‘Precision Prevention’ for Colorectal Cancer
News Apr 20, 2016
Precision medicine’s public face is that of disease — and better treatments for disease through targeted therapies.
But precision medicine has an unsung partner that could affect the lives of many more people: Precision prevention — a reflection of the growing realization that preventing cancer and other diseases may not be one-size-fits-all.
“Precision medicine has been kind of a buzzword recently, but often when people think about precision medicine, they think about treatment,” said Fred Hutchinson Cancer Research Center biostatistician Dr. Li Hsu, who researches precision prevention for colorectal cancer. “I think it’s just as important if not more important to prevent disease.”
In work presented Monday at the American Association for Cancer Research’s annual meeting in New Orleans, Hsu and other researchers from Fred Hutch, the University of Michigan and other institutes debuted their latest progress in precision prevention — an in-the-works method to predict risk of colorectal cancer that integrates genetic, lifestyle and environmental risk factors.
This research is not yet ready to move into clinical practice, said Fred Hutch epidemiologistDr. Ulrike (Riki) Peters, one of the study authors. But it’s the first attempt at combining so many different areas of colorectal cancer risk factors into one comprehensive risk predictor.
Current risk stratification methods for colorectal cancer screening recommendations are based on age and family history alone. No family history of the disease? Start colonoscopies at age 50. Have an immediate relative who had colorectal cancer? Start at age 40.
But these methods are likely missing many at risk, Peters said. Eighty percent of those with colorectal cancer have no known first-degree family history. And, unlike some cancers, it’s a disease where screening and prevention are tightly linked — colonoscopies can catch premalignant lesions and if those lesions are removed, the patient is likely spared from developing cancer.
“That is a very unique aspect of colorectal cancer,” Peters said.
Even though the disease is highly preventable if caught in the precancerous stages, colorectal cancer is the second leading cause of cancer-related deaths (for men and women combined) in the U.S., topped only by lung cancer. So along with encouraging people to get the recommended colonoscopies, a better sieve to catch those at higher risk of the disease could have an impact both on cancer prevention and on sparing those at low risk unnecessary procedures.
“At the end, what we want to do is to reduce disease burden given limited resources,” said Dr. Jihyoun Jeon, a cancer modeler and epidemiologist at the University of Michigan who presented the risk prediction model in a poster at the AACR meeting. “We want to save resources but also prevent as much [disease] as possible.”
Stitching the risk factors together
The improved risk prediction method was developed using data from more than 18,000 people, approximately 8,400 of whom had colorectal cancer. These data come from two large colorectal cancer studies that Peters leads or co-leads, known as the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO) and the Colorectal Transdisciplinary Study (CORECT).
Peters, Hsu and their colleagues have been working for years to identify the genetics behind colorectal cancer. It was always their goal to use that information to improve risk prediction, Hsu said, but only recently has the team amassed enough links between genes and disease to be able to work on the precision prevention piece of the puzzle.
Using 19 known environmental and lifestyle risk factors for the disease and 64 common genetic variants, the statisticians sorted the more than 18,000 people in their dataset along a continuum of high to low risk of colorectal cancer.
Some of the known risk factors for colorectal cancer include smoking, obesity, a sedentary lifestyle and diets high in red meat and processed meat, Peters said. Increasing folate, calcium, fruit and vegetable intake and, for some, use of aspirin or non-steroidal anti-inflammatory drugs can reduce risk of the disease.
Hsu stressed that the model, which was funded by the National Cancer Institute, currently only provides guidance on when to start regular colonoscopies. The researchers haven’t yet addressed the issue of changing the frequency to more or less than what’s currently recommended — every 10 years (for those at average risk) — as building such models requires additional information such as results from previous screenings. After crunching the numbers, their model spat out recommendations that those in the highest 10 percent of risk start screening at age 44 (for men) or 47 (for women), if they do not have any immediate family with colorectal cancer. In the lowest 10 percent, the model recommended starting colonoscopies at 58 for men and 63 for women.
“If someone is at low risk and would not start screening until age 60, that means you would have one less screening during your lifetime,” Hsu said.
The important question, of course, is whether their risk prediction model will catch and prevent more cases of cancer than current guidelines. The researchers don’t yet know. But Peters is hoping to soon launch a new project to address that question by looking at how their model fares in a large community group, to be conducted with researchers from Kaiser Permanente.
It’s hard to predict when their method could show up in the clinic, the researchers said. They need to first validate that it works in another large dataset of study participants, which they’re planning to start soon, using data from the NCI’s Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial.
And they also need to assess whether the tool, which includes genetic testing, would be cost-effective. They have primary data suggesting that it is but need to do more analyses, Peters said. After which, the next step is to study the model’s effectiveness at predicting cancer risk in a randomized clinical trial, she said.
“This is showing the path where we would like to move toward,” Peters said. “Most [high-risk] people don’t know that they have an increased risk. We would like to inform this better by not making this one size fits all.”
MIT researchers have developed a cryptographic system that could help neural networks identify promising drug candidates in massive pharmacological datasets, while keeping the data private. Secure computation done at such a massive scale could enable broad pooling of sensitive pharmacological data for predictive drug discovery.
Previous work by the International Multiple Sclerosis Genetics Consortium (IMSGC) has identified 233 genetic risk variants. However, these only account for about 20% of overall disease risk, with the remaining genetic culprits proving elusive. A new study has tracked down four of these hard-to-find genes.READ MORE