"CRANIOFACIAL AND OBESITY DIFFERENCES IN OSA IN CAUCASIAN AND CHINESE PATIENTS Differences in Craniofacial Structures and Obesity in Caucasian and ChinesePatients with Obstructive Sleep Apnea 1 2 3 4 4 Richard W. W. Lee, MD, PhD ; Sivabalan Vasudavan, BDSc, MDSc, MPH ; David S. Hui, MB, BS, MD ; Tania Prvan, PhD ; Peter Petocz, PhD ;2 1 M. Ali Darendeliler, PhD ; Peter A. Cistulli, MD, PhD1 Centre for Sleep Health and Research, Department of Respiratory Medicine, Royal North Shore Hospital; Woolcock Institute of Medical Research; and2 3 University of Sydney, NSW, Australia; Discipline of Orthodontics, Sydney Dental Hospital, University of Sydney, Australia; Division of Respiratory4 Medicine, Department of Medicine & Therapeutics, The Chinese University of Hong Kong, Hong Kong; Department of Statistics, MacquarieUniversity, NSW, Australia Study Objectives: To explore differences in craniofacial structures and obesity between Caucasian and Chinese patients with obstructive sleepapnea (OSA). Design: Inter-ethnic comparison study. Setting: Two sleep disorder clinics in Australia and Hong Kong. Patients: 150 patients with OSA (74 Caucasian, 76 Chinese). Interventions: Anthropometry, cephalometry, and polysomnography were performed and compared. Subgroup analyses after matching for: (1)body mass index (BMI); (2) OSA severity. Measurements and Results: The mean age and BMI were similar between the ethnic groups. Chinese patients had more severe OSA (AHI35.3 vs 25.2 events/h, P = 0.005). They also had more craniofacial bony restriction, including a shorter cranial base (63.6 ± 3.3 vs 77.5 ± 6.7mm, P < 0.001), maxilla (50.7 ± 3.7 vs 58.8 ± 4.3 mm, P < 0.001) and mandible length (65.4 ± 4.2 vs 77.9 ± 9.4 mm, P < 0.001). These fndingsremained after correction for differences in body height. Similar results were shown in the BMI-matched analysis (n = 66). When matched for OSA2 severity (n = 52), Chinese patients had more craniofacial bony restriction, but Caucasian patients were more overweight (BMI 30.7 vs 28.4 kg/m ,P = 0.03) and had a larger neck circumference (40.8 vs 39.1 cm, P = 0.004); however, the ratios of BMI to the mandible or maxilla size were similar. Conclusions: Craniofacial factors and obesity contribute differentially to OSA in Caucasian and Chinese patients. For the same degree of OSAseverity, Caucasians were more overweight, whereas Chinese exhibited more craniofacial bony restriction. Keywords: Obstructive sleep apnea, ethnicity, craniofacial, obesity Citation: Lee RWW; Vasudavan S; Hui DS; Prvan T; Petocz P; Darendeliler MA; Cistulli PA. Differences in craniofacial structures and obesity inCaucasian and Chinese patients with obstructive sleep apnea. SLEEP 2010;33(8):1075-1080. 9-11 OBSTRUCTIVE SLEEP APNEA (OSA) IS A COMMON of OSA risk in Asian cohorts. Similarly, studies evaluatingDISORDER IN WHICH RECURRENT SLEEP RELAT- the role of obesity suggest that whilst it is a consistent risk fac- ED UPPER AIRWAY OBSTRUCTION CAUSES SLEEP tor for OSA across ethnic groups, the risk attributable to obesity3,6 fragmentation and intermittent hypoxemia. Ethnicity has been differs between racial groups. Despite suffering from a similarsuggested as an important risk factor for OSA. However, de- degree of OSA, patients from Asian groups are generally lessfning the role of ethnicity is complex, as it incorporates ge - overweight compared to their Caucasian counterparts, suggest- netic, environmental, and cultural factors that can, individually ing that ethnicity may differentially infuence the attributionor in combination, infuence the other recognized risk factors of these OSA risk factors. To date, studies that have exploredfor OSA. Whilst earlier prevalence studies suggest that OSA these complex interactions remain limited, especially in a di- 1,2 may be more common in certain ethnic groups, recent data in rect inter-ethnic comparison. Hence the aim of this study was3 4 5 Hong Kong Chinese, Indians and Koreans suggest that OSA to explore the differences in craniofacial structures and obesityprevalence in these ethnic groups are not dissimilar to the Cau- between Caucasian and Chinese patients with OSA. 6 casian populations. Obesity and craniofacial factors are well recognized in the MATERIALS AND METHODS