University of Southern California Dissertations and Theses

A cone beam-CT evaluation of the upper airway for sleep apnea prediction

(USC Thesis Other)

A cone beam-CT evaluation of the upper airway for sleep apnea prediction

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A CONE BEAM-CT EVALUATION OF THE UPPER AIRWAY
FOR SLEEP APNEA PREDICTION

by

Maria Reyes Enciso Cachafeiro

____________________________________________________________________

A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(CLINICAL AND BIOMEDICAL INVESTIGATIONS)

May 2009

Copyright 2009 Maria Reyes Enciso Cachafeiro

ii
DEDICATION
To my Family and Mentors:
Lucien, Pablo and Diego Nocera
La Familia Enciso Cachafeiro
Glenn T. Clark
Stanley Azen

iii
ACKNOWLEDGEMENTS
A special thank you to:
Dr. Glenn T. Clark
Dr. Stanley Azen
Dr. Yang Chai
Michelle Bailey
Dr. Takumi Ogawa
Dr. Yuko Shigeta
Dr. Manuel Nguyen

iv
TABLE OF CONTENTS

Dedication ii
Acknowledgements iii
List of Tables v
List of Figures vi
Abstract vii
Chapter 1: Introduction 1
Chapter 2: Review of Literature 3
Chapter 3: Hypotheses 16
Chapter 4: Subjects and Methods 17
Chapter 5: Results 35
Chapter 6: Discussion 48
Chapter 7: Assumptions 53
Chapter 8: Limitations 54
Chapter 9: Summary 57
Chapter 10: Conclusions 58
Bibliography 60
Appendix 65

v
LIST OF TABLES
Table 1 - Pilot data: Comparison between OSA+ and control groups 32
Table 2 - A priori sample size and power 35
Table 3 - Descriptive statistics for OSA and control groups 37
Table 4 - Comparison of race distribution between OSA and control groups 38
Table 5 - Comparison of ethnicity distribution between OSAs and controls 38

Table 6 - Comparison of CBCT measurements between OSAs and
Controls 40

Table 7 - Comparison of minimum cross-sectional airway location
between OSA and control groups 41

Table 8 - Comparison of minimum cross-sectional airway shape between
OSA and control groups 41

Table 9 – Stepwise multiple linear regression analyses of the risk
factors on the log(AHI) 43

Table 10 - Logistic stepwise regression analysis of the risk factors for OSA 44

Table 11 - Logistic stepwise regression analysis of the risk factors for OSA 44

Table 12 – Screening for OSA with the Berlin Questionnaire and the ARES
questionnaire 45

Table 13 – Sensitivity, Specificity, Positive and Negative Predictive
Value for two questionnaires against ambulatory somnographic data 46

Table 14 – Association between ARES and Berlin questionnaires 46

Table 15 – Association between Family History and OSA 47

vi
LIST OF FIGURES
Figure 1 - Cone-Beam CT NewTom 3G 22

Figure 2 - The oropharyngeal region selected by the user on (A) axial view
and (B) mid-sagittal view of the upper airway (between PNS and the
most anterior inferior point of the second cervical) 23

Figure 3 – Cross-sectional upper airway 24

Figure 4 – 3D airway surface superimposed on sagittal view of a healthy
patient (A) and OSA patient (C); 3D airway surface for a healthy
patient (B) and OSA (D). 24

Figure 5 - Antero-posterior (7.3mm) and lateral dimension (17.7) of the
cross-sectional airway 25

Figure 6– Occlusal plane is defined by the incisal point on mid-sagittal
image (A) and the first molars (B) 26

Figure 7 - Vertical and horizontal soft palate length measurements 28

vii
ABSTRACT
Introduction: Cephalometric and other imaging modalities have been used to
compare Obstructive Sleep Apnea (OSA) patients with controls, but volumetric
measurements of the upper airway are scarce. Purpose: To compare the Cone-Beam
Computerized Tomography (CBCT) scan measurements between patients with OSA
and controls to develop a prediction model for OSA based on imaging and family
history. Methods: 39 OSA patients (Apnea-Hypoapnea Index≥15) and 23 controls
based on ambulatory somnographic assessment were recruited through flyers and
mail at USC School of Dentistry. Each patient answered the Berlin Questionnaire,
collected family member’s Berlin Questionnaires and was imaged with CBCT.
Linear and volumetric measurements of the upper airway were performed and
multivariate logistic regression analysis was used to identify risk factors for OSA.
Results: OSA patients were predominantly male, older, had a larger neck size and
larger Body Mass index than controls. The minimum cross-sectional area and the
lateral dimension were significantly smaller in cases. Conclusions: Age>57 years,
male gender, a positive Berlin (or ARES Questionnaire) and narrow upper airway
lateral dimension were identified as significant risk factors for OSA. This prediction
model for OSA could be implemented based on this work, however further studies
are needed to generalize our findings. Limitations of the study include self-selection
bias for the controls: subjects with a positive family histoy of sleep apnea are more
willing to participate in a study to predict OSA based on family history, so the odds-
ratios of having OSA based on family history were underestimated.
1
CHAPTER 1: INTRODUCTION

Prevalence and medical consequences of sleep apnea
Sleep apnea is a disorder of interrupted breathing during sleep due to collapse
of the airway during breathing when the pharyngeal and tongue muscles relax during
sleep causing an obstructive event. This disorder is usually associated with loud
snoring (though not everyone who snores has this disorder). Sleep apnea also can
occur if the neurons that control breathing malfunction during sleep (known as
central sleep apnea). During an episode of obstructive apnea, the airflow is stopped
at least for 10 seconds, so there is a resulting drop in oxygen saturation levels in the
blood, detectable with oximetry. A related event, hypopnea, is characterized by a
reduction in airflow associated with a decrease in oxygen saturation. When the
person's blood oxygen level falls, the brain responds by awakening the person
enough to tighten the upper airway muscles and open the windpipe. The person may
snort or gasp, then resume snoring. This cycle may be repeated a hundred times a
night. A study released in 1993 by Young et al.
51
estimated that 2% to 4% of the
middle-aged adult population in North America suffers from Obstructive Sleep
Apnea (OSA) and this prevalence increases dramatically with aging. In a follow-up
study in 2008, Young’s group found one in six people over 50 years old had at least
mild apnea, with one-fourth of those cases severe
50
. Data generated by Young’s
team suggest at least 75% percent of severe sleep apnea cases still go undetected.
Researchers estimated that two-thirds of the people diagnosed at the outset of the
2
study with apnea chose not to use the standard of treatment (Continuous Positive
Airway Pressure or CPAP) at least 4 nights/week during the time of the study.
People with severe apnea who went untreated were four times as likely to die as
those without the disorder.
50
Sleep apnea is more prevalent among the obese, in
males, and in older individuals. Snoring, while not as serious as apnea, affects 30%
of the adult population. It is estimated that, in the U.S. alone, $70 billion is lost in
decreased productivity, lost wages, and property damage due to snoring and sleep
disorders. The medical consequences of moderate and severe OSA are severe weight
gain, attention deficit and poor memory, increased sick time, poor job performance,
hypertension, heart disease and higher probability of accidents (sometimes fatal)
either on the job site
35
or while driving
6
. OSA is directly associated with congestive
heart failure and other heart diseases such as pulmonary hypertension, cardiac
arrhythmia, ischemic heart disease and stroke
48
. It is estimated that OSA may be
responsible for 38,000 cardiovascular deaths per year in the United States alone.
These are severe medical consequences. Identifying at-risk individuals as early as
possible improves intervention and reduces morbidity.
The objective of this study is to compare the upper airway characteristics of
OSA patients with a control group, using CBCT 3-D imaging for future prediction of
OSA based on imaging and family history. The study will quantify the lateral and
antero-posterior dimensions of the smallest cross-sectional area in the upper airway,
as well as minimum caliber and volumetric measurements.

3
CHAPTER 2: REVIEW OF LITERATURE

Sleep Apnea definitions
Apnea event is defined in this thesis as cessation of respiratory air flow for a
minimum of 10 seconds. Hypopnea is considered to be present when there is at least
a 50% reduction of the air-flow signal combined with a decrease in hemoglobin
oxygen saturation of at least 1%. The apnea index (AI) is defined as the average
number of apneas per hour of sleep, and the apnea/hypopnea index (AHI) is defined
as the average number of apneas plus hypopneas per hour of sleep. An AHI of less
than 5 events/hour is consider normal, between 5 to 14 is indicative of mild sleep
apnea syndrome, an AHI between 15 and 30 is indicative of moderate apnea, an AHI
31 and greater indicates severe apnea.
50
Children’s apnea severity is tabulated
differently and will not be covered in this Master’s thesis.

Sites of Airway Obstruction
Airway obstruction can occur in many areas of the nasopharynx, oropharynx
and hypopharynx. More commonly, airway obstruction occurs in the oropharynx.
Fatty pads and enlarged parapharyngeal walls reduce the size of the posterior airway,
which increases the chance of obstruction during sleep. An elongated soft palate and
enlarged uvula may further compromise the airway. Decreased muscle tone during
sleep also contributes to airway collapse. The base of the tongue is a common site of
4
hypopharyngeal obstruction in sleep apnea. Patients with a small or receding jaw
(retrognathic) are at increased risk for obstruction. Occasionally, obstruction may be
caused by an enlarged tongue. In this setting, obstruction occurs when the base of the
tongue impinges on the airway just above the glottis.
43

Treatment of OSA
Surgical treatment is irreversible and there are contradictory reports in the
literature about efficacy, side effects and relapse, so it is out of the scope of this
thesis. The two main reversible treatments for OSA are Continuous Positive Airway
Pressure and mandibular advancement dental appliances:
1. Continuous positive airway pressure (CPAP) is the most effective
treatment for obstructive sleep apnea according to the American Academy of Sleep
Medicine. CPAP patients during sleep wear a face mask connected to a pump that
forces air into the nasal passages and keeps the airway open during sleep. In nasal
CPAP, the airway pressure delivered into the upper airway is continuous during both
inspiration and expiration. In BiCPAP, the airway pressure delivered during
inspiration and expiration can be different. CPAP is safe and effective. Daytime
sleepiness improves or resolves. Heart function and hypertension also improve. And,
importantly, the quality of life improves. During full polysomnography in a sleep
laboratory CPAP patients should be monitored to determine the correct pressure.
Many patients at first find the mask uncomfortable, claustrophobic or embarrassing.
CPAP must be used every night, every hour of the night and every nap for life to be
5
effective. Non-compliant patients experience a full return of obstructive sleep apnea
and related symptoms. Unfortunately, compliance is a major problem with only 53%
of the patients using CPAP > 90% of the nights after only 9 weeks of treatment.
44

2. Mandibular advancement appliance (MAD): An oral appliance holds the
jaw and tongue forward and holds the palate up thus preventing closure of the
airway. This small increase in airway size often is enough to control the apneas in
mild to moderate obstructive sleep apnea patients. In a very recent review of the
literature, MAD’s are reported to significantly reduce the AHI from 25 to 14
events/hour with the combined success and response rates of 64%.
11
The advantage
of the dental appliance is that it does not require surgery, and compared to CPAP it is
small, portable, and does not require a machine. Side effects are relatively minor but
frequent with excessive salivation and teeth discomfort the most common ones. One
handicap is that a dental appliance requires natural teeth to fit properly, and as in the
case of CPAP it must be worn every night. A summary of the follow-up compliance
data shows that at 30 months, 56–68% of patients continue to use oral appliance
11
.
In conclusion, according to Hoffstein
11
oral appliances although not as effective as
CPAP in reducing sleep apnea, snoring, and improving daytime function, are
preferred by the patients undoubtedly because they find them to be less cumbersome
than CPAP.

