Differences in skull base thickness in patients with spontaneous cerebrospinal fluid leaks Alkis J. Psaltis, M.D., Ph.D., Lewis J. Overton, M.D., William W. Thomas, III, M.D., Nyssa Flemming Fox, M.D., Caroline A. Banks, M.D., and Rodney J. Schlosser, M.D.
Y P
ABSTRACT
Background: This study was designed to determine if differences in overall skull base thickness exist in patients with spontaneous cerebrospinal fluid (SCSF) leaks and to compare our institution’s 10-year experience with the endoscopic repair of these leaks to the existing literature. A retrospective cohort study was performed in a tertiary rhinologic practice. Methods: A retrospective study by two blinded independent reviewers compared the skull base thickness on computerized tomography (CT) imaging in our SCSF leak patients to nonleaking controls and patients with traumatic CSF leaks. Surgical outcomes were compared with a pooled analysis of the published literature Results: SCSF leaks were seen most commonly in obese, middle-aged women of African American descent. SCSF leak patients were shown to have thinner skull bases in the region of the ethmoid roof, lateral lamella, and anterior face of the sella (p ⬍ 0.05) compared with patients with traumatic leaks and nonleaking controls. No racial difference in skull base thickness was observed when patients were matched according to leak type. The success rate of primary endoscopic intervention in our patients was high at 94% and comparable with the published literature. Conclusion: Differences in skull base thickness exist in SCSF patients, but whether this predisposes to such leaks or is a consequence of the underlying pathophysiology remains unknown. The endoscopic endonasal approach provides a highly effective means of repairing such leaks with success rates comparable with the endoscopic repair of other leak types. Adjuvant measures including weight reduction, lumbar drain, and acetazolamide use may increase success rates and should be considered in their management. (Am J Rhinol Allergy 28, e73–e79, 2014; doi: 10.2500/ajra.2014.28.4002)
T
C
erebrospinal fluid (CSF) leaks occur as a consequence of a breach in the integrity of the skull base and its underlying dura. Untreated, such leaks may result in serious complications such as pneumocephalus, meningitis, or cerebral abscess. A recent metaanalysis revealed that spontaneous CSF (SCSF) leaks may represent the most frequent type of leak, accounting for ⬎40% of all CSF leaks requiring surgery.1 This is a new finding with traumatic leaks historically considered the most common. Reasons for this may reflect a reduction in traumatic leaks from seat belt and airbag usage and improved surgical techniques or may simply represent an increase in diagnosis or prevalence of SCSF leaks. SCSF leaks occur in the absence of a discernible cause. They present with unique clinical and radiological features and historically have been associated with higher recurrence rates. Patients are typically obese middle-aged women, often presenting with coexistent features of benign intracranial hypertension (BIH). Although radiographic findings consistent with increased intracranial pressure, such as empty sella, have been reported, it is unclear if this translates into generalized differences in the skull base thickness in SCSF leak patients. This study compares our institution’s 10-year experience with the management of SCSF leaks to a pooled analysis of previously published literature and examines whether radiological differences in skull base thicknesses exist in this patient group.
O D
O N
METHODS
This study was approved by the Medical University of South Carolina’s (MUSC) Institutional Research Board (IRB 11037).
From the Department of Otolaryngology Head and Neck Surgery, Medical University of South Carolina, Charleston, South Carolina The authors have no conflicts of interest to declare pertaining to this article Address correspondence to Alkis J. Psaltis, M.D., Ph.D., Department of Otolaryngology, Head and Neck Surgery, Medical University of South Carolina, 135 Rutledge Avenue, Charleston, SC 29425 E-mail address: [email protected] Copyright © 2014, OceanSide Publications, Inc., U.S.A.
