The chances are that, if you follow news articles about cancer, you’ll have come across headlines like “Most Cancers Caused By Bad Luck” (The Daily Beast) or “Two-thirds of cancers are due to “bad luck,” study finds” (CBS News). The story – based on research out of Johns Hopkins University – has grabbed widespread media attention. But it’s also raised the ire of science communicators who think that the headlines and stories are, in the words of a couple of writers, “just bollocks”.
With all the coverage of the paper, and the subsequent coverage of the coverage, I was interested in just how off-base the news articles were, and to what extent this was down to lazy reporting.
The paper in question is “Variation in cancer risk among tissues can be explained by the number of stem cell divisions” by Cristian Tomasetti and Bert Vogelstein, published this month in the journal Science. At the heart of the paper the authors look at how stem cell divisions in different tissues correlate with lifetime risk of developing cancer in those tissues. The study shows a clear correlation with the cancer types considered – the faster the stem cells divide in a particular tissue, the greater the chance of developing cancer in that tissue.
The two researchers then tease out the degree that they think random genetic mutations, as opposed to environmental and lifestyle factors, influence cancer risk. They conclude that, out of 31 cancer types considered, 22 were primarily associated with random genetic mutations (they called these “R-tumors” – the R standing for “random”), and nine were associated with environmental factors on top of these random mutations (deterministic tumors, or “D-tumors”).
In the author’s words,
“We refer to tumors with relatively low ERS [“extra risk score”] as R-tumors (R for replicative; green cluster in Fig. 2) because stochastic factors, presumably related to errors during DNA replication, most strongly appear to affect their risk.”
In other words, out of the 31 cancer types studied, the authors’ analysis showed that 70% of them – just over two thirds – were predominantly determined by random mutations and not environmental factors; what the authors term in the paper as “bad luck”.
The inference that many cancers – and even cancer types – cannot easily be prevented by reducing environmental exposures or changing lifestyles, proved to be a media-magnet. Headlines resulted along the lines of
“Cancer Is More Bad Luck Than Bad Behavior, Study Says” (Bloomberg)
“Two-Thirds of Cancer Cases Are Simply Down to Bad Luck” (Gizmodo)
“Two-thirds of adult cancers largely ‘down to bad luck’ rather than genes” (The Guardian)
“Most cancer types ‘just bad luck'” (BBC News)
“Most cancer cases ‘due to bad luck'” (Daily Mail)
And some commentators weren’t amused.
Michael Head for instance tweeted
No, media, two-thirds of #cancers are not ‘due to bad luck’. Crap reporting. Again. http://t.co/cYjcgoAqZv
— Michael Head (@michaelghead) January 2, 2015
In response to many of the headlines and articles, Adam Jacobs (linked to in the tweet above) wrote on his blog The Stats Guy …
A paper published in Science has been widely reported in the media today. According to media reports, such as this one, the paper showed that two thirds of cancers are simply due to bad luck, and only one third are due to environmental, lifestyle, or genetic risk factors.
The paper shows no such thing, of course.
… concluding with
We know that lifestyle is hugely important not only for cancer, but for many other diseases as well. For the media to claim that lifestyle isn’t important, based on a misunderstanding of what the research shows, is highly irresponsible.
Over at The Guardian, the media-questioning was taken up by Bob O’Hara and GrrlScientist under the headline “Bad luck, bad journalism and cancer rates”. Not pulling their punches, they wrote:
The big science/health news story this week is about cancer rates, with news outlets splashing headlines like “Two-thirds of adult cancers largely ‘down to bad luck’ rather than genes” (for example, here) or “Most cancer types ‘just bad luck’” (here). (I’m not even going to look to see what the Daily Mail has to say about this.) But these headlines, and the stories, are just bollocks. The work, which is very interesting, showed no such thing.
At this point my curiosity was piqued (egged on my science bloggers like Ed Yong who similarly questioned the media coverage). Was this just a particularly egregious case of widespread lazy journalism, or did the stories have a common root?
Reading the original paper, the authors were clearly building a case for the majority of the cancers they studied having predominantly random origins. This is particularly clear in figure 2 in the paper (see below) where they cluster cancers into random versus deterministic types. But the language is still somewhat cautious in the paper.
