Similar to NY Prostate Cancer Conference - S. Stone - Session 1: Cell cycle progression genes differentiate indolent from agressive prostate cancer (20)
2. Prognostic Markers and Clinical Utility Prostate cancer has a natural history that is highly variable and difficult to predict. Overtreatment is widely recognized as a financial burden to the health care system and leads to unnecessary morbidity. Inappropriate conservative treatment has consequences, as prostate cancer is the second or third most common cause of cancer death in the western world. Significant clinical need for prognostic biomarkers that can aid physicians in predicting disease outcome.
3. PrognosticBiomarkers Genetic variants BRCA1&2, 8q24, KLK2-3 region Genomic rearrangements TMPRSS2-ERG Genomic instability or copy number variation (GEMCap) RNA expression PCA3 Multi-gene expression signatures
4. Prognostic RNA Signatures Early attempts to develop prognostic signatures in prostate cancer had some success, but were typically done with small cohorts (microarray data which required frozen tissue). Also, many signatures were never validated or not clearly shown to add discrimination beyond clinical parameters. To date, no signature has changed clinical practice. In contrast, prognostic RNA signatures have impacted clinical care in breast cancer.
5. Prognostic Signatures in Breast Cancer Mosley and Keri (BMC Med. Genomics 2008) compared the performance of 6 validated prognostic signatures. While similar in performance, the signatures shared only a few genes in common. However, the signatures all contained several genes related to cell cycle progression (CCP genes). Without contribution of CCP genes, the signatures could not be rigorously associated with outcome. 5
6. Cell Cycle Progression Genes CCP gene encode products that are required during cell cycle progression (mostly M phase). They are regulated, at least in part, at the level of RNA expression (Whitfield et al. Mol. Biol. Cell 2002). Expression levels measure fraction of cells within sampled tissue that are actively dividing. Since CCP genes measure a single biological process, their expression levels should be correlated. 6
7. CCP Gene Selection Selected 126 candidate CCP genes. Designed expression assays and ran candidate genes on FFPE sections from 96 anonymous prostate tumors. Selected 31 genes that were best correlated with mean of 126 candidate genes. 7
8. Question Are CCP genes prognostic in prostate cancer, and do they add significant information to clinical parameters? Prostatectomy cohort from the U.S. Conservatively managed TURP cohort from the UK
9. Prostatectomy Cohort Patient samples were collected between 1985 and 1995 (Swanson et al. Urologic Oncology 2007). Median clinical follow-up of 9.5 years. All patients were evaluated for stage, Gleason score, and post-surgical pathology. At surgery, a majority of patients (66%) were classified as cT2. Median preoperative PSA was 6.6 ng/ml. Median age at surgery was 67 years. Overall, 37% had biochemical recurrence and the median time to recurrence was 2.4 years. 9
10. Prostatectomy Study Design We selected 410 subjects with available tissue. Determined CCP signature score by calculating the mean expression for the entire set of 31 genes, normalized by 15 housekeepers. Primary endpoint was biochemical recurrence. Used Cox proportional hazards model for time-to-recurrence, which incorporates information about censored patients (i.e. those lost to follow-up). p-values were based on likelihood ratio test of reduced model (covariates only) vs. full model (covariates + mean of expression). 10
11. Summary of Univariate Analysis CCP score was predictive of biochemical recurrence (N = 366, 89% of available samples). HR 1.89 per CCP unit (1.54, 2.31), p-value =5.6x10-9 11
14. TURP Cohort Diagnosed with cancer between 1990 and 1996 (Cuzick et al. BJC 2006). Median follow-up was 9.75 years. Median age at diagnosis was 70 years. Exclusion criteria included: age (≥ 76), treatment within 6 months, or metastasis/death within 6 months of Dx. Median PSA was about 10 ng/ml, Median Gleason score 7. 24% had died from prostate cancer at 10-years. 14
15. TURP Cohort Study Design We randomly selected398 patients. Experimental plan was completed and signed prior to any analysis. We calculated CCP score and sent data to QMUL for analysis. CCP score was used to predict death from prostate cancer. Multivariate analysis included PSA and Gleason score. 15
16. Summary of Univariate Analysis CCP score was predictive of prostate cancer specific death (N=337, 85% of available samples). HR 2.9 per CCP unit (2.38, 3.57), p-value =6.1 x 10-22 16
19. Hazard Ratio of CCP on Prostate Cancer Death by Gleason Score Gleason <7 Gleason =7 Gleason >7 Combined .1 .2 .5 1 2 5 10 Hazard ratio 19
20. Clinical Utility of CCP Score Primary clinical utility of CCP score is to aid physicians in selecting appropriate treatment at diagnosis. Does CCP score change patient prognosis in the conservatively managed TURP cohort compared to clinical parameters only for predicting prostate cancer death at diagnosis? 20
22. Conclusions CCP expression score is a strong univariate predictor of disease outcome after surgery and at diagnosis. CCP score is robust to clinical setting and patient composition. CCP score consistently adds significant prognostic information beyond clinical parameters. Studies aimed at extending these results to needle biopsies and additional clinical settings are warranted and ongoing.