E(CJW

oLpvMDDMv

va2Fan

o1

1

ffects of Mandibular Advancement DeviceMAD) on Airway Dimensions Assessed Withone-Beam Computed Tomography

ennifer A. Haskell, John McCrillis, Bruce S. Haskell, James P. Scheetz,illiam C. Scarfe, and Allan G. Farman

Upper airway constriction is an important contributing factor to obstructive

sleep apnea (OSA), which may be treated in a palliative manner with mandib-

ular advancement devices (MADs) to increase patency of the airway. It may be

the treatment of choice for affected individuals who cannot use a continuous

positive airway pressure device or who are not candidates for surgical correc-

tion of OSA. The specific distance applied during mandibular advancement,

however, is often arbitrarily determined. This project uses cone beam computed

tomography imaging in patients with OSA to determine a quantifiable relationship

between airway patency and mandibular advancement. This correlation may be

the basis to create an ideal technique to diagnose and treat patients having OSA.

Twenty-six subjects successfully treated for OSA with a MAD received 2 cone

beam computed tomography scans; 1 with and 1 without the MAD. Volumetric,

cross-sectional, and cephalometric measurements were gathered from these

scans. With the use of linear regression statistical analysis, specific predictor pa-

rameters have been identified for volumetric and cross-sectional airway informa-

tion. An average oropharyngeal volume increase of approximately 2800 mm3 was

achieved with MAD therapy. (Semin Orthod 2009;15:132-158.) © 2009 Elsevier Inc.

All rights reserved.

Itauueabsvaattegpapwv

Alumna, School of Dentistry and Graduate School, Departmentf Anatomical Sciences and Neurobiology, University of Louisville,ouisville, KY; Practitioner, Louisville, KY; Clinical Professor, De-artment of Orthodontics, School of Dentistry, University of Louis-ille, Louisville, KY and Div. of Orthodontics, Vanderbilt Universityedical Center, Nashville, TN; Professor, University of Louisville,epartment of Diagnostic Sciences, Prosthodontics, and Restorativeentistry, Louisville, KY; and Professor, Department of Oral andaxillofacial Radiology, School of Dentistry, University of Louis-

ille, Louisville, KY.The present study is currently under IRB approval at the Uni-

ersity of Louisville as of April 20, 2006. It has undergone 2nnual reviews and is not due for another annual review until April1, 2009. Risk to the subjects is from radiation exposure using aDA/CDRH approved device. The short and long-term risks of somaticnd genetic damage at the level used in this study are consideredegligible.

Address correspondence to Dr J. A. Haskell, Department of Orth-dontics, Eastman Dental Center, 625 Elmwood Ave., Rochester, NY4620; E-mail: jennyhaskell@gmail.com or bruce@haskellbraces.com

© 2009 Elsevier Inc. All rights reserved.1073-8746/09/1502-0$30.00/0

cdoi:10.1053/j.sodo.2009.02.001

32 Seminars in Orthodontics, Vol 15, N

n the present study, the authors sought todetermine a quantifiable relationship be-

ween mandibular advancement performed withremovable orthotic device and the resulting

pper respiratory airway dimensions and vol-me. Upper airway dimension has been consid-red a contributing factor to obstructive sleeppnea (OSA). This condition has been treatedy continuous positive airway pressure (CPAP),oft-tissue surgery, orthognathic surgical ad-ancement of the mandible or removable oralppliances directed at mandibular protrusion,nd patency of the pharyngeal airway to preventhe lumen from collapsing during sleep. Tradi-ionally, upper respiratory airway space has beenvaluated by the use of cephalometric radio-raphs; however, this method results in superim-osition of all bilateral structures of the skullnd only provides a 2-dimensional (2D) antero-osterior (AP) linear dimension. In this study,e analyzed a series of images obtained from 2olumetric data sets generated with the use of

one beam computed tomography (CBCT): with

o 2 (June), 2009: pp 132-158

aas(Sdmm

O

Stsiblidwgosc

eAafgtmwaaa�tmcclptmssdebt

C

TtOondpdoOaptbt

idiswHsmcctbsIptfce

tOwed

voocsm

133Mandibular Advancement and CBCT

nd without the use of a mandibular orthoticdvancement device. All images used in thistudy were generated with the i-CAT® CBCTXoran Technologies, Ann Arbor, MI/Imagingciences International, Hatfield, PA). This mo-ality allows an identification of skeletal land-arks seen in 2D cephalometry and also permitseasurements of airway space as a 3D volume.

verview of OSA

leep apnea is defined as a decrease in respira-ion, yielding hypoxia and hypercapnia duringleep. It can be caused by many factors, includ-ng those of either neurologic origin or physicallockage of the airway. This study focused on the

atter, more common, OSA. OSA physically lim-ts the amount of air that a person can inhaleuring sleep. It occurs in 4% of men and 2% ofomen. Many authors agree that polysomno-raphic data are needed for definitive diagnosisf OSA; however, there are several diagnosticymptoms and consequences of the disease, in-luding snoring as a very major indicator.

OSA causes episodic apneic or hypopneicvents during sleep, defined by the Americancademy of Sleep Medicine as 10 seconds ofbsent or decreased airflow. These events arerequently characterized by electroencephalo-raphic arousal and decreased oxygen satura-ion of hemoglobin. Severity of OSA is deter-

ined by an apnea/hypopnea index (AHI),hich measures the average number of thesepneic events per hour. Mild OSA is defined asn AHI between 5 and 15, moderate between 15nd 30, and severe OSA is defined as an AHI30. Although OSA is caused by an occlusion of

he upper airway, the level of obstruction inost OSA cases is the oropharynx.2 This is

aused by sleep-induced relaxation of the mus-les attached to the soft tissues that make up theumen of the oropharynx. It is now thought thatatency of this region is dependent on the ac-

ion of the oropharyngeal dilator and abductoruscles. The airway is subject to collapse when

ubatmospheric intraluminal pressure during in-piration overwhelms the stabilizing force pro-uced by these muscles.3 In OSA, the apneicvents are accompanied by a physical effort toreathe better, whereas in central sleep apnea,

here is no such effort to overcome the apnea. O

onsequences of OSA

he most serious and deadly consequences arehe cardiovascular diseases that can arise fromSA. Hypertension, tachycardia, increased risk

f cerebrovascular accidents, daytime hypercap-ia, atrial fibrillation, and even coronary arteryisease all have been associated with OSA. Hy-ertension is more prevalent in younger or mid-le-aged OSA sufferers than those aged 60 yearsr older.4 The pathogenesis of these effects ofSA is still being researched, but it is generally

ccepted that the intermittent hypoxia and hy-ercapnia that is caused by the apneic episodesriggers homeostatic compensation events in theody that lead to cardiovascular disease overime.

Many authors have investigated the character-stics, incidence, and pathophysiology of the car-iovascular consequences of OSA. Hypertension

s found in 50-90% of OSA patients.5 It has beenhown that nocturnal blood pressures of patientsith severe OSA can reach up to 240/120 mmg.5 This increase in blood pressure at night is a

pecific characteristic of OSA, because most nor-ally functioning individuals will have a de-

rease in blood pressure during sleep. Anotherharacteristic of OSA-related hypertension ishat the organs affected are more often therain and heart, whereas the kidneys tend to bepared, unlike in other forms of hypertension.6

t has also been noted that OSA hypertensiveatients have a greater blood plasma volumehan other hypertensive patients. Kawata et al.7

ound that daytime hypercapnia, an unexpectedonsequence, has an incidence approximatelyqual to 14% of OSA patients.

The Wisconsin Sleep Cohort Study showedhat there is a statistical relationship betweenSA severity and hypertension severity, and Ka-ata et al.6 showed that AHI and, to a lesserxtent, body mass index (BMI) are predictors ofaytime hypercapnia severity.

Beyond hypertension and its related cardio-ascular damage, atherosclerosis is also a sequelaf OSA. Patients with OSA have an increase ofxidative stress and a decrease in antioxidantapabilities.7,8,9 Oxidative stress also may be as-ociated with the increased amounts of inflam-atory cytokines and interleukins found in

SA, which in turn lead to an enhanced rate of

va

tdTa

D

Pmsmtmbpdoa

btohhncTmnn

nttgbadrttohtOtd

n

tafb

foBkgAtotv

coasmTamtteptaOFoiwtncv

M

Aobetdbcg

134 Haskell et al.

ascular damage, atherosclerosis and coronaryrtery disease.4

Airway obstruction can be caused by a multi-ude of different types of blockages on any of theifferent levels of the upper or lower airway.his study focused upon the anatomy of the orond velopharyngeal areas.9

iagnostic Problems in OSA

olysomnography is a sleep study that measuresany physiological variables associated with

leep and is used to diagnose OSA. Commoneasurements include oxygen saturation, elec-

rocardiography, air flow, respiratory effort, limbovement, eye and jaw muscle movement, and

rain electrical activity. Although polysomnogra-hy is considered the gold standard for OSAiagnosis, there are preliminary techniques thatften go ignored to determine whether to sendpatient for a sleep study.Although OSA is a common disease, it has

een estimated that between 80% and 90% ofhose with it go undiagnosed. In fact, many casesf what were initially thought to be “primaryypertension” were later found to be secondaryypertension caused by OSA. Rahaghi and Bas-er10 investigated the cause of this great defi-iency by interviewing known OSA patients.hey found that there was an average of 87onths that elapsed between the patient firstoticing a symptom of OSA, and being diag-osed with this condition.

Because of the large percentage of undiag-osed sufferers, preliminary examinations of pa-

ients should include screening for sleep apneao help reduce the number of OSA patientsoing untreated. There are several ways this cane done with varying ease and efficiency. Fabernd Grymer11 reviewed several techniques foriagnosis of OSA and rated them for ease, accu-acy, ability to be standardized, and cost-effec-iveness. One technique includes visual inspec-ion of the nose and pharynx to asses anybvious anatomic discrepancies. This method,owever, does not allow visualization inferior to

he oropharynx, so will not identify all cases ofSA. A more telling review of anatomic struc-

ures can be done with fiberoptic endoscopyuring wakefulness for a more detailed view.

