Abstract Crosstalk and compensatory circuits within cancer signaling networks fundamentally limit the activity of targeted therapies. Combination drug regimens are thus required to fully inhibit an oncogenic network. However, rationally designing optimal combinations, given the large number of targeted agents available remains a challenge. Previously, we developed a multi-scale systems model of ERBB2-high breast cancer to quantitatively interrogate relationships between biomarkers (proxies for the mode of network activation) and combination drug efficacy. We found that targeting the ERBB2-ERBB3 heterodimer with a combination of ERBB2 inhibitors (trastuzumab and lapatinib) and an ERBB3 inhibitor (MM-111) was more effective at inducing tumor regression than the combination of an AKT and MEK inhibitor (MK2206 and GSK1120212), and was significantly better tolerated. However, the model was parameterized based on data collected from one cell line. Here we have performed profiling of an 18-cell line panel to assess whether our model and results are extendable to other ERBB2-high cancers. We monitored cell viability and signaling events upon AKT and MEK inhibitor treatments in the presence or absence of the ERBB3 ligand heregulin in these cell lines. While all cells were ERBB2+, we observed widely variable phenotypic and signaling dependencies on the PI3K/AKT and MAPK/ERK pathway activities across the panel. At the phenotypic level, cells primarily depend on either AKT or ERK signaling in basal conditions. Interestingly, upon heregulin stimulation some cells lines switch pathway dependency from AKT to ERK. Adaptive feedback circuits downstream of ERK and AKT were identified in all cell lines, though the identity and strength vary extensively. ERBB3 signaling and total AKT were consistently up-regulated to various degrees upon AKT inhibitor treatment. In contrast, multiple ERBB receptors as well as other RTKS, AKT signaling, and total ERK were upregulated in the response to MEK inhibition. While it is plausible that signaling pathways beyond AKT and ERK are modulating cell viability, we are able to quantitatively describe cell growth regulation based on AKT and ERK pathway activity using quantitative logic-based modeling framework. Our results will help us better understand how signaling events are decoded by cancer cells into phenotypic responses, and enable in silico drug combination screening across molecularly and functionally heterogeneous cancers. Citation Format: Jinyan Du, Daniel Kirouac, Johanna Lahdenranta, Ryan Overland, Matthew Onsum, Charlotte McDonagh. In silico design of biomarker-optimized drug combinations in ERBB2+ cancers. [abstract]. In: Proceedings of the Third AACR International Conference on Frontiers in Basic Cancer Research; Sep 18-22, 2013; National Harbor, MD. Philadelphia (PA): AACR; Cancer Res 2013;73(19 Suppl):Abstract nr A10.