Cancer causation: the Darwinian downside of past success?
Mel Greaves Lancet Onco/2002; 3: 244-51

MG is Professor of Cell Biology and Director of the Leukaemia Research Fund Centre
for Cell and Molecular Biology
at the Institute of Cancer Research, London, UK.

Correspondence: Professor Mel Greaves, Institute of Cancer Research,
Chester Beatty Laboratories, 237 Fulham Road,
London SW3 6JB, UK.
Tel: +44 (0) 2078783823.
Fax: +44 (0) 2073520266.

Causal mechanisms in all diseases are diverse and multifactorial, but medical scientists, as pragmatists, inevitably focus on limited or circumscribed components of pathogenetic puzzles. In cancer, epidemiologists have traditionally sought to incriminate exposures; geneticists uncover inherited susceptibility; and molecular biologists deconstruct the proximal mechanisms of cell transformation. Molecular epidemiology promises to deliver new insights in terms of gene-environment interactions. Each of these endeavours has undeniably provided rich dividends and insights into cancer causation, but are these likely to be sufficient as a coherent explanation of our vulnerability to cancer? I suggest that the biological plausibility of causal mechanisms would benefit from a historical, evolutionary perspective. The essential argument is that genes or gene variants and phenotypic traits that were adaptively selected in the past as advantageous now contribute crucially to cancer because of their mismatch with current environmental and social circumstances. The risk attributes of skin pigmentation and some dietary factors in cancer can be plausibly interpreted within this context. A case is made here for a Darwinian perspective on breast and prostate cancers, for which current understanding of causation is limited and contentious.

Cancers are a collective of some hundreds of cellular disorders of differing origins and degrees of malignancy, but one essential feature is shared-the territorial expansion of a mutant clone. Mutations, accumulated sequentially within descendent subclones over years or decades, drive the disease via Darwinian or natural selection of cells ( figure 1) .This process is essentially the same as evolutionary divergence or serial dominance of species of microbes, plants, and animals in ecological niches.1,2 The common ingredients are an increase in numbers by reproduction ( cell proliferation), genetic diversification within the expanded clone (further gene mutations or deletions in the stem-cell compartment), environmental pressures and competition (for tissue space and nutrients), and emergence through these bottlenecks of anew , dominant subclone. And the process succeeds partly because the evolution of multicellular creatures, including ourselves as accidental products, has for some 500 million years and longer retained the more ancient proclivity of single cells to clone themselves extensively and emigrate.2 An evolutionary or Darwinian perspective helps rationalise the inherent risk features of stem cells, the protracted latency and variable dynamics of cancer, its colonial tendencies, and its eventual therapeutic intransigence.2 The whole process comes down to natural selection. In this essay, I advance the view that the evolutionary principles of Darwinian adaptation and selection operate also at another level in cancer-in causation.

Figure 1. Clonal evolution of a cancer. All cancers evolve by Darwinian principles: clonal proliferation, genetic diversification within the clone, and selective pressure enabling mutant subclones to bridge the bottlenecks (such as anoxia, restricted space and nutrients, apoptosis imposition). Each colour in the figure represents a cell (and its descendent clone) acquiring the first (blue) or additional, sequential mutations. Grey represents dying cells. This diagram greatly simplifies the extensive genetic diversity, complex population structure, and highly variable dynamics of cancer clones. N, normal stem cells.


Mechanistically, mutations that collectively corrupt cellular control pathways represent the proximate causal mechanism.³ However, in a broader context the 'causes' of cancer include those exposures, genetic characteristics, and physiological factors that induce mutations or influence the probability of mutations. We now have a general audit of exogenous and endogenous exposures that can lead directly or indirectly to cancercausing mutations. Included on the list are ionising radiation, genotoxic chemicals, microbial agents, chronic inflammation, dietary factors, and hormones.4 In some cases, the evidence is incontrovertible: the causal connection between tobacco carcinogens and lung cancer is overwhelmingly the most important in terms of worldwide burden.7 Other clear-cut examples are the links between asbestos fibres and mesothelioma and papillomaviruses and cervical cancer." We are now also beginning to identify many inherited variations in genes that either directly lead to cancer as highly penetrant mutants or, much more commonly, indirectly modify risk of cancer via their influence on the efficiency of the body's mechanisms for carcinogen detoxification, antimicrobial immune response, or repair ofDNA damage.89 The Human Genome Project and its offspring, the Human Cancer Genome Project, will greatly assist the definition of cancerrisk genotypes and help endorse the plausibility of candidate exposures.10 But genetics cannot explain it all; cancer risk can also be modified by other factors, particularly dietary intake.6lll'

