Oncolytic viruses are replication-competent viruses which selectively infect and kill cancer cells often derived from wild-type viruses that could otherwise be pathogenic and infect healthy cells. Tumour selectivity and benignity towards healthy cells often results from artificial selection on viruses over several generations or genetically engineering viruses before inoculation. However, successful oncolytic viruses deplete the supply of tumour cells upon which they depend for proliferation. As the tumour undergoes remission, such viruses could be maladapted in an environment with scarce tumour cells. In such situations, intuition suggests that strong selective pressure exists for the virus to evolve to exploit healthy cells. Their large population sizes and rapid-generation times allow many viral populations to undergo extremely rapid evolutionary change. The prospect of once therapeutic, repication-competent viruses evolving normal-cell tropism has been noted briefly by several authors (e.g., Russel 1994, Galanis 2006). Yet host-range shifts by successful oncolytic viruses away from tumour cells to healthy cells are rarely, if ever, observed empirically, despite the potential for a reversion to wild-type strains. The mismatch between our intuition that viral populations can rapidly evolve to switch from a depleted cancerous cell host to an abundant normal host, and the empirical lack of evidence for such adaptive change in oncolytic viruses, raises the question of whether ecological processes and constraints could prevent the evolution of viral strains that are virulent for healthy cells.
Results/Conclusions
Using a separation of time scales argument, we show that in the absence of trade-offs between virulence and transmission of the virus, mutant viral strains which infect both cancer and healthy cells have a higher per-capita growth rate than cancer-specialist cells whenever the long-term equilibrium of healthy cells is larger than the long-term equilibrium of cancer cells. Moreover, healthy-cell specialists are increasingly favored as the relative density of cancer cells to healthy cells declines. Finally, we use numerical results to show that a severe trade-off between lysing time and burst size of oncolytic viruses can mitigate, or at least delay, the evolution of generalist oncolytic virus strains even as the density of cancer cells decreases. This occurs because the potential fitness benefits to switching to a more abundant host are offset by the severe costs in either lysing time or bursting size. As the strength of this trade off is relaxed, oncolytic viruses continue to readily evolve towards exploiting healthy cells.
