Original Article
Carotid Cavernous Fistulas and Dural Arteriovenous Fistulas of the Cavernous Sinus:
Validation of a New Classification According to Venous Drainage
Giuseppe Leone1,2, Leonardo Renieri2, Alejandro Enriquez-Marulanda3, Adam A. Dmytriw4, Sergio Nappini2,
Antonio Laiso2, Giuseppe Buono5, Mariano Marseglia5, Adriana Iuliano6, Mario Muto1, Francesco Briganti5,
Salvatore Mangiafico2, Nicola Limbucci2
BACKGROUND: Multiple carotid cavernous fistula (CCF)
classifications have been proposed. However, they lacked
predictive factors for the clinical presentation, natural
history, and hemorrhagic risk. Our aim was to externally
validate a new classification according to venous drainage
(i.e., the Thomas classification [TC]) to assess its relationship with symptoms, endovascular treatment, and
outcomes.
-
METHODS: We performed a multicenter retrospective
review of CCFs at 2 major academic institutions. The CCFs
were classified using the Barrow classification (BC) and
TC systems.
-
RESULTS: The data from 94 patients with a diagnosis of
CCF were collected during a study period 23 years, 4
months. Of these 94 patients, 89 had undergone CCF treatment and 5 had experienced spontaneous occlusion.
Complete occlusion was achieved in 89.9% of the treated
patients. Complications occurred in 5.3% of the patients,
including permanent deficits in 2.1%. TC type 4 was
associated with cortical symptoms compared with type 2
(P [ 0.003) and type 3 (P < 0.001). The BC was not able to
detect significant differences among the symptom types.
Significant differences were found using the TC for the
transarterial-only, transvenous anterior-only, and transvenous posterior-only approaches (P < 0.001, P [ 0.03, and
P [ 0.001, respectively). The transvenous posterior and
transvenous anterior approach were significantly associated with type 2 and 3 TC, respectively. Excluding direct
CCFs, the BC was not related to the treatment approach. No
significant differences in the outcomes were found. However, a trend toward a lower occlusion rate for TC type 4
compared with type 3 was observed.
CONCLUSION: The TC provided useful information
regarding the fistula anatomy and venous hemodynamics,
which correlated with the clinical symptoms and treatment
strategy.
-
-
Key words
- Carotid cavernous fistula
- Classification
- Venous drainage
Abbreviations and Acronyms
BC: Barrow classification
CCF: Carotid cavernous fistula
CS: Cavernous sinus
DAVF: Dural arteriovenous fistula
ECA: External carotid artery
ICA: Internal carotid artery
TC: Thomas classification
From the 1Department of Neuroradiology, Antonio Cardarelli Azienda Ospedaliera di Rilievo
Nazionale, Naples, Italy; 2Neurovascular Interventional Unit, Careggi University Hospital,
WORLD NEUROSURGERY 128: e621-e631, AUGUST 2019
INTRODUCTION
C
arotid cavernous fistulas (CCFs) are acquired vascular lesions, representing an abnormal connection by direct
shunt or dural branches of the internal carotid artery
(ICA) and/or the external carotid artery (ECA) with the cavernous
sinus (CS). Indirect connections include dural arteriovenous fistulas (DAVFs) of the CS. The CCFs types are disparate. Hemodynamic classifications have traditionally separated CCFs into highflow and low-flow fistulas. In contrast, etiological classifications
have distinguished between spontaneous and traumatic lesions.1,2
Florence, Italy; 3Department of Neurosurgery, Boston Medical Center, Boston,
Massachusetts, USA; 4Division of Diagnostic and Therapeutic Neuroradiology, Toronto
Western Hospital, University Health Network, Toronto, Ontario, Canada; 5Unit of
Interventional Neuroradiology, Department of Advanced Biomedical Sciences, Federico II
University of Naples, Naples, Italy; and 6Ophthalmology Unit, Department of Neuroscience
and Reproductive and Odontostomatological Sciences, Federico II University of Naples,
Naples, Italy
To whom correspondence should be addressed: Giuseppe Leone, M.D.
[E-mail: g.leonemd@gmail.com]
Citation: World Neurosurg. (2019) 128:e621-e631.
https://doi.org/10.1016/j.wneu.2019.04.220
Journal homepage: www.journals.elsevier.com/world-neurosurgery
Available online: www.sciencedirect.com
1878-8750/$ - see front matter ª 2019 Elsevier Inc. All rights reserved.