pathogenesis of OSA. It is also becoming clear that the bal- ance and interaction between these two risk factors is crucial Subjects 7,8 in the development of OSA. It is likely that such interaction Patients in both ethnic groups were referred initially for theis strongly infuenced by ethnicity. Cephalometric studies have investigation of symptoms suggestive of OSA (snoring, day- suggested that craniofacial factors are important determinants time sleepiness, and/or witnessed apneas) in specialist sleepdisorders clinics. The Caucasian subjects were OSA patients re- cruited from a tertiary referral sleep disorders center in Austra- lia. This center has a longstanding strong clinical and researchSubmitted for publication September, 2009 interest in oral appliance therapy, and offers this treatment asSubmitted in fnal revised form December, 2009 a frst-line alternative to other treatment modalities for OSA.Accepted for publication February, 2010 The patients were recruited on the basis of having had cepha- Address correspondence to: Peter A. Cistulli MD, PhD, Centre for Sleeplometry performed as part of the routine clinical assessmentHealth and Research, Department of Respiratory Medicine, Level 8, Mainfor oral appliance therapy. The Chinese patients were recruitedBlock, Pacifc Highway, St Leonards, NSW 2065, Australia; Tel: +61 29926 8674; Fax: +61 2 9906 6391; E-mail: [email protected] from a center in Hong Kong, and were part of a previous cepha- SLEEP, Vol. 33, No. 8, 2010 Ethnic Difference in OSA Risk Factors—Lee et al 107512 lometric study. Ethnicity was determined by self-report and t-tests. In addition, subgroup analyses with 2 separate match- was confrmed by the study investigators. All patients had OSA ing procedures were performed from the recruited Caucasian2 as determined by polysomnography (apnea-hypopnea Index and Chinese patients: (1) matched for BMI [ ± 1 kg/m ]; (2)[AHI] = 5 /h). matched for OSA severity (AHI [ ± 10 events/h] and MinSaO2 [± 10%]). These were performed by selecting pairs of patients,Polysomnography one from each ethnic group, with similar BMI or OSA severity,Diagnostic polysomnography (PSG) was performed in ac- within the ranges described above. This allowed comparison13,14 cordance with previous studies and recommendations. of subgroups of Caucasian and Chinese patients that did not15 Sleep staging was determined using standardized defnitions. differ in overall BMI or OSA severity, while still maintainingRespiratory variables included chest wall and abdominal a reasonable subgroup sample size after the matching. Allow- movement, diaphragm EMG, nasal airfow and pressure, and ing for the multiple comparisons, a P-value of less than 0.002oxygen saturation by pulse oximetry. Defnitions of all scored was considered signifcant for the cephalometric measurementsevents were consistent at both study sites. Apnea was defned (Bonferroni adjustment for 27 variables). For P-values be- as a cessation of breathing during sleep = 10 sec, with oxygen tween 0.002 and 0.005, these were considered marginally sig- desaturation > 3% and/or associated with an arousal. Hypopnea nifcant. Comparisons between bony and soft tissue dimensionswas defned as = 50% reduction in airfow, associated with an were also made after correction of per meter of subject bodyarousal and/or = 3% reduction in blood oxygen saturation, last- height. Regression modeling was also performed on the entireing = 10 sec. The AHI was calculated as the average number of sample of 150 patients. The natural logarithm of AHI [lnAHI]episodes of apnea plus hypopnea per hour of sleep. Minimum was used as the response variable, since the transformed val- oxygen saturation (MinSaO ) was also measured. Scoring was ues allowed the regression assumptions to be better satisfed.2 performed by experienced sleep technologists. Variables examined included 27 cephalometric measurements,age, BMI, neck circumference, and ethnicity. Initial analysis in- Cephalometric Examinations volved constructing a multiple linear regression model using aA lateral cephalometric radiograph was taken for each sub- stepwise approach to select potential explanatory variables for16 ject at both sites, according to previously described methods. AHI. Subsequent exploratory