LBA23 - Derivation and validation of blood mRNA expression signatures to stratify castration resistant prostate cancer patients and predict poor outcome

Date 30 September 2012
Event ESMO Congress 2012
Session Genitourinary tumors, prostate II
Topics Prostate Cancer
Translational Research
Basic Principles in the Management and Treatment (of cancer)
Presenter David Olmos Hidalgo
Authors D. Olmos Hidalgo1, D. Brewer2, G. Attard1, D. Danila3, J. Clark2, C. Parker4, E. Castro5, M. Fleischer3, A.H.M. Reid1, S. Sandhu6, R.J. Jones7, C.S. Cooper2, H.I. Scher8, J.S. De Bono9
  • 1Prostate Targeted Therapy Group, The Royal Marsden NHS Foundation Trust and Institute of Cancer Research ICR, SM25PT - Sutton/UK
  • 2Prostate Cancer Molecular Carcinogenesis, The Institute of Cancer Research, Sutton/UK
  • 3Genitourinary Oncology Service, Memorial Sloan Kettering Cancer Center, 10065 - New York/US
  • 4Academic Urology, Royal Marsden Hospital NHS Foundation Trust, SM2 5PT - Sutton/UK
  • 5Prostate And Gu Clinical Research Unit, The Institute of Cancer Research ICR, SM2 5NG - Sutton/UK
  • 6Prostate Targeted Therapy Group/ Ddu, Royal Marsden Hospital, SM25PT - Sutton/UK
  • 7Medical Oncology, Beatson West of Scotland Cancer Centre Gartnavel General Hospital, G12 0YN - Glasgow/UK
  • 8Medicine, Memorial Sloan Kettering Cancer Center, 10065 - New York/US
  • 9The Royal Marsden NHS Foundation Trust and Institute of Cancer Research ICR, SM25PT - Sutton/UK


Background: Biomarkers are urgently required to dissect prostate cancer (PrCa) inter-patient disease heterogeneity to improve treatment and accelerate drug development. We analyzed blood mRNA expression arrays to identify metastatic castration resistant PrCa (CRPC) with poorer outcome.
Methods: Whole blood was collected into PAXgeneTM tubes from CRPC patients and PrCa patients selected for active surveillance (AS). In Stage I (derivation test-set) 69 CRPC patients were used as cases and 31 AS patients as controls; in Stage II (validation-set) 70 CRPC patients were evaluated. Whole blood RNA from patients in Stage I was hybridised to Affymetrix U133plus2 microarrays. Expression profiles were analysed using Bayesian Latent Process Decomposition (LPD) to identify RNA expression profiles associated with CRPC subgroups and prognosis. A reduced gene signature was then derived using Random Forest algorithm and later verified (Stage I) and validated (Stage II) by qRT-PCR studies. All p values were corrected for false discovery rate.
Results: LPD analyses of the mRNA expression data divided the evaluable patients in stage-I (n=94) into 4 groups. LPD1 and LPD2 consisted almost entirely of CRPC patients (14/14; 17/18); LPD3 and LDP4 comprised similar proportions of CRPC patients (15/31; 12/21) vs controls (AS patients). LPD1 group patients had features of worse prognosis CRPC and poorer overall survival (OS) than CRPC patients in other LPD groups (p=0.00007). A 9 gene signature (TERF2IP, TMCC2, SNCA, GABARAPL2, RIOK3, TFDP1, HMBS, SLC4A1, STOM) classified patients into this LPD1 group with a very low misclassification rate (1.2%). After verification of the signature by qRT-PCR, LPD1 membership was associated with worse OS in the derivation CRPC cohort (10.7 vs 25.6 months, p=0.00001). Importantly, the prognostic value of this signature was confirmed in the validation CRPC cohort, where LPD1 membership was also associated to poor OS (9.2 vs 21.6 months, p=0.001), and remained an independent prognostic factor in multivariable-analyses for both cohorts. The LPD1 signature was associated with bone marrow disruption (mobilization of early erythroid cells) and decreased immune response.
Conclusion: Gene expression signatures derived from whole Blood genome profiling identify CRPC patients with very poor outcomes.