(215) Genetic modifiers of nutritional status in CF are both unique to CF and shared with the general population

Background: Nutritional status is a key outcome in cystic fibrosis (CF) with undernutrition predicting worse lung disease and earlier mortality. Maintaining body mass index (BMI) above 50^th percentile is recommended for optimal health in CF. On the other hand, obesity is becoming more frequent in the CF population but does not benefit lung function measures. Nutritional status is dependent on treatments, environmental factors, and CFTR function. In the general population, BMI is determined in part by genetic variants. While CFTR genotype is a primary determinant of nutritional status in CF, variation in BMI is largely determined by genes other than CFTR (genetic modifiers). Understanding the genetic contributions to BMI in CF can identify CFTR-independent targets contributing to undernutrition or obesity in CF. We report an initial genome-wide association (GWAS) study for BMI on >4000 individuals with CF in the CF Genome Project (CFGP).

Methods: Common ( >0.5% MAF) variants passing QC were provided by the CFGP. Longitudinal height and weight, demographics, and covariates were ascertained from medical records and the CFF Patient Registry 2017 (CFGP data set). BMI z-score was calculated using CDC reference equations. The trait used in the analysis (AvgBMIz) was derived by calculating the average of average per-quarter BMI z-scores. Data were excluded if they were from the year of death, after the first organ transplantation and after modulator approval. GWAS used linear mixed modeling with fixed effect of study site, first 4 principal components, birth cohort, and pancreatic sufficiency (PI-PS) status. Relatedness was modeled as a random effect.

Results: AvgBMIz was correlated with study site, birth cohort, PI-PS status/CFTR genotype severity, and gastrostomy-tube (G-tube) use. In an analysis restricted to pancreatic insufficient (PI) individuals (n=4101), genome-wide significant association was observed for variants in the FTO gene (rs28567725, P= 3.36e-8; MAF = 0.41) and ADAMTS5 (rs162500, P=4.92e-10; MAF = 0.006). The relatively low frequency ADAMTS5 variant had high effect size (β=-0.7). When including PI and PS individuals (n=4409), the FTO variant was associated (P=4.52e-08) while the significance of the ADAMTS5 variant decreased (P=2.35e-7). To assess how genetic influences on BMI in the general population affect BMI in CF, a polygenic risk score (PRS) was calculated using 941 variants identified in studies of non-CF individuals; this PRS was strongly associated with AvgBMIz while explaining but a fraction of the trait variance (P=4.9e-30; r2=0.03) indicating that other genetic influences remain to be delineated.

Conclusions: The variant in FTO, known to be associated with obesity in the general population, was significantly associated with increased BMI in CF. Other genetic contributors in general population also affect BMI in CF, despite substantial differences in nutritional needs and prevalence of obesity. ADAMTS5 variants are not known to be associated with BMI in the general population, but functional studies suggest a role in liver integrity in a mouse model of high-fat diet-induced obesity. This raises the possibility of a link between nonalcoholic steatohepatitis and diet-induced obesity in CF. In addition, these results suggest that interventions for low and high BMI derived in non-CF populations may be effective in CF. Supported by CFF CUTTIN18XX1, BAMSHA18XX0, and KNOWLE18XX0.

Acknowledgements: This work was supported by CFF [grant numbers CUTTIN18XX1, BAMSHA18XX0, and KNOWLE18XX0]

References: