Gene expression analysis in combination with clinical parameters is effective in predicting relapse after radical prostatectomy

Early prediction of tumour recurrence at the time of radical prostatectomy (RRP) would certainly be of great importance for deciding possible adjuvant therapy in patients with prostate cancer (CaP). Clinical and pathological parameters can correctly predict recurrence in no more than 70% of cases. Gene expression analysis is a powerful method for studying prostate cancer biology. Recent studies have been directed to identifying new genes implicated in CaP initiation and progression and establishing novel methods for validation of differentially expressed genes in cancers at moderate and high risk of progression. The objectives of this study were to evaluate whether (a) the assessment of the expression profiles of a set of genes, directly or indirectly related to cell proliferation state or apoptic activity, can correctly predict relapse of CaP at the time of RRP and (b) a combination of gene expression profile data with standard clinical-pathological parameters can further enhance the prediction potential. The study was conducted on 355 patients with localised CaP, undergoing RRP, but only 23 pairs of prostate specimens (tumour and matched benign tissues of the same gland) were eligible for the analysis. Age, familiarity, PSA, prostate volume, stage and grade were evaluated for prognosis as clinical and pathological parameters. The level of expression of a panel of 8 genes related to polyamine metabolism, cell proliferation and apoptosis were measured in tumours and matched normal tissues. After undergoing RRP, patients were regularly followed up at our center with physical examination (PE) and PSA measurement. After a mean time of 62.3 months 13 out of 23 patients (56.5%) did not recur (Group 1), while the remaining 10 patients (43.5%) had a relapse (Group 2). When patients were classified according to clinical-pathological parameters, the overall correct prediction of recurrence was 87% (Group 1 = 100%, Group 2 = 70%). When patients were classified according to gene profiling alone the overall correct prediction was 82.6% (Group 1 = 84.6%, Group 2 = 80%). When combining gene profiling and clinical-pathological parameters a 95.7% prediction rate was obtained (Group 1 = 100%, Group 2 = 90%)