Experimental Study Of Quantitative Diagnosis Of Pathological Changes In Rabbit Liver Fibrosis Model By CT Perfusion Imaging And MR Diffusion Weighted Imaging

Objectives1. To study and explore the influence of using different contrast agent dosages andinjection rates on the perfusion parameters, and also, whether there were any markedperfusion parameter differences among various slices as well as the left and rightlobes on the same slice in hepatic CT perfusion (HCTP).2. To research and investigate the effect of using various b-values and TR times, aswell as changes in tissue micro-circulation perfusion on ADC values in liver MRimaging with diffusion-weighted imaging (DWI)technique. Furthermore, whetherthere were any differences of measured ADC values among various slices as well asthe left and right lobes on the same slice.3. On the bases the pathological findings and diagnosis of the rabbit liver fibrosismodels (LFM) as well as from the three aspects (pathological scoring and grading,staging as well as the fibrosis severity of the LFM respectively), the correlationbetween the variations of CT perfusion parameter as well as the T1 and T2 values ofMRI with the liver cirrhotic changes of LFMs and their pathological backgroundswere studied.4. All the imaging parameters, the results of liver function tests and the 4 indicatortest for liver fibrosis, as well as those of the pathological scoring and grading, liverfibrosis staging and severity grading were statistically and correlatively studied to find out the correlation between the variations of imaging parameters with the relativelaboratory results and quantitative diagnostic histopathological findings. Thefeasibility and potentials of CTP as well as MR-DWI in the evaluation of liverfunction impairments and the severity of liver fibrosis were also explored.5. Using multiple linear regression analysis, the quantitative diagnostic parameterssuch as the pathological scoring and grading of the LFM, liver fibrosis severityscoring, fibrosis staging were statistically analyzed and the Fisher's lineardiscriminant functions were also established to evaluate: (1) the accuracy of CTP andMR-DWI in the quantitative diagnosis of liver fibrosis; (2) the possibility of thecombination of multi-imaging modalities to non-invasively and quantitativelydiagnose liver fibrosis.Materials and Methods1. Twenty New Zealand white rabbits were randomly divided into three groups. Fourrabbits were designated as normal control group, and, eight rabbits were designated asexperimental groups A and B with four rabbits each respectively. Rabbits of theexperimental groups were repetitively and subcutaneously injected with carbontetrachloride for 8 weeks in group A and 12 weeks in group B. Rabbit liver fibrosismodels (LFMs) were thus made in experimental groups..2. CT perfusion imaging (CTP) was performed on a GE Lightspeed 16-sliced CTscanner with an axial scan mode. Scan technique parameters were as follows:eight-row four-slices, slice thickness 2.5mm, 80kV, 120mA and scan time 1s with 1sintervals. Scanning started at 2s after the injection of contrast agents; continuousscans were performed 45 times, a total scan time was 90s, and the total number ofimages was 180. For the study of the influence of perfusion factors on the variation ofCTP parameters, nine sets of different perfusion factors were designed and integratedby the combination of three different injection rates (1.0ml/s, 1.5 ml/s and 2.0 ml/srespectively) and three different dosages of contrast agents (1.0ml/kg body weight,1.5 ml/kg body weight and 2.0 ml/kg body weight respectively), and wereindividually administered to the rabbits of the normal control group. Experimentalfactors for the CTP groups were set as: injection rate 2.0 ml/s and the dosage ofcontrast agent 1.0ml/kg body weight. The CTP parameters were measured on theAW4.2 workstation and the software of Perfusion 3 liver tumor perfusion mode was used. The abdominal aortic artery was selected as the input artery and portal veintrunk or its main branches as input veins. All parameters were individually measuredfor 5 times at the same sites in the left and right lobes of the liver, and the averageswere counted as the final results. Each chosen ROI should not be less than 60mm2in size. Three slices were randomly chosen as measured objectives.3. Conventional MR scan was performed with a MAGNETOM VISION (Siemens)1.5T MRI system and head coil. Five slices with SE sequence, slice thickness 5mmand 512×512 matrix were made. T2-values were measured on the dual-echo sequencescans (TR=600ms, TE=20ms/80ms), from which two images of the same section withdifferent echo times were obtained. An Evaluation T2 Map was processed andgenerated by the computer, then, T2 values were respectively measured in the rightand left lobes of the liver. For the measurement of T1 values, SE scans of the liver,using TR=300ms and TE=15ms, were first performed, and