novel radiographic measurement algorithm demonstrating a ... · pdf filenovel radiographic...
TRANSCRIPT
Original Research—Sinonasal Disorders
Novel Radiographic MeasurementAlgorithm Demonstrating a Link betweenObesity and Lateral Skull BaseAttenuation
Otolaryngology–Head and Neck Surgery2015, Vol. 152(1) 172–179� American Academy ofOtolaryngology—Head and NeckSurgery Foundation 2014Reprints and permission:sagepub.com/journalsPermissions.navDOI: 10.1177/0194599814557470http://otojournal.org
Shawn M. Stevens, MD1, Paul R. Lambert, MD1, Habib Rizk, MD1,Wesley R. McIlwain, MD2, Shaun A. Nguyen, MD1, andTed A. Meyer, MD, PhD1
No sponsorships or competing interests have been disclosed for this article.
Abstract
Objectives. (1) To describe a validated algorithm for measur-ing tegmen thickness on computed tomography scans. (2)To compare the tegmen thickness in 3 groups: patients withspontaneous cerebrospinal fluid (CSF) leaks, obese controls,and nonobese controls.
Study Design. Retrospective review.
Setting. Patients with spontaneous CSF otorrhea often havehighly attenuated tegmen plates. This is associated with obe-sity and/or idiopathic intracranial hypertension (IIH). No evi-dence exists, however, that objectively links obesity and/orIIH with skull base attenuation.
Subjects and Methods. This was a retrospective review from 2004to the present. Patients with spontaneous CSF otorrhea andmatched obese (body mass index [BMI] .30 kg/m2) and nonobese(BMI \30 kg/m2) controls were selected. Tegmen thickness wasmeasured radiographically. Interrater validity was assessed.
Results. Ninety-eight patients were measured: 37 in the CSFgroup (BMI, 36.6 kg/m2), 30 in the obese group (BMI, 34.6kg/m2), and 31 in the nonobese group (BMI, 24.2 kg/m2).The CSF group had a significantly thinner tegmen comparedto both the obese control (P \ .01) and nonobese control(P = .0004) groups. Obese controls had a thinner tegmenthan nonobese controls (P \ .00001). A significant inversecorrelation was detected between skull base thickness andBMI. Signs/symptoms of IIH were most commonly found inthe CSF group. Good to very good strength of agreementwas detected for measures between raters.
Conclusion. This is the first study to (1) quantify lateral skullbase thickness and (2) significantly correlate obesity with lat-eral skull base attenuation. Patients who are obese withspontaneous CSF leaks have greater attenuation of theirskull base than matched obese controls. This finding sup-ports theories that an additional process, possibly congeni-tal, has a pathoetiological role in skull base dehiscence.
Keywords
spontaneous, CSF leak, obesity, idiopathic intracranial hyper-tension, tegmen, dehiscence
Received July 20, 2014; revised September 12, 2014; accepted October
8, 2014.
Spontaneous cerebrospinal fluid (CSF) otorrhea is a rela-
tively rare phenomenon in which CSF enters the con-
fines of the temporal bone in a manner unrelated to
other causes. Spontaneous CSF otorrhea occurs in 2 popula-
tions: young children and middle-aged adults. Children typi-
cally present after episodes of recurrent meningitis, and the
leak is often associated with congenital anomalies.1-6 In adults,
a diagnosis is often made after myringotomy for a persistent
effusion in obese, middle-aged females.7,8 Two major theories
predominate regarding the pathophysiology in adults. The first
posits that skull base attenuation occurs chronically secondary
to increased intracranial pressure, which in turn is closely
linked to obesity and idiopathic intracranial hypertension
(IIH).9-18 The second theory, suggested by Gacek et al19 and
corroborated by others, suggests that aberrant arachnoid granu-
lations may instigate skull base erosion.18-21 A congenital pre-
disposition may exist in either process.22-24
The radiographic appearance of the skull base in obese
individuals is often one of broad attenuation.18 Quantifiable
evidence correlating obesity, CSF leaks, and skull base
1Department of Otolaryngology–Head and Neck Surgery, Medical
University of South Carolina, Charleston, South Carolina, USA2Department of Otolaryngology–Head and Neck Surgery, Madigan Army
Medical Center, Tacoma, Washington, USA
This article was presented at the 2014 AAO-HNSF Annual Meeting and
OTO EXPO; September 21-24, 2014; Orlando, Florida.
Corresponding Author:
Shawn M. Stevens, MD, Department of Otolaryngology–Head and Neck
Surgery, Medical University of South Carolina, 135 Ashley Avenue, MSC
550, Charleston, SC 29425, USA.
