Increased risk of cancer in dogs and humans: A consequence of recent extension of lifespan beyond evolutionarily determined limitations?

Abstract

Cancer is among the most common causes of death for dogs (and cats) and humans in the developed world, even though it is uncommon in wildlife and other domestic animals. We provide a rationale for this observation based on recent advances in our understanding of the evolutionary basis of cancer. Over the course of evolutionary time, species have acquired and fine‐tuned adaptive cancer‐protective mechanisms that are intrinsically related to their energy demands, reproductive strategies, and expected lifespan. These cancer‐protective mechanisms are general across species and/or specific to each species and their niche, and they do not seem to be limited in diversity. The evolutionarily acquired cancer‐free longevity that defines a species’ life history can explain why the relative cancer risk, rate, and incidence are largely similar across most species in the animal kingdom despite differences in body size and life expectancy. The molecular, cellular, and metabolic events that promote malignant transformation and cancerous growth can overcome these adaptive, species‐specific protective mechanisms in a small proportion of individuals, while independently, some individuals in the population might achieve exceptional longevity. In dogs and humans, recent dramatic alterations in healthcare and social structures have allowed increasing numbers of individuals in both species to far exceed their species‐adapted longevities (by two to four times) without allowing the time necessary for compensatory natural selection. In other words, the cancer‐protective mechanisms that restrain risk at comparable levels to other species for their adapted lifespan are incapable of providing cancer protection over this recent, drastic, and widespread increase in longevity.

INTRODUCTION: THE EVOLUTIONARY LEGACY OF CANCER

According to the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program, cancer describes a group of over 100 pathological conditions that are characterized by uncontrolled cellular proliferation.¹ Even including deaths due to COVID-19 in 2020, cancer is the most common or the second most common cause of death for humans in Europe² and in the United States,^3–5 and it is believed that cancer is also the leading cause of disease-related death for older companion and working dogs in the developed world.^6–12

In a remarkable thesis of how cancer was transformed from an obscure and mystical condition to a medical problem,¹³ Koblenz notes that the death rate from cancer increased by almost 20-fold from the years 1850 to 2009, although it is apparent that a major portion of the cancer epidemic of the 20th century was driven by tobacco.¹⁴ For humans living in the United States in 2020, the lifetime probability of being diagnosed with an invasive cancer is approximately 40% (an approximate lifetime risk of 1 in 2.5).¹⁵ Tumors of the prostate, lung, and colon account for more than 40% of all cases in men, and tumors of the breast, lung, and colon account for 50% of all cases in women. Cancer also claims a large number of domestic dogs’ and cats’ lives.^{7,11,12,16–20} Even while the frequent occurrence of cancer in both of these domestic species is well recognized, the precise incidence of invasive cancers remains to be established.²¹ Registries that systematically account for cancer incidence and mortality of dogs and cats are not standardized, making comparisons among them challenging.²² Furthermore, many (perhaps most) cases of suspected cancers in animals are not definitively diagnosed histopathologically, and there are considerable referral biases in the hospital databases used to obtain estimates of cancer incidence and cancer mortality.²³

Despite these limitations, the Veterinary Cancer Society estimates that one in four dogs will be diagnosed with cancer in their lifetime, and that cancer is the leading cause of death in pets that are beyond middle age.²⁴

As hard as it is to obtain actuarial estimates in domestic animals, assessment of disease and cause-specific mortality rates in wild animal populations is even more challenging.²⁵ The available data suggest that animals in the wild have variable incidence rates of specific cancers, although mortality is often due to loss of fitness, competition, predation, accidents, or human-related causes.²⁶ It is also clear that the overall rate of cancer among species in captivity is variable,²⁷ but none are as high as those seen in domestic dogs and cats.²⁸ Cancers in wild animals where the etiology can be determined are often due to infectious agents^28,29 or the tumors themselves are transmissible,^29–31 and it is notable that such infectious and transmissible cancers represent exceptions to the low rates of cancer observed in wild animals. Conversely, cancers associated with infectious agents (such as feline leukemia virus-associated tumors³²) and transmissible cancers (such as canine transmissible venereal tumors³³) account for only a small fraction of the rather large number of cancers seen in dogs and cats in the developed world.

Comparative biologists have questioned why cancer is so prevalent in domestic dogs and in humans, what accounts for the difference in cancer risk and incidence seen in dogs and humans as compared to wild populations, and what we can do about it. In this perspective, we propose that the apparent excess incidence of cancer in both of these species is due to the fact that dogs and humans are two species that have benefited from social, medical, and technological advances that allow them to live beyond the age that nature intended, as species-specific cancer-protective mechanisms that evolved over millions of years³⁴ cannot address the cancer vulnerabilities acquired through the rapid increase in longevity experienced by modern dogs and humans. The observation that there are substantial differences in the incidence of specific cancer types that afflict domestic pet dogs and humans, despite a closely shared environment, suggests that there are fundamental biological risk factors present in both species that are distinct from environmental exposures.^15,21 This different prevalence of cancer types and the rather poor conservation of genomic driver events in histologically similar cancers of both species should raise a note of caution about “the dog as a model” of cancer causation.³⁵ On the other hand, the shared sensitivity of dogs and humans to age-associated cancers provides unique opportunities to understand the relationships between organismal aging and cancer risk.³⁶ Finally, we advance the concept of risk assessment coupled with strategic prevention as a means to reduce cancer mortality in dogs, and eventually, perhaps also in humans.

