Keywords
Computed tomography - Hounsfield unit - intracranial venous sinus thrombosis - magnetic
resonance imaging
Introduction
Cerebral venous sinus thrombosis (CVST) is predominantly a disease of young adults
(<50 years old), accounting for 1-2% cases of stroke.[1],[2] Most patients with CVST present with nonspecific signs and symptoms and are likely
to undergo nonenhanced head computed tomography (NCT) at presentation. However, NCT
may be normal in up to two-thirds of patients with venous sinus thrombosis.[3] In acute stages of thrombosis, the sensitivity of diagnosing CVST is higher as the
sinuses are hyperdense. The findings on NCT are often subtle and nonspecific in the
early stages which are evident as “hyperdense” venous sinuses and cerebral swelling.[3] Though magnetic resonance imaging (MRI) is considered the imaging modality of choice
for diagnosing CVST, it is not universally available in an acute setting.[4],[5] Moreover, it is also known for artefacts, flow-related void rendering it as a suboptimal
study for diagnosing CVST.
One alternative diagnostic modality could be CT venography (CTV), which is more readily
available compared to MRI. CTV is nearly as good as MRI or may be even better in some
cases.[6] With recent advancement in technology, CTV and MR venography (MRV) have now replaced
digital subtraction cerebral angiography as the diagnostic test of choice for CVST.[4],[7],[8],[9] Hyperdense sign of venous sinuses on NCT scan has a sensitivity of 84% and 95% and
specificity of 96% and 95% with a HU (Hounsfield unit) cut-off value of 65 and 62,
respectively, as shown by some recent studies.[10],[11] Measurement of H:H ratio (average HU vs hematocrit) in addition to venous sinus
attenuation is likely to increase sensitivity.[11] Complications such as venous infarcts and fragmented hemorrhage occur late during
the course of CVST, which are easily diagnosed on NCT; hence, diagnosing CVST at initial
stages remains challenging.
The primary aim of the present study was to evaluate whether increased attenuation
in cerebral venous sinuses in acute condition (less than 3 days) can be used to diagnose
acute CVST and to determine its diagnostic value. The secondary aim was to determine
whether there is possible correlation between HCT and HU.
Early detection of CVST is important as initiation of anticoagulation therapy is thought
to prevent the propagation of thrombus and subsequent venous infarcts and hemorrhage.
This in turn reduces the mortality and morbidity associated with long-term neurological
sequelae.[12]
Patients and Methods
This was a retrospective, observational study conducted between December 2013 and
December 2015 in a tertiary care hospital. The study involved two independent groups.
One group of patients with sinus thrombosis confirmed by MR venography (group A) who
underwent NCT. For group B (nonthrombotic/control), we retrospectively reviewed data
of NCT in 200 patients with complaints of headache and seizures whose study was normal.
Out of the 200 patients, those who underwent MRV but did not show any evidence of
venous sinus thrombosis were included in group B (control). The HU, hemoglobin (HGB),
hematocrit (HCT), and H:H (HU to HCT) ratio of both groups were recorded and compared.
Patients in the age group of 10–80 years of either sex who underwent MRV within 24
hours of NCT were included in both the groups. Only patients who had an acute history
(<3 days) and whose HGB and HCT reports were available within 24 hours of scan were
included in the study group. Patients with a history of trauma, intracranial surgery,
intracranial artefacts, and those who received contrast 24 hours prior to the scan
were excluded.
All patients were imaged on Philips brilliance 64 slice CT scanner. Images were acquired
by using CT parameters of 120 kV; section thickness, 5 mm (reconstruction images slice
thickness, 1 mm); 300 mA; and pitch, 0.675.
We retrospectively reviewed and analyzed the data through radiology information system
PACS (Picture Archiving and Communication System).
In group A patients, HU was measured in individual thrombotic segments. For patients
with multiple sinus thrombosis, average HU of all thrombosed sinuses (hyperdense segments)
on CT was taken for ease of comparison. Each segment is a single venous sinus which
go by the name SSS, transverse sinus, sigmoid sinus, and others.
In a previous study, Black et al. concluded that variation in the size of region of interest (ROI) does not have different
HU's.[13] We used the best fit circle for HU measurement in group B patients at the inferior
aspect of superior sagittal sinus (SSS) and at venous confluence in axial image [Figure 1]A and [Figure 1]B. The other cerebral venous sinuses in group A patients were also measured in axial
image, where it was best seen [Figure 1]C, [Figure 2]A and [Figure 2]B. Minimum, maximum, and average HU were noted. We also calculated the H: H ratio
for each patient to normalize observed blood attenuation with regard to hemoconcentration.
