A multicenter evaluation of the PCA3 molecular urine test: pre-analytical effects, analytical performance, and diagnostic accuracy

BACKGROUND: Measurement of prostate cancer gene 3 (PCA3) mRNA normalized to prostate-specific antigen (PSA) mRNA in urine has been proposed as a marker for prostate cancer. METHODS: We investigated pre-analytical effects, analytical performance, and diagnostic accuracy of a quantitative assay for PCA3. RESULTS: Urine specimens collected without prostate manipulation demonstrated low informative rates. However, specimens collected following digital rectal examinations of 3 or 8 strokes per prostate lobe demonstrated informative rates >94%. Across all urine specimen types, median PCA3 results did not show statistically significant differences (P>0.8). Measurements of controls of known mRNA content demonstrated percent recoveries of 100+/-15% for both PCA3 and PSA mRNAs. PCA3 mRNA total, intra-assay, inter-assay, and inter-site CVs were < or =17.1%, < or =14.0%, < or =9.9%, and < or =3.2%, respectively. Corresponding CVs for PSA mRNA assay were < or =11.5%, < or =8.6%, < or =7.9%, and < or =8.3%. Blinded assay of urines from 72 men with known prostate biopsy outcomes yielded areas under the curve from receiver-operating characteristic analysis of 0.7 at both research sites. Deming regression of individual PCA3 results between sites yielded slope=0.94, intercept=0.48, R=0.9677 (P<0.0001). CONCLUSIONS: The PCA3 assay is insensitive to pre-analytical factors, performs well analytically and correctly classifies a high percent of subjects with known prostate cancer status across research sites.