6
Diagnosis of Obstructive Sleep Apnea
Presently, the diagnosis of OSA predominantly depends on overnight in-
laboratory polysomnography (PSG), which includes the continuous recording of a
number of physiologic variables including airflow, chest/abdominal movements,
electroencephalography, electromyography, and oxygen saturation. This allows
quantification of hypopneas and apneas according to standard criteria. The high cost
of in-laboratory full-night polysomnography, together with long waiting lists for
sleep studies, have led to the development of a variety of ambulatory sleep study
systems.
48
The earliest device was based on pulse oximetry only
5
but it has been
shown to suffer from limited accuracy.
18
Several 4-6 channel based somnographic
methods have been tried as less expensive substitutes for full polysomnography.
Currently there is no recommended method from the American Academy of Sleep
Medicine. In the study presented in this thesis we performed two-night in-home
studies for assessment of OSA with the Apnea Risk Evaluation System (ARES)
Unicorder (Advanced Brain Monitoring, Carlsbad, CA, USA). This device will be
described in the Methods section in detail. In this study, ambulatory sleep devices
were chosen because they are closer to the reality in terms of supine and lateral
position (the patient sleeps in his/her bed with as many pillows as they need), and are
inexpensive and better adapted to clinical research. The data from the ARES
Unicorder has been correlated in two separate research studies to traditional
polysomnography in the sleep laboratory and found to have good to excellent
7
sensitivity and specificity in both. The review of these two studies is presented in the
Methods section.

Predicting Sleep Apnea from Signs, Symptoms and Family History
There are many associated signs and symptoms of a sleep apnea disorder in
adults according to the NIH (NIH #95-3803). These include hypersomnolence,
obesity, especially nuchal obesity (neck size >17 inches in males, >16 inches in
females) and systemic hypertension to name only a few. Unfortunately these signs
and symptoms are usually observed in patients that already have moderate to severe
OSA. One area that has been explored as a predictor of individuals who are at-risk
but not yet exhibiting OSA is a positive familial history of snoring/OSA in their
immediate family. In 1978, Strohl et al.
41
reported the association in the same family
of a sudden infant death syndrome (SIDS) case and several cases of OSAS.
Guilleminault et al.
8
reported the familial association in several generations of SIDS
and Apparent Life-Threatening Events (ALTE) with OSA in adults. According to
McNamara and Sullivan there is significant evidence that familial factors influence
the risk of developing upper airway obstruction in both pediatric and adult patients.
21

OSA in adults has been shown to aggregate significantly within some
families.
9,23,31,34,41
Family studies have shown that relatives of patients with OSA
have a 2-4 fold increased risk of developing OSA compared with controls.
34
It has
been suggested that members of such families are predisposed to developing upper
airway obstruction due to genetic risk factors including obesity, craniofacial
8
morphology, and an abnormality of ventilatory and respiratory muscle control.
9,33,41

The belief that family history is a predictor of OSA in adults is strongly held. For
example at the National Heart, Lung and Blood Institute web site in “Who is at risk
for sleep apnea?”

they categorically state that, “If someone in your family has sleep
apnea, you are more likely to develop it.”
52
In spite of this strong statement coming
from the NHLBI, there is need for a systematic study on family history of
snoring/OSA as a predictor of disease.

Berlin Questionnaire
One questionnaire that captures snoring/OSA symptoms is the Berlin
Questionnaire (Appendix). This is an 11-item questionnaire that includes questions
about risk factors for sleep apnea, including snoring behavior, wake-time sleepiness
or fatigue, and obesity or hypertension. This instrument has been used in multiple
studies of primary care patients collected in primary medical practices. It was
developed in 1996, and its origin and use in primary care has been reported.
14,26
It is
a self-report instrument that is focused on a set of known symptoms and clinical
features associated with sleep apnea. One introductory question and four follow-up
questions concerned snoring, witnessed apneas, and the frequency of such events.
Three questions addressed daytime sleepiness, with a sub-question about drowsy
driving. One question asked for a history of high blood pressure. Patients were to
provide information on age, weight, height, and sex. Body mass index (BMI) was
calculated from the self-reported patient information on weight and height. The
9
questionnaire is available in both English and Spanish versions. Validity of this
questionnaire is described in the Methods section. Scoring the Berlin Questionnaire
for each subject as “high risk” or “low risk” is summarized in the Appendix.

Scoring a Family History
Family history has been shown to be a valuable aid in predicting disease risk
in other disease entities, such as breast cancer, heart disease, colon cancer, and
hypertension. Rather than a dichotomous ranking of positive or negative family
history for a specific disease, disease risk can be discussed in terms of its severity in
the family as well as whether it exists in the immediate family (sibling, parent,
uncles and aunts and grandparents), or in the extended family (cousins, nephews,
etc). For example, research in well-known populations such as breast cancer
49
do not
simply count the number of affected relatives in the family but they also considered
the heterogeneity of the familial information when calculating an individuals relative
risk for breast cancer. However, relative risk implies that you know the risk for an
individual of similar age and gender with similar characteristics (excluding family
history) in the population at large. Another example of a complex family history
assessment is a study by Hunt et al.
13
They calculated the relative risk of developing
future coronary heart disease (CHD) or hypertension between positive and negative
family history families comparing for different definitions of a positive family
history. In a follow up paper Williams et al.
47
used a quantitative family risk score
(QFRS) to calculate the degree of familial aggregation of coronary heart disease
10
(CHD), stroke, hypertension, and diabetes. Of course for OSA, the ability to
calculate a quantitative family risk score is problematic, since 75% of the population
with OSA is not diagnosed. This means that age of “disease” onset, severity of
disease and even recognition of disease will not be evident in most individuals.
However, it is very possible to gather a questionnaire from the patient plus their
immediate family (parents and siblings) regarding whether they exhibit snoring,
sleepiness and interrupted breathing symptoms during sleep. In this study we will use
a common method used to determine those having a positive family history: anyone
with one or more affected first degree relatives will be classified as having a positive
family history. An affected first degree relative is defined as someone classified as
“high risk” by the Berlin questionnaire. We will also compute a family history score
(sum of family members at “high risk”, divided by the number of family members).
The Methods section has an extended description of the family history variables.

Predicting Sleep Apnea using Cephalometric X-rays
Two-dimensionsal cephalometric imaging (almost always taken in the
sagittal plane) has been used to look for anatomic differences in OSA patients for
over 15 plus years. Schwab
38
reviewed the cephalometric literature and reported that
the most commonly reported skeletal abnormalities were (1) mandibular and
maxillary deficiency, (2) reduced dimension of the posterior airway space (measured
at the base of the tongue), (3) enlarged tongue, (4) enlarged soft palate, and a (5)
caudally displaced hyoid. Miles et al.
26
performed a detailed meta-analysis of the
11
literature that compare OSA and control groups and they concluded that the studies
varied widely in quality. Specifically, many of the controls groups were not matched
populations and frequently the sample size was often quite small. The meta-analysis
found that only the distance from the gonion to the gnathion (Go-Gn), a measure of
the lateral dimension of the mandible, consistently predicted OSA.
26
Summarizing
the voluminous cephalometric measurement studies that have tried to correlate facial
morphology with OSA severity lead to the conclusion that there is no consistent
agreement regarding which specific anatomic markers have the best correlation. This
inconsistency is due in part to the fact to the previously mentioned problem with
control groups and design but also because the studies usually use diverse
measurements and techniques. Finally, a major disadvantage of cephalometric
measurements is that the imaging is usually performed in an upright posture and a
lateral view of airway does not appreciate its 3-dimensional form since it cannot
capture the width of the airway. This last point is critical as most authors agree that
the oropharynx and hypopharynx combined represent the collapsible segment of the
airway and thus this area is the logical target for performing comparison of anatomic
features of OSA and non-OSA patients.

3-D measurements of Airway Caliber and Volume
When the actual caliber of the airway has been assessed, clear cut differences
between individuals with and without OSA have been reported by multiple
authors.
2,10
One study looked at 57 male patients with OSA using CT and reported a
12
significant association between airway collapsibility and both soft palate length and
hyoid bone position.
40
Another study, reported on six patients with OSA who were
subjected to paralysis and general anesthesia for measurement of airway
collapsibility.
15
They report that airway collapsibility was reversed with linear
relationship with increasing amount of mandibular advancement, thus establishing a
direct causal relationship between airway volume and collapsibility. Schwab et al,
37

found in awake subjects at the minimum airway area, that the thickness of the lateral
pharyngeal muscular walls rather than enlargement of the parapharyngeal fat pads
was the predominant anatomic factor causing airway narrowing in apneic subjects. In
a following study
39
the researchers found using MRI significantly increased risk of
sleep apnea the larger the volume of the tongue, lateral pharyngeal walls, and total
soft tissue. Trudo et al.
42
studied normal volunteers during wakefulness and sleep
with MRI and found a narrowing in the RP region of the airway due to posterior
movement of the soft palate, thickening of the lateral pharyngeal walls and an
increase in togue oblique distance. Chen et al.
3
found that the anteroposterior and the
lateral diameters of the retropalatal (RP) region, as well as the smallest area of the
RP region, are significantly smaller in subjects with OSA.
These data suggest that 3-dimensional airway assessment collected in supine
position has promise. The remaining question is, “can it be used to predict apnea risk
in the individuals before they develop full clinical cases of OSA?” There are two
basic methods used to capture airway dimension in a 3-D fashion. They include MRI
and CT. While MRI uses no radiation, it has the disadvantage of being relatively
13
slow, claustrophobic and expensive compared to CT and therefore far less likely to
be used as a screening tool. Computerized tomography of the airway has been
performed using helical imaging, cine CT, and cone-beam tomography. Each method
has unique advantages and disadvantages. Helical imaging is more expensive and
radiation is an issue. Cine CT can capture the dynamics of the airway
36
in a “movie”
of the sagittal view but is not commonly found in clinical care so it cannot be used as
a screening tool. Cone-beam CT’s are relatively quick with imaging times between
10 to 90 seconds for a maxillomandibular volume. In addition, they result in a
fraction of the effective absorbed dose of radiation
21,22
compared to medical CT. For
this reason as well as diagnostic yield, cost and access, cone-beam CT is rapidly
evolving into common clinical practice in dentistry for assessment of impacted teeth,
orthodontic evaluation, implants, sinuses, fractures, pathologies and implant
placement. Another important argument underlying our efforts to develop CT based
analytic methods for computing the upper airway volume and other morphological
variables that predict OSA risk is that these analytic procedures can be easily
automated in the future. If so, analytic algorithms that we develop could be applied
to any CT DICOM imaging system and a diagnostic test for OSA risk would become
a routine aspect of all dental and medical 3-D imaging procedures, thus capturing an
even larger potentially OSA at-risk group.

