Vol. 119 No. 1 January 2015

The usefulness of cephalometric measurement as a diagnostic tool for obstructive sleep apnea syndrome: a retrospective study Hyun-Ho Ryu, DDS,a Chul-Hoon Kim, DDS, PhD,b Sang-Myung Cheon, MD, PhD,c Woo-Yong Bae, MD, PhD,d Sang-Ho Kim, MD, PhD,e Soo-Kweon Koo, MD, PhD,f Myoung-Soo Kim, PhD,g and Bok-Joo Kim, DDS, PhDh Busan Saint Mary’s Hospital, Pukyong National University, Dong-A University Medical Center, Busan, Korea

Objective. Obstructive sleep apnea syndrome (OSAS) is a disorder characterized by apnea and hypopnea maintained for over 10 seconds and occurring at least 5 times per hour, with at least 30 episodes during 7 hours of nocturnal sleep. The most important pathophysiology in OSAS is the obstruction of the upper airway during sleep. The aim of this study was to identify the correlations between lateral cephalometric parameters, which seemed to be related to OSAS severities, and polysomnography (PSG) indices and to thus determine the cephalometric parameters reflecting OSAS severity. Patients and Methods. A total of 140 participants (122 males, 18 females) were evaluated by lateral cephalography and PSG. A total of 29 measurements (24 distances and 5 angles) were made on lateral cephalography. Cephalometric and PSG parameters were evaluated statistically to select and validate the cephalometric parameters reflecting OSAS severity. Result. OSAS has a significant relationship with the anatomic deformities of craniofacial and soft tissues. Lateral cephalometry revealed that patients with OSAS have a significant vertical airway length, a regrognathic mandible, a thick uvula, a large tongue, and a long mid-face length. The position of the hyoid bone had a tendency to displace inferiorly and/or posteriorly. Using the discriminant variable combination, including tongue base-posterior nasal spine (T1-PNS), sellaenasioneB point angle (SNB), maximum uvula thickness (Max U), tongue base-tongue tip (T1-TT), and nasion-anterior nasal spine (N-ANS), 102 of 140 (72.9%) patients were correctly assigned to the normal-to-mild and moderate-to-severe apneaehypopnea index (AHI) groups. Conclusions. Lateral cephalometric radiography may be an accessible and suitable tool for evaluation of craniofacial and soft tissue deformities in their correlations with OSAS severity. Further research on the cephalometric parameters reflecting OSAS severity is needed. (Oral Surg Oral Med Oral Pathol Oral Radiol 2015;119:20-31)

Obstructive sleep apnea syndrome (OSAS) is a disorder characterized by apnea and hypopnea maintained for over 10 seconds and occurring at least 5 times per hour, or with at least 30 episodes during 7 hours of nocturnal sleep.1,2 Various symptoms are associated with OSAS, including excessive daytime sleepiness, diminished executive function, oxygen desaturation, cardiovascular disorder, depression, and personality change.3 The estimated prevalence in Western countries is 4% among men and 2% among women.4,5 a Postgraduate, Department of Oral and Maxillofacial Surgery, Dong-a University Medical Center. b Professor, Department of Oral and Maxillofacial Surgery, Dong-a University Medical Center. c Associate Professor, Department of Neurology, Dong-a University Medical Center. d Associate Professor, Department of Otorhinolaryngology, Dong-a University Medical Center. e Professor, Department of Neurology, Dong-a University Medical Center. f Chief, Department of Otorhinolaryngology, Busan Saint Mary’s Hospital. g Associate Professor, Department of Nursing, Pukyong National University. h Assistant Professor, Department of Oral and Maxillofacial surgery, Dong-a University Medical Center. Received for publication Jun 28, 2014; accepted for publication Jul 22, 2014. Ó 2015 Elsevier Inc. All rights reserved. 2212-4403/$ - see front matter http://dx.doi.org/10.1016/j.oooo.2014.07.537

20

The most important pathophysiology in OSAS is the obstruction of the upper airway during sleep. The upper airway is a pliant tube and, as such, is subject to collapse.6 Airflow velocity increases at the site of the obstruction in the respiratory tract, in accordance with the Bernoulli effect.7 Since the tendency for upper airway obstruction in OSAS patients is greatly influenced by the anatomic state of the respiratory tract,8,9 evaluating the structure of the upper airway is essential to make the correct diagnosis of, and provide effective treatment for, OSAS.2 A variety of tools for the identification of the obstruction site, anatomic structure, and pattern have been introduced, all of which have advantages and disadvantages (Table I).10 Although polysomnography (PSG) is

Statement of Clinical Relevance Our aim was to find out whether the use of lateral cephalometric radiography is an accessible and suitable methodology for the evaluation of craniofacial and soft tissue deformities related to obstructive sleep apnea syndrome (OSAS) severity. As lateral cephalometric radiography is an easily used clinical tool in dentistry, early diagnosis of potentially severe OSAS was possible.

