The importance of extrinsic factors in the development of cancers

ABSTRACT

There was recently a debate on the relative importance of intrinsic versus extrinsic factors in cancer risk. We reported that extrinsic risk factors contribute substantially (more than 70–90%) to most common cancers. This finding promotes further research into the causes of cancer and how they could be prevented.

In January 2015, a research paper in Science reported a significant correlation between lifetime risk of 31 types of cancer and the total stem cell divisions in their corresponding tissues,¹ implying that approximately two-thirds of cancer risk variation can be explained by the number of stem cell divisions. Based on this finding, the authors of that study further concluded that a large percentage of the incidences of many human cancers are caused by intrinsic factors of random stem cell mutations rather than extrinsic factors such as environmental exposures, leading to the so-called “bad luck” hypothesis of cancer development. Following the original report, many inspiring discussions have argued against this hypothesis from various points of view^2,3; however, none directly tackled the initial analysis or offered specific alternatives for quantifying the contribution of extrinsic risk factors in cancer development.

In our study,⁴ we started with a thought experiment in which known extrinsic factors are hypothesized to boost the risks of all cancers by 4-fold through the induction of mutations in stem cells. It is clear that in this hypothetical setting, both intrinsic and extrinsic factors contribute to cancer risk, among which the extrinsic factors account for at least 80%. However, if we conduct regression analyses in this thought experiment and follow the logic presented in the Science report,¹ two-thirds of cancer risk variation would still be attributed to intrinsic factors. Clearly, this contradiction negates the effect of attributing the results on the correlation of the variation in risk with stem cell division solely to intrinsic factors.

Next, we re-evaluated the same data using a distinct approach, based on the assumption that cancers with the same number of stem cell divisions should share the same base of intrinsic cancer risk if the relationship between total stem cell division and cancer risk is causal. The implicit assumption here is that various human stem cells would not exhibit 5- to 1,000-fold differences in their intrinsic mutation rates. The results showed that, for many cancers, less than 10% of their risk was attributable to random errors in cell duplication. To further ensure that our results are independent of the specific theory of stem cell origin of cancer, we collected another set of data concerning the total number of tissue cell divisions calculated from homeostatic tissue cell numbers and their turnover rates for several tissues.⁵ The analysis of cancer risk versus total tissue cell divisions reached the same conclusion as that from the total stem cell data. The inference from the new data demonstrates that, irrespective of whether a subpopulation or all dividing cells contribute to cancer, intrinsic factors do not play a major causal role in cancer development. A subsidiary conclusion is that the correlation between stem cell division and cancer risk cannot be taken to causally implicate stem cells since the same relationship was seen with total cell division.

These results are supported by strong epidemiologic evidence; for example, global cancer statistics reveal that many cancers exhibit large geographic variations,⁶ and immigrants moving from countries with lower cancer incidence to countries with higher incidence acquire the higher risk in their new country.⁷ We further analyzed the Surveillance, Epidemiologic and End Results Program (SEER) data⁸ and showed that many cancers have been increasing consistently in incidence and/or mortality over 20 y or more, suggesting that changing environments and external factors such as smoking and air pollutants contribute heavily to these cancers.

We also examined recent studies on mutational signatures in cancer, which are regarded as “fingerprints” left on cancer genomes by different mutagenic processes.⁹ Among approximately 30 distinct signatures, 2 were considered to be mutations resulting from intrinsic processes because of their strong positive correlations with age in the majority of cancers, whereas the other signature mutations were considered to result from extrinsic factors due to a lack of consistent correlation with age.⁹ We dichotomized these signatures according to their intrinsic or extrinsic origins and found that, although a few forms of cancer had >50% intrinsic mutations, the majority of cancers, such as melanoma, lung, bladder, and colorectal cancers, had a large proportions of mutations likely caused by extrinsic factors.⁴

Lastly, we used computational modeling to dissect the contribution of intrinsic processes in the development of cancer based on known gene mutations in cancer and the likelihood that they arise from intrinsic mutation rates.¹⁰ We found that, when 3 or more mutations are required for cancer onset, intrinsic processes are far from sufficient to account for the observed risks, again indicating the small contribution of intrinsic cancer risks in many cancers.

Although these 4 distinct approaches used for extrinsic cancer risk assessment are based on almost orthogonal data, i.e. cancer biology, epidemiology, cancer genomics, and mathematical modeling, they all reached the same broad conclusion that most cancers are attributed largely to external risk factors, with only 10–30% attributed to random mutations, or intrinsic factors. Also, the 4 approaches involved both data- and model-driven quantitative analyses, with and without using stem cell estimations, suggesting that our conclusion is quite robust.

Overall, our research showed that “bad luck”, i.e., random mutations that naturally occur when cells divide, is not a sufficient explanation for cancer development and that extrinsic factors have a predominant influence (Fig. 1). Prevention is therefore key, and individuals, government agencies, and private donors should make this a priority. Our study also provides a solid foundation for the argument that cancer investigators need to double their efforts on discerning external factors, understanding the pathogenesis of cancer, and effective prevention. Our main message is to promote further research into the causes of cancer and how they could be prevented.

Figure 1.

Contribution of intrinsic and extrinsic factors to cancer risks. Here, intrinsic risk is defined as the internal random mutation rate in dividing cells (normal tissue stem cells or other proliferating cells). Extrinsic factors may affect cancer risk through cell divisions (by inducing additional mutations or affecting proliferating rate) and/or other biological functions. *HPV: human papillomavirus.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.