www.journals.elsevier.com/world-neurosurgery
e621
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Figure 1. Thomas classification type 1 and Barrow classification type D
fistula in a sexagenarian with severe diplopia, ptosis, and headache. (A)
External and (B) internal right carotid angiograms in the lateral view
showing a dural carotid cavernous fistula, fed by the external carotid artery
and internal carotid artery dural branches, draining only toward the inferior
The first notable CCF classification, proposed by Barrow et al.,3
was based on the arterial angioarchitecture. However, the main
drawback of the Barrow classification (BC) has been the lack of
predictive factors for the clinical presentation, natural history,
and, importantly, the hemorrhagic risk. Other classifications,
limited to CS DAVFs, have been proposed.4-8 The classifications
by Borden et al.4 and Cognard et al.5 both focused on the drainage
Figure 2. Thomas classification type 2 and Barrow
classification type C fistula in a septuagenarian with
mild chemosis, exophthalmos, and diplopia. (A)
Angiogram after left external carotid artery injection
showing a dural carotid cavernous fistula with low flow,
e622
www.SCIENCEDIRECT.com
petrosal sinus. (C) Because of the patency of the inferior petrosal sinus,
retrograde transvenous catheterization was straightforward. (D) Common
carotid angiogram after coiling of the cavernous sinus showing occlusion of
the shunt.
pattern of DAVFs. However, these classifications have not
distinguished between anterior and posterior drainage in the CS
region, which could be important for treatment planning and
clinical correlation. These classifications, based on the venous
pattern,4-8 albeit excluding direct CCFs, reflect that symptoms
and hemorrhagic risk will be influenced more by the venous
drainage than by the arterial supply.9
draining toward the superior ophthalmic vein and
inferior petrosal sinus. (B) Because the symptoms
were not severe, conservative management was
attempted. The symptoms improved, and the 2-month
follow-up angiogram showed occlusion of the fistula.
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.04.220
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
Figure 3. Thomas classification type 3 and Barrow
classification type B fistula in a sexagenarian with
history of progressive chemosis, ocular pain
exophthalmos, ptosis, and diplopia. Right internal
carotid artery angiogram in (A) lateral and (B) frontal
views showing a high-flow dural carotid cavernous
fistula with drainage into the superior ophthalmic vein
(SOV). A tight stenosis was present at the junction
between the cavernous sinus and the SOV. (C)
Our aim was to validate a classification system for CCFs, recently
proposed by Thomas et al.,10 which was based on the venous
drainage pattern. We used data from a separate cohort of patients
WORLD NEUROSURGERY 128: e621-e631, AUGUST 2019
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Catheterization of the fistula from the thrombosed
inferior petrosal sinus was unsuccessful. The SOV was
surgically exposed and catheterized. The fistulous
compartment of the right cavernous sinus was
embolized with coils and Onyx. (D) A compliant balloon
was temporarily inflated in the carotid syphon to
prevent arterial reflux. Angiogram after embolization of
the right internal carotid artery in the (E) lateral and (F)
frontal views showing occlusion of the shunt.
with CCFs who had not been a part of the original study from 2
major academic institutions in Italy to assess its relationship with
the symptoms, endovascular treatment approach, and outcomes.
www.journals.elsevier.com/world-neurosurgery
e623
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Figure 4. Thomas classification type 4 and Barrow
classification type D fistula in a septuagenarian
complaining of severe headache, right
ophthalmoplegia, ptosis, and bilateral chemosis. Right
external carotid artery angiogram in (A) frontal and (B)
lateral views showing right carotid cavernous fistula.
The fistulous point was located in the medial
METHODS
Patient Characteristics
We retrospectively reviewed the medical records and imaging
studies of all patients who had been treated for CCFs at 2 major
Italian academic institutions from October 1994 to February 2018.
We enrolled 94 consecutive patients with CCFs. The population
was stratified according to age, sex, CCF type using the BC and
Thomas classification (TC),10 symptoms, treatment type,
occlusion rate, morbidity, mortality, and clinical outcomes. The
present study was conducted in accordance with institutional
guidelines and institutional review board approval. All the
patients had provided written informed consent before any
invasive procedure and also for the eventual scientific
publication of patient records and information anonymously.
e624
www.SCIENCEDIRECT.com
compartment of the cavernous sinus at level of the
junction with the intercavernous sinus. The drainage
was toward the superior ophthalmic vein and
sphenoparietal sinus, with massive cortical reflux.
Angiogram of right common carotid artery in (C) frontal
and (D) lateral views showing occlusion of the fistula.
CCF Classification Schemes
According to the BC,3 a CCF was classified as type A when a direct
connection between the ICA and CS was seen; type B or C when
the CCF was fed only by the dural branches of the ICA or ECA
respectively; and type D when the CCF was fed by the dural
branches of both the ICA and the ECA. Using the TC,10 the
CCFs were classified as type 1 when fistula drainage was present
only posteriorly and/or inferiorly through the inferior and/or
superior petrosal sinuses, pterygoid and parapharyngeal plexus
(Figure 1). The CCFs were classified as type 2 when the drainage
was both anterior through the ophthalmic veins and posterior
and/or inferior (Figure 2). The CCFs were classified as type 3
when the fistula drainage was occurring only anteriorly through
the superior and/or inferior ophthalmic veins (Figure 3). The
CCFs were classified as type 4 when cortical drainage was
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.04.220
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
Figure 5. Thomas classification type 5 and Barrow
classification type A fistula in a septuagenarian with
recent trauma resulting in skull base and right orbit
fractures with development of a direct carotid
cavernous fistula. She presented with a severe visual
deficit, bilateral exophthalmos, and ophthalmoplegia.
