Family Studies in Patients with the Sleep Apnea-Hypopnea Syndrome

Methods

Patients

We recruited consecutive new patients referred to the Edinburgh Sleep Laboratory who were found to have the sleep apnea–hypopnea syndrome and who had more than 15 apneas plus hypopneas per hour of sleep in association with at least two major symptoms of this syndrome. To avoid studying the familial pattern of obesity, patients were excluded from the study who had a body mass index greater than 30.0 kg/m². Patients with other identified predisposing factors (including gross retrognathia, hypothyroidism, acromegaly, or neuromuscular disorders) were excluded. None of the index patients was included in our previous investigation [9]. Each patient was asked to give details of all first-degree relatives aged 15 to 75 years, all of whom were then requested to participate in the study.

Control participants were subsequently recruited from an Edinburgh general practice register and were matched on a one-to-one basis for age, sex, height, and weight with each of the 51 relatives studied. Our a priori matching was for 50 pairs, however, because one relative did not complete a symptom questionnaire, an additional pair of participants was studied. All participants gave written informed consent for participation, and local ethics committee approval was obtained for the study.

Questionnaire

Both groups (relatives and controls) completed a sleep symptom questionnaire that included queries on date of birth, collar size, height, weight, alcohol consumption, and the presence or absence of sleep symptoms. Smoking, drug, family, and medical histories were also recorded. Bed partners, where available, were asked about participant's behavior during sleep.

Excessive daytime sleepiness was considered present if the participants averaged two naps per day. Sleeping against their will was considered present if the participants reported irresistible sleep attacks in embarrassing or dangerous situations. Loud snoring was considered present if the bed partner reported that on most nights snoring was of sufficient intensity to disturb others. Nocturnal choking was considered present if the participant woke up with choking or suffocating within the past month. Current alcohol consumption was calculated in units of alcohol per week, and current cigarette smoking was judged to be present if the participant reported smoking at least one cigarette a day during the previous month.

Sleep Study

The 51 relatives and 51 controls each had a one-night sleep study, during which airflow at the mouth and nostrils was recorded by thermocouples; thoracoabdominal movement was recorded by induction plethysmography; ear oxygen saturation was recorded by an Ohmeda Biox 3700 oximeter; and an electroencephalogram, electro-oculogram, and electromyogram were recorded on a 16-channel polygraph (Specialised Laboratory Equipment). Sleep [11] and breathing patterns [12] were manually scored using standard criteria. An arousal was defined as a 3- to 15-second period of return of α or theta waves with or without an increase in submental electromyographic measurements in non-rapid eye movement sleep or with an increase in electromyographic tone in rapid eye movement sleep [13].

Acoustic Reflectometry and Cephalometry

Upper airway areas were measured using an acoustic pulse reflectometer, as we have previously described [14]. All measurements were made through the mouth with the participant seated and at functional residual capacity. Lateral cephalometry was done, using a cephalostat and standard protocol, in both the groups from which standard bony and soft-tissue dimensions were measured [15].

Statistical Analyses

All data were scored by a single observer (RM) blinded to the participant's group. Data are reported as the mean ± SE when normally distributed and as the median and range when non-normally distributed. Differences between groups were assessed by the paired Student t-test, the Wilcoxon matched-pairs signed-rank test, or the chi-square test as appropriate, using SPS-PC [16]. Intrafamily correlation of cephalometric variables was assessed by multilevel analysis of variance [17].

Results

Participants

Ninety-four consecutive new patients with the sleep apnea–hypopnea syndrome were considered. Forty-five patients were rejected because their body mass index exceeded 30.0 kg/m², and 10 were rejected because of other predisposing factors. The remaining 39 eligible new patients with this syndrome were approached, and all provided information on their first-degree relatives. Five had no eligible relatives, whereas 2 families with a total of 3 eligible relatives refused participation. Thus, we tried to recruit all eligible relatives of 32 (31 men) patients in the study. These 32 index patients had a mean age of 51 years (range, 28 to 68 years), a mean body mass index of 26.5 kg/m² (range, 21.4 to 29.9 kg/m²), a mean apnea plus hypopnea frequency of 39/h (range, 16 to 105/h), and a collar size of 40 cm (range, 39 to 44 cm).

