Keywords
cavernoma - developmental venous anomaly - DVA injury
Introduction
Cerebral cavernous malformations (CCMs) and developmental venous anomalies (DVAs)
are classified as low-flow vascular malformations. However, these two lesions differ
clinically. CCMs typically are a symptomatic lesion with seizures, hemorrhage, or
focal neurologic deficits depending on size and location.[1]
[2]
[3] DVAs are classified as benign vascular malformations with very low or nearly no
risk of hemorrhage. The annual risk of bleeding for a DVA ranges from 0.15% to 0.68%.[4]
[5] Furthermore, risk of hemorrhage-related morbidity is exceedingly low with a 0% mortality
rate.[6] DVA is usually accepted as atypical venous drainage of the normal brain tissue.
Venous infarction after significant DVA obliteration has been reported.[7] Therefore, surgical removal or obliteration is generally not recommended for isolated
DVAs.
According to the literature, DVAs associated with CCMs are found at a rate of 14 to
30%.[8]
[9]
[10]
[11]
[12]
[13] According to some articles, 100% of CCMs are associated with DVA.[14] Therefore, these combined lesions are likely the cause of DVAs presenting with hemorrhage.
Nonetheless, surgical resection of CCMs is considered the treatment of choice including
CCMs associated with DVA. Even so, meticulous dissection to avoid injury to the DVA
adjacent to the CCM is still recommended because of the fear of serious adverse consequences.
Conversely, there are reports of cases with intraoperative resection of medullary
veins and partial coagulation of the main draining veins of the DVA without severe
cerebral edema.[15]
Although there is the possibility of impairing a DVA during CCM resection, we have
also had experiences of coagulating a DVA during CCM surgery without complication,
particularly if the DVA could not be detected in presurgical imaging or if epilepsy
requires removal of the hemosiderin-staining brain tissue surrounding the CCM together
with the adjacent DVA. There is still no clinical study in which DVA resection during
CCM removal is compared with DVA preservation. Therefore it is the aim to compare
clinical features and imaging parameters in the patients with and without DVA compromise
after CCM resection.
Patients and Methods
Study Design, Setting, and Participants
A retrospective cohort study was approved by the local institutional review board
for clinical and imaging data collection and a review of the patients operated on
for cavernoma in our hospital from January 1, 2006, to December 31, 2011 was performed.
A total of 67 patients with cavernomas were treated surgically; however, we excluded
all spinal cavernous malformations and all patients with insufficient imaging data
and unavailable clinical follow-up data. Therefore, only 38 patients completely fulfilled
the inclusion criteria. All these 38 cases were operated on for CCMs, confirmed by
the histopathologic results, and had complete radiologic data (pre- and postoperative
magnetic resonance [MR] imaging; [Fig. 1]) and complete clinical information.
Fig. 1 Flowchart of the study algorithm. CCM, cerebral cavernous malformation; DVA, developmental
venous anomaly; MRI, magnetic resonance imaging.
Definition
All 38 CCM patients were divided into three groups by pre- and postoperative MR images:
group I, CCM without associated DVA; group II, impairment and occlusion of the associated
DVA during CCM removal; and group III, preservation of the associated DVA during CCM
removal. To define associated DVA, we used T1-weighted MR images with gadolinium,
T2-weighted MR images, and fluid-attenuated inversion recovery (FLAIR) signal MR images
that demonstrate abnormal venous tubular structures with typical caput medusae characteristics adjacent to the CCM. DVAs located remotely to the CCM were categorized
as group I. Damage, respectively occlusion of the DVA, was defined by the absence
of venous structures adjacent to the CCM in the postoperative MR images compared with
the corresponding preoperative MR images ([Fig. 2]).
Fig. 2 (A) Preoperative magnetic resonance imaging (MRI) (upper left) showed right frontal
cerebral cavernous malformation (CCM) with associated developmental venous anomaly
(DVA), and postoperative MRI (upper right) showed complete resection of CCM and absence
of associated DVA. (B) Intraoperative photograph shows associated DVA (arrow) adjacent
to CCM (asterisks) before (lower left) and after (lower right) DVA was coagulated
and resected.
Data Collection
The following data were collected: age, gender, number of CCMs, location, size recorded
as the average diameter in three dimensions (anteroposterior, left to right, and superoinferior)
from preoperative MRI, pre- and postoperative Karnofsky score, postoperative additional
neurologic deficits including new onset of a focal neurologic deficit on physical
examination. Postoperative peri-resectional edema was calculated as volume of ellipsoid
shape of hyperintensity on FLAIR MRI surrounding the resection area to evaluate the
consequence of venous compromise.
Follow-up Data
The long-term clinical data were evaluated by telephone interview. Nineteen of the
38 patients were followed up. The follow-up interval ranged from 1 to 7 years. The
clinical data collected during the interview included daily functional status matched
to Karnofsky score, seizure control status classified by using the Engel classification,
and postoperative headache via a subjective pain scale score from 0 to 10.
