"original article Facial morphology and obstructive sleep apnea 1 2 3 4Anderson Capistrano , Aldir Cordeiro , Leopoldino Capelozza Filho , Veridiana Correia Almeida , 5 6 3 Priscila Izabela de Castro e Silva , Sandra Martinez , Renata Rodrigues de Almeida-Pedrin DOI: http://dx.doi.org/10.1590/2177-6709.20.6.060-067.oar Objective: This study aimed at assessing the relationship between facial morphological patterns (I, II, III, Long Faceand Short Face) as well as facial types (brachyfacial, mesofacial and dolichofacial) and obstructive sleep apnea (OSA) inpatients attending a center specialized in sleep disorders. Methods: Frontal, lateral and smile photographs of 252 patients(157 men and 95 women), randomly selected from a polysomnography clinic, with mean age of 40.62 years, were evalu - ated. In order to obtain diagnosis of facial morphology, the sample was sent to three professors of Orthodontics trained toclassify patients’ face according to five patterns, as follows: 1) Pattern I; 2) Pattern II; 3) Pattern III; 4) Long facial pattern;5) Short facial pattern. Intraexaminer agreement was assessed by means of Kappa index. The professors ranked patients’facial type based on a facial index that considers the proportion between facial width and heighR t. esults: The multiplelinear regression model evinced that, when compared to Pattern I, Pattern II had the apnea and hypopnea index (AHI)worsened in 6.98 episodes. However, when Pattern II was compared to Pattern III patients, the index for the latter was11.45 episodes lower. As for the facial type, brachyfacial patients had a mean AHI of 22.34, while dolichofacial patientshad a significantly statistical lower index of 10.52 . Conclusion: Patients’ facial morphology influences OSA. Pattern IIand brachyfacial patients had greater AHI, while Pattern III patients showed a lower index. Keywords: Diagnosis. Face. Obstructive sleep apnea.Objetivo: o objetivo deste estudo foi avaliar a associação entre os padrões morfológicos faciais e tipos faciais (braquifacial,mesofacial e dolicofacial) com a Apneia Obstrutiva do Sono (AOS) em pacientes de um centro especializado em dis - túrbios do sono. Métodos: foram utilizadas fotografias faciais de frente, perfil e sorriso de 252 indivíduos, selecionadosaleatoriamente entre pacientes que procuraram uma clínica especializada em polissonografia. Para o diagnóstico morf- o lógico facial, a amostra foi enviada a três professores de Ortodontia treinados na classificação do padrão facial, e cada umrecebeu a orientação para classificar o padrão facial da seguinte forma: 1 = Padrão I, 2 = Padrão II, 3 = Padrão III, 4 =Padrão Face Longa e 5 = Padrão Face Curta. A concordância interexaminadores foi avaliada por meio do Índice Kappa.O diagnóstico do tipo facial foi estabelecido por meio de um índice facial que leva em consideração a proporção entre alargura e altura da face.Resultados: no modelo de regressão linear múltipla, ficou evidenciado que o Padrão II teve acapacidade de agravar o índice de apneia e hipopneia (IAH) em 6,98, enquanto os pacientes do Padrão III tinham esseíndice atenuado em 11,45. Para o tipo facial, os pacientes braquifaciais apresentaram um IAH médio de 22,34, enquantoo grupo classificado como dolicofacial mostrou um índice menor, de 10,52, com significância estatística Co . nclusão: odesenho morfológico facial se mostrou um considerável fator de agravamento ou proteção da SAOS, onde os indivíduosPadrão II e braquifaciais tiveram IAH maiores, enquanto nos pacientes Padrão III esse índice foi reduzido. Palavras-chave: Diagnóstico. Face. Apneia do Sono Tipo Obstrutiva. 1Professor of Occlusion and Orthodontics, Faculdade de Odontologia de RecifeHow to cite this article: Capistrano A, Cordeiro A, Capelozza Filho L, Almei- (FOR-PE), Recife, Pernambuco, Brazil. da VC, Castro e Silva PI, Martinez S, Almeida-Pedrin RR. Facial morphology2MSc in Orthodontics, Universidade Sagrado Coração (USC), Bauru, Sãoand obstructive sleep apnea. Dental Press J Orthod. 2015 Nov-Dec;20(6):60-7.Paulo, Brazil. DOI: http://dx.doi.org/10.1590/2177-6709.20.6.060-067.oar 3Professor, Universidade Sagrado Coração (USC), Undergraduate and Graduateprograms, Bauru, São Paulo, Brazil. » The authors report no commercial, proprietary or fnancial interest in the products4Postgraduate student, Cooperativa dos Odontologistas de Pernambucoor companies described in this article. (COOPE), Recife, Pernambuco, Brazil. 5Student, Sindicato dos Odontologistas no Estado de Pernambuco » Patients displayed in this article previously approved the use of their facial and in-(SOEPE), Postgraduate program in Orthodontics, Recife, Pernambuco, Brazil. traoral photographs. 6Specialist in Sleep Medicine, Universidade de São Paulo (USP), São Paulo, SãoPaulo, Brazil, and Stanford University, Stanford, California, USA. Contact address: Anderson Capistrano E-mail: [email protected] Submitted: December 21, 2014 - Revised and accepted: June 30, 2015 © 2015 Dental Press Journal of Orthodontics 60 Dental Press J Orthod. 