The special issue on cancer and evolution: Lessons learned

Abstract This special issue of evolutionary applications focused on the evolution of cancer has provided a wealth of different viewpoints and results from leaders in the field. Together, these papers emphasize the importance of a broad perspective in order to understand why we and other animals get cancer, how it evolves within an individual, and what we can do about it. We can no longer take reductionist approaches that consider only the cancer cells and their genes. Instead, we need to understand how millions of years of evolution have guided strategies that shape cancer risk, why cancer risk varies across different animals, how cancer risk can vary in a population and be influenced by ecology (and influence this ecology), and of course how cancers evolve within us and the evolutionarily informed strategies to counter their impact. My goal here will be to “bring it all home,” providing a refresher of lessons learned with added kibitzing.


| E VOLUTIONARY UNDER S TANDING OF C AN CER RIS K
Evolution has shaped stem cell pool organizations and dynamics both to optimize tissue function and to prevent the development of malignancies that could impair organismal fitness. Birtwell et al. (This volume Birtwell et al., n.d.) highlight the critical need to determine the distribution of fitness effects of mutations in epithelial stem cells (modeled here for intestinal crypts) in order to understand how competition within a crypt or between crypts dictates cancer risk.
They used an agent-based microsimulation model to demonstrate that competition between crypts (for small stem cell pool sizes) and within crypts (for larger pool sizes) suppresses mutator clones when most non-neutral mutations are deleterious (as they likely are). The higher the proportion of deleterious mutations, the lower the odds of tumor suppressor gene inactivation, coinciding with reduced emergence of mutator clones. This study can help explain how stem cell pool organization (neither too big nor too small), and competition within and between the units, can minimize carcinogenesis. Can their model be used to explain a quandary arising from recent studies of clonal prevalence in normal tissues? While around 1% of colonic crypts are fixed for oncogenic mutations (translating into ~ 100,000 oncogenically activated crypts in an adult) (Lee-Six et al., 2019), this number is much higher for the esophagus and the endometrium in older adults (Martincorena et al., 2018;Moore et al., 2020;Yokoyama et al., 2019), suggesting stronger selection against such clones in the colon.
How does evolution accommodate different life history strategies, including body size and longevity, while mitigating associated risks like somatic decline and cancer? Erten and Kokko (This volume Erten and Kokko, n.d.) explore "ontogenetic management strategies" for somatic cells, and how these strategies differentially evolve in organisms dependent on body size. They model a range of strategies for in silico organisms, varying parameters of somatic cell strategies, Indeed, there has been much debate in the cancer community concerning the basis of cancer incidence patterns: between different tissues, within a lifespan, and across the animal kingdom. Why do cancers that arise in various tissues with very different stem cell pools and, requiring highly variable numbers of oncogenic mutations, show such similar late-life patterns of incidence in humans? And why are animal species with huge variance in body size and lifespans mostly able to delay cancer risk till what would be postreproductive ages in the "wild" (the Peto's Paradox described above)? Rozhok and DeGregori (This volume Rozhok and DeGregori, n.d.) propose that three orthogonal evolutionary processes controlling (a) somatic mutation occurrence, (b) species-specific life history traits (strategies for tissue maintenance and tumor suppression that maximize the odds of reproductive success), and (c) rates of physiological aging determine cancer rates across species and within the lifetimes of individuals (tissue decline in old age promotes selection for adaptive and sometimes oncogenic mutations). They attempt to reconcile what appear to be divergent observations of mutation rates, cancer susceptibility across animals, and similar aging-associated rates for many cancers with very different somatic evolutionary parameters into a unified framework.
Aging and other contexts (like due to cigarette smoking) both increase mutation prevalence (and thus heritable variability) and tissue environmental changes that contribute to dramatic increases in cancer risk (Laconi, Marongiu, & DeGregori, 2020). Gatenby and Brown (This volume Gatenby and Brown, n.d.) describe how normal cells can accumulate mutations, but that these mutations do not promote somatic evolution as the cells only possess the fitness function of the host animal, at least in young healthy tissues. In this manner, a soma serves its one true master-the germline, with the evolved objective of maximizing germline transmission. Only upon insult (inflammation, damage, aging, etc) is there a loss of this tissue control, and the cells (even if temporarily) acquire a self-defined fitness function, akin to a speciation event. Thus, aging and other insults promote cancers by facilitating the transition from host-defined to self-defined fitness. They also provide fascinating insight into the myriad of sources of information content in a cell, beyond its DNA, such as transmembrane ion distributions. Their model adds to a growing appreciation that cancer evolution is about more than just mutations, but requires overcoming hurdles evolved by animals to maintain functional tissues despite mutation accumulation (DeGregori, 2011;Gatenby & Gillies, 2008).