Causal networks

From this cocktail of ingredients, a more comprehensive and plausible causal network for cancer risk emerges to replace the misleading concept of a single cause. As in almost all diseases, the causal mechanism is multifactoriall3 and has components that may operate in a combinatorial way ( eg, inherited alleles in a susceptible genotype) .The importance of gene-environment interaction in disease is now generally recognised.l0 To handle the complexity of cancer in this way requires merging of disciplines and assembly of predictive algorithms. It emphasises the importance in causality of the interplay of exposures, genetic and dietary modifiers, and chance (figure 2). Why chance? Because there is a random element in the way that DNA gets damaged and leads to mutations; only if the 'right' gene is mutated or misrepaired in the 'right' way in the 'right' cell does it matter.'3 And given that our genetic inheritance is itself a parental lottery, and our environmental exposures may be inadvertent and invisible, the whole process of cancer is undeniably imbued with chance, much like evolution in general. This is the way things work, by and large, in biology.
Armed with adequate data on genetic profiles and relevant exposures, we might become able to calculate risk of particular cancers for individuals, although my guess is that such risk estimates will remain crude or within wide confidence intervals, given the multiplicity of unquantifiable modifiers of risk and the prevailing influence of chance.

Figure 2. Cancer roulette. Cancer cause is a consequence of exposures (exogenous or endogenous) that lead to oncogenic mutations in stem cells in different tissues, but actual risk is always subject to both modifiers or modulators of risk (principally genetic and dietary factors) and the ubiquitous role of chance (the roulette wheel).


The evolutionary angle

Would all of this constitute a satisfactory, sufficient, or complete explanation of why we get cancer? In a purely mechanistic sense maybe, but in a wider sense I suggest not. What is missing from this formulation is a historical dimension, covering both social and evolutionary time frames. My view is that we cannot make sense of the apparently high incidence or risk of particular cancers in particular groups of individuals or societies without reference to both our evolutionary and more recent historical past.
This argument is part of the perspective of 'evolutionary' or 'Darwinian' medicine proposed most forcibly and eloquently by physician Randolph Nesse and evolutionary biologist George Williams.14-1- (For an excellent recent review see Steams and Ebert.l8) The innovative idea was that the array of chronic diseases that appears to be associated with affluence-obesity, adultonset diabetes, cardiovascular disease, some infectious diseases, allergies, and cancer-has an evolutionary rationale, as may other medical peculiarities of the human condition, such as fetal-matemal conflicts, the menopause, and ageing. The evolutionary arguments are perhaps most well substantiated for pathogen virulence, resistance to antibiotics, and the emergence ofnew infections.I8I9 For the chronic diseases, including cancer, the underlying concept is of a mismatch in which our rapid elaboration of social behaviour patterns has become dislocated from and has outpaced our Stone Age genetics. If the confrontation is protracted by longevity, eventually, something has to give.

A key part of this argument rests on the premise that certain normal (non-mutant) genes and gene variants or alleles selected in the past because they encoded functions that endowed survival or reproductive advantage now have the potential indirectly to increase cancer risk because of a change in the physiological context in which these same genes are now required to operate. Or, in other words, there is a nature-nurture mismatch. This twist arises as a paradoxical consequence of human success at social engineering, engendering exotic lifestyles driven by social and economic rather than biological imperatives, and that, together with longer lifespans, teases out inherent design limitations of our bodies.
The most straightforward and well-rehearsed cancer example is with skin cancers in pale-skinned caucasians, including both the most common cancer of all, basal-cell carcinoma, and the most territorially aggressive of all, melanoma. The essential difference between white skin and black African skin is in the intracellular amounts and packaging of melanin pigment. Genetic studies on rodents suggest that many genes contribute directly or indirectly to skin coloration, but at some time during the migration of modern human beings northwards from Africa, there must have been selective pressure favouring paler skins ( or something linked to paler skin). The reasons for this development are not entirely clear. The prevailing explanation has to do with toning down to increase ultraviolet-light-dependent biosynthesis of vitamin Din cloudy northern regions,20 but whether this is the whole story is tar from certain.2122 Whatever the exact basis of selection and the particular genes involved, genetic programming for paler skins with a diminished melanin ultraviolet filter became a bonus in survival and reproduction. Now, however, that same advantage confers a risk of melanoma and other less belligerent skin cancers. But it is only a serious risk in the context of two other more recent changes-intermittent, intensive exposure to the sun ( eg, roasting at noon on Mediterranean beaches) and living long enough for a clone to accumulate the required set of mutant credentials. Little or no 'reverse' Darwinian selective pressure can be applied because most individuals who develop lethal melanoma are past normal reproductive age. Having said that, however, melanoma now has the fastest rising rate of any cancer in young adults in some parts of the world, particularly in northern Europe. Furthermore, the very high frequency of lethal melanoma in young black Africans with albinism will have exerted a strong negative selective pressure against transmission of their genetically altered skin pigmentation.22

Diet and cancer

A broadly similar argument can be made for our genetic adaptations for dietary intake and metabolism-and the corresponding risk that contemporary eating habits may harbour for several cancer types.6ll12 Dietary components are complex, and the precise way in which they contribute to cancer risk, positively or negatively, remains unresolved and contentious. On the negative side, there is persuasive evidence that deficiency of foods rich in antioxidants, which protect from DNA damage, is bad news, especially coupled with excess calorie intake and the modern trait of a paucity of physical exercise.1223 Many of us now persistently binge, exercise too little, and live long enough for it to matter. These habits adversely affect the associated risk of several cancer types, believed to include those of the breast, prostate, colon, and pancreas, as well as obesity, adultonset diabetes, and heart disease. The energy surfeit idea is rendered credible by observations on substantially lower cancer rates in rodents under calorie restriction2425 and is endorsed by some epidemiological studies.61 11"3 Potential
mechanisms increasing the probability of malignant transformation include disruption of insulin signalling networks and persistently high concentrations of insulinlike growth factor I (IGF I), an inhibitor of cell death programmes.
This story isn't just about gluttony and sloth; there are also genetic factors at play. The so-called 'thrifty genotype' provides a plausible, evolutionary exp1anation of why we have such a problem with these chronic consequences of overeating and a sedentary existence in modern societies."' The argument is a deceptively simple one-that under Stone Age or preagricultural conditions, there might have been a decided survival advantage (and therefore reproductive advantage) for individuals whose genetic profile best equipped them for rapid insulin release and glucose conservation. This pattern would have accommodated excess intake and storage during transient times of plenty, as insurance against the likelihood of subsequent famine or the travails of long-distance migration. Although we do not yet have adequate insight into the many genes and allelic variants that are almost certainly involved in this pathway, certain human populations seemed to be historically well endowed but are now heavily penalised in this respect-for example, the Arizona Pima Indians, Canadian Inuits, Australian aborigines, and Polynesians. Once again, earlier genetic advantage flips to a flaw because the environmental and behavioural context has changed.

Evolutionary take on breast and prostate cancer

So does the evolutionary argument hold water for any of the other major types of cancers? My guess is that it makes sense for most cancers, albeit in varying degrees,2 but let's just consider two of special interest-those of breast and prostate. The two related issues here are the generally high risk of these cancers in more developed societies and the high susceptibility of certain individuals.
Both cancers are very common in North America and northern Europe, less common in eastern and southern Europe, and even less so in oriental countries.2~2' Immigrants moving from low-risk countries to the USA or Australia acquire host-country degrees of risk.~')"'0 These findings alone provide a powerful argument that causation is linked to lifestyle factors or exposures that are socially or geographically variable (and therefore potentially preventable) ..j",2Y For breast ( and ovarian]l) cancer, the general view is that some significant component of risk is attributable to a chronic consequence of our wholesale deviation from the reproductive lifestyle of Stone Age women:232 earlier menarche, delayed pregnancy or none at all,33-35 and lack of protracted breastfeeding.36 Persistent oestrogenic stimulation could lead to mutational DNA damage indirectly via sustained ( cyclical) proliferation and endogenous oxidative stress, via a genotoxic effect of some oestrogen metabolites, or by a combination of these mechanisms. 'Early' full-term pregnancy not only provides a break to oestrogenic stress but also may remove many atrisk cells by terminal differentiation of luminal epithelium.37 The biological plausibility of these associations is supported by experiments in rodents, in which hormonal mimicry of pregnancy protects against mammary carcinogenesis.38
At some time during the past 15 million years, females of higher primate species acquired genetic underpinning of non-seasonal oestrus. The resultant reproductive traits bequeathed to Homo sapiens are reflected in contemporary hunter-gatherer tribes, whose serial pregnancies and protracted breastfeeding habits provide the breaks to otherwise continuous ovarian cycling for 25 or so years.32 Rates of breast cancer in such groups are very low. Progressively, however, along with socioeconomic improvement, emancipation, and contraception, women have adopted reproductive lifestyles for which, it can be argued, they are historically and genetically ill adapted. Pump priming of the ovaries, uterus, and breast in the absence of the expected outcome of pregnancy continues unabated. Sustained oestrogenic stress is the result and cancerous mutations a probable consequence. Hence, the revealing 300-year-old anecdote of Italian nulliparous nuns (figure 3) whose chosen path of celibacy increases their risk of breast cancer ,39-41 while protecting them from sexually transmitted papillomaviruses and cervical cancer .
The effect of these changes seems to be compounded by modern dietary (and exercise deficit) effects on production of oestrogens and timing of menarche.2342 A very plausible concept is that the lottery of who gets breast cancer is largely a combination of modern reproductive and dietary lifestyles mismatched with genotypic traits that may have originally conferred benefits, plus the longer time available for a confrontation, and, of course, chance.
Could the same type of explanation apply to men and their prostates? There is no unambiguous epidemiological evidence, but a plausible argument can be made for a composite risk (figure 2) arising in the evolutionary context. Part of the explanation, given the very high frequency of occult prostate cancer in both occidental and oriental ageing men,43 could be that persistent activity of the prostate incurs cancerous penalty points, provided the man lives long enough. As far as we know, the prostate gland developed for one thing only-lubrication of sperm passage. The evolutionary rationale of the prostate operating at full potential is unclear, but there may be an adaptive advantage of persistent pump priming ( via testosterone) of the prostate to optimise chances of successful fertilisation of females with non-seasonal, covert oestrus.2 This idea might help explain why male human beings have the biggest prostate gland among mammalsalong with the domesticated dog, tellingly perhaps the only other mammal known to suffer from spontaneous prostatic carcmoma.

