Journal Logo


That ‘Bad Luck’ Science Paper

Insights from cancer biostatisticians & biologists

DiGiulio, Sarah

doi: 10.1097/01.COT.0000462844.84052.49
  • Free

In a much-talked-about study in Science (2015;347:78-81), Cristian Tomasetti, PhD, and Bert Vogelstein, MD, quantified (for the first time) a strong positive correlation between the variation in cancer risk among different organs and the number of stem cell divisions in those organs. Although many of the resulting headlines and think pieces in the lay press overemphasized—and in many cases misinterpreted—the link between “cancer risk” and “luck,” cancer biostatisticians say the paper is a significant step in understanding the biology of what drives cancers to arise and highlights key directions for future research.

Asked for his opinion for this article, Giovanni Parmigiani, PhD, Professor and Chair of the Department of Biostatistics and Computational Biology at Dana-Farber Cancer Institute, said the point of the paper is essentially a very simple one: to show the correlation between the number of stem cell divisions in a tissue and the likelihood of a cancer arising in that tissue. “It's not necessarily a causal relation, but it is a very very important correlation to be aware of. And it is the first time this correlation has been quantified.”

Although he was not involved in this latest work, Parmigiani has worked with both Vogelstein and Tomasetti in the past on earlier genome cancer projects in Vogelstein's lab, as well as other biology mathematical modeling projects.


“The paper represents a genuine breakthrough in our understanding of how cancer arises,” Parmigiani said in a phone interview. “It is the first time that anyone can quantify the effect of the imprecise fidelity of cell division in the oncogenetic process. ... Nobody before had collected so systematically and accurately the number of cell divisions that tissue-specific stem cells undergo during the course of one's lifetime.”

Paper Details

The paper set out to determine—as Vogelstein, Professor of Oncology and Pathology and Director of the Ludwig Center for Cancer & Therapeutics at Johns Hopkins Kimmel Cancer Center and Howard Hughes Medical Institute Investigator; and Tomasetti, Assistant Professor in the Division of Biostatistics and Bioinformatics in the Department of Oncology at Johns Hopkins Kimmel Cancer Center, note in the opening paragraphs—given that hereditary and environmental factors cannot fully explain the differences in organ-specific cancer risk, how else the differences in cancer risk between tumor types could be explained.

Also speaking via telephone, Vogelstein said that before the paper was published there was undoubtedly an understanding that “replicative mutations—a stochastic component or ‘bad luck’”—were involved in cancer development. “That's unquestionable. But this was the first time that this component could actually be measured.”


The findings suggest, he explained, that for the 31 cancer types included in the data, two-thirds of the mutations that are required for cancer appear to be explained by replicative factors that occur simply because cells are dividing.

Based on previous research (including genome-wide analyses) finding that genetic and epigenetic changes in large part drive the development of cancer, the authors theorized that the stochastic effect associated with the lifetime number of stem cell divisions within each tissue are a major contributor to cancer overall and can be more important than hereditary or external environmental factors. The more frequently stem cells in a certain tissue divide, the more likely a cancer is going to develop in that tissue.

As Tomasetti and Vogelstein explained in the paper, they plotted the total number of stem cell divisions during the average lifetime of a human against the lifetime risk for cancer of the tissue type of that tumor. A total of 31 tissue types in which stem cells had been quantitatively assessed were used for the analysis, along with the lifetime cancer incidence risk in the U.S. for those cancer types based on data from the Surveillance, Epidemiology, and End Results database.

The data showed a highly positive correlation—with a linear correlation coefficient of 0.804, suggesting that 65 percent of the differences in cancer risk (between 39 and 81 percent based on the 95 percent confidence interval) among different tissues can be explained by the number of stem cell divisions in those tissues. And, as noted in the paper, that finding suggests that the stochastic effect of DNA replication appears to be the major contributor to cancer in humans.

The authors offer further evidence of their theory by assigning each cancer type in the study an “extra risk score”—defined (for the purpose of the paper) as the product of lifetime risk and the total number of stem cell divisions. A high extra risk score (high cancer risk for a particular tissue relative to the number of stem cell divisions in that tissue) therefore would indicate that environmental or inherited factors should play a relatively more significant role in that cancer's risk.

And the data showed that the tumors with relatively high extra risk scores (including familial adenomatous polyposis and Lynch syndrome colorectal cancers, basal cell carcinoma, and lung cancers in smokers) were those with known links to specific environmental or hereditary risk factors.

Key Points

A key point of the paper was defining replicative mutations as mutations that are caused solely by the process of normal DNA replication, Vogelstein said. “Every time a perfectly normal human cell divides—even in a test tube—it will make about three mutations. That's been measured. It's a side effect of evolution—cells make mistakes.”

Such random replicative mutations are not the only mutations that cause cancer, though, of course; the others are inherited genetic mutations or environmental-associated mutations caused by exposure to carcinogens such as cigarette smoke or sunlight.

GIOVANNI PARMIGIANI, PHD. GIOVANNI PARMIGIANI, PHD: “The paper represents a genuine breakthrough in our understanding of how cancer arises.”

It has been suggested that other stochastic components (i.e., other forms of “bad luck”) also play a role in cancer, he added. “For example, the immune system and epigenetic switches may also play a role. But our research shows that the majority of the stochastic component can be attributed to mistakes made during DNA replication.