Email: [email protected]
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
attenuation is lacking, however. In the present study, we
sought to overcome this by developing a novel radiographic
measurement algorithm for the lateral skull base. Using
temporal bone computed tomography (CT) scans from
patients with spontaneous leaks, obese controls, and nonob-
ese controls, we sought to address 3 hypotheses: (1) current
CT imaging/formatting technology can now achieve precise
and reproducible measurements of lateral skull base thick-
ness, (2) obese patients will have a quantifiably thinner
skull base than nonobese patients, and (3) patients with
spontaneous CSF otorrhea will have more skull base
attenuation than obese patients without leaks. The aims of
the study were to validate the algorithm for interrater relia-
bility and to correlate radiographic findings with demo-
graphic, clinical, and body mass index (BMI) data.
MethodsStudy Approval
Approval was obtained from the Medical University of South
Carolina (MUSC) Institutional Review Board (Pro00023289).
Group Selection
This retrospective chart review from 2004 to 2013 utilized
ICD-9 codes and operative notes. Inclusion criteria for the
primary study group were the following: diagnosis of spon-
taneous CSF otorrhea, surgical intervention performed, age
.18 years, and dedicated high-resolution temporal bone CT
scans available. Demographic data recorded were age, sex,
and race. Moreover, BMI (kg/m2) was calculated based on
height and weight at the time of surgery. Clinical variables
documented included hypertension, diabetes mellitus, osteo-
penia/osteoporosis, IIH, and smoking status. Exclusion cri-
teria were the following: history of temporal bone trauma,
cholesteatoma, neoplasm, prior lateral skull base and/or
mastoid surgery, congenital ear anomaly, and age \18
years.
Control subjects were culled from adult cochlear implant
patients. The subjects were separated according to BMI into
obese (BMI .30 kg/m2) and nonobese (BMI \30 kg/m2)
groups. Inclusion criteria included age .18 years and dedi-
cated high-definition CT scans available. Demographic data,
clinical data, and exclusion criteria were identical to those
of the primary study group.
Imaging and Hardware
Dedicated temporal bone CT scans were obtained with a
standard collimation of 0.625 mm. Images were reconstructed
in axial, coronal, and sagittal planes. The majority of scans
were obtained at MUSC on either a Siemens Somatom
Sensation 16 or Siemens Definition 128 (Siemens Medical
Solutions, Malvern, Pennsylvania). Spatial resolution was
rated as accurate to 0.1 to 0.2 mm using this protocol.
All images were analyzed on standardized personal comput-
ing stations equipped with Intel Core 2 Duo CPUs (Intel
Corporation, Santa Clara, California). Processing speed was 2.3
GHz with 4 GB of available random-access memory (RAM).
Stations were all equipped with Lenovo Think Vision 19-inch,
square, backlit light-emitting diode (LED) and flat panel liquid-
crystal display (LCD) monitors (Lenovo, Raleign, North
Carolina) set to a maximum resolution of 1280 3 1024 dpi.
Software
Reformatted images were analyzed via the proprietary digi-
tal radiology imaging system AGFA Impax 6 (AGFA
Impax, Mortsel, Belgium). Temporal bone thickness was
measured via the partial volume averaging formula (measure-
ment caliper) in this software. Accuracy was rated to 0.1 mm.
CT Measurement Algorithm: Tegmen Tympani
Skull base measurements were made at predefined points.
Measurements of the tegmen tympani (TT) were conducted
within a coronal cut best depicting the malleus, the incus,
and at least 2 turns of the cochlea (Figure 1). Measurements
were made at the thickest and thinnest visible points of the
tegmen plate as well as at a fixed point directly cephalad to
the ossicles. All measures were made on both sides of the
skull base. Due to software limitations of the Impax measure-
ment calipers and the quoted CT spatial resolution, we set an
absolute minimum limit for thickness of 0.4 mm. This limit
applied to dehiscent areas with no measurable bone density.
CT Measurement Algorithm: Tegmen Mastoideum
Measurements were made of the tegmen mastoideum (TM)
in a coronal cut best demonstrating the most lateral curvature
of the posterior semicircular canal (Figure 1). Measurements
included the thickest and thinnest visible points along with a
fixed point at the center of the tegmen plate between the cor-
tical bone and posterior semicircular canal medially.
Measurements were repeated on both sides of the skull base.
The minimum thickness limit was applied.
Additional Calculations for Statistics
Using the above protocol, a total of 12 measurements were
made for each patient (6 of the TT/TM on each side). To facil-
itate statistical comparisons, mean aggregate thickness values
were calculated by averaging the data for the TT and TM on
each side of the skull base. The overall mean skull base thick-
ness was also calculated by averaging the mean aggregate TT
and TM thicknesses of both sides. This value was statistically
correlated with BMI and other demographic/clinical factors.
Validation Measures
The intraclass correlation coefficient (K) was calculated for
the purpose of assessing interrater reliability among 3 raters
at staggered levels of training in otolaryngology–head and
neck surgery. This included a medical student, a fourth-year
resident, and a second-year neurotology fellow. For the
assessment, a total of 39 scans were randomly selected from
the overall study cohort including 13 from each of the 3
study groups. The clinical diagnosis, group allocation,
demographic data, operative notes, and clinical information
were withheld from the raters. Over a period of 2 weeks,
each rater independently performed a full set of measure-
ments as described above on each of the 39 scans (468
Stevens et al 173
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
measurements per rater). The raters were blinded to each
other’s measurements. Results were expressed as the intra-
class correlation coefficient (K) and 95% confidence interval.
