Familial Aggregation and Heritability of Obstructive Sleep Apnea Using Children Probands

Abstract

Study Objectives:

Previous studies suggest the presence of familial aggregation of obstructive sleep apnea (OSA) in adults. However, similar data on childhood OSA are limited. This family study aimed to investigate the heritability and familial aggregation of childhood OSA and to examine whether significant differences existed between patients of normal weight and overweight.

Methods:

Children aged 6 to 18 years were recruited as probands either from attendants to sleep clinic (with habitual snoring) or the community (without habitual snoring). Parents and siblings of the probands were also invited to participate. All participants underwent nocturnal sleep study.

Results:

A total of 229 probands took part, of whom 33 had moderate to severe OSA, 70 had mild disease, and 126 had no OSA. A total of 412 relatives were also recruited. Although the overall heritability of obstructive apnea-hypopnea index (OAHI) was not significant (h² ± SE = 0.03 ± 0.09, P = .37), it was significant in overweight individuals on subgroup analysis (h² ± SE = 0.43 ± 0.24, P = .032). Significant interaction effect of overweight was demonstrated in both heritability and familial aggregation analyses. Bivariate genetic analysis found that the genetic correlation between OAHI and body mass index in overweight individuals (ρ_g ± SE = 0.63 ± 0.18) was significantly different from both 0 (P = .005) and 1 (P = .025).

Conclusions:

The differential results of heritability and familial aggregation of OSA in normal weight and overweight subgroups substantiated the recommendation of separating childhood OSA into normal weight and overweight subtypes. In the overweight subgroup, there may be obesity-independent components involved in the genetic variance of OAHI, although a significant proportion of the genetic variance is shared with obesity.

Citation:

Au CT, Zhang J, Cheung JYF, Chan KCC, Wing YK, Li AM. Familial aggregation and heritability of obstructive sleep apnea using children probands. J Clin Sleep Med. 2019;15(11):1561–1570.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Familial aggregation of obstructive sleep apnea (OSA) has been demonstrated in adult studies but similar evidence for children is limited. It is well known that OSA in obese and nonobese children are different but the effect of obesity on familial aggregation of OSA and heritability of OSA severity index remains unclear. This family study aimed to investigate the heritability and familial aggregation of OSA among children and their first-degree relatives and to examine how obesity would modify the association.

Study Impact: The differential results between overweight and normal weight subgroups support that childhood OSA should be divided into normal weight and overweight subtypes and the normal weight subtype may be considered as a separate disease from adult OSA.

INTRODUCTION

Family studies using adult probands suggest the presence of familial aggregation in obstructive sleep apnea (OSA).^1–6 Studies on aggregation of childhood sleep-disordered breathing are scarce. A questionnaire-based study documents a higher prevalence of OSA-related symptoms in first-degree relatives of children with OSA.⁷ However, the study was limited by the lack of a control group and the OSA status and severity were not ascertained by polysomnography. More recently, a retrospective study that utilized a hospital diagnosis database demonstrated that children with at least one sibling with OSA are at a higher risk of having OSA.⁸ Another study using the same database showed that children with a parent affected by OSA have increased risk of development of the same disorder.⁹ Nonetheless, both studies were retrospective analyses of hospital records that tended to bias toward familial clustering of diseases. The Cleveland Family study documents that moderate OSA (apnea- hypopnea index > 10 events/h) is more prevalent in children with family history of OSA than those without (8.4% versus 1.6%).¹⁰ However, the study recruited adults as probands and the possible difference between parent-offspring and sibling-sibling associations of childhood OSA could not be examined. The Cleveland Family Study also suggested that children with family history of sleep-disordered breathing might have a higher risk of having residual OSA after adenotonsillectomy. The sample size, however, was so small (n = 60) that the association was not statistically significant in both unadjusted and adjusted analyses.¹¹

Because there are substantial differences between childhood and adult OSA in terms of etiology, effects on sleep architecture, behavioral outcomes, and daytime symptoms as well as treatment strategies,¹² it is inappropriate to extrapolate findings from adult studies directly to children. Therefore, it is worthwhile to investigate the possibility of genetic linkage between childhood and adult OSA, and whether disease aggregation existed between siblings.

