Study On Grading Diagnostic Model Of Chronic Pancreatitis Based On Multiple Ordinal Logistic Regression After Factor Analysis

Part 1 Secretin Enhanced Magnetic Resonance Cholangiopancreatography in the Diagnosis of Exocrine Pancreatic Function:A Systematic Review and Meta AnalysisObjective: to evaluate the value of secretin enhanced magnetic resonance cholangiopancreatography(S-MRCP) in the diagnosis of exocrine pancreatic function.Methods: Two authors independently performed a comprehensive search of Pub Med, EBSCO, Ovid, Springer, and the Cochrane Library from January 1990 to September 2013. Metadisc 1.4 was used for data analysis, and the pooled weighted sensitivity, specificity and 95% confidence interval(CI) were calculated. The area under curve(AUC) of summary receiver operating characteristic(SROC) curve was also calculated.Results: 6 studies were included in Meta-analysis. All groups had homogeneity, and a fixation effects regression was developed. The pooled weighted sensitivity with 95% CI were 0.79 [95% CI, 0.68-0.88] and 0.91 [95% CI, 0.86-0.94] respectively. AUC of SROC curve were 91.88%.Conclusions: Secretin-enhanced MRCP has a much high diagnostic accuracy and should be preferred for diagnosis of exocrine pancreatic function.Part 2 Secretin Enhanced MRCP: Assessment of Duct Visualization in Chronic PancreatitisObjective: To investigate Secretin enhanced Magnetic Resonance Cholangiopancreatography(S-MRCP) in improvement of duct visualization and the value of grade diagnosing CP.Methods: This was a prospective study. The present study includes 88 subjects(23 health volunteers and 65 CP). Coronal T2-weighted sequences and thick slab 2D MRCP sequences were performed before and after 0.1 m L/kg secretin at 2 min intervals. Image quality, visualizations of main pancreatic duct(MPD) and branch pancreatic duct(BPD), diameter of MPD and duodenal filling(DF) were analyzed by two radiologists before and after secretin stimulation. The exocrine function of the pancreas was evaluated using DF grades. All subjects were classified based on Cambridge classification and DF grades. The results of visualization of MPD and BPD,grade diagnosing CP were compared with ERCP. The diameters of MPD before and after secretin stimulation and the peak time were compared between all groups of Cambridge classification using the one-way ANOVA test. All imaging findings of MPD and BPD were compared between two groups of DF grades using x2 test.Results: There was significant difference of baseline value(F=15.17, P=0.0001), peak value(F=14.86, P=0.0001), 10 min after secretin stimulation value(F=15.08, P=0.0001) and peak time(F=24.67, P=0.0001) between groups of Cambridge classification. S-MRCP has more sensitivity and specificity than MRCP compared with ERCP reference standards. The Kappa value was 0.55 in MRCP and 0.77 in S-MRCP for grade diagnosing CP based on Cambridge classification. There were significant difference of dilation(x2=5.01, P=0.03), narrowing(x2=3.81, P=0.05), filling defect(x2=6.88, P=0.01) and BPD of pancreatic head(x2=4.26, P=0.04) between groups of DF grades. Using 4 min as the cutoff value, the peak time was the most accurate for differentiating patients with health volunteers from CP and yielded 87% sensitivity and 100% specificity.Conclusions: Secretin-stimulated MRCP moderately improves the duct visualizations of MPD and BPD, and the accuracy of grade diagnosing CP.Part 3 Assessment of Chronic Pancreatitis: The Utility of Diffusion Weighted MR Imaging before and after Secretin StimulationObjective: To explore the diagnostic value of diffusion-weighted MR imaging(DWI) after secretin stimulation for diagnosing chronic pancreatitis(CP).Methods: This was a prospective study. Ninety-nine consecutive individuals underwent secretin DWI and FE-1 testing and included 23 healthy volunteers, 11 risk volunteers, 15 mild CP patients, 14 moderate CP patients and 36 severe CP patients grouped by Cambridge classification about endoscopic retrograde cholangiography(ERCP), CT and ultrasonography. Transverse breath-hold single-shot echo-planar