Early detection of prostate cancer requires significant improvement. Prostate specific antigen (PSA) and other methods fall short in distinguishing benign and malignant prostates, or defining whether the cancer is aggressive or indolent. In contrast, metabolite profiles in non-invasive samples might be better suited to detect and provide insight into the pathophysiology of prostate cancer.
We investigated seminal fluid (SF) and post-ejaculate urethral washings (PEUW) from 155 patients over 5 years. 1D NOESY spectra were acquired at both 500 and 900 MHz. 2D spectra were also obtained to assist metabolite identification. Following initial experiences with glucose contamination in SF samples, the “add-to-subtract” method was used to reduce the confounding influence of glucose. Spectral alignment (icoshift) and data reduction (using standard and optimized bucketing methods) were performed prior to multivariate statistical analysis. Principal components analysis, partial least squares and orthogonal projections to latent structures were used to investigate metabolite correlation for different clinical variables.
Metabolomic analysis was initially confounded due to exogenous metabolites (ethanol, xenobiotics). Appropriate exclusion of confounders allowed identification of known and new metabolites to improve insight into sample and disease processes. Preliminary results have indicated that combining metabolomic data from SF and PEUW promise to improve detection rates. Furthermore, monitoring of clinical progress for patients with initially negative prostate biopsies has allowed for incorporation of changing cancer status with time.
Our results suggest that NMR-based metabolomics is an ideal method to facilitate non-invasive detection of prostate cancer. Detection is improved using clinical and metabolic markers from excreted biofluids, bringing us one step closer to developing an accurate, non-invasive screening test for prostate cancer.