Right internal carotid artery angiogram in (A) frontal and
(B) lateral views showing multiple direct shunting
points, with drainage toward the bilateral inferior
petrosal sinus, superior ophthalmic vein, and right
identified as filling the superficial middle cerebral veins and
perimesencephalic and cerebellar venous system (Figure 4).
Finally, the CCFs were considered type 5 when a high-flow
direct arteriovenous shunt was present between the ICA and CS
(Figure 5).
Two experienced interventional neuroradiologists reviewed the
pretreatment angiograms and classified the CCFs using the BC
and TC. The discrepant cases were resolved by consensus. After
treatment, the follow-up imaging studies were reviewed, and the
CCFs were classified as completely occluded, partially occluded
(the shunt was still present but reduced), or inadequately occluded
(minimal or no reduction). The clinical presentation, including
WORLD NEUROSURGERY 128: e621-e631, AUGUST 2019
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
sphenoparietal sinus with cortical reflux. After failure of
the occlusion balloon test, selective embolization of the
fistula was performed. Coils were simultaneously
placed in the cavernous sinus through the retrograde
transvenous and transarterial approaches. (C) A
compliant balloon was temporary inflated in the carotid
syphon to avoid migration of the coils through the
fistula. (D) Internal carotid artery angiogram showing
the outcome after 2 embolization sessions.
ocular (i.e., loss of vision, ocular pain, glaucoma, chemosis,
exophthalmos), cavernous (i.e., ophthalmoplegia, diplopia, ptosis, anisocoria), and cortical (i.e., intracranial hemorrhage,
seizure, focal neurological deficits) involvement, was correlated
with the TC and BC.
Endovascular treatment was classified as conservative,
transarterial, transvenous, or combined. The transvenous route
was used for anterior CCFs via the ophthalmic veins or posterior/inferior CCFs via the petrosal sinuses and parapharyngeal plexus. The embolization methods included
coiling, a combination of coiling and Onyx injection (Medtronic, Minneapolis, Minnesota, USA), and transarterial
www.journals.elsevier.com/world-neurosurgery
e625
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Table 1. Patient Characteristics (n ¼ 94)
Characteristic
Value
Age (years)
Median
62.5
Interquartile range
54e73
Sex
Female
58 (61.7)
Male
36 (38.3)
Symptoms
Cavernous only
1 (1.1)
Ocular/orbital
25 (26.6)
Ocular/orbital plus cavernous symptoms
58 (61.7)
Cortical
8 (8.5)
Asymptomatic
2 (2.1)
Data presented as n (%), unless otherwise noted.
embolization (i.e., particles, coils, detachable balloons, or a
combination). The occlusion grade, retreatment rate,
morbidity, and mortality were recorded. The endovascular
approach used and occlusion rate were correlated with the type
of fistula according to the BC and TC.
Statistical Analysis
For statistical analysis, the observations were defined as patients,
CCFs, or procedures, as applicable. The categorical variables are
reported as proportions and continuous variables as the mean
standard deviation or median and interquartile range, as
appropriate, according to the distribution of the data. In each
group, categorical variables were compared using the c2 test, and
continuous variables were compared using the Kruskal-Wallis test.
A pairwise multiple comparison adjustment was performed using
the Bonferroni method and Dunn’s test. All statistical analyses
were performed using the Stata, version 14, statistical software
package (StataCorp., College Station, Texas, USA).
RESULTS
Patient Characteristics
Overall, 94 patients were identified. The patient characteristics are
summarized in Table 1. Most of the patients (61.7%) had
combined ocular and cavernous symptoms. Cortical symptoms
were reported in 8 patients (5 with type 4 TC and 3 with type 5
TC). Among the patients with cortical symptoms as their
primary presentation, 4 also had had ocular symptoms and 3
had also had cavernous symptoms. The CCF type according to
the BC and TC is reported in Table 2.
Treatments and Outcomes
Invasive treatment was performed in 89 patients (94.7%), and 5
patients (5.3%) had experienced spontaneous resolution. The
e626
www.SCIENCEDIRECT.com
proportion of patients with spontaneous occlusion resolution was
7.2% including only those with dural CCFs. All 5 patients with
spontaneous resolution had small-size fistulas, without cortical
venous drainage and had presented with few symptoms. Spontaneous occlusion had occurred within a mean period of 50 days
after the diagnosis. In addition, 3 of the 5 patients experienced
transient clinical worsening before spontaneous recovery.
The treatment characteristics are summarized in Table 3.
Complications (transient or permanent symptoms) occurred in 5
patients (5.3%). Treatment-related permanent complications
occurred in 2 patients (2.1%); both were ischemic events. In 1
patient, CS thrombosis and intracerebral bleeding occurred 2
months after unsuccessful transvenous embolization, before a new
treatment attempt but without permanent deficits. No deaths
occurred.
