Background: Perturbation of the PI3K pathway in prostate cancer is common and with the development of targeted inhibitors, reliable identification of pathway activation is critical. This study aims to develop an approach integrating both immunohistochemical (IHC) and gene expression profiling (GEP) data to address the heterogeneity in molecular markers of the PI3K pathway in prostate cancer. Methods: A prostatectomy cohort of men from the Physicians\#8217; Health Study (1982-2004) for whom both IHC and GEP data were available comprised the training cohort. Men with prostate cancer from a separate watchful waiting cohort from Sweden (1977-1999) were used for validation. For each tumor sample, IHC expression of markers in the PI3K pathway including PTEN, pAKT (Ser 473), pS6RP (Ser 240/244), and stathmin were examined on tissue microarrays. Intensity and percent positive was evaluated using automated image analysis. A PI3K score ranging from 0-4 was developed for each man comprised of 1 point assigned for each high IHC expression of pAKT, pS6 and stathmin, and low expression of PTEN. GEP for each sample was determined from formalin-fixed, paraffin embedded tissue using cDNA-mediated annealing, selection, extension, and ligation (DASL) assay for approximately 6000 genes. Using a predefined gene-set developed from several curated and published sets for the PI3K pathway, principal component analysis (PCA) was performed to develop models for predicting the score derived from the IHC. The model developed in the training cohort was then tested in the validation set. Results: There were 73 and 281 men in the training and validation cohorts respectively. The correlations between the staining of the IHC markers was higher for pS6RP, pAKT, and stathmin with correlation coefficients between makers from 0.1 to 0.3; PTEN expression was less correlated with the other markers with correlation coefficients from 0.001 to 0.1. In the training set, fully concordant scores of 0 and 4 were seen in 5% and 12% respectively. An increasing IHC score was statistically significantly related to an increasing Gleason score (p= .015). A set of 185 genes from previously identified sets related to the PI3K pathway significantly distinguished between high and low IHC scores. The first principal component from the PCA was significantly related to the IHC score (p=.006). Data relating the IHC and GEP to clinical data in the validation set will be presented. Conclusion: The majority of prostate tumors do not show clearly concordant patterns of IHC staining for makers in the PI3K pathway. A score for the marker staining pattern on IHC was shown to be related to clinical factors. GEP data, using pre-defined gene sets, can discriminate between the IHC staining patterns and could be useful in better classifying the PI3K pathway status in heterogeneous tumors.

Citation Information: In: Proc Am Assoc Cancer Res; 2009 Apr 18-22; Denver, CO. Philadelphia (PA): AACR; 2009. Abstract nr 2745.

100th AACR Annual Meeting-- Apr 18-22, 2009; Denver, CO

American Association for Cancer Research
2009