Figure 3. Nuns and breast cancer. Examination of the breast by the surgeon Teodorico Borgognoni (1275). An apparent excess of breast cancer among nuns was recognised long ago in Europe. The pioneering Italian epidemiologist Bernadino Ramazzini was perhaps the first to draw attention to this excess in his descriptions of lifestyle, occupation, and disease,394o along with his memorable quote, "You can seldom find a convent that does not harbour this accursed pest, cancer, within its walls". More systematic information on the risks of breast and cervical cancers in nuns was provided by Rigoni-Stern in his survey of cancer deaths in Verona, 1760-1839.239 Given the date of the above painting (1275), breast cancer has probably been common in nuns for many centuries. The painting is from Leiden University (MS Vossios Lat F 3, fol 90v) and has been published previously in a review of the history of breast cancer.41

This cannot be the whole story, however, as the internationally varied incidence rates and the influence of migration testify. There must also be an explanation ofwhy age-specific rates of prostate cancer have apparently increased in the past few decades, independently of screening, which has certainly inflated reported incidence rates, especially in the USA.44 Some feature of modern or western lifestyle, as with breast cancer, may, in combination with longevity, be exacerbating a nature-nurture mismatch. Epidemiological studies have yet to provide clear evidence, but one component again may be diet-Iow intake of antioxidants or phytooestrogens, perhaps. There is some evidence from experimental prostate carcinogenesis in mice that the polyphenolic constituents of green tea are protective.45 An unfavourable ratio of diet to exercise, raising concentrations of IGFl or affecting the testosterone signalling pathway may also be important.46

Another factor promoting the transition of incipient to full-blown cancer could be sexual activity itself, recycling stress in the prostate.2 Male human beings may have been adaptively primed in our early evolutionary history for persistent prostate activity, but persistent sexual activity keeping the prostate in business after 'normal' reproductive age ( ie, over age 50 years) is biologically exotic behaviour, albeit very common and natural in modern societies. Needless to say, this activity itself depends on the continued efficacy of testosterone signalling, and therefore sexual activity could be a surrogate marker rather than a cause of prostate stress. Epidemiological evidence bearing directly on this idea is sparse and ambiguous. Studies on the risk of prostate cancer in celibate Catholic priests, for example, have produced mixed results.47 In a study of sexual-behaviour variables and prostate-cancer risk, there was no overall association with number of partners or frequency of intercourse, but there was an unpredicted (by the researchers) observation of a significant trend towards increased risk with greater frequency of intercourse-for men in their fifties, but not at younger ages.48 This uncomfortable issue has not been adequately investigated.

All in the genes?

