Challenging the axiom: Does the occurrence of oncogenic mutations truly limit cancer development with age?

Abstract

A widely accepted paradigm in cancer research holds that the development of cancers is rate-limited by the occurrence of oncogenic mutations. In particular, the exponential rise in the incidence of most cancers with age is thought to reflect the time required for cells to accumulate the multiple oncogenic mutations needed to confer the cancer phenotype. Here I will argue against the axiom that the occurrence of oncogenic mutations limits cancer incidence with age, based on several observations, including that the rate of mutation accumulation is maximal during ontogeny, oncogenic mutations are frequently detected in normal tissues, the evolution of complex multicellularity was not accompanied by reductions in mutation rates, and that many oncogenic mutations have been shown to impair stem cell activity. Moreover, while evidence that has been used to support the current paradigm includes increased cancer incidence in individuals with inherited DNA repair deficiencies or exposed to mutagens, the pleotropic effects of these contexts could enhance tumorigenesis at multiple levels. I will further argue that age-dependent alteration of selection for oncogenic mutations provides a more plausible explanation for increased cancer incidence in the elderly. While oncogenic mutations are clearly required for cancer evolution, together these observations counter the common view that age-dependence of cancers is largely explained by the time required to accumulate sufficient oncogenic mutations.

The field of cancer research is dominated by the view that oncogenesis is rate-limited by the incidence of oncogenic mutations, and that these mutations are typically advantageous when they occur in the right cell type. Such oncogenic mutations (including activation of proto-oncogenes and deactivation of tumor suppressor genes, whether by genetic or epigenetic mechanisms) are thought to provide cells with various “Hallmarks of Cancer”, including sustained proliferative signaling and resistance to growth suppressive and cell death signals (1). In particular, it is widely accepted that the exponential increase of cancer incidence with age reflects the time required for cells to accumulate sufficient numbers of genetic and epigenetic mutations to confer the cancer phenotype (2–8). This paradigm in part originates with the classic modeling studies of Armitage and Doll (9, 10), which showed that the incidence of cancers increases with around the 6^th power of age, suggesting that the age-dependent accumulation of 6–7 oncogenic mutations is needed for cancer development. The logic is quite simple: Aging → Mutations (including Oncogenic Mutations) → Cancer.

Oncogenic mutations are clearly required for cancer evolution, and increases in genetic/epigenetic diversity in somatic cells associated with aging should contribute to cancer incidence. Increased rates of genomic instability in some cancers can also help promote tumor evolution (1). However, I will argue that the axiomatic attribution of the rising incidence of cancers with age primarily to the accumulation of oncogenic mutations is insufficiently justified, stimulating the following questions: Why is aging associated with increased cancer incidence? Is the association explained by the requirement to increase the accumulation of oncogenic mutations? Or are pre-existing oncogenic mutations largely the substrate upon which age-dependent alterations in selection act? This perspective will not address whether cancer evolution in general involves or requires accelerated mutation accumulation, unless this acceleration were due to aging.

I will describe evidence to challenge the axiom that the occurrence of oncogenic mutations limits cancer incidence with age. I will argue that the age-dependent accumulation of mutations plays a relatively minor role in the increased incidence of cancer with age. Instead, other aging-associated changes, such as alterations in tissues that influence the selection for oncogenic events, largely underlie the aging-association of cancers.

The rate of mutation accumulation over time is maximal during ontogeny (development to maturity)

While the commonly accepted idea is that accumulation of oncogenic mutations with age accounts for the age-dependence of cancer, many if not most mutations appear to accumulate during ontogeny, rather than during adulthood (11–13). The maintenance of self-renewing adult tissues may require relatively few stem cell divisions. Indeed, most telomere shortening in human cells occurs before birth (14). It is estimated that any given hematopoietic stem cell (HSC) will divide on average 5–10 times through the life of an adult mouse (from maturity to 2–3 years of age) (15), and yet one would surmise that the generation of each HSC in a young mouse required far larger numbers of cell divisions (counting from the one-celled zygote). Thus, it is not surprising that a substantial fraction of mutations and epigenetic changes would occur and accumulate during ontogeny (12, 13), followed by a more modest rate of accumulation during tissue maintenance post-maturity (although an accelerated accumulation of mutations/genomic rearrangements in late life has been observed (16, 17), and DNA repair mechanisms, at least as assessed in cell culture, may become impaired in old age (18)). At least as assessed using transgenic reporters in C57BL/6 mice, accumulation of mutations (including genome rearrangements) from maturity through old age is relatively modest (~2–3 fold in some tissues) (12). In particular, the accumulation of mutations in the spleen is negligible from maturity through old age (12, 19) (Figure 1), consistent with the very low division rates for HSC past maturity (15). Since leukocytes in the spleen are derived from HSC and are relatively short-lived, they should serve as adequate proxies for the analysis of mutation accumulation in HSC and hematopoietic progenitors. As the most common malignancies that develop in old C57BL/6 mice are lymphocytic (20), the paucity of aging-dependent accumulation of mutations in the hematopoietic compartment is at odds with the mutation-centric paradigm. That the incidence of most cancers rises late in life (Figure 1), with kinetics that are quite disconnected from the time-dependent accumulation of mutations (maximal during ontogeny) in the tissues from wince these cancers arise, argues that age-dependent acquisition of oncogenic mutations is not a rate-limiting step in tumorigenesis. Indeed, Frank has proposed that cancers that develop in old age may in fact depend on oncogenic mutations accumulated during ontogeny (13).

