Does Androgen Receptor Have a Role in a Subgroup of ER Negative Breast Cancer?
ME Edgerton, S Sanga, E Downs-Kelley, V Cristini. UT MD Anderson Cancer Center, Houston, TX; University of Texas, Austin, TX; UT Houston Health Science Center, Houston, TX
Background: Previously Doane et al and Farmer et al, postulated a subgroup of ER negative breast cancers that demonstrated gene expression profiles of Androgen Receptor (AR) signaling. The series were both small, with 6/22 and 10/41 patients that demonstrated the AR molecular profile in each publication. The two groups were not rigorously compared in the original reports, which were published nearly simultaneously.
Design: We combined the gene expression data from both sources, performed cross-platform normalization, and updated the probe definitions to reduce institutional bias and improve signals. We compared the presumed AR signaling subgroups across the two institutions to determine if they were equivalent, used other network inference methods to test the hypothesis that they demonstrate expression patterns that support AR signaling, and built a model to identify more patients and examine survival.
Results: We were able to build a model of presumed AR positive molecular profiles from one data source (e.g. Doane) and use it to predict the same population in the other data source (e.g. Farmer) with high accuracy. Combined data from Doane and Farmer supported AR as a signaling network that was active in that subgroup. We identified more data with presumed AR signaling from published ER negative data in the public domain that had survival data. We built a Kaplan-Meier curve for this patient population that showed poor survival; however, all of the patients for whom we could identify survival data and who expressed AR signaling network were also her-2neu positive.
Conclusions: Our results show that 1) the subgroups identified by Doane and Farmer have the same molecular profile; 2) network inference methods support a role for AR signaling in this subgroup; 3) survival data examined thus far may be biased by her-2neu status as all AR signal positive cases identified with survival data were also her-2neu positive. Doane and Farmer each reported only half of the patients with an AR signaling profile were her-2neu positive, while the other half were triple negative; however, they did not publish survival data. Previous IHC work by Ellis et al supports AR as a positive prognostic marker in the triple negative phenotype. This work supports a potential role for therapeutic intervention using AR inhibition in breast cancer.
Category: Special Category for 2009 - Pan-genomic/Pan-proteomic approaches to Cancer
Monday, March 9, 2009 1:45 PM
Platform Session: Section G, Monday Afternoon