O C
Retrospective Review of SCSF Leaks at MUSC and Skull Base Measurement The charts of all patients undergoing endoscopic repair of any type of CSF leak between 2002 and 2012 were reviewed by two independent observers. Patients with SCSF and traumatic leaks were identified. Information recorded included demographics, body mass index (BMI), leak location, clinical presentation, diagnostic workup, repair method, use of adjuvant therapies, complications, and success rate of repair. Available computed tomography (CT) scans of these patients and a random selection of non-CSF leak patients were downloaded from the radiology server, deidentified, and randomly loaded for review. Two blinded independent observers reviewed all images recording skull base thickness at predetermined bilateral locations 6 mm off midline. The aim of this study was to determine whether SCSF leak patients had overall differences in skull base thickness compared with other individuals. Measurements were made in the sagittal plane at the (1) thinnest portion of the posterior table of each frontal sinus, (2) ethmoid roof at the midpoint between the posterior aspect of the frontal recess and the front face of the sphenoid sinus, and (3) thinnest part of the anterior face of the sella turcica. Coronally, observers measured (1) thickness of the ethmoid roof, (2) the height and thickness of the lateral lamella immediately posterior to the orbital globe, and (3) the thinnest portion of the sphenoid sinus lateral wall 6 mm posterior to the sphenoid rostrum (Fig. 1). Measurement of the leak site itself was intentionally avoided so as not to confound the primary purpose of this part of the study. In addition, it was thought that displacement of the skull base from prolapsed meninges or a traumatic fracture may introduce measurement artifact.
Systematic Review and Pooled Analysis of Previous Published Literature A search of all studies published through January 1, 2012, reporting outcomes after endoscopic repair of CSF leaks, was conducted using Pubmed, Medline/Old Medline, and Cochrane Central databases. Two authors performed the search independently to identify potential articles using key words “cerebrospinal fluid leak or fistula,” “CSF leak or fistula,” “benign intracranial hypertension,” “idiopathic cerebrospinal fluid leak,” “endoscopic sinus surgery,” “ESS,” “FESS,”
American Journal of Rhinology & Allergy
Delivered by Ingenta to: Economics Dept IP: 95.181.177.57 On: Tue, 28 Jun 2016 16:51:56 Copyright (c) Oceanside Publications, Inc. All rights reserved. For permission to copy go to https://www.oceansidepubl.com/permission.htm
e73
Y P
T
O N
O C
Figure 1. (A–E) Shows where the predetermined measurements were taken on coronal and computed tomography (CT) scans.
“endoscopic repair,” “transnasal sinus surgery,” and “complications.” Article references were scanned to identify additional studies missed in the primary search. Inclusion required a full-text English article, with at least five human patients undergoing purely endoscopic surgical repair of a SCSF leak with documented follow-up to assess repair success. Studies were excluded if heterogeneous in leak type, information could not be reliably extracted, leaks were managed nonsurgically, insufficient follow-up, or the same cohort of patients was used in a later larger series. Figure 2 summarizes the search strategy used. Initial data extraction was performed independently and then entered into a standardized database after discussion and consensus resolution of discrepancies.
O D
Statistical Analysis
Skull Base Thickness. The intraclass correlation coefficient for the two rater analysis was performed using AgreeStat 2011 software (Advanced Analytics, LLC, Gaithersburg, MD). Using GraphPad prism software Version 5.02 (GraphPad Software, Inc., La Jolla, CA), groups were compared with one-way ANOVA testing with post hoc Bonferroni multiple comparison test on the average measurements at each site. Systematic Review Pooled Analysis. Statistical analysis was performed using StatsDirect software Version 2.7.8. (StatsDirect, Cheshire, U.K.). Each variable of interest was analyzed independently to derive a summary measure across all published studies for which data were available. Given the likelihood of variability between studies, a random effects model (DerSimonian-Laird) was used, with heterogeneity
e74
assessed by the I2 statistic. The possibility of publication bias was evaluated using Begg’s and Egger’s tests, as well as visual inspection of the funnel plot. Where possible, results are described in accordance with the PRISMA guideline for reporting meta-analyses, with 95% confidence intervals. A value of p ⬍ 0.05 was considered significant for all statistical tests.