Of a total of 108 eligible first-degree relatives, 91 agreed to participate in the study and 17 refused or could not be contacted. All 91 relatives completed sleep questionnaires. Fifty-one relatives were randomly selected and matched one to one with control participants. No differences were noted between the matched and unmatched relatives in any of the sleep-related symptoms, but the 51 matched relatives were younger (36 years [SE, 2 years] compared with 44 years [SE, 3 years]; P = 0.02). The matched relatives consumed less alcohol (8 units/wk [SE, 1 unit/wk] compared with 12 units/wk [SE, 2 units/wk]; P = 0.04) than did the 40 unmatched relatives, and proportionately fewer matched relatives were current smokers (14 compared with 19; P = 0.02). Fifty-five controls were approached, and 51 controls agreed to participate. The characteristics of the two groups are given in Table 1. No difference was noted between relatives and controls in smoking habits, alcohol consumption, or collar size.

Sleep Questionnaire and Sleep Studies

Sleep symptom questionnaire data were available in 50 matched pairs. More relatives than controls reported loud snoring, daytime sleepiness, and nocturnal choking (Table 1).

When compared with controls (Figure 1), relatives had higher apnea plus hypopnea frequencies (median of 13/h for relatives compared with 4/h for controls; P < 0.001), and more relatives than controls had greater than 5, 10, and 15 apneas plus hypopneas per hour of sleep (P < 0.001) (Figure 1). The relatives also had more arousals per hour than did controls (median of 30/h for relatives compared with median of 17/h for controls; P < 0.001) (Figure 2), greater sleep disturbance with more light sleep (stages 1 plus 2, 209 min [SE, 7 min] for relatives compared with 179 min [SE, 8 min] for controls; P = 0.006), and less slow-wave sleep (stages 3 plus 4, 78 min [SE, 5 min] for relatives compared with 91 min [SE, 5 min] for controls; P = 0.03).

No significant differences were noted between groups in total time in bed, sleep period time, or total sleep time. Relatives had more 2% oxyhemoglobin desaturations per hour than did controls (6/h [SE, 1/h] for relatives compared with 3/h [SE, 1/h] for controls; P = 0.04). Relatives also had higher 3% desaturations per hour than did controls (4/h [SE, 1/h] for relatives compared with 2/h [SE, 1/h] for controls; P = 0.04), but no significant difference was noted in the frequency of 4% desaturations per hour between the relatives and controls (2/h [SE, 1/h] for relatives compared with 1/h [SE, 0.4/h] for controls; P = 0.1).

Acoustic Reflectometry

Relatives had decreased total pharyngeal volume (19 mL [SE, 1 mL] for relatives compared with 23 mL [SE, 1 mL] for controls; P = 0.01) and decreased glottic cross-sectional area (1.7 cm² [SE, 0.1 cm²] for relatives compared with 1.9 cm² [SE, 0.1 cm²] for controls; P = 0.05), but significant differences were not noted in oropharyngeal junctional, average, or maximal pharyngeal cross-sectional areas between the groups.

Cephalometry

Lateral cephalometric measurements (Table 2, see Figure 3 for location of features) were only available in 36 matched pairs because not all participants volunteered for these radiographs. The relatives had retroposed maxillae and mandibles with shorter mandibles and longer soft palates with wider uvulae (Table 2). No significant difference was noted in these cephalometric variables between generations (P = 0.13) or between different families with the sleep apnea–hypopnea syndrome (P > 0.3). Multilevel analysis of variance showed 80% intrafamily correlation of bony angle abnormalities, regardless of the generation level.