The problem with both of these latter tech-

iques is that the patient is awake, and the struc- m

ures may change position during sleep causingn obstruction that goes unnoticed during wake-ulness. This is why a definitive diagnosis cannote made without a sleep study.11

Obesity is one of the most important riskactors of OSA. Obesity is defined as a BMI of 30r greater by the National Institute of Health.MI is calculated by dividing one’s weight inilograms by the square of height in meters toive a normalized measure a person’s body mass.lthough obesity is a very important risk factor,

he direct relationship between BMI and severityf OSA is poorly understood. One study foundhat the correlation between BMI and AHI isery low (r � 0.23).12

Fogel et al.12 set out to determine more spe-ific predicting factors than BMI within thebese population. They found that several vari-bles were positively correlated with severity ofleep apnea, such as airway collapsibility, and aore AP-oriented than laterally oriented airway.he latter finding is likely related to the result-nt mechanical disadvantage of the genioglossususcle. The function of this muscle is to pull the

ongue anteriorly, which increases patency ofhe airway. If the length of the muscle is short-ned, it does not have as great of an ability toerform this function, thus allowing the tongueo stay in a retruded position (i.e., blocking theirway). This is an example of the obstruction inSA at the level of the oropharynx. Additionally,

ogel et al.12 found that the pharyngeal volumef an individual experiencing more severe OSA

s lower when the lungs are at residual volume orhen there is less air in the lungs. This means

hat, just after exhalation, the danger of an ap-eic event is greatest and that there is a negativeorrelation between lung volume and OSA se-erity.

ost Common Types of Obstruction

lthough most obstructions in OSA occur at theropharynx, another common point of airwaylockage is the nasopharynx. Hypertrophied ad-noids are a common source of obstruction ofhe upper airway. This type of obstruction mayirectly cause OSA if the individual is not able toreathe through the mouth during sleep, or itan result in mouth-breathing. If a patient is stillrowing, mouth breathing can alter the develop-

ent of skeletal structures in a growing individ-

uamfiTo

glairmotojpea

iRObfgffiOctiitst

illijrptosp

l

clraot

T

C

TsimOttgosmoemtfpa

eotmsu

F

135Mandibular Advancement and CBCT

al.13 Major et al.13 found that there was at best,n r � 0.68 correlation between linear measure-ents of the upper airway in a 2D cephalometriclm and the diagnosis of upper airway blockage.hey suggest that the cephalogram should be usednly as a screening tool for airway obstruction.

The mechanism that decreases oropharyn-eal patency in cases with those skeletal anoma-ies associated with OSA has been highly debatednd researched. A skeletal Class II configurations caused by the maxilla being too far anteriorelative to the mandible. It may be caused by theandible being positioned too far posteriorly,

r simply underdeveloped. It may also be due tohe maxilla being positioned too far anteriorly,r any combination of these problems in either

aw. In skeletal Class II patients with OSA, theroblem is likely caused by a posteriorly-ori-nted mandible that is displacing the soft tissuesttached to it, impinging on the airway space.1

Johal et al.14 investigated the specific anatom-cal anomalies statistically contributing to OSA.etrognathia was found a contributing factor inSA, with patients having a shorter mandibularody length, with a significantly shorter distancerom their posterior pharyngeal wall to the lin-ual surface of their lower incisors. A shorter APace length in general also was a contributingactor. Therefore, their findings support bimax-llary surgical advancement as treatment for

SA, a treatment successfully used in severeases of OSA for the last decade.15 Many inves-igators also agree that a lower-positioned hyoids also a contributing factor.12,15,16 This anomalys related to posteriorly-displaced tongue becausehe muscles that connect the tongue to the hyoid,uch as the hyoglossus, would pull the tongue pos-eriorly when the hyoid is more inferior.

Soft tissue anomalies can also be a contribut-ng factor to OSA. Johal et al.16 found that aarger soft palate, a smaller pharynx, and aarger tongue were soft-tissue contributors. Anncrease in palate length was significant in sub-ects with OSA. The distance between the poste-ior aspect of the soft palate and the posteriorharyngeal wall was on average two-thirds that of

he control subjects. A large tongue size is an-ther contributing factor as it also decreases theize of the pharynx by causing the tongue to belaced posteriorly.15

The pharyngeal dilator muscles seem to be

ess effective in preventing pharyngeal lumen s

ollapse in these soft-tissue anatomical anoma-ies. Sher et al considered etiological factors byegions of the pharynx, the retropalatal regionnd the retrolingual region.17 There may be anbstruction in either or both of these regions ofhe pharynx.

reatment of OSA

ontinuous Positive Airway Pressure (CPAP)

he general aim of all treatment modalities inleep-breathing disorders is to facilitate breath-ng and thereby reduce the risk of increased

orbidity. Because there can be many causes ofSA, there are also several different treatment

ypes. They all focus on preventing collapse ofhe lumen of the pharynx during sleep. Theold standard for initial treatment is home usef a device called CPAP18 because it seems to beomewhat of a “cure-all” OSA treatment. Noatter where the obstruction in the pharynx

ccurs, this treatment has been shown to be veryffective in most OSA patients. In this treatmentethod, the patient wears a mask at night (Fig 1)

hat is attached to a machine that continuouslyorces air through the entire airway. This keeps theharynx patent due solely to increased air pressures if it is being blown up like a balloon.

This method has been shown to be highlyffective, but there are several negative aspectsf this type of treatment. The most important ishat many patients cannot tolerate the treat-

ent.18 Many complain of not being able toleep with a mask on their face or that it feelsnnatural to have air blown down their throat all

igure 1. CPAP treatment during sleep. (Color ver-

ion of figure is available online.)

tCndit

S

AmtTtgpms

hstrhsmorahceishtltw

gtatoTdtmbs

imrtp

batobbcoampwroffi

fnmobssdo

M

Fcsmabitdmdmsp

136 Haskell et al.

he time. Still others, who can fall asleep with thePAP in place, will rouse in the middle of theight and remove the device. Other reasons foriscontinuing CPAP therapy have been primar-

ly related to issues of nasal dryness and conges-ion, and difficulty adapting to the pressure.

urgical Treatments

lthough it has proven to be very effective inany cases of OSA, not all patients who are able

o tolerate CPAP will respond to the treatment.hese are often more severe OSA sufferers. For

hese patients, another treatment option is sur-ery. The guidelines for OSA surgery state that arerequisite for surgery candidates is that theyust be nonresponsive to CPAP or other non-

urgical OSA treatments.19

There are many different types of surgery thatave been used to treat OSA. They range fromoft tissue to osteotomy surgeries. The older ofhe 2 is soft-tissue surgery. The uvulopalatopha-yngoplasty (UPPP) was introduced in 1981 andas been the most popular form of soft tissueurgical treatment of OSA. In this type of treat-ent, part of the soft palate and surrounding

ropharyngeal tissues are surgically resected toeduce their potential for obstructing the upperirway during sleep. This type of treatment hasad about a 50% success rate.17,20 Sher, et al.17

oncluded that a low success rate is likely if thetiology of the problem is a pharyngeal narrow-ng or collapse. The OSA may simply not re-pond to a reduction of the soft palate alone. Itas also been anecdotally reported that this

reatment, even with postsurgical success, hasead to relapse after several years when the softissue has grown back, or the patient has gainedeight.

Because of the limitations of soft-tissue sur-ery in the treatment of OSA, orthognathicreatments have been used. Placing the anteriorspect of the mandible more anteriorly pulls theongue forward and away from the posterior wallf the oropharynx, thus opening the airway.hese procedures include inferior sagittal man-ibular osteotomy and maxillomandibular os-

eotomy and advancement. Much more recently,axillary and mandibular expansion has also

een found to be effective in the treatment of

leep apnea.1,19 a

An orthognathic surgical treatment for OSAs bimaxillary, or maxillomandibular, advance-

ent. This procedure advances both jaws ante-iorly with the use of bilateral sagittal split os-eotomy on the mandible and a Le Fort Irocedure on the maxilla.

Bimaxillary surgical advancement of the jawsrings forward all soft tissue attached to themnd the hyoid bone. MAD devices only advancehe mandible. The difference is that the dentalcclusion can be controlled in the surgicallyimaxillary advancement because both jaws areeing moved. This way, the original occlusionan be maintained, or a malocclusion of skeletalrigin corrected. This procedure is aggressivend it can correct OSA due to obstructions atany different levels of the oropharynx. It also

revents the immediate postoperative period oforsening OSA that is found in soft-tissue pha-yngeal surgeries. Because this area has not beenperated upon, there is no edema that can causeurther obstruction. Rather, the swelling is con-ned to the soft tissue of the face.19

Another surgical treatment that has beenound to coincidentally improve OSA but wasot initially used as a primary OSA treatment isandibular and maxillary transverse distraction

steogenesis. When used in conjunction withimaxillary advancement, this treatment can re-ult in a significant increase effect on the dimen-ions of the oropharynx,1 as a decreased lateralimension of the airway is a very common etiol-gy of OSA in obese patients.

andibular Repositioning

or patients with mild-to-moderate OSA whoannot tolerate CPAP treatment, orthognathicurgery may be too aggressive a form of treat-ent. An option for these patients is a remov-

ble oral appliance that repositions the mandi-le forward. The success of this type of treatment

s based on a somewhat similar response of theissues to that of orthognathic surgery. Theseevices have their effect because of the attach-ent of the mandible to the tongue, pharyngeal

ilator muscles, and indirectly the soft palate. Byoving the mandible forward, it brings these

tructures that make up the lumen of the oro-harynx forward as well, thereby increasing the

irway space.21-28

pmcrOpcmo

puecdotttm

3iptostacfewt

stp

itOnntsh

RO

C

TdlawrtftbabmthoF

s

F(

137Mandibular Advancement and CBCT

Many authors now agree that mandibular re-ositioning treatment with an oral appliance hasany advantages and should be among the first

onsidered choices for treatment on a wide va-iety of patients, including patients with severeSA, if an optimal amount of advancement isossible.22,29-31 Fransson et al.30 reported a greatapacity for the patient to adapt to usage, withinimal negative temporomandibular disorders

r dental changes after a 2-year period.The Food and Drug Administration has ap-

roved the use of these devices only for individ-als 18 years and older. There are many differ-nt types of appliance design, but no definitiveriteria exist either for the ideal amount of man-ibular protrusion or for the amount of verticalpening associated with it. The success rates ofhese devices vary. Most often, a patient must beitrated to the amount of protrusion that is op-imal for that individual. This process involves

uch trial and error.The removable Herbst appliance (Figs 2 and

) is one type used for mandibular advancementn patients with OSA. This appliance is com-osed of 2 acrylic splints, one that fits over the

eeth in each arch, which are attached to eachther by a pistonlike connector. This pistonerves to push the mandible forward relative tohe maxilla, while a rod slides inside a tube tollow the patient to open and close. This devicean be adjusted by the patient to more or lessorward extension, which is called titrating toffectiveness. Movement of the mandible for-ard is done in small increments with the pa-

ient is instructed to continue until the OSA

igure 2. Herbst appliance: anterior view. (Color ver-

uion of figure is available online.)

ymptoms have diminished, or to stop the ac-ivation it if too much discomfort in the tem-oromandibular joint is felt.