14
Upper airway segmentation, mesh modeling and volume calculation
The issue of airway volume determination involves segmentation from CT (or
MRI) which is an extremely well known area of research. In the area of segmentation,
original methods involved the time-consuming task of manually tracing structures
from slice to slice. This process is now possible with significantly less interaction from
the user. Programs such as 3D Slicer
7
, Mimics (Materialise N.V., Heverlee, Belgium),
Amira (TGS Inc., 5330 Carroll Canyon Road, San Diego, CA 92121-3758), CTPak
(University of Western Ontario) and Analyze 5.0 (AnalyzeDirect, Inc., 11425 Strang
Line Road, Lenexa KS 66215 USA) provide semi-automatic image-processing-based
segmentation and modeling from CT slices. These programs use the Generalized
Marching Cubes algorithm to create a polygonal wireframe mesh of the segmented
slices and export a 3-dimensional model in STL or other formats. Once the upper
airway or region of interest segmented, the area or volume (if multiple slices) can be
computed automatically. A very active research area is how to automatically segment
any craniofacial area; in particular we are interested in the upper airway region.
Previous reports concentrated on segmentation of the intrathoracic airway trees based
on fuzzy logic, rule-based methods, seeded region growing, deformable models or
shape-based interpolation. Segmentation of the upper airway from MRI is semi-
automatic in Liu et al.
20
User intervention is needed to select a rectangular region of
interest, to select the first and last slice defining the upper airway and multiple seeds or
voxels contained in the region of interest. In prior work we have implemented an
algorithm similar to Liu’s described in the Methods section and validated.
15
Significance
The important axiom relevant to our research is that “prevention is generally
far better use of health care resources than treatment of a chronic disease.” To
prevent, one must be able to predict and one argument for trying to predict OSA in
the incipient and usually subclinical stage of the disease is that atherogenesis starts at
a very early stage after sleep apnea onset, perhaps even during the first night. By the
time the patients are diagnosed with the syndrome, usually at the age of 47-50 years
when symptoms become apparent, they have already accumulated substantial
atherosclerotic insults which may be irreversible. Diagnosis at this age may be too
late.
16
Thus, to prevent cardiovascular morbidity, sleep apnea diagnosis and
treatment should be made as early as possible! If this can be done, then treatment can
be rendered thereby avoiding congestive heart failure and other heart diseases such
as pulmonary hypertension, cardiac arrhythmia, ischemic heart disease and stroke
48
.
Early intervention can also help those at risk to avoid weight gain, increased sick
time, poor job performance and accidents (sometimes fatal) either on the job site or
while driving. The open question is, “how can we predict OSA?” Our answer to this
question is contained in our hypothesis described in Chapter 3.

16
CHAPTER 3: HYPOTHESES
Primary hypothesis
• OSA patients (AHI≥15 events/hour assessed by ambulatory somnography)
have smaller upper airways than controls
Primary Null-hypothesis
• There is no significant difference in upper airway anatomy between OSA
patients (AHI≥15) and controls
Secondary hypothesis
• There is an association between OSA status and positive family history
• There is a significant association between having a “high risk” Berlin
questionnaire and OSA
• There is a significant association between having a “high risk” ARES
questionnaire and OSA
Secondary Null-hypotheses
• There is no significant association between OSA status and positive family
history
• There is no significant association between “high risk” Berlin questionnaire
and OSA status
• There is no significant association between “high risk” ARES questionnaire
and OSA status.

17
CHAPTER 4: SUBJECTS AND METHODS
Study design
This is a prospective case-control study since the outcome is relatively rare
(sleep apnea) and there is a second group of controls. The study exposure (CBCT
measurements and family history) was assessed on all the subjects at recruiting time.

Subjects
This prospective study included 39 OSA patients (5 female and 34 male,
mean age: 58.1 ± 11.2 years/old) and 23 controls (9 female and 14 male, mean age:
47.3± 13.8 years/old) enrolled between June 2006 and August 2008 (see recruitment
section below for details). All patients had been referred for a CBCT scan at the
Redmond Imaging Center, School of Dentistry, University of Southern California.
The control subjects were patients presenting for TMJ related purposes or snorers
whose AHI<15 events/hour. Data such as age, gender, BMI, race/ethnicity, blood
pressure, and number of siblings and parents was recorded. Each participant
answered the Berlin Questionnaire in Appendix and asked his siblings/parents to do
the same. After signing informed consent all subjects had a baseline ambulatory
sleep study described below to assess the outcome (OSA cases defined by an
AHI≥15 events/hour). Table 3 presents descriptive statistics of OSA and control
groups. This prospective study was approved by the Institutional Review Board of
the University of Southern California HSC-051050.
18
The inclusion criteria for OSA-cases were patients referred for sleep apnea;
they can be either male or female and of any ethnicity or race, but they must:
1. Be older than 15 years, so most of craniofacial development will be finished.
2. Give consent to access their 3D CBCT data
3. Be able to gather the Berlin Questionnaire (Appendix) from at least two members
of their immediate family (parents and siblings).
4. Be able to successfully utilize the ARES Unicorder for two-nights and have a
AHI≥15 (sign of OSA).
Our “control” group were patients who participated in the study and had
AHI<15, so they were recruited by the same means, flyers, mail and oral
communication. The benefit to all the subjects is that they received a $150 value
CBCT scan, as well as a radiology report about their scan, and a ambulatory
somnographic assessment (using the ARES Unicorder), if they collected their
immediate family history (parents and siblings) using the Berlin Questionnaire.
Inclusions for controls:
1. To be able to gather a Berlin Questionnaire from members of their immediate
family (parents and siblings);
2. To be able to successfully utilize the ARES device for two nights and have
AHI<15/hr.
3. Give consent to access their 3D cone-beam CT data
Exclusions for both controls and OSA-cases: Subject will be excluded if they:
19
1. are younger than 15
2. are unable to read and converse in either English or Spanish (or their parent/legal
guardian is unable if the subject is a minor), because in this case filling out the
questionnaire and/or following instructions for the in-home sleep study are
compromised.
3. Prior craniofacial surgery related to OSA.

Recruitment
All patients at Dr Clark’s Westwood clinic regardless of age, gender, race or
ethnicity were sent a letter explaining the study, the inclusion criteria and the
procedures involved in the study (sleep study, family questionnaires and CBCT
scan). We sent 328 letters, and 34 of those contacted through letters finally
completed the study. The most common exclusion criteria was the fact that subjects
had less than 2 family members alive. The other 30 subjects included in the study
were recruited orally or through flyers at the USC School of Dentistry (USC Faculty,
staff, students or CE course attendants). Of the 64 patients who signed the informed
consent only one did not complete the sleep study and one had an unusable CBCT,
leaving us with 62 subjects with CBCT plus sleep study data. Of the 63 subjects
with a complete sleep study, 4 subjects (3 controls and 1 case) had no family
members or had completely lost contact with them, and 5 subjects (1 control and 4
cases) never returned ANY family questionnaires leaving us with 54 records
including family history and sleep study.
20
Diagnosis of OSA
In this study two-night in-home studies for assessment of OSA were
performed with the Apnea Risk Evaluation System (ARES) Unicorder (Advanced
Brain Monitoring, Carlsbad, CA, USA) for all subjects. The ARES Unicorder
measured oxygen saturation, pulse rate, airflow, respiratory effort, snoring levels,
head movement, and head position from a wireless recorder self applied with a single
strap to the forehead. Reflectance oximetry is used to obtain the SpO2 and pulse rate
signals. Respiratory effort is derived from the measurement of changes in forehead
venous pressure acquired using a combination of photoplethysmography and changes
in surface pressure of the reflectance oximetry sensor, and head movement. Airflow
is obtained via a nasal cannula and a pressure transducer. A calibrated acoustic
microphone is used to acquire quantified snoring levels (dB). Accelerometers are
used to measure head movement and derive head position. The recorder was
designed to be easily affixed by the patient, and provide alerts during the study if
poor quality airflow or SpO2 is detected so the device could be adjusted. The
diagnostic sensitivity of in-lab ARES for diagnosing OSA using an AHI cut-off of 15
per hour was 95% and specificity was 94%, with a positive likelihood ratio (LR+)
=17.04, and negative likelihood ratio (LR-) = 0.06. For in-home ARES data the
sensitivity was 85% and specificity 91% (LR+ = 9.34, LR- = 0.17).
1
In the second
study, the concurrent in-laboratory comparison yielded a sensitivity of 97.4, a
specificity of 85.6, a positive predictive value of 93.6, and a negative predictive
21
value of 93.9; in-home comparison sensitivity, specificity, positive predictive value,
and negative predictive value were 91.5, 85.7, 91.5, and 85.7, respectively.
46

Cone-Beam Computed Tomography
All subjects were radiographically evaluated in supine position with
Frankfort plane perpendicular to the floor during their awake periods. This study
evaluated the 3D images obtained by a dentomaxillofacial volumetric imaging
system. This system was a cone-beam CT (Newtom QR 3G; QA sri, Via Silvestrini
20, 37135 Verona, Italy) and it acquires 360 images at 1-degree intervals in 36
seconds, with a resolution of 512*512 pixels and 12 bits per pixel (Figure 1). The
pixel size was 0.36mm with the 12” sensor. The NewTom® 3G CBCT has
specifically been designed for imaging of the maxillofacial complex. The overall
absorbed dose from this scan (57μSv) is roughly equivalent to a full mouth periapical
series of x-rays (33-84μSv).
21

CBCT Measurements
Scans were imported in the vWorks 5.0™ software (Cybermed, Korea). To
define the volumetric region of interest (VOI), we first selected a midsagittal image
of the airway and used the following planes as our upper and lower VOI borders:
The upper border was defined by a plane drawn parallel to the Frankfort plane and
going through the most distal point of the bony hard palate; the lower border was set
by a plane drawn parallel to the Frankfort plane and going through the most anterior-
22
inferior point of the second cervical vertebrae (see Figure 2A). To define the
anterior-posterior and lateral borders of our VOI we used the following convention:
as shown in Figure 2B, we selected in an axial view of the airway a large enough
square area that will always contain the airway within its borders (the software
allows the user to display up and down the axial images to check that the airway is
always contained in the square area). Between the upper and lower borders and
within the limiting axial square borders the airway VOI was determined. The actual
limits of the segmented airway involved one blinded operator tracing the soft tissue-
air interface for each axial slice using the Vworks software segmentation tools:
thresholding, region growing, dilation and erosion (Figure 3).