Figure 1 (A-C): A 23-year-old male patient presented with headache; plain CT scan
shows superior sagittal sinus thrombosis (HU, 77.38) in A. A is zoomed to show the
best fit ROI in SSS, which is showing thrombosis (B) and left transverse sinus thrombosis
(HU, 70) in C
Figure 2 (A and B): A 21-year-old male patient presented with altered sensorium. Unenhanced
CT scan of head shows thrombosis in right transverse sinus (HU, 74.5) and right sigmoid
sinus (HU, 75) in A and left transverse sinus (HU, 77.8) thrombosis in B
Statistical analysis
Baseline parameters such as age, sex ratio, HGB, HCT, and average HU, between patients
with and without thrombosis were compared using two-tailed unpaired t-test. To assess
the correlation between HGB and HCT with measured CT attenuation linear regression
analysis was used. Two-tailed unpaired t-test was used to compare H:H ratio between
patients with and without sinus thrombosis.
Results
After meeting inclusion and exclusion criteria, 36 patients (59 thrombotic sinuses)
were included in group A, and out of 200 patients, 40 were included in group B. Analysis
was carried out as shown in [Table 1].
Table 1
Comparison of baseline parameters, HU and H: H ratio values between both the Groups
|
Parameters
|
Group A (36)
|
Group B (40)
|
P
|
|
Avg age (range, SD)
|
31.7 (17-61, 10.7)
|
40.4 (10-78,19.2)
|
0.03
|
|
Male
|
30.2 (17-53, 8.5)
|
39.2 (12-70,17.8)
|
|
|
Female
|
37 (22-61, 16.1)
|
41.8 (10-78, 21.1)
|
|
|
Sex (M: F)
|
28:8
|
22:18
|
0.03
|
|
HGB (range, SD)
|
14.8 (8.7-20.5, 2.7)
|
12.6 (6.7-18.6, 2.3)
|
0.00
|
|
Male
|
15.7 (8.9-20.5, 2.2)
|
13.1 (6.7-18.6, 2.6)
|
|
|
Female
|
11.8 (8.7-15.3, 2.18)
|
12 (7.5-14.7, 1.66)
|
|
|
HCT (range, SD)
|
46.2 (27.4-61.3, 8.2)
|
38.8 (19.1-55.9, 7)
|
0.00
|
|
Male
|
49 (31.9-61.3, 6.18)
|
40.1 (19.1-55.9, 8.4)
|
|
|
Female
|
36.4 (27.4-46.8, 7)
|
37.3 (27.2-47.8, 4.8)
|
|
|
HU (range, SD)
|
73.7 (62.3-80.7, 3.8)
|
48.6 (31.2-64.1, 7.7)
|
0.00
|
|
Male
|
74.7 (69.6-80.7, 3.3)
|
49.2 (31.2-61.8, 8.2)
|
|
|
Female
|
70.4 (62.3-75.9, 3.97)
|
47.9 (33-64.1, 7.29)
|
|
|
H:H ratio (range, SD)
|
1.64 (1.24-2.41, 0.29)
|
1.27 (0.95-2.38, 0.23)
|
0.00
|
|
Male
|
1.54 (1.24-2.18, 0.19)
|
1.26 (0.95-2.38, 0.29)
|
|
|
Female
|
1.98 (1.51-2.41, 0.34)
|
1.28 (1.01-1.5, 0.12)
|
|
In patients without sinus thrombosis (group B), the age group ranged from 10 to 78
years (average 40.4 years and SD 19.2). Average age of males was 39.2 years and that
of females was 41.8 years. In patients with sinus thrombosis (group A), age group
ranged from 17 to 61 years (average 31.7 years and SD 10.7). Average age of male patients
was 30.2 years whereas in female patients it was 37 years.
HU values were higher in group A patients (73.7 vs 48.6) ranging from 62.3 to 80.7.
Except for 3 patients in group A, all patients had HU values above 70. HU values in
group B ranged from 31.2 to 64.1 [Figure 3]. ROC curve analysis of HU shows an area under the curve of 1.0 with cut-off of 70
as well as sensitivity of 92% and specificity of 100%.
Figure 3: Scatterplot of sinus attenuation (HU average) vs HCT comparing both groups
(○ = Group A, □ = Group B)
Average hematocrit value was higher in group A patients, 46.2 (SD 8.2) whereas it
was 38.8 (SD 7) in group B patients. H:H ratio in patients without thrombosis ranged
from 0.95 to 2.38 (mean 1.27), and in patients with thrombosis it ranged from 1.24
to 2.41 (mean 1.64).
There was statistically significant difference between both the groups in terms of
age, sex, HGB, HCT, average HU, and H:H ratio (P < 0.05). Linear regression analysis showed positive correlation between HGB and HCT
with CT attenuation (HU) among both the groups (P < 0.05).
