Is cancer a matter of luck?


In 2015, Tomasetti and Vogelstein published a paper in Science containing the following provocative statement: “… only a third of the variation in cancer risk among tissues is attributable to environmental factors or inherited predispositions. The majority is due to “bad luck,” that is, random mutations arising during DNA replication in normal, noncancerous stem cells.” The paper—and perhaps especially this rather coy reference to “bad luck”—became a flash point for a series of letters and reviews, followed by replies and yet further counterpoints. In this paper, I critically assess Tomasetti and Vogelstein's argument, discuss the meaning of “luck” (or, better: “chance”) in the context of the debate, and use this case study to address larger questions about methodological criteria for causal explanations of population level patterns in biomedicine.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. 1.

    As we will see, there is some disagreement among the critics concerning the most serious flaw in Tomasetti and Vogelstein’s argument. These differences at least in part overlap with disciplinary specialty, suggesting different pragmatic interests, methodological commitments, and perhaps also different views about causation, causal selection, and explanation, drove this debate (see also, e.g., O’Rourke et al. 2016; Brigandt 2013).

  2. 2.

    Unfortunately, when biologists talk of this or that class of event as being a matter of “chance,” they are sometimes unclear about whether they intend to refer to our epistemic state (e.g., given our knowledge of various initial conditions, we can at best assign the events some probability), or a state of the world (i.e., the events are in fact not determined). Nonetheless, there are some contexts in which—at least with respect to relevant (known) macro-causal variables (e.g., environmental exposures, such as UV radiation)—biologists do have a very good estimate of the probability of various classes of outcome (e.g., melanoma). For the purposes of my discussion here, we can speak of these as “objective probabilities.”

  3. 3.

    Though there are some regions of the genome more prone to error than others, and some changes to the genome that tend to accelerate the onset of disease more quickly than others (Roberts and Gordenin 2014; Salk et al. 2010; Martincorena and Campbell 2015). For further discussion of the ways in which mutation is random, see, Merlin (2016).

  4. 4.

    I take Tomasetti and Vogelstein to be using the term “stochastic” in the sense that per base pair somatic mutations occur at a regular rate, sometimes called the “background” mutation rate. There are different causes of this outcome, or means by which perfect replication fails to come about. For instance, errors due to random polymerase misincorporation are estimated to be 7.6 × 10−10 (± 3.8 × 10−11) per base per cell division (Tomasetti et al. 2013). As we will see, some readers interpret their use of the term “stochastic” as “random,” in the sense that, given the same initial conditions, all causes of mutation, and all outcomes (e.g., base pair changes, inversions, deletion, chromosomal duplications, etc.) are equiprobable. This seems unlikely, however, as it’s widely known that some causes or error, and some types of mutation, are more common than others. We can predict which elements of the genome are most subject to error, yielding mutations of some types more often than others (Kimsey et al. 2015; Kunkel 2009; Fromme and Verdine 2004; Collins 2005; Gold 2017). Biologists sometimes use the term “stochastic” to refer to processes whose outcomes can be modeled as a random sampling procedures, but Tomasetti and Vogelstein make no such claim (at least explicitly) here.

  5. 5.

    Strictly speaking, “Spearman’s rho = 0.81; P < 3.5 Å ~ 10 − 8) (Fig. 1). Pearson’s linear correlation 0.804 [0.63 to 0.90; 95% confidence interval (CI)] was equivalently significant (P < 5.15 Å ~ 10 − 8).”

  6. 6.

    There is a great deal more to say about IBE, whether it really is two forms of argument or one, whether all IBE begs the question, or instead can be represented warranted inference, such as by appeal to relative likelihood or Bayesian formal models of confirmation. I set these (endless) debate to one side here, but see Plutynski (2011) for a review of the history of these debates over IBE’s form and warrant. See the “Appendix”, however, for two different ways of formally reconstructing the argument. The latter draws on Schupbach’s formal representations of IBE as a relative likelihood argument (2016). Many thanks to Jonah Schupbach for his feedback and suggestions.

  7. 7.