Although childhood OSA is regarded as a heterogeneous condition with protean manifestations,¹³ two major subtypes based on patients’ characteristics have been hypothesized.¹⁴ The first subtype is associated with lymphoid tissue enlargement in the absence of obesity, whereas the other is primarily associated with obesity¹⁵ that is believed to be phenotypically more akin to the adult form of OSA.¹⁴ A previous study suggests that there is substantial overlap between the genetic basis of OSA and obesity,¹⁶ and that the genetic susceptibility of OSA has both obesity-dependent and obesity-independent components, emphasizing the importance of examining obese and nonobese individuals separately. From a nosologic perspective, the familial aggregation pattern of these two subtypes may serve as further support to differentiate childhood OSA into distinct groups.

This study aimed to address the following research questions: (1) Is OSA in children associated with OSA in their parents and/or siblings? (2) Is familial aggregation of OSA present in both those of normal weight and overweight? We hypothesized that first-degree relatives of children with OSA would have a higher risk of the development of OSA and that their disease state would be more severe, that is, higher obstructive apnea-hypopnea index (OAHI), than those of children without OSA. In addition, we speculated that there would be differences between overweight and normal weight subgroups in familial aggregation of OSA and heritability of OAHI, and that overweight had a significant modification effect on familial aggregation of OSA and heritability of OAHI.

METHODS

Study Population

This was a prospective case-control family study. Children and adolescents age 6 to 18 years were recruited as index subjects (probands). Those with habitual snoring and/or other symptoms suggestive of OSA identified from our sleep disorder clinic were recruited as suspected OSA case probands. Those without habitual snoring were recruited from a cohort established by a previous epidemiologic study, which assessed the relationship between sleep disturbance and neurohormonal dysregulation in a random sample of primary and secondary school students,¹⁷ to serve as suspected control probands. Probands were confirmed to be case or control according to their overnight polysomnography (PSG) results (see next paragraphs for definitions). The exclusion criteria for both case and control probands included previous treatment for OSA, genetic syndrome, congenital or acquired neuromuscular disease, obesity secondary to an underlying cause, and craniofacial abnormalities. All first-degree relatives (parents and siblings) of probands were invited to participate. History of habitual snoring of relatives who refused to take part in this study was also obtained to examine for possible response bias. Written consent and verbal assent were obtained from the parents and probands respectively. This study was approved by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee (CREC-2008.438).

Collection of Data

All participants underwent evaluation of clinical profile including completion of demographic and OSA questionnaire, anthropometric measurements, pubertal assessment, and overnight PSG (in-laboratory PSG for children and home PSG for adults). On the day of PSG, participants were admitted to our sleep laboratory in the afternoon. Overweight in children was defined as body mass index (BMI) ≥ 85th percentile (corresponding to a z score of 1.036) of the local reference,¹⁸ whereas for adult participants, overweight was defined as BMI ≥ 25 kg/m².

Polysomnography

A model Siesta ProFusion III PSG monitor (Compumedics Telemed, Abbotsford, Victoria, Australia) was used to record the following parameters: electroencephalogram (F4/A1, C4/A1, C3/A2, O2/A1), bilateral electrooculogram, electromyogram of mentalis activity, and bilateral anterior tibialis. Respiratory movements of the chest and abdomen were measured by inductance plethysmography. Electrocardiogram and heart rate were continuously recorded from two anterior chest leads. Arterial oxyhemoglobin saturation was measured by an oximeter with finger probe. Respiratory airflow pressure signal was measured via a nasal catheter placed at the anterior nares and connected to a pressure transducer. An oronasal thermal sensor was also used to detect the absence of airflow. Snoring was measured by a snoring microphone placed near the throat. Body position was monitored via a body position sensor.

