When it comes to cancer-causing mutations, a new paper argues that most arise from random cell divisions, with the environment second and inherited mutations a distant third, according to this chart based on cancers in British women. But many scientists don’t embrace this perspective.

C. Tomasetti et. al., Science 355, 6331 (24 March 2017)

Debate reignites over the contributions of ‘bad luck’ mutations to cancer

How much of cancer is due to random “bad luck”? More than 2 years ago, a pair of researchers brought that question to prominence when they tried to sort out environmental versus inherited causes of cancer. They examined the extent to which stem cell divisions in healthy cells—and the random mutations, or “bad luck” that accumulate—drive cancer in different tissues. Their effort, which implied that cancer was harder to prevent than hoped and that early detection was underappreciated, sparked controversy and confusion. Now, the researchers are back with a sequel: a new paper that aims to parse “bad luck” risks by cancer type, and that brings in cancer data from other countries.  

Cancer is largely a genetic disease, in which a medley of mutations accumulates to the point that a few cells reach a state of unchecked growth. Though some of the mutations may be more powerful than others, this new study focuses on apportioning cancer-causing mutations generally to one of three categories. Ultimately, its authors conclude, across 32 cancer types, 66% of cancer-promoting mutations arise randomly during cell division in various organs throughout life, 29% trace to environmental causes, and 5% are inherited. But the new paper appears unlikely to resolve, and may even reignite, the furor that was lit 2 years ago.

Many researchers Science spoke with took issue with the authors’ hard numbers for various reasons. “The whole idea of viewing cancer as a whole is pretty alien to us, the causes of different types are so different,” says Noel Weiss, an epidemiologist at the Fred Hutchinson Cancer Research Center in Seattle, Washington. And “what seems random today will not seem so random in the future.”

In January 2015, mathematician Cristian Tomasetti and cancer geneticist Bert Vogelstein, both at Johns Hopkins University in Baltimore, Maryland, described in Science their effort to untangle cancer’s randomness from inherited or environmental factors like sun exposure and smoking. Various organs contain stem cells, from which many cancers are thought to arise. Key to the pair’s original analysis was how quickly those stem cells are thought to replicate in different organs and how many such stem cells there are. The researchers matched those data against U.S. cancer rates in various tissue types. They found that the more stem cells and the more rapidly a particular organ’s stem cells replicate, the higher the risk of cancer in that tissue.

For example, because of this cells in the colon are much more likely to turn cancerous than those in the nearby duodenum, the researchers argued. Additional number crunching suggested that two-thirds of the variability in cancer risk across tissue types could be explained by these stem cells and the “bad luck” mutations that flourish there.

Many scientists took issue with the paper and how reporters communicated it, in part because they felt it overemphasized the randomness of cancer and downplayed the value of trying to prevent it. The study was also widely misunderstood as suggesting that two-thirds of cancer cases were attributable to random mutations.

In the current installment, published today in Science, the researchers wanted to look beyond how bad luck contributed to variations in U.S. cancer rates, and tackle two questions. First, would what they reported 2 years ago hold up outside the United States? They focused on 17 cancers—a mix of common and rare, such as colon cancer and the bone cancer osteosarcoma—for which incidence data were available across 69 countries, and for which they already had information on stem cell numbers and replication rates. (The researchers assumed these held steady in different populations because of basic human biology.) They found virtually the same correlation between cancer rates and the rates of stem cell division in those tissues—suggesting that the two-thirds figure holds up globally.

Their second question took a different tack. Could they break down the absolute proportion of random mutations driving different cancer types? Teasing out just what proportion of cancer-causing mutations is random had “never been addressed before,” Tomasetti notes. Along with graduate student Lu Li, the pair applied what was known about inherited contributions and environmental influences on mutation rates. They also drew on the Cancer Research UK database, which has in-depth epidemiological and other data about cancers in the United Kingdom. In pancreatic cancer, for example, their results suggested that just 5% of mutations were inherited, 18% were due to environmental factors like smoking, and the remaining 77% were the result of random mutations. For prostate cancer, they estimated that a whopping 95% of mutations that drive the disease were randomly acquired. In all, based on the U.K. data for 32 cancers, the researchers estimated that 66% of mutations driving cancer were due to “bad luck.” (Again, they emphasize, that doesn’t mean two-thirds of cancer cases are from those “bad luck” mutations.)