then, another set of scansof the liver were rescanned once more using the same technical parameters except forchanging the TR time to 3000ms. An Evaluation T1 Map was created fromprocessing the two images of same section by the system computer softwares, and, T1values were respectively measured on the T1 value map. The method and sites ofmeasurement were similar to those measured in CT. When plain scans wereperformed, T1WI contrast enhancing scans were repeatedly undergone with a dosageof 0.2ml/kg Gd-DTPA at 30 and 60 min after the injection of contrast media. Usingthe same method as stated above, the T1 values at 30 (T1/30) and 60 (T1/60) minpostcontrastly were respectively measured. Then, the deduction of plain T1 minusT1/30 and plain T1 minus T1/60 were calculated. Magnetization rate at 30 minpostcontrastly (T1/30-P) was created from plain T1 deduced by T1/30 and thendivided by plain T1; similarly, magnetization rate at 60 min postcontrastly (T1/60-P)was derived from plain T1 minus T1/60 and then divided by plain T1.4. MR Diffusion-weighted imaging (MR-DWI) scan was performed with a SONATAmaestro class 1.5T (Siemens) MR scanners, body phased-array coil and parallelacquisition techniques were applied. The scanning techniques were as follows:spin-echo single-shot echo-planar sequence, five slices per scan, 5mm slice thicknessand lmm gaps, eight times acquisition, matrix 128×128, gradient field 40m/Tm andeffective bandwidth 1502Hz. Two different sequences were used for MR-DWI: Sequence one, using TR=3500ms, TE=76ms. b=0/200/500s/mm2 and scanning time3min19s; Sequence two, TR=4500ms, TE=78ms, b=0/300/600s/mm2 and 4min18sscanning time. After the scans were completed, four different ADC maps were createdby the system through automatic calculation. The ADC-values were measured oneach ADC map, those were ADC11 (b=200s/mm2), ADC12 (b=500s/mm2), ADC21(b=300s/mm2) and ADC22 (b=600s/mm2). The method of measuring ADC-valueson each slice and both lobes of liver was similar to that stated before.5. The animals were sacrificed within a week after complete CT and MRexaminations. Blood samples were acquired from the central artery of rabbit's ear forliver biochemical function tests and liver fibrosis four-indicator tests. After a generalobservation of the configuration and color, the liver was dissected at the leve ofgallbladder and two pieces of liver tissue specimens with an approximate size of0.3cm×0.3cm×1.0cm were obtained from each lobe of the liver. The sites of the liverspecimens taken were required to be as much as consistent to those of the imagingparameters measured. Tissue slices of the specimens were made with the stains ofconventional H&E as well as for fibrosis (Gomori's dyeing). A professor ofhepatology (who did not know any details of the study) was asked to individuallyevaluate the histopathological slices and make diagnoses for each rabbit on hepaticinflammation gradings as well as fibrosis stagings. The pathological slices of eachanimal were independently examined and then scored and staged according to therelative findings and referring to the internationally approved criteria onhistopathologic scoring and staging system, ISHAK. Results of the liver specimen ofthe left and right lobes in the same rabbit were obtained from individual evaluationand independently and statistically counted. The scoring system for the evaluation ofhepatic inflammatory process was ranged from 1 to 18 scores, while the one forfibrosis staging was ranged from StageⅠtoⅥ. Every time when the animals from theexperimental group were sacrificed, a rabbit from the control group wassimultaneously killed as a control animal.6. The study and analysis of our research were made from the three aspects inaccordance with the results of histopathological diagnoses, pathological gradingscores (scoring 1-6 marks designated as class A, 7-12 marks as class B, 13-18 marksas class C), fibrosis staging (stageⅠ-Ⅵ), fibrosis severity (stageⅠandⅡfor mild fibrosis, stageⅢandⅣfor moderate fibrosis and stageⅤandⅥfor severe fibrosis).The research contents involved the correlation of the variations of the parametersmeasured from CTP and MRI with the results of hepatic function tests, hepaticfibrosis four-indicator test and the quantitative diagnosis of histopathology in thethree aspects. The study of the correlation of CTP parameters with T1-, T2-values andADC-values were also included.7. For statistical analysis, statistical package of SPSS 13.0 and one-way analysis ofvariance (Analysis of variance, ANOVA) were used. P＜0.05 indicated that thedifference was statistically significant. Multiple stepwise linear regression analysismethod was used for the discrimination analysis of CT perfusion parameters, MRimaging parameters (T1-, T2-values and ADC values) in the quantitative diagnosis ofhepatic pathological changes of LFMs. Fisher's quantitative diagnostic discriminantfunctions were then established.Results1.3 of the 16 rabbits