Issue Navigator

Volume 10 No. 04
Earn CME
Accepted Papers

Scientific Investigations

Comparison of BMI, AHI, and Apolipoprotein E ε4 (APOE-ε4) Alleles among Sleep Apnea Patients with Different Skeletal Classifications

Jason J. Roedig, D.M.D.1; Barbara A. Phillips, M.D., M.S.P.H, F.A.A.S.M.2; Lorri A. Morford, Ph.D.1; Joseph E. Van Sickels, D.D.S.1; Gabriel Falcao-Alencar, M.S.1; David W. Fardo, Ph.D.3; James K. Hartsfield, D.M.D., M.S., M.M.Sc., Ph.D.1,2; Xiuhua Ding, M.S.3; G. Thomas Kluemper, D.M.D., M.S.1
1College of Dentistry, University of Kentucky, Lexington, KY; 2College of Medicine, University of Kentucky, Lexington, KY; 3College of Public Health, University of Kentucky, Lexington, KY


Study Objectives:

This case-control study investigated whether variations within the APOE-ε gene were associated with having a convex facial profile (skeletal Class II) compared to exhibiting a straight or concave facial profile (Class I or Class III) among patients with obstructive sleep apnea (OSA). Associations between the apnea-hypopnea index (AHI) and body mass index (BMI) scores for these OSA patients were also examined in the context of facial profile.


OSA patients with an AHI ≥ 15 were recruited from a sleep clinic and classified by facial and dental occlusal relationships based on a profile facial analysis, lateral photographs, and dental examination. Saliva was collected as a source of DNA. The APOE1-4 allele-defining single nucleotide polymorphisms (SNPs) rs429358 and rs7412 were genotyped. A χ2 analysis was used to assess Hardy-Weinberg equilibrium and for association analysis (significance at p < 0.05). ANOVA and Fisher exact test were also used.


Seventy-six Caucasian OSA patients participated in the study—25 Class II cases and 51 non-Class II cases. There was no association of the APOE4 allele with facial profile among these OSA patients. Class II OSA patients had significantly lower BMIs (30.7 ± 5.78) than Class I (37.3 ± 6.14) or Class III (37.8 ± 6.17) patients (p < 0.001), although there was no statistical difference in AHI for Class II patients compared with other groups.


OSA patients with Class II convex profile were more likely to have a lower BMI than those in other skeletal groups. In fact 20% of them were not obese, suggesting that a Class II convex profile may influence or be associated with OSA development independent of BMI.


Roedig JJ; Phillips BA; Morford LA; Van Sickels JE; Falcao-Alencar G; Fardo DW; Hartsfield JK; Ding X; Kluemper GT. Comparison of BMI, AHI, and apolipoprotein E ε4 (APOE-ε4) alleles among sleep apnea patients with different skeletal classifications. J Clin Sleep Med 2014;10(4):397-402.

Obstructive sleep apnea (OSA) is a common but underdiagnosed disorder characterized by repetitive episodes of complete or partial upper airway obstruction leading to cessation (apnea) or intermittent reduction (hypopnea) of airflow.1 These episodes usually last a minimum of 10 seconds and result in loud snoring, potential decreases in oxygen saturation, and chronic daytime sleepiness. Potentially fatal consequences of this disorder include hypertension, pulmonary hypertension, heart failure, nocturnal cardiac dysrhythmias, myocardial infarction, and ischemic stroke.2 In addition to adverse cardiovascular outcomes, people with OSA are have an increased risk of motor vehicle crashes.2

The gold standard for diagnosing OSA is a history and physical examination and an overnight sleep study (polysomnography).2 Although obesity is the phenotype most commonly associated with OSA, variations in craniofacial structure are also associated with the condition.36 When OSA is caused or influenced by a variation in anatomy, the obstruction may originate from either soft tissues (including musculature) or from skeletal components.2 One of the most common craniofacial risk factors is a retrognathic mandible (retruded relative to the maxilla in the sagittal plane).4 The maxilla and mandible may both be retruded relative to the cranial base in the sagittal plane, resulting in bi-maxillary retrusion. The retrognathic mandible and/or bi-maxillary retrusion contribute to a craniofacial relationship referred to as skeletal class II by the dental/orthodontic community.