For unattended home PSG, a portable, 16-channel sleep recorder (MediPalm, Braebon Medical Corporation, Canada) was used to collect data same as those recorded in the sleep laboratory. Data collected from either attended or unattended PSG were interpreted by a registered sleep technologist. Adult and pediatric PSG results were reviewed by two different technologists (JYC and CTA, respectively) who were blinded to family information. An adequate overnight PSG was defined as one with total sleep time longer than 6 hours.

Respiratory events including obstructive apneas, mixed apneas, central apneas, and hypopneas were scored based on the recommendations from The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications.¹⁹ Different scoring criteria were employed for pediatric and adult participants, in accordance with the manual. Specifically, a respiratory event was scored when it lasted 10 seconds or longer in adults (> 18 years), but for children (2–18 years), an event would be scored when it lasted two breaths or more irrespective of its duration. Central and obstructive hypopneas were separately scored. OAHI was defined as the total number of obstructive apneas, mixed apneas, and obstructive hypopneas per hour of sleep. The diagnostic criteria of OSA were age dependent. For children and adolescent (6–18 years), OSA was defined as OAHI ≥ 1 event/h. For adults, the diagnostic cutoff was OAHI ≥ 5 events/h. Moderate to severe OSA was defined as OAHI ≥ 5 events/h and ≥ 15 events/h for children and adults, respectively.

Statistical Analysis

Continuous data were presented as mean ± standard deviation unless otherwise specified. Categorical data were shown as number (proportion). Group comparisons were tested by one-way analysis of variance or t test, and chi-square test for continuous and categorical data respectively. Generalized estimating equation was used to test whether relatives of case probands had higher OAHI than relatives of control probands while adjusting for age, sex, pubertal status, and BMI. To examine the possible modification effect of overweight on the association between probands’ OSA and relatives’ OAHI, the interaction terms OSA_(proband) × overweight_(proband) and OSA_(proband) × overweight_(relatives) were also tested in the generalized estimating equation. We also performed analyses separately for father-child, mother-child, and sibling pairs. The statistical significance level was set as a value of P < .05. SPSS 20.0 for Windows (SPSS Inc., Chicago, Illinois, USA) was used for all tests except for heritability analysis.

Estimation of Heritability for Traits

The additive and narrow sense heritability of OAHI was calculated by restricted maximum likelihood methods in the Sequential Oligogenic Linkage Analysis Routines (SOLAR) program.²⁰ All analyses were adjusted for age, age-squared, age × sex, sex and BMI. This method aimed to distinguish the variance of specific phenotype accounted by genetic (σ_g²) and environmental (σ_e²) components, which consisted of common environmental factors, measurement errors, and nonadditive genetic factors.²⁰ The total variance of specific phenotype could be expressed as σ_p² = σ_g² + σ_e². The narrow sense heritability (h²) of total phenotypic variance was h² = σ_g²/σ_p².

Because variance components analysis is sensitive to outliers and nonnormal distributions, OAHI was inversely normal transformed using a rank-based procedure implemented by SOLAR before analysis to minimize the possible artifacts induced by skewed distributions.²¹ Previous study demonstrated that this transformation does not induce correlations between relatives or lead to inflated estimates of heritability.²²

To understand the association between obesity and OSA severity in families, bivariate genetic analysis was also performed to estimate the genetic (ρ_g) and environmental (ρ_e) contributions for the correlation between OAHI and BMI:

where h₁² and h₂² referred to the heritability of OAHI and BMI, respectively.

To further assess the modification effect of overweight on heritability of OAHI, we adopted the method suggested by Diego et al²³ It tested whether there was a significant difference in the genetic variance between overweight and normal weight subgroups, or if the genetic correlation between OAHI in overweight individuals and OAHI in normal weight individuals was significantly different from one (ie, if there were different sets of genes acting in overweight and normal weight subgroups), or both. In other words, there were two null hypotheses, namely:

where σ_{g (overweight)}² and σ_{g (normal weight)}² were the genetic variance in overweight and normal weight individuals, respectively, and ρ_{g (overweight, normal weight)} was the genetic correlation between OAHI in overweight and normal weight subgroups.