Cancer is an ancient disease

To understand the apparent cancer epidemic, it is essential to understand the foundational pathogenesis of cancer and its relationship to evolution across the animal kingdom. Cancer arises from the malignant transformation of a single cell that undergoes uncontrolled proliferation, but the disease is confined to multicellular animals.^37,38 In single-cell organisms such as yeast, uncontrolled proliferation might lead to exhaustion of resources that will cause some individuals to die, but others can remain dormant until the environment is once again favorable for growth and reproduction.^39,40 In contrast, because multicellular animals rely on specialization of tissues and organs, they cannot tolerate the damage and dysfunction created by excessive, uncontrolled cellular proliferation even in a single organ or tissue. Thus, cancer is not a modern physiological defect, but rather it is a vulnerability that is rooted deeply in vertebrate evolutionary history⁴¹ and nearly every multicellular animal is at risk for developing cancer.^26,37,38,42 The ancient origins of cancer are supported by the increasing examples of cancers identified in the fossil record.^43–47 While a precise estimate of cancer incidence and prevalence in antiquity is impossible, it has mostly been assumed that it was rare.^48,49 This is partly due to the fact that the fossil record is biased toward tumors of bone (primary and metastatic), which overall represent only a fraction of the cancer burden in modern humans.⁵⁰ Historians and scientists have offered different interpretations for this assumed rarity, ranging from lack of exposure to carcinogens that are ubiquitous in modern societies to support it, to the technical and methodological pitfalls in the analysis of paleopathological samples to refute it,^48,51 to the focus on literature and paleopathological specimens originating primarily from ancient Egyptian and Greek societies and neglecting the literature from Chinese and other significant ancient societies.⁵²

Ewald has also proposed that the rarity of cancer in antiquity (and in wildlife) could be due to increased mortality of individuals with cancer where such evidence of cancer-related mortality or cancer-related morbidity contributing to mortality would disappear due to predation (consumption of the tumorous tissue).⁵³ The evidence of cancer could also be erased by environmental effects on the cancerous tissues, directly related to the fragility caused by the cancer, or to the lack of specificity of the pathological changes caused by the cancer.⁵³ Ewald also suggested that parasites that cause cancer (where parasites are defined broadly to include multicellular, cellular, and subcellular replicative agents that live in or on a host organism and negatively affect the evolutionary fitness of the host) have existed in human and animal populations for millennia, but their impact may be underappreciated by the emphasis of mutations as the root cause of cancer.⁵³ It is interesting that dogs appear to be a privileged species in this regard: with the exception of canine papillomaviruses that cause benign warts, neither DNA nor RNA oncoviruses that fulfill Koch's postulates have been definitively identified in canids, despite sporadic reports of retrovirus or herpes virus particles in dog lymphomas, leukemias, and peripheral blood lymphocytes.

Cellular replication and the creation of aged (permissive) microenvironments are major risk factors for cancer

Malignant transformation, an essential step in the pathogenesis of cancer, is caused by mutations that alter the sequence or structure of DNA.⁵⁴ Cancer, thus, results in part from a collection of somatic changes leading to inappropriate cell division or survival, coupled with the failure of tumor-suppressive mechanisms, which together provide cells with a clonal growth advantage. Examples of clonal somatic changes include mutations, copy number changes, and epigenetic structural DNA changes that lead to oncogenic activation. A consequence of this paradigm is that anything that increases somatic changes can also lead to increased cancer risk. A further requirement for cancer formation is that the transformed cells must have a permissive environment where they can grow and develop the complex anatomical structures that create tumors.^54–56 Recently, Laconi and colleagues proposed that, “the tumor-suppressive potential of youth is more potent than previously realized, limiting cancers through half a century of human life even in the face of increased DNA mutations and highly perturbed tissue environments,”⁵⁷ and much additional work suggests that organismal aging and the associated development of cellular senescence, proinflammation, and waning immunity appear to be major mechanisms that allow for the development of these permissive environments.^58–64

Pathologically significant mutations are propagated during the process of DNA replication, as nondividing cells will not pass on the cancer phenotype to progeny cells. The earliest examples that cemented the association between mutations, proliferation, and cancer included the increased cancer risk observed in individuals who had occupational exposures to mutagens,^65,66 habitual tobacco users,⁶⁷ or individuals who had mutations in genes involved in DNA damage repair.⁶⁸ Similarly, elevated cancer rates can be found in individuals with germline mutations in DNA polymerases (POLE/POLD1).⁶⁹

Companion dogs generally do not have the same level of exposures to industrial (occupational) or social (tobacco) mutagens as humans, their hair coat protects them from cancer-causing ultraviolet radiation from the sun, and controlled breeding practices help to significantly reduce the likelihood for true heritable cancer syndromes to become established. In a singular situation where a syndromic cancer (canine renal cystadenoma and nodular dermatofibrosis, associated with a mutation in the folliculin gene) arose in a canine pedigree, it was rapidly identified and characterized,^70,71 providing tools to eliminate it from the breeding population. Some studies have identified associations between certain animal cancers and environmental exposures. For example, studies done by Reif and colleagues 30–40 years ago reported associations between exposure to environmental tobacco smoke and cancers of the respiratory tract in dogs.^72,73 However, these findings have not been replicated in more recent studies.⁷⁴ Similar lack of replication has confounded definitive associations between other environmental exposures and dog cancers.^75–77

On the other hand, consistent associations between certain cancer types and dog breeds suggest enrichment of risk alleles that can inform heritable risks associated with cancer.^21,78–87 While there are multiple examples of genomic regions and of coding and regulatory variants associated with canine tumors in specific breeds,^{85,86,88–90} virtually all of these cancers occur in dogs that are past middle age, accounting for variations in breed-specific aging. Furthermore, many factors in the germline seem to contribute to breed-specific risk for cancer in dogs,^91,92 underscoring the complexity of the disease, and/or the incomplete penetrance of the putative risk alleles.

