Abstract P2-09-34: An mRNA-based method to measure PI3K activity in cancer tissue using a computational pathway model to assess FOXO transcriptional activity

Introduction The PI3K signaling pathway is frequently active in breast cancer, and therapeutic inhibitors have been developed. However, it has proven difficult to correctly predict treatment response. We developed a method that measures functional activity of the PI3K pathway using a computational model that infers transcriptional FOXO activity (downstream of PI3K) from expression levels of its target genes. In principle, PI3K pathway activity inhibits transcriptional FOXO activity, hence inactive FOXO is indicative of active PI3K. Method We developed a knowledge-based computational model to infer transcriptional FOXO activity from cancer tissue mRNA expression levels, using a Bayesian network approach (Verhaegh et al., Cancer Res 2014). Model calibration was done on microarray data from HUVEC cells with inducible FOXO3.A3-ER (GSE16573). Results The FOXO model was biologically validated with in-house microarray data from independent breast cancer cell lines. ER positive, PIK3CA E545K mutant MCF7 and triple negative MDA-MB-231 cells were stably transduced with a doxycycline inducible FOXO3.A3 expression vector, allowing controlled induction of FOXO3 protein activity. FOXO activity was determined to be low in untreated and 20% FBS treated MCF7 cells, and high after doxycycline, LY294002, and combination treatment. Next, we tested our FOXO model on independent MCF7, BT-20 and MDA-MB-453 cell line data treated with EGFR inhibitor erlotinib (GSE30516), showing an increase of FOXO activity upon treatment, due to reduced PI3K pathway activity (combined Wilcox rank sum test p = 7.8x10 −5 ). We further analyzed independent publicly available data from breast cancer patients. FOXO was generally active in healthy breast tissue. Compared to healthy breast tissue, FOXO activity was higher in normal-like and luminal A breast cancer samples (p = 1.9x10 −6 and 0.025, resp.), and lower in luminal B samples (p = 4.2x10 −7 ). In addition to the above mechanism for regulating FOXO activity, literature suggests that FOXO can also be activated by cellular oxidative stress, which is often associated with PI3K signaling. This may be assessed using expression levels of the FOXO target gene SOD2, which is differentially expressed between the two FOXO activity modes, and whose function is to reduce oxidative stress. Public data shows an increasing percentage of elevated SOD2 levels among FOXO-active samples with increasing breast cancer aggressiveness: 7% in normal-like, 5% in luminal A, 18% in luminal B, 31% in HER2-enriched and 74% in basal like breast cancer. Conclusion Our computational model to measure PI3K activity using FOXO target gene mRNA levels was able to measure increased FOXO activity in multiple cancer cell lines after PI3K inhibition. FOXO activity was measured high in healthy breast tissue and in normal-like and luminal A breast cancer, and lower in luminal B, indicating PI3K activity in the latter group. In more aggressive subtypes, FOXO activity was increasingly accompanied by high SOD2 expression, suggesting oxidative stress with associated PI3K activity as the FOXO activating mechanism. Clinical utility for improved response prediction and monitoring of PI3K pathway inhibitors is being investigated with clinical partners. Citation Format: Verhaegh W, van Ooijen H, Hornsveld M, Dam C, Eijkelenboom A, Dou M, Velter R, Burgering B, van de Stolpe A. An mRNA-based method to measure PI3K activity in cancer tissue using a computational pathway model to assess FOXO transcriptional activity [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-09-34.