RESULTS

Proband Characteristics

A total of 229 probands, including 82 overweight and 147 normal weight children, were recruited, of whom 139 were recruited from the community whereas 90 were recruited from the clinic. All participants were ethnic Chinese and consanguinity was not present in this cohort. Each family is independent of each other. There were no significant differences in OAHI between clinic-based and community-based individuals within each OSA severity group (Table S1 in the supplemental material). As expected, overweight individuals had significantly higher OAHI and prevalence of OSA than the normal weight individuals (Table S2 in the supplemental material). A total of 103 probands were found to have OSA, of whom 33 had moderate to severe disease and 70 had mild disease. The remaining 126 probands had no OSA. A significantly greater proportion of male sex and higher BMI z score were found in those with OSA compared to their counterparts without OSA (Table 1).

Table 1.

Characteristics of probands and their relatives across different OSA severities of probands.

Characteristics of First-Degree Relatives

A total of 412 first-degree relatives, including 310 adults (132 overweight) and 102 children (32 overweight), were recruited. The overall response rate was 61.5% (412 out of a total of 670 first-degree relatives). The recruitment details are shown in Figure 1. Prevalence of habitual snoring among recruited relatives was similar to that of all relatives (Table 2). There were no significant differences in OAHI or prevalence of OSA or moderate to severe OSA between parents of case and control probands (Table 1). However, siblings of probands with moderate to severe OSA had significantly greater OAHI (P = .028) and a higher prevalence of OSA (P = .028) than those of probands with OAHI < 5 events/h (Table 1).

Figure 1. Flowchart of sample recruitment (probands and first-degree relatives).

The percentages in the blankets represented the response rates out of the total number of the corresponding first-degree relatives. The overall response rate was 61.5% (412 out of a total of 670 first-degree relatives). FDR = first-degree relatives, MS = moderate to severe, OSA = obstructive sleep apnea.

Table 2.

Proportion of relatives having habitual snoring.

Family Aggregation of OSA and the Effect of Overweight

The generalized estimating equation found that relatives of case probands did not have a significantly higher OAHI than relatives of control probands (estimated marginal mean: 4.0 events/h versus 3.7 events/h, P = .59). Further analysis found that relatives of probands with moderate to severe OSA had a higher OAHI than relatives of probands without moderate to severe OSA (estimated marginal mean: 5.0 events/h versus 3.5 events/h, P = .044) (Table S3 in the supplemental material). The interaction effects of probands’ overweight (P = .11) (Figure 2) and relatives’ overweight (P = .083) (Figure 3) on the association were both insignificant. Nevertheless, pairwise comparisons showed that among families of overweight probands, relatives of probands with moderate-to-severe OSA had a higher OAHI than relatives of probands with OAHI < 5 events/h (estimated marginal mean: 6.6 events/h versus 3.8 events/h, P = .022) (Figure 2). Similarly, among families with overweight relatives, relatives of probands with moderate to severe OSA also had a higher OAHI than relatives of probands with OAHI < 5 events/h (estimated marginal mean: 6.9 events/h versus 3.6 events/h, P = .029) (Figure 3). Similar findings were not observed in families of normal weight probands and normal weight relatives (Table S3).

Figure 2. Illustration of the interaction effect between probands’ moderate to severe obstructive sleep apnea and probands’ overweight on relatives’ obstructive apnea hypopnea index.

EMM = estimated marginal mean, OAHI = obstructive apnea-hypopnea index.

Figure 3. Illustration of the interaction effect between probands’ moderate to severe obstructive sleep apnea and relatives’ overweight on relatives’ obstructive apnea hypopnea index.

EMM = estimated marginal mean, OAHI = obstructive apnea-hypopnea index.

Subgroup analysis for father-child pairs showed that the interaction effect of probands’ overweight on the association between probands’ moderate-to-severe OSA and fathers’ OAHI was significant (P = .025) (Figure 2). Pairwise comparison demonstrated that fathers of overweight probands with moderate-to-severe OSA tended to have a higher OAHI than fathers of overweight probands with OAHI < 5 events/h (estimated marginal mean: 18.1 events/h versus 9.9 events/h, P = .060), whereas no significant difference was found between fathers of normal weight probands with and without moderate-to-severe OSA (estimated marginal mean: 8.4 events/h versus 12.2 events/h, P = .22) (Figure 2 and Table S4 in the supplemental material).

