Poster T2 - Exploiting therapy response dynamics to overcome fibroblast-mediated drug resistance in HER2+ breast cancer

Introduction: Breast cancer is the most commonly diagnosed cancer and is the second most deadly form of cancer in women and will account for an estimated 310,000 new diagnoses and 42,000 deaths this year. Human epidermal growth factor 2 overexpressing (HER2+) breast cancer accounts for 15-20% of patients diagnosed with invasive breast cancer. Recurrence and resistance to HER2-targeted therapies remain current treatment challenges and an estimated 57% of patients do not achieve pathologic complete response. The tumor microenvironment provides supportive cues that regulate tumor therapy response. Stromal fibroblasts are the most abundant cell type in the tumor microenvironment and their presence is associated with poor survival outcomes and HER2-targeted therapy resistance. In preclinical models, fibroblasts reduce tumor cell sensitivity to the HER2-targeted therapies by sustaining proliferation and survival signaling via paracrine factors. The impact of fibroblasts on tumor cell growth and death dynamics during treatment remains poorly understood. To address this need, we employed an experimental-computational approach in which we calibrated a mathematical model to experimentally-measured tumor growth and death dynamics to predict how different these dynamics affect the number of residual tumor cells after treatment and develop strategies to improve treatment efficacy. An improved understanding of these dynamics may unlock new therapeutic strategies for overcoming fibroblast-mediated therapy resistance.

Materials and

Methods: To interrogate the impact of fibroblasts on therapy response dynamics of breast tumor cells, we cultured a panel of nine HER2+ breast tumor cell lines either in monoculture (2000 cells per well in 96-well plate) or coculture with fibroblasts at various seeding densities of AR22 mammary fibroblasts (500-4000 cells per well) and treated these cultures with the HER2-targeted therapy lapatinib (0-3µM) for four days. Tumor cells were engineered to express the nuclear label H2B-GFP and incubated with the cell death marker Sytox Deep Red to facilitate detection of live and dead tumor cells. The number of live and dead tumor cells was measured every four hours using time-lapse microscopy to capture the growth/death trajectories of tumor cells. An ordinary differential equation (ODE)-based mathematical model that captures the impact of lapatinib and fibroblasts on tumor cell growth and death rates was calibrated using these experimental data to predict tumor cell fate for a specified lapatinib concentration, treatment duration, and fibroblast seeding density. Four variants of cell death dynamics in this mathematical model (constant, increasing, decreasing, or transient death rate with time) were fitted to each cell line. Akaike information criterion, a metric for model fit, was compared for each model and the best-fitting model was selected for each cell line. Simulations were performed in the treatment space (lapatinib dose, duration, and fibroblast density) to identify cell line-specific conditions in which fibroblasts protected tumor cells from death and to evaluate conditions to maximize tumor cell death in the presence of fibroblasts.

Results, Conclusions, and Discussions: Results
Coculturing tumor cells with fibroblasts reduced tumor cell sensitivity to lapatinib at doses greater than 100nM for five of the nine HER2+ breast tumor cell lines: BT474, EFM192, HCC1569, HCC202, and MDA361. The impact of fibroblasts on lapatinib-mediated growth inhibition and cytotoxicity was cell line specific. For HCC1569, HCC202, and MDA361 cells that were primarily growth inhibited by lapatinib, fibroblasts reduced growth inhibition, whereas for BT474 and EFM192 tumor cells fibroblasts reduced the cytotoxic effects of lapatinib (Fig.1A). The protective effects of fibroblasts on tumor cells increased with fibroblast seeding density. We fitted four variants of an ODE-based model that considers the impact of fibroblast-mediated protection and lapatinib on tumor growth dynamics to the experimentally measured tumor growth and death trajectories (Fig.1B). Treatment response dynamics varied across cell lines and were captured by our model (Fig.1C). The transient death rate and increasing death rate models had the lowest AIC for EFM192 and MDA361, respectively, indicating these models best explained the death dynamics of these cell lines. We performed four-day treatment simulations for each cell line and varied lapatinib dose and fibroblast seeding densities and measured the number of surviving tumor cells at the end of the treatment simulation. We found that above a sufficient fibroblast seeding density (specific to each cell line), no cell death was induced by lapatinib even at high concentrations (Fig.1D). Additionally, increasing treatment duration was effective at overcoming this resistance for some cell lines.

Conclusions
We evaluated the impact of fibroblasts on HER2+ breast tumor cell response dynamics to the HER2-targeted therapy lapatinib and found that fibroblasts mitigate the growth inhibitory and cytotoxic effects of lapatinib in a cell line dependent manner. We developed a predictive mathematical model for each cell line to identify conditions that limit lapatinib efficacy and develop cell line specific strategies to maximize therapy response.
Discussion
Developing treatment approaches to overcome fibroblast-mediated therapy resistance is imperative to improving outcomes for HER2+ breast cancer patients. To this end, an understanding of how fibroblast density and tumor death dynamics contribute to poor therapy response is key to informing more effective therapeutic strategies.

Acknowledgements (Optional):