| 1.Purpose:1.1 To characterize intravoxel incoherent motion(IVIM)parameters in different regions of interest(ROIs)of rabbit VX2 liver tumors during follow-up 4 weeks.1.2 To investigate the value of different quantitative models of diffusion-weighted multiparametric imaging(DW-MPI),including intravoxel incoherent motion(IVIM)model,diffusion kurtosis imaging(DKI)model and stretched-exponential model(SEM)for monitoring the longitudinal parameters changes of rabbit liver VX2 carcinoma and for correlating the perfusion-related imaging parameters to vascularity of tumor tissue.2.Materials and Methods:This study was approved by the institutional animal care and use committee.2.1 Animal ModelEighteen(male=11 and female=7)New Zealand White rabbits weighing between 2.5 and 3.5 kg were used.Prior to tumor implantation,the VX2 tumor strain was transplanted into the hind thigh muscles of tumor carrier rabbits.The study rabbit was sedated with an intramuscular injection of 2 mg per kilogram of Xylazine hydrochloride(Sumianxin II,Jilin,China)and intravenous injection of 30 mg per kilogram of sodium pentobarbital.After preparing and draping the abdomen in standard surgical fashion,a midline abdominal incision was made and an approximately 5-mm tunnel in the left medial lobe of the liver was made and approximately 1-mm3 tumor fragments of were placed through the tunnel and gently pushed into the liver parenchyma.Eighteen separate tumor fragments were implanted into the livers of 18 rabbits.VX2 liver tumors were incubated for 12-14 days after tumor implantation.2.2 MRI AcquisitionPrior to MR imaging,all rabbits fasted for 4-6 hours with free access to water.MR imaging was performed in the supine position,using a 1.5-T MR imaging system(Achieva 1.5 T,Philips Healthcare,Best,Netherlands)using a specialized rabbit coil.The animals were sedated with Xylazine hydrochloride and sodium pentobarbital.After routine localization images were obtained,transverse T2-weighted fast spin-echo images(repetition time msec/echo time msec,2100/100;section thickness,2.5 mm;matrix,148 × 115)and T1-weighted images using a gradient-echo sequence(410/10;section thickness,2.5 mm;matrix,148 x 120)were acquired.Subsequently,DW images were obtained by using a free-breathing single-shot echo-planar imaging pulse sequence with diffusion gradients applied in three orthogonal directions with the following parameters:TR/TE,～1310/65 ms;section thickness,3 mm;field of view,10 × 10 cm;matrix,68 × 64;number of excitations(NEX),6;and 2 sets of b-factors were performed for different diffusion models.The first set contained 12 b values(b= 0,10,20,30,40,50,75,100,150,300,500,and 800 s/mm2)for ADC and IVIM analysis and the second one contained 6 b values(b= 0,400,800,1200,1600,and 2000 s/mm2)for DKI and SEM analysis.2.3 Post-processing and data analysisDW imaging data quantification and analysis were conducted by using a manufacturer-supplied software(PRIDE DWI Tool,version 1.5,Philips Healthcare)and workstation(Extended Workspace,Philips Healthcare)to extract the ADC and non-Gaussian diffusion models parameters,including the true diffusion coefficient(D),the pseudodiffusion coefficient(D*),the perfusion fraction(f),the diffusion coefficient(DDKI),diffusional kurtosis(KDKI),the distributed diffusion coefficient(DSEM)and the stretching parameter(αSEM).ADCs were automatically calculated by using all b values with a monoexponential fit.The IVIM parameters were calculated based on the following equation[5]:S(b)/S(b=0)=(1-f)· exp(-bD)+ f exp(-bD*),where S(b=0)is the mean signal intensity of the region of interest(ROI)for a b value of 0,and S(b)is the signal intensity for higher b values.DKI and SEM parameters were counted based on the following 2 equation:S(b)=S(b=0)exp(-b DDKI + b2 DDKI2·KDKI/6)[6],and S(b)= S(b=0)exp(-b DSEM)αSEM[7],respectively.All parametric maps were created on a pixel-by-pixel basis for each animal.2.4 ROIsTwo radiologists(H.W.and M.H.,with seven and sixteen years of experience in liver MR,respectively)jointly reviewed the 18 animals MR examinations.Image J software(National institute of Health,Bethesda,MD)was used to manually draw ROIs on each lesion at the level of maximum transverse diameter of lesions.Conventional T2-weighted MR images and DW images(b=150s/mm2)were used as references to determine lesion areas on corresponding ADC maps and IVIM parametric maps(Fig 1).The area that visibly had the most restricted diffusion(MRDA)on the ADC map was also determined.Three free-hand ROIs were drawn on the ADC maps,including the whole lesion,the MRDA and peripheral area of the lesion(Fig 2).The ROIs were then copied from the ADC map and placed on the other parametric maps for D,D*,and f.The average of the three ROIs measurements for each parameter was used as a representative value.For the current data analysis,the PRIDE software will calculate the goodness of fit,R2,pixel by pixel.R2 values of ROIs collected were confirmed falling within the scope of 0.8 and 0.9.The data would be