Figure 2 from Tomasetti and Vogelstein (2015). Cancer types are clustered by those where stochastic (replicative) factors dominate (green), versus those where environmental and inherited factors are substantial (blue). ERS – the “adjusted risk score” – is the product of the lifetime risk and the total number of stem cell divisions (log10 values). From the paper: “The adjusted ERS (aERS) is indicated next to the name of each cancer type. R-tumors (green) have negative aERS and appear to be mainly due to stochastic effects associated with DNA replication of the tissues’ stem cells, whereas D-tumors (blue) have positive aERS. Importantly, although the aERS was calculated without any knowledge of the influence of environmental or inherited factors, tumors with high aERS proved to be precisely those known to be associated with these factors.”
The associated press release from Johns Hopkins University is more direct. Under the headline “Bad Luck of Random Mutations Plays Predominant Role in Cancer, Study Shows”, the press release states
“By [the authors’] measure, two-thirds of adult cancer incidence across tissues can be explained primarily by “bad luck,”
At this point, the press release is referring to the role that random events play in determining whether a cancer will develop. As the release clarifies,
Using statistical theory, the pair calculated how much of the variation in cancer risk can be explained by the number of stem cell divisions, which is 0.804 squared, or, in percentage form, approximately 65 percent.
In other words, they conclude that random genetic mutation “bad luck” as stem cells divide is an important factor underlying the numbers of cancer cases observed – and as a result the lifetime risk of developing cancer.
The release goes on to note:
Finally, the research duo classified the types of cancers they studied into two groups. They statistically calculated which cancer types had an incidence predicted by the number of stem cell divisions and which had higher incidence. They found that 22 cancer types could be largely explained by the “bad luck” factor of random DNA mutations during cell division. The other nine cancer types had incidences higher than predicted by “bad luck” and were presumably due to a combination of bad luck plus environmental or inherited factors.
This directly mirrors the findings presented in the paper; that of the cancers studied, 70% were largely explainable by random mutations during cell division.
Comparing this to the headlines above, the media articles, release and paper align surprisingly well. “Bad luck” is the authors’ phrase, and they do emphasize the dominance of random genetic events in the majority of cancers, and cancer cases.
In this respect, it’s hard to be too tough on on the media coverage – sure, some of the stats may have got a little twisted, but the dominant message seems to have its roots in the paper and the institutional (and author-sanctioned) press release.
So is there a problem here, or have the media actually done good, contrary to perceptions from some quarters?
From my reading of the paper, the press release and the media coverage, this isn’t as straight forward as it might seem. Certainly, it seems that many reporters made an honest effort to faithfully represent what the authors were saying. And yet, science reporting is more than just reporting the facts – it’s also contextualizing those facts in a way that is useful to readers and society more generally.
Going back to Adam Jacobs’ piece, it’s worth repeating his conclusion:
We know that lifestyle is hugely important not only for cancer, but for many other diseases as well. For the media to claim that lifestyle isn’t important, based on a misunderstanding of what the research shows, is highly irresponsible.
If you take the stance – as he does – that environmental and lifestyle factors are critical to determining good and bad health (and as a public health professor, its a stance I am professionally expected to take), news articles that imply we don’t need to worry so much about the pollution we emit, the chemicals we expose people to or the way we live our lives, can be seen as highly irresponsible unless backed up by rock solid evidence. They open the door to an abdication of responsibility when it comes to environmental health. Why spend a fortune on preventing environmental emissions when they don’t matter? Why undergo cripplingly expensive product safety testing if ingredients don’t really cause cancer? Why support inconvenient regulatory agencies if all they do is cripple commerce without preventing cancer and other diseases?
This is a valid fear, backed up by a long history of environmental health disasters. And it’s a fear that requires researchers and research institutions to take at least some responsibility for how they pitch and promote their work.
In the case of this paper, it’s hard to see clear evidence of bad reporting. There is a lack of balance and contextualization though that, it seems, has its roots in the original paper.
