Blood-based biomarkers for detecting aggressive prostate cancer at time of biopsy

Abstract

4637 Background: To date, we have applied our unique methodology (the Sentinel Principle) to identify blood-based gene expressed biomarkers for several diseases including osteoarthritis, bladder cancer and psychiatric disorders. It is well known that new biomarkers for prostate cancer detection are needed, particularly aggressive forms of prostate cancer. Our objective was to identify gene expression signatures and to characterize a set of biomarkers from whole blood to identify patients with aggressive forms of prostate cancer at the time of prostate biopsy. Methods: We conducted a two staged study. The first stage was for gene discovery using microarrays (Affymetrix U133Plus2.0 GeneChips) among 47 patients (all Caucasian) who underwent a prostate biopsy for prostate cancer. We included 16 patients (cases) diagnosed with aggressive forms of prostate cancer defined as having a histologic grade of Gleason Score 7 or more and 31 patients with no evidence of cancer at biopsy (controls). The second stage was a validation study (108 patients) using real-time RT-PCR methods. Logistic regression was used to assess the ability of linear combinations of specific transcripts to distinguish cancers from controls. Results: In the first stage, we identified 1661 probes that were significantly different in blood gene expression profiles between cases and controls (p<0.05). In the second stage, real-time RT-PCR assays validated four genes among 33 cases and 75 controls: three up-regulated genes: A 1.4 fold, p<0.001, B 1.2 fold, p=0.03, C 1.3 fold, p=0.08; and one down-regulated gene D 0.70, p=0.008. Linear combination of these 4 genes discriminated the cases from controls with a sensitivity of 90% and specificity of 53% (AUC: 0.80, 95% C.I.: 0.713∼0.871). The performance of PSA (4 ng/ml cut-off) in the same 108 sample set was 91% sensitivity and 13% specificity (AUC: 0.67, 95% CI: 0.75∼0.98). The performance of combining the four blood biomarkers with PSA yielded an AUC of 0.86 (95% C.I.: 0.77∼0.91) with a sensitivity 90% and specificity 70%. Conclusions: When used with PSA, these four blood-based biomarkers improved the overall accuracy to identify patients with prostate cancer at biopsy. These results would have important implications for new biomarkers discovery for prostate cancer detection. [Table: see text]