J Cancer 2016; 7(3):297-303. doi:10.7150/jca.12771 This issue

Research Paper

Predicting Prostate Biopsy Results Using a Panel of Plasma and Urine Biomarkers Combined in a Scoring System

Maher Albitar1, Wanlong Ma1, Lars Lund2, Ferras Albitar1, Kevin Diep1, Herbert A. Fritsche3, Neal Shore4 ✉

1. NeoGenomics Laboratories, Irvine, CA;
2. Departments of Urology, Odense University Hospital, Odense, Denmark;
3. Health Discovery Corporation, Savanah, Georgia;
4. Carolina Urologic Research Center, Myrtle Beach, SC, USA.

This is an open access article distributed under the terms of the Creative Commons Attribution (CC BY-NC) License. See http://ivyspring.com/terms for full terms and conditions.
Albitar M, Ma W, Lund L, Albitar F, Diep K, Fritsche HA, Shore N. Predicting Prostate Biopsy Results Using a Panel of Plasma and Urine Biomarkers Combined in a Scoring System. J Cancer 2016; 7(3):297-303. doi:10.7150/jca.12771. Available from https://www.jcancer.org/v07p0297.htm

File import instruction


Background: Determining the need for prostate biopsy is frequently difficult and more objective criteria are needed to predict the presence of high grade prostate cancer (PCa). To reduce the rate of unnecessary biopsies, we explored the potential of using biomarkers in urine and plasma to develop a scoring system to predict prostate biopsy results and the presence of high grade PCa.

Methods: Urine and plasma specimens were collected from 319 patients recommended for prostate biopsies. We measured the gene expression levels of UAP1, PDLIM5, IMPDH2, HSPD1, PCA3, PSA, TMPRSS2, ERG, GAPDH, B2M, AR, and PTEN in plasma and urine. Patient age, serum prostate-specific antigen (sPSA) level, and biomarkers data were used to develop two independent algorithms, one for predicting the presence of PCa and the other for predicting high-grade PCa (Gleason score [GS] ≥7).

Results: Using training and validation data sets, a model for predicting the outcome of PCa biopsy was developed with an area under receiver operating characteristic curve (AUROC) of 0.87. The positive and negative predictive values (PPV and NPV) were 87% and 63%, respectively. We then developed a second algorithm to identify patients with high-grade PCa (GS ≥7). This algorithm's AUROC was 0.80, and had a PPV and NPV of 56% and 77%, respectively. Patients who demonstrated concordant results using both algorithms showed a sensitivity of 84% and specificity of 93% for predicting high-grade aggressive PCa. Thus, the use of both algorithms resulted in a PPV of 90% and NPV of 89% for predicting high-grade PCa with toleration of some low-grade PCa (GS <7) being detected.

Conclusions: This model of a biomarker panel with algorithmic interpretation can be used as a “liquid biopsy” to reduce the need for unnecessary tissue biopsies, and help to guide appropriate treatment decisions.

Keywords: RNA, Cell-free, Gleason, scoring, high-grade, algorithm.