Discussion
Hyperattenuation of cerebral sinuses is particularly seen in acute stages of thrombosis.
The increase in attenuation occurs due to clot retraction, elimination of water, and
increased concentration of red blood cells (RBC) and hemoglobin, thus resulting in
high HU values. As time passes, RBC and hemoglobin gets degraded, and hence, the attenuation
gradually decreases over several days.[14] Therefore, there is normalization of HU values in subacute and chronic stages of
CVST.
In our study, we included only those patients who had acute symptoms (less than 3
days) and measured HU density in venous sinuses. Average HU value in our study was
73.7 (SD 3.8) in group A and 48.6 (SD 7.7) in group B. To our knowledge, no study
have been done to provide a definitive cut-off value of HU for CVST on NCT. Black
et al. considered 70 HU as a cut-off for suspecting thrombosis, but they also concluded
that further studies needed to confirm the diagnosis.[13] In our study, we observed that CT HU of >70 confirms thrombosis in group A patients
compared to group B whose HU values were less than 64.1 (average 48.6).
As per our study, HU >70 is highly valuable for diagnosing thrombosis (sensitivity
92% and specificity 100%), and further investigation is not required to confirm the
diagnosis. Similarly, HU value of <64 is very specific in ruling out thrombosis. Drawback
of these HU value occurs when it is in the intermittent range, which happens as the
stage of thrombosis moves to subacute or chronic phase. We may misinterpret HU values
as normal, and in such cases, it is advisable to perform venography.
Recent studies showed sensitivity (84% and 95%) and specificity (96% and 95%) with
an HU cut-off value of 65 and 62, respectively.[10],[11] Black et al. noted that patients with CVST often had a HU of >70 and those without had a HU of
<70; H: H ratio of >2 also showed a good correlation with CVST. They also demonstrated
a correlation between the patient's HCT and venous sinus density (HU).[13] In a recent study by Ali Alsafi et al., 768 venous sinuses were analyzed; 46 with CVST and 720 without. The average HU
of vessels containing a thrombus was 68 ± 1.56, which was significantly higher than
that of normal vessels 52 ± 0.28.[15]
In our study, H:H ratio showed a value of 1.64 ± 0.29 in patients with thrombosis
and 1.27 ± 0.23 in patients without thrombosis. Black et al. found mean H: H values of 2.20 in patients with CVST and 1.44 in patients without
CVST.[13] Buyck et al. study also showed higher mean H:H value in patients with acute CVST (1.91 vs 1.33).[11] Compared to other studies, a little overlap of H:H ratio values were seen between
the groups in our study [Figure 4],[11],[13] whereas HU values showed almost no overlap between the groups [Figure 5].
Figure 4: Box plot showing average H:H ratio values compared between both the groups
Figure 5: Box plot showing average sinus attenuation values (HU) compared between
both the groups
ROC curve analysis of H:H ratio showed an area of 0.90 and with an optimum cut-off
of 1.40 for sensitivity of 86% and specificity of 82.5%. A threshold value of 1.50
gives a high specificity of 95% but with low sensitivity (64%). A study by Besachio
et al. reported a higher H: H ratio value of 1.7 for similar sensitivity and specificity.[10] An H: H ratio of more than 1.52 suggests a strong likelihood of a clot (sensitivity
95%, specificity 100%), as recommended by Buyck et al.[11] Black et al. suggested a threshold value of 1.8 to suspect the presence of thrombosis.[13]
In most centres in an acute setting, clinician prefers plain CT of the head. Early
detection and initiation of treatment is mandatory in these cases. MRI is time consuming
and moreover patient cooperation is important. Thus, if acute CVST is diagnosed accurately
in plain CT scan itself, we can save valuable time and start the required treatment
at the earliest which can further reduce the complications or disease progression.
As the age of thrombosis increases, and if the treatment is not started early, there
is a high chance of extension of thrombus across multiple venous sinuses and venous
hemorrhagic infarct may also occur. This leads to significant morbidity to the patient.
Limitations of the study
This was a retrospective study, and the sample size in the thrombotic group was less.
However, the number of patients were more compared to several previous studies. No
data is available regarding hydration and medication status of patients. Interobserver
variability in measuring ROI was not calculated.
Conclusion
In case of an acute setting, plain CT is specific for the diagnosis of acute CVST,
but as the age of the clot progresses, density of venous sinuses reduces. Based on
our study, we recommend 70 HU as the cut-off value (sensitivity 92%, specificity 100%)
for diagnosis of CVST, and there is no need for further confirmatory studies such
as venography.
It is important to note that a normal HU value does not exclude the presence of thrombosis
in venous sinuses. In patients with possible risk factors for thrombosis or unexplained
symptoms, further conclusive imaging studies such as CTV or MRI should be done despite
a negative head CT.
