Re: Serum proteomic patterns for detection of prostate cancer.

Petricoin et al. (1) proposed that assessment of serum proteomic patterns with mass spectrometry and combination of these patterns with a bioinformatic algorithm can detect prostate cancer with approximately 95% sensitivity and 70%–80% specificity. These figures seem favorable when compared with common biochemical testing for prostate cancer, which currently includes analysis of total prostate-specific antigen (PSA) and percent free PSA in serum. Similar approaches have been proposed by the same group for the diagnosis of ovarian and other cancers (2). The authors speculate that the discriminatory proteomic patterns originate from proteins (including cytokines, chemokines, metabolites, or enzymatic cleavage products) released by the prostate or its microenvironment during blood percolation through the organ. However, a limitation of this technology, also discussed elsewhere (3), is that the authors offer no evidence in regard to the identity of their seven most discriminatory molecules. In a previous publication (4), the authors contend that the identity of these proteins is not important in the context of their method. However, the identity of such proteins, as well as their relative concentrations in serum, is of paramount importance for understanding why they offer discriminatory power for detecting prostate cancer and how they originated in the circulation. For example, to derive their algorithm, Petricoin et al. (1) used serum samples from patients with biopsyproven prostate cancer and total PSA levels of more than 4 ng/mL and from patients with no evidence of disease and total PSA levels of less than 1 ng/mL. It would have been interesting to know whether the authors’ method identified this internal control and highly discriminating molecule (i.e., PSA) as one of the discriminatory proteins in serum. Such a finding would confirm that their method is sensitive enough to detect proteins at a concentration level of 1 ng/mL or higher. Their method of extracting serum prior to analysis (i.e., by hydrophobic interaction with a C16 column),