Back to Program


Utility of Prostate Size, Patient Age, and Presence of Atypical Small Acinar Proliferation (ASAP) in Prior Biopsy in Selecting Patients for Repeat Biopsy: Results from a Multi-Center Prospective Prostate Biopsy Cohort
Simpa S. Salami, MD, MPH1, John T. Wei, MD, MS2, Meredith M. Regan, ScD3, Douglas Scherr, MD4, Javed Siddiqui, MS2, Michael Kearney, MD1, Robert Eyre, MD1, Gary Kearney, MD1, Elizabeth Genega, MD1, Mark Rubin, MD4, Martin G. Sanda, MD1.
1Beth Israel Deaconess Medical Center/Harvard Medical School, Boston, MA, USA, 2University of Michigan, Ann Arbor, MI, USA, 3Dana-Farber Cancer Institute, Boston, MA, USA, 4Weill Cornell Medical College, New York, NY, USA.

Background:
The decision to send a patient with a previously negative biopsy for repeat biopsy presents a challenge to the urologist, and predictive models that can be used for this purpose are lacking. We sought to evaluate the usefulness of pre-biopsy clinical parameters in predicting the risk of prostate cancer in a cohort primarily composed of men undergoing repeat biopsy.
Methods:
Men undergoing prostate biopsy at 6 American urology practices from 3 states that comprise a Clinical Validation Center of the National Cancer Institute Early Detection Research Network were prospectively enrolled. This analysis was limited to only patients undergoing a repeat biopsy without a previous diagnosis of prostate cancer. A multivariate logistic regression predicting prostate cancer based on significant pre-biopsy clinical parameters was developed. The performance of the model was evaluated by estimating sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).
Results:
A total of 450 men undergoing repeat biopsy were identified, of which 27% were diagnosed with prostate cancer. 21% of men diagnosed with cancer had a previous abnormal biopsy. In a multivariate model predicting the risk of prostate cancer, age, natural log of prostate specific antigen density (logPSAD), and ASAP were associated with an increased risk (all p<0.05) while number of previous negative prostate biopsies was associated with decreased risk (p<0.05) of prostate cancer. The finding of previous high grade PIN, Atypia, Abnormal DRE, and PSA Velocity were not significant predictors of cancer on repeat biopsy. The AUC for the multivariate model was better than PSA alone (Figure 1; AUC=0.71 vs. 0.53). At 80% sensitivity, specificity was increased from 22% (based on serum PSA alone) to 45% by the model.
Conclusions:
A multivariate predictive model comprising age, prostate size (optimized as logPSAD), number of previous negative prostate biopsies, and previous finding of ASAP may provide a useful platform for carefully selecting patients for repeat biopsy.
Funding: NIH EDRN U01 CVC (Sanda MG, PI)


Back to Program