Strength of agreement relative to K can be interpreted as the
following: \0.20 = poor; 0.21-0.40 = fair; 0.41-0.60 = mod-
erate; 0.61-0.80 = good; and 0.81-1.00 = very good.25
Statistics
All analyses and graphs were performed with Sigma Plot 12.5
(Systat Software Inc, San Jose, California) and MedCalc
13.3.0.0 (MedCalc Software bvba, Ostend, Belgium). Disease
information and demographic variables were summarized by
means of summary statistics. Continuous variables were sum-
marized by the mean 6 standard deviation. Nominal variables
were summarized by the frequency and percentage. All contin-
uous variables were tested for normal distribution as deter-
mined by the Kolmogorov-Smirnov test. Comparisons of
outcomes (nominal variables) were performed using the Fisher
exact test or x2 test with Yates correction. For continuous vari-
ables, comparisons were made using an independent t test. A
correlation model was used to determine the relationship
among variables such as skull base thickness, age, and sex. A
value of P \ .05 was considered indicative of statistical
significance.
Results
A total of 42 patients were identified with a diagnosis of
spontaneous CSF otorrhea. Of these, 5 patients were
excluded for inadequate temporal bone CT scans, leaving
37 patients for the CSF group. Only 31 total patients were
identified from the control cohort who were obese. As the
number of nonobese controls was relatively large, 31 con-
secutive patients were selected to numerically match the
obese controls. Of these, only 1 obese control failed to meet
inclusion criteria, leaving 30 patients for the obese control
group and 31 patients for the nonobese control group.
The demographic and clinical information of each group
is summarized in Table 1. The mean ages of the CSF group,
obese control group, and nonobese control group were 60.9
years, 57.0 years, and 62.2 years, respectively. The mean
Table 1. Group Demographics and Clinical Data.
Spontaneous
CSF Leak
Obese Control
Group (BMI .30 kg/m2)
Nonobese
Control Group (BMI \30 kg/m2)
Statistical
Significance
Total (N = 98), n 37 30 31 N/A
Mean age, y 60.9 57.0 62.2 —
Mean BMI, kg/m2 36.6 34.6 23.7 —
Sex, n
Male 7 13 17 —
Female 30a 17 14 P = .028a
Race, n
White 21 22 24 —
Black 16 8 7 —
Hypertension, n 29a 18 12 P = .023a
Diabetes mellitus, n 12 7 7 —
Osteopenia/osteoporosis, n 4 1 1 —
Current/former smoker, n 11 8 9 —
Idiopathic intracranial hypertension, n 5 0 0 —
Abbreviations: BMI, body mass index; CSF, cerebrospinal fluid; N/A, not applicable.aAll 3 groups were similar for all measures except for the CSF group in which female sex and hypertension were significantly more common.
Figure 1. Measurements: (A) Left tegmen tympani, demonstrating the fixed (0.07 cm), thickest (0.15 cm), and thinnest/dehiscent (arrow-head) points. (B) Right tegmen mastoideum, demonstrating the thinnest (0.04 cm), fixed (0.05 cm), and thickest (0.19 cm) points.
174 Otolaryngology–Head and Neck Surgery 152(1)
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
BMIs of the groups were 36.6 kg/m2, 34.6 kg/m2, and 23.7
kg/m2, respectively. The CSF group and obese control group
did not significantly differ from each other in terms of BMI.
The CSF group had a significantly higher proportion of both
women and patients with hypertension. All diagnoses of IIH
were in the CSF group, but the relatively small number was
not enough to reach statistical significance. The groups did
not differ in any other clinical factor.
When reviewing temporal bone scans, the lateral skull base
of patients in both the CSF group and obese control group had
a similar, broadly attenuated appearance with a moth-eaten
quality to the bone (Figure 2). There was generally a paucity
of pneumatized bone cephalad to the otic capsule structures
and ossicles. Findings were usually bilateral and involved both
the TT and TM. Scans in nonobese controls typically demon-
strated thick and well-defined tegmen plates with a notable
amount of pneumatized bone buffering the inner ear.
Objectively, skull base thickness differed significantly
between each of the groups bilaterally (Figure 3). The non-
obese controls had the thickest skull bases, with a mean of
1.204 6 0.047 mm and 1.337 6 0.048 mm on the left and
right sides, respectively. This was significantly thicker than
skull base measurements in both the obese control group
(left: 0.981 6 0.037 mm; right: 1.008 6 0.038 mm; P \.00001) and CSF group (left: 0.862 6 0.040 mm; right:
0.811 6 0.033 mm; P \ .0004).