    There is a rich philosophical literature on the variety of senses of “stem cell” in the biological literature, methodological challenges facing experimental work on stem cells, and on the role of cancer stem cells in cancer biology, which deserve attention, but which I cannot review in any depth here. However, I refer the reader to the excellent work of Fagan (2013), and Laplane (2015).

  8. 8.

    In the first of the two formalizations of the argument in the “Appendix”, this background knowledge are premises 1–7.

  9. 9.

    It is this author’s view that there is a middle ground between these competing “paradigms.” (Plutynski 2018, 2020) That is, one can take both tissue organization and mutation to play important roles in cancer causation. An integrative approach that attempts to draw upon multiple perspectives is preferred. See also Malaterre (2007), Marcum (2009) and Bertolaso (2016).

  10. 10.

    Indeed, as Batlle and Clevers (2017) has argued, our biological understanding of “stemness” itself is a moving target. Whether and how cell hierarchies present in tumors are organized, such that they yield higher or lower rates of cancer, is still not well-understood, given that our experimental work on this question is indirect. Variation in tumor niches may differentially affect whether and how often tumor cells acquire the stemness phenotype (see also: Laplane 2015; Laplane and Solary 2019).

  11. 11.

    Regarding whether objective chances (or all talk of probability) can be reduced to propensities or relative frequencies, I do not think it’s necessary for me to take a stance in service of diagnosing the conceptual and methodological confusions at issue in this debate. However, I endorse a broadly pluralist view about probability, akin to that defended by Suárez (2020, 2017), and am a realist about the objectivity of macro-level probabilities. If they can be measured, are robust, and predictive, they are objective, whether they are multiply realized or reducible (see, e.g., Sober 2010 for a discussion of the “reality of macro-probabilities” in evolution, for an analogous case).

  12. 12.

    One could well argue that Doll and Hill’s evidence was more than sufficient to establish that smoking caused lung cancer by 1950, but “sufficiency” in my view depends upon what one wishes to use such causal information for. We can have better and worse evidence, and better or worse reasons to regulate (e.g., saving people’s lives versus making a profit). There is of course a great deal more to say here about the pragmatic and value-laden character of inference in epidemiology and public health (see, e.g., Reiss 2015; Broadbent 2011a, b; Plutynski 2018).

  13. 13.

    As they later acknowledge (2017), etiology is an independent matter from relative effectiveness of primary and/or secondary prevention. That is, whether or not a primary or secondary preventive policy is warranted has far more to do with whether the measures in question are practically effective than whether they ought to be so in principle.

  14. 14.

    In this sense, statistical explanations in biology can be said to be “autonomous” from those at other temporal or spatial scales. As might be expected, then, there is a debate about Woodward’s virtues of stability and proportionality, and whether they indeed favor “higher level” explanations. While considerations of space prohibit addressing these matters at length here, see Shapiro and Sober (2012); Weslake (2013); Franklin-Hall (2016), and Woodward’s reply (2018, forthcoming).

  15. 15.

    Many philosophers argue that the causes that matter are the “ultimate” or “foundational” ones—those that are the concerns of physicists. If indeed the fundamental sciences are deterministic, then nature as a whole is deterministic. On this view, the question of whether cancer is a matter of “chance” or luck is simply ill-conceived. The problem with such arguments is that they trade on matters that no empirical evidence so far can decide. Whatever you make of a priori arguments for causal determinism, the jury is still out for those of a more empiricist bent. Even our best physical theories leave open the question of whether causal determinism is true (Hoefer 2016). Moreover, it’s in principle possible that regularities at the macrolevel—or the domains that concern us with respect to explanation and prediction in biology—are only weakly constrained by micro-scale regularities (see, e.g., Ismael 2016, 2017; Batterman 2011).