The mean follow-up time was 42 months (range, 24e60). The
complete occlusion rate after the first embolization treatment was
69.7%. Of the 89 patients, 18 (20.2%) required retreatment.
Excluding those patients with spontaneous resolution, after all the
treatments, occlusion was complete in 80 patients (89.9%), partial
in 6 (6.7%), and inadequate in 3 patients (3.4%). In addition,
clinical complete resolution occurred in 81 patients (86.2%) and
partial resolution in 7 (7.4%). However, the symptoms were unchanged in 3 patients (3.2%) and had worsened in 3 patients
(3.2%). None of the CCFs recurred.
Classification of CCFs and Association with Patient
Characteristics
Using the TC, we found a significant association with patient age
(P ¼ 0.01; Table 2). After adjustment for multiple comparisons, we
found a significantly lower patient age for those with type 5
compared with those with type 2 (P ¼ 0.002) and type 3
(P ¼ 0.004). The BC was also able to detect a significant
difference according to patient age (P ¼ 0.005; Table 4). After
adjustment for multiple comparisons, we found a significantly
lower age among those with type A compared with those with
type B (P ¼ 0.0007) and type D (P ¼ 0.007). The TC system was
also able to detect significant differences in the symptoms
(Table 2). After adjustment for multiple comparisons, we found
type 4 CCFs had a greater association with cortical symptoms
compared with type 2 (P ¼ 0.003) and type 3 (P < 0.001). Also,
type 4 CCFs were significantly less associated with ocular and/or
orbital symptoms with or without cavernous symptoms
compared with type 2 (P ¼ 0.003) and type 3 (P < 0.001).
Patients with type 1 CCFs had significantly more cavernous-only
symptoms compared with those with type 3 (P ¼ 0.001) and
type 5 (P ¼ 0.01). However, the low number of type 1 fistulas
limited the clinical significance of these results. We found no
significant differences among the patients’ symptoms using the
BC (Table 4). Finally, we found no significant differences in sex,
asymptomatic presentation, and spontaneous resolution of the
CCF using either classification system.
Classification of CCF and Association with Treatment Approach
Significant differences according to the TC were detected for the
transarterial-only, transvenous anterior-only, and transvenous
posterior-only approaches (P < 0.001, P ¼ 0.03, and P ¼ 0.001,
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.04.220
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Table 2. Carotid Cavernous Fistula Type According to Thomas Classification
Thomas Classification
Variable
Type 1 (n [ 4)
Type 2 (n [ 14)
Type 3 (n [ 41)
Type 4 (n [ 10)
Type 5 (n [ 25)
P Value
Age (years)
75 (65e78)
69.5 (60e78)
67 (56e72)
62.5 (54e65)
55 (38e62)
0.01*
Female sex
2 (50.0)
1 (7.1)
17 (41.5)
4 (40.0)
12 (48.0)
0.13
Cavernous only
1 (25.0)
0 (0.0)
0 (0.0)
0 (0.0)
0 (0.0)
<0.001*
Ocular/orbital with or without
cavernous symptoms
3 (75.0)
14 (100.0)
40 (97.6)
5 (50.0)
21 (84.0)
<0.001*
Symptoms
Cortical
0 (0.0)
0 (0.0)
0 (0.0)
5 (50.0)
3 (12.0)
<0.001*
Asymptomatic
0 (0.0)
0 (0.0)
1 (2.4)
0 (0.0)
1 (4.0)
0.9
Spontaneous resolution
1 (25.0)
1 (7.1)
3 (7.3)
0 (0.0)
0 (0.0)
0.24
Transarterial only
2 (50.0)
1 (7.1)
6 (14.6)
1 (10.0)
17 (68.0)
<0.001*
Transvenous anterior only
0 (0.0)
1 (7.1)
14 (34.1)
1 (10.0)
2 (8.0)
0.03*
Transvenous posterior only
1 (25.0)
10 (71.4)
14 (34.1)
8 (80.0)
4 (16.0)
0.001*
Combined transvenous
0 (0.0)
1 (7.1)
4 (9.8)
0 (0.0)
0 (0.0)
0.42
Open surgical
0 (0.0)
0 (0.0)
0 (0.0)
0 (0.0)
2 (8.0)
0.23
0 (0.0)
1 (7.1)
0 (0.0)
1 (10.0)
3 (12.0)
0.26
Treatment approach
Complications
Occlusion after first embolization
0.61
Complete
2 (66.7)
10 (76.9)
26 (68.4)
5 (50.0)
19 (76.0)
Partial or incomplete
1 (33.3)
3 (23.1)
12 (31.6)
5 (50.0)
6 (24.0)
0 (0.0)
3 (23.1)
8 (21.0)
3 (30.0)
4 (16.0)
Retreatment
Occlusion on follow-up after all treatment
0.79
0.012*
Complete
1 (33.3)
12 (92.3)
36 (94.7)
8 (80.0)
23 (92.0)
Partial or inadequate
2 (66.7)
1 (7.7)
2 (5.3)
2 (20.0)
2 (8.0)
Clinical outcome
0.59
Resolution
4 (100.0)
12 (85.7)
38 (92.7)
8 (80.0)
19 (76.0)
Partial resolution
0 (0.0)
1 (7.1)
2 (4.9)
0 (0.0)
4 (16.0)
Unchanged
0 (0.0)
0 (0.0)
1 (2.4)
1 (10.0)
1 (4.0)
Worse
0 (0.0)
1 (7.1)
0 (0.0)
1 (10.0)
1 (4.0)
Data presented as median (interquartile range; P value calculated using Kruskal-Wallis test) or n (%; P value calculated using c test).