Figure 1. Comparison of the time-courses of mutation accumulation with onset of malignancies in the hematopoietic system in mice.

Stylized curves represent the numbers of mutations detected in spleens (gray) (12, 19) and lymphoma incidence (black) (20) in C57BL/6 mice.

Oncogenic mutations are frequently detected in non-diseased tissues

Oncogenic mutations are necessary, but not sufficient, for tumorigenesis. Many barriers to tumor progression exist and it is well known that multiple oncogenic mutations are required for the emergence of clinically-detectable cancers (21, 22). Still, if the incidence of oncogenic mutations were a rate-limiting step in tumorigenesis, one would not expect an abundance of oncogenic mutations in the absence of tumorigenesis. Nonetheless, cell clones with mutational or epigenetic inactivation of the PTEN or INK4A tumor suppressor genes are frequently found in histologically normal endometria and breast (respectively) of cancer free women (23, 24), far outpacing the incidence of the corresponding cancers. Furthermore, the presence of TEL-AML1 and AML1-ETO translocations in blood cells of newborns is ~100-fold greater than the risk of the associated leukemias (25). Perhaps most surprisingly, histologically advanced microscopic tumors are detected in many tissues of adult humans (22, 26), but which appear to be mostly held in check by unknown mechanisms. In addition, even though it is thought that the incidence of chronic myeloid leukemia (CML), which increases exponentially late in life, is limited by the occurrence of the initiating Bcr-Abl translocation (27), in frame Bcr-Abl fusions are detected in leukocytes of ~1 in 3 healthy individuals (28, 29), the vast majority of which will fortunately never develop this leukemia (despite persistence of the translocation in leukocytes for long enough to suggest an HSC origin (30)). Notably, CML in chronic phase is thought to be a simple leukemia (a myeloproliferative disorder), with Bcr-Abl as the only recurrent mutation (31). In mouse models, the expression of Bcr-Abl provides a much greater advantage to progenitor cells in an aged as compared to young hematopoietic system, leading to increased clonal expansion and leukemogenesis (32). Thus, at least in this mouse model, the occurrence of the same oncogenic mutation, Bcr-Abl, results in very different outcomes dependent on the age of the target tissue.

The evolution of multicellularity has not been accompanied by decreases in DNA mutation rates

If increased acquisition of oncogenic mutations is the primary driver of oncogenesis, organisms that are more likely to acquire oncogenic mutations should be more likely to get cancer, and larger/longer-lived animals would have been expected to have evolved mechanisms to limit mutation accumulation. Thus, it might seem surprising, but mutation rates are higher in larger and more complex organisms like mammals than they are in prokaryotes, single celled eukaryotes, and simpler multicellular organisms (11). Additionally, within mammals, somatic mutation rates are even higher than those for the germline. For example, the average somatic mutation rate for humans across four tissues (~1 × 10⁻⁹/site/cell division) is ~17-fold higher than the germline rate, and surprisingly, several-fold higher than rates for S. cerevisiae and E. coli (11). While some studies have shown that the efficiency of DNA excision repair among mammals is proportional to lifespan and/or body size (at least for fibroblasts exposed in vitro to ultraviolet light; UV) (33, 34), these differences have not been shown to coincide with similar changes in mutation accumulation in tissues in vivo and UV-induced excision repair may be less relevant for small nocturnal mammals (in fact, recognition of UV-induced cyclobutane pyrimidine dimers is suppressed in rodent cells (35)). Indeed, while the disparate methods used muddy comparisons (and highlight the need for direct measurements of mutations in mammals with age), rates of mutation accumulation in somatic tissues are similar between rodents and humans (11). So of all of the proposed evolved tumor suppressive mechanism that keep cancer rates sufficiently low in multicellular organisms long enough to promote reproductive success (36), improvements in DNA repair do not appear to have been harnessed during the evolution of bigger, more complex and longer-lived animals. DNA fidelity mechanisms were apparently already “good enough” to limit cancer through reproductive years, and the evolution of tumor suppressive mechanisms with increasing multicellularity did not require further refinement.