Importantly, the association between cellular replication and cancer is not a new concept. As far back as the early 1900s, Theodor Boveri's work on cell division had led to the hypothesis that cell division created cancer risk.⁹³ Even to this day, one of the most universal prognostic indicators within tumor tissue is the level of Ki-67 protein, which is often used to gauge the proliferation rate, and by extension the aggressiveness of a tumor.⁹⁴

Peto's paradox

An implication of the association between replication and cancer is that increased longevity (more cell divisions) and increased size (more cells) should result in elevated cancer risk and greater cancer incidence. From this paradigm, Sir Richard Peto, a statistician well known for helping to uncover the association between lung cancer and smoking, identified a paradox⁹⁵ and articulated the problem,⁹⁶ which was subsequently named “Peto's paradox” by Nunney.⁹⁷ Peto's paradox states that at the species level, the incidence of cancer does not appear to correlate with the number of cells in an organism. In other words, the predicted linear association between large size of an organism (which is partly due to greater numbers of cells, attained by increased cell division) and cancer is inapparent across the multitude of species in the animal kingdom²⁷; extremely large animals, especially those with long lifespans, which undergo far more replications (and therefore would be expected to have an excessively high replicative cancer risk) do not have higher cancer rates than smaller animals, which need less replication to attain their adult size and that generally have shorter lifespans (Figure 1). As detailed below, various investigators have since proposed that cancer-protective mechanisms acquired over the course of evolution provide satisfactory answers to Peto's paradox.

FIGURE 1. Cancer risk and Peto's paradox. Illustration of Peto's paradox, which describes the absence of the predicted relationship between animals’ (A) size and (B) longevity relative to their cancer risk

CANCER-PROTECTIVE MECHANISMS IN LONG-LIVED SPECIES

Several decades passed between Peto's original observation and the initial formulation of a potential solution, which came to light through the study of large and/or long-lived species of animals that have evolved mechanisms allowing for the minimization of cancer risk (see references^{26,34,37,58,97–105} for previous, outstanding discussions on this subject).

As species have evolved, unique adaptations that protect them from cancer have arisen multiple times, and often, independently from each other. These cancer-protective mechanisms address different aspects of vulnerability and they become part of each species’ evolutionary history. As illustrated below, different species have “attacked” different aspects of tumor formation, including reducing the likelihood that detrimental mutations are incorporated into their genomes, altering their metabolism, making the tissue microenvironment inhospitable to tumor growth, and probably others. These changes have occurred stochastically over millions of years, but they must have been subject to strong positive selection,¹⁰⁶ with the consequent effect of giving rise to adaptive traits that protect species from cancer⁹⁷ while allowing them to occupy niches where they can invest energy to become large, have a long lifespan, or both.¹⁰⁷ It is especially intriguing that in their evolutionary history, several tumor suppressor genes in hominoids have been subject to evolutionary selection such that the wild type (reference) alleles are distinct from variants found in apes, and presumably in the common ancestor to these primate species.¹⁰⁶ Indeed, the ancestral wild type alleles for some of these genes such as APC and TP53 (as found in some modern apes), or BRAF and TNFAIP3 (as found in Neanderthals and Denisovans, respectively) are found to be deleterious in modern humans, exclusively as somatic variants, and closely associated with cancer risk.¹⁰⁶

It is reassuring that many tumor-suppressive mechanisms are highly conserved across vertebrate (and oftentimes invertebrate) species. For example, the products of the TP53 and the PTEN genes, and the enzymes that maintain telomere length, are highly conserved mechanisms that suppress tumor formation across many different species.^108–116 Replicative senescence is another widespread, primordial mechanism that seems to protect animal species against longevity-associated somatic modifications.^117,118 In most cells, senescence programs are activated upon reaching a predetermined number of replication events, as well as by somatic mutation events, and many genes involved in regulation of senescence have been directly implicated in tumor suppression.¹¹⁹

Specific examples of species-specific cancer-protective mechanisms have been described in the elephant family, bowhead and other whales, and in different families of long-lived mole rats,³⁴ while more are being deduced from recently sequenced genomes of other long-lived species such as Amazon parrots,¹²⁰ white sharks,¹²¹ and whale sharks.¹²² Each of these species appears to have acquired or “fine-tuned” unique mechanisms that allow them to have long and relatively cancer-free lives. For example, the TP53 gene underwent a series of retrotranspositions in the elephant lineage, after the split from its last common ancestors that in the extreme resulted in more than 20 copies of this gene in modern African elephants.^123,124 While not all of the elephant TP53 genes are likely to be active, the orthologous elephant p53 protein was shown to be a more potent inducer of the p21cell cycle kinase inhibitor and to have greater pro-apoptotic activity than the human p53 protein. The specific mechanisms of the elephant p53's enhanced apoptotic ability, including it to be delivered as a novel therapeutic for cancer in dogs and humans remains a subject of intense investigation (Dr. Joshua Schiffman, personal communication). The function of the elephant p53 retrogenes remains unclear, but when transfected into human cells, the retrogenes were upregulated and translated upon treatment with chemotherapy and radiation,¹²³ and in separate experiments, Vazquez and colleagues showed that one or more of the retrotransposons may induce expression of at least one pseudogene (LIF-6) that seems to have been refunctionalized to promote apoptosis in elephant cells, but not in those of other mammalian species.¹²⁵ And yet, duplication of TP53 is not unique to elephants, nor does it guarantee long life. Rats, wallabies, and some small bats also have duplicated this gene over the course of evolution without consequent gains in size or longevity.¹²⁴

