| BackgroundThe pathology of nonalcoholic fatty liver disease (NAFLD) ranges from simple steatosis to nonalcoholic steatohepatitis (NASH). It has become the most common chronic liver disease all over the world. Some can develop into cirrhosis even hepatic cell cancer. The prognosis of NASH-related cirrhosis is worse than hepatitis-related cirrhosis. Early diagnosis and prompt treatment of NAFLD is of great importance in clinical practice because it can improve liver fibrosis and prevent cirrhosis.There have been many examinations on the diagnosis and assessment of the severity and the potential treatment options for NAFLD. Liver biopsy is considered to be the reference standard for the diagnosis of NAFLD and NASH. However, liver biopsy is not feasible in all patients with NAFLD in view of the invasiveness of the procedure and its sampling variability. Therefore, an invasive technique which can accurately evaluate the severity of NAFLD and treatment are mostly wanted. Diffusion-weighted imaging (DWI) can detect molecular diffusion in vivo. Intravoxel incoherent motion (IVIM) further separates estimation of microcirculation in the capillaries and in molecular diffusion. CT perfusion (CTP) can provide liver blood flow, blood volume, hepatic arterial perfusion and portal vein perfusion. There have been some studies on cirrhosis using either IVIM or CTP. This study was divided into three parts, to assess the microscopic details about tissue architecture and hemodynamic changes in NAFLD using IVIM and CTP.The objective of the study is to:(1) Investigate the correlation between pseudo diffusion coefficient D*, perfusion fraction f derived from IVIM DWI and BV, BF derived from CTP; Verify the feasibility and accuracy of IVIM DW-MRI in evaluating tissue diffusion and microcirculation. (2) Explore an ideal animal model of NAFLD which is similar with human pathology and proper for imaging assessment. (3) Assess the diagnostic performance of IVIM and CTP parameters in NAFLD and the correlation between IVIM and CTP parameters and severity of NAFLD.Part One The feasibility and correlation of intravoxel incoherent motion diffusion-weighted imaging and CT perfusion in normal rabbit liver1 ObjectiveAssess the diagnostic performance of IVIM and CTP parameters in NAFLD and the correlation between IVIM and CTP parameters and severity of NAFLD.2 Materials and methods2.1 Subjects8 healthy New Zealand white rabbits (weighting 2.3-2.7kg) provided by Guangdong Medical Lab Animal Center2.2 Equipment and reagentsPhilips Achieva 3.0T MRI scannerDedicated rabbit coil manufactured by Shanghai Chenguang medical technologies Co., LTDSiemens Definition AS+multi-slice CTSu Mian Xin Π and 3% pentobarbital sodium solutionIohexol 300 mgl/ml2.3 Methods2.3.1 MRIscanMR sequences consist of conventional transverse T2 weighted imaging (T2WI) and IVIM-DWI. The parameters of transverse T2WI were following:repetition time (TR)/echo time (TE) 365.58/80.00 ms, echo train length (ETL) 16, slice thickness/gap 2.0/0.2 mm, field of view (FOV) 150×150, Matrix 152×146, receiver bandwidth 184 Hz, NEX 6. Multi-b diffusion weighted MR images (b=0,10, 20,30,50,80,100,200,400,800 s/mm2) were acquired using single-shot echo-planar imaging pulse sequence with free-breathing. The parameters were following:TR/TE 1743.36/45.17 ms, slice thickness/gap 3.0/0.3 mm, FOV 130 X 130mm, matrix 64×65, receiver bandwidth was 2994 Hz/pixel, NEX 8, and fat was suppressed using spectral presaturation inversion recovery (SPIR). Total acquisition time was 13 min.2.3.2 CTscanAbdominal plain CT scan was performed after MRI scan. The parameters were following:80 kVp,130 mAs, acquire mode 128 X 0.6mm, pitch 0.6, FOV 150mm, matrix 512×512, reconstruction slice thickness/gap 2.0/0.0 mm, kernel B30f. For perfusion imaging, a Dynamic 4D mode was performed 2 seconds after the intravenous bolus injection of 1 ml/kg of contrast material and continued for 50 cycles (1.5 sec/cycle). The contrast material was injected at a rate of 1 ml/s through