Cross-section

the “gold standard” for the identification of individuals with OSAS, quantification of its severity, medical management guidance, and determination of the success of treatment modalities, it does not identify the site of obstruction or predict surgical results. Development of diagnostic tools for correct detection of the pattern of upper airway obstruction is, however, important because during sleeping periods, this pattern is dynamic, and its direction is variable and three-dimensional.11 Dynamic means of obstruction site evaluation are sleep videofluoroscopy (SVFS) and drug-induced sleep endoscopy (DICE). The advantages of SVFS include direct observation of the dynamic anatomy and the obstructive site during drug-induced sleep and its easy availability; the disadvantages are exposure to high doses of radiation and superimposition of normal anatomic structures. DICE is an inexpensive, accessible technique for dynamic evaluation of the airway in multiple positions in drug-induced sleep. However, both SVFS and DICE require sedation to attain sleep during the procedure, and the differences in reproducibility between natural sleep and drug-induced sleep remain controversial. Lateral cephalometric radiography also is widely used for evaluation of upper airway obstruction in patients with OSAS. Although there are some disadvantages, such as studying a three-dimensional object with a two-dimensional picture, superimposing structures, and having the patient awake and in the upright position, lateral cephalometric radiography is a noninvasive, inexpensive, widely available, and technically easy method for evaluation of skeletal and soft tissue abnormalities contributing to obstruction and, indeed, has been widely used in the examination of patients with OSAS. Also, this tool can be used to evaluate the obvious abnormalities of the maxillary and mandibular position for sagittal airway computed tomography (CT). The correlations between lateral cephalometric parameters and OSAS-related indices (e.g., AHI, RDI, etc.) have been analyzed in numerous studies.12-17 Unfortunately, the widely known parameters associated with OSAS are numerous, taking all of them into consideration is time consuming, and in any case, not all of them correctly coincide with PSG results. Also, there are no finely designed methods for the identification of potential patients with OSAS on the basis of routine lateral cephalometric radiography. Development of such methods would provide a very valuable diagnostic tool for distinguishing between those with OSAS and those without OSAS. The aims of this study were to identify the correlation between lateral cephalometric parameters seeming correlated with OSAS severities and PSG indices and, further, to determine the cephalometric parameters reflecting OSAS severity.

  þ þ þ þ   þ  þ þ þ þ  þ        þ þ þ  þ þ þ þ þ þ þ þ      PSG, polysomnography; CT, computed tomography; MRI, magnetic resonance imaging.

Allow PSG In-expensive

þ þ þ þ   þ þ  þ  þ þ þ  þ

Allow sedation

þ þ    þ þ þ

Dynamic No radiation

þ  þ þ þ þ þ 

Method

Visual inspection Fiberoptic endoscopy Cephalometry Fluoroscopy CT MRI Acoustic reflections Manometry

Non- invasive

 þ  þ þ þ þ þ

Technical ease Supine exam possible Widely available

Table I. Advantages and disadvantages of available methods for evaluation of obstruction site in patients with OSAS

 þ   þ þ þ 

ORIGINAL ARTICLE Ryu et al. 21

Evaluate surrounding tissue

OOOO Volume 119, Number 1

ORAL AND MAXILLOFACIAL SURGERY 22 Ryu et al.

PATIENTS AND METHODS Patients This study employed a retrospective cross-sectional descriptive design. All of the participants were selected on the basis of the data on patients who had visited the Sleep Center of Dong-A University Medical Center in Busan, Korea, between January 2009 and June 2013. The inclusion criteria were as follows: (1) complaint of sleep-related disorder (e.g., snoring, choking, and gasping for breath during sleep), (2) undergoing overnight PSG, and (3) having undergone lateral cephalography. All of the selected participants met all three of these criteria. The exclusion criteria were as follows: (1) less than 10 teeth in each jaw, (2) severe periodontitis, (3) diagnosis of temporomandibular joint disorder, and (4) use of any medication for disorders associated with cognition impairment, such as dementia or mental retardation. After excluding 212 out of the initial pool of 352 patients on the basis of inappropriate data or refusal to participate, 140 (122 males, 18 females) were included in the analysis. This study was approved by the institutional review board of Dong-A University Medical Center in Busan, Korea. METHODS Anthropometric evaluations The body mass index (BMI, Kg/m2), or Quetelet index, was determined as the individual’s body mass (kilograms) divided by the square of his or her height (meters). The neck circumference (NC) and waist circumference (WC) also were measured relative to stature (inches). Polysomnography. Polysomnography (PSG) was conducted at the Sleep Center of Dong-A University Medical Center using Alice Sleepware (Philips Respironics, Andover, MA) or Embla RemLogic 2.0 (Embla, Broomfield, CO). Apnea is defined by the American Academy of Sleep Medicine (AASM) as the cessation of airflow through the nostrils and mouth for at least 10 seconds.18 Hypopnea is defined as a recognizable transient reduction of breathing for 10 seconds or longer and a decrease of airflow by at least 30% as associated with oxygen desaturation of 4% or more.19,20 As an index of obstructive breathing disorder, the apneaehypopnea index (AHI), as the measure of the frequency of apnea and hypopnea, is evaluated. Also, the lowest saturation of peripheral oxygen (LSpO2) during sleep is measured. Lateral cephalometric radiography. RADIOGRAPHIC POSITIONING AND PREREQUISITES. Lateral cephalography was performed, with the orbital-auricular plane (F-H plane) parallel to the floor and the subject in the upright position at a 165-cm focal film distance and a 15-cm film distance from the sagittal plane using the PaX300 C Plus (Vatech, Hwaseong, Korea; 15). A discriminant function analysis (DA) was performed with the selected cephalometric parameters

ORAL AND MAXILLOFACIAL SURGERY 24 Ryu et al.

OOOO January 2015

predictors tend to have large weights. Nevertheless, since discriminant loadings have been considered more valid information compared with discriminant function coefficients, the interpretation of the present findings was based on  0.30, as suggested by Hair et al.22 The group centroid is the mean value of the discriminant score for a given category of dependent variable. The cutoff is the mean of the two centroids, and can be calculated as: Cutoff value ¼ ðCA NB þ CB NA Þ=ðNA þ NB Þ21

Fig. 6. Airway-related cephalometric measurements. NASu: Narrowest airway space from PNS to UT; NASt: Narrowest airway space from UT to EpT (epiglottis); PAS(PL): Posterior airway space on palatal line (ANS-PNS); PAS(OL): Posterior airway space on occlusal line (upper incisal tipemesiobuccal cusp of maxillary first molar); PAS(ML): Posterior airway space on mandibular line (Go-Me).

and validated to determine the correlation between the cephalometric parameters and OSAS severity and the corresponding utility for screening patients who needed further evaluation and treatment. DA can serve the same function as multiple linear regression by determining a linear equationelike regression that will predict the group to which a case belongs. There are two methods of selecting discriminant variables: (1) simultaneous or direct method, and (2) the stepwise method.21 In the present study, the simultaneous or direct method was applied to analyze the correlation between the cephalometric parameters and PSG indices and, thereby, select meaningful discriminant variables. The form of the function or equation was as follows: D ¼ a þ W1 X1 þ W2 X2 þ W3 X3 þ ::::: þ Wi X21 i

where D ¼ discriminate function a ¼ constant W ¼ discriminant coefficient X ¼ value of the independent variable i ¼ number of discriminant variables The discriminant score (Z) is the score of the discriminant function formula (D) substituting the values of the measured cephalometric parameters of the participants. The Ws in the regression equation are unstandardized discriminant coefficients analogous to the b-weight. Standardized discriminant coefficients also can be used like the b-weight in regression. Good

where C ¼ the group centroid and N ¼ the number of cases in a group. In the present study, the case was assigned to the normal-to-mild AHI group if the discriminant score (Z) of the function was less than or equal to the cutoff value, whereas it was assigned to the moderate-tosevere AHI group if was greater than the cutoff value. To verify the utility of the population classification, we performed a cross-validation. Statistical analysis. To answer the research questions, the SPSS WIN 18.0 program (SPSS Inc., Chicago, IL) with Pearson bivariate correlation analysis and discriminant function analysis was utilized in this study. Before conducting the discriminant analysis, the underlying assumptions were confirmed. A significance level of 0.05 was predefined in all cases. RESULTS Demographic characteristics of patients The patients numbered 140 (male: 122, female: 18). Their ages ranged from 18 to 74 years (male: 18-74, female: 21-66), with a mean of 45.9  13.10 years (male: 45.38  13.27, female: 49.44  11.52). Their average BMI was 26.44  3.59, neck circumference 15.28  1.27 inches, waist circumference 36.11  3.63 inches, AHI 24.23  21.3, and LSpO2 79.93  11.94 (Table II). Correlations between anthropometric measurements and polysomnographic parameters. AHI was positively correlated with weight (r ¼ .32; P < .01), BMI (r ¼ .34; P < .01), NC (r ¼ .31; P < .01) and WC (r ¼ .29; P < .01), and negatively correlated with LSpO2 (r ¼ .62; P < .01). LSpO2 was found to be correlated with weight (r ¼  .21; P < .05), BMI (r ¼ .31; P < .01), NC (r ¼ e.26; P < .01), and WC (r ¼ .27; P < .01) (Table III). Correlations among cephalometric measurements, polysomnographic indices, and anthropometric measurements. SKELETAL-RELATED PARAMETERS. Anterior facial height (N-Me) had a correlation with AHI (r ¼ .21; P < .05), as did posterior facial height (S-Go) (r ¼ .21; P < .05). Mid-face length (N-ANS) had a correlation with both AHI (r ¼ .22; P < .01) and LSpO2 (r ¼ .17;

OOOO Volume 119, Number 1

ORIGINAL ARTICLE Ryu et al. 25

Table II. Demographic characteristics of patients (n ¼ 140)

Age (years) Height (meter) Weight (kg) BMI (kg/m2) NC (inch) WC (inch) AHI LSpO2

Total (n ¼ 140)

Male (n ¼ 122, 87.14%)

Female (n ¼ 18, 12.86%)

M  SD

M  SD

M  SD

45.90 170.73 77.21 26.44 15.28 36.11 24.23 79.93

       

13.10 8.11 12.56 3.59 1.27 3.63 21.3 11.94

45.38 172.63 79.30 26.60 15.51 36.51 25.45 79.23

       

13.27 6.51 11.68 3.57 1.08 3.50 21.87 12.37

49.44 157.88 63.03 25.34 13.69 33.41 15.98 84.67

       

11.52 5.92 8.60 3.66 1.33 3.42 14.63 6.92

M, mean; SD, standard deviation; BMI, body mass index; NC, neck. circumference; WC, waist circumference; AHI, apneaehypopnea index; LSpO2, lowest saturation of peripheral oxygen.