As argued above, tissue microenvironments and how they change during life and following insults exert substantial influences on somatic evolution and cancer risk, notwithstanding the common adherence to a mutation-centric explanation of cancer risk (e.g., Tomasetti, Li, & Vogelstein, (2017)). Solary and Lapane (This volume Solary and Lapane, n.d.) describe how clones often driven by putatively oncogenic mutations and, even genetically complex carcinomas, accumulate in our tissues (even dominating tissues) as we age. But what determines whether these clones contribute to cancer or, as is the case for the VAST majority of these clones, not?
The authors describe how normal (youthful) tissue architecture can impair cancer development and the many tissue disturbances that can promote the malignant evolution of these clones. A normal tissue microenvironment not only restrains oncogenesis by suppressing selection for malignant phenotypes, but can even normalize cells with malignant genotypes (i.e., stifling the malignant phenotype that would otherwise result). Malignant clonal emergence often requires tissue disruptions that result from wounding, UV light, therapies, other extrinsic exposures (e.g., from smoking), obesity, and of course aging-related tissue decline. Commonalities of these contexts include increased inflammation and reduced cell competition, and often stromal cell senescence and gut dysbiosis. Cancers evolve to manipulate their own microenvironment, altering stromal, immune, soluble and matrix components, and luckily the vast majority of oncogene-driven clonal expansions fail to successfully do so. The ubiquitous presence of oncogenic mutations in our tissues raises important questions about the forces controlling cancer development and has additional implications toward early detection.
For cancer prevention, while avoiding mutagenic exposures is of course still very much advisable, more efforts should be invested in developing interventions that alter tissue environments to be less cancer promoting. Strategies to target malignant cells need to be complemented by interventions that control the microenvironment and thus the direction of somatic evolution.
Racial disparities in cancer risk and outcomes are well described are significantly younger and smoke less than European Americans (EA). Despite being younger and having smoked less (on average), AAs developed adenocarcinomas that exhibited more nonsilent mutations, exhibited an increase in the CS4 mutation signature associated with cigarette smoking, and displayed more mutations in known cancer genes, than those in EAs. Effectively, each pack of cigarettes smoked appears to result in more mutations and earlier cancer development in AAs relative to EAs. This study shows that germline ancestry can impact mutational processes and likely somatic selection landscapes, although it is mysterious why such effects are not evident for lung squamous cell carcinomas. What is clear is that our evolutionary past has substantial influences on somatic evolutionary processes in our bodies.

| AN E VOLUTIONARY TANGO -THE DYNAMI C S OF C AN CER CELL S WITHIN US AND WITH OTHER ENTITIE S
The advent of single-cell genomic technologies, including single cell RNAseq and DNAseq, has identified marked cellular heterogeneity with cancers, both phenotypic and genotypic (Lipinski et al., 2016;Marusyk, Janiszewska, & Polyak, 2020). This heterogeneity is driven by variable selective forces throughout the malignancy, influenced by microenvironmental variables including pH, oxygen, nutrients, immune cells, and other stromal cells, and even competition with other cancer cells. Notably, drift also plays a significant role, particularly at later stages of cancer evolution (Sun, Hu, & Curtis, 2018;Williams et al., 2018). Robert Noble et al. (This volume Noble et al., n.d.) leveraged computational modeling to better understand the conditions that dictate when cellular heterogeneity predicts future cancer growth. As they review, higher clonal diversity sometimes predicts poor outcomes for patients, likely due to the increased adaptability provided by diverse phenotypic variants in the face of challenges, including from therapies. But then sometimes it does not, perhaps due to selective sweeps by highly malignant clones or the general unpredictability of cancers. Their spatial, stochastic model follows tumor evolution in a 2D grid. They show that clonal diversity early in cancer evolution predicts higher growth rates later, while the opposite is true for diversity late in cancer evolution, perhaps due to clonal interference and the lack of selective sweeps. It is important to measure diversity across the tumor, not only at the edge, and to consider the mutation rate of the tumor, consistent with analyses of kidney cancers (Turajlic et al., 2018). To the extent that computational modeling can predict real cancer evolution, these results suggest that leveraging tumor heterogeneity for prognosis will require careful consideration of when and where this diversity is measured.