Current Knowledge/Study Rationale: Although BMI is strongly correlated with OSA, there are individuals with other risk factors. This study was done to see if APOE-ε gene variants and/or facial profile were associated with differences in AHI or BMI in OSA patients.

Study Impact: A convex facial profile is associated with OSA even in non-obese patients, indicating that facial profile could be an important tool in risk assessment. Other non-obesity OSA factors such genetic variation need to be further explored.

The nomenclature of a Class II skeletal relationship has evolved from the original classification of malocclusions developed by Dr. Edward Angle in the 1890s.7 The original dental classifications of Class I, II, and III specifically described the variable sagittal relationship between the permanent maxillary and mandibular first molars bilaterally. Dr. Angle, however, did not take into account the sagittal relationship of the maxilla and mandible themselves or the facial profile. Subsequently it became clear that variation in the sagittal relationship between teeth was usually a reflection of the corresponding sagittal relationship between the maxilla and mandible. The classification system that was originally designed to describe occlusal relationships was “borrowed” to describe similar patterns of balance and imbalance between the jaws, often with placing the descriptor “skeletal” in front of the terms Class I, Class II, or Class III.

When the jaws are in harmony in the sagittal plane, thereby providing facial and occlusal balance, this is considered or described to be skeletal Class I. When the mandible is positioned posteriorly in its relationship to the maxilla in the sagittal plane, it is described as a skeletal Class II relationship. This imbalance can be a result of a protruded maxilla, a retruded mandible or a combination of both. The majority of Class II relationships are due to a retruded and/or small mandible, where the maxilla is in a normal position relative to the cranial base. However, the maxilla may be retruded as well.8 Finally, when the mandible is positioned too far forward in its relationship with the maxilla, whether due to maxillary deficiency or mandibular protrusion (or both), it is referred to as a skeletal Class III relationship. The prevalence of each skeletal class depends upon the population of interest, but is roughly 60% Class I, 35% Class II, and 5% Class III for growing individuals of western European descent.7

While environmental factors may contribute to the development of OSA, evidence suggests that it may have a familial component. OSA family members often share similar jaw relationships and/or body mass indices (BMIs).3,9 The underlying causes of OSA are clearly variable, with data from clinical and epidemiologic studies indicating that genetic factors influence the expression of OSA.9 The apolipoprotein E (APOE) gene, specifically allele ε4, has been associated with OSA in two previous cohort studies involving a Caucasian population.10,11 However, other studies did not replicate these findings.12,13 Thus one aim of this case-control study was to investigate whether genetic variations of the APOE gene, especially the presence of the APOE4 allele, are associated with OSA in Class II patients compared to non-Class II (i.e., Class I or Class III) OSA patients. This study also investigated whether the severity of sleep apnea (as assessed by AHI) and/or the average body mass index (BMI) scores associated with OSA in Class II patients were significantly different when compared to the scores of non-Class II OSA patients.


Patient Recruitment

Ethics approval for this study, in the form of a Full Medical Review, was provided by the University of Kentucky Internal Review Board (IRB Protocol #12-0557-F6A). All study participants completed the informed consent/assent process and authorized the release of their health information. Patients 13 years of age and older were recruited from the University of Kentucky Good Samaritan Hospital Sleep Center during routine follow-up of OSA.

Determination of Skeletal Malocclusion

All participants underwent a profile facial analysis, the taking and review of lateral photographs, and dental examination to determine skeletal malocclusion. Although lateral cephalometric radiographs are used regularly to analyze the skeletal classification of orthodontic patients, they are not a routine part of an OSA patient evaluation, and hence were not available for any of the subjects. In this study, skeletal malocclusion was determined by dental exam, facial profile analysis, and lateral photograph based on specific criteria (Table 1). Greater emphasis was placed on the facial analysis to determine skeletal malocclusion than dental exam data, largely due to the number of missing teeth observed in some patients. Tooth loss may alter dental occlusion classification, but not facial classification. Patients who had complete dentures, a craniofacial syndrome and/or central sleep apnea syndrome (CSAS) were excluded from the study.

Criteria for the determination of skeletal classification from a subject's photographs and dental examination


table icon
Table 1

Criteria for the determination of skeletal classification from a subject's photographs and dental examination

(more ...)