  1. Aktipis CA, Boddy AM, Jansen G, Hibner U, Hochberg ME, Maley CC, Wilkinson GS (2015) Cancer across the tree of life: cooperation and cheating in multicellularity. Philos Trans Royal Soc B: Biol Sci 370(1673):20140219

    Article  Google Scholar 

  2. Batlle E, Clevers H (2017) Cancer stem cells revisited. Nat Med 23(10):1124

    Google Scholar 

  3. Batterman RW (2011) The tyranny of scales.

  4. Bechtel W (2018) The importance of constraints and control in biological mechanisms: insights from cancer research. PhilosSci 85(4):573–593

    Google Scholar 

  5. Bertolaso M (2016) Philosophy of cancer. Springer, Dordrecht

    Book  Google Scholar 

  6. Bozic I, Antal T, Ohtsuki H et al (2010) Accumulation of driver and passenger mutations during tumor progression. ProcNatlAcadSci USA 107(43):18545–18550

    Article  Google Scholar 

  7. Brigandt I (2013) Integration in biology: philosophical perspectives on the dynamics of interdisciplinarity. Stud HistPhilosBiol Biomed Sci 44:461–465

    Article  Google Scholar 

  8. Broadbent A (2011a) Inferring causation in epidemiology: mechanisms, black boxes, and contrasts. In: Illari PMK, Williams J (eds) Causality in the sciences. Oxford University Press, Oxford, pp 45–69

    Chapter  Google Scholar 

  9. Broadbent A (2011b) Epidemiological evidence in proof of specific causation. Leg Theory 17(4):237–278

    Article  Google Scholar 

  10. Clevers H (2018) Bordeaux workshop in philosophy of cancer. Pers Commun

  11. Collins AR (2005) Antioxidant intervention as a route to cancer prevention. Euro Jour of Can 41(13):1923–1930

    Article  Google Scholar 

  12. Davy-Smith G, Relton CL, Brennan P (2016) Chance, choice and cause in cancer aetiology: individual and population perspectives. Intl J Epidemiol 45(3):605–613

    Article  Google Scholar 

  13. Doll R, Hill AB (1950) Smoking and carcinoma of the lung. BMJ 2(4682):739

    Article  Google Scholar 

  14. Doll R, Hill AB (1954) The mortality of doctors in relation to their smoking habits. BMJ 1(4877):1451

    Article  Google Scholar 

  15. Duesberg P, Rasnick D (2000) Aneuploidy, the somatic mutation that makes cancer a species of its own. Cell Motil Cytoskelet 47(2):81–107

    Article  Google Scholar 

  16. Fagan M (2013) Philosophy of stem cell biology: knowledge in flesh and blood. Palgrave Macmillan, New York

    Book  Google Scholar 

  17. Franklin-Hall LR (2016) High-level explanation and the interventionist’s “variables problem.” Br J PhilosSci 67(2):553–577

    Article  Google Scholar 

  18. Fromme JC, Verdine GL (2004) Base excision repair. Adv Protein Chem 69:1–41

    Article  Google Scholar 

  19. Gold B (2017) Somatic mutations in cancer: Stochastic versus predictable. Mutat Res Genetic Toxicol Environ Mutagen 187:37–46

    Article  Google Scholar 

  20. Gröbner SN, Worst BC, Weischenfeldt J et al (2018) The landscape of genomic alterations across childhood cancers. Nature 555(7696):321–327

    Article  Google Scholar 

  21. Haldane JBS (1964) A defense of beanbag genetics. PerspectBiol Med 7(3):343–360

    Article  Google Scholar 

  22. Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next generation. Cell 144(5):646–674

    Article  Google Scholar 

  23. Hill AB (1965) The environment and disease: association or causation? Proc R Soc Med 58:295–300

    Google Scholar 

  24. Hochberg ME, Noble RJ (2017) A framework for how environment contributes to cancer risk. EcolLett 20(2):117–134

    Google Scholar 

  25. Hoefer C (2016) Causal determinism. In: Zalta EN (ed) The stanford encyclopedia of philosophy. Accessed Spring 2020

  26. Hu C, Hart SN, Bamlet WR et al (2016) Prevalence of pathogenic mutations in cancer predisposition genes among pancreatic cancer patients. Cancer EpidemiolPrevBiomark 25(1):207–211

    Google Scholar 

  27. Ismael JT (2016) How physics makes us free. Oxford University Press, Oxford

    Book  Google Scholar 

  28. Ismael JT (2017) An Empiricist’s guide to objective modality. In: Slater MH, Yudell Z (eds) Metaphysics and the philosophy of science: new essays. Oxford University Press, Oxford, pp 109–125