If these ideas are correct, they might be endorsed by genetic studies. Such studies might also tease out the difference between the general susceptibility that exists at population level in modern western societies as a consequence of behavioural changes from the increased susceptibility inherited by some individuals but not others. We might expect, for example, that inherited genetic polymorphisms that impinge on the oestrogenic and androgenic signalling pathways, and perhaps on insulin circuitry also, would confer altered risk of breast and prostate cancer for certain individuals in excess of the risk derived from the nature-nurture mismatch referred to earlier.
Constitutive genetic factors clearly do contribute very significantly to risks of breast and prostate cancers, around 10% of cases occurring in a familial pattern.49 This clustering has been attributed to the inheritance of dominant or highly penetrant mutant genes including BRCAl and BRCA2. Inheritance of such genes can increase breast-cancer risk up to a 50-80% probability. These dangerous genes have origins, as mutant varieties, hundreds or a few thousand years ago. The most plausible explanation for the high prevalence of individual BRCAl or BRCA2 mutations in certain populations ( eg, Ashkenazi Jews50) is not via any adaptive or 'useful' selection but via the historical founder effect-ie, migrant colonisation of an isolated geographic region by a very small number of people including one individual in whose germ cell the mutation first arose.5l As long as there was no adverse effect on reproductive fitness, there would be no strong selective pressure against mutation carriers (female or male), provided that carriers reproduced before they developed cancer and died.
Other studies suggest, however, that the prevalence of familial pedigrees of breast and prostate cancers underestimates the contribution that inherited genes contribute to risk, not only because new mutant genes can arise in the germline at each generation. Some insight into the contribution of inherited genetics to cancer risk or that of any other disease can be deduced from comparison of frequency or concordance of disease in identical or monozygotic twins versus dizygotic (fraternal) twins. This type of analysis, introduced by Darwin's cousin Francis Galton 125 years ago, has proved very productive in medicine, though interpretation of data is not without difficulty. In the largest analysis of this type for adult cancer, involving a cohort of 44 788 Scandinavian twin pairs, the outcome varied with cancer type.52 For lung cancer, monozygotic twins showed very little, if any, increase in concordance over dizygotic twins, whereas for prostate and breast cancers, the differences were very large. Although these data are subject to some limitations,5354 they suggest that around 40% of prostate-cancer risk and just under 30% of breast-cancer risk can be ascribed to inherited susceptibility. Analysis of risk of contralateral breast cancer and risk in twins has provocatively suggested that most cases of breast cancer may have an important genetic risk component.55
The general interpretation of these data is not, or should not be, that breast and prostate cancers are deterministic genetic diseases like cystic fibrosis or betathalassaemia. Neither is it very likely that the one in ten women in more developed countries at risk of breast cancer over their lifetime were all born harbouring silent mutant genes lurking with malign intent. A more likely explanation is that some normal and common variants ot certain genes can, with moderate probabilities, amplify still further the increased risk of these cancers-ie, a racheting up of the gene-environment mismatch. Several inherited gene variants probably have to operate in concert to constitute a risky genotype. If this interpretation is correct, we face a conundrum-why are risky genotypes tor breast and prostate cancers apparently so common?

Genetic susceptibility: winners become losers?

The explanation may well lie in some part of gene-environment-behavioural interactions with evolutionary overtones. It will doubtless become clearer once we identify all the genes involved and what the proteins they encode actually do, hence the relevance to cancer of the Human Genome Project. To date, studies of candidate gene polymorphisms in breast cancer have yielded somewhat equivocal or inconsistent results.56,57 Arguably the most informative molecular epidemiological data so far for prostate and breast cancers implicate variants in genes with products involved in the sequential synthesis, metabolism, or cellular receptor binding and signalling of androgens and oestrogens.58 64 The evidence currently is more persuasive for prostate than for breast cancer. Significantly, most, if not all, alleles that seem to be associated with increased risk of prostate or breast cancer are known or predicted to encode protein functions that increase activity in the hormonal signalling pathway.58-64 Several, though not all, of these alleles are very common and are more prevalent in ethnic groups with well recognised high risks ( eg, African Americans and prostate cancer).65 Some alleles may be associated with advanced rather than localised disease, thus underpinning late, promotional events.64 All cell signalling networks are regulated by positive as well as negative controls; short polymorphic polyglutalline repeats of the androgen receptor, which increase receptor signalling and prostatecancer risk,61 have the opposite effect in postmenopausal women of decreasing risk of breast cancer, possibly by counteracting oestrogenic signalling.62 In this context, another relevant finding is that the BRCAl gene, which is mutationally inactivated (in one copy) in carriers, encodes a protein that normally represses transcription of the oestrogen receptor.66 The breast-cancer risk in women who carry BRCAl or BRCA2 is high but variable and may itself be modified by coinheritance of polymorphic alleles at other loci. Rebbeck and colleagues found that risk was increased by the presence of long polyglutalline repeats in a gene, AIBI, that increases oestrogenic transcriptional signalling in the mammary gland.67
Altogether, these data clearly point in the direction of genetic control of sex-hormone signalling contributing to variation in risk susceptibility. Critics may argue that these preliminary genetic susceptibility screens with individual candidate genes have given low odds ratios for risks and, in some cases, inconsistent results. That is hardly surprising, particularly since what is likely to be critical is the genotypic variation underpinning a complete signalling pathway or network rather than any individual component. Neither should we expect that only one pathway influences risk of mutational changes in prostate or breast stem cells. Alleles affecting oxidative-stress or DNA-repair pathways are likely to contribute to risk. Vitamin D influences proliferation and apoptosis of prostatic cells, and preliminary genetic data implicate alleles of the vitamin D receptor in risk of advanced prostate cancer .64
Why now would such genotypes constitute an increased risk for cancer? In a superficial sense, the answer is obvious: many, if not all, breast and prostate cancers start their own evolutionary journey being driven by oestrogen and testosterone, respectively. Further, hormone- independent, progression of prostate cancers may be facilitated by ligand-independent triggering of amplified androgen receptors.68 As human beings pass the 5O-year barrier, sex hormone concentrations decline, and the idea that those of us with the more persistently increased hormonal signalling are most at risk seems to make sense. This idea fits with the concept of sustained hormonal stress and accords with the experimental data on tullours induced in animals by long-term administration of sex hormones69,70 and with the anecdotal evidence of breast -cancer development in trans-sexual men injected with high doses of oestrogens.71
These preliminary genetic data and arguments still beg the question of the historical or evolutionary rationale for the apparently high prevalence of risky genotypes. Stephen Gould and some other prominent evolutionary biologists are sceptical of ascribing everything that exists in the biological world to previous adaptive advantage and selection, and any such supposition does require exallination.72 Nevertheless, a Darwinian interpretation would make sense here, although it would be difficult to substantiate. The speculative argument I advance is that selective advantage in our historical past accrued from chance acquisition of gene variants that endowed substantial metabolic advantages ( under prevailing environmental conditions)-vitallin D synthesis, energy storage, or greater reproductive success. Nothing favours Darwinian selection more than survivability and being out front in the mating game.
Women genetically well endowed with enhanced fertility, coupled perhaps with efficient energy storage capacity ( via alleles in insulin circuitry and fat metabolism), would undoubtedly have in past times been at a distinct advantage in terms of passing on their genes. Men with gene variants facilitating more effective prostate priming coupled perhaps not only with fertility, but with mate-attracting potency, hunting ability, and survivability, might be expected to accrue reproductive advantage to be passed on to their descendants. But in contemporary societies, these same beneficiaries may be those most at increased risk of cancer via persistent proliferative activity of breast or prostate stem cells, oxidative stress, and DNA damage, particularly now that lifespans are longer . Evolutionary counter-pressures can exert little or no influence if the major downside or trade-off ( cancer) arises in predominantly ageing and reproductively inactive individuals. The argument is not that all human genes or alleles that are associated with increased cancer risk have a historical rationale in adaptive selection, but that many such genes will have this rationale.
We will not have a wholly satisfactory grasp of the causal mechanisms for breast and prostate cancer until we fully appreciate that cause, in reality, is an interactive network of proximal and distal events. We need to improve in developing algorithms that compute the compound risk derived from genotypic variation, diet, reproductive physiology , and behaviour, and allow for the ubiquitous role of chance. The complex interaction of our unique genotypes and our natural or concocted environments and lifestyles dictates risk of many cancers. How this risk translates into actual rates of disease that vary in time and place may begin to make sense only in the context of our evolutionary and social history.