The bowhead whale genome hints at unique metabolic adaptations, including mutations in UCP1, which encodes the mitochondrial uncoupling protein-1 of brown fat, as well as altered insulin signaling. Additionally, the bowhead whale genome includes cancer-protective genes under positive selection, including ERCC1 (excision repair cross-complementing rodent repair deficiency, complementation group-1; a member of the nucleotide excision repair pathway) and duplication of PCNA (proliferating cell nuclear antigen; a gene involved in DNA repair).¹²⁶ Other whales, including humpbacks, minkes, and toothed whales seem to have developed segmental duplications in tumor suppressor genes that are species-specific and are also under positive selection,^104,127 consistent with the interpretation that these events occur over long evolutionary periods and are unique to each species and dictated by their niche.

Rodents show extreme variation in lifespan, which can be used to identify potential cancer-protective mechanisms. Across multiple short- and long-lived rodent species, five amino acid substitutions in SIRT6 account for most of the variation in DNA double-strand break repair efficacy, with increased SIRT6 activity also showing a direct correlation with lifespan among these species.¹²⁸ Another example comes from mole rats, the longest lived rodents alive today, which can reach 20 to more than 30 years of age. African mole rats of the family Heterocephalidae have developed a series of distinct cancer-protective mechanisms that modify the organization of stromal matrix proteins, making the microenvironment inhospitable to cancer,¹²⁹ while the distantly related blind Middle Eastern mole rats of the family Spalacidae have developed a hypersensitive cell-death response to mitigate the excessive drive to proliferate that is associated with events that lead to malignant transformation.¹³⁰ Perhaps the most widely recognized cancer-protective mechanism in mole rats is the abundance of high molecular weight hyaluronic acid in the African naked mole rat, possibly due to increased production of hyaluronan by the HAS2 gene product with concomitant reduced activity of the enzymatic machinery that degrades hyaluronic acid.^129,131 This stands in sharp contrast to the hyaluronic acid content in, for example, human tumors, where hyaluronidases cleave hyaluronic acid into small molecular weight species that promote inflammation and tumor growth.¹³²

Together, these examples indicate that large and long-lived animals have evolved diverse and sometimes convergent mechanisms that have enabled large investments in reproduction and high energy requirements over long and relatively noncancer-prone lifespans.⁹⁷ These varied solutions to the cancer problem, in turn, create species-specific barriers with limitations on cancer protection determined by the idealized lifespan and body size acquired through these millions of years-long evolutionary processes. And thus, exceeding these barriers through technological innovation to achieve increased lifespans, as is the case with “caninity” and humanity, or via artificial selection (selection for size in the absence of selection for compensatory protective mechanisms), as is the case in dogs, generates conditions that are cancer prone.

We should note that Ujvari and colleagues proposed an alternative hypothesis that, while selective breeding (and inbreeding) and the consequent attenuation of genetic diversity result in the accumulation of deleterious genetic variants in domestic animals, the absence of natural selection could theoretically allow these animals to invest more resources to support existing anticancer mechanisms and potentially develop new ones.¹³³ While this is an intriguing premise, humans have not significantly increased the lifespan of domestic animals other than dogs and cats, and in our opinion, rare anecdotes of spontaneous cancer regression, for example, in dogs with bone cancer, fail to support strong maintenance or widespread development of novel cancer-protective mechanisms in the accelerated time scale of domestication. On the contrary, the excess of cancers seen in modern dogs (and in humans), and the patterns of their association with age, would strongly suggest that there has been neither sufficient time nor selective pressure to allow for evolution of adaptive mechanisms to reduce the risk of cancer that comes with the creation of aged cellular environments, increased exposures, and the accumulation of somatic mutational events in these populations.

Breaking through the evolutionary lifespan adaptation is a major contributor to cancer risk in humans and in dogs

One theme that has been underappreciated, in our opinion, is the dramatic gains in expected lifespan that have occurred in the past 150–200 years in humans and over the past ∼50 years in dogs, and the associated risk of cancer that is attributable to aging. According to Finch,¹³⁴ life expectancy from birth in humans doubled in the 5–7 million years from the time the hominid lineage split from the common ancestor shared with chimpanzees until modern hunter-gatherers. It is worth noting that the process of speciation in apes and hominoids (presumably similar to the process of speciation in other animals) included unique adaptations in sequence, and perhaps in function of tumor suppressor genes.¹⁰⁶ In their study of 120 oncogenes and tumor suppressor genes in seven hominoid species, including two extinct species, Neanderthal and Denisovan, Kang and Michalak also found opposing selection pressures operating on these two classes of genes, with tumor suppressor genes being under weaker purifying selection than oncogenes.¹⁰⁶ The authors hypothesize that the selective pressures accounting for these differences could come from the activity of oncogenes, where a single variant can be detrimental to the individual, in contrast to tumor suppressor genes, which generally maintain activity even in a state of heterozygosity, and where loss of both alleles is usually necessary for cancer-prone phenotypes to arise.¹⁰⁶

The precise lifespans of intermediate species during human evolution are unknown, although it is apparent that early modern humans and Neanderthals had a larger proportion of older adults than prior Homo species and Australopithecines.¹³⁴ Human life expectancy doubled again in the 150 years between 1860 and 2010 during industrialization, and it increased further in the contemporary era with increasing access to hygiene, medical care, and other advantages of complex societies¹³⁵ (Figure 2A).