DNA repair deficiency, mutagens, and cancer: complicated relationships

The increased cancer incidence associated with inherited DNA repair deficiencies or exposure to DNA damaging agents is often cited as key support for the argument that time-dependent accumulation of oncogenic mutations is responsible for the rise of cancer rates with aging (e.g. (2, 4, 6, 37)). The logic seems simple: agents that increase mutation frequency also increase cancer incidence. Though easily understood, this rationale bypasses important characteristics of diseases associated with inherited DNA repair defects, particularly cancer-promoting characteristics that extend beyond increased frequencies of oncogenic mutations. For example, ATM (ataxia telangiectasia mutated) deficiency also reduces the fitness (see Box 1 for definitions) of HSC, increases reactive oxygen species, alters metabolism, promotes inflammation, and decreases immune function (37–40), all of which could contribute to cancer evolution at multiple levels. These pleotropic effects of ATM loss emanate both from impaired DNA repair as well as non-repair functions of ATM. Thus, for DNA repair deficiencies, it is difficult to assign the blame for increased cancer to a particular consequence of the genetic defect.

Box 1. Glossary.

Adaptive- increases fitness (e.g. a mutation that increases cellular fitness would be adaptive); note that whether a mutation is adaptive should be context dependent.

Adaptive Landscape- relationships between genotype and cellular fitness; in evolutionary biology, these landscapes describe how changes in genotype (and the corresponding phenotype) influence organismal fitness, but should also describe similar relationships for cells.

Cell fitness- a measure of the ability of a cell to pass its genotype on to future cell generations; cell fitness is not simply a measure of cell duplication rates; for example, certain stem cells need to limit cell cycling to maintain themselves within the stem cell niche (necessary to remain as stem cells)

Niche- for a stem cell, this refers to the environmental factors (other cells, matrix and soluble factors) that influence the properties of the cell

Moreover, an increase in mutation rates does not always confer increased cancer incidence. Heterozygous mutation of the DNA polymerase δ (at L604G and L604K) in mice increases mutation rates 4–5 fold in embryonic fibroblasts, with an even larger increase in the frequency of chromosomal aberrations (>17 fold), but without increasing the incidence of cancer (41). In particular, since the L604G/+ mice have normal lifespans and cancer incidence, increased mutation rates in these mice appear to have been uncoupled from any changes in both cancer and overall physiology (contrasting with ATM deficient mice). In contrast, heterozygous L604K mutation leads to a shorter lifespan, and cancers in L604K/+ mice show accelerated progression (but with similar tumor incidence to +/+ mice). One could conclude that L604K accelerates tumor development by either increasing genetic diversity or by altering selective pressures; the absence of a similar acceleration in the L604G/+ mice would argue for the latter, although the two explanations need not be mutually exclusive. Although in vivo mutation rates have not been determined for L604K/+ and L604G/+ mice, the increases in L604 DNA polymerase δ mutant mice are likely to eclipse the modest accumulation of mutations in mice from maturity to old age (12). Thus, increasing mutation rates is apparently not sufficient for increased tumorigenesis. Still, other loss-of-function mutations in DNA polymerases do increase cancer incidence in mice (42), indicating that there is either a threshold increase in mutations required to increase cancer rates over background or that other effects of these mutations on mouse physiology may promote the increased cancer incidence.

Analogous concerns can be raised for associations between exposure to DNA damaging agents and cancer, as these agents (and the resulting DNA damage) similarly cause pleotropic effects (reduced progenitor cell fitness, increased inflammation, increased cell turnover, decreased immune function, etc.) (43). Thus, the extent to which radiation, chemotherapy treatments, and other mutagenic exposures increase cancer rates by inducing oncogenic mutations cannot currently be determined.

