80)%. Pretreatment D* value of normal group was (122.12±25.51)* 10-3mm2/s. Pretreatment ADC value of normal group was (1.35±0.06) × 10-3mm2/s. Pretreatment D value of NAFL group was (1.14±0.23) × 10-3mm2/s. Pretreatment f value of NAFL group was (18.65±8.50)%. Pretreatment D* value of NAFL group was (114.24±46.05) × 10-3mm2/s. Pretreatment ADC value of NAFL group was (1.30±0.35) × 10-3mm2/s Pretreatment D value of borderline group was (1.11±0.26) × 10-3mm2/s. Pretreatment f value of borderline group was (23.98±5.56)%. Pretreatment D* value of borderline group was (106.02±49.21) × 10-3mm2/s. Pretreatment ADC value of borderline group was (1.23±0.19) × 10-3mm2/s. Pretreatment D value of NASH group was (0.98±0.15) × 10-3mm2/s. Pretreatment f value of NASH group was (28.19±5.80)%. Pretreatment D* value of NASH group was (118.54±30.42) × 10-3mm2/s. Pretreatment ADC value of NASH group was (1.14±0.15) × 10-3mm2/s. The pretreatment D value of NASH group was obviously lower than that of normal (P< 0.05). The pretreatment f value of NASH group was higher than that of normal and NAFLD (P< 0.05). Pretreatment D* and ADC value showed no significant difference between normal, NAFL, borderline and NASH group (p=0.851, P= 0.168).Conclusions1.High fat high cholesterol diets could successfully induce NAFLD rat model.2.D derived by IVIM model is significantly lower than ADC derived by mono-exponential model, and the repeatability of D is better than that of ADC. The repeatability of D value is the best, followed by f, D*.3. Perfusion fraction extracted from IVIM diffusion-weighted imaging may help in the differentiation of NASH from simple steatosis. IVIM DWI can be utilized as a noninvasive assessment of nonalcoholic fatty liver. The IVIM model is better than the single index model in diagnostic efficacy. |
PDF Full Text Request