    Chapter  Google Scholar 

  29. Kimsey IJ, Petzold K, Sathyamoorthy et al (2015) Visualizing transient Watson–Crick-like mispairs in DNA and RNA duplexes. Nature 519(7543):315

    Article  Google Scholar 

  30. Kimura M (1968) Evolutionary rate at the molecular level. Nature 217(5129):624–626

    Article  Google Scholar 

  31. Kunkel TA (2009) Evolving views of DNA replication (in) fidelity. Cold Spring HarbSymp Quant Biol 74:91–101

    Article  Google Scholar 

  32. Laplane L (2015) Cancer stem cells. Harvard University Press, Cambridge

    Google Scholar 

  33. Laplane L, Solary E (2019) Philosophy of biology: towards a classification of stem cells. Elife 8:e46563

    Article  Google Scholar 

  34. Malaterre C (2007) Organicism and reductionism in cancer research: towards a systemic approach. Int Stud PhilosSci 21(1):57–73

    Google Scholar 

  35. Marcum JA (2009) Cancer: complexity, causation, and systems biology. MedicinaStoria 9(17–18):267–287

    Google Scholar 

  36. Martincorena I, Campbell PJ (2015) Somatic mutation in cancer and normal cells. Science 349(6255):1483–1489

    Article  Google Scholar 

  37. Mayr E (1963) Animal species and evolution. Harvard University Press, Cambridge

    Google Scholar 

  38. Merlin F (2016) Weak randomness at the origin of biological variation: the case of genetic mutations. In: Pence C, Ramsey G (eds) Chance in evolution. University of Chicago Press, Chicago, pp 176–195

    Google Scholar 

  39. Nelson CM, Bissell MJ (2006) Of extracellular matrix, scaffolds, and signaling: tissue architecture regulates development, homeostasis, and cancer. Annu Rev Cell Dev Biol 22:287–309

    Article  Google Scholar 

  40. Noble R, Kaltz O, Nunney L, Hochberg ME (2016) Overestimating the role of environment in cancers. Cancer Prev Res 9(10):773–776

    Article  Google Scholar 

  41. Nowak M, Waclaw B (2017) Genes, environment, and bad luck: explaining cancer risk in the statistical sense. Science 355(6331):1266–1267

    Article  Google Scholar 

  42. Nunney L (2018) Size matters: height, cell number and a person’s risk of cancer. Proc R Soc B 285(1889):20181743

    Article  Google Scholar 

  43. Nunney L, Muir B (2015) Peto’s paradox and the hallmarks of cancer: constructing an evolutionary framework for understanding the incidence of cancer. Philos Trans R Soc B BiolSci 370(1673):20150161

    Article  Google Scholar 

  44. O’Rourke M, Crowley S, Gonnerman C (2016) On the nature of cross-disciplinary integration: a philosophical framework. Stud HistPhilosSci Part C Stud HistPhilosBiol Biomed Sci 56:62–70

    Article  Google Scholar 

  45. Plomin R, Daniels D (1987) Why are children in the same family so different from each other? Behav Brain Sci 10:1–16

    Article  Google Scholar 

  46. Plutynski A (2011) Four problems of abduction: a brief history. HOPOS: J IntSocHistPhilosSci 1(2):227–248

    Google Scholar 

  47. Plutynski A (2018) Explaining cancer: finding order in disorder. Oxford University Press, New York

    Book  Google Scholar 

  48. Plutynski A (2020) Cancer modeling: the advantages and limitations of multiple perspectives. In: Massimi M, McCoy CD (eds) Understanding: Perspectivism Scientific challenges and methodological prospects. Taylor & Francis, New York

    Google Scholar 

  49. Reiss J (2015) A pragmatist theory of evidence. PhilosSci 82(3):341–362

    Google Scholar 

  50. Roberts SA, Gordenin DA (2014) Hypermutation in human cancer genomes: footprints and mechanisms. Nat Rev Cancer 14(12):786–800

    Article  Google Scholar 

  51. Rozhok AI, Wahl GM, DeGregori J (2015) A critical examination of the “bad luck” explanation of cancer risk. Cancer Prev Res 8(9):762–764

