Discussion
This manuscript reports polygenic effects of multiple genetic variants as well as modulating effects of environmental exposures on risk of recurrent wheeze and asthma during early childhood. First, we determined GRS for recurrent wheeze and asthma based on four genetic variants located on chromosome 17q12-21. These loci were previously associated with childhood-onset asthma in the largest published GWAS to date.(15) In addition, we observed that breastfeeding until one year reduced prevalence of recurrent wheeze, especially among children with moderate or low genetic risk but not among those with high genetic risk. This suggests that breastfeeding at one year of life reduces risk of recurrent wheeze but does not offset high genetic risk. In contrast, NO2 exposure appeared to increase the risk of respiratory outcomes, particularly among children with moderate or high genetic risk but did not affect those in the low GRS group. Overall, our findings support the notion that asthma is a polygenic disease that is modulated by interactions with environmental exposures, which could account in part for the missing heritability. Moreover, our results indicate that while asthma may be difficult to diagnose in preschool-aged children, risk of respiratory outcomes can be detected during early infancy using recurrent wheeze between ages 2-5 years.
Our GWAS of recurrent wheeze identified 98 associated SNVs, albeit all are in strong linkage disequilibrium (LD, r2> 0.8), suggesting that they represent a single associated locus. Of these, 19 have been previously correlated with asthma or a related trait (e.g., allergy, atopic march, hay fever and eczema) and only two are coding. For example, SNV rs2305479 is a missense variant located within the GSDMB gene, resulting in an amino acid substitution (Glycine to Arginine at position 304/416 of the Gasdermin-B protein), and has been previously implicated in inflammatory processes. An earlier study reported that rs2305479 leads to abnormal sulfatide transport, which might compromise the integrity of the epithelial cell barrier and promote inflammatory processes in inflammatory bowel disease (IBD).(37) In addition, rs11078928 codes for a splice acceptor, which could result in an alternative mRNA isoform. Further experiments are needed to determine the biological effects of these genetic variants as well as others in the same associated LD block.
In addition to SNVs associated with recurrent wheeze on chromosome 17, which is a well-established locus for childhood asthma, we identified that SNV rs145454327 on chromosome 1 was also associated with asthma by age 5 years.(15) This intronic variant was previously associated with childhood-onset asthma in the largest GWAS to date.(15) Furthermore, chromatin state model analysis of fetal lung tissue identified that this variant could potentially alter the motif for 4 transcription factors, GATA-4, Tbp, FOXP1, and Sox8. GATA-4 have been previously implicated with pathogenesis in bronchial asthma (38) and FOXP1 is known to regulate epithelial cell fate and regeneration during lung development.(39)
While our GWAS of recurrent wheeze and asthma both identified associations with loci previously associated with asthma, there was no overlap of associated loci. These results may reflect the phenotypic heterogeneity of asthma as well as challenges in clinical diagnosis among young children before age 7 (when spirometry is difficult and the use of methacholine challenge tests are not typical).(16,17) In addition, we combined definite and possible asthma into one case group for GWAS, which may have further affected the accuracy of asthma diagnosis by age 5. However, the association of SNPs on chromosome 17q12-21, a well-replicated locus for childhood-onset asthma, with recurrent wheeze in this study indicates that asthma risk may be assessed among young children, even prior to the diagnosis of asthma.
In addition to individual SNV associations with recurrent wheeze and asthma, we identified the polygenic effects of four genetic variants correlated with both recurrent wheeze (p = 1.53e-08) and asthma diagnosis (p=9.39e-08). While the four variants identified in the GRS analysis are non-coding, located either within intronic and 3’UTRs, these have been previously identified to have potentially regulatory effects. For example, the three risk variants associated with increased prevalence of recurrent wheeze (rs3816470, rs8076131, and rs12603332) are known loci that alter the expression of genes (eQTLs), as reported by the Genotype-Tissue Expression (GTEx) project. Similarly, Tonchevaet al. reported that the protective SNV rs3902920 alters the expression of ORMDL3 , a well-established asthma locus.(40) All three risk variants identified in the GRS are correlated with increased expression of ORMDL3/GSDMB in lung cells and immune cells (leukocytes) while the protective variant is correlated with decreased expression of the same genes. Studies have found that mice expressing increased levels of human ORMDL3 or GSDMB have an asthma phenotype characterized by increased airway responsiveness and increased airway remodeling in the absence of airway inflammation.(41) This suggests that the four variants used in calculating GRS in this study may affect wheeze and asthma by regulating the expression of these well-established asthma genes.
