Construction And Validation Of Prostate Cancer Prediction Model Based On Clinical Indicators Such As Free Prostate Specific Antigen Density

Objective: Prostate biopsy is the gold standard for diagnosing prostate cancer,because of its invasive nature and complications,many studies have been devoted to building predictive models to assist in the diagnosis of prostate cancer.Existing studies have shown that free prostate specific antigen density(fPSAD)is positively correlated with the risk of prostate cancer,but there are no predictive models in patients undergoing prostate biopsy to show whether fPSAD is an effective predictor of prostate biopsy for the diagnosis of prostate cancer.Therefore,this study constructed and validated a prediction model based on clinical indicators such as fPSAD to predict the probability of prostate cancer in prostate biopsy patients.Methods: Clinical data of patients who underwent ultrasound-guided prostate biopsy at a tertiary class A hospital from January 2017 to June 2022 were collected retrospectively.Patients’ information was collected including age,body mass index(BMI),history of hypertension,diabetes,coronary heart diseases,personal history of smoking and alcohol consumption,radio of neutrophil to lymphocyte,total prostate specific antigen,free prostate specific antigen,fPSAD,prostate specific antigen density(PSAD)and postoperative pathologic results.Patients from January 2017 to December 2020 were grouped into a model constructing group,and patients from January 2021 to June 2022 were grouped into a validation group.The baseline data of the two groups with prostate cancer and non-prostate cancer were analyzed,then LASSO regression was used to screen variables.Using multivariate logistic regression to fit and construct predictive model.The receiver operating characteristic curve,consistency test and clinical decision curve were used to evaluate the overall value of the model.Results: After inclusion and exclusion criteria,there were 465 patients in the model constructing group and 238 patients in the validation group.The established model predictors included 7 factors containing age,BMI,history of hypertension,diabetes,free/total prostate specific antigen(f/tPSA),fPSAD andPSAD.The area under the ROC curve of the model is 0.808(95%CI=0.769-0.847,P<0.001)and the optimal cut-off value was 49%.Its sensitivity and specificity were 68.9% and 81.1%.The ROC curve analysis of the model and each predictor showed that the diagnostic value of the model was better than each predictor,andPSAD and fPSAD also had good diagnostic value.Through external validation,the model has good discrimination and the area under ROC is 0.800(95%CI=0.743-0.858,P<0.001).The unreliability test P=0.582 in the consistency test indicates that the model has a strong calibration degree.The decision curve shows that if the cut-off value of the model is in the range of 10%～95%,it indicates that the model has good clinical applicability.The model’s cut-off value in this article is 49%,which is within the range.Conclusion: 1.Age,BMI,history of hypertension,diabetes,f/tPSA,fPSAD,andPSAD may be used as predictive factors for diagnosing prostate cancer in patients undergoing prostate biopsy.2.The diagnostic predictive model constructed by combining the above clinical features has high accuracy and good clinical application value,and the resulting model can help clinicians and patients in the preoperative decision-making of prostate biopsy.