FIGURE 2. Increases in human lifespan overlap with the majority of human cancer deaths. (A) Humans had a relatively stable life expectancy of 30–40 years until the industrial revolution. The upper boundary of this life expectancy represents the ancestral evolved cancer barrier. However, in the 100 years between 1850 and 1950, human life expectancy increased by approximately 50% across the world, and it did so again in the 60 years between 1950 and 2010. That trend continued through the present (latest measurements shown for 2017). (B) The overall cancer death rates per 100,000 people do not have a proportional relationship with age in the United States. Instead, cancer death rates are vanishingly rare, proportionately, in individuals under 15 years of age, they become apparent with a marginal increase between the ages of 40 and 50 after individuals cross the cancer barrier, and then increase dramatically thereafter. Individuals over 60 years of age account for the vast majority of cancer deaths in the United States (United States data from 2017). This is representative for most of the developed world. (C) Lung cancer death rates per 100,000 men also do not have a proportional relationship with age. Despite decades of persistent exposure to myriad potent carcinogens in tobacco products, lung cancer is extremely rare in smokers under the age of 40. Lung cancer becomes more apparent in smokers by the age of 40 after individuals cross the cancer barrier and continues to increase geometrically until the age of 80–85 (United Kingdom data from 2015–2017). This is also representative for most of the developed world

Thus, human lifespan was molded to a niche over a 5–7-million-year period where there were strong evolutionary selective pressures, including selection for cancer-protective mechanisms.¹⁰⁶ Because the niche occupied by our hominid ancestors was different from that occupied by other great apes, we cannot infer their lifespan from that of modern nonhuman great apes. But we can estimate it was shorter than pre-industrial modern humans or 20th century foragers, whose life expectancy was 30–40 years,^134,136 although this was heavily influenced by high infant mortality. Nevertheless, even if we accept an age somewhere between 35 and 45 years as the evolutionarily determined lifespan for humans, fewer than 5% of cancer cases are diagnosed in humans from birth to 35 and less than 10% from birth to 45, making the expected cancer incidence in this population comparable to that seen in apes and monkeys¹³⁴ (Figure 2B,C).

Like humans, the lifespan of canids in the wolf family was molded to a niche over a 5–7-million-year period under strong evolutionary selective pressures. We cannot be certain of the expected lifespan of the canid wolf ancestor, but an estimate of approximately 3.5–5 years is possible based on studies of modern wild wolves and feral dog or village dog populations.^106,137,138 And yet, in contrast to what is observed in wolves and in feral dogs and village dogs even today, the expected lifespan of companion pet dogs in much of the developed world has increased by a factor of two to four (to an average of 9–14 years, depending on breed, according to information available from the American Kennel Club¹³⁹) in the ∼50 years since dogs were brought into the household and integrated as members of the family in the latter part of the 20th century. Interestingly, the eventual selective breeding for form and function that occurred over the past 15,000 years, and the purposeful derivation of modern dog breeds over the past 100–400 years, might have influenced the canine lifespan through enrichment of alleles that encode for small size.^140,141 But, following the human example, if we accept 3–5 years as the evolutionarily determined lifespan for primordial dogs evolving over a period of ca. 6 million years, the available data reproducibly indicate that fewer than 5%–10% of cancer cases are diagnosed in modern dogs from birth to 3–5 years,^{12,18,142,143} making the expected cancer incidence in this population comparable to that seen in captive wolves under 5 years of age¹⁴⁴ (no estimates of cancer exist for wolves or other related canid species in the wild) and to that which has been estimated for most animals in the absence of human influences.¹⁰⁷ It is also worth noting that wolves and other wild canids that are maintained in captivity and that reach advanced ages also exhibit increased susceptibility to some of the most common cancers seen in modern domestic dogs.^144,145

Hence, we advance the premise that a major component of the elevated cancer risk seen in domestic dogs and in humans when compared to other animals is due to the shattering of the life-expectancy barrier that was evolutionarily determined: essentially “living longer than nature intended.” The longevity that both species enjoy at the present time has dramatically increased (two to four times), compared to the lifespan to which each species had adapted over its evolutionarily history (an “evolutionarily determined lifespan”), through social, medical, and technological advancements, over very short periods of time, and with artificial selection playing an increasingly important role, especially in the case of dogs.^58,98

A recent study of cancer in nondomestic mammals kept in zoological collections is consistent with the interpretation that living longer than nature intended is a major risk factor for cancer. While captivity might introduce its own set of risks for cancer or premature mortality, such as exposure to anthropogenic pollutants, stress, etc., it also removes common causes of mortality for many species, such as competition, predation, accidents, or human-related causes. Consequently, animals in zoos would be expected to outlive some of their counterparts in the wild. Vincze and colleagues did not specifically address gains in longevity in their study, but cancer-related mortality was still lower than 10% for almost 80% of the species in this study (and it was zero for almost 25%).²⁷ As a group, cancer mortality was highest in carnivores. The authors propose that this might be due to dietary factors (consumption of meat), but it is possible that carnivores are also more amenable than noncarnivores to greater life extension in captivity.