analysis involved plotting lnA- Radiographs were taken at end-expiration, with the head in the HI against each of the potential predictors, with ethnicity as anatural position. This was achieved by asking the subjects to grouping variable, to identify signifcant two-way interactionslook into their own pupils refected in a mirror located at eye between ethnicity and the cephalometric or obesity variables.level. Cephalometric analysis was performed in a standard- Finally, linear models were constructed including ethnicity andized fashion to examine anatomical variables previously used its interactions in order to examine the differential impact of the13 for OSA investigation. The landmarks and measurements 2 ethnic groups on OSA severity. obtained are shown in Figure 1. All cephalograms were handtraced by a single investigator (SV). Where a bilateral landmark RESULTS presented as 2 images, the average of the 2 was used, with a One hundred ffty patients with OSA (74 Caucasian, 76 Hongtemplate of the bilateral structures employed to ensure accurate Kong Chinese) were recruited for this study. Baseline charac- reproduction. The cephalograms were scanned and digitized teristics of the comparison groups are presented in Table 1. Theusing custom analysis software (Dolphin Imaging Program, two ethnic groups had similar mean age, gender proportions,Premium Version 10.0 CA, USA). The mean enlargement fac- neck circumference and BMI, but the Chinese patients had moretor for the cephalograms was 11% for the Caucasian group and severe OSA (mean AHI 35.3 ± 26.1 vs 25.2 ± 16.3 events/hr,13% for the Chinese group. While this did not affect the angular P = 0.005; MinSaO 75.5 ± 14.7 vs 85.3 ± 6.2%, P < 0.001).2 measurements, the linear measurements were transformed by The Chinese patients also had more craniofacial bony restrictionthe respective enlargement factors using the imaging software (Table 1). Specifcally, they had a shorter cranial base [SN] (63.6to allow direct comparison. ± 3.3 vs 77.5 ± 6.7 mm, P < 0.001), midface length [Co-A] (82.7Measurement error was assessed in 20 randomly selected ± 4.7 vs 96.2 ± 8.0 mm, P < 0.001), maxilla [ANS-PNS] (50.7 ±patients from both groups, whose radiographs were re-traced, 3.7 vs 58.8 ± 4.3 mm, P < 0.001) and mandible [Go-Me] (65.4re-digitized, and re-measured by the same examiner under the ± 4.2 vs 77.9 ± 9.4 mm, P < 0.001). These cephalometric mea- same conditions one month later. The coeffcient of variation surements remained shorter in the Chinese patients after correc- was calculated. tion for differences in body height. The hyoid position appearedlower relative to the mandibular plane [MP-H] in the Cauca- Anthropometric Measurements sian patients (22.4 ± 6.7 vs 19.0 ± 6.0 mm, P = 0.001), but thisThe anthropometric measurements of obesity included neck difference was no longer signifcant when corrected for bodycircumference (cm), height (m) and weight (kg), enabling cal- height. Other cephalometric skeletal differences are summarized2 culation of the body mass index (BMI) (kg/m ). in Table 1; these include a greater SNA, ANB, Y-Axis and SN- PP angle, and a smaller gonial angle in the Chinese patients. TheStatistical Analyses Chinese patients had a relatively smaller airway space (PAS),Data were analyzed using a statistical package (SPSS for and their soft palate and tongue size were also smaller (Table 1). Windows Versions 14, Chicago IL, USA). Comparisons of the In the BMI-matched subgroup analysis, 66 pairs of Cau- cephalometric, anthropometric, and polysomnographic data casian and Chinese patients were compared. The mean BMI2 2 between 2 ethnic groups were performed using the unpaired was 29.7 ± 4.7 kg/m in the Caucasians and 29.6 ± 4.6 kg/mSLEEP, Vol. 33, No. 8, 2010 Ethnic Difference in OSA Risk Factors—Lee et al 1076N S CoPNS ANSBaA PhwsptC2Ht T P LAFH GoB PASHGnMeEbC4Figure 1—Defnitions of cephalometric landmarks and measurements Anatomical Landmarks: ANS (Anterior nasal spine)—tip of the median sharp bony process of the palatine bone in the hard palate. A Point—deepestmidline point on the maxillary alveolus between ANS and the maxillary alveolar crest. B Point—deepest midline point between the mandibular alveolar crestand the gnathion. Ba (Basion)—most inferior