excluded and not be involved in statistical analysis if the R2 values were less than 0.8.2.5 Histologic EvaluationAll 18 animals were euthanized after MR imaging.After euthanasia,the livers were harvested for necropsy.Liver specimens were fixed in a 10%buffered formaldehyde solution,sliced at the same MR section thickness(3 mm)through and parallel to the maximum section of tumors,and embedded in paraffin for histologic examination.Histologic slices were stained with hematoxylin-eosin(H-E)and immunohistochemical staining for the specific endothelial antigen CD34(a monoclonal mouse antibody against CD34,DAKO,Denmark)to determine the mean vessel density(MVD)of the tumor.First,each slide was imaged at low magnification(×40)to identify hot spots(where MVD was highest.Then counting the CD34-stained microvessels at high magnifiation(×200)and the mean of five measurements in the hot spots was used as MVD.Histologic slices were digitized with an optical magnification image acquisition system(Nikon,Japan).These specimens were consulted with a consultant pathologist(Y.L.,with more than 20 years experience in gastrointestinal oncology).Intensive tumor cell area,sparse cell area and necrosis were evaluated.2.6 Statistical AnalysisA P value of less than.05 was considered indicative of a statistically significant difference.Statistical analyses were performed by using SPSS(version 20.0;SPSS,Chicago,Ill)and Medcalc software.2.6.1 ReproducibilityTen rabbits were selected at random and the measurement of peripheral area at any time points was calculated to evaluate the reproducibility of ADC and IVIM parameters.The intraclass correlation coefficient(ICC)was calculated to derive the data variability for the 2 observers.All 18 rabbits at week 3 were chosen and the measurements of eight parameters were calculated to evaluate the reproducibility.The intraclass correlation coefficient(ICC)was calculated to derive the data variability for the 2 observers.The reproducibility of measurements was also evaluated with the Bland-Altman method.The mean absolute difference(bias)and the 95%confidence interval of the mean difference(limits of agreement,LOAs)between the first and second DW imaging series were compared.2.6.2 Serial Measurements of ADC and IVIM ParametersFor the different ROIs of tumors(n = 18),serial changes in ADC and IVIM parameters at different time points were compared using the Friedman test.In cases of statistical significance,further comparisons were performed by using the Wilcoxon signed rank test.In addition,to determine whether there were differences in ADC and IVIM parameters in three ROIs of tumors at each time point,the Kruskal-Wallis test or the Mann-Whitney test comparison was used.2.6.3 Serial Measurements of DW-MPI ParametersSerial changes of all eight parameters at three time points were compared by using the Friedman test.In cases of statistical significance,further comparisons were performed by using the Wilcoxon signed rank test.Moreover,to determine whether there were differences in diffusion-related parameters,ADC D,DDkI,and DSEM,at each time point,the Kruskal-Wallis test or the Mann-Whitney test comparison was used.2.6.4 CorrelationsTo analyze the ability of each imaging parameter to predict the vascularity of the tumor,univariate linear regression analysis was performed with ADC,fand D*from week 2 to week 4 as predictive variables and MVD as response variables.A multivariate linear regression was performed to identify whether the linear combination of these variables could improve the prediction of MVD.In this analysis only predictive variables showing highest correlation in the univariate analysis were included resulting in the following linear models:MVD = β0 +β1 ADC +β2f+β3 D*The betas in equation are the estimated regression coefficients.For each regression analysis the correlation coefficient r2 was determined.3.RESULTS3.1 Reproducibility3.1.1 Reproducibility of ADC and IVIM ParametersIn ten rabbits that underwent repeat MR measurements,ADC,D and f values showed good reproducibility,with ICC between the 2 observers of 0.978(95%CI:0.912-0.995),0.964(95%CI:0.854-0.991)and 0.920(95%CI:0.679-0.980),respectively.D*showed a relatively good reproducibility,with ICC of 0.797(95%CI:0.184-0.950).3.1.2 Reproducibility of DW-MPI ParametersAll 18 rabbits at week 3 underwent repeat MR measurements,with ICC ranging from 0.745 to 0.967(Table 1).The Bland-Altman analysis for repeated measurements is shown in Fig 2.ADC,D,f,DSEM,and DDKI values showed excellent reproducibility.D*,αSEM and KDKI,however,showed a relatively good reproducibility.ICC of D(0.967,95%CI:0.865-0.993)was ranked first in the table and ICC of KDKI(0.745,95%CI:0.160-0.872)was at the bottom of the table.3.2 Parameter Measurements3.2.1 Serial Measurements of ADC and IVIM Parameters in the Different ROIsAlthough measurements of ADC and D(median with ranges in parentheses)in different ROIs demonstrated a gradual decrease from week 2 to week 4(expect D measurements in peripheral