Newly developed risk outcomes can help identify the risk of childhood obesity

The newly developed risk outcomes synthesize genetic information into an easy-to-interpret indicator that can help clinicians identify young children who are most at risk of developing obesity.

The study, led by Penn State researchers, used new statistical methods to establish evaluation criteria using data collected from young children. The study also shows that stable results are achievable from studies that are an order of magnitude smaller than typical genetic studies when comprehensive data are collected over time and used in conjunction with powerful statistical tools.

About 18% of children in the United States are obese and 6% are severely obese. If we can identify the most at-risk children, we could prevent the development of obesity in the first place. In this study, we created risk outcomes based on genetic information that clinicians could potentially use to identify young children who would benefit most from intervention strategies. “

Sarah Craig, Assistant Professor of Biology, Penn State

This study is part of a larger project called INSIGHT (Interventional Nurses Begin Babies to Grow on Healthy Paths), coordinated through Milton S. Hershey Penn State Health Medical Center, where researchers and clinicians work together to identify biological and social risk factors for obesity and the impact of responsive parental interventions in the early life of the child. The research team collected longitudinal data -; periodically 8 times between birth and three years of age -; including weight, height and behavioral and environmental variables – of nearly 300 children. They also collected a blood sample for genetic analysis from each of the children, which served as a basis for preparing the risk results. The team published its results in an article published in the journal Econometrics and Statistics.

Risk results -; called “polygenic risk outcomes” because they are based on many genetic sites in the genome -; distills vast genetic information in an easy-to-understand number. Typically, the results include information from a number of single nucleotide polymorphisms (SNPs) or locations in the genome where individual letters of the DNA alphabet may vary in people who are most associated with the metrics of interest -; in this case growth rates and obesity.

“Previous attempts to create polygenic risk outcomes for obesity have been developed using genetic information from adults or older children and include between 100 and two million SNPs,” said Katerina Makova, a professor of biology and life sciences at the Department of Life Sciences. Verne M. Willaman in Penn. condition. “Such high numbers are challenging and potentially costly for consistent reproduction, especially in a clinical setting. We have created two evaluation options with much less SNP – one with 24 and one with 5 – which can nevertheless provide valuable information to researchers and clinicians. “

The research team used new statistical techniques from an area called functional data analysis to identify SNPs most associated with obesity, which were then included in the results.

“Unlike many genetic studies that collect data for one measurement, such as body mass index – BMI and at one point in time, we took advantage of the longitudinal data collected over time,” said Francesca Chiaromonte, a professor of statistics. and Huck’s Department of Life Sciences Statistics at Penn State. “Several measurements of weight and height over time give a growth curve for each child and we can analyze the shapes of the curves for children in our cohort using functional data analysis. We took advantage of this richer data at every step of the analysis. “

Genetic data yields millions of SNPs to be analyzed, and the team used several techniques to narrow the group down to SNPs, most closely related to growth curves and obesity measures.

“We first assessed the impact of each SNP individually on obesity-related measures as a way to eliminate those that are clearly unrelated,” said Anna Kenny, a Penn State statistics student at the time of the study and now a postdoctoral fellow. researcher at the University of California, Berkeley. “Some studies have decided to stop at this step, but we have narrowed the scope even further by looking at all other SNPs at once and eliminating those that did not seem to have an impact when considered with others.”

This process yielded 24 SNPs, which the researchers included in a polygenic risk assessment. The results based on the growth curves turned out to be related to other, more frequently used measures; they are higher in children with higher conditional weight gain-; change in weight gain in the first 6 months-; and rapid weight gain in infants -; predictor of obesity later in life.

The research team further narrows the group to five of the most stable SNPs -; SNPs that had the greatest impact, even when they disturbed the data. Of these five SNPS, they created a second result that can be used as a simpler alternative.

“Although the result of 24 SNPs is more powerful than the result of 5 SNPs, we confirmed that both are useful measures for the risk of obesity and we believe that each can be used in a clinical environment,” said Matthew Reimher, Associate Professor of statistics in Penn State. “A result that requires the introduction of less SNP should make it easier to produce in clinics.”

In particular, the results of this study also predict obesity in older children and adults, which the research team confirmed using publicly available datasets. However, results from other studies based on information on adult obesity are not translated into young children in this study.

“This suggests that the genetic signals associated with obesity that we see in early childhood are crucial throughout life,” said Ian Paul, a professor of pediatrics and public health at Penn State College of Medicine. “However, with age, people begin to show other parts of their genetic makeup. Results based on early signals appear to be more stable throughout a person’s life. This underscores the need for more research that focuses on identifying risk and preventing obesity in young children, especially during the “first 1,000 days” covering pregnancy and the first two years after birth. “

The study also shows that smaller studies that deeply characterize individuals and take advantage of functional data analysis techniques can be a powerful alternative to typical large-scale genetic research.

“These techniques can open doors to smaller laboratories with fewer resources,” Craig said. “By working carefully and rigorously to collect longitudinal data from more targeted cohorts and using powerful statistical techniques, you can still find useful information with a survey that is an order of magnitude less than typical GWAS surveys.

In addition to Craig, Makova, Chiaromonte, Kenny, Reimher, and Paul, the research team includes Junley Lin, a research associate at Penn State at the time of the study; Leann Birch, the late professor of food and nutrition at the University of Georgia, who helped lead INSIGHT; Jennifer Savage, director of the Center for Childhood Obesity Research and an associate professor of food science at Penn State; and Michele Marini, research technologist and statistician at the Center for the Study of Childhood Obesity in Penn State.


Reference in the magazine:

Craig, SJC, et al. (2021) Building a polygenic risk of childhood obesity using functional data analysis. Econometrics and statistics.

Source link