A multivariate index analyte (MIA) assay for differentiating aggressive from nonaggressive prostate cancer.

Abstract

e588 Background: Biomarkers that can assist clinicians and patients to proceed when PSA and /or DRE are equivocal. Such biomarkers should establish both sensitivity and specificity for prostate cancer detection in order to improve go-forward decisions to perform prostate biopsy. Following the successful use of a three-protein marker panel to increase the specificity of prostate cancer detection1 we have now used the same technology to examine whether an MIA assay can assist in differentiating aggressive from non-aggressive cancer in prostate cancer patients. Methods: Samples from patients with either aggressive prostate cancer or non- aggressive prostate cancer were obtained from two sources. The cohort criteria comprised of serum samples where blood was drawn from patients with adenocarcinoma and a PSA greater than or equal to 2ng/mL. All men were Caucasian with the exception of 3 who were African American. Non-aggressive prostate cancer was defined as having a Gleason score of 6 (n = 35) and aggressive prostate cancer was characterized as Gleason score 7 and above (n = 69). Biomarker levels were determined using a plate based ELISA for GPC-12 and a bead-based MIA assay for the other markers. Results: By using biostatistical analysis (Simplicity Bio, Switzerland) two models were identified that were able to differentiate between aggressive and non-aggressive prostate cancer. One consisted of a combination of 5 analytes and the other used 6 analytes. Model 1 containing PSA and GPC-1 plus 4 analytes produced a combined sensitivity of 81% and specificity of 78% (AUC 0.81). The second model comprising of GPC-1 with an additional 4 analytes achieved a sensitivity of 72% with a specificity of 76% (AUC 0.76). Both models had a p value of less than 0.05. By itself PSA was a poor predictor of prostate cancer with a sensitivity of 58% and specificity of 43% (AUC 0.55). Conclusions: The analytes identified by the two statistical models demonstrate potential utility for using the combined markers as a new means of differentiating aggressive prostate cancer from non-aggressive cancer. An additional study to further validate these models is currently being constructed.