Prostate Cancer Gene Expression Adds Prognostic Information beyond Clinical and Pathological Criteria Such as Stage, Gleason Score, PSA, AUA Criteria, and CAPRA Score.
Sara Falzarano, Cristina Magi-Galluzzi, Tara Maddala, Diana Cherbavaz, Fredrick L Baehner, Carl Millward, Eric Klein. Cleveland Clinic, OH; Genomic Health, Inc, Redwood City, CA
Background: Prognosis of localized prostate cancer is estimated using clinical and pathological features including PSA, stage, Gleason score (GS), AUA criteria, and CAPRA score. These criteria do not yet fully account for individual tumor biology. We hypothesized that quantitative gene expression would add prognostic information beyond that provided by clinical and pathological features.
Design: All pts with clinical stage T1/T2 prostate cancer treated with RP at a single institution from 1987 to 2004 were identified (n∼2,600). A cohort sampling design was used to select 127 pts with clinical recurrence (cR) and 374 pts without cR after RP. Surgical GS and clinical data were centrally reviewed. RNA was extracted from 6 manually dissected 10 µm FFPE tissue sections obtained from RP specimens and expression of 732 cancer-related and reference genes was quantified using a standardized RT-PCR assay. Univariate/multivariate analyses of cR, PSA recurrence, and prostate cancer−specific survival (PCSS) were performed using Cox PH regression.
Results: Blocks from 431 pts were evaluable by pre-specified criteria. Median f/up was 5.8 yrs. Pts were mostly Caucasian (83%), <70 yrs old (93%), clinical stage T1 (66%), had baseline PSA <10 ng/mL (82%), and had surgical GS≤7 (87%). In univariate analysis, many genes were significantly (unadj. p<0.05) associated with cR, PSA recurrence, and PCSS, 295, 235, and 203 genes respectively. The number of associated genes was much greater than that expected by chance and the associations were very strong, many with hazard ratios greater than 1.7 per standard deviation change in gene expression. Many genes remained significantly associated with each of the outcomes in multivariate analyses adjusting for pathologic T-stage, tumor specimen Gleason pattern (GP), GS, AUA group and CAPRA score (221, 194, 99, 244, and 271 genes respectively). For the strongest 20 genes, the magnitude of association was diminished by less than 20% after adjustment for stage, GP, AUA, and CAPRA, and less than 50% after adjustment for GS.
Conclusions: This genomic study was notable for the large number of cR events, the use of a standardized quantitative assay, and rigorous central review of pathology and clinical data. These results indicate that quantitative gene expression using RT-PCR adds prognostic information beyond traditional clinical and pathological features.
Category: Genitourinary (including renal tumors)
Monday, February 28, 2011 1:45 PM
Platform Session: Section A, Monday Afternoon