Clusters of metastasizing cells have been observed in humans
and animal models, and thought to increase metastatic cell survival and seeding of distant sites (Cheung & Ewald, 2016). Campenni et al. By some estimates, at least 15% (and likely more) of cancers have infectious origins (Ewald & Swain Ewald, 2013). These include cancers that are directly caused by viruses (e.g., liver, head and neck, and almost all cervical carcinomas) and others that emanate from the inflammation that accompanies pathogen infection (e.g., Helicobacter pylori and stomach cancers) (Fernandes et al., 2015;Lin, King, & Chung, 2015;Wu et al., 2010). While most people with these patho- Pathogens have indirectly evolved to abrogate cancer defense mechanisms, as these host mechanisms can often serve to limit pathogen persistence. As has become increasingly clear in recent years, microbes can be our friends. The authors revise their barrier theory to incorporate how a healthy microbiome can protect against cancers, adding to the list of their essential contributions to our well-being (synthesis of certain vitamins, proper digestion, water absorption, barrier function, and immune regulation) (Rook & Dalgleish, 2011). crease inflammation, which can promote cancer evolution. In contrast, a healthy microbiome produces short-chain fatty acids like butyrate that promote barrier function and suppress inflammation.
A Western diet high in red meat favors the unhealthy microbiota, while a high-fiber diet provides fuel for butyrate-producing bacteria.
Mutualists can also antagonize cancer-causing pathogens. In addition to evidence that your microbiota influences your cancer risk, the microbiota can influence immune function to substantially impact responses to anti-cancer checkpoint therapies. However, we are far from understanding these connections, with different studies implicating different bacterial species as key determinants of cancer risk or immune responses. As always, the reality is more complex than what (necessarily) reductionist laboratory experiments suggest.

| MED I CINE A L A DARWIN
An evolutionary understanding of cancer has the potential to transform how we prevent, manage, and treat this disease. A human cancer can be composed of on the order of a trillion cells, and thus virtually every possible mutation or gene loss/gain will be present within this population, particularly when one considers increases in mutation prevalence in cancer cells. Similarly, pathogens can also have high population sizes and high mutation rates. Thus, whether for cancer or a disease-causing pathogen, resistance often develops in response to treatments designed to eradicate the problem, and evolutionary-informed strategies are required to limit the develop- results from computational modeling, mouse models of cancer, and clinical trials have indicated that seeking the MTD may be misguided, not only due to excessive damage to normal tissues but also because such strategies will lead to fixation of therapy-resistant cancer cells (West et al., 2020). While evolutionary-informed strategies, such as adaptive therapy, can keep a cancer at bay for longer (and with less toxicity to the patient), the cost may be the abandonment of an attempt to cure (Hansen, Woods, & Read, 2017). So how to choose?
Hansen and Read (This volume Hansen and Read, n.d.) explore this quandary. They model aggressive therapy with the intent to cure, but with the risk of earlier progression of drug resistant disease, versus containment strategy to manage resistance which uses competition to keep resistant cells at bay but where cure is unlikely. They show that this decision will depend on the probability of cure and the extent to which containment can delay resistant relapse, which are themselves dependent on mutation rate, cancer cell turnover kinetics, initial tumor burden, and the number of initial therapy-resistant cells. These parameters are used to place patients on a cure-progression plane; with clinical validation, and with the caveat that patients and their cancers are highly complex and thus difficult to parameterize, this tool could be useful for decision making. Finally, it would be interesting to add in additional components-how damage to the tissue microenvironment dependent on the intensity of therapy could engender selection for more aggressive cancer phenotypes and how differences in immune parameters can guide decisions.