Returning to the dog example, canine osteosarcoma provides a good example to illustrate the association of size and cancer susceptibility, and a solution to Peto's paradox.⁹⁸ This disease occurs only rarely in humans, but it is quite common in dogs.¹⁴⁶ Because the natural history of the disease in both species is similar, canine osteosarcoma has been long proposed as a surrogate disease model to better understand its human counterpart.^147,148 In addition to an anatomical predilection for the long bones and shared patterns of metastasis, a few molecular traits are conserved between canine and human osteosarcoma. An initiating event, such as TP53 mutation, seems to be required, which allows replication of chaotic genomes and clonal outgrowth of tumors with highly disrupted and heterogeneous genomes.¹⁴⁹

However, important epidemiologic differences coexist with the similarities between human and canine bone tumors. For example, the reported age-adjusted incidence of osteosarcoma in humans is between 0.2 and 4.5 per million, with most affected patients being children, adolescents, and young adults,¹⁵⁰ whereas in dogs, this disease is most commonly observed in adult and elderly individuals from large and giant breeds, with a lifetime risk of approximately one in five at the extreme.^{146,148,151–153} The strong association between size and risk of osteosarcoma in dogs illustrates a similar, albeit weaker, association between risk of osteosarcoma and size in humans, where it is more common in males than in females, and in children who are taller and have high birthweights.^154–156

Of additional interest, larger dogs age more rapidly than smaller dogs and have shorter lifespans.¹⁵⁷ This pattern of rapid aging and shorter lifespan also aligns well with the risk for bone cancer (Figure 3), where a high rate of cellular replication is necessary to form large bones (larger than the norm for the species as it evolved), where aged tissues might provide more permissive environments for the development of cancer, and where the protections afforded by modern veterinary healthcare, vaccination against lethal pathogens, appropriate nutrition, and shelter have more than doubled life expectancy for dogs living as companions in the human environment.

FIGURE 3. Canine size, lifespan, and osteosarcoma risk. Larger dogs have shorter life expectancy and increased risk for osteosarcoma than smaller dogs. Specifically, using the German shepherd dog breed as the reference, great danes have an odds ratio (OR) of 5 for development of osteosarcoma, whereas miniature schnauzers have an OR of 0.2 for development of osteosarcoma.153 Overall, the highest risk breed has an OR >200-fold greater than the lowest risk breeds for development of this disease (reproduced from Makielski et al. Risk factors for development of canine and human osteosarcoma: a comparative review. Vet Sci. 2019;6(2):48. Copyright: the authors)

The association between increased body size and increased cancer risk is observed, not only with osteosarcoma where the effect manifests in the extreme, but also with other cancers.^{7,12,17,19,158} And while the major variant alleles associated with size determination in the dog (including GHR, HMGA2, IGF1, IGF1R, SMAD2, STC2, LCORL, and CDK4)¹⁵⁹ have not been found to confer independent cancer risk, this may be due to the fact that variants are fixed in purebred dogs. To understand the potential significance of genomic traits that regulate size in dog aging and cancer, future studies should include multiple breeds and/or mixed breed dogs to unmask the potential contribution of fixed alleles.^159,160

Altogether, these observations suggest that Peto's paradox is no paradox at all, when applied within a single species, and removing the impact of evolution within a niche. That is, Peto's principle (as opposed to Peto's paradox) would predict that increased body size should be associated with increased cancer risk in dogs. Thus, extension of this concept leads to the notion that longevity is intricately associated with cancer risk, but only when such longevity exceeds the evolutionarily adapted lifespan for that species. The observation that small dogs live longer than large dogs and have lower cancer rates, thus conforms with Peto's principle.^{12,98,161,162}

Furthermore, these relationships are not completely static, so if replicative senescence is a cancer-protective mechanism, the earlier onset of senescence (as would occur in animals with shorter lifespans) that is observed in large dogs could be related to their increased cancer risk over the same timescale as their smaller counterparts. Untangling the connections between senescence, cancer risk, and lifespan in different dog breeds should generate important insights into how better to understand cancer in humans and identify people with increased risk earlier in their lifetime.^{21,36,58,84,162,163}

After considering these concepts, a “simple model” is that, within a given species over the course of a lifetime, cancer-preventive mechanisms may change from protective states ensuring evolutionary fitness to less-protective states once they no longer provide advantage. In other words, anticancer defenses in old animals will be under less selective pressure than in younger animals, and costly defenses are likely to be maintained at a lower level. In reality, this is likely to be far more complicated as progenitor cells need extreme protection from somatic damage, while damage might be tolerated (and is observed) in terminally differentiated cells.¹⁶⁴ This threshold of age-related vulnerability,⁵⁸ which is more readily observed after individuals cross cancer-protective barriers for their species, can also explain the prolonged latency seen in humans between carcinogen exposures and the onset of cancer. For example, habitual use of tobacco products usually starts in early adolescence, but the onset of tobacco-related cancers, on average, occurs after the age of 60,¹⁶⁵ which is beyond the predicted cancer-protective barriers for humans (Figure 2C). Intriguingly, the apparent evolution of a permissive environment in the aged lungs may be independent of exposure to tobacco-derived toxicants, although there does appear to be an incontrovertible association between the overall increased frequency of lung cancer and increased mutagenesis mediated by these compounds.¹⁶⁶