point on the anterior margin of the foramen magnum in the median plane. Go (Gonion)—most lateral externalpoint at the junction of the horizontal and ascending rami of the mandible. Gn (Gnathion)—most antero-inferior point on the bony mandibular symphysis. H(Hyoidale)—most antero-superior point on the body of the hyoid bone. Me (Mentum)—lowest point on the bony outline of the mandibular symphysis. MP(Mandibular plane)—line joining Me and Go. N (Nasion)—most anterior point of the fronto-nasal suture. PNS (Posterior nasal spine)—tip of the posteriorspine of the palatine bone of the hard palate. spt (soft palate tangent)—tangent point on a line parallel to the long axis of the soft palate at the maximumwidth. Phw (Posterior pharyngeal wall)—point on the posterior pharyngeal wall at the same horizontal level as spt. S (Sella)—the center of the sella turcica. Bony Dimensions: SN—anterior cranial base length. LAFH—lower anterior face height (ANS-Me). AFH—anterior face height (N-Me). PFH—posterior faceheight (S-Go). Go-Me—mandibular length. ANS-PNS—maxillary length. Co-A—midface length. Overjet—horizontal distance between the upper and lowercentral incisors measured parallel to the occlusal plane. It is measured from the labioincisal edge of one upper central incisor to the labial surface of thecorresponding lower central incisor with the upper and lower teeth in centric occlusion. Overbite—vertical distance between the incisal edge of the uppercentral incisor and the incisal edge of the lower central incisor. MP-H—perpendicular distance from the MP to H. Soft Tissue Dimensions: RPAS—width ofnasopharynx (Phw-spt). PAS—distance between the posterior pharyngeal wall and the dorsal surface of the base of the tongue, measured on the line thatintersects Go and B point. PNS-P—posterior nasal spine to the tip of the soft palate. Mx Soft Palate—maximal soft palate thickness. Tongue Length—lengthof the tongue. Tongue Height—maximal height of the tongue. Angular Measurements: BaSN—cranial base angulation in the mid-sagittal plane. SNA— angle from S to N to A Point. SNB—angle from S to N to B Point. ANB—angle from A Point to N to B Point. Y-Axis—facial axis (GnSN). Gonial Angle—angleformed by the posterior border of the mandible and the mandibular plane. CVT-SN—angulation of the cervical spine (C2-C4) with the cranial base (SN).SN-PP—angulation of the cranial base (SN) with the palatal plane. SN-OP—angulation of the cranial base (SN) with the occlusal plane. SN-MP—angulationof the cranial base (SN) with the mandibular plane. PP-MP—angulation of the palatal plane with the mandibular plane. in the Chinese patients. There were equal numbers of males In the OSA severity-matched subgroup analysis, 52 pairsin each group. Similar to the comparison with the entire pa- of Caucasian and Chinese patients were compared (Table 2).tient cohort, the Chinese patients continued to have more se- The mean AHI and MinSaO were similar between the 2 eth- 2 vere OSA when matched for BMI (mean AHI 33.8 ± 24.5 vs nic groups. When matched for OSA severity, the Caucasians2 24.9 ± 15.8 events/h, P = 0.02; MinSaO 76.2 ± 13.7 vs 85.3 ± were more overweight (BMI 30.7 ± 57 kg/m vs 28.4 ± 4.32 2 6.3%, P < 0.001). They also had more craniofacial bony restric- kg/m , P = 0.03) and had larger neck circumference (40.8 ±tion for the same degree of obesity (data not shown). 3.5 cm vs 39.1 ± 2.6 cm, P = 0.004). The Chinese patientsSLEEP, Vol. 33, No. 8, 2010 Ethnic Difference in OSA Risk Factors—Lee et al 1077casians and the OSA severity matched Chinese patientsTable 1—Baseline patient demographics and cephalometric characteristics (Table 2). Caucasian Chinese Mean PThe explanatory model for OSA severity was developed(n = 74) (n = 76) difference value using the entire cohort of 150 patients and it can be describedAge (years) 48.5 ± 11.0 49.5 ± 11.0 - NS by the following equations (developed from the one model,containing signifcant terms for interaction of ethnicity withGender (% male) 79.7 82.9 - NS BMI, SN-OP, and PNS-P): Neck circumference (cm) 40.9 ± 3.4 40.1 ± 3.5 0.7 ± 0.6 NS Weight (kg) 92.9 ± 17.0 80.1 ± 16.4 12.8 ± 2.7 < 0.001 Caucasian Height (m) 1.74 ± 0.08 1.64 ± 0.08 0.10 ± 0.01 < 0.001 lnAHI = 2.61 – 0.050*Co-A + 0.045*ANS-PNS +2 BMI (kg/m ) 30.8 ± 