A study conducted with removable Herbstn patients with OSA compared the before-reatment and after-treatment AHI values inSA sufferers and found that the Herbst sig-ificantly improved the AHI by up to 34 ap-eic events. The cephalometric analysis of

hese patients revealed that those who re-ponded the most had a shorter mandible-to-yoid distance.21,22,23,30,31,32,33

adiographic and Other Imaging of theropharynx

ephalometrics in OSA Imaging

he imaging of the upper airway space has tra-itionally been accomplished with the use of

ateral cephalometric radiography (Fig 4). Andvantage of this type of imaging is that it isidely used and readily available. It also uses aelatively low radiation dosage. However, imagesaken from a lateral viewpoint give only 2D in-ormation. This information is valuable becausehe AP dimension is that which is most likely toe changed with mandibular protrusion. It is notn ideal imaging technique for an OSA studyecause any changes in the medial-lateral di-ension cannot be measured. This is an impor-

ant dimension because previous researchersave found that a decreased lateral dimension isften present in patients with OSA.12 The only

igure 3. Herbst Appliance: posterosuperior view.Color version of figure is available online.)

seful measurements that can be made in ceph-

amvmmoepyl

rpmOfnedptcwwssspa

O

Trc

porto

aantwbVaaagimacci

mewrwttb

spbcdMiots

C

Boipv

FwRa

138 Haskell et al.

lometry for OSA and all other 2D imagingethods are linear or angular, and therefore

olume cannot be accurately assessed. Further-ore, cephalometry produces problems withagnification, distortion, and superimposition

f structures.21 Another limitation of cephalom-try in diagnosing and treating OSA is that theatient must be upright. Observation of the phar-nx when the patient is supine, as they are moreikely to be during sleep, cannot be visualized.9

Johal and Conaghan14 attempted to find aelationship between maxillary morphology inatients with OSA by using cephalometrics. Theyade cephalometric lateral radiographs of 13SA patients and 18 control subjects. They

ound that their population did not have a sig-ificant difference in frequency of Class II skel-tal patterns. All subjects with OSA had reducedistance from the posterior nasal spine (PNS) toosterior pharyngeal wall, indicating a constric-

ion in the airway by a retruded maxilla. Theranial base was shorter and the palatal angleas significantly more obtuse in male patientsith OSA only. This finding indicates that the

oft palate was further inclined into the airwaypace in these instances. In addition, there wereignificant differences in palatal height at theremolar and molar points, suggesting that pal-tal vaulting contributes to OSA.

ther OSA Imaging

o overcome the limitations of the static, up-ight images found in cephalometry, fluoros-

igure 4. Lateral cephalometric skull radiographith airway outlined in red (with permission fromMO, Corp., Denver, CO). (Color version of figure isvailable online.)

opy has been used to observe changes in the f

harynx on patients in a supine position.21 Flu-roscopy also has the advantage of being able toecord the dynamic changes in the pharynx ashe mandible is moved, as opposed to static viewsf the pharynx in different positions.21

For volumetric and area measurements of theirway, the technique of acoustic reflection haslso been used for nearly 25 years. In this tech-ique a rhinometer and pharyngometer wave

ube and nosepiece or mouthpiece generate soundhich is reflected in the airway and recordedy microphones within the wave tube. In 2002,iviano34 published a review article on theccuracy, reproducibility, and usefulness ofcoustic reflection in OSA airway assessment. Itsccuracy has been compared favorably to radio-raphic and magnetic resonance (MR) 3D imag-ng and it has the unique ability to localize the

inimum or maximum cross-sectional area of theirway. The limitation of acoustic reflection is thathanges in the airway cannot be localized to spe-ific anatomic structures as they can in 3D imag-ng.

In 2002, Sanner et al.8 conducted an experi-ent using MR imaging (MRI) to image the

tiology of pharyngeal obstruction in patientsith OSA. They found that in the patients notesponding to treatment, the area of obstructionas in the velopharynx only, whereas other pa-

ients who responded to treatment had obstruc-ions in the glossopharynx (oropharynx) only oroth the glossopharynx and velopharynx.

Although MRI may be more useful in under-tanding the role of soft-tissue in OSA than com-uted tomography (CT), it may be unwarrantedecause of the extremely high cost. In manyases, it would actually be less expensive to tryifferent treatments for OSA than to use theRI to predict effectiveness. An answer to 3D

maging of the airway may lie in CBCT technol-gy, which does not show clear delineations be-ween soft tissues, but clearly shows the airwaypace and related skeletal structures.

onventional CT Versus CBCT

oth conventional CT and CBCT have a sourcef X-rays that spin around the subject, collecting

nformation from all directions, which a com-uter then uses to compile a 3D image. A con-entional CT has a fan-shaped beam originating

rom a high-output rotating anode generator

tfori3ott2Cit

C

R

Aotmppattiap1

ws

I

Afhaptsv0cI(Ja

2crtaas

R

Bm

Fmn

Fs

139Mandibular Advancement and CBCT

hat moves around the subject many times in theorm of a spiral, projecting through the subjectnto a series of solid-state image detectors ar-anged in a 360° array to collect several slices ofmages that the computer later stacks to form aD image.34 The CBCT has a cone-shaped beamriginating from a low-energy fixed anode tubehat is projected through the subject to an at-ached single solid-state or amorphous siliconD panel detector that rotates with the beam.BCT can collect a much greater amount of

nformation from the subject in a single rotationhan the conventional CT.34,35

BCT Technology

adiation Dosage

main advantage of using CBCT to image theropharynx is the relatively low dosage of radia-ion. Several studies have been done to deter-

ine the exact radiation dosage to differentarts of the body of different types of radiogra-hy.36-38 It has been found that the general dos-ge of CBCT dosimetry is up to 50 times lesshan spiral CT. CBCT uses only minimal radia-ion, equal to 7 panoramic exposures, or approx-mately 3.5 days of background exposure. This is

huge reduction from spiral CT, which has arobability of causing stochastic effects in up to

igure 5. Lateral skull view of a 3D reconstructionade by 3DVR from the iCAT (Imaging Sciences Inter-ational). (Color version of figure is available online.)

45.6 examinations per million taken in the v

hole body, or up to 199 per million in thealivary glands.36,37

mage Quality and Accuracy

lthough soft tissue is not clearly delineatedrom other soft tissue on CBCT, it clearly showsigh contrast between bone, teeth, empty space,nd soft tissue in general. It is ideal to show theatency of the airway, related to the position ofhe hard tissue structures of the skull.17,34 Thepatial resolution is also much greater than con-entional CT, with a voxel resolution between.09 and 0.25 mm.38 Some programs that canreate these 3D images are 3DVR (Danaher/maging Sciences International, Hatfield, PA)Figs 5 and 6), Anatomage (Anatomage Inc., Sanose, CA), and Dolphin 3D imaging (Dolphin Im-ging/Patterson Dental, Chatsworth, CA).

CBCT images can be used to make accurateD simulations of lateral cephalometrics, APephalometrics, panoramic images, and arthrog-aphy.40 The resulting images are more accuratehan those traditionally made using lateral ceph-lometric radiographs. These CBCT imageschieve this accuracy by isotropic images recon-tructed with cone beam technology.39

econstructing and Measuring the Airway

ecause the value of measuring the airway di-ensions in patients with OSA has been recog-

igure 6. Anterior airway space view of a 3D recon-truction made by 3DVR made by iCAT CBCT. (Color

ersion of figure is available online.)

nsctaaMpsmtSspiviat

aFpaPvasnst7apt

OwMootoOtdttosg

E

Palsclm

swiMOadunm

e1damtosvtdacctn

T

Twwptgtocfir

140 Haskell et al.

ized, there have been many studies that mea-ure the airway in this population. The mostommon measurements comparing OSA pa-ients with normal airways made in 3D imagingre minimum surface area of the oropharynx,nd the AP and lateral lengths of this area.41

ayer et al.42 found that the airway in OSAatients with greater BMI tends to be morepherical, with a shorter lateral length. Severalorphometric techniques have been developed

o obtain precise measurements on 3D images.hi et al.43 created an algorithm to calculateeveral parameters based on the contrast of theixels of the airway to the surrounding soft tissue

n CBCT images. They measured total airwayolume, smallest trans-sectional area, largest sag-ttal view airway area, and smallest cross-areasnd anterior-posterior distances of the retropala-al and retroglossal space.