in the experimental groups were accidentally dead, and theresearch experiments were accomplished with the remaining 13 rabbits. Theexperimental livers looked pale and grey in color, with white and thin membranouscoatings as well as millet-like nodules on the surface of the livers, the rims of thelivers became blunt and stiff livers were found in gross inspection of the experimentallivers. Histopathologically, degeneration and necrosis of the hepatocytes, and, variousextent of inflammatory process as well as collagen fiber deposits were seen in theextracellular spaces of the lesions. The pathological hepatic findings of the 13experimental rabbits were all compatible to the diagnostic criteria of histopathologyof liver fibrosis. It was found that the pathological diagnoses of fibrosis severitygrading were not unanimous between the two different hepatic lobes in four rabbits.2. The results of the normal rabbit liver CTP study found that the measured results ofBF, BV, BFA, BFP and MTT were significantly different in case of various injectionrates and dosages of contrast agents were used. The values of BF, BV, BFA and BFPelevated as the injection rates and dosages of contrast agent increased, while thevalues of MTT were found descending in case that the injection rate was accelerated.Statistically, the values of HAl showed no significant differences when the CTPtechnical parameters were changed. The variations of the measured results of BV and MTT values between both lobes of the liver were of great significance, the BV andMTT values of the left lobe were higher than those of the right lobe. The differencesof measured results of all perfusion parameters, in different slices, revealed nostatistical significance3. Differences in ADC-values of MR-DWI were of great statistical significance incase that different scanning parameters and modes were used. When high b-valueswere used, the measured ADC-values were significantly lower than those when lowerb-values were applied. The ADC-values would be shown to increase when the TRtimes were prolonged, however, the ADC-values were not affected by TR timeprolongation in case that high b-values were used.4. In the study of the results of CTP from the LFMs and normal rabbit livers, it wasfound that the measured results of the values of BF, BV, HAI and BFP in LFMs weresignificantly lower than those in the normal control group, and statistically, there weresignificantly differences in the results of pathological diagnosis. In pathologicalscoring and grading, the values of BF, BV and BFP were gradually declined as thepathological scoring and grading were going up, and, as soon as the values decreasedmore remarkable, the scoring and grading ascended synchronously. However, theHAI and BFA values were significantly declined in grade A and B, while they beganto increase in grade C. in fibrosis staging and sub-staging, the values of BF, BV,BFA, BFP and HAI were gradually declined as the staging and sub-staging werepromoted, and the HAI and BFA values showed no increase in the late stages.5. In the results of conventional MR scans, it was showed that T1- and T2-valueswere significantly higher in LFMs than those in the normal rabbits, T2-valuesobviously increased as the pathological changes deteriorated, and the T1-valuessignificantly prolonged in the early stage, while slightly shortened in later stages. Thedifferences of T1-values in the post-contrast scans between the experimental andcontrol rabbits were of no statistical significance.6. In the study of the MR-DWI results of LFM, it was found that the four ADC- andBP-values of LFMs were significantly lower than those of the normal rabbits, andtheir differences were of great significance. The ADC- and BP-values declined moresignificantly along with the progress of the pathological changes.7. Using statistical correlation analysis methods, the study of the experimental results showed that the T2-, T1-, ADC- and BP-values revealed a significant correlation withthose of the liver function and hepatic fibrosis four-indicator tests. There werenegative correlation between the ADC-values and the values of HAI and BFA,however, there were no correlation between the ADC-values and the values of BF, BV,BFP, PS. There were significant negative correlation between the parameters of ADC,BF, BV, BFA, BFP with the results of pathological scoring and grading, as well asfibrosis staging (P＜0.01). There were significant positive correlation between theT2-values with the results of pathological scoring and grading as well as fibrosisstaging (P＜0.01).8. Using multiple linear regression analysis and stepwise discriminant analysis for thestudy of the CTP and MR parameters, the results of liver function tests as well as theresults of hepatic fibrosis four-indicator test in the evaluation of pathologicalquantitative diagnosis of LFM, Fisher's discriminant functions were established. Theaccuracy ratios of pathological