Table III. Correlations between anthropometric measurement and polysomnographic parameters Weight Height BMI NC WC AHI LSpO2

Weight

Height

BMI

NC

1 .54y .82y .75y .72y .32y .21*

1 .04 .34y .19* .06 .09

1 .67y .73y .34y .31y

1 .76y .31y .26y

WC

AHI

LSpO2

1 .62y

1

1

.29y .27y

BMI, body mass index; NC, neck circumference; WC, waist circumference; AHI, apneaehypopnea index; LSpO2, lowest saturation of peripheral oxygen. *P < .05. y P < .01.

P < .05) SNB and A pointenasioneB point angle (ANB) were each found to have a correlation with AHI (r ¼ .23; P < .01 [SNB], r ¼ .20; P < .05 [ANB]). Of these, N-ANS was included as a discriminant variable owing to its correlation with both AHI and LSpO2 (Table IV). UVULA-RELATED PARAMETERS. Max U had a correlation with both AHI (r ¼ .19; P < .05) and LSpO2 (r ¼ .22; P < .01), whereas uvula length (PNS-UT) and uvula angle (U ang) did not show any correlation with either AHI or LSpO2. Accordingly, Max U was included as a discriminant variable (Table IV). HYOID-RELATED PARAMETERS. Most of the cephalometric parameters related to the hyoid were correlated with both AHI and LSpO2. However, H-A and H-MnP were correlated with only AHI (r ¼ .29; P < .01 and r ¼ .21; P < .05, respectively). H-B and H-C3, however, were correlated with both AHI (r ¼ .28; P < .01 and r ¼ .25; P < .01, respectively) and LSpO2 (r ¼ .28; P < .01 and r ¼ .20; P < .05, respectively), and so were selected as discriminant variables (Table IV). TONGUE-RELATED PARAMETERS. All of the cephalometric parameters relating to the tongue had a correlation with PSG parameters. Length of tongue (T1-TT), T1-A, T1-B, and T1-PNS were positively correlated with AHI and negatively correlated with LSpO2. Accordingly, T1-TT, T1-A, T1-B, and T1-PNS were included as discriminant variables (Table IV).

AIRWAY-RELATED PARAMETERS. Narrowest airway space (NASt) was negatively correlated with LSpO2 (r ¼ .23; P < .01). This (first subject) result did not reflect the facts that a narrow anteroposterior dimension of the upper airway might be related to OSAS severity and that the parameters (second subject) measured in a similar location did not show the same correlation. Accordingly, the airway-related parameters were excluded as discriminant variables (Table IV). As for SNB, even though it had no correlation with LSpO2, we selected it as a discriminant variable, simply because it has been widely used as a cephalometric parameter, and the author considered mandibular position to be a critical parameter reflecting OSAS severity. In summary, the following nine parameters were selected as discriminant variables: N-ANS, SNB, Max U, H-B, H-C3, T1-TT, T1-A, T1-B, and T1-PNS. Validation of selected cephalometric parameters. A variety of discriminant analyses, including the selected nine discriminant variables, were conducted to predict whether a participant belonged to the normal-to-mild AHI group or the moderate-to-severe AHI group. The combination of discriminant variables, including N-ANS, SNB, Max U, T1-TT, and T1-PNS, showed the most discriminant function (Figure 7). Evaluations of assumptions of linearity, normality, and singularity revealed no threat to the discriminant function analysis.

ORAL AND MAXILLOFACIAL SURGERY 26 Ryu et al.

OOOO January 2015

Table IV. Correlations among cephalometric measurements, polysomnographic indices, and anthropometric measurements Polysomnographic Skeletal-related S-N S-Go Go-Me S-Ba Ba-PNS ANS-PNS N-Me N-ANS ANS-Me SNA SNB ANB Go-Me/S-N Uvula-related PNS-UT MaxU U ang Hyoid-related H-A H-B H-MnP H-C3 Tongue-related T1-TT T1-A T1-B T1-PNS Airway-related NASu NASt PAS(PL) PAS(OL) PAS(ML)