Ogawa et al.41 defined the upper and lowerspects of the airway by 2 lines parallel to therankfort Horizontal; 1 through the most distaloint of the hard palate, and 1 through the mostnterior-inferior point of the second vertebrae.revious authors41,43 measured the total airwayolume, the smallest cross-sectional area, andnterior-posterior and lateral lengths of themallest cross-sectional area. Ogawa et al. alsooted the general shape of the smallest cross-ectional area as being closest to rounded, ellip-ical, square, or concave. They found that in0% of OSA cases, the smallest cross-sectionalrea of the airway was below the occlusallane, and more often had an elliptic shape ofhis area.

Peh et al.44 measured the airways of ChineseSA patients with conventional CT because theyished to assess the airway in a supine position.ost of the Asian OSA population (62%) is not

bese. It is believed that craniofacial morphol-gy plays a more major role in OSA etiology inhis population than does soft-tissue morphol-gy. There is also a higher incidence of severeSA in the nonobese patients in this popula-

ion.45 Peh et al.44 found statistically significantifferences between OSA and control subjects in

he hyoid position, the nasal cavity length, theongue length (vallecula to tongue base distance),ropharyngeal airway space, posterior airwaypace, and the velopharyngeal and hypopharyn-

eal cross-sectional areas. m

ffect of Mandibular Advancement on OSA

revious studies have concluded that graduallydvancing the mandible forward to an optimalevel have yielded good responses from theupine patient via a reduced AHI, which isalculated upon the basis of electroencepha-ographic, elector-oculographic, and electro-

yographic measurements.25-32

There have been several studies that mea-ured the oropharynx dimensions in patientsith OSA, but most of the studies using 3D

maging, such as CBCT, conventional CT, orRI have compared OSA airways to their non-SA counterparts.41 Of the studies evaluating

ctual changes in the pharynx caused by man-ibular advancement, most of the studies havesed cephalometry or other 2D imaging tech-iques that do not show the transverse or volu-etric dimensions of the pharynx.In 2004, Ogutcen-Toller et al.2 conducted an

xperiment looking at changes in the airway in5 snoring subjects who had not previously beeniagnosed with OSA. They made a customizedcrylic block for each subject that advanced eachandible 3 mm less than maximum protrusion

o an average advancement of 2.39 mm andpen 7 mm by interincisal distance. By using CTcans, 1 with and 1 without the mandibular ad-ancement device, they measured the cross-sec-ional area of the airway. The only dramaticallyiffering measurement between the appliancend nonappliance scans, was the minimumross-sectional area of the airway, which in-reased by 60 mm2 or 72%. It was not known ifhe subjects had OSA, as snoring by itself doesot indicate sleep apnea.

he Present Study

he present study used CBCT to image the air-ay of successfully treated OSA patients with andithout a mandibular advancement device inlace, and measured parameters which addresshe patency of the oropharyngeal, velopharyn-eal, and hypopharyngeal airway. It was in-ended to try to understand the effect of OSArthotic appliances on airway dimensions, in-luding cross-sectional areas and volume bynding cephalometric and 3D anatomical cor-elates for the efficacy of mandibular advance-

ent treatment in specific cases of OSA.

M

Thaei(gb1snosHwpbalRvC

I

Ts

Fswtoabtt

I

Ttapw3OoppTtpw(t

141Mandibular Advancement and CBCT

ethods

wenty-six patients diagnosed with OSA whoad been previously treated with mandibulardvancement device (MAD) therapy by a gen-ral dental practitioner with advanced trainingn the diagnosis and treatment of OSA patientsJ.M.) were recruited for the study followinguidelines designated by the intuitional reviewoard of the University of Louisville. There were7 men and 9 women in this study group. Eachubject underwent polysomnography to diag-ose their OSA, defined as an AHI greater thanr equal to 5. Each subject was treated with theame type of MAD, specifically a removableerbst appliance, to act as a constant. The MADas titrated to effectiveness according to theatient feeling relief of symptoms. This was doney the patient adjusting the pistons to a greaternd greater extent until the symptoms were noonger noticed. Each patient was referred to theadiology and Imaging Science Division, Uni-ersity of Louisville School of Dentistry forBCT, and informed consent was obtained.

maging

wo CBCT scans were performed with the Clas-ic iCAT on each awake subject seated upright.

Figure 7. Orientation screen for Dolphin 3D Imagin

or both scans, the 22 cm, extended field of viewetting was used during a 20 � 20 seconds scanith a 0.4-mm resolution. One was made without

he appliance, serving as a control and the sec-nd was made with the removable Herbst OSAppliance in place. The image detector andeam were positioned to maximize coverage ofhe upper airway to approximately the level ofhe fifth cervical vertebra.

mage Generation

he CBCT volumetric datasets were imported inhe format of single-file DICOM files into a pre-lpha version 11.0 of Dolphin 3D Imaging (Dol-hin Imaging and Management Solutions, Chats-orth, CA), a software package that reconstructsD images and facilitates making measurements.nce imported, the 3D reconstructions were

riented so the Frankfort horizontal plane wasarallel to the axial plane. The mid-sagittallane was oriented to the midline of the subject.he coronal plane was oriented so that it passed

hrough both the left and the right porionoints. In cases of asymmetry, the orientationas made as close as possible to these guidelinesFig 7). Once the image was oriented properly,he software was used to create a 2D simulated

g. (Color version of figure is available online.)

lrowTptwsaebtfis

dusoAaasewfiS

I

C

TlmtptcMf

wotoclt

1

142 Haskell et al.

ateral cephalometric image with the use of aay-sum technique. The options were set to anrthogonal projection type and a 100-mm ruleras placed in the cephalometric image (Fig 8).he airway analysis tool was used to define theortion of the airway of interest. This portion is

he velopharynx and the oropharynx, definedith the superior border being the edge of the

oft palate to the posterior of the pharynx (par-llel to FH). The inferior border is the tip of thepiglottis on a plane parallel to FH. The borderetween the velopharynx and the oropharynx ishe occlusal plane. The area of interest was de-ned by a clipping box and seeds in the airwaypace (Fig 9).

Once the portion of the airway of interest wasefined, the airway analysis tool calculated vol-me and cross-sectional areas of differentlices of the airway. It also generated JPG filesf cross-sectional areas of interest (Fig 10).xial-view images of the smallest cross-sectionalrea (SmCa), largest cross-sectional area (LgCa),nd area of the slice at the level of the anterioruperior aspect of C2 (C2Ca) were saved andxported into Image J for analysis. This processas performed for all subjects for both CBCTles, with and without the device, i.e., S1-In and1-OUT.

Figure 8. “Build X-rays” tool in Dolphin 3D Imagin

mage Analysis and Data Generation

ephalometric Measurements

he lateral cephalogram simulated image al-owed the measurement of classic cephalometric

easurements with the Dolphin Program. A cus-om cephalometric analysis was created for theurpose of this study. The measurements wereested by measuring the same image on anotherephalometric analysis package, RMO Joe (Rockyountain Orthodontics, Inc., Denver, CO), and

ound to be the same.Two series of cephalometric measurements

ere made. The first series described the anat-my of the subject on the non-moving aspect ofhe craniofacial complex. The second series hadne or more points on the mandible that showhange when the Herbst was in place. The fol-owing specific measurements were made due tohe accompanying rationale:

Anatomic Descriptive Measurements

. Cranial deflection N-S-Ba: this angle denotesthe relative angulation of the posterior andanterior cranial base. The measurement ofthis kyphosis may impact upon the nasopha-ryngeal airway. A more acute angulation maycompress the airway by having a skull shape

g. (Color version of figure is available online.)

2

3

4

of in

Fo

143Mandibular Advancement and CBCT

that is more vertical than elongated (norm:129.6 degrees; standard deviation [SD]: 5 de-grees).

. Ba–S–PNS: this may be used to determine thehorizontal position of the hard and soft pal-ate. The more acute the angulation, the morepotential for diminished airway the AP plane(norm: 63 degrees; SD: 2.5 degrees).

. Linder-Aronson (Rocky Mountain Orthodon-tics, Diagnostic Services Handbook CourseSyllabus, 1989, Denver, CO) (1): the distancefrom PNS to the nearest adenoid tissue in a

Figure 9. Pink area denotes defined airway portion

igure 10. Axial slice of CBCT scan exported to Image

f figure is available online.)

line from PNS to basion. This is consideredan excellent, but limited expression of airwayavailability in the horizontal AP plane. It is anindicator of potential airway obstruction(norm: dependent on age, sex, ethnic origins,size of tissue mass).

. Linder-Aronson (2): another indicator of po-tential airway obstruction, measured by thedistance from PNS to the nearest adenoidtissue in a line from PNS perpendicular tosella-basion (norm: Also dependent on age,sex, ethnic origins, size of tissue mass).

terest. (Color version of figure is available online.)

nk shows cross-sectional area of airway. (Color version

J. Pi

Atu

1

2

3

4

5

6

7

8

9

1

Esm

A

Twsmacta(cTausw“

t

F

144 Haskell et al.

Measurements of Mandibular Change (Withand Without MAD)

ll of these were altered with the movement ofhe mandible forward and obliquely downwardsing the MAD.2

. Lower facial height: ANS-Xi-PM: Describesthe vertical relationship of the mandible andmaxilla. Low values indicate a skeletal deep-bite, a factor known to be contributory toOSA (norm: 45 degrees; SD: 4.0 degrees) (Xiis mandibular foramen).

. Posterior facial Height: sella-gonion: De-scribes the vertical dimension of the ramus tothe cranium (in mm).

. Ramal height: From gonion to CF: measuredfrom the posterior border of the pterygo-maxillary fissure to the angle of the mandible(norm: 54.8-mm; SD 3.3 mm).

. Anterior facial Height: nasion—menton (inmm).

. Saddle angle: nasion-sella to sella-articulare: .Low values could indicate a forward positionof the mandible. The relative position of themandible in space may well have an influenceon the airway patency and /or the relativeadvancement required for a MAD to be effec-tive (norm: 123.0 degrees; SD: 3.0 degrees).

. Facial axis: measured by the angle formed bythe plane CC to gnathion and the basion—nasion plane. The placement of the Herbstwill show the oblique change in angulation ofthe mandible. The degree of up and forwardor down and back positioning may be impor-tant to evaluate effective OSA symptom re-duction (norm: 90 degrees; SD: 3.5 degrees).