    Article  Google Scholar 

  52. Salk JJ, Fox EJ et al (2010) Mutational heterogeneity in human cancers: origin and consequences. Annu Rev Pathol 5:51–75

    Article  Google Scholar 

  53. Schupbach JN (2016) Inference to the best explanation, cleaned up and made respectable. In: McCain K, Poston T (eds) Best explanations: new essays on inference to the best explanation. Oxford University Press, Oxford, pp 39–61

    Google Scholar 

  54. Shapiro LA, Sober E (2012) Against proportionality. Analysis 72(1):89–93

    Article  Google Scholar 

  55. Sober E (2010) Evolutionary theory and the reality of macro-probabilities. In: Eells E, Fetzer JH (eds) The place of probability in science. Springer, Dordrecht, pp 133–161

    Chapter  Google Scholar 

  56. Sornette D, Favre M (2015) Debunking mathematically the logical fallacy that cancer risk is just “bad luck.” EPJ Nonlinear Biomed Phys 3(1):10

    Article  Google Scholar 

  57. Sonnenschein C, Soto AM (2008) Theories of carcinogenesis: an emerging perspective. Semin Cancer Biol 18(5):372–377

    Article  Google Scholar 

  58. Soto AM, Sonnenschein C (2005) Emergentism as a default: cancer as a problem of tissue organization. J Biosci 30(1):103–118

    Article  Google Scholar 

  59. Soto AM, Sonnenschein C (2011) The tissue organization field theory of cancer: a testable replacement for the somatic mutation theory. BioEssays 33(5):332–340

    Article  Google Scholar 

  60. Suárez M (2020) Philosophy of probability and statistical modelling: Cambridge elements in the philosophy of science. Cambridge University Press, Cambridge

    Book  Google Scholar 

  61. Suárez M (2017) Propensities, probabilities, and experimental statistics. In: Massimi M, Romeijn J, Schurz G (eds) EPSA15: selected papers: European studies in philosophy of science, vol 5, pp 335–45

  62. Tomasetti C, Vogelstein B, Parmigiani G (2013) Half or more of the somatic mutations in cancers of self-renewing tissues originate prior to tumor initiation. ProcNatlAcadSci 110(6):1999–2004

    Article  Google Scholar 

  63. Tomasetti C, Vogelstein B (2015) Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science 347:78–81

    Article  Google Scholar 

  64. Tomasetti C, Li L, Vogelstein B (2017) Stem cell divisions, somatic mutations, cancer etiology, and cancer prevention. Science 355:1330–1334

    Article  Google Scholar 

  65. Various authors. 2015 Letters on cancer risk. Science. 347(6223), 728–729. (Ashford, Nicholas A, Bauman, P, Brown, HS, et al Cancer risk: role of environment; Potter, JD, Prentice, RL Cancer risk: tumors excluded; Wild, C, Brennan, P, Plummer, M, et al Cancer risk: role of chance overstated; Song, M, Giovannucci, EL Cancer risk: many factors contribute; Gotay, C, Dummer, T, Spinelli, J Cancer risk: prevention is crucial; O'Callaghan, M Cancer risk: accuracy of literature.

  66. Weinberg CR, Zaykin D (2015) Is bad luck the main cause of cancer? JNCI: J Natl Cancer Inst 107(7)

  67. Weslake B (2013) Proportionality, contrast and explanation. Australas J Philos 91(4):785–797

    Article  Google Scholar 

  68. Woodward J (2010) Causation in biology: stability, specificity, and the choice of levels of explanation. BiolPhilos 25:287–318

    Google Scholar 

  69. Woodward J (2018) Explanatory autonomy: the role of proportionality, stability, and conditional irrelevance. Synthese 1–29

  70. Woodward J (forthcoming) Causation with a human face: normative theory and descriptive psychology

  71. Wu S, Powers W, Zhu Y, Hannun (2016) Substantial contributions of extrinsic risk factors to cancer development. Nature 529:43

    Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Anya Plutynski.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Below is the link to the electronic supplementary material.