Anthropometric

AHI

LSpO2

Weight

Height

BMI

NC

WC

.01 .20* .11 .11 .00 .13 .21* .22y .15 .07 .23y .20* .07

.02 .13 .12 .00 .03 .08 .10 .17* .03 .07 .12 .06 .00

.22* .41y .34y .23y .18* .24y .28y .29y .12 .03 .15 .15 .19*

.25y .50y .27y .37y .14 .17* .39y .36y .22* .03 .18* .18* .11

.08 .15 .20* .03 .11 .18* .08 .11 .01 .02 .05 .05 .14

.20* .40y .30y .21* .17 .26y .32y .28y .21* .03 .13 .12 e.16

.06 .29y .28y .14 .12 .1 .18* .13 .12 .02 .11 v.11 v.17*

.16 .19* .03

.15 .22y .02

.25y .31y .12

.17* .18* .06

.19* .25y .01

.36y .29y .03

.22* .21* .05

.29y .28y .21* .25y

.16 .28y .10 .20*

.44y .59y .39y .51y

.36y .31y .33y .39y

.29y .48y .25y .35y

.47y .57y .37y .61y

.37y .53y .29y .46y

.29y .32y .21* .35y

.17* .19* .17* .21*

.45y .40y .57y .33y

.29y .25y .35y .22*

.34y .30y .44y .24y

.40y .43y .53y .37y

.36y .36y .45y .38y

.01 .23y .11 .07 .18*

.15 .26y .11 .16 .29y

.11 .04 .01 .11 .07

.11 .28y .13 .12 .29y

.07 .33y .09 .02 .30y

.06 .30y .09 .12 .29y

.07 .03 .10 .01 .08

ANS, anterior nasal spine; PNS, posterior nasal spine; UT, uvula tip; MaxU, maximum uvula thickness; U ang, uvula angle; MnP, mandibular plane; TT, tongue tip; EpT, epiglottis tip; NAS, narrowest airway space; PAS, posterior airway space; AHI, Apneaehypopnea index; LSpO2, lowest saturation of peripheral oxygen. *P < .05. yP < .01.

The following discriminant functions effectively distinguished between the two groups: Eigen value (¼ 0.21), Wilk lambda (¼ 0.83), c2 (¼ 26.13) (P < .000; Table V). A canonical correlation is a multiple correlation between predictors and the discriminant function. This discriminant function explained 17.6% (¼ 0.422) of the AHI variance (Table V). Five input variables were identified as discriminant predictors scoring higher than  0.30. T1-PNS (0.78) was the most important discriminating factor. The patients belonging to the moderate-to-severe AHI group tended to have higher T1-PNS, T1-TT, Max U, and N-ANS and a lower SNB (Table VI). The unstandardized canonical discriminant function coefficients considered in the regression analysis are

shown in Table VII. Accordingly, the discriminant function formula (D) was: D ¼ 7:056  0:172½T1  PNS  0:153½SNB þ 0:141½Max U  0:048½T1  TT  0:030½N  ANS The group centroid is shown in Table VIII. The cutoff value was calculated as: 59  0:391 þ 81  0:536 59 þ 81 ¼ 0:145

Cutoff value ¼

If the discriminant value (Z) is more than e0.145, the selected participant can be assigned to the

OOOO Volume 119, Number 1

ORIGINAL ARTICLE Ryu et al. 27

Table VI. Discriminant loadings and standardized canonical discriminant function coefficients Predictor variable

Discriminant loadings

Standardized canonical discriminant function coefficients

T1-PNS SNB Max U T1-TT N-ANS

.776 .554 .370 .362 .311

.977 .564 .306 .362 .169

ANS, anterior nasal spine; Max U, maximum uvula thickness; T1, Tongue base; TT, tongue tip; PNS, posterior nasal spine.

Table VII. Unstandardized canonical discriminant function coefficients

Fig. 7. Selected discriminant variables. N-ANS (mid-face length): Distance from N to ANS; SNB: Angle between S-N and N-B; Max U (maximum uvula thickness): Maximum thickness of uvula perpendicular to uvula length; T1-TT (length of tongue): Distance from T1 to TT; T1-PNS: Distance from T1 to PNS.

Table V. Canonical discriminant functions among AHI and discriminant variables Discriminant analysis function Normal-to-mild versus moderateto-severe

Eigen Canonical ChiP value correlation Wilks l square df value .21

.42

.83

26.13

5

.000

moderate-to-severe AHI group, whereas if it is less than or equal to 0.145, the selected participant can be assigned to the normal-to-mild AHI group. Classification matrices for two-group discriminant analysis. Of the 140 participants, 102 (72.9%) were correctly assigned to the normal-to-mild and moderateto-severe AHI groups. The cross-validation of the produced discriminant function with a hold-out sample showed a hit ratio of 70.7% (Table IX; Figure 8). The proportional chance criterion was applied to validate this discriminant function.21 The calculated value was [(81/140)2 þ (1 e 81/140)2] ¼ 0.512. Because the accuracy of the discrimination (72.9%) was more than 51.2%, it could be concluded that the discriminant function analysis, including N-ANS, SNB, Max U, T1-TT, and T1-PNS, showed a high reliability.

DISCUSSION A variety of studies have been conducted to demonstrate the correlation between craniofacial and/or soft tissue abnormalities and OSAS-related parameters

Predictor variable T1-PNS SNB MaxU T1-TT N-ANS (Constant)

Unstandardized canonical discriminant function coefficients .172 .153 .141 .048 .030 7.056

ANS, anterior nasal spine; Max U, maximum uvula thickness; T1, tongue base; TT, tongue tip; PNS, posterior nasal spine.

using lateral cephalometric radiography. Skeletal measurements commonly involve the position of the hyoid, facial length, and the maxilla and mandible relative to the skull base. The most commonly used soft tissueerelated cephalometric parameters are tongue size, uvula length, uvula thickness, and posterior airway space (PAS). According to Riley et al., in normal patients, the hyoid bone was positioned 15.4  3 mm below the mandibular plane, the soft palate length was 37  3 mm, and the PAS, determined by a line drawn from point B through the gonion (Go) intersecting the base of the tongue and the posterior pharyngeal wall, was 11  1 mm.23,24 However, these normal measurements do not accurately reflect OSAS severity under clinical conditions. The present study conducted comprehensive measurements of cephalometric parameters seemingly associated with OSAS severity and correlated OSAS severity with the cephalometric measurements. In the skeletal-related measurements, anterior facial height (N-Me), N-ANS, and posterior facial height (S-Go) were positively correlated with AHI. Because N-ANS was correlated not only with AHI but also with LSpO2, we considered it to be the most valuable parameter. This finding was in contrast to those of other studies. Tangugsorn and Ahmet demonstrated, for example, that OSAS patients have a reduced mid-face length.25,26 However, these studies compared OSAS

ORAL AND MAXILLOFACIAL SURGERY 28 Ryu et al.

OOOO January 2015

Table VIII. Group centroid of discriminant function Group

Centroid

Normal-to-mild (n ¼ 59) Moderate-to-severe (n ¼ 81)

.536 .391

patients and healthy controls, not OSAS severity. Rather, as elongated facial morphology is related to airway length, a longer face seemingly correlates with OSAS severity. The anteroposterior position of the mandible is another crucial factor because the mandible has soft tissue that can obstruct the posterior pharyngeal airway directly during sleep. A positive correlation between mandibular retrognathia and OSAS severity has been reported by some authors,27-30 whereas other studies have found no evidence of mandibular retrognathia.12,31,32 In the present study, SNB was negatively correlated with AHI, and ANB was positively correlated. It seems, therefore, that retrognathic mandible has a correlation with OSAS severity. A posteriorly positioned mandible seemed to replace the tongue and hyoid inferoposteriorly in affecting the airway dimension. ANB does not indicate a true anteroposterior position of the mandible relative to the skull base but, rather, a mandibular position relative to the maxilla. In view of the mandibular anteroposterior position, SNB is more valuable than ANB. And whereas SNB had no correlation with LSpO2, we selected it nonetheless, simply because it has been a widely used cephalometric parameter, had a higher correlation than did ANB, and showed a positive discriminant loading in a preliminary analysis. Max U was correlated with AHI and LSpO2. A thick uvula indicates a higher chance of collapse of the nasopharyngeal airway. Although a normal length of soft palate has already been implicated,23,24 maximum uvula thickness had a higher correlation with OSAS severity than did length in this study. Most of the cephalometric parameters related to the hyoid and tongue were correlated with both AHI and LSpO2. These results reflect the fact that the positions of the hyoid and tongue are related to the PSG parameters. This finding was consistent with those of previous investigations.13,30,32-36 Inferoposterior displacement of the hyoid and tongue could cause narrowing of the upper airway and, thus, would be related to OSAS severity. Also, T1-PNS, which can represent the length of the upper airway, showed the highest correlation with AHI. The longer the length of upper airway, the greater is the chance of collapse there. The sizes of the soft palate and tongue, which comprise the anterior pharyngeal wall, were correlated with AHI and LSpO2. This means that posterior and/or inferior displacement of soft tissue, including the soft

palate and tongue, during sleep could pose a greater risk of narrowing of the upper airway. Moreover, if a patient with a thick soft palate and a large tongue has a retrognathic mandible, there will be a greater tendency toward obstruction of the upper airway. Most airway-related parameters, including the narrowest airway space and PAS, did not have any correlation with PSG parameters. Since lateral cephalometric radiography is performed with the patient in the upright posture, this condition cannot be deemed to accurately reflect the upper airway of patients in the lying-down posture during sleep. Additionally, because the muscle tone of the soft tissue around the upper airway ordinarily relaxes during sleep, it is difficult to reproduce the same condition while a subject is awake. Most studies evaluating correlations between cephalometric measurements and OSAS-related parameters have focused on the airway dimension seen on cephalograms. Due to difficulties in accurate measurement and the limitations of imaging superimposition and twodimensional imaging, airway-related parameters do not show consistent correlations with OSAS-related parameters. The majority of patients with OSAS display obstruction of the respiratory tract at the rear of the soft palate and/or tongue; accordingly, many sleep with their mouths open and complain of dehydration of the oral cavity due to mouth breathing during sleep. Therefore, evaluation of the lateral cephalometric parameters in the mouth-open state, as well as comparison between the mouth-open state and the mouth-closed state in lateral cephalometry, might be meaningful. Accordingly, further studies on the correlation between lateral cephalometric measurements in the mouth-open state and various OSAS-related parameters should be conducted. Many investigators have attempted to use regression analysis to identify, from lateral cephalometric measurements, the appropriate formula for evaluation of OSAS severity.37-40 Other studies have performed statistical analysis to evaluate and validate the prediction formula associated with anthropometry and clinical parameters.41,42 In the present study, through a correlation analysis of cephalometric measurements relative to PSG parameters, we isolated nine discriminant variables including N-ANS, SNB, Max U, H-B, H-C3, length of tongue (T1-TT), T1-A, T1-B, and T1-PNS. The discriminant function formula arrived at was D ¼ 7.056 þ 0.172[T1-PNS] e 0.153[SNB] þ 0.141 [Max U] e 0.048[T1-TT] e 0.030[N-ANS]. T1-PNS was the factor that contributed the most, followed by, in order, SNB, Max U, T1-TT, and N-ANS. As noted above, the length of the upper airway seemed to be a critical factor for prediction of OSAS severity. Even though hyoid-related factors had a significant relationship with PSG parameters, they were excluded

OOOO Volume 119, Number 1

ORIGINAL ARTICLE Ryu et al. 29

Table IX. Classification matrices for two-group discriminant analysis Predicted group membership Groups

Normal-to-mild n (%)

Moderate-to-severe n (%)

Total (n ¼ 140)

Original

Normal-to-mild Moderate-to-severe

44 (74.6) 23 (28.4)

15 (25.4) 58 (71.6)

Cross-validation*

Normal-to-mild Moderate-to-severe

44 (74.6) 26 (32.1)

59 (100.0) 81 (100.0) 72.9% 59 (100.0) 81 (100.0) 70.7%

Hit ratio 15 (25.4) 55 (67.9) Hit ratio *Cross-validation was performed only for those cases.

Fig. 8. Histograms showing distribution of discriminant scores for normal-to-mild and moderate-to-severe apneaehypopnea index (AHI) groups.

because they had shown negative discriminant functions in a preliminary analysis. The inferior and posterior positions of the hyoid seemed to be influenced by mandibular position and/or the size of soft tissue (e.g., tongue) and to be indirectly related to the PSG indices. By using five discriminating factors, 102 of 140 patients (72.9%) were correctly assigned to normal-tomild and moderate-to-severe AHI groups. The proportional chance criterion was 0.512 (51.2%). These discriminant models could potentially be useful in prioritizing patients for further evaluation (e.g., by PSG).

CONCLUSIONS We conducted a retrospective study of comprehensive correlations among cephalometric measurements, PSG indices, and the effectiveness of cephalometric parameters for OSAS screening. OSAS has a significant relationship with anatomic deformities of craniofacial and soft tissues. Of these, the length of the upper airway (T1-PNS) and SNB seemed to affect OSAS severity, whereas the anteroposterior dimension of the upper airway did not show any significant correlation with PSG indices. It seems that lateral cephalometric radiography, albeit inaccurate and nonreproducible for actual sleep situations, is an accessible and suitable tool

ORAL AND MAXILLOFACIAL SURGERY 30 Ryu et al.

for evaluation of craniofacial and soft tissue deformities related to OSAS severity. Further research on cephalometric parameters that reflect OSAS severity is needed in order to develop advanced diagnostic modalities for OSAS. This study was supported by the Dong-A University research fund.

OOOO January 2015

17.

18.

19. 20.

REFERENCES 1. Brouillette RT, Thach BT. A neuromuscular mechanism maintaining extrathoracic airway patency. J Appl Physiol. 1979;46: 772-779. 2. Tiner BD, Waite PD. Surgical and nonsurgical management of obstructive sleep apnea. In: Peterson LJ, Indresano AT, Marciani RD, Roser SM, eds. Principles of Oral and Maxillofacial Surgery. 6th ed. Philadelphia, PA: J.B Lippincott Company; 1992:1531-1548. 3. Sunitha C, Aravindkumar S. Obstructive sleep apnea: clinical and diagnostic features. Indian J Dent Res. 2009;20: 487-491. 4. Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328:1230-1235. 5. Centers for Medicare and Medicaid Services, US Department of Health and Human Services. Decision Memo for Continuous Positive Airway Pressure (CPAP) Therapy for Obstructive Sleep Apnea (OSA) (CAG-00093 R2). Available at: http://bit.ly/ hEQDIk. Accessed March 2010. 6. Patil SP, Schneider H, Schwartz AR, Smith PL. Adult obstructive sleep apnea: pathophysiology and diagnosis. Chest. 2007;132: 325-337. 7. Schwab RJ, Kuna ST, Remmers JE. Anatomy and physiology of upper airway obstruction. In: Kryger MH, Roth J, Dement WC, eds. Principles and Practice of Sleep Medicine. Philadelphia, PA: Saunders; 2005. 8. Goldberg AN, Schwab RJ. Identifying the patient with sleep apnea: upper airway assessment and physical examination. Otolaryngol Clin N Am. 1998;31:919-930. 9. Guilleminault C, Stoohs R, Clerk A, et al. A cause of excessive daytime sleepiness. The upper airway resistance syndrome. Chest. 1993;104:781. 10. Faber CE, Grymer L. Available techniques for objective assessment of upper airway narrowing in snoring and sleep apnea. Sleep Breath. 2003;7:77-86. 11. Won Tae-Bin. Contemporary methods of upper airway evaluation in obstructive sleep apnea patients. Korean J Otorhinolaryngol Head Neck Surg. 2013;56:7-13. 12. deBerry-Borowiecki B, Kukwa A, Blanks RH. Cephalometric analysis for diagnosis and treatment of obstructive sleep apnea. Laryngoscope. 1988;98:226-234. 13. Maltais F, Carrier G, Cormier Y, Series F. Cephalometric measurements in snorers, non-snorers, and patients with sleep apnoea. Thorax. 1991;46:419-423. 14. Riha RL, Brander P, Vennelle M, et al. A cephalometric comparison of patients with the sleep apnea/hypopnea syndrome and their siblings. Sleep. 2005;28:315. 15. Lyberg T, Krogstad O, Djupesland G. Cephalometric analysis in patients with obstructive sleep apnoea syndrome. I. Skeletal morphology. J Laryngol Otol. 1989;103:287. 16. Partinen M, Guilleminault C, Quera-Salva MA, et al. Obstructive sleep apnea and cephalometric roentgenograms. The role of

