Introduction: Obstructive sleep apnea (OSA) is a common disorder associated with increased risk of cardiovascular diseases and mortality. Recent observational admixture mapping analysis and gene enrichment analysis linked genes and pathways associated with heme metabolism to OSA traits, but the causal contribution is unclear. Iron and heme metabolism is associated with OSA comorbidities and carotid body ventilatory control mechanisms. In this study, we use gene expression data to examine whether iron and heme related pathways have a causal effect on OSA.
Methods: We performed Mendelian randomization (MR) analyses considering the expression level of 15 candidate Gene Ontology pathways as exposures and four OSA traits as outcomes, including the apnea hypopnea index (AHI), and three nocturnal oxygen saturation measurements. In discovery analysis, we performed two-sample MR using local expression trait loci (cis-eQTLs) from the Genotype-Tissue Expression (GTEx) portal and published genome-wide association summary statistics for OSA traits (N>19,000). Significant pathways were then followed-up by one-sample MR using high coverage DNA and RNA sequencing data from the Multi-Ethnic Study of Atherosclerosis generated by the NHLBI Trans-Omics for Precision Medicine (TOPMed) project.
Results: Discovery analysis identified putative causal associations between up-regulated heme biosynthetic process pathway on increased AHI and overnight hypoxemia (minimum P-value=0.018 across OSA traits). These associations were supported in European and Hispanic/Latino Americans but not in African Americans in replication analysis, consistent with prior ancestry-specific associations between a heme-related gene and OSA traits.

Conclusion: This study suggested a causal association between heme biosynthetic processes and OSA severity, suggesting novel biomarkers and possibly treatment targets. Future work is needed to identify the mechanisms for this association and to exclude reverse causality.

Support: This study was supported by Sleep Research Society Foundation Career
Development Award 018-JP-18 to H.W. and NHLBI R35HL135818 to S.R. Molecular data for the TOPMed MESA (phs001416) was supported by NHLBI.

RATIONALE
Current guidelines do not sufficiently capture the heterogenous nature of asthma leading to suboptimal management and treatment strategies. A more comprehensive classification of asthma into biologically meaningful subgroups is needed. Given its position on the central biological dogma, as the ‘ome closest to phenotype, reflecting genetics, environment and their interactions, metabolomics represents a novel and compelling approach to accurately identifying asthma endotypes; i.e. subtypes defined by their functional or pathobiological mechanisms, with the potential for clinical translation.
METHODS
We performed plasma metabolomic profiling of 1155 asthmatic children from the Genetics of Asthma in Costa Rica Study (GACRS) across four profiling platforms covering a broad range of the metabolome. We generated patient similarity networks for each platform that connected asthmatics via edges representing patient-to-patient similarity in their metabolomic profiles (controlling for age, sex and body mass index). We then fused the four platform-specific networks using Similarity Network Fusion and performed spectral clustering on the resulting fused similarity network to identify metabo-endotypes. We explored phenotypic and clinical differences across the metabo-endotypes using ANOVA and chi-square tests. For validation we recapitulated the endotypes in an independent population of asthmatic children (CAMP, n=911) to determine whether the same between-endotype differences were observed. Finally, we identified metabolomic and genomic drivers of validated metabo-endotype membership, meta-analyzing findings from GACRS and CAMP.
RESULTS
We identified five metabo-endotypes in GACRS and observed significant differences across metabo-endotypes in asthma-relevant phenotypes pre-bronchodilator (p-ANOVA=8.3x10-5) and post-bronchodilator (p-ANOVA=1.8x10-5) FEV1/FVC ratio; use of inhaled (p-ANOVA=5.0x10-13) and oral (p-ANOVA=0.007) steroids, and airway hyperresponsiveness to methacholine (AHR PC20<16.8, p-ANOVA=8.4x10-7)p-ANOVA=8.4x10-7). Furthermore, there was a significant difference in eosinophil count (p-ANOVA=0.009). The recapitulated metabo-endotypes in CAMP displayed significant differences (p<0.05) in many of the same phenotypes. (Table 1). The “most-severe” asthma endotype was defined by the lowest FEV1/FVC ratio, and the highest prevalence of AHR and oral steroid usage. It was characterized by higher levels of glycerolipids and lysophospholipids, as well as lower levels of nucleotides relative to the other endotypes. These findings suggest dysregulation of pulmonary surfactant homeostasis in the “most-severe” endotype and were supported by the genetic analysis, which found that members of this endotype were more likely to carry variants in key pulmonary surfactant regulation genes such as BACH3 (meta-analyzed p=5.2x10-4) and BMP3 (meta-analyzed p=8.5x10-4).
CONCLUSIONS
These findings demonstrate that clinically meaningful endotypes can be derived and validated using metabolomic data, and that interrogating the drivers of these metabo-endotypes can help understand their pathophysiology and identify therapeutic targets.