Fitness, selection and cancer evolution: an alternative model

For organismal evolution, natural selection works on heritable diversity, and major periods of speciation (such as the Cambrian Explosion) were likely due to altered environmental selection pressures rather than increases in mutation rates. Analogously, numerous investigators have stressed the importance of the microenvironment in cancer development and the critical role of altered selection (21, 22, 44–49). Dramatic changes in tissue microenvironments occur with age, including stromal changes and increased inflammation (50, 51). These age-dependent changes should substantially alter adaptive landscapes (relationships between genotype and cellular fitness; Box 1), which describe how mutational changes can be adaptive, maladaptive or neutral in a context-dependent fashion. Alterations in adaptive landscapes in old age should promote selection for particular oncogenic mutations from within the standing genetic/epigenetic variation (Figure 2), whether it arose from endogenous (oxidative damage, replicative errors, etc.) or exogenous (exposure to environmental carcinogens) insults. Indeed, recent studies indicate that the frequency of cells with clonally-expanded genomic rearrangements increases substantially after 50–60 years of age in humans, correlating with cancer risk, which could reflect alterations in the adaptive landscape, increased rates of genomic alterations, and/or decreased stem cell polyclonality (52–54). Notably, in another study, 5 of the 6 detected clonally-expanded chromosomal abnormalities were present in both bladder and blood, suggesting an early embryonic origin of the events (55). Mutations that arise during ontogeny (or anytime after) but were neither adaptive nor maladaptive at the time, may be adaptive in the new landscape, thereby conferring a selective advantage and promoting clonal expansion (36, 48). Context-dependent selection leading to expansion of the oncogenically mutated clone would then greatly increase the likelihood of acquisition of secondary oncogenic mutations in cells that harbor an initiating lesion. Moreover, some of these oncogenic events selected for by the age or carcinogen altered adaptive landscapes could then contribute to increased genomic instability, providing more fuel for selection to act upon. While new mutations that accumulate with age should increase the cellular variation subject to selection, this alternative model does not depend on age-dependent accumulation of mutations to explain increased cancer incidence in old age.

Figure 2. Selection-centric model.

This model posits that aging is largely associated with cancer due to alterations in selection for oncogenic mutations. The weight of the arrow reflects the proposed contribution to cancer incidence.

To understand cancer evolution, we should consider why large and long-lived multicellular organisms like ourselves are so good at not getting cancer (36). For example, what tumor suppressive mechanism could allow for mammals as diverse as mice and blue whales to largely avoid cancer through their reproductive years (Peto’s Paradox)? As argued above, the commonly accepted view that cancer incidence is rate-limited by the occurrence of oncogenic mutations does not appear to be consistent with the common presence of oncogenic mutations in normal tissues, with the most rapid accumulation of mutations during ontogeny, and with the lack of reductions in somatic mutation rates during the evolution of complex multicellularity. We have proposed that cancer avoidance through reproductive years is dependent on the same basic principle that governs the avoidance of other hallmarks of aging: investments are made in tissue maintenance to the extent that provides the best return in terms of reproductive success. Thus, we have argued that the maintenance of tissue stem and progenitor cell fitness is inherently tumor suppressive, as high cellular fitness should disfavor selection for phenotype-altering somatic mutations (see (36, 48) for a full description of this “Adaptive Oncogenesis” model). Of course, other mechanisms, such as alterations in how telomeres are maintained (56, 57), could also contribute to similar tumor suppression through reproductive years for species with hugely different sizes and lifespans.

If we again consider HSC, given that hematopoietic malignancies are common in mice and that HSC are the best characterized stem cells, it is striking that mutations defined as oncogenic (activation of an oncogenic pathway, either by tumor suppressor gene deletion or oncogene expression) typically exhibit a common phenotype in HSC: loss of self-renewal (Table 1). For this table, I have assembled all published reports that I could find which describe oncogenic mutations engineered in mouse HSC under reasonably physiological contexts (i.e. in young unperturbed bone marrow at steady-state). It is notable that even mutations, such as in PTEN (58, 59), which increase proliferation (leading to initial expansion of short-term progenitors), impair HSC maintenance. In fact, a common effect of oncogenic mutations in HSC is to increase cell cycling (60), which likely contributes to loss of self-renewal: HSC maintenance necessitates an appropriate level of quiescence. Thus, we would expect that these mutations, should they occur in an individual HSC, would lead to clonal exhaustion by differentiation. Finally, it is notable that many of these mutations have been shown to be advantageous in vitro. For example, β-catenin activation increases HSC self-renewal and expansion in vitro (61). Animals did not evolve stem cells that would be well adapted to in vitro culture, and certain oncogenic events can be adaptive under such stressful conditions. The studies summarized in Table 1 provide support for the model that stem cells occupy a local fitness peak on the adaptive landscape, such that changes in phenotypic parameters will be rarely advantageous and typically disadvantageous (Figure 3).