*Statistically significant (P < 0.05).
2
respectively; Table 2). After adjustment for multiple comparisons,
we found that type 5 CCFs were significantly more likely to be
associated with a transarterial approach compared with type 2
(P < 0.001), type 3 (P < 0.001), and type 4 (P ¼ 0.002). Also,
type 5 was significantly less associated with the transvenous
posterior-only approach compared with type 2 (P ¼ 0.001) and
type 4 (P < 0.001). Posterior transvenous treatment was more
common for type 2 than type 3 fistulas (P ¼ 0.015) and for type 2
fistulas compared with the remaining population (P < 0.05;
considering both all types of fistulas and dural fistulas only). The
anterior transvenous approach was more common for type 3
WORLD NEUROSURGERY 128: e621-e631, AUGUST 2019
fistulas than for the remaining population (P < 0.001). A trend
was found for using the anterior transvenous approach for type 3
fistulas compared with type 2 fistulas (P ¼ 0.05).
The BC was able to detect differences for the transarterial-only
and transvenous posterior-only approaches (Table 4). After
multiple comparison adjustment, type A was associated with
greater rates of the transarterial approach compared with types
B and D (P < 0.001 for both comparisons). Additionally, type B
was associated with greater rates of the transvenous posterioronly approach compared with type A (P < 0.001). No other significant differences were found when comparing the treatment
www.journals.elsevier.com/world-neurosurgery
e627
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Table 3. Treatment Characteristics (n ¼ 94)
Characteristic
n (%)
Spontaneous resolution
5 (5.3)
n ¼ 89
Treatment
Transarterial only
27 (28.7)
Transvenous anterior only
18 (19.1)
Transvenous posterior only
37 (39.4)
Transvenous anterior plus posterior
5 (5.3)
Combined arterial plus posterior
Embolic material used
2 (2.1)
n ¼ 89
compared with the other fistula types. A significant difference in
the occlusion rate at follow-up (excluding spontaneous resolution)
was detected using the TC (Table 2). After multiple comparison
adjustment, type 1 was associated with lower complete occlusion
rates compared with type 3 (P ¼ 0.001) and type 5 (P ¼ 0.005).
However, the low number of type 1 fistulas limited the clinical
impact of these results. Among the other types, type 4 had the
lowest complete occlusion rate (80%), without significant
differences compared with the other fistula types (P ¼ 0.36).
Using the BC, no significant differences were noted for the
occlusion rates at follow-up after all treatments (Table 4). The
occlusion status after the first embolization and retreatment was
not significantly different using either scale. The clinical
outcome also was not significantly different using either scale.
Coils
64 (73.6)
Acrylic glue
2 (2.3)
DISCUSSION
Particles
4 (4.6)
Coils plus Onyx
12 (13.8)
Coils plus particles
1 (1.1)
Onyx plus particles
1 (1.1)
The BC and TC are the only classification systems that include
both direct and dural fistulas.3,10 Other classifications, such as the
Cognard classification, have included only dural fistulas.5
Classifying only indirect fistulas is understandable from a
pathological viewpoint. Albeit direct and indirect fistulas are
different diseases with different etiologies, an anatomy-based
classification that includes both dural and direct fistulas, such
as the BC and TC, could be useful because both types of CCFs can
have overlapping symptoms, venous patterns, and therapy
options.
The mainstay of CCF treatment has been transvenous embolization, with transarterial embolization rarely performed and, in
most cases, reserved for direct CCFs.11 Transarterial embolization
of indirect low-flow CCFs will generally be cumbersome owing to
the small size, complex anatomy, and multiplicity of arterial
feeders. Therefore, transarterial embolization has been typically
used only to reduce arterial inflow before transvenous occlusion
for high-flow indirect CCFs and as a viable alternative after failure
of transvenous attempts. Although coils and particulate agents
have been used, these agents, used alone, cannot provide permanent occlusion of the fistula.
Transvenous access to the CS can be from the posterior
approach (usually from the inferior petrosal sinus) or anterior
approach (usually from the superior ophthalmic vein), or combined. Because the main difference between the BC and TC is that
former was based on the arterial anatomy and the latter on the
venous drainage pattern, we believe the TC system might be more
useful during therapeutic planning.