Subgroup analysis for mother-child pairs found that the main effect of probands’ moderate to severe OSA on mothers’ OAHI was significant. Mothers of children with moderate to severe OSA had a higher OAHI than those of children with OAHI < 5 events/h (estimated marginal mean: 4.4 events/h versus 1.8 events/h P = .006) (Table S5 in supplemental material). In addition, the effect of the interaction term mothers’ overweight × probands’ moderate-to-severe OSA on mothers’ OAHI was also significant (P = .016). Pairwise comparison showed that overweight mothers of probands with moderate-to-severe OSA had a higher OAHI than overweight mothers of probands with OAHI < 5 events/h (estimated marginal mean: 9.8 events/h versus 2.1 events/h, P = .005), whereas no significant difference was found between normal- weight mothers of probands with and without moderate to severe OSA (estimated marginal mean: 1.7 events/h versus 1.5 events/h, P = .62) (Figure 3 and Table S5). However, the interaction effect of probands’ overweight was insignificant (P = .32) (Figure 2 and Table S5).

Subgroup analysis for sibling pairs revealed that the main effect of probands’ moderate-to-severe OSA on sibling’s OAHI was not significant (P = .082) (Table S6 in the supplemental material). The effect of the interaction term probands’ overweight × probands’ moderate to severe OSA on siblings’ OAHI were also insignificant (P = .115). Nevertheless, pairwise analysis showed that siblings of overweight probands with moderate to severe OSA had a higher OAHI than siblings of overweight probands with OAHI < 5 events/h (estimated marginal mean: 5.0 events/h versus 1.2 events/h, P = .047), whereas no significant difference was found between siblings of normal weight probands with and without moderate to severe OSA (estimated marginal mean: 1.2 events/h versus 0.9 events/h, P = .69) (Figure 2 and Table S6). The interaction effect of siblings’ overweight on the association between probands’ moderate to severe OSA and siblings’ OAHI was insignificant (P = .40). The small sample size (n = 4) in the group of overweight siblings of probands with OAHI ≥ 5 events/h should be noted (Figure 3 and Table S6).

Heritability of OAHI and Effect of Overweight

Post hoc power calculations revealed that the sample sizes of overall (n = 641), normal weight (n = 395) and overweight subgroups (n = 246) provided a power of 80% to detect a heritability of 0.21, 0.32 and 0.52, respectively, with a type I error of 5% (Figure S1 in the supplemental material). Heritability of OAHI was not significant in overall analysis (h² ± SE = 0.03 ± 0.09, P = .37). Subgroup analysis revealed that heritability of OAHI was only significant in overweight (h² ± SE = 0.43 ± 0.24, P = .032) but not in normal weight individuals (h² = 0) (Table 3). Results from interaction analysis showed that the genetic standard deviation of OAHI in overweight individuals was significantly different from that of the normal weight individuals (0.49 versus 0.02, P = .041), so that the null hypothesis (1) σ_{g (overweight)}² = σ_{g (normal weight)}² can be rejected, suggesting a significant modification effect of overweight on the heritability of OAHI. However, the genetic correlation of OAHI in overweight and OAHI in normal weight subgroups was not significantly different from one, meaning that the null hypothesis (2) ρ_{g (overweight, normal weight)} = 1 cannot be rejected.

Table 3.

Heritability of obstructive apnea-hypopnea index and body mass index and bivariate analysis of obstructive apnea-hypopnea index and body mass index.

Bivariate analysis found that the genetic correlation between OAHI and BMI was not significant in normal weight subgroup (P = .12) and overall analysis (P = .24). In overweight individuals, the genetic correlation (ρ_g = 0.63 ± 0.18) was significantly different from both 0 (P = .005) and 1 (P = .025) (Table 3).

