论文部分内容阅读
Background: The natural history of prostate cancer is highly variable and difficult to predict.Improved tools are needed to match treatment more appropriately to a patients risk of progression.Therefore,we developed an expression signature composed of genes involved in cell cycle progression(Prolaris)and tested its utility in prostate cancer.Methods: We developed an expression signature composed of 31 cell cycle progression and 15 housekeeper genes.An expression score(Prolaris score)was derived as the mean of all cell cycle progression genes.The signature was tested at disease diagnosis in two conservatively managed cohorts from the UK(N=337 and 349),after radical prostatectomy in two cohorts from the U.S.(N=366 Scott & White Hospital,TX and 413 USCF,CA),and after external beam radiation therapy(N=141)in a cohort from Durham VA Medical Center.All studies were retrospective.Results: The cell cycle progression signature was a highly significant predictor of outcome in all five studies.In conservatively managed patients,the Prolaris score was the dominant variable for predicting death from prostate cancer in univariate analysis(p = 6.1 x 10-22 after diagnosis by TURP,and p = 8.6 x 10-10 after diagnosis by needle biopsy).In both studies,the Prolaris score remained highly significant in multivariate analysis making it a stronger predictor of disease-specific mortality than other prognostic variables.After prostatectomy,Prolaris predicted biochemical recurrence(BCR)in univariate analysis(S&W p = 5.6 x 10-9; UCSF p= 2.23 x 10-6)and provided additional prognostic information in multivariate analysis(S&W p = 3.3 x10-6; UCSF 9.5 x10-5).After radiation therapy,Prolaris predicted BCR(Phoenix)in univariate(p=0.0017)and multivariate analysis(p=0.034).In all five studies the HR per unit change in the Prolaris score was remarkably similar,ranging from 1.89 to 2.92,indicating that the effect size for the Prolaris score is robust to clinical setting and patient composition.Conclusions: The Prolaris test predicts prostate cancer outcome in multiple patient cohorts and diverse clinical settings.In all cases,it provides information beyond clinicopathologic variables to help differentiate aggressive from indolent disease.