RESULTS MUSC Series Radiological Comparison of Skull Base Thickness. CT scans from 32 SCSF leak patients, 25 traumatic leak patients, and 36 controls were available for review. Table 1 shows the comparative demographical information for each group. A Cohen’s -score of 0.62 was obtained for all measurements made by the two observers, signifying substantial interobserver agreement.2 Compared with other groups, SCSF leak patients had significantly thinner skull bases at the lateral lamella, ethmoid roof, and anterior wall of sella (p ⬍ 0.05). No significant difference was observed in lateral lamella height or skull base thickness at the posterior table of the frontal sinus or the lateral sphenoid sinus wall (p ⬎ 0.05) [Fig. 3; Table 1]). When patients were matched according to leak type and race, no racial difference in skull base thickness was observed other than the anterior sella thickness in the control nonleak population. In this group, white patients had significantly thicker skull bases in this region than black patients (p ⬍ -.05). No significant difference in the
January–February 2014, Vol. 28, No. 1
Delivered by Ingenta to: Economics Dept IP: 95.181.177.57 On: Tue, 28 Jun 2016 16:51:56 Copyright (c) Oceanside Publications, Inc. All rights reserved. For permission to copy go to https://www.oceansidepubl.com/permission.htm
No English abstract (1)
Abstracts reviewed (135)
Previous meta-analysis (1)
T
Not all CSF leaks (1)
O N
Y P
O C
Figure 2. Search strategy for systematic review of literature relevant to the endoscopic repair of spontaneous cerebrospinal fluid (SCSF) leaks.
Table 1 Summary of demographic and radiological skull base thickness for different patient groups
Demographics No. of patients Age (yr) Gender Race Radiological measurement; mean (mm) and SD (mm) Frontal post–table thickness Ethmoid roof thickness Sagittal Coronal Lateral lamella thickness Lateral lamella height Lateral sphenoid thickness Anterior wall of sella thickness
O D
SCSF Leak
Traumatic CSF Leak
Controls
32 55 F ⫽ 30; M ⫽ 2 B ⫽ 25; W ⫽ 7
25 43 F ⫽ 11; M ⫽ 14 B ⫽ 14; W ⫽ 11
36 45 F ⫽ 29; M ⫽ 7 B ⫽ 16; W ⫽ 20
16.9 ⫾ 5.4
17.6 ⫾ 10.8
15.7 ⫾ 15.4
7.2 ⫾ 3.7 9.6 ⫾ 4.5 7.9 ⫾ 1.7 4.9 ⫾ 2.1 10.5 ⫾ 4.4 8.8 ⫾ 3.9*
7.1 ⫾ 2.5 7.8 ⫾ 2.8 7.8 ⫾ 2.5 5.1 ⫾ 1.8 9.2 ⫾ 3.0 37.8 ⫾ 2.5
5.3 ⫾ 3.1* 7.3 ⫾ 3.7 6.8 ⫾ 2.1* 5.6 ⫾ 2.7 9.2 ⫾ 8.4 5.5 ⫾ 2.1*
*Measurements in SCSF leak patients that were statistically significantly different (p ⬍ 0.05) compared with controls and traumatic leak patients. SCSF ⫽ spontaneous cerebrospinal fluid; B ⫽ black; W ⫽ white.