Statistical Analysis
The Fisher exact test was used to compare two groups and the chi-square test for three
groups for categorical variables. For continuous variables, comparison between groups
was performed with analysis of variance with least significant difference post hoc
test comparison between three groups or unpaired t test for two groups. Pearson correlation analysis was used to evaluate correlation
between continuous variables. Statistical significance was set at p < 0.05 for a 95% confidence interval. All statistics were performed with SPSS v.20.0
software (IBM Corp., Armonk, New York, United States).
Results
During a 6-year period from January 2006 to December 2011, 67 consecutive patients
with cavernous malformation were treated surgically in our hospital. According to
the requirements and methods of our study, only 38 patients were included in this
evaluation. The main reason for exclusion was the lack of adequately comparable pre-
and postoperative image data. The mean patient age was 39 years; 21 patients (55.3%)
were female and 17 were male (44.7%) ([Table 1]). Thirty patients (78.9%) had a single CCM lesion; eight (21%) had multiple CCM
lesions. Twenty-two patients (57.9%) had CCMs located in supratentorial noneloquent
areas, 9 (23.7%) in supratentorial eloquent areas, 4 (10.5%) in the cerebellum, and
3 (7.9%) in the brainstem. Twenty-four CCM patients (63.1%) had no associated DVA;
in 10 patients (26.3%), associated DVA was compromised during surgery, and in 4 patients
(10.5%) the associated DVA was preserved. There were no significant differences in
age (p = 0.87), gender (p = 0.14), average CCM size (p = 0.30), number (p = 0.73), and average preoperative Karnofsky score (p = 0.39) between the groups. In the group with preserved DVA, significantly more brain
stem locations were seen (p = 0.13; [Table 1]).
Table 1
Preoperative characteristics
|
Group I
No DVA
(n = 24)
|
Group II
Impairment of DVA (n = 10)
|
Group III
Preservation of DVA (n = 4)
|
Total
(n = 38)
|
p
|
|
Age, mean ± SD, y
|
40.5 ± 19.5
|
36.7 ± 27.0
|
39.2 ± 20.7
|
39.1 ± 20.1
|
0.87
|
|
Gender, n patients (%)
|
0.14
|
|
Male
|
11 (45.8)
|
6 (60)
|
0 (0)
|
17 (44.7)
|
|
|
Female
|
13 (54.2)
|
4 (40)
|
4 (100)
|
21 (55.3)
|
|
|
Size, mean ± SD, mm
|
19.5 ± 11.5
|
17.3 ± 9.2
|
10.7 ± 2.6
|
18.0 ± 10.5
|
0.30
|
|
Number
|
0.73
|
|
Single, no. of patients (%)
|
18 (75)
|
9 (90)
|
3 (75)
|
30 (78.9)
|
|
|
Multiple, no. of patients (%)
|
6 (25)
|
1 (10)
|
1 (25)
|
8 (21.1)
|
|
|
Location, no. of patients (%)
|
.013[a]
|
|
Supratentorial noneloquent area
|
16 (66.7)
|
5 (50)
|
1 (25)
|
22 (57.9)
|
|
|
Supratentorial eloquent area
|
6 (25)
|
2 (20)
|
1 (25)
|
9 (23.7)
|
|
|
Cerebellum
|
1 (4.2)
|
3 (30)
|
0 (0)
|
4 (10.5)
|
|
|
Brainstem
|
1 (4.2)
|
0 (0)
|
2 (50)
|
3 (7.9)
|
|
|
Preoperative Karnofsky score, mean ± SD
|
91.25 ± 6.12
|
92 ± 4.2
|
87.5 ± 5.0
|
91.05 ± 5.6
|
0.39
|
Abbreviations: DVA, developmental venous anomaly; SD, standard deviation.
a Significant.
[Table 2] shows the postoperative results. There were no statistical significance between
the three groups for the patients who had additional neurologic deficits (p = 0.050). However, the subgroup analysis demonstrated that the group with preserved
associated DVA had more additional neurologic deficit than the group with compromised
associated DVA (75% versus 10%; p = 0.041). There were no significant differences between the groups without associated
DVA (group I) and impaired DVA (group II) (37.5% versus 10%; p = 0.215). There were no significant differences between the three groups concerning
the average postoperative Karnofsky score (88.33 ± 9.17, 92.0 ± 6.32, and 90.0 ± 8.16,
respectively; p = 0.51). Patients with preserved associated DVAs (group III) had the lowest volume
of postoperative peri-resectional edema (2.48 ± 1.48 cm3) compared with patients with impairment of the associated DVA (group II) (8.16 ± 3.78
cm3) and patients without associated DVA (group I) (8.90 ± 9.75 cm3). However, statistical analysis showed no significant difference between these groups
(p = 0.35).