The impact of these ideas is more profound among species. Extension of this concept indicates that small species that are highly vulnerable to predation and rely on explosive growth strategies, such as wild mice, have short lives, and the cost of cancer-protective mechanisms might decrease their overall fitness.¹⁶⁷ The elevated incidence of cancer in laboratory mice, which far exceed the evolutionarily determined lifespan for this species, is consistent with this observation and argues that exceeding this lifespan barrier in any species will probably result in excess cancer when compared to the same population inhabiting within its evolutionary niche. And the opposite may also be true: “ancient” species or species that are less vulnerable to predation and that have had relatively stable body sizes over “long periods” of evolutionary time such as turtles and alligators should have, and likely did evolve, stronger cancer-protective mechanisms,.^34,a

This concept was also attributed to Dr. Jay Olshansky (University of Chicago) in a Vox article, authored by Joseph Stromberg and published on June 16, 2014 (https://www.vox.com/2014/6/16/5796732/do-naked-mole-rats-have-the-secret-to-long-healthy-lives)

These examples support the ideas that cancer is influenced by evolutionary pressures⁵⁸ and that increased risk associated with modern increases in lifespans may be responsible for the observed increased incidence of cancer in human and canine populations. Large dogs are dually affected as artificial selection for large size seems to have had the unintended consequence of increasing cancer risk, perhaps in part by increasing the replicative risk as large stature requires more cells.^12,162 Figure 4 illustrates this concept using bowhead whales, African elephants, African naked mole rats, ancestral humans, modern humans, ancestral dogs, and modern dogs. Each of these species has achieved an evolutionarily derived balance between tumor-suppressive mechanisms and senescence mechanisms that have selected for lifespans that are consistent with their respective evolutionary pressures. Modern humans and modern domestic dogs are outliers in that their expected lifespan now is more than twice as long as that to which each species had adapted over millions of years of evolution, and for both species, the vast majority of cancers occur in individuals that have crossed this evolutionarily adapted lifespan. Neither species has had the time, nor has it been subjected to selective pressures, to establish a new evolutionary balance that resets its cancer barrier.

FIGURE 4. Increasing canine and human risk due to surpassing the species’ respective evolutionarily determined lifespan barriers. (A) Multicellular species, represented here by bowhead whales, African elephants, African naked mole rats, ancestral humans, modern humans, ancestral dogs, and modern dogs, have achieved an evolutionarily derived balance between tumor-suppressive mechanisms and senescence mechanisms that have selected for lifespans that are consistent with their respective evolutionary pressures. Recent technological advances have led to lifespans in humans and dogs that are more than twice as long as those to which each species had adapted over millions of years of evolution. The vast majority of cancers occur after individuals cross this evolutionarily adapted lifespan (dashed line) for (B) humans and (C) dogs. Neither species has had the time, nor has it been subjected to selective pressures, to establish a new evolutionary balance that resets its cancer barrier

If cancer is virtually inevitable, what is the solution

Cancer is virtually, but not entirely, inevitable. Even with the risks of replication-induced mutations, size, and longevity, only about 25%–40% of all humans and dogs will develop cancer in their lifetime. Incredible breakthroughs have been made in treatments for advanced human cancers, all of which raise hope and might find future applications in dogs.

Nevertheless, the best solution to cancer is to prevent it entirely. Decreasing environmental exposure to carcinogenic agents has been extensively described and pursued in humans.¹⁴ It may be possible to engineer cancer risk out of the genome by increasing endogenous anticancer mechanisms, or potentially by transferring protective strategies utilized by large and long-lived organisms,¹⁶⁸ although this could raise serious ethical dilemmas. In dogs, strategies to decrease environmental exposure to carcinogenic agents are less likely be an effective method to reduce cancer, as the exposure of dogs to such environmental carcinogens is generally quite limited, and at the same time, such exposures might be unavoidable.

In humans, early identification of cancer via screening of at-risk populations has led to effective treatment and decreases in mortality. To extend this approach to dogs, we are developing diagnostic tests to identify canine cancer as early as possible.¹⁶⁹ However, testing for early detection and for risk assessment alone is insufficient. In order to be ethical and valuable, the information from such testing must be actionable. So, our goal is to pair each test with a potential preventive solution that can eliminate the cancer at its earliest stage. One example is the Shine-On project^169,170 (Figure 5), where we have developed a blood test to detect rare events associated with the presence (or risk) of hemangiosarcoma (a malignant tumor of blood vessel-forming cells that is relatively common in dogs), and which is meant to be used once dogs reach an age where risk for this cancer is meaningful (i.e., older than 6 years).^{12,142,171,172} Another example from our group is a serum exosomal mRNA signature associated with the presence of osteosarcoma¹⁷³ that is currently being investigated for early detection in at-risk breeds of dogs. By design, these early detection and risk assessment tests would be paired with a preventive strategy, for example, the targeted drug EGF bispecific angiotoxin (eBAT).^174–176 This preventive solution is based on eBAT's safety profile, which justifies its use in otherwise healthy animals, and on its proposed mechanisms of action to simultaneously eliminate cancer stem cells and disrupt the tumor niche.^174–177 In this way, eBAT may be effective in preventing or delaying the emergence of cancer in healthy dogs determined to be at high risk by eliminating emerging tumors before they ever have a chance to develop. This proof-of-concept in dogs would open the door to develop similar approaches for people.¹⁷⁸ We recognize that these approaches may not be the only or the definitive answer to cancer. But we are convinced that these approaches, and others like them, will be part of the solution. As we tilt the balance toward successful prevention, though, we must remember that it is our duty to support graceful aging. We must resist the siren's song of “longevity at any cost.” It is not wise, and it is not evolutionarily sound.