. SNB: the angular position of the mandible tothe cranium as measured from sella—na-sion—B point (norm: 80 degrees; SD: 3.7mm).

. ANB: measured by the angle formed by theplanes nasion—point A and nasion—point B(norm: 2.0 degrees).

. Ramus–Xi position: measured by the angleformed by the planes CF—Xi (mandibularforamen) and Frankfort horizontal. It de-scribes the horizontal position of the ramusunaffected by vertical change (norm: 76 de-grees; SD: 3.0 degrees).

0. Horizontal movement: anterosuperior as-pect of C- 3 to pogonion: the distance as

measured in mm to show the forward dis- fi

placement of the mandible using the MAD(Fig 11).

ach of the aforementioned cephalometric mea-urements was made 3 times for each imageade from each subject.

irway Measurements

he airway analysis tool in Dolphin 3D Imagingas used to find the volume (Fig 12) and cross-

ectional areas of the airway. It is able to find theinimum cross-sectional area in the part of the

irway that had been defined, and the other 2ross sections were found by scrolling throughhe axial slices. The volume and cross-sectionalreas from the smallest cross-sectional areaSmCa), largest cross-sectional area (LgCa), andross-sectional area at C2 (C2Ca) were recorded.he airway was defined 3 times, and the volumend cross-sectional areas were found for eachnique airway definition. Each defined airwaypace can vary based on the sensitivity settings,here the “clipping box” and the imagingseeds” were placed.

Finally, for each airway definition, the loca-ion of the smallest cross-sectional area was

igure 11. Horizontal movement. (Color version of

gure is available online.)

nps

tbsMoAon

d

Fi

Fo

145Mandibular Advancement and CBCT

oted as either above or below the occlusallane. The rationale for using these 3 cross-ections is explained below:

(SmCa): the smallest cross section (Fig 13) isthe point where the airflow is theoreticallymost constricted. This point is a limitingfactor in airflow.

(LgCa): the largest cross section shows themaximum amount the airway has ex-panded

(C2Ca): this cross section is measured at aconstant level (Fig 14), whereas the loca-tion of the smallest and largest can shift.

igure 12. Airway volume reconstructed and measures available online.)

igure 13. Smallest cross-sectional area marked on vol

nline.)

This measurement will show how the areaat the same point will change.

It was noted for each scan whether the loca-ion of the smallest cross section is above orelow the occlusal plane, as some authors havehown that patients are more likely to respond toAD therapy if the smallest cross section of the

ropharynx is located below the occlusal plane.value of (1) is given if it is below, and a value

f (2) is given if it is above. This is the onlyonparametric data measured in the study.

The images exported to Image J, a publicomain Java image processing program, at

Dolphin software, version 11. (Color version of figure

tric reproduction. (Color version of figure is available

d in

ume

hmoia(Ecoot

otiem

12345

F rea i

F(

146 Haskell et al.

ttp://rsbweb.nih.gov/ij/, are used to make lineareasurements of airway cross-sections. The scale

n Image J is calibrated to the ruler marks on themage. A linear measurement in mm is then madet the largest AP dimension and the largest lateralL) or transverse dimension (Fig 15). Microsoftxcel (Microsoft, Seattle, WA) is then used to cal-ulate the L:AP ratio to quantify the shape of thepening. The larger the ratio, the more elongatedr elliptic is the shape of the cross section, whilehe smaller the ratio, the more circular. The shape

igure 14. Arrow points to cross section of C-2. Pink a

igure 15. Airway area of largest cross section (LgCa

L-LgCa) linear dimensions of this cross section shown in

f the airway is important because previous au-hors have described that it influences the collaps-bility and expandability of the airway. In summary,ach cross section has the following measure-ents:

. area

. lateral dimension in mm (L)

. anteroposterior dimension in mm (AP)

. L:AP ratio

. Measurements of airway cross section

s (C2Ca). (Color version of figure is available online.)

own in pink. Anteroposterior (AP-LgCa) and lateral

) sh yellow. (Color version of figure is available online.)

S

Omcwda(rcrmcr

dclctmccmd

fi

t

aeewwwt

R

TdopclsdtCdo

tlCmmftpsu

taf

147Mandibular Advancement and CBCT

tatistical Analysis

nce all 3 readings of each measurement wereade, the mean was found, the difference cal-

ulated between the mean measurements madeith and without the appliance in place, and theata were imported into SPSS, a statistical datanalysis software package produced by SPSS Inc.Chicago, IL). SPSS was used to perform linearegression statistical analysis. A reliability intra-lass correlation was also computed using theaw data, and comparing the 3 measurementsade, rather than using the averages. It was

omputed for all variables used in the linearegression analyses.

For the descriptive analysis, means, standardeviations, skewness, and kurtosis values wereomputed for all parametric variables. Multipleinear regression was then used to assess theontribution of each predictor in determininghe value of the outcome measure for the para-

etric outcome variables. Linear regression washosen to evaluate whether the changes in theephalometric measurements based on the move-ent of the mandible with the Herbst could pre-

ict the outcomes of the changes in the pharynx.The predictor variables in this study were the

nal changes measured in:

Posterior facial height,Ramal height;Anterior face height;Saddle angle;Facial axis;Ramus position;Horizontal movement.

he outcome measures were changes in:

VolumeSmCa: smallest cross-sectional areaAP-SmCa: AP linear dimension of the smallest

cross-sectional areaL-SmCa: lateral linear dimension of the small-

est cross-sectional areaL:Ap-SmCa: ratio of the lateral to the anteri-

or-posterior linear dimensions of the small-est cross-sectional area

LgCa: largest cross-sectional areaAP-LgCa: AP linear dimension of the largest

cross-sectional areaL-LgCa: lateral linear dimension of the largest

cross-sectional area f

L:AP-LgCa: ratio of the lateral to the AP lin-ear dimensions of the largest cross-sec-tional area

C2Ca: cross-sectional area at the level of C2AP-C2Ca: AP linear dimension of the cross-

sectional area at the level of C2L-C2Ca: lateral linear dimension of the cross-

sectional area at the level of C2, andL:AP-C2Ca: ratio of the lateral to the ante-rior-posterior linear dimensions of thecross-sectional area at the level of C2.

The usefulness of the set of predictor vari-bles was determined by testing the null hypoth-sis that R2 was equal to zero. If the null hypoth-sis was rejected, the set of predictor variablesas examined to determine which 1 or onesere statistically significant predictors. In thisay, the predictor variables having an impact on

he outcome measure were identified.

esults

he linear regression showed that there were 6ependent, or “outcome,” variables measuredn the volumetric scan that were found to beredictable by independent variables. These out-ome variables were Volume of the oropharynx,argest cross-sectional area (LgCa), the cross-ectional area at C2 (C2Ca), the lateral linearimension of the cross section at C2 (L-C2Ca),he AP linear dimension of the cross section at2 (AP-C2Ca), and the ratio of these 2 linearimensions (L:AP-C2Ca). One predictor for eachf these outcome variables was found.

For volume, largest cross-sectional area (LgCa),he cross-sectional area at C2 (C2Ca), and theateral linear dimension of the cross section at2 (L-C2Ca), the predictor was horizontal move-ent. This means that of all the cephalometricseasured, the only significant predictor was how

ar anteriorly the mandible had been pulled byhe appliance. Therefore, an algorithm could beroduced to predict each of these outcome mea-urements based on a given amount of mandib-lar advancement.

For the AP linear dimension of the cross sec-ion at C2 (AP-C2Ca), and the ratio of the lateralnd AP linear dimensions, (L:AP-C2Ca), 2 dif-erent cephalometric predictor variables were

ound. Saddle angle was found to be able to

pc

vscwms

m1

vvbm

w

T

ns made

T

*

148 Haskell et al.

redict (AP-C2Ca) and facial axis was a signifi-ant predictor for (L:AP-C2Ca).

Table 1 shows the reliability or consistencyalues of each variable used in the linear regres-ion analysis. This statistical analysis shows howonsistently all the measurement in this studyere made. It was computed by the use of the 3easurements made of each variable. The value

hown is the intraclass correlation of the average

able 1. Reliability or Consistency Values of Each Va

n

Facial angle, FH-NPO 26Facial axis Ricketts, NaBa-PtGn 26Saddle-sella-angle, SN-Ar 26Posterior face height, SGo 26Anterior face height, NaMe 26Total face height, N-Gn 26Posterior facial height, Go-CF 26horzMovmt 26Volume(mm3) 26SmCa (mm2) 26AP-SmCa (mm) 26L-SmCa 26Ratio-SmCa 26LgCa 26AP-LgCa 26L-LgCa 26L:AP-LgCa 26C2Ca 26AP-C2Ca 26L_C2Ca 26L:AP-C2Ca 26

is number of subjects. Reliability statistics show level of conhows a very high level of consistency in the measurements

able 2. Descriptive Statistics of Change

n Me

Facial angle, FH-NPO 26 0.8Facial axis Ricketts, NaBa-PtGn 26 �1.0Saddle-sella-angle, SN-Ar 26 �10.8Posterior face height, SGo 26 3.5Anterior face height, NaMe 26 7.4Total face height, N-Gn 26 8.1Posterior facial height, Go-CF 26 3.4horzMovmt 26 3.9Volume(mm3) 26 2792.7SmCa (mm2) 26 43.1AP-SmCa (mm) 26 0.5L-SmCa 26 2.4L:AP-SmCa 26 0.1LgCa 26 71.4AP-LgCa 26 0.8L-LgCa 26 3.6L:AP-LgCa 26 0.1C2Ca 26 77.6AP-C2Ca 26 1.0L_C2Ca 26 4.2L:AP-C2Ca 26 0.1

These values of Kurtosis are reported by SPSS where the norm

easures. The range of possible values is from 0 to, where 1 is the greatest amount of consistency.