All lateral photographs were oriented and standardized using Dolphin Imaging software (version 10.5; Dolphin Imaging & Management Solutions, Chatsworth, CA). The head of each participant was oriented with the soft tissue correlate of Frankfort Horizontal (Ala-Tragus) parallel to the floor. Three measurements were calculated on each participant: facial divergence (distance between soft tissue pogonion and the line perpendicular to ala-tragus registered from soft tissue glabella), facial convexity (angle formed by intersection of the soft tissue glabella-subnasale line and subnasale-soft tissue pogonion line), and position of the maxilla (distance between subnasale and the line perpendicular to ala-tragus registered at soft tissue glabella). These linear and angular measurements were measured and compared with published soft-tissue norms.14 Since greater emphasis was placed on facial analysis, these measurements were used to reinforce the clinical profile facial analysis of each participant.

Following the completion of this evaluation, 2 dental practitioners (the one who categorized the subjects in this report [JR], and one additional practitioner) independently assessed 10 additional OSA patients to determine the accuracy of the methodology described in Table 1 for determining the skeletal classification. There was 100% agreement on the skeletal classification among the examiners using this system of classification.

Determination of AHI and BMI Measurements

For purposes of this study, OSA was defined by an apnea plus hypopnea index (AHI) ≥ 15/h; patients with ≥ 5 central apneas/h were excluded from participation. The AHI for each subject was determined during an overnight, in-laboratory diagnostic sleep study in the accredited University of Kentucky Good Samaritan Hospital Center. The AHI was calculated using standard criteria.15

BMI was calculated in the clinic using the following equation: BMI = Weight (lb) / (Height (in) × Height (in)) × 703, where 703 is a conversion factor related to measurements made in pounds and inches versus kilograms and meters. Standard weight status categories associated with BMI ranges for adults were used.16

DNA Collection, Isolation, and Single Nucleotide Polymorphism (SNP) Genotyping

Saliva (2-4 mL) was collected and genomic DNA isolated from all study participants using the Oragene-DNA Saliva Collection Kits (DNA Genotek Inc., Ottawa, Ontario, Canada). Purified DNA was suspended in 10 mM Tris-HCl, 1 mM EDTA pH 8.0 (Fisher Scientific, Fair Lawn, NJ), and DNA concentrations were measured on the NanoDrop-1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE). The SNPs rs429358 and rs7412 were selected based upon their reported association with OSAS. SNP analysis of the purified genomic DNA was performed utilizing Taqman-based SNP Genotyping Assay Kits and reagents (Life Technologies/Applied Biosystems, Carlsbad, CA) in the Roche LightCycler 480 Instrument (Roche Applied Science, Indianapolis, IN).

Statistical Methods

ANOVA was used to compare BMI and AHI among the skeletal classifications and by genotype. A χ2 analysis was used to access Hardy-Weinberg equilibrium (HWE) (significance at p < 0.05), and a Fisher exact test was applied to analyze the association between APOE4 and skeletal classification of OSAS patients (significance at p < 0.05). All analyses were conducted using R software (version: 2.15.3) or the Excel 2010 (Microsoft Corporation) software program.


Seventy-six Caucasian patients with OSA were recruited to this study—52 males and 24 females for a male-female ratio of 2:1. Within this cohort, there were 37 skeletal Class I subjects (CL I), 25 skeletal Class II subjects (CL II), and 14 skeletal Class III subjects (CL III). The average age, BMI, and AHI among the OSA subjects by facial classification were determined (Figure 1). There was no statistical difference in age among the different skeletal classes. The AHI was lowest for skeletal Class II patients (33.6 ± 11.9), followed by skeletal Class III (39.5 ± 17.2), and was highest for skeletal Class I (46.0 ± 29.1); but the 3 groups were not statistically different (p = 0.2). However, differences in BMI were statistically significant among the skeletal classes examined (p < 0.001). Overall, the skeletal Class II OSA patients had a lower BMI than Class I and Class III patients, although on average they were still overweight (Table 2).

Average age, BMI and AHI grouped by facial category in OSA patients.


Figure 1

Average age, BMI and AHI grouped by facial category in OSA patients.