“The theory explains why many cancers may not be preventable by changing lifestyle or environment,” Vogelstein said.

View from a Cancer Stem Cell Expert

Commenting for this article in a phone interview, cancer stem cell biologist Quintin Pan, PhD, Associate Professor and Research Director of the Head and Neck Oncology Program at The Ohio State University, agreed that the key takeaways from the paper have big implications for his field. The findings support the theory that normal stem cells are the cells of origin for cancer, he said.


“The cell origin for cancer has been a mystery—and it is still a mystery,” he explained. The stem cell hypothesis posits that a unique cell population, cancer stem cells are the only cells that can initiate and repopulate a tumor. And, Pan added, the Vogelstein paper suggests the intriguing idea that normal stems cells through accumulated mutations may be the cell of origin for cancer stem cells—providing the first evidence to support the suggestion that normal stem cells can be reprogrammed into cancer stem cells through mutations.

“The data presented in this paper suggest that mutations in normal stem cells may convert those stem cells into cancer stem cells. ... I think the advance [from this paper] is that perhaps it's the mutations in the normal stem cell compartment that can lead to subsequent replication of these mutated stem cells leading to an increase in cancer risk.”

QUINTIN PAN, PHD. QUINTIN PAN, PHD: “As a cancer biologist with a focus on anti-cancer therapeutics, I am interested to know: What are the key mutations?”

The next step for the research is to further validate the findings and then reveal the driver mutations in the normal stem cells that promote oncogenesis, Pan added. “As a cancer biologist with a focus on anti-cancer therapeutics, I am interested to know: What are the key mutations? And which mutations are indispensable to promote cancer? These key pieces of the puzzle are needed so we can begin to design therapies that can target these critical genetic abnormalities and hopefully cure cancer.”

Implications for Cancer Biologists

Right now, the key implications are for cancer biologists, Parmigiani said: “Don't take for granted that if we work at it hard enough we will be able to pin down cancer-wide genetic theories or environmental and behavioral factors [that cause cancer]. There could potentially be a hard core of randomness that we will never be able to get rid of.”

Parmigiani said the findings do affect his work in cancer biostatistics and computational biology: “In terms of understanding mechanisms of variability and mechanisms of cancer progression, the paper opens up many new avenues for thinking about cancer evolution, especially from a mathematical standpoint. And it gives us a key through which we can try to interpret the effect of genetic variation or environmental exposures that are known to cause cancer.

“We can now try to see whether cancers are developing based on this channel of the number of stem cell divisions—or they are acting independently. In a search for understanding the mechanism of cancer, I think this is a very important guiding light.”

But, he added: “This research is about understanding the biology [of cancer]. There's a lot of work that we can do before we can go back to the community of oncologists and say, this is what this really means.

DOMINIK WODARZ, PHD. DOMINIK WODARZ, PHD: “Understanding the key forces that drive the development of cancer [as shown in the Tomasetti and Vogelstein paper], is important for achieving better prevention strategies.”

“Down the line I hope this will reinvigorate the efforts that we make for early detection,” he added

Implications for Prevention

In a “News & Views” article published in Nature in response to the paper, Dominik Wodarz, PhD, Associate Professor of Ecology and Evolutionary Biology in the School of Biological Sciences at the University of California, Irvine, and Ann Zauber, PhD, a biostatistician in the Department of Epidemiology and Biostatistics at Memorial Sloan Kettering Cancer Center, note that a key takeaway from the research is recognizing the significance of chemoprevention to slow the evolutionary processes of cancer development, as well as the generation of early-detection techniques (Nature 2015;517:563-564).

“The Tomasetti and Vogelstein paper highlights the importance of evolutionary mechanisms for the development of cancer; and further suggests that interfering with the evolutionary processes at work using specific drugs could prevent or delay cancer—chemoprevention,” Wodarz added in an email.

The challenge is that screening is currently difficult for many cancers, especially rare cancers, Wodarz and Zauber noted in their article. To be effective, screening needs to be used in settings where the benefits achieved outweigh the risks of complications from the screening and also outweigh the burden of false positives that cause excess testing and worry.

“If random genetic changes are more relevant drivers of carcinogenesis, then biomarkers and early detection methods will have to be developed to prevent cancer mortality in the general population.”

Another important takeaway from the Tomasetti and Vogelstein paper is to be aware of the degree of chance in cancer development in terms of recognizing who is at risk for various types of cancer, Zauber added via email. “We need to allow for imperfection in identifying all who are at risk for a given cancer. We use information of who is at higher risk and needs further surveillance—but we cannot forget that many (or most) of the rest of cancer cases are probably arising in those without immediately identifiable risk factors. We must be aware of this large random component to cancer development and be ready to follow up with a wider group of at-risk individuals who might be at risk.

ANN ZAUBER, PHD. ANN ZAUBER, PHD: “We need to allow for imperfection in identifying all at risk for a given cancer. We use information of who is at higher risk and needs further surveillance—but we cannot forget that many (or most) of the rest of cancer cases are probably arising in those without immediately identifiable risk factors.”

“The challenge is how to accomplish a larger surveillance without causing harms from un-needed testing.”

Insights from George Sledge...

He shares how the paper relates to what he tells his patients (OT 3/10/15 issue).

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
    Home  Clinical Resource Center
    Current Issue       Search OT
    Archives Get OT Enews
    Blogs Email us!