Figure 1 - Cone-Beam CT NewTom 3G.

23
From the segmented airway images, the software automatically computed the
surface mesh shown in Figure 4 and the volume of the oropharyngeal region in mm
3
.
The program also automatically reports the cross-sectional area in mm
2
for each
axial slice, so to find the slice with the minimum airway locus the operator has to
visualize all the slices in the upper airway region and select the smallest. Once we
identified the slice with the smallest computed cross-section area, we measured its
antero-posterior (AP) and lateral (L) dimensions (Figure 5).

Figure 2 - The oropharyngeal region selected by the user on (A) axial view and
(B) mid-sagittal view of the upper airway (between PNS and the most anterior
inferior point of the second cervical).

(A) (B)
24
Figure 3 – Cross-sectional upper airway.

Figure 4 – 3D airway surface superimposed on sagittal view of a healthy patient
(A) and OSA patient (C); 3D airway surface for a healthy patient (B) and OSA
(D).
(A) (B)

25

Figure 5– Antero-posterior (7.3mm) and lateral dimension (17.7) of the cross-
sectional airway.

To report the location of the narrowest locus in the airway some researchers
use the retropalatal (RP: between PNS and the tip of the uvula) and retroglossal (RG:
below the tip of the uvula) regions as intraoral landmarks.
8,18
All our subjects
presented the smallest caliber in the retropalatal region and presented soft palates
with different sizes, so in this study we used the occlusal plane to define a third
((C) (D)
26
intraoral reference allowing to classify the airway in the upper occlusal (UO) and the
lower occlusal (LO) regions.

Figure 6– Occlusal plane is defined by the incisal point on mid-sagittal image
(A) and the first molars (B).

(A)

(B)

27
Using these data we were able to verify:
• the total airway volume
• of the oropharyngeal region (Figure 4),
• the smallest cross-section area (Figure 3),
• the anteroposterior (AP) and lateral (L) dimensions of the smallest
cross-section area (Figure 5),
• the relationship between AP and L (AP/L),
• the localization of the smallest cross-section area in the upper (UO) or
lower (LO) oropharyngeal region using the oclusal plane defined in
Figure 6,
• the configuration of the smallest cross-section area in the upper airway
as rounded, elliptic, square or concave,
• the vertical (VSP) and horizontal (HSP) soft palate length (Figure 7).
HSP is the distance from the posterior nasal spine to the vertical line
going through the most posterior contour of soft palate (in the mid-
sagittal plane). VSP is the distance from the horizontal line going
through the posterior nasal spine to the tip of soft palate.
• the total airway length from PNS to the most anterior-inferior border of
the 2nd vertebrae,
• the ratio between the vertical soft palate length and the airway length
(VSP/airway_length).
28
Identification of the dentoskeletal landmarks and subsequent measurements
were manually performed by a single investigator for both groups. Twenty patients
were randomly selected and their images re-sliced and re-measured by the same
examiner after a 60-day interval. All data was stored on the Research computer
located in room 4112B at the USC School of Dentistry.

Figure 7 – Vertical and horizontal soft palate length measurements.

Questionnaires
Cases and controls answered the same questionnaire (Appendix) including
personal demographics and the Berlin Questionnaire itself. In the demographics
section we collected for each subject and its family members: gender (female/male),
29
age in years and months, weight in pounds, height in feet and inches, systolic and
diastolic blood pressure, alive mother (yes/no), alive father (yes/no), number of
brothers and number of sisters. The Berlin Questionnaire contains 11 questions
divided in three categories of questions related to snoring/sleeping patterns. One
introductory question and four follow-up questions concerned snoring, witnessed
apneas, and the frequency of such events. Three questions addressed daytime
sleepiness, with a sub-question about drowsy driving. One question asked for a
history of high blood pressure (BP). The exposure assessment was not verified in any
way by the researcher and were based only in participant’s self-assessment.
The answers to the questions were converted to dichotomous or categorical
variables for the analysis. The scoring system has been described in Appendix. For each
subject and family member we obtain a “high risk” or “no risk” score for OSA. We
defined the dichotomous variable FBQ as 1 if at least one family member is at high-risk
for OSA, and 0 otherwise. We also computed the FBS as the number of family members
with a positive Berlin Questionnaire divided by the number of family members (parents
and siblings). Calculations were repeated with a dichotomous variable FB=1 if FBS≥0.5
and 0 otherwise.

Statistical Analyses
Descriptive statistics (means and standard deviations for continuous variables
and frequencies and percentages for dichotomous and categorical variables) were
obtained for each measurement. Histograms and Kolmorogov-Smirnov tests were
30
used to check for normality of all variables. To assess differences between means of
normal continuous variables we used the independent t-test, and the Wilcoxon
Signed Ranks Tests was used to compare non-parametric continuous variables from
independent samples. Chi-square test was used to compare categorical variables (i.e.
shape of the airway), proportions or dichotomous variables such as gender. If the
prevalence of a variable was less than 5, the Fisher's exact test was used. Since
overall AHI values failed to pass the Kolmogorov-Smirnov test for normality, the
AHI values were log-transformed. The transformed data were found to be normally
distributed, and then used for multiple linear regression analyses. We assessed

the
correlations between continuous variables and AHI using the nonparametric

Spearman rank test because AHI was not normally distributed. Potential associations
between log(AHI) and individual risk

factors were assessed for significance with a
2

test for binary

variables and standard regression techniques for both continuous and
categorical variables. The linear regression assumptions of linearity,
homoscedasticity and normality of the residuals were successfully evaluated.
Multiple linear regression with stepwise selection of variables was performed to
evaluate the relation of log(AHI) and those variables, which showed a significant
association in the univariate model or for comparison with prior work. Median data
was used to categorize the continuous variables selected in the final model, and
stepwise logistic regression analysis was carried out to assess the adjusted odds-
ratios of all risk factors in the final model on OSA status. All variables left in the
final model were significant at the 0.15 level. Regression analyses were repeated
31
with stepwise regression and minimum Mallow's Cp. Primary data management was
performed using the Statistical Package for the Social Sciences (SPSS-PC+ for
Windows; SPSS; Chicago, IL). Statistical analyses were performed using SAS
System for Windows (Version 9.0 or later, SAS Institute, Cary, NC). A p-value less
than 0.05 will be used to determine statistical significance.

Reliability of Measurements
To test the reliability of the CBCT measurements, 20 patients were randomly
selected and their images re-measured by the same examiner after a 60-day interval.
Statistical analysis of the difference between the duplicate measurements was
conducted by deriving the Intraclass Correlation Coefficient.

Sample size calculations
To determine how many OSA and non-OSA subjects might be needed to be
able to distinguish them using both the Berlin Questionnaire and the 3-D CT
measures we performed the following calculations. Using the pilot data in Table 1,
we calculated the equal number of cases and controls necessary for 80% power to
find a statistically significant (two-sided alpha=.05) difference between the
measurements of each of the CT variables: AP, L, volume, minimum area.
Assuming that 32.3% of subjects will be classified as high-risk by the
Subject Berlin questionnaire, and using Netzer’s data
27
showing that the
unadjusted odds ratio for an AHI>15 given high-risk status is 6.3, we calculate a
32
p2=.75, and a sample size of 21 cases and 21 controls to provide 80% power with a
5% two-sided significance level. A sample size of 39 cases and 23 controls would
provide us with 92.6% power.
For each of the CBCT variables and the Berlin questionaire the required
sample size is shown in the Table 2. In addition, the table shows the power that
would be provided for each variable with our final sample size of 39 cases and 23
controls. The final sample size gave us at least an 82% power on all variables except
volume, which should be considered only in secondary analysis.

Table 1 – Pilot data: Comparison between OSA+ and controls.
Variable OSA
Average ± SD
Controls
Average ± SD

P value†
Gender F=2, M=8 F=2, M=8 ------
Age 52.9±14.7 45.4±19.5 0.496
BMI (kg/m
2
) 29.5±9.05 23.1±3.05 0.034
Volume (cm
3
) 4868.4±1864.0 6051.7+1756.4 0.054
Min. area (mm
2
) 45.8±17.5 146.9+111.7 0.011
AP (mm) 4.57±1.2 7.76+3.31 0.009
L (mm) 11.6±4.5 16.2+6.8 0.104
† Non-parametric Wilcoxon test.

33

Secondary Analysis Sample Size
To compute the sample size required to distinguish OSA from non-OSA
subjects using the immediate family history Berlin score, we next utilized the
number of adults with a positive (or at “high risk” for OSA) score with the Berlin
questionnaire derived from the Netzer et al., (2003)
27
study. They reported that
32.3% of screened subjects in a broad range of primary care settings in USA,
Germany and Spain had a high pretest probability for OSA. Therefore, for FBQ
(family Berlin Questionnaire, dichotomous variable), we expect 32.3% of the
controls to have a positive family history. If we expect family history to be
associated with OSA with an odds ratio of 4.0, p2=0.656, 80% power to detect an
increased risk of OSA associated with family history at a 5% two-sided significance
level can be attained with a sample size of 35 cases and 35 controls. A sample size
of 35 cases and 19 controls would provide us with 66.1% power (39 cases and 23
controls gives a power of 73.2%).
For family history score (FHS = positive family members / number of
family members), the mean score for controls will be 0, but we expect a score of
about .5 for the cases. Assuming a standard deviation for both cases and controls of
about .75, 80% power to detect a difference between FHS for the cases versus the
controls at a 5% two-sided significance level can be attained with a sample size of 36
cases and 36 controls. A sample size of 35 cases and 19 controls would provide us
with 64.8% power.
34
In conclusion, we do not have enough sample size to get to 80% power in our
secondary analysis of family history data, however we can use this pilot study to
compute the final sample size needed for a future study.

Table 2 - A priori sample size and power.
Variable Sample size
for 80% power
(α = .05)
Power for sample size of 39
cases and 23 controls
(α = .05)
Volume N1=N2=37 .71
Min area N1=N2=10 .99
AP N1=N2=10 .99
L N1=N2=25 .82
Berlin Questionnaire N1=N2=21 .93

35
CHAPTER 5: RESULTS

Descriptive statistics
Table 3 presents the subjects’ gender and mean age, Body Mass Index (BMI),
neck circumference and somnographic variables defined in Chapter 2 (Apnea-
Hypoapnea Index and Apnea Index). The two groups were statistically significantly
different in mean age, BMI and neck circumference, with the cases older, heavier
and with thicker necks than controls as expected. The control’s mean age was 10
years lower than the mean age of the cases.
Cases were mostly males (34/39), and gender distribution was statistically
significantly different from controls (p=.0167). Control’s mean neck circumference
was 2 inches narrower than cases. At baseline the two groups were statistically
significantly different in apnea severity as measured by apnea events per hour
(p<.0001) and apnea-hypoapnea index (p<.0001). This is to be expected as OSA
cases were patients with an AHI≥15/hr and otherwise the subject was deemed a
control. There were no statistically significant differences in race or ethnicity
distribution between the two groups (Tables 4, 5).