In contrast to a classification system based on the arterial
angioarchitecture, such as the BC,3 we found that the TC types
significantly correlated with the patients’ symptoms. The pattern
of symptoms is dependent on the functional and anatomical
features, such as the shunt velocity, direction of the venous
drainage, and pressure within the venous structures.10
Therefore, because the symptoms are strictly related to the
venous drainage pattern,11 a classification using the venous
drainage pattern could be of greater importance for the
management CCFs. Also, any novel classification should not
disregard this anatomical aspect.
Concerning the patient characteristics, type 5 and type A CCFs
were significantly associated with a younger patient age. Our
Detachable balloon
3 (3.4)
Complications
5 (5.3)
Occlusion after first embolization
n ¼ 89
Complete
62 (69.7)
Partial
18 (20.2)
Inadequate
9 (10.1)
Retreatment
18 (20.2)
Treatment number
n ¼ 89
1
71 (79.8)
2
17 (19.1)
3
1 (1.1)
Occlusion after all treatments
Complete
n ¼ 89
80 (89.9)
Partial
6 (6.7)
Inadequate
3 (3.4)
Clinical outcome
n ¼ 94
Complete resolution
81 (86.2)
Partial resolution
7 (7.4)
Unchanged
3 (3.2)
Worse
3 (3.2)
approach between the other types of fistulas using the BC. We
found no significant differences for the combined transvenous or
open surgical approach using either scale.
Classification of the CCF and Association with Outcomes
Complications occurred in 5.3% of the patients (12% of direct
fistulas). However, no significant differences were observed
e628
www.SCIENCEDIRECT.com
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.04.220
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
Table 4. Carotid Cavernous Fistula Type According to Barrow Classification
Barrow Classification
Variable
Type A (n [ 25)
Type B (n [ 10)
Type C (n [ 6)
Type D (n [ 53)
P Value
Age (years)
55 (38e62)
73.5 (65e76)
70.5 (56e80)
64 (56e73)
0.005*
Female sex
12 (48.0)
3 (30.0)
3 (50.0)
18 (34.0)
0.56
Cavernous only
0 (0.0)
0 (0.0)
0 (0.0)
1 (1.9)
0.85
Ocular/orbital with or without cavernous symptoms
21 (84.0)
9 (90.0)
5 (83.3)
48 (90.6)
0.83
Cortical
3 (12.0)
1 (10.0)
0 (0.0)
4 (7.5)
0.79
Asymptomatic
1 (4.0)
0 (0.0)
1 (16.7)
0 (0.0)
0.05
0 (0.0)
2 (20.0)
0 (0.0)
3 (5.7)
0.11
Symptoms
Spontaneous resolution
Treatment approach
Transarterial only
17 (68.0)
0 (0.0)
3 (50.0)
7 (13.2)
<0.001*
Transvenous anterior only
2 (8.0)
0 (0.0)
1 (16.7)
15 (28.3)
0.06
Transvenous posterior only
4 (16.0)
8 (80.0)
2 (33.3)
23 (43.4)
0.004*
Combined transvenous
0 (0.0)
0 (0.0)
0 (0.0)
5 (9.4)
0.25
Open surgical
2 (8.0)
0 (0.0)
0 (0.0)
0 (0.0)
0.13
3 (12.0)
1 (10.0)
0 (0.0)
1 (1.9)
0.24
Complications
Occlusion after first embolization
Complete
Partial/inadequate
Retreatment
0.48
19 (76.0)
7 (87.5)
4 (66.7)
32 (64.0)
6 (24.0)
1 (12.5)
2 (33.3)
18 (36.0)
4 (16.0)
0 (0.0)
0 (0.0)
14 (28.0)
Occlusion on follow-up after all treatment
0.13
0.26
Complete
23 (92.0)
7 (87.5)
4 (66.7)
46 (92.0)
Partial
2 (8.0)
1 (12.5)
2 (33.3)
4 (8.0)
Resolution
19 (76.0)
8 (80.0)
5 (83.3)
49 (92.4)
Partial resolution
4 (16.0)
0 (0.0)
0 (0.0)
3 (5.7)
Unchanged
1 (4.0)
1 (10.0)
1 (16.7)
0 (0.0)
Worse
1 (4.0)
1 (10.0)
0 (0.0)
1 (1.9)
Clinical outcome
0.16
Data presented as median (interquartile range; P value calculated using Kruskal-Wallis test) or n (%; P value calculated using c2 test).
*Statistically significant (P < 0.05).
findings are similar to those reported by Grumann et al.,12 likely
because direct CCFs are often the consequence of trauma, which
frequently occur in younger persons. We did not detect any
differences when stratified by sex, asymptomatic presentation,
or spontaneous resolution using the TC or BC.