Supplementary material 1 (DOCX 15 kb)

Appendix: formalizing Tomasetti and Vogelstein’s IBE

Appendix: formalizing Tomasetti and Vogelstein’s IBE

First version

  • Premise 1. Genomic changes occur simply by chance during DNA replication.

  • Premise 2. The endogenous (somatic) per base pair mutation rate of all human cell types appears to be nearly identical.

  • Premise 3. There are a large number of somatic mutations known to exist in cancer cells.

  • Premise 4. Among such mutations are mutations to genes that make a causal contribution to the “hallmark” behaviors of cancer cells (e.g., oncogenes and tumor suppressor genes. The mechanisms by which these genes affect regulatory pathways in the cell, e.g., detection of errors in replication, or halting and initiating the cell cycle (cell birth and death) are in many cases well-understood (Hanahan and Weinberg 2011).)

  • Premise 5. Cancer incidence by and large increases as we age.

  • Conclusion 1. Somatic (i.e., acquired, rather than inherited) mutations are causally responsible for the initiation of a tumor. (from 1 to 5).

  • Premise 6. Stem cells—those cells in a tissue or organ that can self-renew and are responsible for the development and maintenance of the tissue's architecture—have the capacity to initiate a tumor, if and when they acquire sufficient number of mutations.

  • Premise 7. Stem cells make up a small proportion of the total number of cells in a tissue; and each tissue type has a different number and division pattern.

  • Premise 8: The more often a cell turns over, the more mutations they are likely to acquire (by premises 1–2).

  • Conclusion 2. There should be a strong, quantitative correlation between the lifetime number of divisions among a particular class of cells within each organ (stem cells) and the lifetime risk of cancer arising in that organ. (from 1 to 8).

  • Premise 9. There is a 65% correlation between rates of stem cell division in a given tissue type and lifetime cancer risk in that tissue type.

  • Conclusion 3. The differences in lifetimes risk across tissue types are caused primarily by “luck”—stochastic acquisition of mutations in stem cells.

Second version (drawing upon Schupbach 2016):

… the hypothesis that offers the most powerful potential explanation of some proposition will be the one that makes that proposition the most likely. In Bayesian terms, the hypothesis judged to provide the best explanation will have the greatest corresponding likelihood of any explanatory hypothesis considered. This result clarifies the nature of the reason that favors the most explanatory hypothesis over those that are explanatorily inferior. A hypothesis’s likelihood … is positively related to its overall probability in light of the evidence…

So, in this case the argument above may be separated into two parts: a rehearsal of the relevant background knowledge, and an assessment of the relative likelihood. The background knowledge in this case is: rates of turnover of stem cells in all tissue vary across tissues/organs, rates of mutation are constant, and cancer is (in large part) a product of the acquisition of a series of mutations over our lifetimes, there’s a 65% correlation between rate of turnover of stem cells and average cancer incidence in any given tissue.

So, here’s the observation (e): there are orders of magnitude of difference in cancer incidence across different tissues and organs.

What explains this?

  • H1: rate of turnover of stem cells in varies across tissues/organs.

  • H2, H3, etc.: exogenous factors have differential effect across tissues/organs

  • Background b: rates of mutation are constant, and cancer is (in large part) a product of the acquisition of a series of mutations over our lifetimes; also perhaps: higher turnover rates increase the chances of getting the cancer-causing series of mutations (is that right?), etc.

Argument, v2

  • E: There are orders of magnitude difference between cancer incidence across different tissues/organs (e.g., bone cancer and brain cancer are very rare, whereas skin cancer is relatively common, etc.)

  • Out of the available potential explanations of E, H1 is the best in the sense of being the most powerful AND being the best confirmed by background evidence; formally, H1 has a much higher degree of explanatory power over E than any of the other H's, which coincides with the probabilistic claim that Pr(E|H1&b) >  > Pr(E|Hi&b) for all i ≠ 1. AND H1 is more probable in light of b than any of the other H's: Pr(H1|b) > Pr(Hi|b) for all i ≠ 1.

  • Therefore, H1.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Plutynski, A. Is cancer a matter of luck?. Biol Philos 36, 3 (2021).

Download citation


  • Cancer
  • Explanation
  • Causation
  • Probability
  • Chance