Establishment And Verification Of Risk Prediction Model For Prostate Cancer Based On SWE And MpMRI

Objective:Based on shear wave elastography(SWE)and multi-parameter magnetic resonance imaging(mpMRI),a model for predicting the risk of prostate cancer(PCa)at the first biopsy was established,and the predictive efficiency of different models was compared and analyzed to help urologists make clinical decisions and avoid unnecessary puncture biopsies.This study is divided into three chapters:Chapter one: The clinical data of patients were collected,and the corresponding imaging indexes were obtained based on SWE and mpMRI.The differences of clinical data and imaging indexes between PCa patients and non-PCa patients were analyzed,and the predictive efficiency of each univariate index to PCa was evaluated.Chapter two: The independent predictors of PCa occurrence were screened by multi-factor stepwise Logistic regression analysis,and the Logistic regression models without SWE index and without SWE index were established based on the independent prediction factors.The prediction efficiency of the two models was compared to explore whether the efficiency of the prediction model was improved after adding SWE quantitative index.Chapter three: Based on the results of multi-factor Logistic regression analysis,a Nomogram prediction model with SWE quantitative index is established,and the prediction efficiency of the model is evaluated and verified,and its clinical practical value is discussed.Chapter one1.Materials and methodsA total of 128 patients suspected of PCa and ultrasound-guided prostate biopsy in our hospital were divided into PCa group(n = 44)and non-PCa group(n = 84).Independent sample t-test,Mann-Whitney U test and chi-square test were used to compare whether there were significant differences between the two groups,and the prediction efficiency of each single factor index was analyzed by receiver operating characteristic(receiver operating characteristic curve,ROC)curve.The difference was statistically significant(P < 0.05).2.ResultsThere were significant differences in prostate cluster calcification,total prostate specific antigen,free prostate specific antigen,prostate specific antigen density,prostate volume,maximum difference of Young’s modulus,average difference of Young’s modulus,minimum difference of Young’s modulus and PI-RADS score between PCa group and non-PCa group.There was no significant difference in age,body mass index,puncture route and needle number(P > 0.05).ROC curve analysis showed the presence or absence of intra-prostate cluster calcification,total prostate specific antigen,free prostate specific antigen,prostate specific antigen density,prostate volume,maximum difference of Young’s modulus,average difference of Young’s modulus,minimum difference of Young’s modulus and area under the curve for diagnosis of PCa by PI-RADS score(area under the curve).AUC)were 0.607,0.803,0.742,0.838,0.721,0.959,0.954,0.918 and 0.903,respectively.The AUC of the maximum difference of Young’s modulus was higher than that of other indexes.3.Brief summaryIn this study,based on SWE and mpMRI analysis,patients with clustered calcification in the prostate,smaller prostate volume,higher density of total prostate specific antigen,free prostate specific antigen,prostate specific antigen,maximum difference of Young’s modulus,average difference of Young’s modulus,minimum difference of Young’s modulus and higher PI-RADS score were at greater risk of developing PCa.Through the analysis of ROC curve,it is found that the above indexes have significant prediction efficiency for the occurrence risk of PCa,and the difference between the maximum value of Young’s modulus is the best.Chapter two1.Materials and methodsThe objects included are the same as the first chapter.The independent influencing factors of PCa were screened by multi-factor stepwise Logistic regression analysis,and the Logistic regression models without SWE index and SWE index were established respectively.The ROC curve of each model was drawn,and the difference between the AUC of the two models was tested by Delong test.The prediction efficiency of the regression model with SWE index compared with the regression model without SWE index was analyzed by net reclassification index(NRI)and comprehensive discriminant improvement index(IDI).2.ResultsMultivariate stepwise Logistic regression analysis showed that prostate specific antigen density,mean difference of Young’s modulus and PI-RADS score were independent predictors of the risk of PCa.The AUC of the regression model without SWE index and SWE index was 0.941 and 0.984 respectively.Delong test showed that there was significant difference in AUC between the two models.NRI and IDI analysis showed that the prediction efficiency of adding SWE index was significantly better than that of not adding SWE index(NRI=1.320,P < 0.001 and IDI=0.164,P < 0.001).3.Brief summaryIn this study,it was found that prostate specific antigen density,average difference of Young’s modulus and PI-RADS score were independent predictors of PCa.Based on these independent predictors,two Logistic regression models are established.The AUC,sensitivity and specificity of the logistic regression model with SWE index are 0.984,0.932 and 0.929 respectively,and its AUC and sensitivity are better than those of the regression model without SWE index.Chapter three1.Materials and methodsBased on the independent predictors obtained from multi-factor stepwise Logistic regression analysis in chapter 2,a Nomogram prediction model for predicting the risk of PCa is established,and the goodness of fit of the model is evaluated by calibration curve,the distinguishing ability of the model is evaluated by ROC curve,and the clinical application value of the model is evaluated by decision curve analysis(DCA).2.ResultsThe Nomogram prediction model is constructed based on the independent prediction factors.The sensitivity and specificity of the model AUC=0.984 are 0.932 and 0.929 respectively.The calibration curve shows that the prediction risk of the model is consistent with the actual risk.The analysis of decision curve shows that the clinical rate of return of the model is higher than that of "total intervention" and "non-intervention" in almost all predicted probability thresholds,suggesting that the Nomogram model has a good clinical net return in predicting the risk of PCa.3.Brief summaryIn this study,based on the independent predictors obtained by multi-factor stepwise Logistic regression analysis,a Nomogram model for predicting the risk of PCa is established.The accurate curve shows that the model has good calibration and differentiation ability.The analysis of decision curve shows that the Nomogram model has good clinical practical value in almost all prediction probability threshold range.ConclusionsBased on SWE and mpMRI analysis,we found that patients with clustered calcification,smaller prostate,higher density of total prostate specific antigen,free prostate specific antigen,prostate specific antigen,maximum difference of Young’s modulus,average difference of Young’s modulus,minimum difference of Young’s modulus and higher PI-RADS score had a higher risk of developing PCa.Through the analysis of ROC curve,it is found that the above indexes have significant prediction efficiency for the occurrence risk of PCa,and the difference between the maximum value of Young’s modulus is the best.In this study,it was found that prostate specific antigen density,average difference of Young’s modulus and PI-RADS score were independent predictors of PCa.Based on these independent predictors,two Logistic regression models are established.The AUC,sensitivity and specificity of the logistic regression model with SWE index are 0.984,0.932 and 0.929 respectively,and its AUC and sensitivity are better than those of the regression model without SWE index.In this study,based on the independent predictors obtained by multi-factor stepwise Logistic regression analysis,a Nomogram model for predicting the risk of PCa was established.The calibration curve shows that the model has good calibration and discrimination ability.Decision curve analysis shows that the Nomogram model has good clinical practical value in almost all prediction probability threshold range.The model established in this study can provide a reference for clinicians to accurately assess the risk of PCa and formulate a reasonable diagnosis and treatment plan.