skull base thickness of patients with and without radiological evidence of an empty sella was identified. Demographics, Clinical Features, and Diagnostic-Related Features. Fortyfive MUSC patients with 63 fistula sites were identified. Table 2 summarizes the demographic, clinical, and radiological features obtained from chart and image review. Thirty-one (69%) patients were
black. Mean BMI was 35.1 kg/m2 (range, 25.64–52.93 kg/m2) with 36 (80%) patients who were obese (BMI ⬎ 30 kg/m2) and 28 (62%) who were morbidly obese (BMI ⬎ 35kg/m2). The ethmoid roof/cribriform plate and sphenoid sinuses were the most common leak sites accounting for 27 (42%) and 26 (41%) of all leaks, respectively. Of the sphenoid sinus leaks, 20 (77%) were located in the lateral recess with 6
American Journal of Rhinology & Allergy
Delivered by Ingenta to: Economics Dept IP: 95.181.177.57 On: Tue, 28 Jun 2016 16:51:56 Copyright (c) Oceanside Publications, Inc. All rights reserved. For permission to copy go to https://www.oceansidepubl.com/permission.htm
e75
Ethmoid Roof (sagittal view) Ethmoid Roof (coronal)
p
Y P
ABSTRACT
Background: This study was designed to determine if differences in overall skull base thickness exist in patients with spontaneous cerebrospinal fluid (SCSF) leaks and to compare our institution’s 10-year experience with the endoscopic repair of these leaks to the existing literature. A retrospective cohort study was performed in a tertiary rhinologic practice. Methods: A retrospective study by two blinded independent reviewers compared the skull base thickness on computerized tomography (CT) imaging in our SCSF leak patients to nonleaking controls and patients with traumatic CSF leaks. Surgical outcomes were compared with a pooled analysis of the published literature Results: SCSF leaks were seen most commonly in obese, middle-aged women of African American descent. SCSF leak patients were shown to have thinner skull bases in the region of the ethmoid roof, lateral lamella, and anterior face of the sella (p ⬍ 0.05) compared with patients with traumatic leaks and nonleaking controls. No racial difference in skull base thickness was observed when patients were matched according to leak type. The success rate of primary endoscopic intervention in our patients was high at 94% and comparable with the published literature. Conclusion: Differences in skull base thickness exist in SCSF patients, but whether this predisposes to such leaks or is a consequence of the underlying pathophysiology remains unknown. The endoscopic endonasal approach provides a highly effective means of repairing such leaks with success rates comparable with the endoscopic repair of other leak types. Adjuvant measures including weight reduction, lumbar drain, and acetazolamide use may increase success rates and should be considered in their management. (Am J Rhinol Allergy 28, e73–e79, 2014; doi: 10.2500/ajra.2014.28.4002)
T
C
erebrospinal fluid (CSF) leaks occur as a consequence of a breach in the integrity of the skull base and its underlying dura. Untreated, such leaks may result in serious complications such as pneumocephalus, meningitis, or cerebral abscess. A recent metaanalysis revealed that spontaneous CSF (SCSF) leaks may represent the most frequent type of leak, accounting for ⬎40% of all CSF leaks requiring surgery.1 This is a new finding with traumatic leaks historically considered the most common. Reasons for this may reflect a reduction in traumatic leaks from seat belt and airbag usage and improved surgical techniques or may simply represent an increase in diagnosis or prevalence of SCSF leaks. SCSF leaks occur in the absence of a discernible cause. They present with unique clinical and radiological features and historically have been associated with higher recurrence rates. Patients are typically obese middle-aged women, often presenting with coexistent features of benign intracranial hypertension (BIH). Although radiographic findings consistent with increased intracranial pressure, such as empty sella, have been reported, it is unclear if this translates into generalized differences in the skull base thickness in SCSF leak patients. This study compares our institution’s 10-year experience with the management of SCSF leaks to a pooled analysis of previously published literature and examines whether radiological differences in skull base thicknesses exist in this patient group.
O D
O N
METHODS
This study was approved by the Medical University of South Carolina’s (MUSC) Institutional Research Board (IRB 11037).
From the Department of Otolaryngology Head and Neck Surgery, Medical University of South Carolina, Charleston, South Carolina The authors have no conflicts of interest to declare pertaining to this article Address correspondence to Alkis J. Psaltis, M.D., Ph.D., Department of Otolaryngology, Head and Neck Surgery, Medical University of South Carolina, 135 Rutledge Avenue, Charleston, SC 29425 E-mail address: [email protected] Copyright © 2014, OceanSide Publications, Inc., U.S.A.