DISCUSSION

Familial aggregation was documented in families of children with moderate to severe OSA where first-degree relatives of probands with moderate to severe OSA had significantly higher OSA severity than those of probands with mild or no OSA. The significant findings associated with OAHI ≥ 5 events/h but not OAHI ≥ 1 events/h suggested that the genetic effect could only be revealed in families of children with more severe disease, whereas environmental factors might play a more important part in families of children with mild OSA. Furthermore, interaction analysis revealed that overweight significantly modulated the familial aggregation pattern.

The heritability of OAHI was only significant in overweight individuals. Genetic variance of OAHI among overweight individuals was also found to be significantly greater than that among normal weight individuals, suggesting a significant modification effect of overweight on the heritability of OAHI. However, we failed to reject the null hypothesis of ρ_{g (overweight, normal weight)} = 1, that is, the genetic correlation between OAHI in overweight and normal weight subgroups being 1. Therefore, the differential findings in overweight and normal weight subgroups could not be entirely accounted for by the presence of different sets of genes interacting in each subgroup. Instead, the difference may be explained by either a stronger genetic component in the overweight subgroup, or a stronger environmental component in the normal weight subgroup, or both.

Additional findings were obtained from analysis of father-proband, mother-proband, and sibling pairs. Results from father-proband pairs found that the effect of probands’ moderate to severe OSA was only significant in pairs with overweight probands but not pairs with normal weight probands. In contrast, fathers’ overweight did not significantly modify the association between probands’ moderate to severe OSA and fathers’ OAHI. This can partly be attributed to sampling bias. The prevalence of OSA in all fathers was 75% (117 of 157) and up to 66% (49 of 74) in normal weight fathers, both were much higher than the general population prevalence. The differential effects of overweight on OSA were diluted by including a sizeable proportion of OSA fathers who were normal weight. Participant recruitment was entirely voluntary, individuals with symptoms suggestive of OSA were more willing to participate. As prevalence of OSA in adult males is higher than adult females and children, the probability of a family taking part in this study would very much depend on the father’s decision. This may explain the high OSA prevalence found in fathers but not in mothers or the siblings.

The main effect of probands’ moderate to severe OSA was only significant in mother-proband pairs but not father-probands and sibling pairs. Moreover, the effect of probands’ moderate to severe OSA on relatives’ OAHI in families of normal weight probands was only significant in mother-proband pairs (Figure 2). This suggested that the maternal component for familial aggregation of OSA may be stronger such that the effect could also be seen in the normal weight subgroup. This is consistent with a previous study that showed that familial transmission of snoring was maternally mediated.²⁴ Another study also showed that only maternal but not paternal history of adenotonsillectomy was associated with habitual snoring in children.²⁵

For proband-sibling pairs, although the interaction effect of probands’ overweight and siblings’ overweight were both insignificant, pairwise comparisons suggested that the effect of probands’ moderate to severe OSA was more apparent in pairs involving overweight probands or overweight siblings, implying that sibling aggregation was only significant in the overweight subgroup. Compared to parent-child aggregation, sibling aggregation was more significant in unadjusted analyses, but the superiority disappeared in adjusted analyses. The possible stronger sibling aggregation may be attributed to common underlying pathophysiology in childhood OSA. Sibling aggregation of childhood OSA has been reported by only one other study, which found that siblings of individuals with OSA have a much higher chance of development of the disease than those of children without OSA.⁸ Parents of a child with OSA are likely to be more aware of the related symptoms and thus, more ready to seek advice and evaluation if other offspring display similar symptoms. A study suggests that significant sibling aggregation of OSA is also present in adults,²⁶ likely mediated by sibling aggregations of upper airway dimensions,²⁷ upper airway soft tissue size,²⁸ and craniofacial structures,²⁹ predisposing a higher sibling risk for OSA.