“Of course these are estimates,” said Tomasetti at a press conference on Wednesday. “It’s the best that can be done today. It’s a paradigm shift in how we think about cancer.”

But scientists across different specialties offered mixed reviews of the followup work, and many of their critiques echoed those advanced 2 years ago. “[The authors] make a point that heredity is associated with a certain percentage of cancer, and environment is associated with a certain percent, and this is probably true,” says Anne McTiernan, a physician and epidemiologist at the Fred Hutchinson Cancer Research Center. “But then to assume that the rest is because of stem cell divisions and chance … we just don’t know.” McTiernan was skeptical 2 years ago and remains so now.

“I think why people struggle with this, it’s a very reductionist approach to a complex problem,” says Richard Gilbertson, a pediatric oncologist and cancer biologist at the University of Cambridge in the United Kingdom. The genesis of cancer is incredibly intricate and only partially understood, he believes.

To make their calculations, the authors had to rely on a number of assumptions, several of which were questioned by outside researchers. For example, Gilbertson says, he doesn’t dispute that mutations in proliferating stem cells are the basis of cancer, and his own experiments in mice support this. But the authors assume that stemness is a “fixed” entity—a cell is either a stem cell or not—and take a “mutation-centric” view to cancer’s genesis, he says. The mice that Gilbertson has studied suggest that the big picture is a lot more complicated. In his lab, cells in, say, an animal’s liver that have stem cell potential can accumulate mutations but simply sit, harmless—and then, when he induces tissue damage to the organ, those cells “wake up” and turn cancerous. “Stem cell function is fluid, and can be activated by tissue damage independent of mutations,” Gilbertson says. "Understanding the elements that drive cancer, and it's not just mutations, is critical if we are to prevent it."

Tomasetti doesn’t disagree, but he suggests his research is being misunderstood. “We never argued in the paper that mutations are all that is needed for cancer to occur,” he wrote to Science. “We are not assessing the proportion of cancers explained” by random mutations, but the proportion of mutations that are random—a distinction that’s easy to miss despite its importance.

But Gilberton and others remain unconvinced. Tomasetti’s argument is “missing the point. … Cancer is far more complex than mutations and proliferation,” Gilberton suggests. This paper, he and other say, seeks to distill it down to those two drivers to the exclusion of everything else.

The argument is more than just dizzying semantics. At its core, Tomasetti and Vogelstein want their readers to think about something important: What do the findings mean for tackling cancer on the ground?  And though this paper agrees that about 40% of cancers are preventable—the same number embraced by most epidemiologists—it also carries implications for how to prevent them or detect them early enough for better treatment. For example, the authors believe that far more research into early detection strategies is needed.

In part that’s because the work suggests that right now, cancers caused only by random mutations are unpreventable—and that’s something about which Walter Willett, an epidemiologist at Harvard’s T.H. Chan School of Public Health in Boston, takes a more nuanced view. Harkening back to Gilberton’s point about thinking beyond mutations, he speculates that something as simple as weight loss or stopping hormone replacement therapy in menopause might inhibit random mutations from ultimately causing a life-threatening tumor.

If nothing else, it’s clear Tomasetti and Vogelstein are inspiring debate. Martin Nowak, who studies mathematics and biology at Harvard University, praises the authors for pushing the question of how much of cancer is due to random “bad luck” mutations, and for using statistics to probe the disease. But Nowak thinks we’re a long way from having the solution in hand, and worries about putting too much stock in the numbers being advanced right now. “It will take many years to settle” what proportion of cancer-causing mutations are random, he says. Not to mention what to do about them.