the auricular vein using a power injector. The CT perfusion imaging scanning parameters were as follows:dynamic 4D-mode scan with 0.3 second per 360° rotation; 80 kVp; 100 effective mAs; detector configuration 32 X 1.6 mm; FOV 150 mm. The CT scans were then reconstructed into 2 mm slice thickness with no gap; a 512×512 image matrix.2.3.3 Post processing of IVIM dataIVIM data were archived to the workstation (Extended work Space) and ADC map using 10b values were generated. Diffusion coefficient D, pseudo diffusion coefficient D* and perfusion fraction f were calculated with an in-house program developed basing MatLab. ADC and IVIM parameters were measured with Image J software. An oval ROI with size of 0.25 cm2 was drawn in ADC map of liver right lateral lobe, right medial lobe, left medial lobe and left lateral lobe respectively.8 rabbits with 32 ROI were measured. The ROIs were copied to IVIM parameter maps and measured.2.3.4 Post processing of CTP dataCT perfusion data were archived to a 3D workstation (Synngo MMWP). The steps of perfusion calculation include motion correction, define artery, define portal vein and spleen, and generation of perfusion parameters map of maximum intensity projection (MIP), blood flow (BF), blood volume (BV), hepatic arterial liver perfusion (ALP), portal venous liver perfusion (PVP), hepatic perfusion index (HPI). An oval ROI with size of 0.25 cm2 were placed in keeping with IVIM parameter map and measured.2.3.5 Statistical analysisAll measurement data were described as X±S. The reproducibility of CTP and IVIM parameters were tested using intraclass correlation coefficient (ICC),95% confidence interval (CI) and coefficient of variability (CV). The Bland-Altman map was generated. CV≤10% was thought as excellent reproducibility,10-20% was good,20-30% was acceptable,>30% was poor. All the stastical analysis were performed using SPSS 22.0 or GraphPad Prism 6.01. P<0.05 was considered as statistical significant.3 Results3.1 The reproducibility of CTP parametersThe repeated CTP scan showed the results of BF 27.76±4.59 ml/100ml/min and 27.83±4.81 ml/100ml/min, BV 9.20±2.20 ml/100ml and 9.37±2.11 ml/100ml, ALP 15.41±7.98 ml/100ml/min and 19.80±10.65 ml/100ml/min, PVP 86.54±42.30 ml/100ml/min and 73.16±43.45 ml/100ml/min, HPI 18.94±6.42% and 24.04±9.63%。 BF showed the best reproducibility with ICC0.8797 and CV 7.72%; BV showed good reproducibility with ICC 0.7268 and CV 15.03%; Both PVP and HPI showed acceptable reproducibility with ICC 0.9259 and CV 22.93%,0.8166 and 26.80%, respectively; ALP showed poor reproducibility with ICC 0.7940 and CV 35.44%.3.2 The reproducibility of IVIM parametersThe repeated IVIM-DWI scan showed the results of D 0.92±0.18×10-3 mm2/sec and 0.90+0.18×10-3 mm2/sec, D* 34.61±8.50×10-3 mm2/sec and 33.26 ±8.20×10-3mm2/sec,f 30.78±7.68% and 29.09±5.89%。 D showed the best reproducibility with ICC 0.9368 and CV 6.86%; Both D* and f showed good reproducibility with ICC and CV 0.8623 and 12.25%,0.8789 and 11.19%, respectively.3.3 The correlation between CTP and IVIM parametersThe correlation between BF and D*, BV and f were tested using Pearson method. The correlation coefficient were 0.155 and 0.018 with P value 0.222 and 0.886, respectively. There were no significant correlation found between the two groups.4 ConclusionAmong all the CTP parameters, BF showed the best reproducibility, BV showed good reproducibility, the reproducibility of PVP and HPI were acceptable, while ALP showed poor reproducibility. All IVIM parameters showed better reproducibility than CTP parameters, D showed the best reproducibility, D* and f showed good reproducibility.There were no definitely correlation between BF and D*, BF and f value.Part Two The establishment of animal model with nonalcoholic fatty liver disease1 ObjectiveTo establish an ideal animal model of NAFLD which is similar with human pathology and proper for imaging assessment by means of high fat diet. Observe the natural course of NAFLD by the comparison of