The finding of the significant modification effect of overweight on the heritability and familial aggregation of childhood OSA provided further support to separate childhood OSA into normal weight and overweight subtypes.¹⁴ The pathophysiology of OSA in overweight children is more similar to that in adults, whereas that in normal weight children is more unique and specific to children, as adenotonsillar hypertrophy is much less prevalent in adults. In addition, our differential results of OAHI heritability between overweight and normal weight subgroups suggested that the variance of OAHI in overweight could be explained significantly by a genetic component, whereas an environmental component would play a more important role in normal weight children. Despite this, heritability of OSA could not be totally dismissed in normal weight children because our study design did not allow us to accurately document whether parents had adenotonsillar hypertrophy and/or OSA in their childhood. Previous studies on the natural history of childhood OSA showed that a substantial proportion of children with OSA would have spontaneous resolution with time.^30–32 Therefore, an adult without OSA at the time of assessment could have had the condition as a child.

Bivariate genetic analysis revealed that the genetic correlation between OAHI and BMI was significant in overweight individuals. This suggested the presence of genes contributing to both obesity and OSA among the overweight individuals. At the same time, it was also demonstrated that the genetic correlation was significantly different from 1, implicating that there may be obesity independent components involved in the genetic variance of OAHI, although a substantial proportion of the genetic variance is shared with obesity. Previous studies have demonstrated significant heritability of upper airway dimensions,²⁷ size of the upper airway soft tissues,²⁸ craniofacial structures,²⁹ and respiratory arousal threshold.³³ An earlier study also found that young adult asymptomatic offspring of fathers in whom OSA was diagnosed had impaired respiratory response to externally applied inspiratory resistive loads during non-rapid eye movement sleep compared to offspring of unaffected parents, indicating that muscle responsiveness and ventilatory control might be inherited.³⁴ These are all probable risk factors involving in the nonobesity-related pathway leading to the development of OSA.

Our study had several limitations. First, the sample size was not sufficient for more detailed subgroup analyses to investigate the sex differences in the familial aggregation pattern, to see whether the familial aggregation is maternal mediated or paternal mediated, and to test the possible three-way interaction between probands’ OSA, probands’ overweight, and relatives’ overweight on relatives’ OAHI. The study was also unable to investigate the effects of puberty as only 29% of pediatric participants were in puberty at the time of assessment. The heritability of OAHI was 0 and 0.09 ± 0.19 (P = .32) for the prepubertal and pubertal subgroups, respectively. Second, as previously mentioned, there was a probable sampling bias toward families with members who were more likely to have OSA, especially for fathers. Nevertheless, subgroup analysis for mothers, who were believed to be subjected to less severe sampling bias, also demonstrated significant association of probands’ and mothers’ OSA. Last, our study was a cross-sectional study. It will be interesting to examine whether children without OSA whose father and/or mother have the condition would have a higher risk for incident OSA when they become adults. Longitudinal follow-up of the current cohort would provide this important information. Despite these limitations, our study used a prospective study design that is scientifically more robust than family history in identifying familial aggregation of a disorder by directly examining the available probands and their relatives using overnight PSG as an objective measurement of OSA severity.

In conclusion, this family study showed that familial aggregation of OSA and heritability of OAHI was significant in overweight individuals, and that overweight had a significant modification effect on the familial aggregation and the heritability. Overweight children with OSA have been shown to have a higher risk of residual disease after adenotonsillectomy³⁵ and a higher risk of persistent disease or disease worsening in their adulthood as compared to counterparts of normal weight,^31,36 suggesting overweight and normal weight patients may need different treatment strategies. Our findings from this family study substantiate the idea of categorizing childhood OSA into normal weight and overweight subtypes. The findings also support that the normal weight subtype may be considered as a different disorder from adult OSA. In addition, the results from the bivariate genetic analysis also implicated that both obesity-related and obesity-independent components may be involved in the genetic variance of OAHI in overweight patients.

DISCLOSURE STATEMENT

Work for this study was performed at The Chinese University of Hong Kong. This work was supported by the Research Grants Council of the Hong Kong Special Administrative Region, China [CUHK471210]. The authors report no conflicts of interest.

ABBREVIATIONS

BMI

body mass index

OAHI

obstructive apnea-hypopnea index

OSA

obstructive sleep apnea

PSG

polysomnography

SOLAR

Sequential Oligogenic Linkage Analysis Routines