36
Table 3 -Descriptive statistics for OSA and control groups.
OSA patients
Mean ± SD
(N=39)
Controls
Mean ± SD
(N=23)
T-Test
p-value
Gender 34M : 5F 14M : 9F 0.0167†
Age (yrs) 58.1 ± 11.17
Range:[29.3-80.5]
47.3 ± 13.79
Range:[24.7-68.1]
0.0013
Neck Circumference
(inch)
16.4 ± 1.44 14.6 ± 1.25 <.0001
Body Mass Index
(kg/m
2
)
27.9 ± 3.89
Range:[21.5-36.0]
24.6 ± 3.37
Range:[18.2-33.5]
0.0016
AHI (events/hour) 32.6 ± 16.80
Range:[15-79]
7.0 ± 3.21
Range:[2-13]
<.0001
Apnea Index
(events/hour)
13.8 ± 17.11
Range:[0-68]
1.2 ± 1.37
Range:[0-4]
<.0001
† Chi-square

37
Table 4 - Comparison of race distribution between OSA and control groups.
Race OSA patients
(N=39)
Controls
(N=23)
Chi-square
p-value
1=American Indian or
Alaska native
0 1

0.235
2=Asian 9 2
3=Native Hawaiian or
other Pacific islander
0 0
4=Black or African
American
1 0
5=White 29 20

Table 5 - Comparison of ethnicity distribution between OSAs and controls.
Ethnicity OSA patients
(N=39)
Controls
(N=23)
Chi-square
p-value
1=Hispanic 6 1
0.185
2=Non Hispanic 33 22

38
Cone-beam CT Imaging
Table 6 presents the results of the independent t-tests (or Wilcoxon test for
non-parametric variables) comparing cases and controls. The smallest cross-section
area (p=0.015) and the lateral dimension (p=0.015) showed statistically significant
group differences. There were no statistically significant group differences in mean
AP, airway length, vertical and horizontal soft palate, and total volume or in median
AP/Lateral and VSP/airway ratios (p>0.05). Considering the AP and L relationship
of the narrowest slice of the airway, the OSA patients presented a slightly less
elliptical shape (AP/L=0.38±0.19) than the non-OSA subjects (AP/L=0.32±0.14), but
no significant differences were found between groups.
The location of the smallest cross-section area was retropalatal in all subjects
(above the tip of the uvula). The position of the minimum cross-sectional airway as
UO or LO showed no significant differences (p=0.929, Table 7). There was no
significant difference between groups in shape of the smallest cross-sectional airway
(p=.367), however only one control over 23 had a different shape than transverse
compared to 6/39 OSA patients. (Table 8)
39
Table 6 – Comparison of CBCT measurements between OSAs and controls.
Parameter

OSA group
Mean ± SD
(N=39)
Control group
Mean ± SD
(N=23)
T-TEST
P-value
Minimum Area (mm
2
) 70.6 ± 41.71 99.0 ± 45.59 0.015
AP (mm) 5.4 ± 1.83 6.00 ± 2.47 0.257
Lateral (mm) 15.9 ± 5.77 19.7 ± 5.51 0.015
AP/Lateral 0.38 ± 0.19 0.32 ± 0.14 0.220‡
Airway Length (mm) 44.85 ± 4.80 45.35 ± 5.90 0.717
Vertical Soft Palate (mm) 34.8 ± 6.92 34.4 ± 7.81 0.845
Horizontal Soft Palate (mm) 22.2 ± 4.13 21.3 ± 4.28 0.425
VSP/Airway Length 0.78 ± 0.15 0.76 ± 0.17 0.705‡
Volume (cm
3
) 7.79 ± 3.48 8.71 ± 3.24 0.306
FBQ (score) 0.31 ± 0.37 0.40 ± 0.31 0.221‡
‡ Wilcoxon independent samples non-parametric test

40
Table 7 - Comparison of minimum cross-sectional airway location between OSA
and control groups.

Location of minimum
cross-sectional airway
OSA patients
(N=39)
Controls
(N=23)
Chi-square
p-value
Upper Occlusal 14 8
0.929
Lower Occlusal 25 15

Table 8 - Comparison of minimum cross-sectional airway shape between OSA
and control groups.

SHAPE of minimum cross-
sectional airway
OSA patients
(N=39)
Controls
(N=23)
Chi-square
p-value
1=rounded 4 1

0.367
2=transverse 33 22
3=square 2 0
4=AP 0 0

Predicting OSA with CBCT measurements
In this section we explore the linear relationships between the CBCT
variables and possible confounders BMI, age and gender. After adjusting for age and
gender, we found a significant correlation between BMI and lateral dimension of the
41
airway (r= -.56, p=.0005), as well as BMI and minimum cross-section (r= -.36,
p=.034) in OSA cases but not in controls. Similar correlation was found after
adjusting for BMI and gender between age and lateral dimension (r=.46, p=.005),
and age and volume of the airway (r=.42, p=.0113) in cases only using the Spearman
correlation coefficient. In the linear regression models we adjusted for age, gender
and BMI based on prior knowledge and the significant differences between groups at
baseline (Table 3).
Stepwise linear regression analyses were performed with log(AHI) as the
dependent variable. Explanatory variables included age, gender, BMI, neck
circumference (common confounders for OSA and significantly different at baseline
between groups), plus the CBCT variables (airway volume, min. area, airway length,
AP, L, VSP, HSP). We repeated our analysis with the ARES questionnaire (model1,
Table 9) and the Berlin Questionnaire (model2, Table 9). For each model we
performed a stepwise forward regression analysis and minimum Mallow’s “Cp”
criteria. All variables left in the model were significant at the 0.10 level. Both
model1 and model2 converged to a 4-variable model (Table 9). No other variable
met the 0.15 significance criteria to enter the model. Both model1 and model2 had
almost identical adjusted R-square (50.8% for model1 and 50.7% for model2), and
similar “Cp” value (1.703 for model1 and 1.286 for model2). For comparison, the
adjusted R-square of a simple model with age, gender and BMI was 30.6%.
Stepwise logistic regression analysis was carried out to assess the adjusted
odds-ratios of all risk factors in the final model1 (Table 10) and model2 (Table 11)
42
on OSA status. Age was categorized using median age (57 years old), as well as the
lateral dimension (<17 mm). Male gender, age >57 years, a “high risk” ARES and a
lateral dimension<17mm were predictive of OSA (Table 10). Male gender, age >57
years, a positive Berlin Questionnaire and a lateral dimension < 17mm were
predictive of OSA in model2 (Table 11).

Table 9 – Stepwise multiple linear regression analyses of the risk factors on the
log(AHI).
STEPWISE LINEAR
REGRESSION
MODEL 1 †
STEPWISE LINEAR
REGRESSION
MODEL 2 §
Risk factors β P-value β P-value
Gender (female) -0.95 <.0001 -1.08 <.0001
Age 0.01 .0479 0.03 <.0001
ARES Q.

0.86 .0001 ------- -------
BERLIN Q. ------- ------- 0.69 .0002
Upper Airway Lateral
dimension (mm)
-0.06 .0001 -0.06 <.0001
† Multivariate ANOVA F-test p<0.0001, adjusted R square= 0.508, Cp=1.703
§ Multivariate ANOVA F-test p<0.0001, adjusted R square= 0.507, Cp=1.286.

43
Table 10 - Logistic stepwise regression analysis of the risk factors for OSA
Risk factors Adjusted OR Chi-square
p-value
[95% C.I.]
Male 14.8 .003 [2.5 , 89.1]
Age ≥ 57 years 6.3 .018 [1.4 , 29.0]
ARES Questionnaire
(High risk)
8.2 .010 [1.6 , 41.0]
Upper Airway Lateral
dimension < 17 mm
4.1 .053 [0.98, 17.4]

Table 11 - Logistic stepwise regression analysis of the risk factors for OSA
Risk factors Adjusted OR Chi-square
p-value
[95% C.I.]
Male 23.7 .002 [3.3 , 170.7]
Age ≥ 57 years 14.4 .002 [2.7 , 77.7]
Berlin Questionnaire
(High risk)
8.4 .008 [1.8 , 40.4]
Upper Airway Lateral
dimension < 17 mm
5.9 .024 [1.3 , 27.8]

Screening for OSA
Two questionnaires were used in this prospective study that could be
predictive of OSA, one is the 11-question on Appendix, the Berlin Questionnaire.
The second one is mandatory before performing ambulatory somnographic
assessment with the ARES Unicorder (ARES questionnaire). In Table 12, we
compared with Chi-square test the two groups and found a significant association
between a “high risk” ARES questionnaire and OSA status (p=.003), but not for the
44
Berlin questionnaire (p=.095). The Sensitivity, Specificity, Positive and Negative
Predictive Values for the two questionnaires against ambulatory somnographic data
are presented in Table 13. We repeated the analysis with two cutpoints for OSA
(AHI≥15) and (AHI≥10). The ARES questionnaire had higher sensitivity, lower
specificity, similar PPV and higher NPV than the Berlin questionnaire. Both
questionnaires were significantly associated (p=.0043, Table 14).

Table 12 – Screening for OSA with the Berlin Questionnaire and the ARES
questionnaire.
Questionnaire OSA patients
(N=39)
Control group
(N=23)
Chi-square
p-value
(OR) [95% C.I. OR]
Berlin 27/39
(69.2%)
11/23
(47.8%)
0.095
(OR=2.45) [0.9,7.1]
ARES 35/39
(89.7%)
13/23
(56.5%)
0.003
(OR=6.7) [1.8,25.3]

45
Table 13 – Sensitivity, Specificity, Positive and Negative Predictive Value for
two questionnaires against ambulatory somnographic data.

Parameter BERLIN ARES
AHI≥15 AHI≥10 AHI≥15 AHI≥10
SENSITIVITY 69.2% 71.1% 89.7% 88.9%
SPECIFICITY 52.2% 64.7% 43.5% 52.9%
PPV 71.1% 84.2% 72.9% 83.3%
NPV 50% 45.8% 71.4% 64.3%

Table 14 – Association between ARES and Berlin questionnaires.

ARES Q. + ARES Q. - McNemar’s
p-value (OR)
[95% CI OR]
Berlin Q. + 34 4 0.0043
(OR=6.07)
[1.62,22.6]
Berlin Q. - 14 10

Family history as a predictor of OSA
Every subject enrolled in the study was given sufficient questionnaires and
self-stamped envelopes and was asked to get his/her parents and siblings to answer
the Berlin questionnaire. Repeated attempts by telephone and mail were made after
1, 6 months and 1 year to contact the subject asking to get those questionnaires. Only
46
4 cases and 4 controls never complied. In this study, there was a significant
association between having at least one family member at “high risk” for OSA
(FBQ+) and OSA status. Due to self-selection bias the controls in this study have a
higher prevalence of family members with sleep apnea than the cases (see Table 15),
so the Odds-ratios are underestimated (OR=0.03). However, there was no significant
difference in the proportion of subjects with a family history score [number of family
members with BQ+ / total family members] ≥ 0.50 between groups. The reader is
referred to the Limitations section for a discussion of the biases in the study.