area at week 3 were equal to that of at week 4),only the ADC value for the whole lesion and MRDA and D value for MRDA among different time points had significant differences(p = 0.000,p = 0.024,and p = 0.001,respectively).Then further comparisons made by using Wilcoxon signed rank test showed that ADC for the whole lesion and D for MRDA at week 4 were significantly lower than the corresponding values at week 2(p= 0.000 and p = 0.014,respectively)and week 3(p= 0.014 and p = 0.012,respectively).Between week 2 and week 3,however,both ADC for the whole lesion and D for MRDA showed no statistical significance.The ADC for MRDA was also statistically different between week 2 and week 4(p =0.011),while there was no difference between week 3 and week 2 or week 4.Serial measurements of f in different ROIs at three time points illustrated a downward trend that was also statistically different(Table 1).Compared with week 2,f for all three ROIs markedly decreased at week 3(although these were not statistically significant),and continuously dropped at week 4(p= 0.004,p = 0.002,and p = 0.018,respectively).Moreover,statistical significances of f values in those ROIs were also showed between week 3 and week 4(p= 0.001,p = 0.000,and p =0.000,respectively).D*values,however,in the all three ROIs from week 2 to week 4 were proved no significant differences.3.2.2 The ADC and IVIM parameters in three ROIs of tumors at each time pointThe Kruskal-Wallis test demonstrated significant differences in D,f and ADC values among the whole lesion,MRDA and peripheral area at week 2,3 and 4.However,there were no differences of D*values among in those ROIs at the same time points(Table 1).The comparisons of ADC,D and/values for three ROIs demonstrated there was statistically different between the whole lesion and MRDA as well as between MRDA and peripheral area at the each observing time point with the Kruskal-Wallis test,while no significant differences of ADC,D and/values were found between the whole lesion and peripheral area at any time points by using the Mann-Whitney test.3.2.3 Serial Measurements of DW-MPI ParametersStatistical significances of ADC，f and KDK1 were found from week 2 to week 4(p= 0.000,p = 0.000,and p = 0.002,respectively).Although measurements of ADC demonstrated a slight decrease at 2-4-week follow-up,up-and-down trends in the number off/and KDKl were shown.Then further comparisons showed that there were significant differences of ADC，f and KDKI between week 2 and week 4(p= 0.000,p =0.004,and p = 0.017,respectively)as well as between week 3 and week 4(p= 0.014,p = 0.001,and p = 0.012,respectively).Between week 2 and week 3,however,all three parameters showed no statistical significance(p= 0.102,p = 0.500,and p =0.064,respectively).Furthermore,serial measurements of D,D*,DDKI，Dsem,andαSEM at the same period from week 2 to week 4 illustrated no statistical differences.3.2.4 The comparisons and P values of diffusion-related parametersThe comparisons and P values of diffusion-related parameters,ADC,D,DDKl,and DSEM,are shown in Table 3.There were significant differences between ADC and D,ADC and DDKl,D and DDKl，D and DSEM at week 2,3 and 4.However,no statistical differences between ADC and DSEM,DDKI and DSEM were found at each time point.3.2.5 CorrelationsA univariate linear regression analysis was performed for perfusion-related parameter with MVD as independent variable.The r2for ADC,fand D*with MVD are listed in table 4.ADC and f showed highest correlation with MVD at week 4(r2=0.307,p = 0.017 and r2= 0.402,p<0.01，respectively).D*,however,did not show a statistically significant correlation with MVD at any time point.Multivariate linear regression analysis was performed with ADC,f and D*at week 4 as predictive variables and MVD as response variable(table 5).Analysis revealed significance only for the regression coefficient estimates off(p = 0.001)and ADC(p = 0.002).3.3 Histologic EvaluationFor rabbits euthanized within an hour after MR imaging,gross liver specimen demonstrated small necrosis began to appear in the center of tumor at week 3 after tumor implantation.H-E slides of the peripheral rim of tumor showed the presence of cluster of intensive tumor cell area,the sparse cell area which is close to mucoid degeneration,and some sparse distributions of necrosis.4.Conclusion:Peripheral area of tumor could be used as an alternative to the whole lesion as ROIs to assess the VX2 liver tumor and the period between week 2 and week 3 should be regarded as objective observing time for monitoring the evolution of liver tumor with IVIM DW imaging.D、D*、DOKI DSEM and asEM were identified as the most promising parameters for monitoring changes of VX2 carcinoma and ADC and f showed highest correlation with MVD at week 4.This study validates a preclinical platform to investigate the value of the different diffusion parameters as individual or combined predictors for cellularity and vascularity which are key parameters in living cancer tissues. |
PDF Full Text Request