Table 2 shows descriptive statistics for eachariable used in this analysis in 26 subjects. Allalues reflect the measurements of differenceetween the “with device in place” and “without”easurements.Table 3 shows the number of subjects that fell

ithin a range of Z scores for some key variables.

le Used in the Linear Regression Analysis

Without Device With Device

0.971 0.9740.992 0.9840.952 0.9670.990 0.9910.995 0.9800.995 0.9790.990 0.9750.996 0.9930.995 0.9990.990 0.9950.981 0.9860.997 0.9950.982 0.9710.978 0.9960.946 0.9620.989 0.9770.976 0.9430.997 0.9980.983 0.9890.995 0.9930.974 0.984

ncy of the measurements. All values being greater that 0.90.

SD Skewness Kurtosis*

6 1.6908138 0.582 0.4716 2.1843394 �0.632 0.1039 5.2694548 �0.968 1.1954 1.8280853 0.107 �0.4887 3.5100639 0.684 �0.1575 3.8600343 0.699 0.1985 1.6987300 0.688 0.9770 3.5692267 0.261 0.2532 4380.9077 1.739 5.7892 86.1480279 0.125 3.6637 2.5016975 �0.124 0.5843 4.7274698 �1.043 4.0897 0.8896943 0.864 5.1801 61.6807120 0.500 0.8541 2.0577940 �0.311 0.0230 6.0296583 0.864 1.2329 0.5128242 0.320 2.2532 111.2364667 1.195 5.9554 2.0505118 �0.534 2.0281 4.4366847 �0.416 2.1737 0.5043837 0.714 1.474

riab

siste

an

538491026794961585893461217993599487576973076410653919232564564121313846275669237904

equals zero.

TtitschctstaL

roasvoe(aisi

D

TpMwjp

v2wimm

wtuaarM

pa

Fb

T

1

149Mandibular Advancement and CBCT

he Z score is the number of standard deviationshe value is away from the mean. This table isllustrated in Figures 16, 17, and 18. They showhe distribution of subjects for each range of Zcores. Specifically, Figures 16, 17, and 18 showhange in volume of velopharynx and oropharynx,orizontal mandibular movement, and change inross-sectional area at 3 different levels, respec-ively. The results of the linear regression analysishowed at least one statistically significant predic-or variable only for the following outcome vari-bles: volume, LgCa, C2Ca, L-C2Ca, AP-C2Ca,:AP-C2Ca.

Table 4 shows the R2 value of each significantegression test, and the P value found for theverall significance of the individual outcome toll the predictor variables. The overall test forignificance of the predictors on an outcomeariable showed a P value greater than 0.05 in 4f these tests. However, a predictor variable inach of the 6 analyses was statistically significantP � 0.05). This unexplained “fluke” in thenalysis is likely due to the SPSS program round-ng during the analysis of the variables. Table 5hows which predictors were found to be signif-cant in each linear regression test.

igure 16. The z score distribution of volume change

able 3. Distribution of Z Score Ranges

Variable �3 � Z � �2 �2 � Z � �1 �

Horizontal movement 0 4Volume 0 2SmCa 1 1LgCa 1 2C2Ca 1 0

ars instead a smooth curve. (Color version of figure is av

iscussion

his study was the first to show a sample of OSAatients successfully treated with a removableAD in 3D imaging. The authors of previousork with MAD and 3D imaging46,50 used sub-

ects who only snored or had untreated OSAatients with conventional CT, and MRI.

This was also the first study to use CBCTolumetric imaging of OSA patients. PreviousD studies were limited to cephalometrics,hereas previous CT and MRI studies were lim-

ted to cross-sectional area and linear measure-ents. These studies could not produce volu-etric measurements.Recent advances in CBCT technology in soft-

are packages has allowed these volumetric datao be collected from CBCT scans. This studysed pre-released software from Dolphin (pre-lpha version 11). Technological advances suchs these have allowed progress to be made inesolving and predicting the efficacy of OSAAD treatment in this study.This study was also the first to attempt to

redict changes in airway from changes in ceph-lometric measurements before and after the

is graph is similar to a bell-curve, but is presented in

Z � 0 0 � Z � 1 1 � Z � 2 2 � Z � 3 3 � Z � 4

8 11 2 1 02 10 1 0 15 6 2 1 01 8 3 1 04 9 1 0 1

. Th

1 �

111

ailable online.)

MentcsalIcwgatf

S

Ta0ascifuehpt

FH4

Fcl

150 Haskell et al.

AD is used. All other cephalometric studiesxamined static anatomy as predictors of a diag-osis of OSA. This study found that it is possible

o predict the volume gained, the amount ofross-sectional area gained at the largest crossection, the cross-sectional area gained at C2,nd the lateral linear dimension gained at thisevel from the distance the mandible is advanced.t was also the first study to relate an establishedephalometric measurement to changes in the air-ay. It found that the AP linear dimensionained at C2 is predictable by the saddle angle,nd the amount of shape change (from elliptico round, or vice versa) is predictable by theacial axis measurement.

igure 17. Z score distribution of change in horizonterbst how many standard deviations away from the amm. (Color version of figure is available online.)

igure 18. This shows the Z score distribution of chanross section (SmCa), largest cross section (LgCa), and

evels of the pharynx respond to a MAD. (Color version o

tatistical Conclusions

he intraclass correlation for every variable an-lyzed in this study was equal or greater than r �.95. This result shows an extremely high reli-bility of the measurement process used in thistudy. This study related dynamic alterations inephalometric measurements to specific changesn the airway. The distance the mandible is movedorward can be used to predict the gain in vol-me, largest cross-sectional area, and all param-ters of the airway at C2. That is to say, this studyas discovered that it is theoretically possible toredict the amount an airway will increase fromhe number of millimeters the MAD is activated.

vement. This shows how many subjects titrated theirge amount titrated in this population. The mean was

cross-sectional area at the 3 levels measured: smallests section at C2 (C2Ca). It compares how the different

al movera

ge incros

f figure is available online.)

wapsoemaetpc

lacipamgstmbeaiooo

rpttamsMa

aatsidpst

pcooti

twnM

salradpatcwpta

tcp2h

T

T

151Mandibular Advancement and CBCT

Only a limited number of outcome variablesere found to be predictable, however. Nothingbout the smallest cross-sectional area could beredicted, and only the area at the largest crossection could be predicted. It is likely that the C2utcomes were the most predictable becauseach measurement of the C2 cross section re-ained at the same level before and after appli-

nce placement. Because the smallest and larg-st cross-sections could be at any vertical level ofhe pharynx after the use of the appliance, theredictors in this study may have been insuffi-ient for such complexly changing variables.

Two of the most interesting findings from theinear regression test involved preexisting ceph-lometric measures. The first is that of all theephalometric changes measured in this exper-ment, only the change in saddle angle couldredict the AP dimension of the airway changest C2. The saddle angle is generally used as aeasure of the location of the condyle in the

lenoid fossa in relationship to the basion-na-ion plane. In this study, however, it measureshe change in the location of the condyle as it

oves forward and downward with the mandi-le. The mean change in Saddle Angle in thisxperiment was approximately �11 degrees. An-tomically, this means that the mandible is mov-ng both downward and forward. The result is anblique movement, which has predicted the APpening at C2, situated at the middle of theropharynx.

The second significant predictor cephalomet-ic measurement is facial axis, which was the onlyredictor of the change in the ratio of the L: AP athe level of C2. This ratio reveals the shape ofhe cross section of the airway rather than anbsolute dimension. The larger the value, theore elongated or elliptic the shape, while the

maller the value, the shape is more spherical.oving the jaw down the facial Axis with the

able 4. Significant Outcome Variables

Outcome Variable R2 P Value

Volume (mm3) 0.556 0.043LgCa (mm2) 0.649 0.008C2Ca (mm2) 0.534 0.059L-C2Ca (mm) 0.501 0.091AP-C2Ca (mm) 0.468 0.132L: AP-C2Ca 0.473 0.125

ppliance tended to predict an opening of the

irway towards a more elliptic shape. This is ingreement to previous studies46,47 in whichhe authors found that the increase in cross-ectional area was significant only because of thencrease in transverse dimension, not the APimension. No one has previously related thisarticular observation of “roundness” of airwayhape with a normalized oblique movement ofhe mandible.

Furthermore, the use of the appliance ap-eared to drive the facial axis in nearly everyase towards a more normal measure, regardlessf whether the patient was originally more openr closed in jaw orientation. This explains whyhe mean of the change appears small (�1), buts still significant.

The skewness of the volume and the area ofhe cross section at C2 have a positive value,hich indicates that very few people respondedegatively in the measured parameters to theAD therapy.Most of the airway outcome measurements

howed a very high value of kurtosis. Kurtosis ismeasure of “skinniness” of a bell-curve. A high

evel in this study means that the distribution ofesponse to MAD therapy was close to the mean,nd those who responded away from the meanid not stray very far. The authors have inter-reted this to indicate that the change in theirway is highly predictable in response to MADherapy. In addition, the mean values of airwayhange were overwhelmingly positive values,hich further leads to the conclusion that mostatients responded not only predictably, but posi-

ively (see distribution of the Z scores in Table 2nd the bar graphs in the Results section).

In accordance with the literature reported,his study has shown that MAD therapy in-reased the patency of the velopharynx and oro-harynx. The mean increase of volume was792.7 mm3. Although the average increase inorizontal movement of the mandible was 3.99

able 5. Significant Predictor Variables

Outcome Variable Predictor Variable P Value

Volume (mm3) Horizontal movement 0.014LgCa (mm2) Horizontal movement 0.001C2Ca (mm2) Horizontal movement 0.044L-C2Ca (mm) Horizontal movement 0.011AP-C2Ca (mm) Saddle Angle 0.042

L: AP-C2Ca Facial axis 0.042

mthslcn

VU

Twar(

ofiottcBuwvtsvc

rs

bast

1

2

3

Csfpppatwtfiofs

Ffi

152 Haskell et al.

m with a standard deviation of approximatelyhe same value, the average increase in total faceeight (or vertical change) was 8 mm with amaller standard deviation. That is to say, despite aarge variation in horizontal movement, whenombined with a vertical increase, it yielded a sig-ificant and predictable volumetric improvement.

isualizing Changes in the Airway andnderstanding Nonresponders

he following images show comparisons of air-ay volume between an overresponder (Fig 19),normal responder (Fig 20) and a subject who

esponded with a negative change in volumeFig 21).