It is note-worthy that while GWAS of asthma did not yield associations with loci on chromosome 17, our GRS analysis indicates that SNVs on this chromosome may be used to assess risk of asthma as well as recurrent wheeze. This may be explained by the modest effects of each individual SNV on chromosome 17, which could not be detected in an univariate analysis (i.e. GWAS) of single SNVs but the additive effects of these variants could be detected in our GRS analysis.
In addition to polygenic effects of multiple genetic variants, we determined that breastfeeding until 12 months and NO2exposure during the first 6 months of life significantly interact with genetic factors to modulate risk of recurrent wheeze during early childhood. Specifically, we observed lower prevalence of recurrent wheeze among infants who were breastfed at 12 months compared to those who were not breastfed until 12 months. The protective effects of breastfeeding on wheeze in the CHILD study was previously reported by Azad et al. (42) In the current study, this protective effect was detected in children with modest or low genetic risk but not in the high GRS group, which suggests that high genetic risk offsets the protective effects of breastfeeding at one year of life. In contrast to breastfeeding, we detected higher prevalence of recurrent wheeze among infants exposed to higher NO2 in the first 6 months of life. The detrimental effects on NO2 on recurrent wheezing during the first two years of life in the CHILD study was previously reported.(43) However, our study shows that this effect is most evident among children with high genetic risk, whereby those exposed to high NO2 have a 2.7-fold higher wheeze prevalence than those exposed to low NO2.
This GRS study is the first to use results from the largest GWAS of childhood-onset asthma published to date, which reported distinct association signals between childhood vs. adult-onset disease, suggesting different mechanisms of disease.(15) Prior GRS studies used summary statistics from GWAS of both adult- and childhood-onset asthma. Thus, our study uniquely focused on early-onset disease associations, resulting in an associated GRS based on four genetic variations on chromosome 17, which was the most significant peak from the childhood-onset asthma GWAS. We observed more than a 2-fold higher prevalence of recurrent wheeze and asthma between the high vs. low GRS groups. Moreover, compared to single associated SNVs, the GRS accounts for higher heritability estimates of recurrent wheeze and asthma (h2 = 2% vs. 1.4%). In fact, the GRS based on the additive effects of risk alleles at these four loci accounts for a similar heritability as 31 of the most replicated asthma variants (h2 = 2.5%). While a heritability estimate of 2% may seem low, it was estimated that all common variants account for up to 14% of asthma heritability.(5) Furthermore, we demonstrated that genetic risk interacts with environment exposures, which accounts for some of the missing heritability of asthma. In fact, this interaction accounts for up to 3.6% of the heritability of recurrent wheeze, which is an increase from considering the GRS alone.
While this manuscript reports a novel GRS of asthma and recurrent wheeze that is based on additive effects of childhood-onset asthma genes and identifies novel interactions with environmental exposures, our study has several limitations. First, while our GRS analysis utilized results from an independent GWAS of childhood asthma, the largest to date, replication of our GRS results in an independent asthma cohort is needed to validate our findings. A major challenge for replication is the identification of another cohort with both genetics and environmental exposures data in order to assess the effects of gene-environment interactions on childhood asthma. A second limitation is that our results are in silico -based and experimental studies are needed to test the functional impact, if any, of the identified variants on chromosomes 1 and 17. Given the large LD block on chromosome 17, which consists of numerous SNVs spanning multiple genes, identification of casual variants will be challenging. Finally, we have yet to assess additional environmental exposures as well as the long-term effects (i.e. beyond the first few year of life) on respiratory outcomes among the children as they grow. For example, methacholine challenge tests are ongoing in the CHILD Cohort Study at the age 8 clinical visits, which is ongoing and will result in more accurate diagnosis of asthma.