Table 1. Oncogenic mutations typically impair HSC maintenance.

The effects of oncogenic mutations on HSC analyzed in genetically engineered mice. Data summarized for effects on HSC were, at least in large part, derived from analyses of unperturbed young mice (HSC at steady-state).

Genotype	Pathway Deregulated	Effect on HSC	Corollary	References
Rb cKO	CDK/Rb/E2F	↓self-renewal	↑cell cycle	(74–76)
p21CIP1−/−	CDK/Rb/E2F	↓self-renewal	↑cell cycle	(77)
PTEN cKO	PI3K/AKT/mTOR	↓self-renewal	↑differentiation; ↑cell cycle	(58, 59)
p16INK4A−/−	CDK/Rb/E2F	↓self-renewal	↑cell cycle	(71)
ATM−/−	DDR/metabolism	↓self-renewal	↑ROS	(78)
p53−/− & cKO	p53	no effect		(72, 73)
GSK3 knockdown & GSK3b−/−	Wnt/β-catenin mTOR	↓self-renewal; loss of HSC	↑cell cycle	(79)
APC cKO	Wnt/β-catenin mTOR	↓self-renewal	↑differentiation; ↑ROS	(80)
TSC1−/−	mTOR	↓self-renewal	↑differentiation; ↑ROS	(81, 82)
LKB cKO	mTOR/AMPK	↓self-renewal	↑apoptosis ↑cell cycle	(83)
EGR1−/−	various	↓self-renewal; HSC exhaustion	↑cell cycle; ↑mobilization	(84)
FBW7 cKO	Notch; Myc; Cyclin E	↓self-renewal; ↓CRC	↑cell cycle; ↑Myc; ↑apoptosis	(85, 86)
MEN1 cKO	Interacts with	↓hematopoietic	↓↓HSC function	(87)
(MENIN)	MLL	output; ↓CRC	under stress
c-CBL−/−	Tyrosine kinase signaling	↑HSC#; ↑CRC	↑cell cycle; ↑STAT5 activation	(62)
Ikaros LOF mt	Ikaros	↓self-renewal; loss of HSC	impaired differentiation	(88–90)
β-catenin activation	Wnt/β-catenin/mTOR	loss of HSC; ↓differentiation	↑apoptosis ↑cell cycle	(91–93)
Myc o/e	Myc	↓self-renewal	↑cell cycle	(94)
Rheb2 o/e	mTOR	↓self-renewal	↑cell cycle	(95)
Kras(G12D)	Ras/MAPK	↑CRC; ↓↓HSC #	↑cell cycle	(66)
Bcr-Abl (inducible)	Ras, AKT, STAT, others	↓HSC #; ↑CRC	↑cell cycle; ↑differentiation	(67)

cKO, conditional knockout; o/e, transgenic overexpression; CDK, cyclin-dependent kinase; CRC, competitive repopulating capacity.

Figure 3. The Goldilocks Rule for stem cells.

A. Young healthy stem cells are proposed to possess parameters (cell cycle, differentiation, interactions with the niche, etc.) that are near optimal (“just right”) for maintenance as a stem cells. Stem cells are presumed to occupy a local fitness peak on the adaptive landscape; cancer cells in the same tissue could occupy a higher peak, but transitions to this peak would require passage through lower fitness states on the landscape (see (36)). Thus, acquisition of a single oncogenic mutation would typically be disadvantageous, by changing parameters from their optimum (see Table 1). B. For old or damaged stem cells, parameters are suboptimal or abnormal, and the stem cells no longer possess optimal or near optimal fitness. Changes in parameters could result from both cell-autonomous events (damage to the stem cells) or from non-cell autonomous changes (such as degradation of the niche or systemic changes). These changes in the stem cell pool can lead to selection for oncogenic events that are adaptive to this context.

While the Adaptive Oncogenesis model posits that oncogenic mutations should rarely be advantageous within young, fit stem cell pools, there are potential exceptions. First, c-CBL−/− mice exhibit increased numbers of HSC, and these HSC exhibit increased cycling and greater reconstitution potential in competitive bone marrow transplantation experiments (62) (Table 1). C-CBL is an E3 ligase that downregulates tyrosine kinase signaling. Gain-of-function mutations and translocation of c-CBL are implicated in several cancers including myeloid neoplasms (63, 64), and c-CBL−/− mice exhibit tissue hyperplasia (65). It will be interesting to determine if mutation of c-Cbl in an isolated HSC indeed proves advantageous in a young healthy mouse (as opposed to a mouse deficient in c-CBL in all tissues). Second, the induction of a KRAS^G12D mutation in mice leads to competitive expansion of the hematopoietic clones (including HSC) expressing activated K-Ras, despite a dramatic loss of functional HSC numbers (66). As noted by the authors, conditional activation of KRAS^G12D occurs in many (if not all) tissues, and thus K-Ras^G12D expression in non-hematopoietic tissues could alter the microenvironment for HSC (and thus the adaptive landscape), which could also explain the reductions in HSC numbers. Another possible exception is for Bcr-Abl. Reynaud et al showed that activation of Bcr-Abl expression in unperturbed mice results in reduced HSC numbers, apparently by increased differentiation to more committed myeloid progenitors (67), which nicely supports our model that young unperturbed HSC favor the status quo (the youthful phenotype). These results are also consistent with previous studies which indicate that Bcr-Abl promotes differentiation of human HSC, inhibiting self-renewal (68, 69), and that selection for Bcr-Abl is context dependent (32, 70). However, in the Reynaud et al study, following transplantation into irradiated recipient mice, Bcr-Abl provides a competitive advantage to HSC (67). Irradiation clearly alters the bone marrow microenvironment (and thus the stem cell niche), which could impact upon the adaptive landscape and promote selection for Bcr-Abl mutation. Of course, the alternative explanation is that some oncogenic mutations can be advantageous even in young healthy HSC pools, and it is the small size of this stem cell pool (together with other hurdles to tumorigenesis, such as the need for multiple oncogenic mutations) combined with the inability of these oncogenes to initiate cancer in more committed hematopoietic progenitors that sufficiently limits leukemias initiated by these oncogenes until older ages.

Just as high cellular fitness should prevent the fixation of phenotype-altering mutations, the converse should also be true: reductions in progenitor cell fitness with aging or other insults should increase selection for oncogenic mutations adaptive to the particular context (Figure 3). For example, while INK4A (encodes the p16 cyclin-dependent kinase inhibitor) mutation reduces the self-renewal of young HSC (Table 1), p16 loss actually increases the self-renewal of old HSC (which exhibit self-renewal defects) (71), and thus we would expect that p16 loss would be adaptive in old HSC pools. Similarly, while loss of p53 does not provide an advantage within young healthy hematopoietic pools, p53 mutation is potently selected for within HSC and more committed progenitor pools following irradiation of mice (72, 73). Finally, we have shown that Bcr-Abl is adaptive in old hematopoietic progenitor pools, but not young, by restoring kinase signaling pathways that are reduced in old progenitors (32). Thus, just as maintenance of fit stem cell pools should be tumor suppressive by disfavoring phenotype-altering mutations, reductions in the fitness of stem cell pools (such as during aging or following irradiation) should increase selection for particular oncogenic mutations adaptive to the altered context.

Conclusions

The current paradigm that the occurrence of oncogenic mutations with age is rate limiting for cancer development has provided a framework for a large body of cancer research, particularly for the field of carcinogenesis. I have raised questions to challenge this paradigm: if cancer were rate-limited by the occurrence of oncogenic mutations with age: Why would cancers increase exponentially late in life given that mutation accumulation rates are maximal during ontogeny? Would we expect the frequency of oncogenic mutations in tissues to far outpace the rates of corresponding cancers? Why have mutation rates not decreased during the evolution of larger and longer-lived species? Why would oncogenic mutations impair stem cell maintenance? In addition, I have argued that the cause-and-effect relationships between inherited DNA repair deficiencies (or mutagen exposure), oncogenic mutations and cancer incidence are far from established. There are important implications for a revised understanding of the relationships between aging, carcinogens and cancer incidence. From a practical standpoint, perhaps we should be more concerned about how aging, environmental exposures, and therapies impact on the overall tissue landscape, especially given that prevention of changes in adaptive landscapes is probably more feasible than the prevention of the occurrence of mutations. Whereas limiting the incidence of cancers through maintenance of healthier tissues might turn out to be a reasonable preventative approach, meaningful developments in this area will necessitate challenging the current dogma.