(more ...)

Descriptive statistics for OSA patients as a function of skeletal classification and sex


table icon
Table 2

Descriptive statistics for OSA patients as a function of skeletal classification and sex

(more ...)

One-fifth of all skeletal Class II OSA subjects (2 men and 4 women) had normal BMIs ranging from 20.7 to 24.5. There were no OSA subjects with Class I or III skeletal classifications with normal BMIs. Female patients with Class II had the lowest average BMI measurements (28.2 ± 5.6 kg/m2), while female OSA subjects with Class III had the highest average BMI (42.5 ± 7.6 kg/m2). While there were no statistically significant differences observed with AHI (p = 0.6) among the skeletal classes, it is noteworthy that the cohort of skeletal Class I and III subjects contained 7 subjects (27%) and 3 subjects (21.4%) with AHIs over 50, respectively. By comparison, there were no AHI scores over 50 in the Class II population with OSA.

Genotyping frequencies for rs429358 and rs7412 maintained HWE in the study population (Table 3A and 3B). Likewise, there were no associations with the genotypes at either SNP with AHI or BMI (data not shown). The haplotype created by the combined nucleotides found at rs429358 and rs7412 on the same chromosome can be imputed and the imputed haplotype is commonly used to determine an individual's APOE allele composition (Table 4A).17 The APOE allele composition for all OSAS subjects by skeletal class was determined (Table 4B). Within this cohort, the ε33 allele occurred with the highest frequency in all skeletal classifications, followed by the ε34 allele.

Genotypic distribution of SNP rs2429358 among patients with OSA by skeletal classification


table icon
Table 3A

Genotypic distribution of SNP rs2429358 among patients with OSA by skeletal classification

(more ...)

Genotypic distribution of SNP rs7412 among patients with OSA by skeletal classification


table icon
Table 3B

Genotypic distribution of SNP rs7412 among patients with OSA by skeletal classification

(more ...)

Haplotypes of APOE at rs429358 + rs7412 have previously been shown to be predictive of an individual's APOE allele complement.17


table icon
Table 4A

Haplotypes of APOE at rs429358 + rs7412 have previously been shown to be predictive of an individual's APOE allele complement.17

(more ...)

Number of OSA patients by APOE allele combination grouped by skeletal classification


table icon
Table 4B

Number of OSA patients by APOE allele combination grouped by skeletal classification

(more ...)

When a patient is heterozygous at both SNPs (CT and CT), however, the actual subject's haplotype cannot be imputed with 100% certainty. Only one patient in our subject population was heterozygous for both SNPs, and that patient was excluded for the statistical analysis involving APOE alleles. There was no statistical difference (p = 0.7) in the number of individuals with no APOE4 alleles compared to individuals with one or two APOE4 alleles, when compared by skeletal classification (Table 4C). When comparing the average BMI and AHI among individuals with one or more APOE4 alleles vs. individuals with no APOE4 alleles, there was no statistical difference in BMI (p = 0.8) or AHI (p = 0.9), respectively (Table 5).

Number of OSA patients with no APOE ε4 alleles compared to OSA patients with one or more APOE4 alleles grouped by skeletal classification


table icon
Table 4C

Number of OSA patients with no APOE ε4 alleles compared to OSA patients with one or more APOE-ε4 alleles grouped by skeletal classification

(more ...)

Average BMI and AHI ± 1 SD grouped by lack of APOE4 alleles, or the presence of at least one APOE4 allele.


table icon
Table 5

Average BMI and AHI ± 1 SD grouped by lack of APOE-ε4 alleles, or the presence of at least one APOE-ε4 allele.

(more ...)


The most notable finding of this study was that OSA patients with skeletal Class II (convex facial profile), had a significantly lower BMIs than the non-Class II OSA patients. This finding suggests that the skeletal Class II phenotype may be associated with or directly influence OSA development irrespective of an individual's BMI. Unlike the apparent synergistic association between changes in the PHOX2B gene nucleotide sequence and Class III malocclusion in some but not all children and OSA,18 we found no evidence to support a synergistic association between APOE4 allele and Class II (or Class I or Class III) in our adult OSAS subjects

The focus of this study was to look at factors other than BMI that could be associated with OSA, since there are individuals affected with OSA who are not obese. The factors we choose to investigate were a skeletal Class II facial profile relationship, and the presence of an APOE4 allele, which has been in some studies associated with OSA. We investigated whether having a Class II facial profile and an APOE4 allele either by themselves, or in combination, were associated with a lower BMI in OSA patients.

