Phase I trials in melanoma: Optimizing order and timing of combination therapy
Alexander Anderson (Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center & Research Institute)
(April 14, 2014 3:00 PM - 3:50 PM)
Metastatic melanoma is known to be resistant to standard chemotherapy. During the last few years, targeted therapeutic approaches have emerged as the dominant treatment choice, primarily because they target tumor cells that harbor specific genetic mutations. However, even these targeted drugs have limited long term success in treating metastatic melanoma patients, since eventually resistance emerges. Surprisingly, when these treatments are given in combination much better treatment responses are observed. However, little is known about why the combinations are more successful. Our recent experimental results suggest a possible mechanism, in that these treatments differentially induce autophagy (the process of self-digestion by a cell to temporarily extend its life under stressful conditions) in tumor cells. To better understand how autophagy induction might facilitate better treatment response, we developed a mathematical model comprising of a system of ordinary differential equations that explain the dynamics of melanoma cells under different treatments. Specifically, we incorporated an autophagy cell population and examined how this population affects treatment success. Model parameters, such as the growth and death rates with and without treatments, were estimated by comparing model predictions with in vitro experimental data. Model results show that the combination therapy is effective in controlling tumor population over an extended period of time. The resistance, however, eventually emerges driven in part by the autophagy population. To overcome this resistance, we applied a drug that targets the autophagy population and were able to show that additional administration of this drug inhibited growth of the resistant population. In order to place these results in a more clinically relevant setting (e.g. clinical tumor volume doubling times), we generated a small cohort of virtual patients by varying model parameters to capture the diversity of disease response observed in the clinic. Parameters varied include initial proportion of different cell types, net growth rate, autophagy rate, and cell death rates. We then applied 10 different treatment schedules that were composed of different combinations (order and duration) of AKT inhibitor, Chemotherapy and autophagy inhibitor to this virtual patient cohort. This effectively allowed us to implement a “virtual clinical trial” or phase i trial with our model and select the optimal therapeutic approach across a range of patients.