Reduced CD147 expression is linked to ERG fusion-positive prostate cancers but lacks substantial impact on PSA recurrence in patients treated by radical prostatectomy.

The extracellular matrix metalloproteinase inducer CD147 has been suggested as a prognostic marker in prostate cancer. CD147 expression was analyzed by immunohistochemistry on a tissue microarray containing 11,152 prostate cancer specimens. Results were compared to tumor phenotype, biochemical recurrence, ERG status and deletions on PTEN, 3p13, 6q15 and 5q21. CD147 expression was strong in benign prostatic glands and often reduced in prostate cancers. CD147 immunostaining was found in 71.7% of 7628 interpretable cases. CD147 staining was considered strong in 34.6%, moderate in 24.3% and weak in 12.8% of cancers while 28.3% did not show any CD147 reactivity. Reduced CD147 staining was strongly associated with both TMPRSS2-ERG-rearrangement and ERG expression (p<0.0001 each). Within the subgroups of ERG positive and negative cancers, deletions of PTEN, 3p13, 6q15 and 5q21 were unrelated to the CD147 expression status. Decreased CD147 expression was significantly linked to high preoperative PSA values, high Gleason grade, advanced tumor stage (p<0.0001 each), and positive lymph node involvement (p=0.0026) in all cancers. There was a marginal, but statistically significant, association of reduced CD147 expression with early biochemical recurrence (p=0.0296). The significant reduction of CD147 expression in ERG positive prostate cancer provides further evidence for marked biological differences between "fusion type" and "non-fusion type" prostate cancer. Despite a weak association with PSA recurrence, CD147 cannot be considered a relevant prognostic biomarker.

SPINK1 expression is tightly linked to 6q15- and 5q21-deleted ERG-fusion negative prostate cancers but unrelated to PSA recurrence.

BACKGROUND: The serine peptidase inhibitor, Kazal type 1 (SPINK1) has been suggested to define an aggressive molecular subtype of ERG-fusion negative prostate cancer. It was the aim of this study to further study the clinical relevance of SPINK1 expression and its relationship with other key genomic alterations of prostate cancer. METHODS: A tissue microarray containing more than 10,000 prostate cancers with clinical follow-up was used for immunohistochemical SPINK1 analysis. Data on ERG status as well as PTEN, 6q, 5q, and 3p deletions were available for comparison. RESULTS: SPINK1 expression was absent in benign prostate glands and detectable in 5.9% of 9,503 interpretable prostate cancers. Presence of SPINK1 expression was markedly more frequent in ERG negative (10.4%) than in ERG positive cancers (0.3%; P < 0.0001). However, SPINK1 expression was unrelated to tumor phenotype and biochemical recurrence in all cancers and in the subgroup of ERG negative cancers. Further subgroup analyses revealed, however, that--within ERG negative cancers--SPINK1 expression was significantly linked to deletions at 6q15 (P < 0.0001) and 5q21 (P = 0.0042). CONCLUSIONS: Our results exclude SPINK1 as a relevant prognostic prostate cancer biomarker. However, the data demonstrate that SPINK1 overexpression is tightly linked to the small subsets of 6q15- and 5q21-deleted ERG negative prostate cancers. These findings support the concept of molecularly defined subtypes of prostate cancers.