Search strategy and selection criteria

Relevant publications (up to the end of 2001) were screened by systematic review of the major cancer-research journals (both review journals and those with experimental papers). A PubMed search was used with the key words "Darwinian cancer", "breast cancer, evolution", and "prostate cancer, evolution". Papers were selected on the criteria of direct relevance, a personal judgment on the quality of study design, and strength of conclusion. In some cases, review articles are quoted for convenience rather than primary data sources.


I thank the Institute of Cancer Research and Leukaemia Research Fund for their support, Alan Ashworth, Mitch Dowsett, Jill Ross, Randy Nesse, Freda Alexander, and Barbara Deverson and Chris Priest for help in preparing the paper.


1 Nowell PC. The clonal evolution of tumor cell populations.
Science 1976; 194: 23-28.

2 Greaves M. Cancer: the evolutionary legacy. Oxford: Oxford
University Press, 2000.

3 Hanahan D, Weinberg RA. The hallmarks of cancer.
Cell 2000;lOO: 57-70.

4 Doll R, Peto R. The causes of cancer. Oxford: Oxford University

5 Schottenfeld D, Fraumeni JF, (eds). Cancer epidemiology and prevention,
211d ed. New York: Oxford University Press, 1996.

6 Peto J. Cancer epidemiology in the last century and the next decade.
Nature 2001; 411: 390-95.

7 Peto R, Chen ZM, Boreham J. Tobacco: the growing epidemic.
NatMed 1999; 5: 15-17.

8 Perera FP. Molecular epidemiology: on the path to prevention?
J N atl Cancer I nst 2000; 92: 602-12.

9 Ponder BAJ. Cancer genetics.
Nature 2001; 411: 336-41.

10 Clayton D, McKeigue PM. Epidemiological methods for studying genes and environmental factors in complex diseases.
Lancet 2001 ;
358: 1356-60.

11 World Cancer Research Fund in association with the American Institute for Cancer Research.
Food, nutrition and the prevention of cancer: a global perspective. Washington:
American Institute for Cancer Research, 1997.

12 Griffiths K, Adlercreutz H, Boyle P, et al. Nutrition and cancer.
Oxford: Isis Medical Media, 1996.

13 Dubos R. Mirage ofhealth. New York: Harper and Brothers, 1959.

14 Nesse R, Williams G. Evolution and health: the new science of
Darwinian medicine. London: Weidenfeld & Nicolson, 1995.

