The Study Of Magnetic Resonance Imaging In The Preoperative Diagnosis And Differentiation Of Pancreatic Neuroendocrine Neoplasms

Part 1 The Value of magnetic resonance imaging features in the grading differentiation of pancreatic neuroendocrine neoplasmsObjective To access the value of magnetic resonance imaging(MRI)features in the grading of pancreatic neuroendocrine neoplasms(PNENs).Methods From January 1,2012 to December 30,2016,a total of 99 patients(114 lesions)who underwent preoperative MR imaging were included.All cases were pathologically proved to be PNENs.Tumor grading was based on the 2010 WHO classification which included G1,G2,G3.MR imaging features included lesion location,size,boundary,cystic degeneration,pancreatic duct dilatation,vessel or organs invasion,liver metastasis,and enhancement pattern.Statistical analysis methods including χ2 test,variance analysis,sensitivity,specificity,positive predictive value and negative predictive value were used.Results According to the number of lesions,there were 69 tumors of G1,38 tumors of G2,7 tumors of G3.Ill-defined boundary,cystic degeneration,peri-pancreas invasion,bile duct dilatation,liver metastasis,hypo-hypo intensity in both arterial and portal venous phases were more common in higher grade PNENs,especially in G3.When at least two of the five criteria(ill-defined boundary,size≥3cm,vessel or organs invasion,liver metastasis,hypo-hypo enhancement pattern)were used,sensitivity and specificity of differentiating G3 from G1-2 were 100%,87%.When at least four of the five criteria(well-defined boundary,size≤2cm,no vessel or organs invasion,no liver metastasis,hyper-hyper enhancement pattern),sensitivity and specificity of differentiating G1 from G2-3 were 91%,49%.Most G1 and G2 were characterized by persistent high enhancement.Conclusions MR imaging features varies in different tumor grade of PNENs.Ill-defined boundary,cystic degeneration,peri-pancreas invasion,vessel or organs invasion,liver metastasis,hypo-intensity helped to identify higher grade PNENs,especially G3.Most G1 and G2 were characterized by persistent enhancement.Part 2 The value of quantitive preoperative DW-MRI in the grading of pancreatic neuroendocrine neoplasmsObjective To assess the value of the quantitive parameters of preoperative MRI for the grading of pancreatic neuroendocrine neoplasms(PNENs).Methods Between 2012 and 2016,all resected PNENs with preoperative MR imaging and diffusion weighted imaging(DWI)were analyzed.Tumor grading was based on the 2010 WHO classification.The parameters evaluated included tumor size,minimum apparent diffusion coefficient value(ADCmin),mean ADC value(ADCmean)and Ki67 index.Receiver operating characteristic curve(ROC)was used to determine the cut off values and the sensitivity and specificity of prediction.Spearman correlation and binary logistic regression analysis were adopted to identify the association between tumor size,ADC value and pathological parameters.Results There were 87 tumors in total(48 G1,32 G2,7 G3).The size was significantly bigger as grade increased(size: 1.40 cm G1,2.92 cm G2,3.80 cm G3,P<0.001).ADCmin and ADCmean value were significantly lower as grade increased(ADCmin: 1.215±0.46×10-3 mm2/s G1,0.993±0.48×10-3 mm2/s G2,0.644±0.29×10-3 mm2/s G3,P=0.004;ADCmean: 1.402±0.47×10-3 mm2/s G1,1.281±0.48×10-3 mm2/s G2,0.886±0.24×10-3 mm2/s G3,P=0.023).Size,ADCmin,and ADCmean were all included in the binary logistic regression analysis,thus the predicted value(PRE)was included in the ROC analysis.The area under ROC curve(AUC)of PRE1,size,ADCmin and ADCmean to differentiate G1 and G2-3 were 0.829±0.05,0.835±0.06,0.702±0.060 and 0.629±0.06 respectively.For further analysis,the AUC of size and PRE1 was not significantly different(P=0.65),and the AUC between the other groups was significantly different.The