Few papers have stimulated bloggery like the one on the randomness of cancer by Bert Vogelstein and Cristian Tomasetti, which Science published January 2 (paywall). That doubtless has something to do with the fact of Vogelstein. It’s hard to overstate his stature in cancer research; he’s known for his work on tumor suppressor genes and much else. When Vogelstein speaks, people listen.
Bloggers bashed journalists, the press release, the paper itself, particular statistical approaches to cancer, and minimizing the role of lifestyle in the disease(s). Turns out, too, that people disagree about the meaning of the word “luck.”
It’s not too much to say, really, that at bottom this was an epistemological debate about the nature of the universe. Before I wade into the heavy stuff, though, the paper itself:
No, it did not say that two out of three cancers are caused by bad luck, as most headline writers and many journalists reported. Whatever “bad luck” means.
To my mind, inheriting faulty DNA-repair genes is bad luck, and so is being a smoker long before smoking’s dangers were known. But the resulting cancers would be classified today as, respectively, genetic and environmental rather than random or sporadic or a matter of chance or bad luck. Among the blogging discussions of this paper the terms usually mean one of two somewhat different things: “bad luck” designates either a cancer due to random (postconception) mutation or a cancer that is not preventable.
What the authors of the paper meant by “bad luck”–the term appeared in the paper’s abstract and the scare quotes are theirs–is “random mutations arising during DNA replication in normal, noncancerous stem cells.” That’s not your everyday definition of bad luck.
Some tissues are much more prone to cancer than others, and the researchers’ idea is that these cancer differences can be explained by the number of stem cell divisions characteristic of a tissue. The more cell divisions, the more chance for errors in DNA copying that lead to the untrammeled cell growth that is cancer. The paper’s Editor’s Summary concluded, “Remarkably, this ‘bad luck’ component explains a far greater number of cancers than do hereditary and environmental factors.”
Bloggers try to explain what the paper means–and doesn’t mean
Statistician David Spiegelhalter, who blogs at Understanding Uncertainty, said the paper claimed that “around two-thirds of the variation in incidence rates is explained by chance mutations of stem-cells.” The authors, he said, conclude in their abstract that “’only a third of the variation in cancer risk among tissues is attributable to environmental factors or inherited predispositions’, which may be a fairly reasonable statement to make about population rates in different tissues, but of course says nothing about variation in risks between individuals, and certainly does not say that two-thirds of cases are just luck.”
The anonymous blogger at Plumbum (a statistician? an oncologist?) put it somewhat differently: “I emphasize that ‘two thirds of cancer types’ is not at all the same as ‘two thirds of cancer cases’.”
Some bloggers were furious that in emphasizing the random nature of many cancers, the reports seem to suggest that lifestyle didn’t matter. A heartfelt example was Henry Scowcroft’s at Cancer Research UK’s Science Blog. Reports about the paper at places like the BBC “appeared to contradict the message that many organisations have been trying to hammer home (including us): that although there are no guarantees, we can stack the odds of avoiding cancer in our favour if we embrace a healthy lifestyle.” (Emphasis his.) That’s a huge shame, he said, and to prove it cited this tweet:
“THANKYOU BBC!! Smoking doesn’t cause cancer. It’s just bad luck.”
OK, that’s scary. Unless the twit was just being facetious.
In their severe blast at the media in the Guardian, statistician Bob O’Hara and GrrlScientist said the paper’s data “suggest there is a relationship between risk of cancer and number of cell divisions. But it says nothing about the proportion of cancers due to cell division.” Spiegelhalter likewise noted that media treatment of the paper frequently conflated the population rate for particular cancers with individual risks, and that even when the journalist made the distinction, the headlines usually did not. For this Spiegelhalter blamed the Vogelstein-Tomasetti paper itself more than the media.
He also noted that the idea that most cancers are random and unpreventable is not new. Cancer Research UK late last year announced that 40% of cancers are preventable. Which means that 60% are not. Pretty similar, really, to the Vogelstein-Tomasetti estimate.
At Evolving Economics, Jason Collins blamed much of the journalists’ confusion on the initial press release from the researchers’ institution, Johns Hopkins, and noted that a corrected release issued later didn’t really explain why the first one was misleading. (Gary Schwitzer pointed out at Health News Review that a misleading press release doesn’t excuse misleading journalism. Science and medical journalists are supposed to go way beyond the press release, reading the paper itself carefully and consulting other sources before writing.)
