[1892] Integrated Signature of PI3K Pathway Activation in Prostate Cancer.

Neil E Martin, Jennifer A Sinnott, Stephen Finn, Giuseppe Fedele, Meir J Stampfer, Ove Andren, Sven-Olof Andersson, Lorelei Mucci, Massimo Loda. Dana Farber Cancer Institute & Brigham and Women's Hospital, Boston, MA; Örebro University Hospital, Sweden; Harvard School of Public Health, Boston, MA

Background: The PI3K pathway is commonly dysregulated in prostate cancer (PCa). Tumor heterogeneity and the complexity of the pathway have made it challenging to develop reliable signatures of PI3K pathway activation. We used an integrated approach of 4 immunohistochemical (IHC) markers and gene expression profiling (GEP) data to characterize activation of the pathway across more than 400 tumors.
Design: Men incidentally found to have PCa and enrolled in a watchful waiting cohort from Sweden (1977-1999) were used as the training cohort with a prostatectomy cohort of men from the Physicians' Health Study (1983-2004) used for validation. For each tumor, we determined IHC expression of PTEN, pAKT (Ser 473), pS6RP (Ser 240/244), and stathmin on tissue microarrays using automated image analysis. We created a putative “PI3K score” based on the sum of the quartiles of abnormal staining for each marker. GEP for each sample was determined using the DASL platform for approximately 6100 genes.
Results: There were 304 and 99 men in the training and validation cohorts respectively. In pooled analyses, PTEN (p=0.03), and stathmin (p=0.005) IHC staining were significantly associated with the development of lethal disease. High PI3K scores were significantly associated with Ki67 in both training and validation sets (p<0.01). In the validation cohort, high PI3K scores were borderline significantly associated with increasing Gleason and death or metastatic disease (p=0.05 for both). Using prediction analysis of microarrays, 48 genes were identified in the training set which could significantly distinguish between low and high PI3K scores. When applied to the validation set, the area under the ROC curve was 74% (54-91%) for correct classification of the two states. Gene set enrichment analysis identified a prior in vivo signature of pathway activation as being significantly enriched in the genes segregating high and low PI3K score tumors.
Conclusions: The combined staining pattern of PTEN, pAKT, pS6RP and stathmin appears to be significantly related to proliferation and clinical features of PCa. GEP data supports the hypotheses that these markers capture activation of the PI3K pathway. A signature of PI3K pathway activation could be used in clinical trials investigating agents targeting pathway components.
Category: Special Category - Pan-genomic/Pan-proteomic approaches to Cancer

Monday, February 28, 2011 1:00 PM

Poster Session II # 236, Monday Afternoon


Close Window