This is not a criticism of the paper. But it’s very easy for the significance of research that begins to challenge the status quo to be inappropriately amplified in the media. As I noted in a recent article in Nature nanotechnology,
“when surprising new insights emerge on possible material health risks, where does the responsibility lie for ensuring that new research is conducted on material safety, without this research influencing consumers and regulators before there is plausible justification for action? Or to put it more succinctly, how can we encourage exploratory risk research without it prematurely impacting consumer and regulatory decisions?”
This refers to research on engineered nanomaterials, but the point is just as relevant here: it’s extremely easy for exploratory research to take on the aura of authoritative, actionable knowledge through the lens of the media.
So where does responsibility to temper such amplification lie? Clearly there needs to be responsible reporting at every point in the communication chain. But by the very nature of amplification, care is needed at the source of a story to help ensure that the final reporting is both accurate and responsible (an issue I look at more closely here)
In this case, it was perhaps inevitable that research indicating environmental factors may not be as important as previously thought in causing cancer would lead to “just bad luck” headlines. But those headlines draw explicitly on the language used in the paper and the press release.
Would the media coverage have been different if the work was pitched differently? It’s hard to tell – but in this instance I’d certainly be hesitant to put all the blame on bad journalism.
Paper: Variation in cancer risk among tissues can be explained by the number of stem cell divisions (2015)
Cristian Tomasetti and Bert Vogelstein. Science, Vol. 347 no. 6217 pp. 78-81 DOI: 10.1126/science.1260825
Updated January 4 to include Figure 2 from Tomasetti and Vogelstein (2015)
Thanks, Andrew. The report I saw in the Daily beast this morning said the authors had a large caveat for smoking and its link to lung cancer.
So, recognizing this is something of a “one off,” the authors seem to say that while environmental factors have an affect, these may be only one-third of the risk?
Yes – they clearly say that in some cases environmental factors are important – these are the cancers where their “bad luck” predictions don’t hold.
I fear that the internet attack dogs went for the wrong victim in this case. The paper adds weight to similar estimates for the contribution of chance that have been around for years, but which the authors of all the attacks seem to be unaware. Please read George Davey Smith’s excellent John Snow lecture for a good summary.
I’m baffled by the indignation engendered by suggestion that chance plays a big part in your fate. Life is stochastic, to quite a large extent. On twitter, Alice Roberts made an interesting analogy.
“Prof Alice Roberts @DrAliceRoberts
@david_colquhoun I’m struck by similarity in resistance to accepting role of chance in our individual lives & health – and in evolution”
It’s odd that the sceptics, in this case, are behaving a bit like creationists, or those who believe that it’s your own fault if you get ill.
Thanks for the comments David – am including the link to George Davey Smith’s paper: http://www.dcscience.net/Davey-Smith-2011.pdf
Beyond questions of blame here, there are two deeply-rooted narratives that have been touched in this dialogue:
1. Bad companies, bad people and bad actions cause cancer; and
2. The media cynically sensationalize and misreport science
I suspect that, because of this, the coverage has raised ire because it seems to challenge #1 and seems to support #2. And what we get as a result is a discussion about dogmas, not data.
“The media cynically sensationalize and misreport science”
While not trying to exonerate the Daily Mail from misreporting science, I fear that the truth is worse than that. In many cases, it is the press release from the journal, or from the university PR department that sensationalises the science (and since the authors will normally approve these releases, they must accept some of the blame). I have given several examples e.g. at
http://www.dcscience.net/2014/11/02/two-more-cases-of-hype-in-glamour-journals-magnets-cocoa-and-memory/
In this particular case, though, I’m on the other side. I was astonished when Adam Jacobs made the assertion “We know that lifestyle is hugely important not only for cancer” because that is precisely what we don’t know (and I was pleased to get the support of the oncologist and skeptic, David Gorski, on that). In order to justify this claim, he chose one of the papers that I’d previously singled out as being one of the most ghastly hyped diet papers I’d encountered. See the discussion at
http://www.statsguy.co.uk/are-two-thirds-of-cancers-really-due-to-bad-luck/
Thank you David Colquhoun for commenting from the saner side of this feeding frenzy.
The first attack on the media that I saw came from someone who did not even bother to tell their readers that much of the hype and overstatement they complained of in the media coverage of this paper was in the press release that heralded the publication in the journal Science, itself a powerful PR machine. As you said, it is highly unlikely that this press release got out without researcher clearance.
