Nilanjan Chatterjee sees a more effective, risk-based approach to preventing disease.
From his vantage at the intersection of biostatistics, epidemiology and genetics, Nilanjan Chatterjee seeks to improve upon decades-old approaches to screening and prevention for diseases like breast cancer.
Consider a recent JAMA Oncology paper that the Bloomberg Distinguished Professor co-authored with Paige Maas, PhD ’14. The scientists showed that women whose genes and family history conspire to maximize their risk of breast cancer could nonetheless experience the greatest reduction in absolute risk by modifying lifestyle-related factors like body mass index (BMI), alcohol consumption, menopausal hormone therapy (MHT) and smoking.
The hope, says Chatterjee, is that such findings could one day allow doctors to offer women personalized recommendations for breast cancer screening and prevention.
How did you come to work at the nexus of so many fields?
I was trained in mathematical statistics and probability in India, and I got interested in biostatistical work while doing my PhD at the University of Washington. But my real education in public health, genetics and epidemiology started at the National Cancer Institute.
At NCI, I was involved in designing and analyzing genome-wide association studies for many different cancers. That gave us an unprecedented opportunity to understand the genetic basis of complex diseases. For example we learned that cancers are associated with many, many genetic variants called single nucleotide polymorphism markers, or SNPs. I was also interested in understanding the interaction between genetic susceptibility and other, nongenetic factors, such as lifestyle, diet and behavior. How do these nongenetic or environmental factors add together with genetic factors? Most recently, I became interested in what that meant from a public health point of view—and that meant looking at gene-environment interactions in terms of risk reduction. What if women could change some of their nongenetic risk factors to reduce their risk of breast cancer? How different might the impact of those nongenetic risk factors be, based on their genetic profiles?
How does that play out in your JAMA Oncology paper?
We tried to show what the risk reduction would be if women followed a healthy lifestyle given their genetic profile, as defined by 92 different SNPs tied to breast cancer susceptibility, and a few other nonmodifiable factors like family history. And what we found was that women with the highest genetic risk could experience the biggest reduction in their absolute risk by modifying certain lifestyle factors.
That’s a very important observation since in the past many studies stopped at evaluating whether genetic and environmental factors jointly affected disease risk in a multiplicative fashion. They didn’t evaluate the consequences of such models for risk reduction. But because precision prevention is going to be a part of future precision medicine initiatives, we need to do more of these types of analyses to develop more individualized approaches to disease prevention.
It’s really important to understand that a healthy lifestyle has broad beneficial impact for everyone. But some women who have more risk due to genetic risk factors actually gain even more by making healthy lifestyle choices, which might have implications for motivating more people to do the right thing.
A healthy lifestyle has broad beneficial impact for everyone. But some women who have more risk due to genetic risk factors actually gain even more by making healthy lifestyle choices, which might have implications for motivating more people to do the right thing.
Do each of those modifiable, nongenetic risk factors carry the same weight?
No. In the paper, we considered the four factors simultaneously, and also one factor at a time. Altogether, we estimated that close to 28 percent of breast cancer is preventable if you follow a healthy choice for all of those four factors; but a lot of the benefit is going to come from reducing MHT, alcohol and BMI. MHT was the most important, then BMI, then alcohol and then smoking. Smoking has a much smaller effect on breast cancer, though it obviously has a much bigger effect on lung cancer and other cancers.
What implications could this have for recommendations about breast cancer screening?
The current recommendation by the U.S. Preventive Services Task Force is that women should start screening for breast cancer at age 50. But why do we use age alone to decide when a person should start screening? Age is the strongest risk factor for a lot of chronic diseases, including breast cancer; but there are a lot of other risk factors, like reproductive and lifestyle factors, family history and genetics. So a more rational approach would be to use age and other risk factors, as well.
We show that if you use those other risk factors, 16 percent of women will at age 40 have the same risk as a 50-year-old woman, and 30 percent of women at age 50 will have the same risk as a 40-year-old woman. So you can see that if we use this risk-based approach, then the formal recommendations are going to change substantially.
But we want to make sure that these risk models are validated first in independent studies. And we have to think about the cost. Age is easy, and nongenetic factors can be evaluated by a simple questionnaire. But a genetic test has a real cost.
Thanks to technology, however, the cost of genetic testing is going down dramatically. You can probably get those 92 SNPs tested for less than $50 now; and in the future, the economics will completely change, because genotyping and sequencing will become so cheap that this data will be routinely available for everybody. And the nice thing about genetic testing is that if you do the whole genome, you can use the same test to evaluate the risk of breast cancer, Type II diabetes and a whole bunch of other things. So there’s an economy of scale.
Could this kind of personalized, risk-based approach to screening and prevention in fact be used for other diseases?
Our study is the first of its kind, but it can be applied to any disease. It provides the intellectual framework for asking these kinds of public health-related questions, and for presenting the data in a way that can be used in a clinical setting.