A comparison of skull base thickness controlling for obe-
sity was also made between the obese control group and CSF
group (Figure 4). This was accomplished by removing all
nonobese patients with CSF leaks (6 excluded; n = 31 obese
patients with CSF leaks). Between the 2 groups, patients with
CSF leaks had significantly thinner skull bases (left: 0.833 6
0.038 mm; right: 0.814 6 0.036 mm) than controls (left:
0.981 6 0.037 mm; right: 1.008 6 0.038 mm; P \ .01). The
mean BMI of this ‘‘obese only’’ CSF group was 37.7 kg/m2.
There remained no significant difference in BMI, however,
between the obese CSF and obese control groups.
When correlation modeling was performed, a significant
inverse correlation was detected between skull base thickness
and BMI (r = –0.466, P = .000001) (Figure 5). No significant
correlation was demonstrated between skull base thickness and
age (r = 0.065, P = .644) or any of the other demographic/clin-
ical parameters from Table 1 (data not shown).
Figure 2. Subjective attenuation: Coronal computed tomography scans depicting tegmen tympani from the 3 groups. The nonobese con-trol (NOC) group was normal, the obese control (OC) group was attenuated, and the cerebrospinal fluid (CSF) leak group was dehiscent.
Mean Skull Base Thickness (mm)
Skul
l Bas
e Th
ickn
ess
(mm
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
LeftRight
Obese Control Obese CSF
‡
Figure 4. Obese controls versus obese patients with cerebrosp-inal fluid leaks: mean aggregate thickness comparison. Thicknessdiffered significantly (P \.01) on both sides of the skull base (z).Black, left; gray, right. Error bars are the standard error of themean.
Mean Skull Base Thickness (mm)
Non-ObeseControl
Skul
l Bas
e Th
ickn
ess
(mm
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
LeftRight
CSF Group(All)
Obese Control
*ⱡ †
Figure 3. Mean aggregate thickness comparisons: significant differ-ences (P \.001) were detected between nonobese/obese controls(*), nonobese controls/patients with cerebrospinal fluid (CSF) leaks(z), and obese controls/patients with CSF leaks (y). Black, left;gray, right. Error bars are the standard error of the mean.
Stevens et al 175
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
When the intraclass correlation coefficient (K) was used
to assess interrater reliability, the results were ‘‘good’’ to
‘‘very good’’ among all 3 raters when measuring both obese
and nonobese groups. The results were ‘‘poor’’ to ‘‘good’’
among all 3 raters when measuring individual segments of
the tegmen for the CSF group. Results were ‘‘good’’ to
‘‘very good’’ when comparing mean aggregate measures for
this group (Table 2).
Discussion
In developing the measurement algorithm described above,
consideration was given to the limitations of the Impax soft-
ware package and scanner technology. As the resolution of
CT formatting and the Impax caliper is rated to 0.1 mm,
precision was felt to be adequate for detecting submillimeter
differences between groups. To this end, the absolute mini-
mum measurement convention of 0.4 mm requires addi-
tional discussion. The application of this convention was
considered necessary to overcome software limitations pre-
venting reliable measures \0.4 mm (the calipers cannot be
drawn this small). As nearly all subjects in the CSF group
and some obese controls had skull base dehiscences, use of
this convention rather than 0.0 mm likely masks a more pro-
nounced difference in thickness than that reported. In turn,
the minimum thickness value also likely prevents unneces-
sary skewing of the data toward an attenuated state by pre-
venting 0.0 mm being entered for any measurement
between 0.1 to 0.3 mm. Detection of a significant difference
between groups using our more conservative approach may
lend more support to the hypotheses than the use of a 0.0-
mm minimum value.
Table 2. Interrater Validation Measures.
Measure Intraclass Correlation Coefficient (95% Confidence Interval) Strength of Agreement
Obese control group: left side
Mean aggregate 0.8185 (0.5552 to 0.9368) Very good
Tegmen mastoideum 0.7154 (0.3020 to 0.9009) Good
Tegmen tympani 0.8505 (0.6334 to 0.9479) Very good
Obese control group: right side
Mean aggregate 0.7417 (0.3669 to 0.9101) Good
Tegmen mastoideum 0.7922 (0.4907 to 0.9277) Good
Tegmen tympani 0.6586 (0.1630 to 0.8811) Good
Nonobese control group: left side
Mean aggregate 0.8076 (0.5111 to 0.9363) Very good
Tegmen mastoideum 0.8585 (0.6405 to 0.9531) Very good
Tegmen tympani 0.7750 (0.4283 to 0.9255) Good
Nonobese control group: right side
Mean aggregate 0.8995 (0.7445 to 0.9667) Very good
Tegmen mastoideum 0.7823 (0.4469 to 0.9279) Good
Tegmen tympani 0.8207 (0.5445 to 0.9406) Very good
CSF group: left side
Mean aggregate 0.6775 (0.1804 to 0.8932) Good
Tegmen mastoideum 0.1441 (–1.2654 to 0.7323) Poor
Tegmen tympani 0.5584 (–0.1689 to 0.8619) Moderate
CSF group: right side
Mean aggregate 0.8139 (0.5270 to 0.9383) Very good
Tegmen mastoideum 0.3316 (–0.7691 to 0.7909) Fair
Tegmen tympani 0.7306 (0.2869 to 0.9157) Good
Abbreviation: CSF, cerebrospinal fluid.