Even the abstract in Science contained some of the crimes against humanity that so upset the rabid hordes. No, the evil scribblers did not suddenly conjure up the “bad luck” bit.
It is interesting that many of the scientific experts who weighed in failed to do what they demand of journalists, dig a bit deeper and find the evidence to support your story. I wonder how many of the critics did what I did when a first saw their complaints, which was to rush off to the source of the story to see what it had said. That immediately told me that many flaws in the reporting owed more to the sources than to the journalists.
But why bother to dilute your bile with facts when it is much more convenient to hammer away with the same tired old “media doesn’t get science” line? Why not behave just like the rightly loathed and, as someone else has said, possibly carcinogenic, Daily Mail, and write something that fits your own agenda rather than the facts?
I can’t be bothered to plough through all the tosh out there on this one, so I have not found out if there are any comments about the peer review of the paper, in particular, of the statistical analysis. Given “dodgy statistics” is up there with plagiarism and cooked up data when it comes to retracted papers – I just made up that statistic – it is a bit rich to criticise journalists, as some of the comments have, for not being experts in statistics.
There are many stories out there where journalists do get things wrong. By constantly gunning for stories that owe much to the tenor of the original material, the twitterloonies fall into the “crying wolf” camp.
Now, had they criticised the media for churnalism, parroting garbage fed to them by a PR machine, I might have joined in the fun and games.
I won’t be able to say this quite right, statistically, but this randomness is then (part of) the reason why cancers show up later in life. Yes? More “throws of the dice, ” so to speak.
This would make sense if the probability of genetic mutations correlates with cell divisions – the more division cycles, the greater the cumulative chance of a harmful mutation occurring
what I find interesting is how people respond to cancer news, as though that is the ONLY bad news that occurs in healthcare? Many people, including many many healthcare providers consider cancer to just be the worst, but really most chronic degenerative diseases are pretty awful…and most of them appear to be due to random luck as well…I suppose it comes down to we will all die of something.
but no one wants to believe that.
Hi Andrew,
I’m one of the reporters who wrote about the story. Thank you for an evenhanded look at the study, the press release, and media reports. One observation I’d add is that the impending New Year’s holiday probably made it hard to get independent evaluations. (It certainly did in my case, although I eventually succeeded).
Adam Jacobs made a misleading statement about the random mutation risk hypothesis as presented in the study:
“The problem is that it applies only to explaining the variation in cancer risk from one tissue to another. It tells us nothing about how much of the risk within a given tissue is due to modifiable factors. You could potentially see exactly the same results whether each specific type of cancer struck completely at random or whether each specific type were hugely influenced by environmental risk factors.”
But the authors addressed this point, through the ERS method you quoted above. Maybe the ERS method is flawed, but Jacobs’ blog post doesn’t even acknowledge its existence, let alone attempt to refute it.
The media reports usually stressed that even a one-third risk from environmental factors is still significant. So Jacobs’ closing line: “For the media to claim that lifestyle isn’t important, based on a misunderstanding of what the research shows, is highly irresponsible,” is simply false. Even Jacobs’ link to the Independent article on the study belies that statement.
Best,
Bradley
Thanks to you, and to Michael Kenward for defusing some of the hysteria.
I haven’t seen your report, but it sounds from your comment that you’ve delved deeper than many of the critics.
I’m quite baffled about why there should be such a strong reaction against the idea that chance plays a substantial role in your fate. That, after all, is how evolution works. And the idea was formulated quite clearly by none other than Richard Peto in 1977.
I think the reaction stems from a fear that the public will behave irresponsibly if told chance plays a predominant role in cancer. While that may or may not be true, its a separate issue than the study’s scientific validity. Just because a scientist personally dislikes how a study may be interpreted is not an argument against its accuracy.
The study itself includes statements like “Thus, the stochastic effects of DNA replication appear to be the major contributor to cancer in humans.” That doesn’t say 2/3 of all cancers come from random mutation, of course. I’d like to see that question specifically tackled, using the study data, to get the proportion more precisely quantified. While the press release did give the 2/3 number as applying to all cancers, I recognize that’s no excuse.
