Establishment of an Individualized Predictive Model to Reduce the Core Number for Systematic Prostate Biopsy: A Dual Center Study Based on Stratification of the Disease Risk Score

Zeyu Chen; Min Qu; Xianqi Shen; Shaoqin Jiang; Wenhui Zhang; Jin Ji; Yan Wang; Jili Zhang; Zhenlin Chen; Lu Lin; Mengqiang Li; Cheng Wu; Xu Gao

doi:10.3389/fonc.2021.831603

Establishment of an Individualized Predictive Model to Reduce the Core Number for Systematic Prostate Biopsy: A Dual Center Study Based on Stratification of the Disease Risk Score

Front Oncol. 2022 Feb 14:11:831603. doi: 10.3389/fonc.2021.831603. eCollection 2021.

Authors

Zeyu Chen¹, Min Qu¹, Xianqi Shen¹, Shaoqin Jiang^{1

2}, Wenhui Zhang¹, Jin Ji¹, Yan Wang¹, Jili Zhang¹, Zhenlin Chen², Lu Lin¹, Mengqiang Li², Cheng Wu³, Xu Gao¹

Affiliations

¹ Department of Urology, Shanghai Changhai Hospital, Second Military Medical University, Shanghai, China.
² Department of Urology, Fujian Union Hospital, Fujian Medical University, Fuzhou, China.
³ Department of Health Statistics, Second Military Medical University, Shanghai, China.

Abstract

Purpose: To establish an individualized prostate biopsy model that reduces unnecessary biopsy cores based on multiparameter MRI (mpMRI).

Materials and methods: This retrospective, non-inferiority dual-center study retrospectively included 609 patients from the Changhai Hospital from June 2017 to November 2020 and 431 patients from the Fujian Union Hospital between 2014 and 2019. Clinical, radiological, and pathological data were analyzed. Data from the Changhai Hospital were used for modeling by calculating the patients' disease risk scores. Data from the Fujian Union Hospital were used for external verification.

Results: Based on the data of 609 patients from the Changhai Hospital, we divided the patients evenly into five layers according to the disease risk score. The area under the receiver operating characteristic (ROC) curve (AUC) with 95% confidence intervals (CI) was analyzed. Twelve-core systemic biopsy (12-SBx) was used as the reference standard. The SBx cores from each layer were reduced to 9, 6, 5, 4, and 4. The data of 279 patients with benign pathological results from the Fujian Union Hospital were incorporated into the model. No patients were in the first layer. The accuracies of the models for the other layers were 88, 96.43, 94.87, and 94.59%. The accuracy of each layer would be increased to 96, 100, 100, and 97.30% if the diagnosis of non-clinically significant prostate cancer was excluded.

Conclusions: In this study, we established an individualized biopsy model using data from a dual center. The results showed great accuracy of the model, indicating its future clinical application.

Keywords: PI-RADS; multicenter study; multiparameter MRI; predictive models; prostate biopsy.