Skull Base Thickness and BMI
BMI15 20 25 30 35 40 45 50 55
Skul
l Bas
e Th
ickn
ess
(mm
)
0.1
1
10
r = –0.466p = 0.000001
Figure 5. Correlation between skull base thickness (mm) andbody mass index: all study subjects were included. A significantnegative correlation was detected. r, correlation coefficient.
176 Otolaryngology–Head and Neck Surgery 152(1)
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
The total number of measurement samples per subject
also was intentionally kept high (6 per side/12 per subject) to
more accurately reflect aggregate thickness. The initial build
of the algorithm incorporated a solitary measurement of the
thinnest point on the tegmen. It was quickly discovered, how-
ever, that most subjects with CSF leaks and obese controls
had at least 1 ‘‘thinnest point’’ at the minimum thickness
limit. Detecting a difference between groups was therefore
impossible. Additionally, as the leading pathoetiological
theory for the development of spontaneous CSF otorrhea
entails elevated intracranial pressure and broad skull base
attenuation, a larger sampling of measurements was felt to
more accurately reflect the underlying pathophysiology.
Using these methods, our interrater validation assessment
revealed good to very good strength of agreement for aggre-
gate thickness measures between observers at all levels of
training (Table 2). The lower strength of agreement noted
for CSF group measures was likely due in part to the soft-
ware’s minimum thickness limitations and to the highly
variable presentation of the thickest point in patients with
CSF leaks. When aggregate measures were considered
though, overall agreement was much better.
Although many studies have performed radiographic assess-
ments of skull base attenuation, the vast majority of investiga-
tors have done so in a solely qualitative manner.9,11,23,26-30
Within the ear, nose, and throat literature, little quantita-
tive, measurement-driven data exist to detect differences
between groups.23,27,31,32 To our knowledge, only 1 prior
study has made quantifiable radiographic measurements
of the lateral skull base.23 The best precedent for objec-
tive radiographic investigations ultimately came from the
oral-maxillofacial literature, which heavily utilizes cepha-
lometric evaluations.33-43
Many such studies have used ‘‘smallest detectable differ-
ence’’ (SDD) calculations to further validate CT-based mea-
surement algorithms.33,35-37 The SDD is a statistical
estimate of the smallest significant amount of change that
can be detected by a given measure regardless of how reli-
ably that measure can be reproduced. When we applied
SDD calculations to our data, it was found that the differ-
ence in skull base thickness between each study group was
indeed larger than the calculated SDD. This finding strongly
suggests that the significant differences detected herein are
indeed real differences. Appendix A (available at www.oto-
journal.org) provides further detail on SDD calculations.
Although the pathophysiology underlying adult sponta-
neous CSF otorrhea is not fully understood, this condition is
associated with obesity and IIH.9-18 Our data support these
associations and are consistent with commonly cited demo-
graphic patterns. The mean BMI of our CSF group was 36.6
kg/m2, and the mean age was 60.9 years. Eighty-one percent
(30/37) of the patients were female. Five of 37 (13.5%) had
a formal diagnosis of IIH. Because CSF opening pressure
was not routinely measured, the number of individuals with
IIH was likely higher than reported.
The data also demonstrate a significant correlation between
obesity and skull base attenuation. To our knowledge, this is
the first report of a statistically significant inverse relationship
between skull base thickness and BMI. It is also the first
description of a significant difference in skull base attenuation
between obese and nonobese subjects. As BMI is an indirect
measure of obesity and not related to intracranial processes,
the reasons for the above findings cannot be entirely explained
by these data alone.
Some inferences are possible, however, when consider-
ing the physiology associated with the obese habitus. The
latter causes elevated central venous pressure, leading in
turn to elevated intracranial pressure. Also, IIH is known
to be more prevalent in obese individuals.44,45 Over time,
this steep pressure gradient across the entirety of the dura-
bone interface likely causes gradual and broad bony skull
base erosion. Subjectively, the majority of our obese sub-
jects fit such a description. Other authors have noted simi-
lar findings, and the quantitative measures here support
their prior reports.13-15,18,23,26-30 The lack of an inverse
correlation between skull base thickness and age, however,
does not fit the temporal aspect of the theory. This may be
due to the fact that changes in skull base thickness over
time may be too small to detect with currently available
technology.