21. 22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

anatomic upper airway abnormalities in the definition of abnormal breathing during sleep. Chest. 1988;93:1199. Guilleminault C, Riley R, Powell N. Obstructive sleep apnea and abnormal cephalometric measurements. Implications for treatment. Chest. 1984;86:793. American Academy of Sleep Medicine. International Classification of Sleep Disorders. In: Diagnostic and Coding Manual. 2nd ed. Westchester, IL: American Academy of Sleep Medicine; 2005. Meoli AL, Casey KR, Clark RW, et al. Hypopnea in sleepdisordered breathing in adults. Sleep. 2001;24:469-470. Raj R, Hirshkowitz M. Effect of the new Medicare guideline on patient qualification for positive airway pressure therapy. Sleep Med. 2003;4:29-33. Taek-sub O. Data Analysis for Social Science. 3rd ed. Seoul, Korea: Nanam; 1988:371-403. Hair HF Jr, Anderson RE, Tantham RL, et al. Multivariate Data Analysis with Readings. 4th ed. Englewood Cliffs, NJ: Prentice Hall; 1995. Riley R, Guilleminault C, Herran J, Powell N. Cephalometric analyses and flow-volume loops in obstructive sleep apnea patients. Sleep. 1983;6:303. Riley R, Guilleminault C, Powell N, Simmons FB. Palatopharyngoplasty failure, cephalometric roentgenograms, and obstructive sleep apnea. Otolaryngol Head Neck Surg. 1985;93:240. Tangugsorn V, Skatvedt O, Krogstad O, Lyberg T. Obstructive sleep apnoea: a cephalometric study. Part I. Cervico-craniofacial skeletal morphology. Eur J Orthod. 1995;17:45-56. Tangugsorn V, Krogstad O, Espeland L, Lyberg T. Obstructive sleep apnea (OSA): a cephalometric analysis of severe and nonsevere OSA patients. Part I: Multiple comparison of cephalometric variables. Int J Adult Orthodon Orthognath Surg. 2000;15:139-152. Hochban W, Brandenburg U. Morphology of the viscerocranium in obstructive sleep apnoea syndromedcephalometric evaluation of 400 patients. J Craniomaxillofac Surg. 1994;22:205-213. Jamieson A, Guilleminault C, Partinen M, Quera-Salva MA. Obstructive sleep apneic patients have craniomandibular abnormalities. Sleep. 1986;9:469-477. Series F, St Pierre S, Carrier G. Effects of surgical correction of nasal obstruction in the treatment of obstructive sleep apnea. Am Rev Respir Dis. 1992;146:1261-1265. Andersson L, Brattstrom V. Cephalometric analysis of permanently snoring patients with and without obstructive sleep apnea syndrome. Int J Oral Maxillofac Surg. 1991;20:159-162. Battagel JM, L’Estrange PR. The cephalometric morphology of patients with obstructive sleep apnoea (OSA). Eur J Orthod. 1996;18:557-569. Zucconi M, Ferini-Strambi L, Palazzi S, Orena C, Zonta S, Smirne S. Habitual snoring with and without obstructive sleep apnoea: the importance of cephalometric variables. Thorax. 1992;47:157-161. Pracharktam N, Hans MG, Strohl KP, Redline S. Upright and supine cephalometric evaluation of obstructive sleep apnea syndrome and snoring subjects. Angle Orthod. 1994;64: 63-73. Battagel JM, Johal A, Kotecha B. A cephalometric comparison of subjects with snoring and obstructive sleep apnoea. Eur J Orthod. 2000;22:353-365. Frohberg U, Naples RJ, Jones DL. Cephalometric comparison of characteristics in chronically snoring patients with and without sleep apnea syndrome. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1995;80:28-33. Ozbek MM, Miyamoto K, Lowe AA, Fleetham JA. Natural head posture, upper airway morphology and obstructive sleep apnoea severity in adults. Eur J Orthod. 1998;20:133-143.

OOOO Volume 119, Number 1 37. Bacon WH, Krieger J, Turlot JC, et al. Craniofacial characteristics in patients with obstructive sleep apneas syndrome. Cleft Palate J. 1988;25:374. 38. Hoekema A, Hovinga B, Stegenga B, et al. Craniofacial morphology and obstructive sleep apnoea: a cephalometric analysis. J Oral Rehabil. 2003;30:690. 39. Dempsey JA, Skatrud JB, Jacques AJ, et al. Anatomic determinants of sleep-disordered breathing across the spectrum of clinical and nonclinical male subjects. Chest. 2002;122:840. 40. Kubota Y, Nakayama H, Takada T, et al. Facial axis angle as a risk factor for obstructive sleep apnea. Intern Med. 2005;44:805. 41. Rowley JA, Aboussouan LS, Badr MS. The use of clinical prediction formulas in the evaluation of obstructive sleep apnea. Sleep. 2000;23:929-938.

ORIGINAL ARTICLE Ryu et al. 31 42. Sharma SK, Malik V, Vasudev C, et al. Prediction of obstructive sleep apnea in patients presenting to a tertiary care center. Sleep Breath. 2006;10:147-154.

Reprint requests: Bok-Joo Kim, DDS, PhD Department of Oral and maxillofacial surgery Dong-a Univ. medical center Dongdaesin-dong 3-ga, Seo-gu Busan Korea [email protected]