Multiple studies examining OSA patients have identified several disorder-associated craniofacial features, including a retrognathic (Class II facial pattern) mandible (Table 6).1,4,19,20 Thirty-three percent of our sample was classified as facial Class II. This is higher than the 13% reported in the adult population.7 Interestingly, six of the OSA patients in this study had a normal BMI, and all of them were facial Class II. Conversely, 70 patients (92%) of our sample were classified as either overweight or obese. It is not surprising then that although all three facial classes were represented in the obese patients, the Class II patients had a significantly lower average BMI than did the skeletal Class I and III patients.

Summary of common craniofacial features observed in OSA patients.


table icon
Table 6

Summary of common craniofacial features observed in OSA patients.

(more ...)

Even though there was no statistically significant difference in AHI among the facial types, the Class II group had the lowest average AHI value while the average values for the Class I and Class III groups were approximately the same. This suggests that although all of the subjects were diagnosed with OSA, the Class II group tended to have a less severe manifestation of it, although there was a great deal of variation. The presence of a Class II facial type then appears to be associated with OSA, even in the absence of being overweight or obese.

Since not all skeletal Class II facial subjects develop OSA, and there are some Class II subjects who are not overweight or obese who develop OSA, there appear to be factors other than BMI that influence the development of the condition. This is supported by OSA being a disorder of complex etiology with multiple risk factors that interact to determine the phenotype. In addition to environmental risk factors, there is a strong genetic (or at least familial, which can also include familial environmental cofactors) basis for OSA, likely due to the summation of small to moderate effects from a large number of genetic loci.21,22 It has been estimated that 35% to 40% of the variance in AHI seen in OSA patients can be attributed to genetic factors.22

Some genetic studies have found an association between the apolipoprotein E4 allele and OSA. In contrast, however, several other studies have found no association. Among the middle-aged adult participants in the Wisconsin Sleep Cohort Study (WSCS), the presence of an APOE4 allele was associated with a two-fold increase in odds of OSA.11 In the Sleep Heart Health study APOE4 allele was associated with increased risk of OSA, particularly in individuals under age 65.10 However, the Honolulu-Asia Aging Study and the Finnish study on obstructive sleep apnea/hypopnea patients found no association with OSA and APOE4.12,23 A meta-analysis and meta-regression found that the hypothesis that the APOE4 can be causally associated with obstructive sleep apnea could not be supported in the published literature.24 However, none of these OSA studies investigated the APOE4 allele with the presence of a skeletal Class II facial type, particularly in individuals with a normal BMI. Our study found no association between APOE4 allele and OSA patients' BMI, AHI, or skeletal classification. However, it should be noted that in regard to genetic research, our study is under-powered due to a small sample size if the effect of the APOE4 allele is small.


This was not an industry supported study. This project was funded in part by the National Institutes of Health through the University of Kentucky College of Dentistry Center for Biomedical Research Excellence (P20RR020145), and the E. Preston Hicks Endowed Chair. The authors have indicated no financial conflicts of interest.


The authors thank the staff at the University of Kentucky Good Samaritan Sleep Center for their help in patient recruitment and data collection in this study.



Albajalan OB, Samsudin AR, Hassan R, authors. Craniofacial morphology of Malay patients with obstructive sleep apnoea. Eur J Orthod. 2011;33:509–14. [PubMed]


Chokroverty S, author. Overview of sleep & sleep disorders. Indian J Med Res. 2010;131:126–40. [PubMed]


Young T, Peppard PE, Gottlieb DJ, authors. Epidemiology of obstructive sleep apnea: a population health perspective. Am J Respir Crit Care Med. 2002;165:1217–39. [PubMed]


Chi L, Comyn FL, Mitra N, et al., authors. Identification of craniofacial risk factors for obstructive sleep apnoea using three-dimensional MRI. Eur Respir J. 2011;38:348–58. [PubMed]