In addition to supporting a theory of pressure-induced
erosion caused by obesity, our data also seem to suggest
that subjects with CSF leaks were affected by an additional
pathophysiological process. This was evidenced by that
group having significantly thinner skull bases than obese
patients without leaks. One potential explanation for this is
bias introduced by our measurement algorithm. Given the
preponderance toward skull base dehiscence in subjects
with CSF leaks, the lack of measurable bone may have
skewed the data relative to controls. As discussed previ-
ously, the inclusion of a minimum thickness measure does
render this explanation somewhat less likely. An additional
cause may be hormonal in nature, as female sex was signifi-
cantly overrepresented in the CSF group. While it is diffi-
cult to assess this directly, other indicators of the influence
of sex such as osteoporosis and osteopenia (more common
in women) did not appear to differ significantly between
groups.
One further explanation may be one of congenital predis-
position. A growing conceptual trend in the literature holds
that certain individuals may have a genetic predisposition to
accelerated bone loss or perhaps an aberration in tegmen
ossification. In 2 cadaveric studies, between 6% to 34% of
all temporal bones had at least 1 tegmen dehiscence, while a
smaller subset (1%-6%) had multiple dehiscences and poor
bone quality.46,47 Another large cadaveric study, investigating
the prevalence of superior semicircular canal dehiscence
across age groups, found tegmen thickness to be uniformly
thin in neonates/infants. Thickness gradually increased until 3
years of age and then plateaued. Interestingly, approximately
2% of adult subjects in this study had tegmen thicknesses
that differed little from the neonatal period.22 This potential
congenital predisposition to skull base attenuation has been
corroborated in other studies and case reports.23,27,48-52 Newer
Stevens et al 177
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
investigations linking superior semicircular canal dehis-
cence and spontaneous CSF otorrhea (rate of concomi-
tance, 15%-55%) as well as the data reported here suggest
that at least a portion of patients with spontaneous leaks
may be susceptible to more accentuated skull base thinning
than would be explained by the pathophysiology of an
obese habitus alone.26,28
Conclusion
The data gathered in this study supported the hypothesis
that our novel measurement algorithm and current imaging
technology would allow for precise and reliable quantifica-
tion of lateral skull base thickness. The data also indicated
that a significant correlation between obesity and skull base
thickness exists and that patients suffering from spontaneous
CSF otorrhea have significantly thinner skull bases than do
matched obese controls. While reasoning for the latter find-
ing is not directly clarified by our data, indirect evidence
may suggest the existence of a congenital pathoetiological
process that augments tegmen thinning already known to
occur in this cohort.
Acknowledgments
The authors acknowledge the Medical University of South
Carolina’s Department of Radiology for providing invaluable input
that helped lead to our development of the novel skull base mea-
surement algorithm.
Author Contributions
Shawn M. Stevens, conception of design, acquisition of data, analy-
sis of data, interpretation of data, drafting article, and final approval;
Paul R. Lambert, conception of design, analysis of data, interpreta-
tion of data, critical revision, and final approval; Habib Rizk, acquisi-
tion of data, analysis of data, interpretation of data, critical revision,
and final approval; Wesley R. McIlwain, acquisition of data, interpre-
tation of data, drafting article, and final approval; Shaun A. Nguyen,
analysis of data, interpretation of data, drafting article, and final
approval; Ted A. Meyer, conception of design, analysis of data, inter-
pretation of data, critical revision, and final approval.
Disclosures
Competing interests: None.
Sponsorships: None.
Funding source: None.
Supplemental Material
Additional supporting information may be found at http://otojournal
.org/supplemental.
References
1. Foyt D, Brackmann DE. Cerebrospinal fluid otorrhea through
a congenitally patent fallopian canal. Arch Otolaryngol Head
Neck Surg. 2000;126:540-542.
2. Jegoux F, Malard O, Gayet-Delacroix M, Bordure P, Legent F,
Beauvillain de Montreuil C. Hyrtl’s fissure: a case of sponta-
neous cerebrospinal fluid otorrhea. AJNR Am J Neuroradiol.
2005;26:963-966.
3. Mong S, Goldberg AN, Lustig LR. Fallopian canal meningo-
cele: report of two cases. Otol Neurotol. 2009;30:525-528.
4. Rupa V, Job A, Rajshekhar V. Adult onset spontaneous CSF
otorrhea with oval window fistula and recurrent meningitis:
MRI findings. Otolaryngol Head Neck Surg. 2001;124:344-
346.
5. Teo DT, Tan TY, Eng SP, Chan YM. Spontaneous cerebrosp-
inal fluid otorrhoea via oval window: an obscure cause of
recurrent meningitis. J Laryngol Otol. 2004;118:717-720.
6. Teufert KB, Slattery WH. Cerebrospinal fluid leak of the fallo-
pian canal. Ear Nose Throat J. 2013;92:E20-E23.
7. Sanna M, Fois P, Russo A, Falcioni M. Management of
meningoencephalic herniation of the temporal bone: personal
experience and literature review. Laryngoscope. 2009;119:
1579-1585.
8. Semaan MT, Gilpin DA, Hsu DP, Wasman JK, Megerian CA.
Transmastoid extradural-intracranial approach for repair of
transtemporal meningoencephalocele: a review of 31 consecu-
tive cases. Laryngoscope. 2011;121:1765-1772.