O C
Retrospective Review of SCSF Leaks at MUSC and Skull Base Measurement The charts of all patients undergoing endoscopic repair of any type of CSF leak between 2002 and 2012 were reviewed by two independent observers. Patients with SCSF and traumatic leaks were identified. Information recorded included demographics, body mass index (BMI), leak location, clinical presentation, diagnostic workup, repair method, use of adjuvant therapies, complications, and success rate of repair. Available computed tomography (CT) scans of these patients and a random selection of non-CSF leak patients were downloaded from the radiology server, deidentified, and randomly loaded for review. Two blinded independent observers reviewed all images recording skull base thickness at predetermined bilateral locations 6 mm off midline. The aim of this study was to determine whether SCSF leak patients had overall differences in skull base thickness compared with other individuals. Measurements were made in the sagittal plane at the (1) thinnest portion of the posterior table of each frontal sinus, (2) ethmoid roof at the midpoint between the posterior aspect of the frontal recess and the front face of the sphenoid sinus, and (3) thinnest part of the anterior face of the sella turcica. Coronally, observers measured (1) thickness of the ethmoid roof, (2) the height and thickness of the lateral lamella immediately posterior to the orbital globe, and (3) the thinnest portion of the sphenoid sinus lateral wall 6 mm posterior to the sphenoid rostrum (Fig. 1). Measurement of the leak site itself was intentionally avoided so as not to confound the primary purpose of this part of the study. In addition, it was thought that displacement of the skull base from prolapsed meninges or a traumatic fracture may introduce measurement artifact.
Systematic Review and Pooled Analysis of Previous Published Literature A search of all studies published through January 1, 2012, reporting outcomes after endoscopic repair of CSF leaks, was conducted using Pubmed, Medline/Old Medline, and Cochrane Central databases. Two authors performed the search independently to identify potential articles using key words “cerebrospinal fluid leak or fistula,” “CSF leak or fistula,” “benign intracranial hypertension,” “idiopathic cerebrospinal fluid leak,” “endoscopic sinus surgery,” “ESS,” “FESS,”
American Journal of Rhinology & Allergy
Delivered by Ingenta to: Economics Dept IP: 95.181.177.57 On: Tue, 28 Jun 2016 16:51:56 Copyright (c) Oceanside Publications, Inc. All rights reserved. For permission to copy go to https://www.oceansidepubl.com/permission.htm
e73
Y P
T
O N
O C
Figure 1. (A–E) Shows where the predetermined measurements were taken on coronal and computed tomography (CT) scans.
“endoscopic repair,” “transnasal sinus surgery,” and “complications.” Article references were scanned to identify additional studies missed in the primary search. Inclusion required a full-text English article, with at least five human patients undergoing purely endoscopic surgical repair of a SCSF leak with documented follow-up to assess repair success. Studies were excluded if heterogeneous in leak type, information could not be reliably extracted, leaks were managed nonsurgically, insufficient follow-up, or the same cohort of patients was used in a later larger series. Figure 2 summarizes the search strategy used. Initial data extraction was performed independently and then entered into a standardized database after discussion and consensus resolution of discrepancies.
O D
Statistical Analysis
Skull Base Thickness. The intraclass correlation coefficient for the two rater analysis was performed using AgreeStat 2011 software (Advanced Analytics, LLC, Gaithersburg, MD). Using GraphPad prism software Version 5.02 (GraphPad Software, Inc., La Jolla, CA), groups were compared with one-way ANOVA testing with post hoc Bonferroni multiple comparison test on the average measurements at each site. Systematic Review Pooled Analysis. Statistical analysis was performed using StatsDirect software Version 2.7.8. (StatsDirect, Cheshire, U.K.). Each variable of interest was analyzed independently to derive a summary measure across all published studies for which data were available. Given the likelihood of variability between studies, a random effects model (DerSimonian-Laird) was used, with heterogeneity
e74
assessed by the I2 statistic. The possibility of publication bias was evaluated using Begg’s and Egger’s tests, as well as visual inspection of the funnel plot. Where possible, results are described in accordance with the PRISMA guideline for reporting meta-analyses, with 95% confidence intervals. A value of p ⬍ 0.05 was considered significant for all statistical tests.