2015 Nov-Dec;20(6):60-7Capistrano A, Cordeiro A, Capelozza Filho L, Almeida VC, Castro e Silva PI, Martinez S, Almeida-Pedrin RR original article INTRODUCTION important clinical evidence of quick, inexpensive andIn the last 20 years, Dentistry has discussed cases of simple diagnosis of a disease that significantly impairssnoring and obstructive sleep apnea (OSA). Within an patients’ quality of life. 2-7  interdisciplinary approach, neurologists, otolaryngolo- The literature has demonstrated a relationshipgists, physiotherapists, speech therapists and clinicians between craniofacial dimensions and upper airwayhave all recognized the importance of assessing these structures in patients with OSA. These results givepatients from the point of view of Dentistry, not only support to the potential role facial measurements playin terms of therapeutic control, but also in prevent - in the anatomical phenotype of OSA. ing it by means of treating potential malocclusions that  Craniofacial morphology seems to be among thecould increase the risk of airway disorders, particularly predisposing factors for the development of OSA.when they are associated with predisposing factors Mandibular deficiency and increased anterior-inferior1 7  such as obesity, hypertension and aging. facial height highlight such possibility. The majority7-12  Obstructive sleep apnea (OSA) is a common disor- of publications used cephalometric measurementsder associated with snoring, upper airway collapse at to define craniofacial morphology, which may causesleep, oxygen desaturation and fragmented sleep. It is considerable doubt and contradiction. In an attemptalso associated with cardiovascular morbidity, risk of to avoid it, the present study aims at assessing a diag- 1  car accidents and general mortality. OSA diagnosis is nosis system that considers the genetic determinismnot simple, since polysomnography requires patient’s of craniofacial morphology, with specific methodsmonitoring by a specialist at a sleep laboratory, which centered around a diagnosis concept based on facial13,14renders the procedure relatively expensive and diffi - patterns, in addition to studying its relationshipcult. In order to simplify diagnosis, a study aimed at with obstructive sleep apnea in adult patients. assessing differences in craniofacial phenotype amongcaucasian patients with OSA. The study found thatdetailed anatomic data, such as facial width, distance MATERIAL AND METHODS between eyes, as well as mandibular length and chin- Sample calculation neck angle, were useful in predicting OSA with a For sample size calculation, alpha = 5% and the2 sensitivity index of 86%. power of the test was of 80%. Calculation was basedAnatomical airway narrowing is among the etio- on multiple linear regression analysis, with AHI aslogical factors of snoring and OSA. It consists in the dependent variable; and sex, age, BMI, facial typesoft tissues excess, macroglossia and retrognathism. (three categories) and facial pattern (five categories)This condition causes great resistance that hinders as independent variables, thereby totaling 11 predic- air flow and engenders negative intraluminal pressure tor variables. Effect size was set at 0.10 and minimalwhile inspiring, thereby favoring breathing collapse. sample size was of 178 cases. Calculation was carried15  The risk of developing this disorder significantly in - out using the software developed by Soper, in 2014. creases with weight gain, aging, increased neck cir- cumference and alcohol consumption. The following Sample selection systemic conditions also appear as predisposing fac - The fnal sample comprised 252 patients with a meantors: systemic hypertension, untreated hypothyroid- age of 40.62 (from 18 to 62 years old) and a mean BMI3 ism, acromegaly and nasal obstruction. of 28.74 ± 4.73). A total of 157 men and 95 women whoOSA recognition in the overall population re- were referred to a center specialized in sleep disorders and1 mains low and most patients are not diagnosed. polysomnography. Patients’ main complaint involvedThus, there is a critical, clinical need to develop bet- snoring, insomnia, restless nights, chronic pain, memoryter methods that allow OSA recognition and diag- defcit, bruxism and daytime sleepiness. The researchnosis. The present study was conducted to assess the project from which the present study originated was ap - relationship between facial morphological pattern, proved by Universidade Sagrado Coração (USC) Insti- within a contemporary context of genetic determin - tutional Review Board under protocol #412.260. All pa- ism, and OSA. This relationship might stand for an tients signed an informed consent form. © 2015 Dental Press Journal of Orthodontics 61 Dental Press J Orthod. 2015 Nov-Dec;20(6):60-7original article Facial morphology and obstructive sleep apnea Afer interviewing an average of 900 patients, some tern (Fig 1). The photographs were inserted in a Pow- TM  of them were excluded based on the following criteria: erPoint slide presentation and sent via WeTrans- TM  » BMI over 40. fer to three experienced orthodontic professors.» Impaired posterior occlusal support. Each examiner was advised to classify patients’ facial» Patients with a beard that hindered facial analy - pattern according to the following: 1) Pattern I; 2 ) Pat- sis. tern II; 3) Pattern III; 4) Long Facial Pattern; 5) Short» Patients using continuous positive airway pres- Facial Pattern. Examiners did not have access to pa- sure (CPAP) or intraoral appliance and who were be- tients’ reports and, for this reason, were unaware ofing subjected to the examination, so as to assess the those with and without OSA. therapeutic effectiveness of the devices. Examiners reached