Institute of Information Science, Academia Sinica

Treatments of cancer or infectious diseases are often described by a metaphor of war.  The two parties -- human and disease -- attempt to win the war by destroying or eradicating the opponents.  On the one hand, a vast amount of resources have been devoted to identifying novel drug targets and designing new drugs.  On the other hand, the large population of tumor or pathogen cells can almost always evolve into phenotypes resistant to those new drugs.  Furstrated with the outcomes of escalated rivalries, some scientists start questioning the adequacy of the war metaphor and contemplating alternative treatment strategies.  Tumors typically consist of a heterogeneous cellular population with differential drug responses.  The current paradigm of personalized cancer therapy is to sequentially apply the most efficacious drugs of the dominant subpopulation of cancer cells.  This approach is at best suboptimal as it ignores the pivotal roles of some minor subpopulations and aims for myopic objectives (such as reduction of the dominant subpopulation) rather than the long-term benefits (such as the lifespans of patients).  To overcome these shortcomings, we propose new approaches of "strategizing" cancer therapies by exploiting the power of drug combinations and predictions of population dynamics.  Outcomes of "virtual clinical trials" on simulated data demonstrate the clear superiority of the strategized treatments to the current paradigm of personalized medicine treatments.  The results suggest a promising new direction for cancer therapies.