The Application Of PSMA PET Imaging Combined With Multi-parametric MRI In The Diagnosis Of Prostate Cancer

Part 1 Establishment and efficacy evaluation of a comprehensive model for ¹⁸F-PSMA-1007 PET/MR diagnosis of prostate cancerObjective:To study the value of ¹⁸F-PSMA-1007 PET/MR in the differential diagnosis of benign and malignant prostate lesions.Methods:Retrospective analysis of the clinical and imaging data of patients who underwent ¹⁸F-PSMA-1007 PET/MR examination from September 2021 to February 2023 with pathological results of prostate biopsy.Clinical information included age,body mass index（BMI）,prostate specific antigen（PSA）.Metabolic parameters included maximun standardized uptake value（SUVmax）,standard standardized uptake value（SUVstd）,peak standardized uptake value（SUVpeak）,mean standard uptake value（SUVmean）and SUVmean of mediastinal blood pool,spleen and parotid gland.MR image features included location,signal,shape,boundary,capsule,PI-RADS V2.1 score and apparent diffusion coefficient（ADC）of lesions.Prostate cancer Molecular Imaging Standardized Evaluation（PROMISE）was used to evaluate the expression of prostate-specific membrane antigen（PSMA）receptor in prostate cancer lesions based on molecular images（mi PSMA score）.Independent sample T-test,Mann-Whitney U test orχ²test were used to compare prostate cancer group with non-prostate cancer groups on single factor comparison.The collected clinical information,PET metabolic parameters and MR image features were analyzed by binary Logistic regression,and the independent predictive factors obtained by each group were integrated to establish a comprehensive model.The diagnostic efficacy of independent predictors,comprehensive model and mi PSMA score were evaluated by receiver operating characteristic（ROC）curve.Results:1.70 patients were included,aged from 56 to 88 years old,with an average age of69.13±7.13 years.53 patients were prostate cancer,other patients were non-prostate cancer.2.In univariate analysis,there were significant differences in age,BMI,PSA,SUVmax,SUVmean,SUVstd,SUVpeak,SUVmean ratio of lesion to mediastinal blood pool,SUVmean ratio of lesion to spleen,SUVmean ratio of lesion to parotid gland,location,shape,boundary,capsule,PI-RADS V2.1 score and ADC between the two groups（all P<0.05）.The multivariate analysis showed that there was no independent predictor of prostate cancer in clinical information.In PET metabolic parameters,the SUVmean ratio of lesion to spleen was an independent predictor of prostate cancer.In MR image features,ADC was an independent predictor of prostate cancer.3.The ROC curve showed that the AUC of SUVmean ratio of lesion to spleen was0.825（95%CI=0.715～0.905）,the best cut-off value was 0.91,the sensitivity was 58.5%,and the specificity was 100.0%.The AUC of ADC was 0.887（95%CI=0.788～0.950）,the best cut-off value was 774.5×10^-6mm²/s,the sensitivity was 77.4%,and the specificity was88.2%.The AUC of comprehensive model was 0.912（95%CI=0.820～0.967）,the sensitivity was 84.9%,and the specificity was 94.1%.The AUC of mi PSMA score was0.887（95%CI=0.788～0.950）,the sensitivity was 77.4%,and the specificity was 100.0%.Conclusion:PSMA PET/MR is effective in the diagnosis of prostate cancer.Taking the SUVmean ratio of lesions to spleen>0.91,ADC≤774.5×10^-6mm²/s as the threshold,the AUC for the diagnosis of prostate cancer was 0.912,which was better than mi PSMA score.Part 2 Study on the ¹⁸F-PSMA-1007 PET/MR in risk stratification of prostate cancerObjective:To study the value of ¹⁸F-PSMA-1007 PET/MR in the risk stratification of prostate cancer.Methods:Retrospective analysis of 53 patients with prostate cancer who underwent ¹⁸F-PSMA-1007 PET/MR examination from September 2021 to February 2023（same as Part 1）.According to the diagnosis and treatment guidelines for prostate cancer,patients were divided into high risk group and medium and low-risk group.T-test,Mann-Whitney U test orχ²test were used to analyze clinical information,metabolic parameters and MR image features.The collected clinical information,PET metabolic parameters and MR image features were analyzed by binary Logistic regression,and the independent predictors for the diagnosis of high-risk prostate cancer were obtained.Draw the ROC curve of independent predictors,calculate the optimal threshold,and obtain AUC.Results:1.There were 37 patients in high risk group and 16 patients in medium and low risk group.2.In univariate analysis,there were significant differences in SUVmax,SUVmean,SUVpeak,SUVstd,SUVmean ratio of lesion to mediastinal blood pool,SUVmean ratio of lesion to spleen,SUVmean ratio of lesion to parotid gland,location,shape,capsule,PI-RADS V2.1 score and ADC between the two groups（all P<0.05）.The multivariate analysis showed that there was no independent predictor of high-risk prostate cancer in PET metabolic parameters,while ADC was an independent predictor of prostate cancer in MR image features.3.According to the ROC curve,ADC≤774.5×10^-6mm²/s was used as the diagnostic threshold for high-risk prostate cancer.The diagnostic sensitivity and specificity were 86.5%and 68.7%respectively,and the AUC was 0.758.Conclusion:PSMA PET/MR has a certain value in the risk stratification of prostate cancer.Taking the ADC≤732.06×10^-6mm²/s as the threshold for high-risk prostate cancer,and its AUC is 0.758.