9. Goddard JC, Meyer T, Nguyen S, Lambert PR. New consid-
erations in the cause of spontaneous cerebrospinal fluid otor-
rhea. Otol Neurotol. 2010;31:940-945.
10. LeVay AJ, Kveton JF. Relationship between obesity, obstruc-
tive sleep apnea, and spontaneous cerebrospinal fluid otorrhea.
Laryngoscope. 2008;118:275-278.
11. Prichard CN, Isaacson B, Oghalai JS, Coker NJ, Vrabec JT.
Adult spontaneous CSF otorrhea: correlation with radiographic
empty sella. Otolaryngol Head Neck Surg. 2006;134:767-771.
12. Rosenfeld E, Dotan G, Kimchi TJ, Kesler A. Spontaneous cer-
ebrospinal fluid otorrhea and rhinorrhea in idiopathic intracra-
nial hypertension patients. J Neuroophthalmol. 2013;33:113-
116.
13. Schlosser RJ, Bolger WE. Significance of empty sella in cere-
brospinal fluid leaks. Otolaryngol Head Neck Surg. 2003;128:
32-38.
14. Schlosser RJ, Wilensky EM, Grady MS, Bolger WE. Elevated
intracranial pressures in spontaneous cerebrospinal fluid leaks.
Am J Rhinol. 2003;17:191-195.
15. Schlosser RJ, Woodworth BA, Wilensky EM, Grady MS,
Bolger WE. Spontaneous cerebrospinal fluid leaks: a variant
of benign intracranial hypertension. Ann Otol Rhinol Laryngol.
2006;115:495-500.
16. Stucken EZ, Selesnick SH, Brown KD. The role of obesity in
spontaneous temporal bone encephaloceles and CSF leak. Otol
Neurotol. 2012;33:1412-1417.
17. Vivas EX, McCall A, Raz Y, Fernandez-Miranda JC, Gardner
P, Hirsch BE. ICP, BMI, surgical repair, and CSF diversion in
patients presenting with spontaneous CSF otorrhea. Otol
Neurotol. 2014;35:344-347.
18. Wang EW, Vandergrift WA 3rd, Schlosser RJ. Spontaneous
CSF leaks. Otolaryngol Clin North Am. 2011;44:845-856, vii.
19. Gacek RR, Gacek MR, Tart R. Adult spontaneous cerebrosp-
inal fluid otorrhea: diagnosis and management. Am J Otol.
1999;20:770-776.
20. Manes RP, Ryan MW, Marple BF. A novel finding on com-
puted tomography in the diagnosis and localization of
178 Otolaryngology–Head and Neck Surgery 152(1)
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from
cerebrospinal fluid leaks without a clear bony defect. Int
Forum Allergy Rhinol. 2012;2:402-404.
21. Yew M, Dubbs B, Tong O, et al. Arachnoid granulations of
the temporal bone: a histologic study of dural and osseous
penetration. Otol Neurotol. 2011;32:602-609.
22. Carey JP, Minor LB, Nager GT. Dehiscence or thinning of
bone overlying the superior semicircular canal in a temporal
bone survey. Arch Otolaryngol Head Neck Surg. 2000;126:
137-147.
23. Hirvonen TP, Weg N, Zinreich SJ, Minor LB. High-resolution
CT findings suggest a developmental abnormality underlying
superior canal dehiscence syndrome. Acta Otolaryngol. 2003;
123:477-481.
24. Minor LB, Solomon D, Zinreich JS, Zee DS. Sound- and/or
pressure-induced vertigo due to bone dehiscence of the super-
ior semicircular canal. Arch Otolaryngol Head Neck Surg.
1998;124:249-258.
25. Altman DG. Practical Statistics for Medical Research. 1st ed.
New York: Chapman and Hall; 1991.
26. Allen KP, Perez CL, Isaacson B, Roland PS, Duong TT, Kutz
JW. Superior semicircular canal dehiscence in patients with
spontaneous cerebrospinal fluid otorrhea. Otolaryngol Head
Neck Surg. 2012;147:1120-1124.
27. Crovetto M, Whyte J, Rodriguez OM, Lecumberri I, Martinez
C, Elexpuru J. Anatomo-radiological study of the superior
semicircular canal dehiscence radiological considerations of
superior and posterior semicircular canals. Eur J Radiol. 2010;
76:167-172.
28. El Hadi T, Sorrentino T, Calmels MN, Fraysse B, Deguine O,
Marx M. Spontaneous tegmen defect and semicircular canal
dehiscence: same etiopathogenic entity? Otol Neurotol. 2012;
33:591-595.
29. Connor SE. Imaging of skull-base cephalocoeles and cere-
brospinal fluid leaks. Clin Radiol. 2010;65:832-841.
30. Lee MH, Kim HJ, Lee IH, Kim ST, Jeon P, Kim KH.
Prevalence and appearance of the posterior wall defects of the
temporal bone caused by presumed arachnoid granulations and
their clinical significance: CT findings. Am J Neuroradiol.