PZ Myers gave a thoughtful look at how the study can be used to improve cancer prevention and care: j.mp/pzmyerscancer
Thanks Bradley – from my experience it’s hard to get timely academic input/comment at the best of times; can’t imagine the additional challenges of trying over a holiday period!
Indeed. Not related to this particular saga, I have known researchers to put out a press release and then to disappear for a long vacation without leaving contact details. Do that and you have no grounds to complain about sloppy coverage.
I noticed that one of the more thorough “journalistic” articles on the paper appeared in Science itself. I assume that the writer had earlier access to the paper than lesser mortals.
Makes me wonder if they ever evaluated stochastic rates of cell mutation as the response variable and the environmental variables as the covariates in their regression. In other words, what percent of this labelled “bad luck” is explained by environmental variables? Are these seemingly random mutations perhaps dependent, to a high extent, on environmental variables?
I was wondering much the same thing. Do they determine ‘randomness’ vs environmental factors through cellular isolation from exposures to external radicals and so on? How does this work?
@Bradley J. Fikes
Thanks very much for drawing my attention to PZ Myers blog on this topic.
http://freethoughtblogs.com/pharyngula/2015/01/03/cancer-bad-genes-or-bad-luck/
It’s the best piece I’ve read on the topic.
Isn’t one of the underlying problems in the interpretation of the results of Vogelstein and Tomasetti’s work the very definition of cancer? For a medical doctor, cancer means metastatic disease (or at least disease that has a high probability of becoming metastatic). This is a clinical, reductionist definition. But to address the question of the origins of cancer, one needs to broaden this definition to biology. And this is a completely different story, much more fuzzy, much more complex and usually a very long one. Just think about in situ tumors, all the undiagnosed cancers, spontaneous remissions, etc.
Vogelstein and Tomasetti only looked at clinically diagnosed neoplasms, the visible tip of the “plasms” iceberg. But the causes of cancer are buried in the huge invisible part. And there is one thing we know for sure now: the predominant protective role played by the immune system ( the spectacular results obtained by non-targeted immunotherapies illustrate this well). We also know that the state of the immune system depends strongly on environmental and behavioral factors.
If so, isn’t the status of the immune system THE MAJOR link between environment and behavior and the risk of cancer?
Vogelstein and Tomasetti’s work did not look at any of this. This is OK, I guess because the scope of the paper is quite limited. But the message relayed by the media is wrong and dangerous since it devalues prevention and health promotion.
It’s like saying that the more times an airplane takes off and lands, the higher the risk of crashing. And to remain alive we need to fly, Of course it’s true. But it’s a very partial view, because what prevents airplane crashes are ALSO all the other smaller accidents that build up to air disasters.Vogelstein and Tomasetti’s work only looked at crash statistics.
Thanks for reading my long comment.
One shall not ignore about the influence of grants’ providers and lack of data validation within the cash-strapped research communities. We should not blame the media for doing what a university like J.H. should not have done, i.e., throwaway the principles of further scientific discussion for the sake of PR. As it has happened in many cases in the past, the same PR may come back to them as eggs on the face of their reputation.
Cancer has many faces, and even as today, no one can say simply because it happens in certain parts of body that is necessary the same thing – further away from even being a “type.”
Are the authors sure the statistical/mathematical models they have used is the final verdict within the scientific communities -both alive and in the future [if yes, why they keep printing new texts with no end to them?]… Or, are the cell-biologists job is done by simplifying all kinds of cell divisions under a single biological system?
One thing a true scientist must know is in any analysis, the null hypothesis provides no guaranty that her or his results are right! And that is even if the validity of data are kosher.
And by the way, life itself is a random phenomena within a random medium, influenced by random ambient that is constantly infused by random environmental factors coming from random directions. It is the height of stupidity to try explain “random” by a definite number.
Thak you for the comments on the paper and the resulting media headlines. But even if 2/3 of the examined 31 cancert ypes seem to depend on random (genetic) effects a statement such as “Two-Thirds of Cancer Cases Are Simply Down to Bad Luck” is simply wrong and clearly misleading. The incidence of each cancer has to bee taken in consideration (breast-, colon-, prostate cancer is be far more frequent than duodenum cancer).