5.7 29.7 ± 5.2 1.0 ± 0.9 NS 0.028*PFH + 0.011*BMI – 0.0022*SN-OP – 0.0035*PNS-P. AHI (events/hour) 25.2 ± 16.3 35.3 ± 26.1 -10.1 ± 3.6 0.005 Chinese MinSaO (%) 85.3 ± 6.2 75.5 ± 14.7 9.7 ± 1.9 < 0.001 2lnAHI = 0.50 – 0.050*Co-A \u0007 + 0.045*ANS-PNS +Cephalometric bony dimensions (mm) 0.028*PFH + 0.041*BMI + 0.012*SN-OP + 0.024*PNS-P. SN* 77.5 ± 6.7 63.6 ± 3.3 13.9 ± 0.9 < 0.001 LAFH 75.1 ± 8.8 70.8 ± 5.5 4.3 ± 1.2 0.001 This model suggests that Co-A, ANS-PNS, PFH, andAFH 131.7 ± 12.3 122.3 ± 7.1 9.4 ± 1.6 < 0.001 BMI were independent predictors for OSA severity in bothPFH 87.3 ± 9.9 84.3 ± 6.5 3.0 ± 1.4 NS Caucasian and Chinese. However, a one-unit increase inGo-Me* 77.9 ± 9.4 65.4 ± 4.2 12.4 ± 1.2 < 0.001 BMI resulted in a larger positive contribution to lnAHI forANS-PNS* 58.8 ± 4.3 50.7 ± 3.7 8.2 ± 0.7 < 0.001 the Chinese patients (coeffcient for BMI was 0.041 com - Co-A* 96.2 ± 8.0 82.7 ± 4.7 13.4 ± 1.1 < 0.001 pared to 0.011 for Caucasians). The SN-OP and PNS-P alsoOverbite† 2.6 ± 3.6 0.9 ± 2.8 1.7 ± 0.5 0.002 contributed positively to OSA severity in the Chinese, butOverjet 4.2 ± 2.4 3.9 ± 2.4 0.3 ± 0.4 NS not in the Caucasians. Overall, the model had an adjusted2 MP-H 22.4 ± 6.7 19.0 ± 6.0 3.5 ± 1.0 0.001 r of 0.39. The coeffcient of variation was < 5% for all the cepha - Cephalometric soft tissue dimensions (mm) lometric variables on repeated measures in a subset of pa- RPAS 10.6 ± 4.0 8.8 ± 2.6 1.8 ± 0.5 0.002 tients. PAS† 10.8 ± 4.5 8.2 ± 3.6 2.6 ± 0.7 < 0.001 PNS-P* 40.0 ± 4.3 33.8 ± 4.8 6.2 ± 0.7 < 0.001 DISCUSSION Mx Soft palate* 11.4 ± 1.8 9.6 ± 1.5 1.8 ± 0.3 < 0.001 This study aimed to explore the differences in craniofa- Tongue length* 89.4 ± 7.6 67.3 ± 11.2 22.1 ± 1.6 < 0.001 cial structures and obesity as risk factors for OSA in Cau- Tongue height* 40.0 ± 5.0 31.3 ± 4.3 8.7 ± 0.8 < 0.001 casian and Chinese patients. We found that craniofacialCephalometric angular measurements ( ) ° structures and obesity contributed differentially to OSA inBaSN 126.7 ± 5.9 128.2 ± 6.1 -1.5 ± 1.0 NS these two ethnic groups. For the same degree of obesity,SNA 82.3 ± 3.3 84.5 ± 4.6 -2.2 ± 0.7 0.001 Chinese patients had more severe OSA and more craniofa- SNB 79.3 ± 3.6 79.5 ± 4.5 -0.2 ± 0.7 NS cial bony restriction. When OSA severity was similar, Cau- ANB 3.0 ± 3.3 5.1 ± 3.2 -2.1 ± 0.5 < 0.001 casian patients were more overweight and had larger neckY-Axis 67.1 ± 4.6 71.6 ± 4.8 -4.4 ± 0.8 < 0.001 circumference, whereas the Chinese patients exhibited moreGonial angle 126.5 ± 6.0 120.6 ± 8.2 5.9 ± 1.2 0.002 craniofacial bony restriction. CVT-SN 108.7 ± 7.8 111.6 ± 8.7 -3.0 ± 1.3 NS Epidemiological evidence strongly supports obesity asbeing one of the most important risk factors for OSA acrossSN-PP 8.3 ± 3.8 10.2 ± 4.6 -1.9 ± 0.7 NS 3,4,6 many ethnic groups. Despite similar disease prevalence,SN-OP 14.5 ± 9.5 18.3 ± 10.9 -3.8 ± 1.7 NS Chinese are less obese compared to communities from Aus- SN-MP 30.8 ± 7.0 30.6 ± 7.5 0.1 ± 1.2 NS 3,6,17-19 tralia and the United States. Our data demonstratedPP-MP 25.1 ± 6.7 24.0 ± 6.6 1.1 ± 1.1 NS that on average Chinese patients with OSA had a BMI 2.32 kg/m (~7.5%) lower than Caucasians with the same degreeSee Figure 1 for defnitions of measurements. Data are presented as mean ±of OSA severity. When BMI was similar, however, ChineseSD. Mean difference values are presented as mean ± SE. BMI, Body masspatients had a mean AHI higher by 8.9 events/h in our co- index; AHI, apnea-hypopnea index; MinSaO , minimum oxygen saturation;2 NS, non-signifcant. Measurements remained signifcantly shorter (*P < 0.001, horts and suffered from more severe oxygen desaturation.