15 Nesse RM, Williams GC. Evolution and the origins of disease.
Sci Am 1998; November: 58-65.

16 Stearns SC, (ed). Evolution in health and disease. Oxford: Oxford
University Press, 1999.

17 Trevathan WR, Smith EO, McKenna JJ, (eds). Evolutionary
medicine. New York: Oxford University Press, 1999.

18 Stearns SC, Ebert D. Evolution in health and disease: work in progress.
Quart Rev Biol2001; 76: 417-32.

19 Ewald PW. Evolution of infectious diseases. Oxford: Oxford
University Press, 1994.

20 Loomis WF. Skin pigmentation regulation of vitamin-D biosynthesis in man.
Science 1967; 157: 501-06.

21 Diamond J. The rise and fall of the third chimpanzee. UK: Radi us
Publishing, 1991.

22 Rees J. Post-genome integrative biology: so that's what they call clinical science.
Clin Med 2001; I: 393-400.

23 Mezzetti M, La Vecchia C, Decarli A, et al. Population attributable
risk of breast cancer: diet, nutrition, and physical exercise.
J Natl Cancer Inst 1998; 90: 389-94.

24 Cohen LA, Choi KW, Wang CX. Influence of dietary fat, caloric restriction,
and voluntary exercise on N-nitrosomethylureainduced mammary tumorigenesis in rats.
Cancer Res 1988; 48: 4276-83.

25 Grasl-Kraupp B, Bursch W, Ruttkay-Nedecky B, Wagner A, Lauer B, Schulte-Hermann R. Food restriction eliminates preneoplastic cells through apoptosis and antagonizes carcinogenesis in rat liver .
Proc Natl Acad Sci USA 1994; 91: 9995-99.

26 Neel JV, Weden AB, Julius S. Type II diabetes, essential hypertension, and obesity as "syndromes of impaired genetic homeostasis": the "thrifty genotype" hypothesis enters the 21 st century. Perspect Biol Med 1998; 42: 44-74.

27 Parkin DM, Pisani P, Ferlay J. Estimates of the world wide incidence of eighteen major cancers in 1985. IntJ Cancer 1993; 54: 594-606.

28 Dhom G. Epidemiology ofhormone-depending tumors. In: Voigt KD, Knabbe C, (eds). Endocrine dependent tumors.
New York: Raven Press, 1991: 1-41.

29 Parkin DM. Global cancer statistics in the year 2000.
Lancet Oncol 2001; 2: 533-43.

30 Bouker KB, Hilakivi-Clarke L. Genistein: does it prevent or
promote breast cancer?
Environ Health Perspect 2000; 108: 701-08.

31 Risch HA. Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone.
J N atl Cancer I nst 1998; 90: 1774-86.

32 Eaton SB, Pike MC, Short RV, et al. Women's reproductive
cancers in evolutionary context. Quart Rev Biol1994; 69: 353-67.

33 Henderson BE, Ross RK, Pike MC, Casagrande JT. Endogenous
hormones as a major factor in human cancer.
Cancer Res 1982; 42:3232-39.

34 Lipworth L. Epidemiology of breast cancer.
Eur J Cancer Prev 1995; 4: 7-30.

35 Wohlfahrt J, Melbye M. Age at any birth is associated with breast
cancer risk. Epidemiology 2001; 12: 68-73.

36 Labbok MH. Effects of breastfeeding on the mother.
Pediatr ClinNorth Am 2001; 48: 143.

37 Russo J, Gusterson BA, Rogers AE, et al. Biology ofdisease:
comparative study of human and rat mammary tumorigenesis.
Lab Invest 1990; 62: 244-78.

38 Guzman RC, Yang J, Rajkumar L, et al. Hormonal prevention of
breast cancer: mimicking the protective effect of pregnancy. Proc
Natl Acad Sci USA 1999; 96: 2520-25.

39 Mustacchi P. Ramazzini and Rigoni-Stern on parity and breast
cancer. Arch Intern Med 1961; 108: 639-42.

40 Ramazzini B. Diseases of workers. Translated from the Latin text of
1713 by Wilmer Cave Wright. New York: Hafner Publishing Co,

41 DeMoulin D. A short history of breast cancer. Dordrecht: Kluwer
Publishers, 1983.

42 Coffey DS. Similarities of prostate and breast cancer: evolution, diet, and estrogens.
Urology 2001; 57: 31-38.

43 Breslow N, Chan CW, Dhom G. Latent carcinoma of prostate at autopsy in seven areas.
Int J Cancer 1977; 20: 680-88.

44 Hsing A W , Tsao L, Devesa SS. International trends and patterns of prostate cancer incidence and mortality.
Int J Cancer 2000; 85:60-67.

45 Gupta S, Hastak K, Ahmad N, et al. Inhibition of prostate carcinogenesis in TRAMP mice by oral infusion of green tea
polyphenols. Proc Natl Acad Sci USA 2001; 98: 10350-55.

