Read our COVID-19 research and news.

As the number of stem cell divisions in a tissue rises, so does the chance of cancer striking that site.

As the number of stem cell divisions in a tissue rises, so does the chance of cancer striking that site.

Data: Tomasetti et al./Science

Bad luck and cancer: A science reporter’s reflections on a controversial story

We reporters—or this one, at any rate—often fail to anticipate which stories will grip readers and which will quickly fade into oblivion. Given that, perhaps I shouldn’t have been surprised that a story I saw off to the printing press in the lull between Christmas and New Year’s engendered more comments than any other I’ve written.

The piece, which appeared online with the headline “The simple math that explains why you may (or may not) get cancer” (and in the magazine’s News section with the headline “The bad luck of cancer”), described a paper published in the 2 January issue of Science. As I and many other journalists explained, the study suggested that simple “bad luck”—random mutations accumulating in healthy stem cells—could explain about two-thirds of cancers, exceeding the risk conferred by environmental and genetic factors combined. One message was that some cancers could not be prevented and that detecting them early was key to combating them.

Readers wasted little time in skewering the authors, mathematician Cristian Tomasetti and cancer geneticist Bert Vogelstein of Johns Hopkins University in Baltimore, Maryland. Their statistics were faulty, some argued; they included many rare cancers and left out several common ones. Earlier today, the International Agency for Research on Cancer, the cancer arm of the World Health Organization, put out an unusual press release stating it “strongly disagrees” with the report. The agency said that “nearly half of all cancer cases worldwide can be prevented.” It charged that the authors’ push for early detection “if misinterpreted … could have serious negative consequences from both cancer research and public health perspectives.”

Reporters, if anything, fared worse. “Please, journalists, get a clue before you write about science,” pleaded an irate column in The Guardian, co-authored by an evolutionary biologist who goes by the Twitter handle @GrrlScientist and statistician Bob O’Hara at the Biodiversity and Climate Research Centre in Frankfurt, Germany.

Given the furor, I wondered: Had I gotten it wrong? Had the authors? Answering these apparently straightforward questions proved surprisingly difficult, exposing the challenges that come with communicating science, and the desire by scientist-authors and reporters to streamline the story they’re trying to tell.

I began with my own story, working backward to the science that spawned it. I’d written that the theory of random mutations in stem cells “explained two-thirds of all cancers.” Immediately, I knew that I had written part of that sloppily, to put it generously: The study didn’t include all cancers. In fact, it didn’t include two of the most common, prostate and breast, because the authors weren’t able to pin down the size of the stem cell compartment or the frequency of stem cell divisions in those tissues. Although my piece subsequently noted the number of cancer types in the study, I should have stressed the omissions early on.

Still, was “two-thirds” referring to the number of cases of cancers the study did include, as I and other journalists had suggested—or to something else? Journalists like numbers that abridge a study down to a bullet point. I’d wondered immediately if this two-thirds finding might be one such nugget. Tomasetti had explained to me in a lengthy interview that “if you go to the American Cancer Society website and you check what are the causes of cancer, you will find a list of either inherited or environmental things. We are saying two-thirds is neither of them.” I’d run the text of my "two-thirds" sentence by him prior to publication and he had had no objections (he had other clarifications).

Last week, we spoke again. Tomasetti had received more than 200 e-mails. Parents of children who had died of cancer were grateful that it might have occurred entirely by chance, suggesting that there was nothing they could have done. Biologists and statisticians were disputing his conclusions or simply surprised that so much of cancer might be random.

“We did not claim that two-thirds of cancer cases are due to bad luck,” Tomasetti told me gently. What the study argued, he explained, was that two-thirds of the variation in cancer rates in different tissues could be explained by random bad luck (a point made by others). What exactly did that mean, I wondered? Tomasetti, chatting by phone, had me draw some graphs to help me understand. By the end of the hour, I still wasn’t sure I grasped the essence.

