Genome-Wide Association Studies in Obstructive Sleep Apnea. Will We Catch a Black Cat in a Dark Room?

Giant strides have been made during the last 3 decades in deciphering genetics of rare diseases such as cystic fibrosis or Huntington disease. These strides were based on linkage analysis in families affected by a rare disease, using genetic markers across the genome. However, the linkage analysis uniformly failed in complex diseases such as cancer, cardiovascular diseases, and obstructive sleep apnea.

The common disease–common variant hypothesis (1) argues that genetic susceptibility to common disease is a result of variations in the genome that occur at high frequency in the population, but have small effects on disease expression. Genome-wide association studies (GWASs) seek associations between common gene variants, usually single-nucleotide polymorphism (SNPs), and specific diseases. The first successful GWAS published in 2005 identified a common variant in the complement factor H gene as a potential cause of age-related macular degeneration (2). During the subsequent 8 years, thousands of SNPs were identified in nearly 2,000 published human GWASs (3), resulting in significant progress in understanding diabetes, obesity, and other highly prevalent complex diseases.

The GWAS design has been applied to rare and common sleep disorders. Studies in narcolepsy and restless leg syndrome identified genes conferring susceptibility and contributing to the pathogenesis of these disorders (4–6). In contrast to restless leg syndrome and narcolepsy, sleep disordered breathing (SDB) is a heterogeneous and complex disease. Multiple mechanisms including upper airway structural properties and neuromuscular control, ventilatory control, and sleep–wake regulation influence the overall expression of the disease (7). Nongenetic factors including obesity, age, and comorbid cardiopulmonary disease can significantly affect susceptibility to SDB. Moreover, SDB severity is multidimensional and can be characterized by type, frequency, and duration of apneic events; severity of gas exchange abnormalities; and presence of arousals. Given the complexity of SDB, prerequisites for a definitive genetics study would be a large number of research subjects and clearly defined phenotypes. Nevertheless, previous studies in the genetics of SDB were limited by small sample sizes and the reduction of SDB dimensions to the AHI (8, 9).

In this issue of the Journal, Cade and colleagues (pp. 886–897) present the first genome-wide SDB study that yielded SNPs reaching acceptable levels of significance (10). Furthermore, the results were independent of body mass index. The authors employed two methods to overcome limitations of existing studies in SDB genetics. First, they achieved a large sample size by pooling data from three separate cohorts. Second, in addition to the traditional apnea–hypopnea index, the authors defined SDB phenotypes by mean nocturnal oxyhemoglobin saturation and mean event duration. The apnea–hypopnea index was associated with a SNP in proximity of the GPR83 gene. GPR83 is expressed in the areas of the brain involved in upper airway neuromuscular and respiratory control, including hypoglossal nucleus, dorsal motor nucleus of vagus, and the nucleus of the solitary tract. GPR83 appears to be involved in regulation of metabolic rate and immune responses. Nevertheless, significance of this finding may be diminished by the opposite effect of this SNP on the apnea–hypopnea index in two pooled cohorts: MESA (the Multi-Ethnic Study of Atherosclerosis) and HCHS/SOL (Hispanic Community Health Study/Study of Latinos).

The most interesting findings were related to another SDB phenotype: the duration of the events. Event duration can be determined by multiple factors, including ventilatory control as well as the severity of hypercapnia and hypoxia during the events. The authors detected signals in the genome near two important pathways: hypoxia inducible factor-1α (HIF-1α) and sterol regulatory element binding protein 1 (SREBP1). HIF-1α is a master regulator of metabolic, respiratory, and cardiovascular responses to hypoxia. Polymorphism in ARRB1 (β-arrestin 1), a protein that can stabilize HIF-1α under hypoxic conditions, is particularly relevant to SDB. HIF-1α mediates hypoxic sensitivity in the carotid bodies, and therefore its up-regulation may affect ventilatory responses to hypoxemia and respiratory stability (11). SREBP1 is a key transcription factor of lipid biosynthesis (12). Intermittent hypoxia, a hallmark manifestation of sleep apnea, induces lipid biosynthesis and may lead to fat redistribution via SREBP1 (13, 14). Polymorphism in insulin-induced gene 2 and phospholipase C β2 may lead to overexpression of SREBP1, which could mediate central fat distribution and, hence, predisposition to SDB and reduced pulmonary reserves. During SDB events, low pulmonary reserves may result in severe gas exchange abnormalities leading to early termination of the SDB events (Figure 1). Thus, HIF-1α and SREBP1 are biologically plausible modifiers of SDB manifestations.

Figure 1.

Event duration and lung volumes determine functional phenotypes. Overexpression of sterol regulatory element binding protein 1 (SREBP1) can cause abdominal adiposity, leading to low lung volumes and severe hypoxemia during apneic events. Stabilization of hypoxia-inducible factor-1α (HIF-1α) in the carotid body during hypoxic events can activate ventilatory responses, shortening event duration and mitigating hypoxemia. As event duration decreases, sleep may become more fragmented.

The work presented by Cade and colleagues has several limitations (10). The authors pooled data from Hispanic populations from three cohorts, which used different recording and genotyping techniques. The recording techniques varied from full polysomnography to more abbreviated home sleep studies, which influenced the acquired phenotype. There was significant ancestral heterogeneity within and between cohorts, including subjects of Mexican, Caribbean, and Central American descent. Nevertheless, these techniques enabled the investigators to achieve a large sample size and consistent genetic findings across study cohorts, especially for event duration. These findings must be reproduced in large, independent cohorts of different ethnicities to be robust and generalizable.

The authors made positive strides in defining phenotypes, which was not previously performed in genetic studies of SDB. Nevertheless, their phenotypes do not sufficiently delineate disease pathogenesis or predict clinical outcomes. Specifically, apnea–hypopnea index, oxygen saturation as measured by pulse oximetry, and event duration are results of intricate interactions among upper airway anatomy, abdominal adiposity, neuromuscular control, and arousal responses. Moreover, the implications of event duration are unknown. Shortening of the event duration may lead to sleep fragmentation. Alternatively, prolongation of the events may exacerbate hypoxemia (Figure 1). Further investigations should lead to the development of robust, readily deployable phenotypes beyond the apnea–hypopnea index, focusing on physiological traits derived from sleep recordings (15) and clinical sequelae such as hypertension (16).

Finally, another limitation of this study, common to all GWASs, is that causality cannot be inferred. GWAS findings can serve as the basis to develop novel animal models of specific SDB phenotypes associated with particular genes. Investigators should employ gain and loss of function approaches focusing on genes candidates with the ultimate goal to reproduce the SDB phenotype in animal models. State-of-the-art tools such as tissue-specific and inducible gene knock-out and overexpression and targeted administration of genes using viral vectors can be deployed.

In conclusion, the work of Cade and colleagues opens new horizons in genetics of SDB (10). However, more work is required to find the black cat in a dark room. Investigators should strive to replicate GWAS findings in several large, independent cohorts; develop robust SDB phenotypes with insights into pathogenesis and clinical sequelae; and develop robust animal models to examine SDB phenotypes conferred by candidate genes.