Multiplex panel of metabolomic markers for early detection of prostate cancer.

Abstract

13 Background: Prostate cancer (PCa) is the second leading cause of cancer deaths among men in US. Prostate specific antigen (PSA) the clinical standard for early detection of PCa is not specific for the disease necessitating development of additional markers. Metabolites are stable, measurable in biological fluids and function as physiological indicators of the individual. Earlier we described the metabolomic alteration in PCa and nominated sarcosine as a tissue marker for PCa progression. Further, elevated sarcosine levels were seen in cancer urine. The current study describes a multiplex panel of metabolomic markers in urine that include sarcosine, for early detection of PCa. Methods: A sensitive liquid chromatography- mass spectrometry based assay was used to quantify relative levels of tissue specific metabolites in urine of PCa patients. The urine specimens examined in this study were collected at University of Michigan from post DRE and pre-biopsy from clinically challenging patients with PSA between 4-10 ng/mL where biopsy was either positive or negative for tumor. The urine supernatants were recovered post-centrifugation, adjusted to have an osmolarity between 350-400 mOsm/kg, spiked with defined amount of internal standards and evaluated by multiple reaction monitoring (MRM) using a triple quadrupole mass spectrometer. The MRM data was examined using random-forest (RF)-to generate a panel of discriminatory compounds that were validated in an independent test set. Results: A panel of 11 metabolites (glutamic acid, sarcosine, spermidine, histidine, methionine, adenosine, inosine, thymine, hydroxyphenyl lactic acid, acetyl valine and lauric acid) when employed in a RF-based 2-stage training model (n=63) delineated PCa and benign with a cross-validated (out-of-bag validation based on 20000 decision trees) accuracy of 76 %. When applied to an independent set of test samples (n=98) from the same cohort the accuracy, sensitivity, specificity, positive predictive value and negative predictive values were 72 %, 77.5 %, 62.5 %, 67.8 % and 73.2 % respectively. Conclusions: The study describes a first-generation multiplex panel of metabolic markers for early detection of PCa in urine supernatants. No significant financial relationships to disclose.