facial type diagnosis through a16 » Craniofacial syndromic patients. facial index (n-gn/zy-zy)  that considers the propor- » Patients with chronic obstructive pulmonary tion between facial width and height, with a meandiseases (CPOC) or neurological or mental disorders value of 88.5 for men and 86.2 for women (Table 1).who were affected by upper airway infection. Measurements were obtained with the aid of Photo - TM  » Patients with history of orthognathic surgery or shop CS4 software (Fig 2). any other type of airway surgery. Patients were classified as mesofacial, brachyfacial» Patients under 18 and above 62 years old. and dolichofacial, as shown in Table 1.» Patients who, for any reason, did not agree intaking part in the research. Error of the method » After facial pattern analysis, other patients were To assess the error of classifying both facial pattern17  also excluded for presenting three different diagnoses. and type, Kappa index was used (Table 2). For facialThe 252 individuals comprising the sample were pattern, the diagnoses of three examiners were crosseddivided into two groups according to polysomnog- for each patient. Patients with three diferent diagno - raphy results. The group without OSA (Group I, 77 ses were excluded from this study. Four patients werepatients) and the group with OSA (Group II, 175 excluded by means of this criterion. As for facial type,patients) which presented an AHI value greater than 30% of the sample was randomly selected, so as to re - five episodes of apnea per hour of sleep, enough to assess patients’ facial proportions afer 30 days. characterize the individuals as having the disease. All assessments reached an agreement value thatFrontal, lateral and smile standardized photographs ranged between strong and nearly perfect. Landis &18  were used to assess patients’ facial morphological pat - Koch’s classification was used as reference. Figure 1 - Example of photograph set-up for diagnosis. © 2015 Dental Press Journal of Orthodontics 62 Dental Press J Orthod. 2015 Nov-Dec;20(6):60-7Capistrano A, Cordeiro A, Capelozza Filho L, Almeida VC, Castro e Silva PI, Martinez S, Almeida-Pedrin RR original article To assess the combined effect of sex, age, BMI,facial type and facial pattern on the AHI value, ste- p wise backward multiple linear regression analysiswas used. Significance level was set at 5% ( p < 0.05)for all tests. All statistical procedures were carried outby means of Statistica version 12 (StatSoft Inc., Tulsa,USA) software. RESULTS Of the 289 patients, 29 were excluded based onthe exclusion criteria. A total of 260 patients wereanalyzed, four of which were excluded for prese-nt ing three different morphological diagnoses, one fornot having polysomnography concluded and three fornot having polysomnography results sent for analysiswithin a reasonable time. Thus, 252 patients were in- cluded in the statistical analysis. In terms of facial morphological pattern and OSAprevalence, patients were classified according to datapresented in Table 3. Data analysis revealed statisticaldifference between Pattern II and long face. DespiteFigure 2 - One of the patients comprising the sample. Image used to illus - trate how facial type was obtained. no relevant statistically significant difference, the pe - r centage of short face individuals with OSA is grea -t Table 1 - Facial index. er than expected, as it totaled 77.8% against 22.2%Men Women for individuals without the disorder. Both percent - Mesofacial 83.4 – 93.6 81.6 – 90.8 ages are significantly near those found for Pattern II,Brachyfacial < 83.4 < 81.6 80.3% and 19.7%, respectively. When facial patternsDolichofacial > 93.6 > 90.8 were assessed by Kruskal-Wallis test, according to theabsolute AHI value, there was statistically significantdifference between Pattern II and long face (Table 4).Table 2 - Agreement percentage and Kappa’s values for intraexaminer agree - ment assessment. Importantly, it is worth noting that the mean AHIvalue for the Pattern III group (11.4) was nearly halfMeasurement % of agreement Kappa Examiner 1 versus Examiner 2 77.8 0.64 that of the Pattern II group (22.51). Examiner 1 versus Examiner 3 77.4 0.62 Table 5 shows no statistically significant differencesExaminer 2 versus Examiner 3 81.4 0.68 for distribution of facial types for groups I and II.Facial type 1 versus Facial type 2 90.4 0.90 Nevertheless, when groups were classified accordingto the AHI value (Table 6), there was statistically sig- nificant difference between brachyfacial patients witha mean AHI of 22.34, and dolichofacial patients withStatistical analysis a mean AHI value of 10.52. Data were arranged in tables and charts using ab- Results reveal that facial pattern influenced thesolute (n) and relative (%) frequencies. Quantitative apnea and hypopnea index (AHI) severity when mul- variables were described by mean and standard devia- tiple linear regression analysis was used. This is be- tion parameters. cause apnea is a multifactorial disorder in which theTo assess the relationship between variables, interaction among variables, such as BMI, sex andKruskal-Wallis, chi-square and Spearman’s correla- age, also influence AHI values. Importantly, multipletion tests were used. linear regression analysis is considered adequate for© 2015 Dental Press Journal of Orthodontics 63 Dental Press J Orthod. 2015 Nov-Dec;20(6):60-7"