different time point and severity.2 Materials and methods2.1 Subjects49 healthy male New Zealand rabbits (weighting 2.2-2.5 kg) provided by the animal experiment department of Guangzhou traditional Chinese Medicine University2.2 Equipment and reagentsBeckman Coulter AU5800 Clinical Chemistry SystemSony NEX-5R Compact Interchangeable Lens Digital CameraOlympus BX-53 MicroscopeJEOL JEM 1400 Transmission Electron MicroscopeHigh fat and cholesterol feed2.3 Methods2.3.1 Duplication methods42 rabbits were distributed into 5 groups randomly, which consisted of control (10), high fat feed 4 weeks (HFD-4W), high fat feed 8 weeks (HFD-8W), high fat feed 12 weeks (HFD-12W), high fat feed 16 weeks (HFD-16W).7 rabbits were fed as supplement group. The rabbits of control group were fed with standard diet, while the rabbits of HFD and supplement groups were fed with high fat diet (standard diet +10% lard+2% cholesterol).2.3.2 Observation of general condition2.3.3 Serum biochemical analysisSerum transaminase (ALT and AST) activities, total cholesterol (TC), triglyceride (TG) and total cholesterol (TC)were determined.2.3.4 Pathology2.3.4.1 Gross pathologyThe size, color and quality were observed. The weight of livers was measured. The hepatosomatic index (HIS) was calculated (wet weight of liver/body weight).2.3.4.2 Histological pathologyHistological pathology include Sudan Ⅲ, HE, Masson and Ag.2.3.4.3 Histological scoringThe severity of NAFLD were graded according to NAFLD activity score (NAS)2.3.4.4 Electron microscopy pathologyMitochondria of hepatic cell, hepatic sinusoid, bile canaliculus and perisinusoidal space were observed.2.3.5 Statistical analysisAll measurement data were described as X±S. The homogeneity of variance was tested by Levene method. One way ANOVA were performed homogeneity of variance, otherwise a robust tests of equality of mean was used. The corresponding multiple comparison methods were LSD-t test and Duntte’s T3 test. All the stastical analysis were performed using SPSS 22.0 or GraphPad Prism 6.01. P<0.05 was considered as statistical significant.2.4 Results2.4.1 General conditionFinal cohort consisted of 41 rabbits.9 rabbits were in control group,8 rabbits in each HFD group. The rabbits in control, HFD-4W and HFD-8W were normal, the rabbits in HFD-4W and HFD-8W presented symptoms such as anorexia, less activity, even diarrhea.2.4.2 NAS groupAll the rabbits in control group scored 0,13 rabbits were grouped into Borderline,19 rabbits were grouped into NASH according to the NAS Score system.2.4.3 Serum biochemical resultsALT, AST and TB among control and HFD groups were not significant different (P value were 0.808,0.911 and 0.103, respectively). TG, TC were found significant different among the groups (P value were 0.003 and 0.000). TG, TC of HFD groups were higher than control group (P<0.05), but no significant difference were found among HFD groups (P>0.05).ALT, AST and TB among control and NAFLD groups were not significant different (P value were 0.055,0.277 and 0.342, respectively). TG, TC were found significant different among the groups (P=0.000). TG, TC of NAFLD groups were higher than control group (P<0.05), but no significant difference were found between Borderline and NASH group (P>0.05).2.4.4 Gross pathologyThe normal liver displayed a maroon color and sharp edge. The model liver were enlarged with yellow color.The maximum diameter and HSI were found significant different among the control and HFD groups (P=0.000). The maximum diameter of HFD groups were larger than control group (P<0.05), no significant difference were found among HFD goups (P>0.05). HSI of HFD groups were larger than control group (P<0.05), HIS of HFD-4W were smaller than HFD-12W and HFD-16W (P<0.05). The maximum diameter and HSI were found significant different among the control and NAFLD groups (P=0.000). The maximum diameter and HSI of HFD groups were larger than control group (P<0.05), but no significant difference were found between Borderline and NASH group (P>0.05).2.4.5 Histological pathologyControl group showed normal hepatic lobule structure, the hepatic plate and sinusoid arranged as radial appearance. The cell nucles located in the center of the hepatic cell. Hepatic cells showed no lipid droplets, ballooning or distinct inflammatory cell infiltration. No apparent collagen fibrosis were displayed by Masson trichrome or Methenamine Silver staining.The histological findings of NFALD model included steatosis, ballooning and lobular inflammation. Mild steastosis showed scattered lipid droplets in the hepatic cell. Moderate steastosis showed multiple lipid droplets in the hepatic cess with Ⅰ and Π zone predominance. Severe steastosis showed the cytoplasm of hepatic cell were almost occupied by the large lipid droplet. Ballooning of hepatic cell mostly located in the Π and Ⅲ zone of hepatic acinar, showed swelling of hepatic cells with narrowing of hepatic sinusoid. Lobular inflammation showed scattered inflammation cells infiltration.Liver fibrosis in this study were not common.5 rabbits were scored F1, appeared as perisinusoidal space fibrosis.3 rabbits were scored F2, appeared as fibrosis extended into the portal area.2.4.5.1 Electron microscopy pathologyThe hepatic cell were swelling. Multiple lipid droplet with varying size were found in the cytoplasm. Mitochondria were swelling. Collagen fiber deposition were found in the intercellular and Disse space.2.5 ConclusionThe study successfully duplicate NAFLD model whose pathology is similar with human being. The duplication method of high fat fed has advantages such as easy to perform, low cost and good reproducibility. The model can develop progressively and mimic the pathophysiological change of human. In addition, the rabbit with NAFLD is an ideal animal model for imaging study.Part Three The correlative study of IVIM, CTP and pathology in nonalcoholic fatty liver disease1 ObjectiveTo assess the diagnostic performance of IVIM and CTP parameters in NAFLD and the correlation between IVIM and CTP parameters and severity of NAFLD.2 Materials and methods2.1 SbujectsSee details in Part two2.2 Equipment and reagentsSee details in Part one and Part two2.3 Methods2.3.1 MRI scan and data post processingMRI scan see the details in Part one. The mean value of four lobes was measured.2.3.2 CT scan and data post processingCT scan see the details in Part one. The mean value of four lobes was measured. The liver CT value divided by the erector spinae CT value were calculated.2.3.3 Statistical analysisAll measurement data were described as x±s. The homogeneity of variance was tested by Levene method. One way ANOVA were performed homogeneity of variance, otherwise a robust tests of equality of mean was used. The corresponding multiple comparison methods were LSD-t test and Duntte’s T3 test. The diagnostic performance was tested using receiver operation curve (ROC). A linear correlation and multiple linear regression were applied to study the relation between CTP, IVIM parameters and NAS score. All the stastical analysis were performed using SPSS 22.0 or GraphPad Prism 6.01. P<0.05 was considered as statistical significant.3 Results3.1 The noncontrast CT findings in normal and NAFLD groupLiver CT value, liver/erector spinae CT value ratio were significant different among normal and HFD groups (P=0.000). Liver CT value, liver/erector spinae CT value ratio in HFD-4W、HFD-8W and HFD-16W were lower than normal group (P<0.05), Liver CT value, liver/erector spinae CT value ratio in HFD-16W was lower than HFD-4W、HFD-8W (P<0.05). Liver CT value, liver/erector spinae CT value ratio were significant different among normal and NAFLD groups (P=0.000), furthermore, NASH group< Borderline group< control group (P<0.05).3.2 The CTP findings and parameters comparisonThe time density curve (TDC) of normal rabbit liver showed a fast-rising and plateau pattern, the TDC of HFD group showed a slow-rising and plateau pattern. The TDC of normal rabbit spleen showed a fast-rising and slow washout