NEW YORK (GenomeWeb) – Much of the heritability of common diseases that's thought to be missing might be explained by shared environmental factors instead, according to a new analysis from a University of Edinburgh team.
Genome-wide association studies have uncovered scores of loci linked to susceptibility to various diseases, but these variants often only account for a portion of the estimated disease heritability. Edinburgh's Albert Tenesa and his colleagues, though, suggested that this missing heritability might not actually be missing, but instead be due to inflated heritability estimates.
He and his team estimated the heritability of a dozen complex human diseases by drawing on data from 1.5 million people from the UK Biobank using multiple approaches. From these analyses, they found that heritability estimated through simple family-based statistical models were inflated by an average of 47 percent, as they reported today in Nature Genetics.
"Our results are relevant when assessing potential for the development of personalized medicine, providing realistic expectations of the value of genetic testing," Tenesa and his colleagues wrote in their paper. "In addition, demonstration of the importance of environmental risk factors that contribute to the aggregation of disease within families motivates research to identify and moderate these factors."
The team gathered self-reported personal and family disease history data collected by the UK Biobank for 12 diseases, including diabetes, depression, Parkinson's disease, and hypertension, on 1.5 million people. They noted that disease prevalence was higher among participants' parents than among the participants themselves or their siblings, reflecting an age-related increase in disease burden. For each disease, the researchers calculated the relative risk to the parents, siblings, and partners of ill biobank participants.
These relative risk estimates for parents and siblings — which combined data from blood and adoptive relatives — were higher than the relative risk estimates for partners for all diseases except hypertension and lung cancer. This indicated to the researchers that a combination of genetic and shared environmental risk factors contribute to disease development for most diseases.
The researchers then turned to three approaches to estimate heritability. Using Falconer's method, Tenesa and his colleagues gauged heritability values for pairs of first-degree relatives, such as of a parent and a child or of siblings. They also determined values for adoptive parents and offspring and for adoptive sibling pairs. This approach does not take shared environmental factors among family members into account, the researchers noted.
After accounting for generational differences in disease prevalence, the researchers reported that the highest heritability value between parent and offspring was for severe depression and the highest heritability value between siblings was for prostate cancer.
Heritability value estimates for parent and offspring were lower than those for siblings for hypertension, diabetes, and prostate cancer, among others, which suggested to the researchers that there are non-additive genetic effects at play or that there are higher environmental similarities between siblings than between parents and children.
At the same time, Tenesa and his colleagues reported that the correlation between partners was high for hypertension, though low between adoptive parents and children. However, the adoptive parent-child regression coefficients were greater than zero for heart disease, chronic bronchitis, and breast cancer. Because of this, the researchers said there might be a number of environmental effects shared among family members. Thus, they argued that heritability estimates based on only blood relatives might not reflect the full complexity of shared environmental effects and could be inflated.
Tenesa and his colleagues also used a structural equation modeling (SEM) approach that accounts for shared familial environment to estimate heritability. These values, they reported, were typically lower than those generated using Falconer's method. Still, for most diseases, genetic effects were the major contributor to disease risk, but for some, such as hypertension, the effect of the shared familial environment was more important than the genetic effects.
Using simulated data, the researchers concluded that there was an overestimation of heritability for all 12 diseases they studied.
In addition, heritability estimates based on SNP data could explain an average of 44.2 percent of the Falconer's method estimates and between 44 percent and 57 percent of the SEM family-based heritability estimates.
For hypertension, though, SNP heritability explains just about all of the SEM heritability, indicating that for a disease where much of the familial environmental factors could be modeled, there might be very little missing heritability.
"The 12 diseases analyzed in this large cohort of individuals show significant but moderate values of heritability and an important impact of shared familial environmental effects and support the case for combining these factors with genetic marker information to improve the performance of disease risk prediction methods," Tenesa and his colleagues wrote.