quantitative diagnosis by CTP and MRI for grading,fibrosis staging and fibrosis severity were as follows: for CTP: 52.0％, 54.1％,60.2％respectively; for MRI: 75.6％, 57.8％, 65.6％respectively. The pathologicalquantitative diagnostic accuracy ratios for the LFMs increased to 86.0％, 83.7％,87.2％respectively, in case that the combination of CTP and MR imaging was applied.The accuracy ratios of the pathological quantitative diagnosis, of LFMs, i.e, thepathological scoring, grading, fibrosis staging and severity, improved to 100.0％,95.3％, 96.5％respectively, if the evaluation was made by the result integration fromthe four methods(CTP, MR imaging, liver biochemical function and hepatic fibrosisfour-indicator tests).Conclusions1. A rabbit liver fibrosis model could be induced with repetitive, subcutaneousinjections of carbon tetrachloride, the hepatic histopathological changes of whichwere very similar to those of human liver fibrosis after hepatitis. This model wasextraordinarily suitable for the correlative study of histopathology and imaging inliver fibrosis.2. The values of BE BV, BFA, BFP and HAl of the LFMs were all markedly lowerthan those of the normal rabbits. These values decreased gradually as thedeterioration of cirrhotic changes of the liver was going on. The values of HAI and BFA would be increasing in late stage of fibrosis, and, these were the tendency andsignals of deterioration of liver cirrhotic changes as well as the transformation intoliver fibrosis. The pathological backgrounds of all these changes were: in case of liverfibrosis, the hepatocytes became degenerated and swelling, there were lots ofcollagen fibers precipitating in the extracellular spaces, these led to the stenoses of theintrahepatic capillary networks. As a result, the intralumen pressure and resistance ofthe blood vessels increased and decrease of hepatic blood flow and perfusionoccurred. In the late stage of fibrosis, as a buffer, the volume of blood fromportalvein decreased, capillary proliferation showed in the hepatic sinuses, and finally,the volume of blood supply from the hepatic artery began to increase.3. When liver fibrosis occurred, T2 and T1 values increased, ADC and BP valuesdecreased. In the case that the pathological changes of liver fibrosis was deteriorating,the increase of T2 values and decrease of ADC values were more noticeable. Thesefacts indicated the interpretation that the swelling and fatty degeneration ofhepatocytes as well as the inflammatory reaction resulted in the increase of the valuesof T1 and T2 of the liver parenchyma. Determinant elements of the variation ofADC-values were water contents and the extent of its freedom in extracelluar spaceof the liver, the changes of swelling hepatocytes and large amounts of collagen fibersprecipitating in the extracellular spaces led to the reduction of water contents as wellas the restriction of water freedom. These were the reasons why the ADC valuesdecreased in the case of liver cirrhotic changes and deterioration of the pathologicalprocess.4. T2-, T1- and ADC values showed good correlation with the results of liverbiochemical function and hepatic fibrosis four-indicator tests, as well as the findingsin pathological quantitative diagnosis, including scoring, grading, fibrosis staging andfibrosis severity of the LFMs. Also, there were good correlation between BF, BV,BFA, BFP with the results of pathological quantitative diagnosis, such as scoring,grading, fibrosis staging and severity of the LFMs. These evidences indicated that theCTP and MR imaging parameters were capable of reflecting the severity of liverfunction damage and the pathological changes of the hepatic fibrosis in LFMs.Therefore, CTP and MR imaging possessed the ability of quantitative diagnosis forhepatic fibrosis. 5. Using the method of multiple linear regression analysis and stepwise discriminantanalysis, the CTP and MR parameters, the results of liver biochemical function andhepatic fibrosis four indicator tests of the LFMs were processed. Then, a Fisher'sdiscriminant function was established. The results were showed that the accuracyratio of pathological quantitative diagnosis with the techniques of one imagingmodality was not satisfactory enough. If the combination of two imaging modalities(CTP and MR imaging) was used, the accuracy ratio would be enhanced and therequirements of clinical diagnosis could be well fulfilled. The diagnostic effect ofpathological quantitative diagnosis of hepatic fibrosis could be remarkably improvedin case that the diagnostic informations from the four examination modalities (CTP,MR imaging, liver biochemical function and fibrosis four-indicator tests) werecollected and integrally evaluated; and so, the invasive liver biopsies might probablybe replaced. In conclusion, it was shown that the possibility of pathologicalquantitative diagnosis of liver fibrosis with non-invasive examination modalitiescould be really accomplished.