Table 15 – Association between Family History and OSA.

Parameter Cases
(N = 35) †
Controls
(N=19) †
Chi-square
p-value (OR)
[95% CI OR]
+ FBQ
(at least 1 family
member BQ+)
17/35
(48.6%)
15/19
(79.0%)
0.03 (OR=0.25)
[0.1, 0.9]
Family Score ≥ 0.5 14/35
(40.0%)
8/19
(42.1%)
0.881 (OR = 0.92)
[0.3, 2.9]
† 4 cases and 4 controls never returned their family member questionnaires, so data
is only available in N=35 cases and N=19 controls.

47
Error Analysis
Linear measurements were highly reproducible with mean Intraclass
Correlation Coefficient of 0.977±0.01 for airway length, AP, lateral dimension, VSP
and HSP. The ICCC was 0.965 for volumetric measurements and 0.979 for
computing the minimum cross-sectional airway. These are highly reproducible
results.

48
CHAPTER 6: DISCUSSION
Demographics
The goal of this study was to assess the differences in upper airway
morphology and family history between Obstructive Sleep Apnea adult patients and
controls. The imaging device chosen for this research is a low radiation dental cone-
beam CT which is becoming very popular in dental offices for dental implant
assessment, orthodontics treatment and planning, and Temporomandibular joint
diagnosis. The two groups of subjects were recruited in the same fashion by mail,
flyers and oral communication at a private dental office and the University of
Southern California School of Dentistry. Every subject who qualified was recruited
regardless of their race, gender, age or BMI and the final classification of their OSA
status was based on a standard medical classification (AHI≥ 15/hr) based on a sleep
report independent of the investigator by a blinded sleep expert, so we do not expect
any investigator bias. However, volunteer bias is clearly a problem in this design as
we had a poor recruitment success, and 79% of the controls had a positive family
history. The imaging and sleep data are quantitative data performed by blinded
operator so no bias is anticipated except measurement bias, however Intraclass
Correlation Coefficient was high. Because these were continuous patients, they were
unmatched by age, gender and BMI as in most of prior studies.
30,24
The only large
case-control study had patients matched by gender and race, but not age or BMI.
39

Cases were mostly males as expected with prevalence of OSA double in males than
females.
51
Cases were 10 years older than controls which is not ideal but consistent
49
with prior studies by Mayer et al.
24
(cases 5 years older than controls) and Okubo et
al.
30
(12 years older). Cases had a larger BMI by 3.3 kg/m
2
, as in Mayer et al. (3
kg/m
2
). That is much better than the study by Schwab who had a BMI difference of
10 kg/m
2
. Sample size was relatively large compared to prior studies except Mayer’s
and Schwab’s but those used MRI as the imaging modality. In this study there were
no significant differences in race or ethnicity between the two groups with a large
majority of our subjects White and non Hispanic.

CBCT Imaging
This is to our knowledge the first study with a large sample size comparing
OSA and controls with CBCT. CBCT has the double advantage compared to MRI
that it is accessible and cheap, however there is radiation involved though lower than
medical CT.
22
CT and CBCT provide better bony tissues delineation compared to
MRI, and the latter provides better soft tissue contrast.
In preliminary results with only 10 cases and 10 controls we had found a
negative significant correlation between BMI and the minimum cross-section area,
and BMI and lateral dimension. This result has been corroborated in this larger
sample with a significant correlation between BMI and lateral dimension of the
airway (r= -.56, p=.0005). This is consistent with Welch’s study were after following
a weight loss program the lateral dimension increased significantly from 6.8mm to
10.4mm.
45
We also found a significant correlation between BMI and minimum
50
cross-section (R= -.36, p=.034) in OSA cases but not in controls which is consistent
with Schwab’s study (R= -0.53; p=.06), and Cosentini’s study (R= -.36; P,.05).
4,36

In this study the cases had a slightly more spherical airway shape than controls
(AP/L=0.38 in cases compared to 0.32 in controls), which is similar to Mayer’s
result of 0.41 in cases and 0.39 in controls for subjects with BMI<27 kg/m
2
. Cases
also had a narrower lateral dimension compared to controls. This is consistent with
Hora et al.
12
who found in awake patients using MRI that the transversal (lateral)
dimension of the airway at the retroglossal level was an independent predictor of
OSAs, and this lateral narrowing of the airway was not associated with increased fat
pad thickness.
1
The authors conclude that the narrowing of the lateral airway is due
to parapharyngeal muscular hypertrophy not the accumulation of fat. Concerning the
smaller cross-sectional area found in cases compared to controls, this is consistent
with Schwab’s study were cases had two times smaller cross-sectional area than non-
apneic patients.
36

Screening for OSA
Because of the study design (case-control study) we cannot compute
prevalence or incidence of OSA, however the odds-ratios are a very good estimate of
the relative risk, if the cases in the study represent the cases in the general population
and the controls in the study represent the controls in the general population. In this
study, a subject having a “high risk” ARES questionnaire (which includes
demographic data, medical history and the Epsworth questionnaire), was 6.7 times
51
more likely to have OSA than a subject with “low risk” or “no risk” score. For a
typical screening tool, sensitivity is the most important accuracy criteria. However,
to convince employers to screen for OSA, specificity is also important because of the
costs associated with false positive cases. In a prior publication by the company it
was reported that the ARES algorithm for assigning OSA risk provided a sensitivity
and specificity of 94% and 79% respectively (PPV=91%, NPV=86%),
17
compared to
the Berlin Questionnaire which reported a sensitivity and specificity of 86% and
77% for predicting an RDI greater than 5.
28
In our sample using a clinical criteria of
AHI>10, the ARES questionnaire had a sensitivity of 88.9%, specificity of 52.9%, a
Positive Predictive Value of 83.3% and Negative Predictive Value of 64.3%,
compared to 71.1%, 64.7%, 84.2% and 45.8% respectively for the Berlin
Questionnaire. Though prior studies by other groups were based in different
populations, sleep studies and not ambulatory somnography, and our sample is
smaller, our results confirm that the ARES questionnaire has better sensitivity, worse
specificity, similar PPV and better NPV than the Berlin. In conclusion, the ARES is
only worse in its ability to identify correctly individuals who truly do not have the
disease.

Predicting OSA
One of our goals is to predict OSA based on imaging and family history. Our
results comparing family history between cases and controls were definitely
influenced by “self-selection bias” underestimating the OR, so we decided not to use
52
the family history data in the final regression model. Because the ARES
questionnaire and the Berlin questionnaire are very similar in function and form we
performed the linear regression twice: model1 included the ARES and model2 the
Berlin questionnaire. Both models gave very similar adjusted R-square and Cp, so
their prediction power is very similar. In conclusion, the final model included age,
gender, Berlin or ARES questionnaire and the lateral dimension of the airway. Based
on the fact that ARES questionnaire has better sensitivity, the model with the ARES
questionnaire is preferred. This model explained 50.8% of the variability of the
severity of OSA [log(AHI)]. After dichotomization of the 4 variables in the final
model using the median as the cut-off point we conclude that in this study a male
was 14.8 times more likely to have OSA than a female; subjects older than 57 years
old were 6.3 times more likely to have the disease than younger subjects; a subject
having a “high risk” score at the Berlin questionnaire was 8.2 times more likely to
have OSA than a subject with low risk, and a subject with an upper airway lateral
dimension of the cross-sectional airway less than 17 mm was 4.1 times more likely
to have OSA.
Prediction for OSA based on this 4-variable model should be implemented in
a large community-based prospective study to confirm these findings. The reader is
referred to the Limitations section for a discussion on possible biases affecting this
study, and the conclusion section for a final summary of the findings.
53
CHAPTER 7: ASSUMPTIONS

1. The distortion and magnification of NewTom® images was statistically
insignificant.
2. Measurements were accurate and reproducible.
3. The main confounders in this study (gender, age and BMI) were considered
in the final model.
4. The OSA cases recruited from Dr Clark’s practice and from USC School of
Dentistry (Faculty, staff and students) through flyers, mail and personal
communication are assumed to represent the OSA population from Western
Los Angeles. The controls are assumed to represent the general population
from Western Los Angeles, however our population suffered from “self-
selection bias” in terms of family history of sleep apnea so the results on
family history are not generalizable.
5. See section on Limitations for specific biases.

54
CHAPTER 8: LIMITATIONS

1. Most of the subjects enrolled were White non-hispanic, so the results in this
thesis should not be generalized to other races or ethnicities, or to patients
younger than 32 or older than 80 year’s old. Most of our subjects were male
so the results may not generalize to women.
2. The study only used Dr Clark’s patients or faculty/staff/students at USC as
opposed to the general population for OSA cases and controls. In particular
our controls (patients with AHI<15/hr) had more chances to have a family
history of OSA than the general population as well as students attending CE
courses on OSA have more chances to have OSA in their families. This
“volunteer bias” will underestimate the association between a positive family
history and OSA, as seen in Table 15.
3. Measurement bias: Because the results are based on self-report and this is a
case-control study, there is a potential for recall bias of the exposure.
Cases may be more likely to recall and report snoring loudness, apneas or
high-blood pressure than controls, because they might think more about their
sleep quality and sleep patterns. This will increase the cell “a” in a 2x2 table,
increase the OR, overestimate the association (this will result in positive bias,
away from the null).
55
4. Because of the type of design, case-control study, the exposure and the
outcome were assessed at the same time, so we cannot assess the temporal
relationship between upper airway changes and OSA.
5. Because the interviewer was not blinded to the outcome status in the majority
of the cases, and exposure information was not validated, observer bias is
quite possible for the Berlin questionnaire and the family questionnaires. The
interviewer may probe for more details on past apneas, snoring loudness or
tiredness after sleeping when interviewing the cases than the controls. This
will increase cell “a”, overestimating the odds-ratios for a positive Berlin
questionnaire (positive bias, away from the null). However the operator
performing the CBCT measurements and the sleep expert creating the sleep
report were blinded to outcome status, so we do not expect observer bias in
those measurements.
6. Selection bias. This study is a very clear case of selection bias as there was a
very low response rate (self-selection bias). Persons with a specific
combination of exposure (family history or craniofacial anomalies) and
outcome may self-select themselves to participate in a case-control study.
The flyer specifically addressed “The University of Southern California’s
School of Dentistry is seeking patients who snore or may have sleep apnea
(pauses of breathing during sleep) for a research study designed to identify
predictive markers of sleep apnea based on family history and CT images of
the throat.”
56
7. There is opportunity for selection bias of controls. The controls were patients
who had no sleep apnea severe enough to qualify to be a case. It could create
bias if those who participated were different from those that did not in terms
of exposures analyzed in the study.