Table 2
Postoperative outcome results
|
Group I)
No DVA
(n = 24)
|
Group II)
Impairment of DVA (n = 10)
|
Group III)
Preservation of DVA (n = 4)
|
p
|
|
Postoperative additional neurologic deficit, no. of patients (%)
|
0.05[a]
|
|
Presence of additional deficit
|
9 (37.5)
|
1 (10)
|
3 (75)
|
|
|
Absence of additional deficit
|
15 (62.5)
|
9 (90)
|
1 (25)
|
|
|
Postoperative discharge Karnofsky score, mean ± SD
|
88.33 ± 9.17
|
92,0 ± 6.32
|
90.0 ± 8.16
|
0.51
|
|
Peri-resectional edema volume, mean ± SD, cm3
|
8.90 ± 9.75
|
8.16 ± 3.78
|
2.48 ± 1.48
|
0.35
|
Abbreviations: DVA, developmental venous anomaly; SD, standard deviation.
a Subgroup analysis demonstrated significance between group (II) and (III) (p = 0.041) but no significance between group (I) and (II) (p = 0.215), (I) and (III) (p = 0.285).
The follow-up results are shown in [Table 3]. There was no significant difference between the groups for average follow-up Karnofsky
score (p = 0.94). For evaluation of seizure control, most of the patients had Engel class
I, and there was no significant difference between the groups as would be expected
(87.5%, 100%, 100%, respectively; p = 0.52). In comparison of headaches, the result showed no significance between the
groups in the average subjective pain score (2.50 ± 2.33, 2.00 ± 2.59, and 5.50 ± 3.53,
respectively; p = 0.24). Contrasted with postoperative additional neurologic deficits, at discharge
the results showed no significant difference between the groups in the follow-up additional
neurologic deficit (p = 0.12).
Table 3
Follow-up results
|
Group I
No DVA
(n = 24)
|
Group II
Impairment of DVA (n = 10)
|
Group III
Preservation of DVA (n = 4)
|
p
|
|
Seizure control, no. of patients(%)
|
0.52
|
|
Engel class I
|
7 (87.5)
|
9 (100)
|
2 (100)
|
|
|
Engel class III
|
1 (12.5)
|
0 (0)
|
0 (0)
|
|
|
Headache status (pain score), mean ± SD
|
2.50 ± 2.33
|
2.00 ± 2.59
|
5.50 ± 3.53
|
0.24
|
|
Follow-up additional neurologic deficit, n patients(%)
|
0.12
|
|
Presence of additional deficit
|
2 (25)
|
2 (22.2)
|
2 (100)
|
|
|
Absence of additional deficit
|
6 (75)
|
7 (77.8)
|
0 (0)
|
|
Abbreviation: DVA, developmental venous anomaly; SD, standard deviation.
To analyze the other factors that could affect the postoperative outcome in CCM patients,
we compared patients with and without postoperative additional neurologic deficits
([Table 4]). There was no significant difference between CCM size and postoperative peri-resectional
edema between the two groups (17.59 ± 9.37 versus 18.21 ± 11.31 mm; p = 0.86, 5.41 ± 5.36 versus 9.40 ± 9.09 cm3; p = 0.15, respectively). Concerning the postoperative Karnofsky score, there was a
significant difference between the group of patients with a postoperative additional
deficit and those without (81.54 ± 8.00 versus 93.60 ± 4.89; p < 0.0001). The overall preoperative Karnofsky score, postoperative discharge Karnofsky
score, and follow-up Karnofsky score are shown in [Fig. 3].
Fig. 3 Karnofsky score evaluated preoperatively, at the time of postoperative discharge,
and as long-term follow-up.
Table 4
Comparison between patients with and without additional neurologic deficits
|
Presence of additional neurologic deficit
(n = 13)
|
Absent of additional neurologic deficit
(n = 25)
|
p
|
|
Size, mean ± SD, mm
|
17.59 ± 9.37
|
18.21 ± 11.31
|
0.86
|
|
Peri-resectional edema volume, mean ± SD, cm3
|
5.41 ± 5.36
|
9.40 ± 9.09
|
0.15
|
|
Postoperative discharge Karnofsky score, mean ± SD
|
81.54 ± 8.00
|
93.60 ± 4.89
|
0.000[a]
|
|
Location, no. of patients (%)
|
0.001[b]
|
|
(1) Supratentorial noneloquent area
|
3 (13.6)
|
19 (86.4)
|
|
|
(2) Supratentorial eloquent area
|
6 (66.7)
|
3 (33.3)
|
|
|
(3) Cerebellum
|
1 (25)
|
3 (75)
|
|
|
(4) Brainstem
|
3 (100)
|
0 (0)
|
|
Abbreviations: SD, standard deviation.
a Significant.
b Significant with subgroup analysis demonstrated significance between group (I) and
(II) (p = 0.007), (1) and (4).