RESULTS MUSC Series Radiological Comparison of Skull Base Thickness. CT scans from 32 SCSF leak patients, 25 traumatic leak patients, and 36 controls were available for review. Table 1 shows the comparative demographical information for each group. A Cohen’s -score of 0.62 was obtained for all measurements made by the two observers, signifying substantial interobserver agreement.2 Compared with other groups, SCSF leak patients had significantly thinner skull bases at the lateral lamella, ethmoid roof, and anterior wall of sella (p ⬍ 0.05). No significant difference was observed in lateral lamella height or skull base thickness at the posterior table of the frontal sinus or the lateral sphenoid sinus wall (p ⬎ 0.05) [Fig. 3; Table 1]). When patients were matched according to leak type and race, no racial difference in skull base thickness was observed other than the anterior sella thickness in the control nonleak population. In this group, white patients had significantly thicker skull bases in this region than black patients (p ⬍ -.05). No significant difference in the
January–February 2014, Vol. 28, No. 1
Delivered by Ingenta to: Economics Dept IP: 95.181.177.57 On: Tue, 28 Jun 2016 16:51:56 Copyright (c) Oceanside Publications, Inc. All rights reserved. For permission to copy go to https://www.oceansidepubl.com/permission.htm
No English abstract (1)
Abstracts reviewed (135)
Previous meta-analysis (1)
T
Not all CSF leaks (1)
O N
Y P
O C
Figure 2. Search strategy for systematic review of literature relevant to the endoscopic repair of spontaneous cerebrospinal fluid (SCSF) leaks.
Table 1 Summary of demographic and radiological skull base thickness for different patient groups
Demographics No. of patients Age (yr) Gender Race Radiological measurement; mean (mm) and SD (mm) Frontal post–table thickness Ethmoid roof thickness Sagittal Coronal Lateral lamella thickness Lateral lamella height Lateral sphenoid thickness Anterior wall of sella thickness
O D
SCSF Leak
Traumatic CSF Leak
Controls
32 55 F ⫽ 30; M ⫽ 2 B ⫽ 25; W ⫽ 7
25 43 F ⫽ 11; M ⫽ 14 B ⫽ 14; W ⫽ 11
36 45 F ⫽ 29; M ⫽ 7 B ⫽ 16; W ⫽ 20
16.9 ⫾ 5.4
17.6 ⫾ 10.8
15.7 ⫾ 15.4
7.2 ⫾ 3.7 9.6 ⫾ 4.5 7.9 ⫾ 1.7 4.9 ⫾ 2.1 10.5 ⫾ 4.4 8.8 ⫾ 3.9*
7.1 ⫾ 2.5 7.8 ⫾ 2.8 7.8 ⫾ 2.5 5.1 ⫾ 1.8 9.2 ⫾ 3.0 37.8 ⫾ 2.5
5.3 ⫾ 3.1* 7.3 ⫾ 3.7 6.8 ⫾ 2.1* 5.6 ⫾ 2.7 9.2 ⫾ 8.4 5.5 ⫾ 2.1*
*Measurements in SCSF leak patients that were statistically significantly different (p ⬍ 0.05) compared with controls and traumatic leak patients. SCSF ⫽ spontaneous cerebrospinal fluid; B ⫽ black; W ⫽ white.
skull base thickness of patients with and without radiological evidence of an empty sella was identified. Demographics, Clinical Features, and Diagnostic-Related Features. Fortyfive MUSC patients with 63 fistula sites were identified. Table 2 summarizes the demographic, clinical, and radiological features obtained from chart and image review. Thirty-one (69%) patients were
black. Mean BMI was 35.1 kg/m2 (range, 25.64–52.93 kg/m2) with 36 (80%) patients who were obese (BMI ⬎ 30 kg/m2) and 28 (62%) who were morbidly obese (BMI ⬎ 35kg/m2). The ethmoid roof/cribriform plate and sphenoid sinuses were the most common leak sites accounting for 27 (42%) and 26 (41%) of all leaks, respectively. Of the sphenoid sinus leaks, 20 (77%) were located in the lateral recess with 6
American Journal of Rhinology & Allergy
Delivered by Ingenta to: Economics Dept IP: 95.181.177.57 On: Tue, 28 Jun 2016 16:51:56 Copyright (c) Oceanside Publications, Inc. All rights reserved. For permission to copy go to https://www.oceansidepubl.com/permission.htm
e75
Ethmoid Roof (sagittal view) Ethmoid Roof (coronal)
p