difference of AUC between size and the other two was statistically significant(P<0.05).Optimal cut-off values for size,ADCmin and ADCmean were 1.65cm(Sensitivity 71%,Specificity 87%),1.087×10-3 mm2/s(Sensitivity 74%,Specificity 58%),and 0.879×10-3 mm2/s(Sensitivity 31%,Specificity 94%)respectively.The AUC of PRE2,ADCmin,ADCmean and size to differentiate G3 and G1-2 were 0.868±0.08,0.821±0.09,0.818±0.09 and 0.752±0.09 respectively.For further analysis,AUC between PRE2 and size was significantly different(P=0.04),and there was no significant difference between the AUC of the other groups(P>0.05).Optimal cut-off value for size,ADCmin and ADCmean was 2.10 cm(Sensitivity 86%,Specificity 64%),0.777×10-3 mm2/s(Sensitivity 86%,Specificity 73%),and 1.047×10-3 mm2/s(Sensitivity 86%,Specificity 68%)respectively.Ki67 was positively correlated with tumor size(r=0.496,p<0.001),and negatively correlated with ADCmin and ADCmean(r=-0.322 and-0.224,P<0.05,respectively).Tumor size was negatively correlated with ADCmin and ADCmean(r=-0.518 and-0.379,p<0.001,respectively).Conclusions Tumor size,ADCmin and ADCmean were useful to differentiate grade of PNENs and correlated with tumor proliferation.In differentiating G1 and G2-3,tumor size may be more useful than ADC values.In differentiating G3 and G1-2,ADC values and tumor size showed similar diagnostic performance.Part 3 The value of intravoxel incoherent motion diffusion weighted MRI on differentiating pancreatic neuroendocrine neoplasms and other pancreatic solid neoplasmsObjective To assess the value of intravoxel incoherent motion diffusion weighted MRI(IVIM DW-MRI)on differentiating pancreatic neuroendocrine neoplasms(PNENs),pancreatic ductal adenocarcinoma(PDAC),solid pseudopapillary neoplasms(SPNs)and normal pancreas(NP).Methods A total of 66 patients were included,including 25 cases with pathologically proven PNENs,19 with PDAC,11 with SPNs.There were 11 cases in NP group.According to 2010 WHO grading criteria,PNENs were subcategorized into grade G1,G2 and G3.All patients underwent abdominal conventional MRI examination and IVIM diffusion weighted imaging with 10 b values on a 3.0T MRI.The data included apparent diffusion coefficient(ADC),slow component of diffusion(Dslow),incoherent microcirculation(Dfast)and perfusion fraction(f).Univariate analysis and two independent sample t test were used for inter-group comparison.The diagnostic performance of the parameters was evaluated by using receiver operating characteristic(ROC)analysis.Results There were 28 lesions with PNENs(19 G1,8 G2,1G3),19 PDAC,11 SPNs and 11 NP.ADC,Dslow,Dfast and f didn’t show statistic significancebetween G1 and G2(all P>0.05).ADC,Dslow,Dfast and f values of PNENs and normal pancreas showed no statistic significance.Dfast and f values of PNENs were significantly higher than those of PDAC(all P <0.001).ADC,Dslow,Dfast and f values of PNENs were significantly higher than those of SPNs(all P<0.05).In ROC analysis,for differentiating PNENs and PDAC,Dfast and f showed larger area under ROC curve(AUC)than ADC and Dslow,and the AUC between Dfast and f did not show statistically significant difference(P=0.33).For differentiating PNENs and SPNs,AUC of ADC、Dslow、Dfast、f was 0.750±0.08、0.769±0.09、0.701±0.09 and 0.776±0.08,respectively(P<0.05),and the AUC between the four parameters showed statistically significant difference.Conclusions IVIM parameters helped to differentiate PNENs and PDAC,especially Dfast and f values,indicating that the main point for differentiatint the two was blood perfusion difference.The differentiation of PNENs and SPNs in routing imaging may sometimes be difficult,but there was difference in tissue blood perfusion and water diffusion that can be identified.IVIM parameters alone may not help to differentiate G1 and G2,PNENs and normal pancreas.