DWI were repeated every 2 min after a 0.1 m L/kg secretin injection. Coronal T2 weighted sequence and conventional MRI were also performed on all subjects.FE-1 was assayed by ELISA. The changes of pancreatic ADC values were observed before and after the injection of secretin. Mean baseline and peak ADCs, and time to peak ADCs were compared between groups, and the correlation with FE-1 was determined. A receiver operating characteristic curve(ROC) identified healthy from the risk and CP patients.Results: Eight patients with severe CP were excluded because the significant atrophy of the pancreatic parenchyma was not evaluated for ADC measurement. The mean baseline and peak ADCs were higher in the heathy volunteers than in those with alcohol intake and CP. The risk, mild and moderate CP groups showed a delayed peak compared to healthy volunteers, but the severe CP weren’t able to detect any ADC peak. The peak ADCs were higher than the baseline ADCs in all groups. There was statistically significant difference of peak ADCs(F =49.56, P <0. 05) between normal group and CP groups, but except between normal group and risk group. There was statistically significant difference of time to peak ADCs(F = 105.31, P <0. 05) between all groups. Using 4.67 min as the cutoff value, time to peak ADCs were the most accurate for differentiating healthy from risk patients and those with evident pancreatitis, yielding 80% sensitivity and 100% specificity. A good correlation between baseline and peak ADCs, time to peak ADCs, and FE-1 was shown(r=0.57、0.72 and-0.84, P<0.01).Conclusions: The peak and time to ADCs may inform improved detection of risk and mild CP. Secretin-enhanced DWI is a noninvasive, convenient and accurate method.Part 4 Study on Grading Diagnostic Model of Chronic Pancreatitis Based on Multiple Ordinal Logistic Regression after Factor AnalysisObjective: To acquire better combination of chronic pancreatitis(CP) by means of establishing a predictive model and elevate the general diagnostic ability of CP.Methods: The present study included 23 health volunteers and 68 CP(18 mild CP, 14 moderate CP and 31 severe CP) grouped by Cambridge classification. All the subjects received breath-hold single-shot echo-planar DWI(b value=0, 500 s/mm2), thick slab 2-dimensional MRCP sequence, coronal T2-weighted sequence, routine pancreas plain scan and enhanced scan before and after 0.1 m L/kg secretin at 2 min intervals. The 14 features were analyzed including 9 continuous variables and 5 binary variables. The measurement data used variable variance analysis. The calculated data used the chi-square test. The single ordinal regression analysis was conducted on the variables with significant difference between groups(P<0.05). The effect of multicollinearity was diagnosed and eliminated by the factor analysis. The multiple ordinal logistic regression was conducted.Results: There wasn’t significant difference between control groups and CP groups in pancreas divisum(x2=3.86, P>0.05). There wasn’t significant correlation between the vacuoles of the pancreatic parenchyma and CP(P>0.05). The effect of multicollinearity was eliminated by the factor analysis among the 9 continuous variables. Three common factors were identified, which were F1(X3、X4、X5、X9、X8), F2(X7、X6) and F3(X1、X2) represented the exocrine function, main pancreatic ducts and pancreatic parenchyma, respectively. Six features were implemented into the logistic regression model, which included F1, F2, F3, X10(the visualizations of branch pancreatic duct after secretin stimulation), X11(pancreatic shape) and X13(the filling defects of main pancreatic duct). Finally, the most appropriate regression model was gotten, which was the scale model of the probit link function. The model diagnostic accuracy for normal, mild CP, moderate CP and severe CP was 96.65%, 100%, 71.42% and 100%, respectively. The total predicting accuracy of the model was 94.5%.Conclusion: The multiple ordinal logistic regression model may accurately predict the CP and can offer valuable references for clinical diagnosis and therapy.