Abstract 2705: A computational algorithm to predict tumor growth and cancer stem cell proportion in-vitro and in-vivo from single-cell observations

Abstract

Abstract We have developed a new mathematical algorithm and corresponding computer software that uses observations of single cell behavior to predict the growth of the cancer stem-like cell proportion in larger cancer cell groups. Cancer stem-like cells (CSCs) have been implicated in ovarian cancer tumor growth, chemotherapy resistance, and disease recurrence. Aldehyde dehydrogenase (ALDH) is a primary discriminator of CSCs in ovarian cancer as well as in other cancer types. Unfortunately, the rarity and ability of CSCs to rapidly differentiate makes them difficult to study in-vitro and in-vivo. Microfluidic capture devices now allow us to grow and evaluate single ovarian cancer cells in isolated culture. We deployed microfluidic devices to evaluate the different self-renewal and asymmetric division patterns of ALDH+ and ALDH(-) ovarian CSCs. We analyzed data gathered from arrays of single cell chamber observations and found that purified ALDH+ cells were more proliferative than ALDH(-) cells in both cell-line (n = 112, p < 0.001) and primary (n = 41, p = 0.008) ovarian cancer specimens. Importantly, ALDH+ cells could produce both ALDH+ and ALDH(-) cells, whereas ALDH- cells were only able to produce ALDH- cells. Based on this hierarchy-defining data, we developed an easy to use computer algorithm on the freely-available software R to predict cancer cell population growth in-vitro and in-vivo. In our algorithm, size changes of cell populations are simulated by drawing from observed events. We compared the predictions from our hybrid microfluidics chip and computational algorithm with validation experiments in-vivo and in-vitro for both cell line and primary ovarian cancer cells. We found a superb correlation between observed and predicted in-vitro total and CSC counts for both cell line and primary ovarian cancer cells (correlation r = 0.98, p < 0.0001). Furthermore, this approach appropriately predicted changes in cell growth in the presence of the CSC-promoting growth factor EGFL6 both in vitro and in vivo, over time frames of up to 28 days. The single cell division based sampling algorithm developed here allows for rapid and inexpensive means to predict in-vivo ovarian tumor growth based on microfluidic chip culture and can easily be adapted to evaluate new therapeutic options across other cancer subtypes. Citation Format: Alexander T. Pearson, Patrick Ingram, Shoumei Bai, Euisik Yoon, Trachette Jackson, Ronald J. Buckanovich. A computational algorithm to predict tumor growth and cancer stem cell proportion in-vitro and in-vivo from single-cell observations. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2705.