In her ScienceInsider mea culpa for her own reporting on the paper, Jennifer Couzin-Frankel acknowledged that she had a hard time grasping just what the researchers were getting at, which is this: “Some tissues are overtaken by cancer more readily than others, and mutations accumulating in stem cells explained two-thirds of that variability.” Her post has been praised for its forthrightness, for example in another post by Schwitzer.
When I looked back at her original piece, though, it didn’t seem to me she had much to apologize for–except, maybe, for writing that the paper said stem-cell mutations “explained two-thirds of all cancers.” In fact the study didn’t include all cancers. Indeed, it excluded two of the most common, breast and prostate cancer, because the authors couldn’t find good data on stem cell divisions in those cancers.
If you’re floundering a little on the statistical blogs, flee to PZ Myers at Pharyngula for clear explanations of what the paper found. But the most comprehensive post I encountered was cancer specialist David Gorski’s at Science-Based Medicine. It’s long, but covers nearly all the points mentioned by other bloggers.
Gorski explained the theory behind the paper, cited several journalistic misinterpretations (noting that many were probably based on the misleading Hopkins press release), and described what the paper found. He also criticized it, for instance the puzzling inclusion of melanoma in the random group although nearly all cases are due to UV light exposure.
Gorski also noted that, oddly enough, even if the paper is (mis)interpreted as showing that 2 out of 3 cancer cases are due to chance, the conclusions are reasonable. In most cases, the cause of a particular cancer is a mystery.
Does that mean that Tomasetti and Vogelstein are “on to something in concluding that stem cell replication over one’s lifetime primarily determines the ‘stochastic’ component of cancer risk for each organ? That remains to be seen, but their preliminary finding makes sense, both from the perspective of producing a result that’s in the ballpark of what we already know based on epidemiology and being biologically plausible based on basic cancer biology.”
Enter epistemology: the meaning of “luck” and “random” and “chance”
Vogelstein and Tomasetti set the stage for misinterpretation and complaints by calling on “bad luck.” They introduced the term in the abstract, guaranteeing that “bad luck” would be part of how the paper was explained to others. The authors defined the term precisely enough, but of course their definition–“random mutations arising during DNA replication in normal, noncancerous stem cells”–isn’t what the rest of the world means by “bad luck.”
Gorski put the problem this way: “[S]ome of the objections to this paper seem to flow from a belief in inflated estimates of just what proportion of cancer is due to ‘environment’ and is therefore potentially preventable. It’s been suggested that cancer biologists might be too fast to blame unknown causes on ‘randomness,’ the assumption being that not knowing something means that we will know it in the future and more prevention will be possible. The problem is that not knowing something doesn’t mean that there’s a realistic way of obtaining that missing knowledge or that even if we obtained that knowledge that we’d be able to do anything with it.”
Myers was not so understanding. In another Pharyngula post, he described eye-opening (for him) Twitter exchanges with people who have “a striking psychological antipathy to the whole idea of random effects.” This came as a surprise to him. His early training was in genetics, he explained, “and there you acquire a strong appreciation for the importance of chance events.”
I can’t help wondering if the fact that Myers is a scientist (an evolutionary biologist) is related to his thinking the belief that everything has a cause is “weird.” Gorski, an oncologist who sees patients, has a different view. Humans, he said, crave explanation.
Gorski described his many uncomfortable moments with patients who want to know why they have cancer. Usually he must tell them he just doesn’t know and there’s nothing they could have done to prevent it. Patients don’t find that answer satisfying. “People—including oncologists—really don’t like the concept of ‘sporadic’ cancer, mainly because humans crave explanation. The default assumption is that everything must happen for a reason and there must be a cause for every disease or cancer.”
That’s exactly what one of Myers’s tweet adversaries said: By definition, “luck” has no cause and is therefore unscientific. Everything has a cause, and therefore in principle the cause ought to be discoverable eventually.
The Myers response: “No matter how hard we work, we will never have a sufficiently detailed explanation of every feature of the universe to negate the importance of chance. . . [physics is saying otherwise] but even if it were found that the universe is completely deterministic, the complexity of the phenomena and the number of parameters means that those kinds of causes are unknowable, and randomness is a good higher-level description of what is going on. So get used to it.”
That may be a rational view of the universe. But I wonder if people who want to believe that events have causes, who need to feel in control of their lives, will find it persuasive.