Malignant tumors are thought to comprise phenotypically diverse clonal lineages in competition for nutrients, space and other resources. Genomic instability produces clonal diversity that natural selection acts upon to promote survival of lineages that acquire "cancer hallmark" phenotypes. Among these hallmarks, the angiogenic switch (uncontrolled production of new blood vessels from pre-existing vasculature) is among the most difficult to explain with an evolutionary narrative. Angiogenesis is a public good. As predicted by modeling studies in public-goods games in economics and evolutionary biology, "cheating" clones that free-ride on vasculature organized by cooperative clones can act as a tumor-on-a-tumor, or "hypertumor," ultimately yielding evolutionary suicide. The significance of this prediction in real tumors remains an open question. Here we show that clonal selection favoring angiogenic free-riders typically exists but is nearly always overwhelmed by stochastic mutational history (i.e. genetic drift). Angiogenic potential is maintained in many, perhaps most, tumors by pressure from genetic drift and pleiotropic interactions among deranged angiogenic and proliferative pathways in cancer cells. These conclusions follow from stochastic simulations of selection acting on both angiogenic and proliferative potential. In a significant fraction of simulations, tumors were rescued from evolutionary suicide when earlier, dormant lineages restored the angiogenic signal after "cheating" lineages became necrotic. However, if cheating lineages arose early and grew rapidly enough, they often drove the tumor to evolutionary suicide. We suggest that this mechanism may explain why early ductal carcinoma in situ frequently regresses spontaneously. At any rate, these results stress the importance of genetic drift in the disease's clinical behavior in individual patients.

angiogenesis; clonal_selection; evolutionary_oncology