46 Cohen P. Serum insulin-like growth factor-I levels and prostate cancer risk: interpreting the evidence.
J Natl Cancer Inst 1998; 90:876-79.

47 Kaplan SD. Retrospective cohort mortality study of Roman
Catholic priests. Prev Med 1988; 17: 335-43.

48 Hayes RB, Pottern LM, Strickler H, et al. Sexual behaviour, STDs
and risk for prostate cancer. Br J Cancer 2000; 82: 718-25.

49 Szabo CI, King M -C. Population genetics of BRCA I and BRCA2.
Am J Hum Genet 1997; 60: 1013-20.

50 Roa BB, Boyd AA, Volcik K, Richards CS. Ashkenazi Jewish population frequencies for common mutations in BRCAl and
BRCA2. Nat Genet 1996; 14: 185-87.

51 Diamond JM, Rotter JI. Observing the founder effect in human
evolution. Nature 1987; 329: 105-06.

52 Lichtenstein P, Holm NV, Verkasalo PK, et al. Environmental and heritable factors in the causation of cancer: analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med 2000; 343: 78-85.

53 Hoover RN. Cancer: nature, nurture, or both. N Engl J Med 2000;
343: 135-36.

54 Risch N. The genetic epidemiology of cancer: interpreting family and twin studies and their implications for molecular genetic
approaches. Cancer Epidemiol Biomarkers Prev 2001; 10: 733-41.

55 Peto J, Mack TM. High constant incidence in twins and other relatives ofwomen with breast cancer. Nat Genet 2000; 26:

56 Dunning AM, Healey CS, Pharoah PDP, et al. A systematic review of genetic polymorphisms and breast cancer risk. Cancer
Epidemiol Biomarkers Prev 1999; 8: 843-54.

57 Clemons M, Goss P. Estrogen and the risk of breast cancer. N Engl
JMed2001; 344: 276-85.

58 Ross RK, Pike MC, Coetzee GA, et al. Androgen metabolism and prostate cancer: establishing a model of genetic susceptibility.
Cancer Res 1998; 58: 4497-504.

59 Feigelson HS, McKean-Cowdin R, Coetzee GA, et al. Building a multigenic model of breast cancer susceptibility: CYP17 and HSDl7Bl are two important candidates. Cancer Res 2001; 61: 785-89.

60 Habuchi T, Liqing Z, Suzuki T, et al. I ncreased risk of prostate
cancer and benign prostatic hyperplasia associated with a CYP17gene polymorphism with a gene dosage effect. Cancer Res 2000; 60: 5710-13.

61 Stanford JL, Just JJ, Gibbs M, et al. Polymorphic repeats in the androgen receptor gene: molecular markers of prostate cancer risk. Cancer Res 1997; 57: 1194-98.

62 Giguere Y, Dewailly E, Brisson J, et al. Short polyglutamine tracts in the androgen receptor are protective against breast cancer in the general population. Cancer Res 2001; 61: 5869-74.

63 Makridakis NM, Ross RK, Pike MC, et al. Association of mis-sense substitution in SRDSA2 gene with prostate cancer in AfricanAmerican and Hispanic men in Los Angeles, USA. Lancet 1999; 354: 975-78.

64 Ingles SA, Ross RK, Yu MC, et al. Association of prostate cancer risk with genetic polymorphisms in vitamin D receptor and androgen receptor. J Natl Cancer Inst 1997; 89: 166-70.

65 Shibata A, Whittemore AS. Genetic predisposition to prostate cancer: possible explanations for ethnic differences in risk. Prostate 1997; 32:

66 Zheng L, Annab LA, Afshari CA, et al. BRCA1 mediates ligand independent transcriptional repression of the estrogen receptor . ProcNatlAcad Sci USA 2001; 98: 9587-92.

67 Rebbeck TR, Wang Y, Kantoff PW, et al. Modification of BRCA land BRCA2-associated breast cancer risk by AIBI genotype and reproductive history. Cancer Res 2001; 61: 5420-24.

68 Visakorpi T, Hyytinen E, Koivisto P, et al. In vivo amplification of the androgen receptor gene and progression ofhuman prostate cancer. Nat Genet 1995; 9: 401-06.

69 Noble RL. The development of prostatic adenocarcinoma in Nb rats following prolonged sex hormone administration. Cancer Res 1977; 37: 1929-33.

70 Kwapiem RP, Giles RC, Geil RG, Casey HW. Malignant mammary tumors in beagle dogs dosed with investigational oral contraceptive steroids. J Natl Cancer Inst 1980; 65: 137-42.

71 Symmers WSC. Carcinoma of breast in trans-sexual individuals after surgical and hormonal interference with the primary and secondary sex characteristics. BM] 1968; 2: 83-85.

72 Rose MR, Lauder GV, (eds). Adaptation. San Diego: Academic
Press, 1996.