In our series, type 1 CCFs were significantly associated with the
presence of cavernous symptoms only compared with types 3 and
5. Hemodynamically, it can be envisaged that posterior-only
venous drainage will increase the pressure in the CS, such that
vascular steal could result in cranial nerve deficits, manifesting as
ophthalmoplegia, ptosis, or anisocoria.10 Type 2 and 3 CCFs were
significantly associated with ocular symptoms because their
WORLD NEUROSURGERY 128: e621-e631, AUGUST 2019
drainage mainly occurs through the ophthalmic veins with
resultant ophthalmic venous hypertension, explaining the
development of orbital congestion, secondary glaucoma, and
optic neuropathy. Reversal of flow from the CCF into the
anterior aspect of the CS and the ophthalmic vein was present
in all our patients with orbital congestion. As expected, type 4
TC CCFs were associated with cortical symptoms. The
importance of the presence of cortical drainage is related to its
association with a severe presentation and a greater
hemorrhagic risk, which has also been stressed in various
classifications of DAVFs4,5 and dural CCFs6-8 but was not
considered in the BC.3
www.journals.elsevier.com/world-neurosurgery
e629
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
In our series, treatment was indicated for patients with severe
progressive symptoms and those with cortical venous drainage.
The reported rates of spontaneous dural CCF regression have
ranged from 3.7% to 90%.13-16 In our series, spontaneous resolution occurred in 5.3% of patients (7.2% including only dural
CCFs). This event occurred within a mean period of 50 days after
the diagnosis. Of the 5 patients, 3 experienced transient clinical
worsening before spontaneous recovery. Spontaneous closure can
occur when the shunt elevates the venous pressure within the sinus to a critical point at which stasis occurs and thrombosis is
facilitated.16,17 No spontaneous resolution occurred in patients
with TC type 4 or 5. A likely confounder was that type 4 CCFs were
promptly treated because of the significant hemorrhagic risk. In
addition, type 5 CCFs almost never spontaneously thrombose
owing to the high-flow shunt.
In its original description, the TC system was significantly
associated with the endovascular treatment approach used. Both
TC type 5 and BC type A, corresponding to direct fistulas, were
associated with the transarterial approach. However, TC types 1
and 2 were significantly associated with the transvenous posterior
approach, and type 3 was significantly associated with the transvenous anterior approach. In contrast, the BC system was not
associated with the different transvenous approaches.10 Also, in
our study, the BC and TC systems were associated with the
transarterial treatment approach for direct CCFs. Transarterial
embolization should be preferred because the large direct shunt
provides easy access to the CS to occlude the fistulous point.18
The transvenous posterior approach was associated with TC type
2 and the transvenous anterior approach with TC type 3 CCFs.
Also, in the study by Thomas et al.,10 the preference was to
treat type 3 CCFs with a transvenous anterior approach only, in
contrast to our series in which both anterior-only and posterioronly transvenous approaches were used equally for the treatment
of this type of CCF. In types 2 and 3 CCFs, either approach can be
successful, because catheterization of the inferior petrosal sinus
will be feasible even if it is occluded.19 Therefore, we believe that
REFERENCES
1. Ellis JA, Goldstein H, Connolly ES, Meyers PM.
Carotid-cavernous fistulas. Neurosurg Focus. 2012;
32:E9.
2. Henderson AD, Miller NR. Carotid-cavernous
fistula: current concepts in aetiology, investigation, and management. Eye Lond Engl. 2018;32:
164-172.
3. Barrow DL, Spector RH, Braun IF, Landman JA,
Tindall SC, Tindall GT. Classification and treatment of spontaneous carotid-cavernous sinus fistulas. J Neurosurg. 1985;62:248-256.
4. Borden JA, Wu JK, Shucart WA. A proposed
classification for spinal and cranial dural arteriovenous fistulous malformations and implications
for treatment. J Neurosurg. 1995;82:166-179.
5. Cognard C, Gobin YP, Pierot L, et al. Cerebral
dural arteriovenous fistulas: clinical and
e630
www.SCIENCEDIRECT.com
the treatment choice for categories is mainly dependent on the
neurointerventionalist’s preference.
We found an occlusion rate of 89.9% after all treatments, with
complete clinical resolution in 86.2%. These results are similar to
those reported in recent series.12,20-22 We found that TC type 1
CCFs were associated with a lower complete occlusion rate
compared with that for type 3 and type 5, although the symptoms
had resolved in all patients with type 1 CCF. It is possible that in a
type 1 fistula, a minimal pressure decrease in the CS might
improve the cavernous symptoms, such as ophthalmoplegia,
ptosis, and anisocoria. However, this class of fistula is rare, and
the small proportion of patients (4.2% in our series) weakened the
statistical power of this assertion.
Study Limitations
The present study was limited by its retrospective nature with all
the inherent bias of such a study design. The sample size was
relatively small, which could have limited our ability to detect
differences between the groups, mainly when performing subgroup analyses. Another limitation was the long temporal paradigm, which included a patient cohort treated during a study
period of >20 years, during which the preferred endovascular
methods changed.