57
CHAPTER 9: SUMMARY

The goal of this study was to compare the Cone-Beam Computerized
Tomography (CBCT) scan measurements between patients with OSA and controls to
predict OSA in the future based on imaging and family history. OSA patients were
predominantly male, older, had a larger neck size and larger Body Mass index than
controls. In conclusion, age>57, male gender, a positive Berlin Questionnaire and
narrow lateral dimension were identified as significant risk factors for OSA. The
lateral dimension of the airway has been negatively correlated with BMI, and neck
circumference. Screening for OSA could be implemented based on this work,
however further studies are needed to generalize our findings.

58
CHAPTER 10: CONCLUSIONS

1. In this prospective case-control study, the cases were almost 10 years older,
were 3.3kg/m
2
heavier and had 2 inches thicker neck circumference. Cases
had a mean AHI of 32.6 events/hour compared to 7.0/hr for the controls.
2. There were no significant differences in race, or ethnicity between the two
groups.
3. The smallest cross-section area (p=0.015) and the lateral dimension
(p=0.015) showed statistically significant group differences.
4. There were no statistically significant group differences in mean AP, airway
length, vertical and horizontal soft palate, and total volume or in median
AP/Lateral and VSP/airway ratios (p>0.05).
5. The location of the smallest cross-section area was retropalatal in all subjects
(above the tip of the uvula), and the position of the minimum cross-sectional
airway as upper or lower occlusal showed no significant differences between
groups (p=0.929).
6. There was no significant difference between groups in shape of the smallest
cross-sectional airway (p=.367).
7. After adjusting for age and gender, there was a significant correlation
between BMI and lateral dimension of the airway, as well as BMI and
minimum cross-section in OSA cases but not in controls.
59
8. After adjusting for BMI and gender, there was a significant correlation
between age and lateral dimension, and age and volume of the airway in
cases only.
9. In this study, subjects having a “high risk” ARES questionnaire were 6.7
times as likely to have OSA (AHI>15 events/hr) as subjects with “low or no
risk” score (p=.003).
10. The ARES questionnaire had a sensitivity of 89.7%, specificity of 43.5%, a
Positive Predictive Value of 72.9% and Negative Predictive Value of 71.4%,
compared to 69.2%, 52.2%, 71.1% and 50% respectively for the Berlin
Questionnaire.
11. Final stepwise linear regression model explained 51.4% of the variability of
the severity of OSA [log(AHI)], and included age, gender, the Berlin
questionnaire [or ARES questionnaire] and the lateral dimension of the upper
airway.
12. In this study, a male has 23.7 times more likely to have OSA than a female,
subjects older than 57 were 14.4 times more likely to have the disease than
younger subjects; a subject having a “high risk” score at the Berlin
questionnaire was 8.4 times more likely to have OSA than a subject with low
risk, and a subject with an upper airway lateral dimension of the cross-
sectional airway less than 17 mm was 5.9 times more likely to have OSA.

60
BIBLIOGRAPHY
1
Ayappa I, Norman RG, Seelall V, Rapoport DM. Validation of a self-applied unattended
monitor for sleep disordered breathing. J Clin Sleep Med 2008; 4:26-37.

2
Bradley TD, Brown IG, Grossman RF, Zamel N, Martinez D, Phillipson EA, Hoffstein V.
Pharyngeal size in snorers, nonsnorers, and patients with obstructive sleep apnea. N Engl J
Med 1986;315:1327-1331.

3
Chen NH, Li KK, Li S-Y, Wong C-R, Chuang M-L, Hwang C-C and Wu Y-K. Airway
Assessment by Volumetric Computed Tomography in Snorers and Subjects With
Obstructive Sleep Apnea in a Far-East Asian Population (Chinese). The Laryngoscope
2002;112: 721-726.

4
Cosentini T, Le Donne R, Mancini D, Colavita N. Magnetic resonance imaging of the
upper airway in obstructive sleep apnea. Radiol Med 2004;108:404-416.

5
Decker MJ, Hoecke PL and Strohl KP. Ambulatory monitoring of arterial oxygen
saturation. Chest 1995;95:717-22.

6
Findley LJ, Unverzagt ME and Suratt PM. Automobile accidents involving patients with
obstructive sleep apnea. Am Rev Respir Dis 1998;138:337-40.

7
Gering D, Nabavi A, Kikinis R, W.Eric L.Grimson, Hata N, Everett P, Jolesz F, Black PM,
Wells WM III. An Integrated Visualization System for Surgical Planning and Guidance
using Image Fusion and Interventional Imaging. In Proceedings of Medical Image
Computing and Computer-Assisted Intervention 1999:809-819. Lauren J. O’donnell, W. Eric
L. Grimson, Ferenc A. Jolesz, Peter M. Black, William M. Wells

8
Guilleminault C, Heldt G, Powell N, Riley R. Small upper airway in near-miss SIDS and
their family. Lancet 1986;8476:402-407.

9
Guilleminault C, Partinen M, Hollman K, Powell N, Stoohs R. Familial aggregates in
obstructive sleep apnea syndrome. Chest 107:1545-1551, 1995.

10
Haponik EF, Smith PL, Bohlman ME, Allen RP, Goldman SM, Bleecker ER.
Computerized tomography in obstructive sleep apnea: correlation of airway size with
physiology during sleep and wakefulness. Am Rev Respir Dis 1983;127:221-226.

11
Hoffstein V. Review of oral appliances for treatment of sleep-disordered breathing. Sleep
Breath 2007;11:1-22.

12
Hora F, Napolis LM, Daltro C, Kodaira SK, Tufik S, Togeiro AM, Nery LE. Clinical,
Anthropometric and Upper Airway Anatomic Characteristics of Obese Patients with
Obstructive Sleep Apnea Syndrome. Respiration 2007;74:517-524.

61
13
Hunt SC , Williams RR, Barlow GK. A comparison of positive family history definitions
for defining risk of future disease. Journal of Chronic Diseases 39(10):809-822, 1986.

14
Johns FR, Strollo PJ Jr, Buckley M, Constantino J. The influence of craniofacial structure
on obstructive sleep apnea in young adults. J Oral Maxillofac Surg 1998;56:596-602.

15
Kato J, Isono s, Tanaka A, Watanabe T, Araki D, Tanzawa H, Nishino T. Dose-dependent
effects of mandibular advancement on pharyngeal mechanics and nocturnal oxygenation in
patients with sleep-disordered breathing. Chest 2000;117:1085-1072.

16
Lavie P. Sleep Apnea Causes Cardiovascular Disease. Pro/Con Editorial. Am J Respir Crit
Care Med 2003;168:1-2.

17
Levendowski D, Olmstead R, Popovic D, Carper DL, Berka C, Westbrook PR.
Assessment of Obstructive Sleep Apnea Risk and Severity in Truck Drivers: Validation of
the Screening Questionnaire. Sleep Diagnosis and Therapy 2007;2(2): 20-30.

18
Levy P, Pepin JL, Deschaux BC, Paramelle B, Brambilla C. Accuracy of oximetry for
detection of respiratory disturbances in sleep apnea syndrome. Chest 1996;109:395-399.

19
Li HY, Chen NH, Wang CR, Shu YH, Wang PC. Use of 3-dimensional computed
tomography scan to evaluate upper airway patency for patients undergoing sleep-disordered
breathing surgery. Otolaryngol Head Neck Surg. 2003;129(4):336-42.

20
Liu J, Udupa JK, Odhnera D, McDonough JM, Arens R. System for Upper
Airway Segmentation and Measurement with MR Imaging and Fuzzy Connectedness. Acad
Radiol 10:13–24, 2003

21
Ludlow JB, Davies-Ludlow LE, Brooks SL, Howerton WB. Dosimetry of 3 CBCT
devices for oral and maxillofacial radiology: CB Mercuray, NewTom 3G and i-CAT.
Dentomaxillofac Radiol. 2006;35:219-26.

22
Mah J, Danforth RA, Bumann A, Hatcher D A. Radiation absorbed in maxillofacial
imaging with a new dental CT. Oral Surg Oral Med Oral Pathol Oral Radiol Endod.
2003;96(4):508-513.

23
Mathur R, Douglas NJ. Family studies in patients with the sleep apnea-hypopnea
syndrome. Ann. Intern. Med. 1995;122:174-178.

24
Mayer P, Pepin JL, Bettega G, Veale D, Ferretti G, Deschaux C, Levy
P. Relationship between body mass index, age and upper airway measurements in snorers
and sleep apnoea patients. Eur. Respir. J., Sep 1996; 9: 1801 - 1809.

25
McNamara F, Sullivan CE. Obstructive sleep apnea in infants: relation to family history of
sudden infant death syndrome, apparent life-threatening events, and obstructive sleep apnea.
J Pediatr 2000;136:318–323.

62
26
Miles PG, Vig PS, Weyant RJ, Forrest TD, Rockette HE Jr. Craniofacial structure and
obstructive sleep apnea syndrome: a qualitative analysis and meta-analysis of the literature.
Am J Orthod Dentofacial Orthop 1996;109:163-172.

27
Netzer NC, Hoegel JJ, Loube DD, Netzer CM, Hay B, Alvarez-Sala R and
Strohl KP. Prevalence of Symptoms and Risk of Sleep Apnea in Primary
Care. Chest, October 1, 124(4): 1406 – 1414, 2003.

28
Netzer NR, Stoohs RA, Netzer CM, Clark K, Strohl KP. Using the Berlin Questionnaire
to identify patients at risk for the Sleep Apnea Syndrome. Ann Intern Med, 1999. 131: p.
485–491.

29
Ogawa T, Enciso R, Shintaku WH, Clark GT. Evaluation of cross-section airway
configuration of obstructive sleep apnea. Oral Surg Oral Med Oral Pathol Oral Radiol
Endod. 2007 Jan;103(1):102-8.

30
Okubo M, Suzuki M, Horiuchi A, Okabe S, Ikeda K, Higano S, Mitani H, Hida W,
Kobayashi T, Sugawara J. Morphologic analyses of mandible and upper airway soft tissue
by MRI of patients with obstructive sleep apnea hypopnea syndrome. Sleep 2006;29(7):909-
15.

31
Pillar G, Lavie P. Assessment of the role of inheritance in sleep apnea syndrome. Am J
Respir Crit Care Med 1995;151:688-691.

32
Posnick J, Bite U, Nakano P, Davis J, and Armstrong D. Indirect Intracranial Volume
Measurement. Plastic & Reconstructive Surgery 1992;89(1):34-45.

33
Redline S, Tishler PV. Familial influences on sleep apnea. In: Saunders NA, Sullivan CE,
eds. Sleep and breathing., 2nd ed. New York: Marcel Dekker, 363-377, 1994.

34
Redline S, Tishler PV, Tosteson TD, Williamson J, Kump K, Browner I, Ferrette V, Krejci
P. The familial aggregation of obstructive sleep apnea. Am J Respir Crit Care Med
1995;151:682-687.