Tomasetti was sympathetic. “There are lots of scientists that need clarification” on this paper, he said, along with some statisticians. He was busy preparing a technical report with additional details, and Johns Hopkins had just put out a press release explainer. “I honestly feel—and that’s what I told the BBC, and you can definitely quote me on this—overall, the reporters who interacted with us made a very honest and sincere effort to be as accurate as possible.”

It was only after more hours spent on interviews that I finally understood the two-thirds figure. Some tissues are overtaken by cancer more readily than others, and mutations accumulating in stem cells explained two-thirds of that variability, Tomasetti and Vogelstein had concluded. It was my “aha” moment, and it came too late—after my original deadline.

I contacted some of the critics. “I just reread your article, and I don’t think it falls into the bad category (at worst, it skirts around the lip without dropping in),” wrote O’Hara, an author of the Guardian piece, in an e-mail—confusing me further, for hadn’t I goofed up? By phone, he explained that one of his quibbles was the word “luck”—present in the paper’s abstract, emphasized by the authors, and highlighted in nearly every news story. It sounded sexy, but O’Hara considered it inaccurate, because virtually all cancer is a product of luck in some sense.

“It’s too easy to blame the media,” said David Spiegelhalter, a biostatistician at the University of Cambridge in the United Kingdom, who had blogged about the story. (“Your article was fine,” he assured me.) In this case, he felt, “the gist of the coverage is very reasonable—most cases of cancer are due to chance.”

That said, errors were made along the way, a fact that didn’t surprise him. “This is incredibly difficult stuff,” Spiegelhalter continued. “I do feel for you. It’s one of those things that’s so superficially simple, and yet the superficial simplicity is not correct.”

The paper’s authors, many felt, were also guilty of trying too hard to craft a simple message. The paper included a visually arresting diagram splitting cancers into green and blue categories. The green were cancers “mainly due” to random mutations—suggesting, the authors wrote, that they were less likely to be prevented by changes in behavior or diet. However, that category included esophageal cancer and melanoma, both of which are thought to have strong links to environmental drivers such as heavy alcohol consumption and sun exposure, respectively.

Melanoma sat just slightly inside the green border—but still, it was green, which left many readers exercised. “They’ve ignored some of the fundamental lifestyle factors,” said Graham Colditz, a cancer prevention specialist at Washington University in St. Louis. “Obviously, they had good intentions,” said Anne McTiernan, a physician and epidemiologist at the Fred Hutchinson Cancer Research Center in Seattle, Washington. But, she continued, the authors assumed that a correlation between the number of stem cell divisions and cancer risk meant one was causing the other, something their data couldn’t prove. Tomasetti agreed that this is correct—but he notes that “all the biology we have on this topic supports” the idea that a buildup of random mutations in healthy cells can initiate cancer. Some scientists argued that the graph carried a missive for prevention, with huge risk gaps between a cancer driven by the environment or genetics—such as lung cancer in smokers or head and neck cancer linked to human papillomavirus—and cancer at the same site without a clear cause.

The nuances were many. Even if they quibble on the details, most would agree that random mutations play a real role in cancer - but so do many other things. Despite the furor, this common ground is shared by both the paper’s authors and its critics. “This is a really fascinating pattern that they’ve observed, but it is a small message,” says Timothy Rebbeck, a cancer prevention specialist at the University of Pennsylvania. “It doesn’t exclude the ability to prevent and treat cancer. It doesn’t exclude our need to better understand the causes of cancer.” The paper’s bottom line wasn’t simple, but the message for me was: Science is complicated, and people care deeply about the biology of diseases that affect their loved ones and themselves. Distilling the story—with space constraints, with a desire for clear writing that will hold readers’ attention and help them understand—carries risks for scientists and for journalists. They are ones I hope never to forget—even if I err now and again.

Revised, 2:53pm, 1/14/15: This story has been revised to remove references to unpublished letters to Science.