Discussion

Our study shows that familial factors affect development of the sleep apnea–hypopnea syndrome. It also shows that first-degree relatives of thin patients with this syndrome have smaller upper airways and a different craniofacial morphologic structure than do matched controls. Our data are consistent with a previously reported symptom questionnaire study [8] describing more daytime sleepiness and apneas in relatives of patients with the sleep apnea–hypopnea syndrome. Sleep studies were not done in that study, and the possibility of relatives being more aware of symptoms could not be excluded. Indeed, a reporting bias in our symptom questionnaire data cannot be excluded. However, this would not account for the objective differences in sleep study results in our study. Our results are compatible with the finding from a preliminary study that increased abnormal breathing occurs during sleep in families of patients with the sleep apnea–hypopnea syndrome [18].

The cephalometric analyses indicate that the relatives of thin patients with the sleep apnea–hypopnea syndrome have retroposed maxillae and mandibles, shorter mandibles, longer soft palates, and wider uvulae than do matched controls, changes that have previously been reported in patients with this syndrome [19, 20]. Patients with this syndrome have narrowed upper airways even in the awake state [21], and all of these abnormal cephalometric variables would predispose them to narrowing of the upper airway. Our study does not prove that differences in bone or soft-tissue structure cause the abnormal breathing during sleep. Although differences in soft palatal and uvular size may reflect increased fat deposition [22], they could be secondary to mucosal and muscular changes resulting from the recurrent upper airway occlusion. In addition, facial remodelling can occur as the result of upper airway anatomical changes in infancy, at least in monkeys [23]. However, the absence of a difference in cephalometric variables between generations and the high intrafamily correlation in cephalometric variables suggest that the bony structural changes probably represent the inherited defect in nonobese patients with this syndrome. A recent report [24] from Japan of an increased prevalence of HLA-A2 in patients with the sleep apnea–hypopnea syndrome would support the concept of an inherited defect.

One potential limitation of our study was that not all of the relatives of the eligible patients had sleep studies. The primary reason for this was the cost of the sleep studies. Each overnight sleep study costs $770 or $1540 per matched relative and control. Our initial power calculation based on our earlier study [9] indicated that 50 pairs were needed for the study. The 51 relatives studied were randomly selected from the 108 eligible relatives. Sleep questionnaire data were available in 91 of 108 (84%) relatives, and the 51 who had sleep studies were no different from the other 40 in terms of any sleep symptom; indeed, because they were younger and consumed less alcohol and fewer cigarettes, they might be expected to show less abnormal breathing during sleep than the other 40 [25, 26]. Further, we have now done sleep studies on 24 of these 40 relatives, and we found that their breathing pattern during sleep was similar to that of the 51 relatives studied, with a median of 11 apneas plus hypopneas per hour of sleep. A potential source of bias in case–control studies is noncomparability of cases and controls. We do not believe that this was a problem because our cases and controls were carefully matched for the important variables known to influence breathing during sleep and because cases and controls were drawn from the same population with the same ethnic mix. Thus, we are confident that our findings of significantly increased symptoms of the sleep apnea–hypopnea syndrome and of abnormal breathing during sleep in relatives of patients with this syndrome are real.

These results cannot be extrapolated to all patients with the sleep apnea–hypopnea syndrome. Many patients with the syndrome are obese, and, in early series, many patients were grossly overweight. In our first series, 85% of patients had a body mass index of more than 30.0 kg/m²[27], but this percentage has been falling progressively with greater recognition that this syndrome also occurs in nonobese patients. During the period of recruitment for this study, exactly 50% of the newly diagnosed patients with the sleep apnea–hypopnea syndrome had body mass indices less than 30.0 kg/m². Thus, the familial factors described in this study are applicable to a substantial proportion of patients with this syndrome. It is probable that in obese patients with this syndrome, facial structure differences may be less important than in the current study, although this remains to be tested. The precise mechanism that determines why one obese person develops the sleep apnea–hypopnea syndrome but another equally obese person does not develop the syndrome is unknown.