Can cancer cells be coaxed into a state that is more sensitive to therapy-a sucker's gambit (Maley, Reid, & Forrest, 2004) can have good pharmacodynamic and pharmacokinetic properties, two highly desired characteristics for a pharmacological agent. They describe anti-cancer agents in development from insects, arachnids, amphibians, and marine organisms, diving into details for one compound each. These include melittin in bee venom (pro-cell death and anti-proliferative through multiple cellular targets), chlorotoxin from the Israeli Deathstalker Scorpion (which selectively binds cancer cells, facilitating cancer detection and targeting), Huchansu from the Chinese Bufo toad (a complex mixture of chemicals with a similarly complex mechanism of action), and trabectedin from the Mangrove Tunicate (which binds and distorts DNA, altering transcription factor binding). While the other toxins are still under investigation, given its efficacy trabectedin was approved by the FDA for treatment of liposarcoma and leiomyosarcoma. Challenges remain for most of these agents, including for efficacy and toxicity, as animals did not evolve to produce these compounds as anti-cancer agents, but as toxins to ward off enemies. Can we learn to better harness these poisons for the benefit of patients?

| VIVE L A D IFFEREN CE (E T LE S P OINTS COMMUNS)
Cancer will impact the fitness of an organism in the wild well before it would cause death under protected conditions (Ujvari, Roche, & Fdr, 2017). While we often consider the impact that a cancer has on an individual, very little attention has been paid to cancer's effects on interspecies interactions. Perret et al. (This volume Perret et al., n.d.) develop a multi-parameter mathematical model to explore the theoretical impact of cancer on predator-prey relationships, such as by affecting run speed, demonstrating complex impacts on ecological dynamics of populations, particularly for predators. Cancer can exert selective pressure on species, such as by conferring increased susceptibility to predation or reduced hunting efficiency, to promote the evolution of resistance strategies, compensatory changes in fertility, and feedback loops between predators and prey. There are many testable predictions from this work, including that loss of predators should increase cancer rates in their prey, and careful field work will be required to test these ideas. such by earlier reproduction, in an attempt to maintain fitness despite cancer pathogenesis (a mechanism of tolerance). Cancers are clearly agents of selection, which can lead to co-existence of the cancers and the host over evolutionary time (such as dogs in response to canine transmissible venereal tumor-CTVT) or increased tolerance/resistance over a few generations (for Devils in response to devil facial tumor disease -DFTD). The dramatic rise in DFTD in the last few decades has allowed documentation of the cascading albeit indirect impact of this tumor on other species and the overall ecosystem. The authors also highlight the impact that human activities have on cancer prevalence, including from habitat destruction, environmental carcinogenic contaminations, and climate change. They emphasize the need for improved strategies of surveillance, investigation, and mitigation. In all, the authors promote the critical need for a multidisciplinary approach to understand cancers in wildlife at multiple levels (overall ecology, population, individual, and tumor) and timescales (across millennia, over generations, within a lifetime) in order to protect our nonhuman brethren-often from ourselves.
Just as Darwin recognized the importance of artificial selection and domestication for understanding natural selection (Darwin, 1876), Thomas et al. (This volume Thomas et al., n.d.) describe how human-driven selection for desired traits can lead to high cancer rates, and more intriguingly how compensatory tumor suppressive mechanisms can be selected for even when such processes might be "too expensive" for wild animals, given greater resource availability, reduced threats and often absent competition associated with domestication. Population bottlenecks, high homozygosity, and linkage disequilibrium with desired traits can also contribute to fixation of oncogenic traits in domesticated animals. We can learn a lot about cancer etiology from the products of thousands of years of animal domestication. Like many birds, sea gulls have evolved long lifespans and presumably also similarly delayed cancer rates. Flight endows creatures with a potent mechanism of predator avoidance, and this lowered extrinsic mortality provided an advantage to a longer lifespan (providing further opportunities for reproduction). Meitern et al. (This volume Meitern et al., n.d.) examined gene expression in whole blood from young and old (>16 years) sea gulls, with a focus on genes involved in cancer. Interesting, the vast majority of changes from young to old involved downregulation of gene expression. In particular, they observed reduced expression of eight cancer-related genes in old birds, and they speculate on how these changes could contribute to cancer risk. While determining how these expression changes might contribute to increased cancer risk in old age would require reverse genetic studies, which is not currently feasible for this species, these pioneering studies highlight the potential of comparative biology using wild animal populations to reveal new mechanisms controlling cancer susceptibility.