CONCLUSION
Knowledge of the CCF drainage pattern is crucial for informed
decision making for endovascular treatment. The TC system
provided useful information on fistula anatomy and hemodynamics, which will correlate with the clinical symptoms in a large
cohort of patients. Our findings suggest that a classification system based on the venous drainage is applicable in clinical practice
and demonstrates a correlation with the symptoms. This could
further aid in planning fistula access during endovascular embolization. A comparison with other available classification systems
could be useful to further validate the TC.
angiographic correlation with a revised classification of venous drainage. Radiology. 1995;194:
671-680.
6. Wenderoth J. Proposal for an improved classification system for cavernous sinus dural arteriovenous fistula (CS-DAVF). J Neurointerventional Surg.
2017;9:220-224.
7. Suh DC, Lee JH, Kim SJ, et al. New concept in
cavernous sinus dural arteriovenous fistula: correlation with presenting symptom and venous
drainage patterns. Stroke. 2005;36:1134-1139.
8. Satomi J, Satoh K, Matsubara S, Nakajima N,
Nagahiro S. Angiographic changes in venous
drainage of cavernous sinus dural arteriovenous
fistulae after palliative transarterial embolization
or observational management: a proposed stage
classification. Neurosurgery. 2005;56:494-502 [discussion: 494-502].
9. Stiebel-Kalish H, Setton A, Nimii Y, et al.
Cavernous sinus dural arteriovenous malformations: patterns of venous drainage are related to
clinical signs and symptoms. Ophthalmology. 2002;
109:1685-1691.
10. Thomas AJ, Chua M, Fusco M, et al. Proposal of
venous drainage-based classification system for
carotid cavernous fistulae with validity assessment
in a multicenter cohort. Neurosurgery. 2015;77:
380-385 [discussion: 385].
11. Ducruet AF, Albuquerque FC, Crowley RW,
McDougall CG. The evolution of endovascular
treatment of carotid cavernous fistulas: a singlecenter experience. World Neurosurg. 2013;80:
538-548.
12. Grumann AJ, Boivin-Faure L, Chapot R, Adenis JP,
Robert PY. Ophthalmologic outcome of direct and
indirect carotid cavernous fistulas. Int Ophthalmol.
2012;32:153-159.
13. Shields CB, Tutt HP. Spontaneous obliteration of
carotid-cavernous fistulas. South Med J. 1981;74:
617-620.
WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2019.04.220
ORIGINAL ARTICLE
GIUSEPPE LEONE ET AL.
14. Briganti F, Caranci F, Leone G, et al. Endovascular
occlusion of dural cavernous fistulas through a
superior ophthalmic vein approach. Neuroradiol J.
2013;26:565-572.
15. Sasaki H, Nukui H, Kaneko M, et al. Long-term
observations in cases with spontaneous carotidcavernous fistulas. Acta Neurochir (Wien). 1988;90:
117-120.
VALIDATION OF NOVEL CCF AND DAVF CLASSIFICATION
direct carotid cavernous fistula. Interv Neuroradiol.
2013;19:344-352.
cavernous sinus dural arteriovenous fistula.
Neuroradiology. 2011;53:983-992.
19. Jia ZY, Song YS, Sheen JJ, Kim JG, Lee DH,
Suh DC. Cannulation of occluded inferior petrosal
sinuses for the transvenous embolization of
cavernous sinus dural arteriovenous fistulas: usefulness of a Frontier-wire probing technique. AJNR
Am J Neuroradiol. 2018;39:2301-2306.
Conflict of interest statement: The authors declare that the
article content was composed in the absence of any
commercial or financial relationships that could be construed
as a potential conflict of interest.
Received 30 January 2019; accepted 24 April 2019
20. Meyers PM, Halbach VV, Dowd CF, et al. Dural
carotid cavernous fistula: definitive endovascular
management and long-term follow-up. Am J Ophthalmol. 2002;134:85-92.
Citation: World Neurosurg. (2019) 128:e621-e631.
https://doi.org/10.1016/j.wneu.2019.04.220
21. Kirsch M, Henkes H, Liebig T, et al. Endovascular
management of dural carotid-cavernous sinus
fistulas in 141 patients. Neuroradiology. 2006;48:
486-490.
Available online: www.sciencedirect.com
17. Seeger JF, Gabrielsen TO, Giannotta SL, Lotz PR.
Carotid-cavernous sinus fistulas and venous
thrombosis. AJNR Am J Neuroradiol. 1980;1:141-148.
18. De Renzis A, Nappini S, Consoli A, et al. Balloonassisted coiling of the cavernous sinus to treat
22. Jung KH, Kwon BJ, Chu K, et al. Clinical and
angiographic factors related to the prognosis of
16. Luciani A, Houdart E, Mounayer C, Maurice J-PS,
Merland J-J. Spontaneous closure of dural arteriovenous fistulas: report of three cases and review
of the literature. AJNR Am J Neuroradiol. 2001;22:
992-996.
WORLD NEUROSURGERY 128: e621-e631, AUGUST 2019
Journal homepage: www.journals.elsevier.com/worldneurosurgery
1878-8750/$ - see front matter ª 2019 Elsevier Inc. All
rights reserved.
www.journals.elsevier.com/world-neurosurgery
e631