2008;29:1704-1707.
31. Banu MA, Rathman A, Patel KS, et al. Corridor-based endona-
sal endoscopic surgery for pediatric skull base pathology with
detailed radioanatomic measurements. Neurosurgery. 2014;10
Suppl 2:273-293, discussion 293.
32. Psaltis AJ, Overton LJ, Thomas WW 3rd, Fox NF, Banks CA,
Schlosser RJ. Differences in skull base thickness in patients
with spontaneous cerebrospinal fluid leaks. Am J Rhinol
Allergy. 2014;28:e73-e79.
33. McNemar Q. Psychological Statistics. 4th ed.New York:
Wiley; 1969.
34. Citardi MJ, Herrmann B, Hollenbeak CS, Stack BC, Cooper M,
Bucholz RD. Comparison of scientific calipers and computer-
enabled CT review for the measurement of skull base and cra-
niomaxillofacial dimensions. Skull Base. 2001;11:5-11.
35. Damstra J, Huddleston Slater JJ, Fourie Z, Ren Y. Reliability
and the smallest detectable differences of lateral cephalometric
measurements. Am J Orthod Dentofacial Orthop. 2010;138:
546.e1-546.e8, discussion 546-547.
36. Guyatt G, Walter S, Norman G. Measuring change over time:
assessing the usefulness of evaluative instruments. J Chronic
Dis. 1987;40:171-178.
37. Kropmans TJ, Dijkstra PU, Stegenga B, Stewart R, de Bont
LG. Smallest detectable difference in outcome variables
related to painful restriction of the temporomandibular joint. J
Dent Res. 1999;78:784-789.
38. Luu NS, Nikolcheva LG, Retrouvey JM, et al. Linear measure-
ments using virtual study models. Angle Orthod. 2012;82:
1098-1106.
39. Moerenhout BA, Gelaude F, Swennen GR, Casselman JW,
Van Der Sloten J, Mommaerts MY. Accuracy and repeatability
of cone-beam computed tomography (CBCT) measurements
used in the determination of facial indices in the laboratory
setup. J Craniomaxillo Fac Surg. 2009;37:18-23.
40. Olszewski R, Zech F, Cosnard G, Nicolas V, Macq B,
Reychler H. Three-dimensional computed tomography cepha-
lometric craniofacial analysis: experimental validation in vitro.
Int J Oral Maxillofac Surg. 2007;36:828-833.
41. Suomalainen A, Vehmas T, Kortesniemi M, Robinson S,
Peltola J. Accuracy of linear measurements using dental cone
beam and conventional multislice computed tomography.
Dentomaxillofac Radiol. 2008;37:10-17.
42. Swennen GR, Schutyser F, Barth EL, De Groeve P, De Mey
A. A new method of 3-D cephalometry, part I: the anatomic
Cartesian 3-D reference system. J Craniofac Surg. 2006;17:
314-325.
43. van Vlijmen OJ, Maal T, Berge SJ, Bronkhorst EM, Katsaros
C, Kuijpers-Jagtman AM. A comparison between 2D and 3D
cephalometry on CBCT scans of human skulls. Int J Oral
Maxillofac Surg. 2010;39:156-160.
44. Friedman DI, Jacobson DM. Diagnostic criteria for idiopathic
intracranial hypertension. Neurology. 2002;59:1492-1495.
45. Hannerz J, Ericson K. The relationship between idiopathic intra-
cranial hypertension and obesity. Headache. 2009;49:178-184.
46. Kapur TR, Bangash W. Tegmental and petromastoid defects in
the temporal bone. J Laryngol Otol. 1986;100:1129-1132.
47. Lang DV. Macroscopic bony deficiency of the tegmen tympani
in adult temporal bones. J Laryngol Otol. 1983;97:685-688.
48. Branstetter BF 4th, Harrigal C, Escott EJ, Hirsch BE. Superior
semicircular canal dehiscence: oblique reformatted CT images
for diagnosis. Radiology. 2006;238:938-942.
49. Masaki Y. The prevalence of superior canal dehiscence syn-
drome as assessed by temporal bone computed tomography
imaging. Acta Otolaryngol. 2011;131:258-262.
50. Paladin AM, Phillips GS, Raske ME, Sie KC. Labyrinthine
dehiscence in a child. Pediatr Radiol. 2008;38:348-350.
51. Williamson RA, Vrabec JT, Coker NJ, Sandlin M. Coronal
computed tomography prevalence of superior semicircular canal
dehiscence. Otolaryngol Head Neck Surg. 2003;129:481-489.
52. Zhou G, Ohlms L, Liberman J, Amin M. Superior semicircular
canal dehiscence in a young child: implication of developmen-
tal defect. Int J Pediatr Otorhinolaryngol. 2007;71:1925-1928.
Stevens et al 179
at SOCIEDADE BRASILEIRA DE CIRUR on January 17, 2015oto.sagepub.comDownloaded from