(p = 0.009) but no significance between (2) and (3) (p = 0.266), (2) and (4) (p = 0.515), (3) and (4) (p = 0.143).
Concerning the CCM location, there was a significant difference between the group
of patients with and without an additional neurologic deficit (p = 0.001). Furthermore, the subgroup analysis showed significance between patients
with a CCM in a supratentorial noneloquent area and patients with a CCM in a supratentorial
eloquent area or brainstem (p = 0.007 and p = 0.009, respectively).
Correlation analysis was performed between continuous variables including postoperative
Karnofsky score, peri-resectional edema, size, and headache. The results showed significant
correlation only between CCM size and peri-resectional edema (Pearson correlation = 0.697;
p < 0.0001).
Discussion
DVAs are currently considered benign vascular malformations; most are detected incidentally.
CCMs are the most often encountered etiology in cases of hemorrhage related to DVAs.
Therefore, some authors have advocated resection of the associated DVAs to reduce
the risk of recurrent CCMs.[16] Nonetheless, most recommendations suggest avoiding injury of the associated DVA.[6]
[17]
[18]
Our study, despite being small, suggests that the impairment of associated DVAs during
CCM resection has no significant effect on clinical outcome. There were no significant
differences between the group of patients with impaired DVA (group II) and patients
without associated DVA (group I) (p = 0.51, p = 0.215, p = 0.24, and p = 0.52, respectively) concerning functional status (Karnofsky score), additional
neurologic deficit, postoperative headache, and seizure control. The only factor showing
a direct effect on clinical outcome was CCM location. This is likely the explanation
for the poor results despite DVA preservation (group III) because in two patients
(50%) the resected CCM was located in the brainstem and in one patient (25%) in a
highly eloquent supratentorial area. The volume of the peri-resectional edema also
demonstrated no significant difference between the groups (p = 0.35). Therefore we concluded that the effect of DVA occlusion on brain parenchyma
during CCM resection was not different to that of CCM resection without DVA occlusion.
Some articles reported massive edema after complete or partial DVA occlusion, especially
in the cerebellar location.[14] However, we have no similar results. Three patients (30%) with cerebellar CCM and
intraoperative occlusion of the associated DVA had no postoperative problems. In our
opinion, DVAs associated with CCMs possibly differ from the symptomatic isolated DVA
that Pereira et al[19] described. The pathomechanism of symptomatic DVAs includes increased inflow into
the DVA that could lead to hemorrhage, restriction of outflow from the DVA leading
to venous congestion, and mechanical compression of the DVA. Nonetheless, CCMs associated
with DVAs are statistically more likely to present with hemorrhage than isolated CCMs
possibly related to increased blood flow. Therefore, this could possibly explain the
result in this study that we have found no significant effect on compromise of the
associated DVAs group. However, we believe it is better to preserve the associated
DVAs as the recommendation. Nevertheless, in CCMs encased by associated DVAs, hemosiderin
stain in the brain parenchyma that needs to be resected, DVA not detectable before
surgery or difficulties with DVA preservation due to surgical circumstances, DVA could
possibly be occluded without severe consequences.
Our study was limited by relatively small sample size and retrospective study design.
Therefore some patients had to be excluded due to lack of adequately comparable imaging
data. Particularly because of the limited number of patients with preserved DVA of
which 50% were located in the brainstem, a statistically relevant interpretation of
the data is not always possible. Furthermore, during the clinical follow-up, only
half of all patients could be interviewed, which further increased the difficulty
of the interpretation of the data. Another limitation was that we used only standard
MRI sequences to evaluate the effect of venous impairment that could be improved by
special MRI sequences including susceptibility-weighted imaging.
It would be premature to claim that a DVA adjacent to a CCM can be occluded without
relevant clinical or radiologically detectable complications. However, the obliteration
of CCM-associated DVAs seems to be less dangerous than expected because the DVA is
probably the major venous drainage of the CCM itself and not as relevant for the venous
drainage of the adjacent brain tissue. The total resection of the CCM will reduce
the influx and overall amount of local blood flow; therefore in many cases the quantity
of venous drainage could be reduced without relevant clinical impairment. Generally,
the neurosurgeon should still try to preserve an associated DVA in CCM surgery, but
the fear that the occlusion of a DVA will result in venous congestion, relevant edema,
or venous infarction and increase of surgical morbidity is not always justified.
Conclusion
The true pathophysiology of DVAs is still unclear. However, our results demonstrated
similar clinical outcome and radiographic parameters in patients with and without
intraoperative occlusion of the DVA during CCM resection. The dogmatic basic principle
to never occlude an associated DVA needs to be evaluated further, and in our opinion additional studies are required.