It is noted that the under-responder was onef the few subjects who had a worsening value ofacial axis from 85 to 83 degrees (normal � 90),ndicating a backwards or clockwise movementf the mandible. This may be contrasted withhe example of the average responder, wherehe facial axis stayed within 1 degree of a positivehange towards a normal cephalometric value.oth subjects had an average amount of mandib-lar advancement, whereas the superresponderas able to be advanced 12 mm (an atypical ad-ancement range). Therefore, although the sta-istical analysis demonstrated that there was noignificant correlation between facial axis andolumetric change in airway, a generalized per-eption by the authors showed that the worst

igure 19. Subject 1 responded in volume change 4

gure is available online.)

esponders had facial axis measurements out-ide the normal cephalometric values.

Although some patients appeared to worsenased on the parameters measured in this study,ll subjects reported an improvement in OSAymptoms. There are several possible explana-ions, which may include:

. Other anatomical parameters not yet de-scribed, related to the complex neuromuscu-lar functions of the oropharynx

. The wakefulness and uprightness of subjectsin this study may have differed from the ap-pearance of the airway on the image com-pared with the image of a patient taken in thesupine position when asleep.

. A placebo response was in effect.

ontrastingly, an anecdotal observation of thistudy supported the findings of Sanner et al.,8 whoound through MRI that if the most collapsibleart of the airway is in the velopharynx alone, theatients would not respond well to MAD. In theresent study’s volumetric evaluation with CBCT,ll the subjects found with the smallest cross-sec-ional area above the occlusal plane (with andithout the device) either decreased or remained

he same in volume. This finding, along with thendings of Sanner et al.,8 implies that a MAD maynly be partially effective or completely inef-ective for OSA sufferers of this type. Thistrengthens the observation that a CBCT scan

dard deviations above the mean. (Color version of

stan

asmtiwa

c

jtaAchear

Fo

Ffi

153Mandibular Advancement and CBCT

nd anatomical evaluation of the airwayhould be performed before initiating treat-ent on OSA patients with a MAD. Differen-

iating the morphology of each patient is crit-cal to select the appropriate treatment,hether it be surgical correction, or the use ofremovable orthotic appliance.Another noteworthy finding in this study in-

luded the evaluation of an 18-year-old female sub-

igure 20. Subject 5 responded the mean amountnline.)

igure 21. Subject 23 responded in volume change

gure is available online.)

ect who exhibited large tonsils. The first illustra-ion (A) in Fig 22 shows the arrow pointing to theirway where the tonsils are impinging upon it.lthough it had an overall increase, the volumetrichange with the use of the appliance was less thanalf the average change, with no increase in small-st cross-sectional area with the use of the appli-nce. Another interesting finding was that the nar-owest area as seen in part B of Fig 22, shifted

lume change. (Color version of figure is available

ndard deviations below the mean. (Color version of

in vo

2 sta

iswuaowna

tcommwOtst

sica

atvLttvscs

A

P

Ftv is ava

154 Haskell et al.

nferiorly. The specific location change of themallest cross-sectional area, vertically up or down,as unpredictable. This change is likely basedpon the subject’s individual morphology. Thislteration in morphology of the airway with the usef an appliance could not have been predictedith previous 2D imaging modalities and couldot be assessed without the use of 3D imaging suchs CBCT.

To explain this subject’s improvement in symp-oms despite the apparent lack of volumetrichange, it may be helpful to recall findings fromther authors. Tsuiki et al.48 found with electro-yographic recordings that an increased neuro-uscular tonicity of the genioglossus occurredith mandibular advancement in patients withSA when they were awake. This reduced the

endency of the upper airway to collapse. If theame process occurred in patients during sleep,

igure 22. Volumetric representations of a subject wonsil impingement. (B) The wheel marks the level oalue, but shifted inferiorly. (Color version of figure

his could explain how a seemingly poor re- p

ponder to MAD therapy based on below averagencrease in volumetric and cross-sectional area,ould actually have an improvement in OSA signsnd symptoms.

The subject with the largest airway change wasdvanced 12 mm. Patients with sufficiently dysplas-ic anatomy requiring a bimaxillary surgical ad-ancement of the jaws as outlined by Conley andegan,1 often have jaw advancement surgery in

his range. As the average volume change was inhe order of 2800 mm3, this patient increased inolume 18,000 mm3! This large increase may beimilar to that seen in surgical OSA corrective pro-edures, and could therefore explain the hugeuccess of surgery for a permanent correction.

natomical Mechanisms

revious authors have suggested that the im-

ery large lingual tonsils. (A) Arrow points to site ofallest cross-sectional area, which has not changed inilable online.)

ith vf sm

rovement in volume in the oropharynx with a

Msmcptptobfedmpigbp

fitotpitplfttaa

mt

Ftws

155Mandibular Advancement and CBCT

AD is caused by a forward repositioning of theoft palate.25-28,43,45-50 The mandibular advance-ent theoretically stretches the soft palate with a

oncomitant stiffening of the wall of the oro-harynx itself. This is accomplished throughhe bracing effect of the lateral wall of the softalate in relation to the base of the tongue viahe palatoglossal arch. A transverse wideningf the airway with advancement is said to occurecause of less movement of the hyoid boneorward than that of the mandible. This isxplained because the posterior belly of theigastric muscle and the infrahyoid musclesostly restrict this inferior bone. Another hy-

othesis is that the transverse widening effects due to a reflex response of the stylopharyn-eus muscle to the drag effect upon hyoidone when using a mandibular advancing ap-liance.

igure 23. This reconstruction of the airway made witonsil, and faucial areas. It is illustrated here as what aings. This is actually an empty space in the orop

tyloglossus and the base of the tongue. (Color version of

The use of CBCT in depicting a mechanismor airway volume enhancement with an orthotics also demonstrated in the present studyhrough examination of the tonsillar area of theropharynx. The presence of large tonsils illus-rates the reduction of available airway and theossible limitation of the degree of volumetric

ncrease. With MAD therapy and the absence ofonsils, a reconstructed 3D pharynx reveals ahenomenon not previously described in the

iterature. There is an area of expansion in theaucial pillar area that appears as appendages onhe pharynx in Figure 23, but are actually depic-ions of a void. Figure 23 shows the enlargementnd “ballooning” of the faucial pillar area usinglower jaw advancement appliance.Figure 23 effectively demonstrates why re-

oval of enlarged tonsils that can impinge uponhis important area of expansion is a frequently

VR shows a great expansion of the base of the tongue,ars to be an appendage shaped like a bird with opennx surrounded by palatoglossus, palatopharyngeus,

h 3Dppe

hary

figure is available online.)

ua

cmfacrpsatuO

F

Ppciysiaiivnrusurgm

iirsyiasdtrtbc

hh

ht

C

LCiasrpWsuavttmottrmu

R

156 Haskell et al.

sed treatment modality for an improvement inirway patency in patients with OSA.1

The position of the smallest and largestross-sectional areas before and after treat-ent was not able to be tracked. However,

rom a clinical standpoint, the greatest pointt which the cross-sectional area increase oc-urs may not be as important as total volumet-ic change when treating OSA. Because mostatients had an increase in volume, it is as-umed that this gain in volume allowed moreirflow, and thus improvement in OSA symp-oms. Certain factors, such as airway tonicity,prightness of posture, and awake-ness in anSA sample could not be controlled.

uture Studies

revious researchers and clinicians have re-orted that the gold-standard for permanentorrection of OSA does not deal with surgicalntervention of the soft tissues of the orophar-nx, with the possible exception of excision ofpace-occupying pathognomonic tissue. Bimax-llary surgical advancement of both jaws is anbsolute correction of OSA.1 The present studyndicated a 4-standard deviation above the meanncrease in volume in one subject who was ad-anced the same degree as a normal orthog-athic correction for OSA. There is a need foresearchers to develop a prospective study eval-ating at 3 dimensional changes occurring withurgery. Once a baseline value for normal vol-me is established in these patients, an algo-ithm that will predict the amount of volumeain from the amount of mandibular advance-ent might be developed.There is some confusion as to which type of

ndividual show a pharyngeal airway shape whichs more elliptic than spherical. The literatureeports that OSA patients have a more sphericalhape than seen in normal patients. No one haset described the reason for this differentiationn morphology in these 2 groups. With the datalready collected in this study, it should be pos-ible to relate a modified maxillofacial ratio in-ex (akin to dolichocephalic vs. brachycephalic)

o the shape of the airway. Clinicians have longeported that nasal airway obstruction leads tohe greatest negative morphologic changes (open-ite, adenoid facies) during growth in the doli-

hocephalic pattern.11 Long-face people may well

ave a narrower or spherical airway than wide-eaded individuals.