FIGURE 5. The Shine-On concept of paired early detection and strategic prevention. Shine-On illustrates the approach of early detection and strategic cancer prevention. In this case, dogs that reach the age of increased cancer susceptibility are screened for the presence of incipient or emerging hemangiosarcoma (a terminal tumor of blood vessel-forming cells) using the Shine-On suspicion (SOS) test. The SOS test uses flow cytometry and artificial intelligence to enumerate rare hemangiosarcoma-associated cells in blood and to assign a level of risk. Dogs at high risk would be eligible for targeted prophylaxis, for example, using the drug eBAT, with the intent to prevent or delay the development of clinical hemangiosarcoma

CONCLUSIONS

Our summation of the literature supports the concept that the increased risk of cancer in humans and dogs is a consequence of recent extensions of lifespan and body size beyond evolutionarily determined cancer barriers. In the short-term, we contend that the inevitability of cancer that accompanies these extensions can and should be managed by early detection and risk assessment. For example, screening tests to detect cancer at its inception, before it has become a clinical entity, could be implemented, and paired with rational, strategic interventions such as eBAT that can eliminate the cancer-initiating cell populations and/or make the environment inhospitable for tumor growth. In the long-term, rather than wait for evolutionary innovation, we should carefully consider whether practical and ethical norms would allow engineering of extended cancer-protective mechanisms into at-risk genomes, at least in dog populations that have been rendered highly vulnerable to cancer by artificial selection.

ACKNOWLEDGMENTS

The authors wish to thank Dr. Don Bellgrau and Dr. Logan Spector for their helpful suggestions and critical review of the manuscript, and Curt McAloney (Curt's Media) for assistance with professional illustrations. The work reported in this manuscript was supported in part by the National Cancer Institute of the National Institutes of Health under award numbers R50CA211249 (Aaron L. Sarver) and R21CA208529 (Jaime F. Modiano) and by the United States Department of Defense Congressionally Designated Medical Research Program under award numbers CA170218 and CA190276 (Jaime F. Modiano). Additional support was provided by grants MOU-02234 and MOU-02806 from the AKC Canine Health Foundation (Jaime F. Modiano) and grant D18CA-050 from Morris Animal Foundation (Jaime F. Modiano). The authors also acknowledge generous support from the Karen Wyckoff Rein in Sarcoma Foundation, the Zach Sobiech Osteosarcoma Fund, and the Team Nat Fund of the Children's Cancer Research Fund, and the Animal Cancer Care and Research Program of the University of Minnesota and its many individual donors. Kelly M. Makielski was supported in part by a postdoctoral fellowship from the institutional NIH training grant in Molecular, Genetic, and Cellular Targets of Cancer (T32CA009138). Jaime F. Modiano was supported in part by the Alvin and June Perlman Endowed Chair in Animal Oncology. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, the Public Health Service, the United States Department of Defense, or any of the other agencies that provided support for this work.

CONFLICT OF INTEREST

Jaime F. Modiano is an inventor on U.S. patent 7910315, “Early Detection of Hemangiosarcoma and Angiosarcoma,” assigned to the Regents of the University of Colorado. Jaime F. Modiano is an inventor on U.S. patent application, “Reduction of EGFR Therapeutic Toxicity,” filed on behalf of the Regents of the University of Minnesota. Jaime F. Modiano and Taylor A. DePauw are inventors on U.S. patent application, “Artificial Intelligence for Early Cancer Detection,” filed on behalf of the Regents of the University of Minnesota.

AUTHOR CONTRIBUTIONS

All authors had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Conceptualization: Aaron L. Sarver and Jaime F. Modiano. Methodology: Aaron L. Sarver, Kelly M. Makielski, Taylor A. DePauw, Ashley J. Schulte, and Jaime F. Modiano. Data curation: Aaron L. Sarver and Jaime F. Modiano. Investigation: Aaron L. Sarver, Kelly M. Makielski, Taylor A. DePauw, Ashley J. Schulte, and Jaime F. Modiano. Formal analysis: Aaron L. Sarver and Jaime F. Modiano. Resources: Aaron L. Sarver, Kelly M. Makielski, Taylor A. DePauw, Ashley J. Schulte, and Jaime F. Modiano. Writing the original draft: Aaron L. Sarver and Jaime F. Modiano. Writing the review and editing: Aaron L. Sarver, Kelly M. Makielski, Taylor A. DePauw, Ashley J. Schulte, and Jaime F. Modiano. Visualization: Aaron L. Sarver, Kelly M. Makielski, Taylor A. DePauw, Ashley J. Schulte, and Jaime F. Modiano. Supervision: Jaime F. Modiano. Funding acquisition: Aaron L. Sarver, Kelly M. Makielski, and Jaime F. Modiano.

DATA AVAILABILITY STATEMENT

The data used to illustrate the premises of this study were derived from the following resources available in the public domain: Life expectancy data (1770–2017) in Figure 2A are from Our World in Data and are available at https://ourworldindata.org/life-expectancy. Cancer death rates by age group in Figure 2B are from Our World in Data and are available at https://ourworldindata.org/cancer. Lung cancer incidence statistics in Figure 2C are from Cancer Research UK and are available at https://www.cancerresearchuk.org/health-professional/cancer-statistics/statistics-by-cancer-type/lung-cancer/incidence#ref-1. Age and cancer risk data in Figure 4B are modeled from the NCI Surveillance Epidemiology and End Results (SEER) Program.¹ Cancer attributed deaths in dogs in Figure 4C are modeled from ref.¹⁴²