What do the authors say about the role of the immune system in influencing whether a cancerous mutation, once it has occurred, develops into a life-threatening tumor? After all, one of the functions of the immune system is to destroy cancerous cells before they grow into large tumors. And much research has shown that environment, lifestyle, and genetics have a big effect on the efficacy of the immune system. So while it might be the case that many cancer cells are created by chance, surely health behaviors, the environment, and genetics still have an important role, mediated through the immune system, in determining how deadly those cancers become. If that’s right, then the headline “2/3 of cancer’s are random” should not be interpreted as “2/3 of life threatening cancers are random.” Rather the headline should be “2/3 of cancerous mutations develop by chance.”
We’ll said Steve, with 60 trillion cells there are always cells not forming correctly. The progression from a transformed cell to a full blown tumour cell is not instant. A genetic predisposition to a cancer can be viewed as just starting further along the line of transformation. The immune system includes transformed cells self destructing or being destroyed by neighbouring cells. This signalling is important to understand and is influenced by environmental factors. One important example is the oldest hormone system of the body called Eicosanoids, which is the signalling gateway to the immune system. Harvard medical school quote ” Eicosanoids may represent a missing link between inflammation and cancer and thus could serve as therapeutic target(s) for inhibiting tumor growth.” One form of Eicosanoid is called Resolvins and these can only be made from 20 carbon (long chain) omega 3 and these end the inflammatory cycle. This mechanism is adversely impacted by excessive omega 6, trans fats, and high insulin levels which can be controlled.
I don’t get how the author is trying to say that the press misrepresented the authors? They CLEARLY stated that “bad luck” is a large factor. So how is reporting this back such a crime? It’s what they said themselves!
Furthermore, what the author of this article misses is that the conclusion of this study is a complete joke. It is pure scientific laziness to say that two-thirds of cancer are caused by “bad luck.” Why don’t we tack on “the wrath of God” while were at it? Basically, the scientists have hit a wall where they can no longer explain something. Yet unlike every legitimate scientist in history — where they simply admit they don’t know and then continue researching — these people instead say that they do have the answer — and it’s due to “bad luck.”
So what exactly is the scientific definition of “bad luck”?
And how on Earth is “bad luck” now considered to be a legitimate, measurable scientific influence? Did they ever consider that perhaps emotional well-being may provide key links?
This is so beyond absurd you honestly couldn’t make it up. Wake up people. You’re being duped.
I’m afraid that you have no comprehension at all of random processes (in this case random errors in DNA replication). At the level of single molecules, everything is random. I think perhaps you should read up about stochastic processes before getting so indignant.
Nothing is random in this universe. Everything is based on cause and effect, whether we happen to understand it yet or not. Electrons, for example, act as particles when observed with a camera. However, when no equipment observes the electrons, they act as waves and particles simultaneously. So even simple observation changes things. But who would know this if it hadn’t been determined through science? If it hadn’t been, someone would come along and call it random luck, simply because they don’t know. It is one thing to say that you don’t understand a cause to an effect, and quite another to say that you do know, and that is because of “luck.” I’m sorry, but that’s just absurd. That is NOT science. Factoring in “luck,” something that has no scientific definition whatsoever, is absolutely irresponsible at best, and fraudulent at worst. Everything is based on cause and effect, whether we happen to understand it yet or not.
Not a good example. Every individual electron moves randomly. I suggest some reading about Brownian motion, or about statistical mechanics.
Things look smooth only when averaged over large numbers of particles.
David,
It depends what meaning is given to “random.” If you mean, not predictable, then sure, many physical processes are deeply random. But that kind of randomness is an epistemic matter, rather than an ontological one. In that sense randomness is a contingent fact about our knowledge (and, beyond that, our cognitive limitations), rather than a fact about nature in general. But if by random you mean “not caused,” then I think the discussion moves onto a more philosophical terrain– God doesn’t play dice, that kind of thing– in which we would have to get some account of what we mean by causation. I suspect the concept of randomness that Mr. Hammond is objecting to is the ontological one rather than the epistemic one. But I also suspect that what the authors mean by random is not “not caused” but “not predictable in a way that could lead to meaningful intervention.”
Mars