Understanding these relationships may alsoelp to explain certain unusual findings related

o the facial axis and volumetric change.

onclusion

ateral airway dimensions of the cross-section at2, total volume, and cross-sectional area gained

n the oropharynx can be predicted from themount of mandibular forward movement. Theaddle angle was a predictor of the linear ante-ior-posterior dimension, while the facial axisredicted the ellipticality of the airway at C2.ith the placement of a MAD appliance, the

mallest airway cross-section may move to annpredictable position, superiorly or inferiorlylong the length of the pharynx. No predictorariables could be found for this cross-section ofhe airway. Therefore, it may be advantageoushat all patients wishing to receive MAD treat-

ent for their OSA receive a CBCT scan inrder to assess the individual appropriateness ofhis form of treatment. This is because whenreating OSA, obtaining an improvement in aestrictive point in the airway may be just as orore important than achieving an overall vol-

me increase.

eferences1. Conley RS, Legan HL: Correction of severe obstructive

sleep apnea with bimaxillary transverse distraction osteo-genesis and maxillomandibular advancement. Am JOrthod Dentofac Orthop. 129:283-292, 2006

2. Ogutcen-Toller M, Sarac YS, Cakr-Ozkan N, et al: Com-puterized tomographic evaluation of effects of mandib-ular anterior repositioning on the upper airway: A pilotstudy. J Prosthet Dent 92:184-189, 2004

3. Abbott MB, Donnelly LF, Dardzinski BJ, et al: Obstruc-tive sleep apnea: MR imaging volume segmentation anal-ysis. Radiology 232:889-895, 2004

4. Sharabi Y, Dagan Y, Grossman E: Sleep apnea as a riskfactor for hypertension (review). Curr Opin NephrolHypertens 13:359-364, 2004

5. Rey S, Valdés G, Iturriaga R: Pathophysiology of obstruc-tive sleep apnea-associated hypertension. Rev Med Chil.2007 October;135(10):1333-42. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18180843?ordinalpos�1&itool�EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_RVDocSum. Epub December 20, 2007. InSpanish

6. Kawata N, Tatsumi K, Terada J, et al: Daytime hypercap-nia in obstructive sleep apnea syndrome. Chest 132:

1832-1838, 2007

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

157Mandibular Advancement and CBCT

7. Barcelo A, Miralles C, Barbe F, et al: Abnormal lipidperoxidation in patients with sleep apnoea. Eur Repir J16:644-647, 2000

8. Sanner BM, Heise M, Knoben B, et al: MRI of thepharynx and treatment efficacy of a mandibular ad-vancement device in obstructive sleep apnoea syndrome.Eur Respir J 20:143-150, 2002http://my3.statcounter.com/project/

9. Yamachi M, Kimura H. Oxidative stress in obstructivesleep apnea: putative pathways to the cardiovascularcomplications. Antioxid Redox Signal. 2008 Jan 4;[Epub ahead of print]

0. Rahaghi F, Basner RC: Delayed Diagnosis of obstructivesleep apnea: Don’t ask, don’t tell. Sleep Breath 3:119-124, 1999

1. Faber CE, Grymer L: Available techniques for objectiveassessment of upper airway narrowing in snoring andsleep apnea (review). Sleep Breath. 7:77-86, 2003

2. Fogel RB, Malhotra A, Dalagiorgou G, et al: Anatomicand physiologic predictors of apnea severity in morbidlyobese subjects. Sleep 26:150-155, 2003

3. Major MP, Flores-Mir C, Major PW: Assessment of lateralcephalometric diagnosis of adenoid hypertrophy and pos-terior upper airway obstruction: A systematic review. Am JOrthod Dentofac Orthop 130:700-708, 2006

4. Johal A, Conaghan C: Maxillary morphology in obstruc-tive sleep apnea: A cephalometric and model study.Angle Orthod 74:648-656, 2004

5. Choi JK, Goldman M, Koyal S, et al: Effect of jaw andhead position on airway resistance in obstructive sleepapnea. Sleep Breath 4:163-168, 2000

6. Johal A, Patel SI, Battagel JM: The relationship betweencraniofacial anatomy and obstructive sleep apnoea: A case-controlled study. J Sleep Res 16:319-326, 2007

7. Sher AE, Schechtman KB, Piccirillo JF: The efficacy ofsurgical modifications of the upper airway in adults withobstructive sleep apnea syndrome (review). Sleep 19:156-177, 1996

8. Zozula R, Rosen R: Compliance with continuous positiveairway pressure therapy: Assessing and improving treat-ment outcomes (review). Curr Opin Pulm Med 7:391-398, 2001

9. Rachmiel A, Aizenbud D, Pillar G, et al: Bilateral man-dibular distraction for patients with compromised airwayanalyzed by three-dimensional CT. Int J Oral MaxillofacSurg 34:9-18, 2005

0. Aboussouan LS, Golish JA, Wood BG, et al: Dynamicpharyngoscopy in predicting outcome of uvulopalato-pharyngoplasty for moderate and severe obstructivesleep apnea. Chest 107:946-951, 1995

1. Battagel JM, L’Estrange PR, Nolan P, et al: The role oflateral cephalometric radiography and fluoroscopy inassessing mandibular advancement in sleep-related dis-orders. Eur J Orthod 20:121-132, 1998

2. Clark GT, Arand D, Chung E, et al: Effect of anteriormandibular positioning on obstructive sleep apnea. AmRev Respir Dis 147:624-629, 1993

3. Lowe AA: Can we predict the success of dental appliancetherapy for the treatment of obstructive sleep apneabased on anatomic considerations? Sleep 16:S93-S95,

1993 (suppl)

4. Tsuiki S, Almeida FR, Lowe AA, et al: The interactionbetween changes in upright mandibular position andsupine airway size in patients with obstructive sleep ap-nea. AJODO. 128:504-512, 2005

5. Tsuiki S, Hiyama S, Ono T, et al: Effects of titratableoral appliance on supine airway size. Sleep 24:554-560,2001

6. Tsuiki S, Lowe AA, Almeida FR, et al: Effects of ananteriorly titrated mandibular position on awake airwayand obstructive sleep apnea severity. Am J OrthodDentofac Orthop. 125:548-555, 2004

7. Tsuiki S, Lowe AA, Almeida FR, et al: Effects of mandibularadvancement on airway curvature and obstructive sleepapnoea severity. Eur Respir J 23:263-268, 2004

8. Rose EC, Staats R, Virchow C, Jr, et al: Occlusal andskeletal effects of an oral appliance in the treatment ofobstructive sleep apnea. Chest 122:871-877, 2002

9. Schmidt-Nowara WW, Meade TE, Hays MB: Treatmentof snoring and obstructive sleep apnea with a dentalorthosis. Chest 99:1378-1385, 1991

0. Fransson AM, Tegelberg A, Johansson A, et al: Influenceon the masticatory system in treatment of obstructivesleep apnea and snoring with a mandibular protrudingdevice: A 2-year follow-up. Am J Orthod Dentofac Or-thop. 126:687-693, 2004

1. Bonham PE, Currier GF, Orr WC, et al: The effect ofa modified functional appliance on obstructive sleepapnea. Am J Orthod Dentofac Orthop. 94:384-392,1988

2. O’Sullivan RA, Hillman DR, Mateljan R, et al: Mandib-ular advancement splint: An appliance to treat snoringand obstructive sleep apnea. Am J Respir Crit Care Med151:194-198, 1995

3. Viviano JS: Acoustic reflection: Review and clinical ap-plications for sleep-disordered breathing (review). SleepBreath 6:129-149, 2002

4. Mah J, Hatcher D: Three-dimensional craniofacial im-aging. Am J Orthod Dentofac Orthop. 126:308-309,2004

5. Hashimoto K, Kawashima S, Araki M, et al: Comparisonof image performance between cone-beam computedtomography for dental use and four-row multidetectorhelical CT. J Oral Sci 48:27-34, 2006

6. Ludlow JB, Davies-Ludlow LE, Brooks SL: Dosimetry oftwo extraoral direct digital imaging devices: NewTomcone beam CT and Orthophos plus DS panoramic unit.Dentomaxillofac Radiolol. 32:229-234, 2003

7. Schulze D, Heiland M, Thurmann H, et al: Radiationexposure during midfacial imaging using 4- and 16-slicecomputed tomography, cone beam computed tomogra-phy systems and conventional radiography. Dentomaxil-lofac Radiol 33:83-86, 2004

8. Farman AG, Scarfe WC, Sukovic P: Clinical applicationsof cone-beam computed tomography in dental practice.J Can Dent Assoc 72:75-80, 2006

9. Hilgers ML, Scarfe WC, Scheetz JP, et al: Accuracy oflinear temporomandibular joint measurements withcone beam computed tomography and digital cephalo-metric radiography. Am J Orthod Dentofac Orthop. 128:

803-811, 2005

4

4

4

4

4

4

4

4

4

4

5

158 Haskell et al.

0. Cevidanes LH, Styner MA, Proffit WR: Image analysisand superimposition of 3-dimensional cone-beam com-puted tomography models. Am J Orthod Dentofac Or-thop. 129:611-618, 2006

1. Ogawa T, Enciso R, Memon A, et al: Evaluation of 3Dairway imaging of obstructive sleep apnea with cone-beam computed tomography. Stud Health Technol In-form 111:365-368, 2005

2. Mayer P, Pépin JL, Bettega G, et al: Relationship be-tween body mass index, age and upper airway measure-ments in snorers and sleep apnoea patients. Eur RespirJ 9:1801-1809, 1996

3. Shi H, Scarfe WC, Farman AG: Upper airway segmenta-tion and dimensions estimation from cone-beam CTimage datasets. Int J Cars. 1:177-186, 2006

4. Peh WC, Ip MS, Chu FS, et al: Computed tomographiccephalometric analysis of Chinese patients with obstructivesleep apnoea. Australas Radiol 44:417-423, 2000

5. Vigg A, Vigg A, Vigg A: Obstructive sleep apnea in a refer-

ral population in India. Sleep Breath 7:177-184, 2003

6. Zhao X, Liu Y, Gao Y: Three-dimensional upper-airwaychanges associated with various amounts of mandibularadvancement in awake apnea patients. AJODO. 20008:661-668

7. Kyung SH, Park YC, Pae EK: Obstructive sleep apneapatients with the oral appliance experience pharyngealsize and shape changes in three dimension. AngleOrthod 75:15-22, 2005

8. Sutthipraporn P, Tanimoto K, Ohtsuka M, et al: Posi-tional changes of oropharyngeal structures due to grav-ity in the upright and supine positions. DentomaxillofacRadiol 37:130-135, 2008

9. Kushida Ca, Morgenthaler TI, Littner MR, et al: Practiceparameters for the treatment of snoring and obstructivesleep apnea with oral appliances: An update for 2005.Sleep 9:240-243, 2006

0. Ogawa T, Enciso R, Memon A, et al: Evaluation of 3Dairway imaging of obstructive sleep apnea with cone-beam computed tomography. Stud Health Technol In-

form 111:365-368, 2005