pattern, the TDC of HFD group showed a slow-rising and plateau pattern.There was no significant difference of ALP among normal and HFD groups (P=0.966). While significant difference were found in BF, BV, PVP and HPI among normal and HFD groups (P value were 0.022,0.049,0.000 and 0.000, respectively). BF and BV of HFD groups were lower than control group (P<0.05). No significant difference were found of BF and BV among HFD groups (P>0.05). The PVP in HFD-4W^ HFD-12W and HFD-16W were significant lower than control group, HPI were higher than control group (P<0.05). PVP and HPI in HFD-8W were not significant different from control group (P>0.05). PVP and HPI of HFD-8W was higher and lower than HFD-16W, respectively (P<0.05).There was no significant difference of ALP among normal and NAFLD groups (P=0.432). While significant difference were found in BF, BV, PVP and HPI among normal and NAFLD groups (P value were 0.017,0.003,0.002 and 0.000, respectively). BV of Borderline group was lower than control, BF and PVP of Borderline and NASH group were lower than control, HPI were significant higher than control group (P<0.05). No significant difference were found in BF, BV, PVP and HPI among Borderline and NASH group (P>0.05).3.3 The IVIM findings and parameters comparisonThe liver of NAFLD group were enlarged. The signal attenuation curve of normal liver displayed a typical bi-exponential pattern. The signal attenuation curve of NAFLD liver displayed a mono-exponential pattern.There were significant difference of D, D*,f and ADC among control and HFD groups (P value were 0.000,0.000,0.008 and 0.002, respectively). D, D*,f of HFD groups were lower than control group (P<0.05), D* of HFD-12W and HFD-16W were lower than HFD-4W (P<0.05). No significant difference were found of D and f among HFD groups (P>0.05). The ADC of HFD-12W was lower than control group (P<0.05).There were significant difference of D, D*, f and ADC among control and NAFLD groups (P value were 0.000,0.000,0.001 and 0.002, respectively). D, D*, f and ADC of NAFLD groups were lower than control group (P<0.05). D and ADC of NASH were lower than Borderline group (P<0.05). No significant difference were found of D* and f between Borderline and NASH group (P>0.05).3.4 The comparison of diagnostic performance among CTP and IVIM parametersD showed the best diagnostic performance with area under curve (AUC) 1.0, followed by f , PVP, HPI, D*, BF, ADC and BF. D, D* and ADC were found be diagnostic in differentiating Borderline from NASH, the AUC were 0.723,0.704 and 0.767, respectively.3.5 The correlation and regression between CTP, IVIM parameters and NAS ScoreThere were no significant correlation between BF, BV, ALP and NAS with correlation coefficient-0.287,-0.186 and 0.132 (P>0.05). The correlation coefficient of PVP and NAS was-0.588, HPI and NAS was 0.594 (P<0.05). The correlation coefficient of D, D*, f, ADC and NAS were-0.627,-0.556,-0.623 and-0.519, respectively.Stepwise multiple linear regression method was used to verify the regression between CTP, IVIM parameters and NAS. Only D and f finally entered the equation. The regression equation was Y=12.730-0.7667X1-0.101X2 (Y is NAS Score, X1 is D, X2 is f) with R2=0.716.4 ConclusionD, D* and f decrease in NAFLD model, which means the impediment of molecular diffusion and microcirculation of liver. BF and BV derived from CTP also decrease in NAFLD. Decrease of PVP with increase of HPI in NAFLD represents the hepatic arterial buffer response common seen in cirrhosis.The diagnostic performance of bi-exponential IVIM parameters are better than mono-exponential parameter ADC. PVP and HPI calculated from CTP also shows good performance. The diffusion coefficient D shows the best diagnostic performance among all the parameters.IVIM parameters especially D and f value can represent the histopathological changes of NAFLD quantitatively. Both of them could serve as a superb imaging biomarker for the diagnosis and clinical outcome of NAFLD. |
PDF Full Text Request