Combined Monogenic, Polygenic Risk Scores Could Change Medical Understanding of Disease
By Allison Proffitt
August 20, 2020 | In a paper published today in Nature Communications, IBM and the Broad Institute reported the first results of their collaboration to develop predictive models for cardiovascular disease. Along with Color, the team explored the interplay between monogenic and polygenic risk, proposing that polygenic risk data will significantly improve risk estimation for those who inherit a monogenic risk variant—in some cases dramatically lowering an individual’s absolute risk for a disease. (DOI: 10.1038/s41467-020-17374-3)
Monogenic risk variants disrupt a physiologic pathway with large effect on disease, a single variant can dramatically change an individual’s risk. Polygenic risk reflects multiple variants of small effect in various pathways that work together to impart risk. Individually, these variants have a small effect on disease risk, but together their impact may be significant.
We don’t clearly understand how the two impact each other. “Can disease risk from a monogenic variant that causes major disruption to a specific pathway be meaningfully modified by polygenic risk factors that involve small perturbations to a wide range of cellular pathways?” the authors write. There is some evidence of this, they argue. The common variant background seems to modify the age of disease onset in several diseases linked to single, impactful variants including Huntington's disease and glaucoma. Common variants also seem to amplify or dampen monogenic risk for more common diseases too, like breast, ovarian, and prostate cancer.
The IBM/Broad/Color group studied 80,928 individuals to examine whether incomplete penetrance of known harmful variants could be partly explained by polygenic background.
They looked at three genomic conditions: familial hypercholesterolemia, hereditary breast and ovarian cancer, and Lynch syndrome. All three are labeled Tier 1 Genomics Applications by CDC’s Office of Public Health Genomics denoting conditions with significant potential for positive impact on public health.
They designed two case-control studies from the UK Biobank and Color Genomics commercial testing laboratory, enriching for disease cases to increase statistical power. They also analyzed an independent cohort from the UK Biobank to assess outcomes for both carrier and noncarriers of monogenic risk variants within the context of contemporary medical care.
Using their findings, they calculated new risk scores for coronary artery disease, breast cancer, and colorectal cancer for these participants.
For instance, under the standard model considering only monogenic variant risk, the presence of a familial hypercholesterolemia variant confers a 3.21-fold increased risk of coronary artery disease.
The team compared 6,432 coronary artery disease cases and 6,420 controls from the UK Biobank to examine the impact of the participants’ polygenic background on coronary artery disease risk and grouped the participants into three groups: low polygenic risk scores, intermediate polygenic risk scores, and high polygenic risk scores.
When considering both sets of risk data, risk scores for participants who had the familial hypercholesterolemia variant then ranged from 1.30-fold for those in the lowest quintile of the polygenic score distribution to 12.61 in the highest quintile.
They tested their reasoning again on an independent cohort of 48,812 unrelated UK Biobank participants. Joint modeling of monogenic variant carrier status and polygenic score indicated substantial gradients in risk of coronary artery disease according to inherited DNA variation that can be assessed from the time of birth. Risk ranged from 4.9% for noncarriers in the lowest percentile of the polygenic score to 77.9% for monogenic risk variant carriers in the highest polygenic score percentile.
The team applied the same model to breast cancer using Color Genomics data and UK Biobank data. When modeled as probability of breast disease by age 75 years, risk among monogenic variant carriers ranged from 12.7 to 75.7% and risk among noncarriers ranged from 3.3 to 29.6%, the authors write.
Colorectal cancer followed the same pattern. Using the same UK Biobank cohort of 48,812 participants, they found that estimated absolute risk of colorectal cancer by age 75 years ranged from 11.3 to 79.7% for carriers and 0.7 to 8.7% for noncarriers.
“Risk conferred by monogenic risk variants, which act by perturbing a specific molecular pathway, can be substantially modified by polygenic background,” the authors write. “From a physiological standpoint, additional study is needed to understand how the major disruptions caused by monogenic variants can be offset by other factors. Yet, the risk for monogenic variant carriers with the lowest polygenic risk scores approached the population average.”
These findings have both scientific implications about disease physiology and clinical implications for genetic counseling, they point out.
In a blog posted at the same time as the paper was published, Kenney Ng, with IBM Research, wrote: “Accounting for polygenic background is likely to increase the accuracy of risk estimation for individuals who inherit a monogenic risk variant, and could provide clinicians with another piece of the puzzle as to the true risk of an individual actually developing a condition. When combined with other clinical information, such as medical history, lifestyle and other risk factors, this could help doctors to better assess prevention plans as well as potential treatments and therapies.”
But this level of risk assessment is laborious, and the authors acknowledge that. It requires high-coverage sequencing of the monogenic risk genes, but also a comprehensive look at the common variant background. Improvements to the cost and accessibility of genome sequencing will enable both, they say. And it would be worth it.
“Ultimately, our findings unveil a silver lining: even if an individual carries a genetic mutation associated with one of these diseases, their absolute risk might not be as set in stone as previously thought,” Ng writes. “In fact, their absolute risk might be nearly equivalent to an individual who doesn’t carry the mutation at all—depending on other factors and mutations within their specific genome.”