Header

UZH-Logo

Maintenance Infos

Creation and internal validation of a biopsy avoidance prediction tool to aid in the choice of diagnostic approach in patients with prostate cancer suspicion


Bhindi, Bimal; Jiang, Haiyan; Poyet, Cedric; Hermanns, Thomas; Hamilton, Robert J; Li, Kathy; Toi, Ants; Finelli, Antonio; Zlotta, Alexandre R; van der Kwast, Theodorus H; Evans, Andrew; Fleshner, Neil E; Kulkarni, Girish S (2017). Creation and internal validation of a biopsy avoidance prediction tool to aid in the choice of diagnostic approach in patients with prostate cancer suspicion. Urologic oncology, 35(10):604.e17-604.e24.

Abstract

INTRODUCTION: To reduce unnecessary prostate biopsies while using novel tests judiciously, we created a tool to predict the probability of clinically significant prostate cancer (CSPC) vs. low-risk prostate cancer or negative biopsy (i.e., when intervention is likely not needed) among men undergoing initial or repeat biopsy.
METHODS: Separate models were created for men undergoing initial and repeat biopsy, identified from our institutional biopsy database and the placebo arm of the REDUCE trial, respectively, to predict the presence of CSPC (Gleason≥7 or>33% of cores involved). Predictors considered included age, race, body mass index, family history of prostate cancer, digital rectal examination, prostate volume, prostate-specific antigen (PSA), free-to-total PSA, presence of high-grade prostatic intraepithelial neoplasia or atypical small acinar proliferation on prior biopsy, number of prior biopsies, and number of cores previously taken. Multivariable logistic regression models that minimized the Akaike Information Criterion and maximized out-of-sample area under the receiver operating characteristics curve (AUC) were selected.
RESULTS: Of 7,963 biopsies (initial = 2,042; repeat = 5,921), 1,138 had CSPC (initial = 870 [42.6%]; repeat = 268 [4.5%]). Age, race, body mass index, family history, digital rectal examination, and PSA were included in the initial biopsy model (out-of-sample AUC = 0.74). Age, prostate volume, PSA, free-to-total PSA, prior high-grade prostatic intraepithelial neoplasia, and number of prior biopsies were included in the repeat biopsy model (out-of-sample AUC = 0.81).
CONCLUSION: These prediction models may help guide clinicians in avoiding unnecessary initial and repeat biopsies in men unlikely to harbor CSPC. This tool may also allow for the more judicious use of novel tests only in patients in need of further risk stratification before deciding whether to biopsy.

Abstract

INTRODUCTION: To reduce unnecessary prostate biopsies while using novel tests judiciously, we created a tool to predict the probability of clinically significant prostate cancer (CSPC) vs. low-risk prostate cancer or negative biopsy (i.e., when intervention is likely not needed) among men undergoing initial or repeat biopsy.
METHODS: Separate models were created for men undergoing initial and repeat biopsy, identified from our institutional biopsy database and the placebo arm of the REDUCE trial, respectively, to predict the presence of CSPC (Gleason≥7 or>33% of cores involved). Predictors considered included age, race, body mass index, family history of prostate cancer, digital rectal examination, prostate volume, prostate-specific antigen (PSA), free-to-total PSA, presence of high-grade prostatic intraepithelial neoplasia or atypical small acinar proliferation on prior biopsy, number of prior biopsies, and number of cores previously taken. Multivariable logistic regression models that minimized the Akaike Information Criterion and maximized out-of-sample area under the receiver operating characteristics curve (AUC) were selected.
RESULTS: Of 7,963 biopsies (initial = 2,042; repeat = 5,921), 1,138 had CSPC (initial = 870 [42.6%]; repeat = 268 [4.5%]). Age, race, body mass index, family history, digital rectal examination, and PSA were included in the initial biopsy model (out-of-sample AUC = 0.74). Age, prostate volume, PSA, free-to-total PSA, prior high-grade prostatic intraepithelial neoplasia, and number of prior biopsies were included in the repeat biopsy model (out-of-sample AUC = 0.81).
CONCLUSION: These prediction models may help guide clinicians in avoiding unnecessary initial and repeat biopsies in men unlikely to harbor CSPC. This tool may also allow for the more judicious use of novel tests only in patients in need of further risk stratification before deciding whether to biopsy.

Statistics

Altmetrics

Downloads

0 downloads since deposited on 05 Dec 2017
0 downloads since 12 months

Additional indexing

Item Type:Journal Article, refereed, original work
Communities & Collections:04 Faculty of Medicine > University Hospital Zurich > Urological Clinic
Dewey Decimal Classification:610 Medicine & health
Uncontrolled Keywords:Early detection of cancer; Prostate biopsy; Prostatic neoplasms
Date:October 2017
Deposited On:05 Dec 2017 16:05
Last Modified:09 Dec 2017 03:54
Publisher:Elsevier
ISSN:1078-1439
Publisher DOI:https://doi.org/10.1016/j.urolonc.2017.06.044
PubMed ID:28781111

Download