Background: There are 35 million men ages 50-74 in the United States. The American Cancer Society recommends that men ages 50 and over be screened for prostate cancer. Each year over 20 million Prostate Specific Antigen (PSA) tests are performed, and over one million of these tests are returned positive indicating a risk of prostate cancer. Eight hundred thousand men who have a positive PSA screening test undergo a biopsy where multiple prostate tissue samples are taken and examined by a pathologist. Only 30% of biopsies are confirmed positive for prostate cancer, indicating that a majority represented unnecessary procedures and costs.
 Aim: Identify, using metabolomics, biomarkers to be used as a diagnostic for determining which men with elevated PSA levels have prostate cancer, and which do not.
 Methods: Metabolomics was performed on post-DRE (digital rectal exam), pre-biopsy urine specimens obtained from men at-risk for prostate cancer on the basis of a positive PSA test or DRE. All of the men subsequently underwent a prostate biopsy and the tissue was scored for presence [biopsy+] or absence [biopsy-] of prostate cancer. Metabolomics was performed on a mass spectrometry-based platform including LC-MS-MS and GC-MS instrumentation and automated peak resolution and compound identification software. Small molecular weight metabolites in the mass range of 80-1000 daltons were identified in a non-targeted fashion. Multiple statistical analyses were performed identifying biomarkers correlated with the clinical diagnoses.
 Results: The results showed that the levels of specific metabolites in the urine samples could be used to discriminate men that had prostate cancer from those that did not (determined by later needle biopsy). Seventy metabolites were identified as significantly changed (Wilcoxon rank sum test, p<.05). One significant metabolite was methylglycine (MG, sarcosine). MG correlates mechanistically with a known higher methylation potential reported during tumor development and progression. The performance of MG as a diagnostic biomarker for prostate cancer was analyzed by ROC analysis; the area under the curve was 0.67. In vitro experiments confirmed the role for sarcosine in tumor progression. Addition of MG to benign prostate epithelial cells rendered them invasive as measured by Boyden chamber matrigel invasion assays. In addition, RNAi mediated knock-down of glycine-N-methyltransferase (an enzyme that catalyzes MG production) reduced both MG levels and cell invasive properties. Another metabolite biomarker of prostate cancer was N-acetylasparate (NAA). This biomarker correlates mechanistically with the known behavior of Prostate Specific Membrane Antigen (PSMA). PSMA is a carboxypeptidase overexpressed in prostate cancer whose product is NAA.
 Conclusion: Metabolomics has been employed to identify new biomarkers for diagnosis of prostate cancer in a clinically challenging cohort comprised of high PSA individuals. These biomarkers are being employed to develop a panel of metabolites with diagnostic potential for this deadly disease.
 Supported in part by the Early Detection Research Network (UO1 CA111275-01 to AMC.), National Institutes of Health (RO1CA133458 to AS., TMR and AMC).

Third AACR International Conference on Molecular Diagnostics in Cancer Therapeutic Development-- Sep 22-25, 2008; Philadelphia, PA