†P < 0.01) in Chinese patients after the correction of per meter of body height. These fndings are consistent with data from other inter-eth - 19,20 nic studies. Furthermore, our data also suggest that forevery unit of BMI increment, it results in a greater increasehad more craniofacial bony restriction (shorter cranial base, in OSA severity in the Chinese. These results suggest that themidface length, maxilla and mandible, etc.), a smaller airway impact of obesity on OSA is greater in the Chinese populations.space (PAS), and smaller soft palate and tongue size (Table 2). However, epidemiological data indicate that the odds of havingWhen comparing the ratio of the BMI to the height-adjusted OSA for each standard deviation of BMI increment is in factmandible size (BMI: Go-Me/Height) or maxilla size (BMI: lower in Chinese patients compared to the Wisconsin cohort3,6 ANS-PNS/height), there was no difference between Cau- (odds ratio 2.42 versus 4.19). Therefore, at a population level,SLEEP, Vol. 33, No. 8, 2010 Ethnic Difference in OSA Risk Factors—Lee et al 1078the risk of OSA attributable to obesity seems lower in ChineseTable 2—Subgroup analysis after matching for OSA severity (AHI andpatients, likely due to the lower prevalence of higher BMI lev- MinSaO ) 2 els. While in clinical OSA cohorts such as in this study, withCaucasian Chinese Phigher prevalence of obesity and craniofacial bony restriction,(n = 52) (n = 52) value the impact of BMI on OSA severity seems more pronounced inAHI (events/hour) 23.4 ± 14.4 23.5 ± 17.3 - Chinese patients. It has also been postulated that craniofacial factors, pharyn- MinSaO (%) 84.5 ± 6.2 83.3 ± 6.7 - 2 geal narrowing, and collapsibility assume greater pathogenicAge (years) 47.8 ± 11.0 50.5 ± 11.1 NS 3,21 signifcance in Chinese subjects. In our study, Chinese pa- Gender (% male) 78.8% 78.8% NS tients clearly had evidence of craniofacial bony restriction thatNeck circumferenceis more marked compared to Caucasians. Most importantly,(cm) 40.8 ± 3.5 39.1 ± 2.6 0.004 this is taking into account differences in body height, unlike10,19 Weight (kg) 92.0 ± 16.8 75.0 ± 12.4 < 0.001 previous inter-ethnic comparison studies. The fndings in - clude maxillary and mandibular restriction, shorter midface Height (m) 1.73 ± 0.09 1.63 ± 0.08 < 0.001 2 and cranial base, in addition to a smaller airway space. Overall,BMI (kg/m ) 30.7 ± 5.7 28.4 ± 4.3 0.03 while these fndings have been reported in cephalometric stud - Cephalometric measurements§ 12,22-24 ies within various ethnic groups, comparative studies be- SN / height* 44.2 ± 2.6 39.2 ± 2.6 < 0.001 tween Caucasian and Chinese patients remain very limited. OurGo-Me / height* 44.4 ± 3.6 40.3 ± 3.0 < 0.001 study further explored the impact of ethnicity on OSA severityANS-PNS / height* 33.6 ± 2.4 31.2 ± 2.5 < 0.001 with statistical modeling. Similar to many previous cephalo- Co-A / height* 55.0 ± 3.7 51.2 ± 2.9 < 0.001 24-26 metric studies in OSA, craniofacial structures and BMI wereMPH / height* 12.6 ± 2.7 11.2 ± 3.5 NS independent predictors of OSA severity in both ethnic groups.PAS / height* 6.5 ± 2.6 4.5 ± 1.6 < 0.001 But in the Chinese patients, the soft palate length (PNS-P) andPNS-P / height* 23.2 ± 2.7 20.2 ± 2.8 < 0.001 the angulation between the cranial base with the occlusal planeMx soft palate / height* 6.6 ± 1.0 5.8 ± 0.9 < 0.001 (SN-OP) had additional positive contribution to OSA severity.Tongue length / height* 51.4 ± 4.2 40.2 ± 6.9 < 0.001 The former suggests that upper airway soft tissues may contrib- Tongue height / height* 23.1 ± 2.7 19.1 ± 2.5 < 0.001 ute to airway collapse more so in Chinese patients, especially inSNA† 82.7 ± 3.2 85.3 ± 4.3 0.001 the presence of craniofacial bony restriction. The higher SN-OPANB† 3.1 ± 3.0 5.4 ± 2.7 < 0.001 could refect a steeper cranial base, which is seen in Chinese10 patients with OSA. Y-axis† 66.8 ± 4.8 71.1 ± 4.8 < 0.001 The balance between the relative size of the craniofacial Gonial angle† 126.3 ± 5.7 120.9 ± 8.2 < 0.001 bony compartment and the amount of upper airway soft tissueRatios of obesity to mandibular and maxillary dimensions or degree of obesity is an important determinant of upper air- BMI : Go-Me/height 0.91 ± 0.17 0.91 ± 0.14 NS 7,8,25 way size and OSA risk. This concept is supported by ourBMI : ANS-PNS/height 0.69 ± 0.12 0.71 ± 0.11 NS fnding of the similar ratios of obesity to mandibular and maxil - lary dimensions in the OSA severity- matched Caucasian andSee Figure 1 for defnitions of measurements. All data are presented asChinese patients. This anatomical balance model also explainsmean ± SD. NS – non-signifcant. §Cephalometric measurements thatthe greater impact of obesity on OSA severity in Chinese pa- were signifcantly different are shown. *Dimensions adjusted for bodytients who have on average smaller craniofacial bony enclosure. height (mm/meter). †Angular measurements in degrees. Our study has a number of important limitations. Whilst thesample size for each ethnic group is modest, there is potentialfor selection bias as patients were recruited from two separate ing was adequate. While there were no normal control groupsclinics. The difference in socioeconomic status, cultural, and available in this study, many previous intra-ethnic studies haveenvironmental factors between Australians and Hong Kong already demonstrated differences compared to non-apneic con- 12,27,28 Chinese may infuence subject presentation or referral pattern trols. Future ethnic comparisons of larger clinical cohortsfor clinical assessment, and possibly their preponderance for and control subjects in both genders would add further to thecertain OSA risk factors. Also, Caucasian patients were those results of this study. Craniofacial assessment with a more con- who had cephalometry performed as part of a clinical assess- venient and higher throughput photographic analysis technique29,30 ment for oral appliance therapy. However, the matching proce- could have great potential for application in this endeavor.dures performed in our analyses should address the stated aims This study was to investigate anatomical differences betweenof this study while minimizing some of the potential selection Caucasians and Chinese in OSA, and therefore other pathophys- bias. Furthermore, age and gender numbers were similar be- iological factors with possible ethnic infuence (e.g., neuromus - tween the ethnic groups in both matching analyses. For OSA cular and ventilatory control) were not specifcally examined.severity matching, while the ranges may seem wide, those pairs Examining detailed polysomnographic data, such as the ratiowho were matched with AHI up to 10 events/h and MinSaO2 of apneas to hypopneas, could provide additional phenotypicup to 10% difference tended to be those with more severe OSA. differences between the two ethnic groups. The defnition ofIn these cases, OSA severity is practically well matched. The ethnicity relies on the assumptions of the subjects and investi- means and standard deviations of the AHI and MinSaO2 for the gators, but this remains the most commonly used approach inCaucasian and Chinese patients suggest that the overall match- epidemiological studies. SLEEP, Vol. 33, No. 8, 2010 Ethnic Difference in OSA Risk Factors—Lee et al 1079In summary, this study adds to the increasing evidence of12. Hui DSC, Ko FWS, Chu ASY, et al. Cephalometric assessment ofcraniofacial morphology in Chinese patients with obstructive sleepthe phenotypic differences between Caucasian and Chineseapnoea. Respir Med 2003;97:640-6. patients with OSA. We found that craniofacial structures and 13. Mehta A, Qian J, Petocz P, Darendeliler MA, Cistulli PA. A randomized,obesity contributed differentially to OSA in these two ethniccontrolled study of a mandibular advancement splint for obstructive sleepgroups. Future work using more sophisticated analysis and im- apnea. Am J Respir Crit Care Med 2001;163:1457-61.14. Sleep-related breathing disorders in adults: recommendations for syndromeaging modalities will help to further defne the interaction ofdefnition and measurement techniques in clinical research. Americanthese anatomical risk factors across ethnic groups and this mayAcademy of Sleep Medicine Task Force. Sleep 1999;22:667-89. have implications in the diagnosis and management of OSA.15. Rechtschaffen A, Kales A. A manual of standardized terminology,techniques and scoring system for sleep stages of human subjects. LosAngeles: Brain Information Service/Brain Research Institute, 1968. ACKNOWLEDGMENTS16. Goldmann L. On-screen computerized cephalometric measurement.We thank Associate Professor Gang Shen for his assistance.Sydney: Discipline of Orthodontics, 1996. This research was sponsored in part by the Australian Society 17. Bearpark H, Elliott L, Grunstein R, et al. Snoring and sleep apnea.of Orthodontists Foundation for Research and Education. A population study in Australian men. Am J Respir Crit Care Med1995;151:1459-65.18. Lam B, Ip MSM, Tench E, Ryan CF. Craniofacial profle in Asian andDISCLOSURE STATEMENT white subjects with obstructive sleep apnoea. Thorax 2005;60:504-10. This was not an industry supported study. The authors have 19. Li KK, Kushida C, Powell NB, Riley RW, Guilleminault C. Obstructiveindicated no fnancial conficts of interest. sleep apnea syndrome: a comparison between Far-East Asian and whitemen. Laryngoscope 2000;110:1689-93.20. Ong KC, Clerk AA. Comparison of the severity of sleep-disorderedREFERENCES breathing in Asian and Caucasian patients seen at a sleep disorders center. 1. Kripke DF, AncoliIsrael S, Klauber MR, Wingard DL, Mason WJ,Respir Med 1998;92:843-8. Mullaney DJ. Prevalence of sleep-disordered breathing in ages 40-64 21. Villaneuva ATC, Buchanan PR, Yee BJ, Grunstein RR. Ethnicity andyears: A population-based survey. Sleep 1997;20:65-76. obstructive sleep apnoea. Sleep Med Rev 2005;9:419-36.2. Redline S, Tishler PV, Hans MG, Tosteson TD, Strohl KP, Spry K. Racial 22. Miles PG, Vig PS, Weyant RJ, Forrest TD, Rockette HE Jr. Craniofacialdifferences in sleep-disordered breathing in African-Americans andstructure and obstructive sleep apnea syndrome--a qualitative analysisCaucasians. Am J Respir Crit Care Med 1997;155:186-92. and meta-analysis of the literature. Am J Orthod Dentofacial Orthop 3. Ip MS, Lam B, Lauder IJ, et al. A community study of sleep- 1996;109:163-72. disordered breathing in middle-aged Chinese men in Hong Kong. Chest 23. Sakakibara H, Tong M, Matsushita K, Hirata M, Konishi Y, Suetsugu2001;119:62-9. S. Cephalometric abnormalities in non-obese and obese patients with 4. Udwadia ZF, Doshi AV, Lonkar SG, Singh CI. Prevalence of sleep- obstructive sleep apnoea. Eur Respir J 1999;13:403-10. disordered breathing and sleep apnea in middle-aged urban Indian men. 24. Hsu PP, Tan AKL, Chan YH, Lu PKS, Blair RL. Clinical predictors inAm J Respir Crit Care Med 2004;169:168-73. obstructive sleep apnoea patients with calibrated cephalometric analysis-5. Kim J, In K, Kim J, et al. Prevalence of sleep-disordered breathing in-a new approach. Clin Otolaryngol 2005;30:234-41. middle-aged Korean men and women. Am J Respir Crit Care Med 25. Dempsey JA, Skatrud JB, Jacques AJ, et al. Anatomic determinants of2004;170:1108-13. sleep-disordered breathing across the spectrum of clinical and nonclinical 6. Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S. The occurrencemale subjects. Chest 2002;122:840-51. of sleep-disordered breathing among middle-aged adults. N Engl J Med 26. Lowe AA, Fleetham JA, Adachi S, Ryan CF. Cephalometric and computed1993;328:1230-5. tomographic predictors of obstructive sleep apnea severity. Am J Orthod 7. Ferguson KA, Ono T, Lowe AA, Ryan CF, Fleetham JA. The relationshipDentofacial Orthop 1995;107:589-95. between obesity and craniofacial structure in obstructive sleep apnea. 27. Hoekema A, Hovinga B, Stegenga B, De Bont LGM. CraniofacialChest 1995;108:375-81. morphology and obstructive sleep apnoea: a cephalometric analysis. J 8. Watanabe T, Isono S, Tanaka A, Tanzawa H, Nishino T. ContributionOral Rehabil 2003;30:690-6. of body habitus and craniofacial characteristics to segmental closing 28. Johal AMA, Patel SI, Battagel JM. The relationship between craniofacialpressures of the passive pharynx in patients with sleep-disorderedanatomy and obstructive sleep apnoea: a case-controlled study. J Sleepbreathing. Am J Respir Crit Care Med 2002;165:260-5. Res 2007;16:319-26.9. Wong ML, Sandham A, Ang PK, Wong DC, Tan WC, Huggare J. 29. Lee RWW, Chan ASL, Grunstein RR, Cistulli PA. Craniofacial phenotypingCraniofacial morphology, head posture, and nasal respiratory resistancein obstructive sleep apnea – A novel quantitative photographic approach.in obstructive sleep apnoea: an inter-ethnic comparison. Eur J OrthodSleep 2009;32:37-45. 2005;27:91-7.30. Lee RWW, Petocz P, Prvan T, Chan ASL, Grunstein RR, Cistulli PA. 10. Liu Y, Lowe AA, Zeng X, Fu M, Fleetham JA. Cephalometric comparisonsPrediction of obstructive sleep apnea with craniofacial photographicbetween Chinese and Caucasian patients with obstructive sleep apnea.analysis. Sleep 2009;32:46-52. Am J Orthod Dentofacial Orthop 2000;117:479-85.11. Lam B, Ooi CGC, Peh WCG, et al. Computed tomographic evaluation ofthe role of craniofacial and upper airway morphology in obstructive sleepapnea in Chinese. Respir Med 2004;98:301-7. SLEEP, Vol. 33, No. 8, 2010 Ethnic Difference in OSA Risk Factors—Lee et al 1080"