Miyao E, Noda A, Miyao M, Yasuma F, Inafuku S, authors. The role of malocclusion in non-obese patients with obstructive sleep apnea syndrome. Intern Med. 2008;47:1573–8. [PubMed]


Susarla SM, Abramson ZR, Dodson TB, Kaban LB, authors. Cephalometric measurement of upper airway length correlates with the presence and severity of obstructive sleep apnea. J Oral Maxillofac Surg. 2010;68:2846–55. [PubMed]


Proffit WR, Fields HW, Sarver DM, authors. Contemporary Orthodontics. St. Louis, MO: Mosby Elsevier, 2007.


McNamara JA Jr, author. Components of class II malocclusion in children 8-10 years of age. Angle Orthod. 1981;51:177–202. [PubMed]


Kryger MH, Roth T, Dement WC, editors. Principles and Practice of Sleep Medicine. Philadelphia, PA: Saunders/Elsevier, 2011.


Gottlieb DJ, DeStefano AL, Foley DJ, et al., authors. APOE epsilon4 is associated with obstructive sleep apnea/hypopnea: the Sleep Heart Health Study. Neurology. 2004;63:664–8. [PubMed]


Kadotani H, Kadotani T, Young T, et al., authors. Association between apolipoprotein E epsilon4 and sleep-disordered breathing in adults. JAMA. 2001;285:2888–90. [PubMed]


Saarelainen S, Lehtimaki T, Kallonen E, Laasonen K, Poussa T, Nieminen MM, authors. No relation between apolipoprotein E alleles and obstructive sleep apnea. Clin Genet. 1998;53:147–8. [PubMed]


Foley DJ, Masaki K, White L, Redline S, authors. Relationship between apolipoprotein E epsilon4 and sleep-disordered breathing at different ages. JAMA. 2001;286:1447–8. [PubMed]


Burstone CJ, Marcotte MR, authors. Problem Solving in Orthodontics: Goal-Oriented Treatment Strategies. Chicago: Quintessence Pub. Co., 2000.


Iber C, Ancoli-Israel S, Chesson A, Quan SF, authors; for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology, and Technical Specifications. Westchester, IL: American Academy of Sleep Medicine, 2007.


World Health Organization. 10 Facts on Obesity. Fact file. Accessed 2013.


Williams-Gray CH, Goris A, Saiki M, et al., authors. Apolipoprotein E genotype as a risk factor for susceptibility to and dementia in Parkinson's disease. J Neurol. 2009;256:493–98. [PubMed]


Lavezzi AM, Casale V, Oneda R, Gioventu S, Matturri L, Farronato G, authors. Obstructive sleep apnea syndrome (OSAS) in children with Class III malocclusion: involvement of the PHOX2B gene. Sleep Breath. 2013;17:1275–80. [PubMed]


Ono T, Lowe AA, Ferguson KA, Pae EK, Fleetham JA, authors. The effect of the tongue retaining device on awake genioglossus muscle activity in patients with obstructive sleep apnea. Am J Orthod Dentofacial Orthop. 1996;110:28–35. [PubMed]


Johns FR, Strollo PJ Jr.; Buckley M, Constantino J, authors. The influence of craniofacial structure on obstructive sleep apnea in young adults. J Oral Maxillofac Surg. 1998;56:596–602; discussion 02-3. [PubMed]


Patel SR, author. Shared genetic risk factors for obstructive sleep apnea and obesity. J Appl Physiol. 2005;99:1600–06. [PubMed]


Kaparianos A, Sampsonas F, Karkoulias K, Spiropoulos K, authors. Obstructive sleep apnoea syndrome and genes. Neth J Med. 2006;64:280–9. [PubMed]


Palmer LJ, Buxbaum SG, Larkin E, et al., authors. A whole-genome scan for obstructive sleep apnea and obesity. Am J Hum Genet. 2003;72:340–50. [PubMed Central][PubMed]


Thakre TP, Mamtani MR, Kulkarni H, authors. Lack of association of the APOE epsilon 4 allele with the risk of obstructive sleep apnea: meta-analysis and meta-regression. Sleep. 2009;32:1507–11. [PubMed Central][PubMed]