35
Roth T, Roehrs TA, Conway WA. Behavioral morbidity of apnea. Sem Respir Med
1998;9:554-9.

36
Schwab RJ, Gefter WB, Hoffman EA, Gupta KB, Pack AI. Dynamic upper airway
imaging during awake respiration in normal subjects and patients with sleep disordered
breathing. Am Rev Respir Dis. 1993 Nov;148(5):1385-400.

37
Schwab RJ, Gupta KB., Gefter WB, Metzger LJ, Hoffman EA, Pack AI. Upper airway and
soft tissue anatomy in normal subjects and patients with sleep-disordered breathing.
Significance of the lateral pharyngeal walls. Am J Respir Crit Care Med. 1995;152(5):1673-
89.

38
Schwab RJ. Upper airway imaging. Clin Chest Med 1998;19:33-54.
63

39
Schwab RJ, Pasirstein M, Pierson R, Mackley A, Hachadoorian R, Arens R, Maislin G,
Pack AI. Identification of Upper Airway Anatomic Risk Factors for Obstructive Sleep
Apnea with Volumetric Magnetic Resonance Imaging. Am J Respir Crit Care Med
2003;168:522–530.

40
Sforza E, Bacon W, Weiss T, Thibault A, Petiau C, Krieger J. Upper airway collapsibility
and cephalometric variables in patients with obstructive sleep apnea. Am J Respir Crit Care
Med. 2000;161:347-352.

41
Strohl KP, Saunders NA, Feldman NT, Hallett M. Obstructive sleep apnea in family
members. N Engl J Med 1978;299:969-973.

42
Trudo FJ, Gefter WB, Welch KC, Gupta KB, Maislin G, Schwab RJ. State-related changes
in upper airway caliber ans surrounding soft-tissue structures in normal subjects. Am J
respire Care Med 1998;158:1259-1270.

43
Victor LD. Obstructive Sleep Apnea. Am Fam Physician 1999;60:2279-86.

44
Weaver TE, Kribbs NB, Pack AI, Kline LR, Chugh DK, Maislin G, Smith PL, Schwartz
AR, Schubert NM, Gillen KA, Dinges DF. Night-to-night variability in CPAP use over the
first three months of treatment. Sleep 1997; 20:278–283

45
Welch KC, Foster GD, Ritter CT, Wadden TA, Arens R, Maislin G, Schwab RJ. A novel
volumetric magnetic resonance imaging paradigm to study upper airway anatomy. Sleep
2002;25(5):532-42.

46
Westbrook PR, Levendowski DJ, Cvetinovic M, Zavora T, Velimirovic V, Henninger D,
Nicholson D.Description and validation of the Apnea Risk Evaluation System: A novel
method to diagnose sleep apnea-hypopnea in the home. Chest 2005; 128:2166-2175.

47
Williams RR, Hunt SC, Heiss G, Province MA, Bensen JT, Higgins M, Chamberlain RM,
Ware J, Hopkins PN. Usefulness of Cardiovascular Family History Data for Population-
Based Preventive Medicine and Medical Research (The Health Family Tree Study and the
NHLBI Family Heart Study). The American Journal of Cardiology 87(2):129-135, 2001.

48
Yamashiro Y, Kryger M. Why should sleep apnea be diagnosed and treated? Clin Pulm
Med 1994; 1: 250-9.

49
Yang Q. Khoury MJ. Rodriguez C. Calle EE. Tatham LM. Flanders WD.Family history
score as a predictor of breast cancer mortality: prospective data from the Cancer Prevention
Study II, United States, 1982-1991. American Journal of Epidemiology. 147(7):652-9, 1998
Apri1.

64
50
Young T, Finn L, Peppard PE, Szklo-Coxe M, Austin D, Nieto FJ, Stubbs R, Hla KM.
Sleep disordered breathing and mortality: eighteen-year follow-up of the Wisconsin sleep
cohort. Sleep 2008 Aug 1;31(8):1071-8.

51
Young T, Palta M, Dempsey J, Skatrud J, Weber S and Badr S. The occurrence of sleep-
disordered breathing among middle-aged adults. N Engl J Med 1993;328:1230-5.

52
www.nhlbi.nih.gov/health/dci/Diseases/SleepApnea/SleepApnea_WhoIsAtRisk.html

65

APPENDIX
BERLIN QUESTIONNAIRE:

CASE #: _______

AGE:
_______ YEARS _______ MONTHS

SEX (check one)
□ FEMALE
□ MALE
HEIGHT :
_________ FT ________INCHES

WEIGHT: __________ POUNDS

ETHNICITY (check one)

□ HISPANIC
□ NON HISPANIC
RACE (check one)
□ American Indian/Alaska Native;
□ Asian;
□ Native Hawaiian / Other Pacific
Islander;
□ Black or African American;
□ White.

Please circle one response per row:

Question Response
1. Do you snore? Yes No Do not
know

66

2. Snoring loudness? Loud as
breathing
Loud as
talking
Louder
than
talking
Very
loud
Do not
know
3. Snoring frequency? Almost
every day
3 to 4
times per
week
1 to 2
times per
week
1 to 2
times per
month
Never or
almost
never
4. Does your snoring
bother other people?

Yes

No

5. If present, how often
have your breathing
pauses during sleep
been noticed?
Almost
every day
3 to 4
times per
week
1 to 2
times per
week
1 to 2
times per
month
Never or
almost
never
6. Are you tired after
sleeping?
Almost
every day
3 to 4
times per
week
1 to 2
times per
week
1 to 2
times per
month
Never or
almost
never
7. Are you tired during
waketime?
Almost
every day
3 to 4
times per
week
1 to 2
times per
week
1 to 2
times per
month
Never or
almost
never
8. Have you ever fallen
asleep while driving?
Yes No
9. Asleep driving
frequency?
Almost
every day
3-4 times
per week
1 to 2
times
per week
1 to 2
times per
month
Never or
almost
never
67

10. Do you have high
blood pressure (>
140/90 mm Hg)?

Yes No Do not
know

11. Has your weight
changed
in the last 5 years?

Increased

Decreased

No change

Scoring: A subject will have a “high risk” score if two or three of the following
categories are scored at risk:
Category 1 (Questions 1-5): In category 1, a

positive score for risk is defined
as frequent symptoms (ie,

"more than three to four times per week" or
"almost every day")

in the questions about snoring (q3) and witnessed apneas
(q5).
Category 2 (Questions 6-9): In category

2, a positive score for risk was
frequent symptoms in two or

more questions about awakening sleepy (q6),
waketime sleepiness (q7),

and/or drowsy driving (q8-9).
Category 3 (Questions 10-11): In category 3, a positive score for risk was
defined as a self-report of high blood pressure and/or of height/weight
information giving a BMI of > 30 kg/m2.

Abstract (if available)

Abstract Introduction: Cephalometric and other imaging modalities have been used to compare Obstructive Sleep Apnea (OSA) patients with controls, but volumetric measurements of the upper airway are scarce. Purpose: To compare the Cone-Beam Computerized Tomography (CBCT) scan measurements between patients with OSA and controls to develop a prediction model for OSA based on imaging and family history. Methods: 39 OSA patients (Apnea-Hypoapnea Index≥15) and 23 controls based on ambulatory somnographic assessment were recruited through flyers and mail at USC School of Dentistry. Each patient answered the Berlin Questionnaire, collected family member’s Berlin Questionnaires and was imaged with CBCT. Linear and volumetric measurements of the upper airway were performed and multivariate logistic regression analysis was used to identify risk factors for OSA. Results: OSA patients were predominantly male, older, had a larger neck size and larger Body Mass index than controls. The minimum cross-sectional area and the lateral dimension were significantly smaller in cases. Conclusions: Age>57 years, male gender, a positive Berlin (or ARES Questionnaire) and narrow upper airway lateral dimension were identified as significant risk factors for OSA. This prediction model for OSA could be implemented based on this work, however further studies are needed to generalize our findings. Limitations of the study include self-selection bias for the controls: subjects with a positive family history of sleep apnea are more willing to participate in a study to predict OSA based on family history, so the odds-ratios of having OSA based on family history were underestimated.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The relationship of right and left sleeping position with apnea hypopnea index (AHI) in patients with obstructive sleep apnea (OSA)

PDF

A cone beam CT evaluation of the maxillary dento skeletal complex after rapid palatal expansion

PDF

Seeing sleep: real-time MRI methods for the evaluation of sleep apnea

PDF

A cone beam-ct evaluation of the availability of bone for the placement of miniscrews

PDF

A cone beam-CT evaluation of the proximity of the maxillary sinus to commonly used TAD sites

PDF

Model-based phenotyping of obstructive sleep apnea in overweight adolescents for personalized theranostics

PDF

Functional real-time MRI of the upper airway

PDF

A cone beam-CT evaluation of the availability of space for complete arch retraction in the mandible with TADs

PDF

Predicting CPAP adherence by a sleep study: a wavelet-based deep learning approach

PDF

The effect of cone beam computed tomography (CBCT) imaging on orthodontic diagnosis and treatment planning

PDF

Cone-beam computed tomography images: applications in endodontics

PDF

The effect of Omalizumab on free IgE and spirometric measurements in allergic asthmatics with highly elevated IgE levels

PDF

Validation of a modified pediatric early warning system (PEWS) score

PDF

An assessment of impact of early local progression on subsequent risk for the treatment failure in adolescent and young adult patients with non-metastatic osteosarcoma

PDF

Evaluation of the current state of management of children with central nervous system tumors in Paraguay through a comprehensive needs assessment

PDF

A penta-dimensional longitudinal analysis of the predictors of compulsive internet use among adolescents using linear mixed model (LMM)

PDF

Proper mesio-distal angulation and bucco-lingual inclination of the whole tooth in three-dimensional space -- a standard for orthodontic patients

PDF

The Filipino family initiative: preliminary effects of an evidence-based parenting intervention offered in churches on parent and child outcomes

PDF

Investigating a physiological pathway for the effect of guided imagery on insulin resistance

Asset Metadata

Creator Cachafeiro, Maria Reyes Enciso (author)

Core Title A cone beam-CT evaluation of the upper airway for sleep apnea prediction

School Keck School of Medicine

Degree Master of Science

Degree Program Clinical and Biomedical Investigations

Publication Date 02/12/2009

Defense Date 12/04/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cone-beam CT,family history,imaging,OAI-PMH Harvest,prediction,sleep apnea

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Azen, Stanley Paul (committee chair), Clark, Glenn T. (committee member), Xiang, Anny Hui (committee member)

Creator Email encisor@gmail.com,renciso@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1974

Unique identifier UC1201106

Identifier etd-Cachafeiro-2574 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-145382 (legacy record id),usctheses-m1974 (legacy record id)

Legacy Identifier etd-Cachafeiro-2574.pdf

Dmrecord 145382

Document Type Thesis

Rights Cachafeiro, Maria Reyes Enciso

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu