Prevalence of subarachnoid haemorrhage among patients with cranial venous sinus thrombosis in the presence and absence of venous infarcts

Abstract

Introduction

In patients with cranial venous sinus thrombosis, the occurrence of subarachnoid haemorrhage in association with haemorrhagic venous infarcts is a well described phenomenon. However, the presence of subarachnoid haemorrhage in patients with cranial venous sinus thrombosis in the absence of a haemorrhagic venous infarct is exceedingly rare.

Methods

We retrospectively reviewed charts and scans of all patients who had cranial venous sinus thrombosis confirmed by magnetic resonance venography at our hospital between September 2004 and May 2015. The presence of subarachnoid haemorrhage was ascertained on fluid-attenuated inversion recovery, susceptibility-weighted imaging and/or unenhanced computed tomography scans by a single experienced neuroradiologist. Statistical analysis was performed using the Statistical Package for Social Sciences version 20. Differences in the proportion of haemorrhagic venous infarcts among patients with subarachnoid haemorrhage versus those without subarachnoid haemorrhage were compared using the chi-square test. A P value of less than 0.05 was considered significant.

Results

A total of 138 patients who had cranial venous sinus thrombosis were included in the study. Seventy-three (52.9%) were women and the median age of subjects was 35 (interquartile range 22–47) years. Venous infarcts and haemorrhagic venous infarcts were noted in 20/138 (14.5%) and 62/138 (44.9%) cases, respectively. Subarachnoid haemorrhage was present in 15/138 (10.9%) cases and, in three cases, subarachnoid haemorrhage occurred in the absence of a venous infarct. Haemorrhagic venous infarcts were more prevalent (P = 0.021) among patients with subarachnoid haemorrhage (11/15) than in those without subarachnoid haemorrhage (51/123).

Conclusion

In patients with cranial venous sinus thrombosis, subarachnoid haemorrhage can occur even in the absence of a haemorrhagic venous infarct. The recognition of cranial venous sinus thrombosis as the underlying cause of subarachnoid haemorrhage is important to avoid misdiagnosis and inappropriate management.

Introduction

Cranial venous sinus thrombosis (CVST) accounts for 1% of all cases of stroke.¹ While this disease can affect patients of any age, young and middle-aged women are most commonly affected.² Most patients are found to have a pro-thrombotic state (such as an inherited thrombophilia) or a risk factor for thrombosis (the most common being oral contraceptive pills).³ Clinical manifestations are often varied and previous studies have reported headache, vomiting, altered mental status, visual disturbances, seizures or focal neurological deficits as the notable clinical features.⁴ These signs and symptoms result from impaired cranial venous drainage due to CVST, which leads to raised intracranial pressure and impaired reabsorption of cerebrospinal fluid into the superior sagittal sinus.⁵ In some cases, CVST can lead to venous infarction of the cerebral parenchyma, which often develops haemorrhagic transformation.^6,7

Non-traumatic subarachnoid haemorrhage (SAH) is secondary to rupture of an intracranial aneurysm in 85% of cases.⁸ Sudden onset of a severe headache (‘thunderclap headache’) is the presenting clinical feature in most cases.⁹ If not diagnosed and managed promptly, patients can rapidly progress to develop drowsiness, stupor and coma.¹⁰ Aneurysmal SAH portends a worse prognosis and population-based studies have reported mortality rates to be as high as 45%.¹¹ Apart from aneurysmal rupture, other reported causes of non-traumatic SAH include arteriovenous malformations, intracranial vascular dissection, intracranial vasculitis, dural venous fistula, metastatic disease and bleeding diathesis.^12,13 CVST as a cause of SAH has also been reported in the literature, although it usually occurs in conjunction with the presence of a haemorrhagic venous infarct (HVI).^14,15

The occurrence of SAH in patients with CVST in the absence of venous infarcts (VIs) has been rarely reported.¹⁴ The importance of recognising CVST as the underlying cause of SAH is that its management is radically different from that of aneurysmal SAH. CVST is a potentially treatable condition and requires prompt anticoagulation to prevent long-term neurological sequelae.^3,16 On the other hand, anticoagulation in cases of aneurysmal SAH can have disastrous consequences.⁸ From a theoretical standpoint, CVST can lead to the development of VI, which may undergo haemorrhagic transformation and progress to SAH.¹⁷ However, the occurrence of SAH in patients with CVST in the absence of VI remains an obscure entity.¹⁸ In the present study, our aim was to determine the frequency of SAH in patients with diagnosed CVST, both in the presence and absence of HVI.

Methods

Our hospital is a 522-bedded tertiary care centre located in the city of Karachi (Pakistan) with an estimated population of almost 27.5 million. A retrospective cross-sectional study was performed after obtaining exemption from formal approval by the institutional ethics review committee. We retrieved all magnetic resonance venography (MRV) reports from the institutional radiology information system containing the words ‘dural sinus thrombosis’, ‘cerebral venous sinus thrombosis’, ‘cranial sinus thrombosis’ or ‘cranial venous sinus thrombosis’. Medical records for all patients with a diagnosis of CVST confirmed by MRV between September 2004 and May 2015 were retrieved. Patients with untraceable medical records or those who were not managed at our institution were excluded from the study. We also intended to exclude patients with CVST who had an alternative diagnosis as the underlying cause of SAH (such as a ruptured aneurysm), although we did not encounter any such cases. Using a pre-designed, structured pro forma, each patient’s chart was systematically reviewed and data relating to demographics, clinical features and subsequent work-up were collected. Personal identifiers or other identifiable information was not recorded. All scans were de-identified prior to re-interpretation for the purpose of this study.

All patients included in the study had undergone MRI and MRV scans on a 1.5 Tesla MRI scanner (MAGNETOM Avanto; Siemens AG, Munich, Germany) with a dedicated head coil. Gadolinium-based contrast (Magnevist; Bayer AG, Leverkusen, Germany) in a dose of 0.5 ml/kg was used for performing contrast-enhanced MRV scans. MRV sequences were performed with the following parameters: TR/TE: 3.48/1.22, field of view: 20 cm × 20 cm, matrix: 320 × 128, bandwidth: 380 Hz/pixel and slice thickness: 1.1 mm. A consultant neuroradiologist with more than 5 years of experience was responsible for re-interpreting the scans and he was blinded to the actual report of the scans at the time of interpretation. The presence of VI, HVI or SAH was specifically noted on MRI scans. For the purpose of this study, SAH was diagnosed using two MRI sequences, i.e. fluid-attenuated inversion recovery (FLAIR) and susceptibility-weighted imaging (SWI) sequences. Hyperintense signals in the subarachnoid space on FLAIR sequences along with corresponding signal dropouts on SWI sequences were considered positive for SAH. CVST was diagnosed on contrast-enhanced MRV images as filling defects in the dural venous sinuses and/or cortical or deep veins provided such filling defects were not accounted for by arachnoid granulations or hypoplastic sinuses. Moreover, the site and extent of CVST on each MRV scan was also recorded for each patient. If a patient had undergone head computed tomography (CT) during the course of hospitalisation, these scans were also interpreted for the presence of SAH and other associated findings. SAH on CT scans was defined as the presence of high attenuation streaks within the subarachnoid space. Cohen’s kappa (κ) statistic was calculated as a measure of agreement between the findings reported in the official MRV report and the findings noted by the neuroradiologist re-interpreting scans (for the purpose of this study).

The Statistical Package for Social Sciences version 20.0 was used for performing statistical analysis. Frequencies were calculated for qualitative variables, while medians (interquartile range) were computed for quantitative variables. For all proportions, 95% confidence intervals (CIs) were also computed. Chi-square (χ²) or Fisher’s exact test was used for comparison of proportions. The Pearson product–moment correlation coefficient was computed to determine the correlation between SAH and VI or HVI. For all comparisons, a P value of less than 0.05 was considered statistically significant.

Results

Our initial search of the institutional radiology information system identified 144 patients with CVST on MRV. Of these, four were excluded as they were managed at another hospital and only underwent follow-up imaging at our centre. Another two were excluded due to untraceable records and missing data. This left a total of 138 patients for inclusion in the final analysis. Among these, 73/138 (52.9%, 95% CI 44.4–61.4%) were women and 65/138 (47.1%, 95% CI 38.6–55.6%) were men. The median age of study subjects was 35 (interquartile range 22–47) years. The most common clinical presentations were headache (n = 47/138, 34.1%, 95% CI 26.0–42.2%), seizure (n = 43/138, 31.2%, 95% CI 23.3–39.1%) and motor weakness (n = 29/138, 21%, 95% CI 14.1–27.9%) as given in Table 1. The most common underlying cause of CVST was malignancy (n = 21/138, 15.2%, 95% CI 9.1–21.3%) followed by meningoencephalitis (n = 17/138, 12.3%, 95% CI 6.7–17.9%) and post-partum status (n = 16/138, 11.6%, 95% CI 6.1–17.1%). In another 52/138 (37.7%, 95% CI 29.4–46.0%) patients, no underlying cause could be identified (Table 1). The superior sagittal sinus (n = 86/138, 62.3%, 95% CI 54.0–70.6%) was the most commonly thrombosed dural sinus followed by the transverse (n = 82/138, 59.4%, 95% CI 51.0–67.8%) and sigmoid (n = 62/138, 44.9%, 95% CI 36.4–53.4%) sinuses. The extension of sigmoid sinus thrombosis into the internal jugular vein was noted in 24/138 (17.4%, 95% CI 10.9–23.9%) cases. Cortical veins and deep veins were thrombosed in 12/138 (8.7%, 95% CI 3.9–13.5%) and 9/138 (6.5%, 95% CI 2.3–10.7%) patients, respectively (Table 1).

Table 1.

Descriptive characteristics of patients^a included in our study (n = 138).

Variables	Values, n (%)	95% CI
Sex
Women	73/138 (52.9%)	44.4–61.4%
Men	65/138 (47.1%)	38.6–55.6%
Clinical presentation
Headache	47/138 (34.1%)	26.0–42.2%
Seizure	43/138 (31.2%)	23.3–39.1%
Motor weakness	29/138 (21.0%)	14.1–27.9%
Vomiting	28/138 (20.3%)	13.5–27.1%
Altered consciousness	25/138 (18.1%)	11.5–24.7%
Fever	18/138 (13.1%)	7.4–18.8%
Decreased vision	10/138 (7.2%)	2.8–11.6%
Predisposing factor
None	52/138 (37.7%)	29.4–46.0%
Malignancy	21/138 (15.2%)	9.1–21.3%
Infection	17/138 (12.3%)	6.7–17.9%
Postpartum	16/138 (11.6%)	6.1–17.1%
CNS or inner ear infection	13/138 (9.4%)	4.4–14.4%
Trauma	9/138 (6.5%)	2.3–10.7%
Auto-immune disease	4/138 (2.9%)	0.0–5.8%
Other	6/138 (4.3%)	0.8–7.8%
Sinuses thrombosedb
Superior sagittal sinus	86/138 (62.3%)	54.0–70.6%
Transverse sinus	82/138 (59.4%)	51.0–67.8%
Sigmoid sinus	62/138 (44.9%)	36.4–53.4%
Internal jugular vein	24/138 (17.4%)	10.9–23.9%
Straight sinus	20/138 (14.5%)	8.5–20.5%
Cortical veins	12/138 (8.7%)	3.9–13.5%
Deep veins	9/138 (6.5%)	2.3–10.7%

CI: confidence interval; CNS: central nervous system.

^aMedian age of included patients was 35 years (interquartile range 22–47 years).

^bSome patients had more than one venous sinus thrombosed.

VI, HVI and SAH were noted in 20/138 (14.5%, 95% CI 8.5–20.5%), 62/138 (44.9%, 95% CI 36.4–53.4%) and 15/138 (10.9%, 95% CI 5.6–16.2%) patients, respectively. Figures 1, 2 and 3 demonstrate examples of VI, HVI and SAH. None of these findings were present in the other 56/138 (40.6%, 95% CI 32.2–49.0%) patients. There was strong agreement between the findings reported in the official MRV report and findings noted by the neuroradiologist re-interpreting scans for the purpose of this study (κ = 0.985 for VI, κ = 1.000 for HVI and κ = 0.961 for SAH, respectively). Reports of CT head were available for 93/138 (67.4%) patients in total. Of the 15/138 (10.9%, 95% CI 5.6–16.2%) patients with SAH noted on MRI, all had undergone CT head and the presence of SAH on CT was confirmed in 14/15 (93.3%, 95% CI 89.0–97.6%) patients. SAH was not noted in any of the remaining 79/138 (57.2%, 95% CI 48.8–65.6%) CT scans. The Pearson product–moment coefficient revealed a statistically significant, positive correlation between SAH and HVI (P = 0.010, r = 0.198). The proportion of patients with SAH who had HVI (11/15, 73.3%, 95% CI 65.8–80.8%) was significantly higher (P = 0.021) than the proportion of patients without SAH who had HVI (51/123, 41.5%, 95% CI 33.1–49.9%) as given in Table 2. This is depicted more clearly with a Venn diagram in Figure 4. Among patients with SAH (15/138), 3/15 (20%, 95% CI 13.2–26.8%) had no evidence of a VI or HVI. In all these cases, SAH was limited to the cortical convexity with sparing of the basal cisterns.

Figure 1.

(a) Axial and (b) maximum intensity projection post-contrast images of magnetic resonance venography demonstrate a filling defect in the straight sinus consistent with thrombosis. (c) Coronal fluid-attenuated inversion recovery images show diffuse symmetrical hyperintense signals in bilateral basal ganglia and thalamic regions. (d) Diffusion-weighted imaging and (e) apparent diffusion coefficient mapped axial images reveal diffusion restriction in the corresponding areas consistent with acute venous infarcts.

Figure 2.

(a) Axial and (b) maximum intensity projection post-contrast images of magnetic resonance venography demonstrate a filling defect in the superior sagittal sinus consistent with thrombosis. (c) Coronal fluid-attenuated inversion recovery and (d) axial T2-weighted images show diffuse, heterogeneous, predominantly hyperintense signals in bilateral parietal lobes. (e) Susceptibility-weighted images show signal dropout in the corresponding areas consistent with bilateral haemorrhagic venous infarcts.

Figure 3.

(a) Axial and (b) maximum intensity projection post-contrast images of magnetic resonance venography reveal a filling defect in the superior sagittal sinus consistent with dural sinus thrombosis. (c) Coronal fluid-attenuated inversion recovery images show subtle hyperintensities in the left parietal lobe sulci representing subarachnoid haemorrhage, which was confirmed on (d) susceptibility-weighted imaging. (e) No subarachnoid haemorrhage was noted involving the basal cisterns.

Table 2.

Frequency of subarachnoid haemorrhage, venous infarct and haemorrhagic venous infarct on magnetic resonance venography among study subjects (n = 138).

Type of infarct	Subarachnoid haemorrhagea		Total
	Present	Absent
Venous infarct	1	19	20
Haemorrhagic venous infarct	11	51	62
No infarct	3	53	56
Total	15	123	138

^aSubarachnoid haemorrhage on magnetic resonance imaging was defined as the presence of hyperintense signals in the subarachnoid space on fluid-attenuated inversion recovery sequences along with corresponding signal dropouts on susceptibility-weighted imaging sequences.

Figure 4.

A Venn diagram depicting the relative proportion of patients with venous infarcts, haemorrhagic venous infarcts and subarachnoid haemorrhage (SAH).

Discussion

In this retrospective cross-sectional study, we systematically reviewed scans of patients with CVST for evidence of SAH and found that SAH was noted in 10.9% of patients. Furthermore, we noted that 44.9% of patients had a HVI and another 14.5% had VI without evidence of haemorrhagic transformation. These results are of considerable interest and may suggest that SAH in conjunction with CVST has been underestimated in previous reports published in the literature.^14,17,18 In a report from India, Panda and colleagues retrospectively reviewed the records of 233 patients who were diagnosed with CVST at their centre and found radiological evidence of SAH in 10 cases (i.e. 4.3%).¹⁹ On the other hand, we found a higher frequency of SAH in our study (i.e. 10.9%). One reason for this observation may be that in our study, all scans were reviewed and re-interpreted systematically to look for evidence of SAH. Subtle SAH in cases of CVST may often be overlooked.^17,18

Most previously published literature on SAH in conjunction with CVST is limited to a few small case series. Kato et al. from Japan described the case of a 52-year-old woman who presented with severe occipital headache and was found to have CVST involving the superior sagittal, straight, transverse and bilateral sigmoid sinuses.¹⁴ Her CT showed evidence of SAH involving bilateral cerebellar sulci and right temporal sulcus. In another report by Jaiser et al. from the United Kingdom, a 53-year-old woman presented with occipital headache and was found to have evidence of small left frontal SAH on unenhanced CT.¹⁸ CT angiography and MRV confirmed a diagnosis of CVST (in the absence of a VI) and excluded intracranial aneurysm. In our study, headache was the presenting feature in 47 (34.1%) cases and malignancy (n = 21, 15.2%) was the most common predisposing factor for thrombosis. No obvious pro-thrombotic state could be identified in 52 (37.7%) cases in our study.

A number of hypotheses have been proposed to explain the occurrence of SAH in conjunction with CVST. The most plausible explanation is that SAH in the setting of CVST can occur due to haemorrhagic transformation of a venous cerebral infarction.²⁰ From the infarcted parenchyma, red blood cells may leak into perivascular Virchow–Robin spaces and pool up in the subarachnoid spaces between cortical gyri. However, in our study, three patients with SAH did not have any evidence of VI on radiological imaging. Likewise, in the series of Panda et al., nine patients with CVST and SAH had no radiological evidence of VI.¹⁹ One reason may be that the evolution of a VI requires time and may not be evident on radiological imaging early in the process. A better explanation may be that SAH in conjunction with CVST is due to an alternative pathophysiological mechanism. Kato and colleagues suggested that CVST can lead to the development of venous hypertension, which may lead to rupture of fragile cortical veins.¹⁴ As cortical veins are valveless and lack smooth muscle in their tunica media, venous hypertension can lead to rupture of these veins with leakage of blood into the subarachnoid space at points where these veins enter into the dural venous sinuses.^21,22

The published literature on CVST and SAH suggests that the distribution of SAH may provide a clue to the underlying aetiology of SAH. In a report of four cases, Oppenheim et al. reported that SAH associated with CVST involved only the cortical sulci and spared the basal cisterns.¹⁹ Chang and Friedman described three cases of isolated cortical vein thrombosis involving the vein of Trolard and all patients had SAH limited to the adjacent cortical sulci with sparing of basal cisterns.²³ This observation was also noted in the case series of Panda et al., who suggested that localised convexity SAH sparing the skull base may be an early sign of CVST.¹⁹ In our study, we noted that patients with thrombosis of superior sagittal, transverse and sigmoid sinuses had VI or HVI in the cerebral or cerebellar hemispheres. On the other hand, thrombosis of straight sinus or vein of Galen caused HVI in basal ganglia and thalamus. Among patients with SAH who had VI or HVI, SAH was limited to the vicinity of HVI. In patients with HVI of the basal ganglia or thalamus, SAH could involve the skull base and extend beyond the sulci of the cerebellar convexities. However, consistent with previous observations, SAH in cases without VI spared the basal cisterns and was limited to the cortical convexity. In all such cases, thrombosis involved the deep cortical veins, which may suggest that SAH was caused by pressure changes in the fragile microvasculature of the central nervous system.

Differentiation between SAH caused by CVST and SAH due to other causes is of considerable clinical importance. CVST is a potentially treatable condition, although delays in diagnosis and management have been shown to lead to a poor neurological outcome.²⁴ Prompt anticoagulation with either unfractionated heparin or low molecular weight heparin is recommended to prevent further clot propagation.²⁵ Recent studies have also explored the role of thrombolysis and endovascular therapies (such as thrombectomy) to reduce clot burden and hasten recovery.^26,27 Notable here is the fact that the presence of HVI or SAH in the setting of CVST is not a contraindication to anticoagulation as haemorrhage in such scenarios is caused by venous hypertension and is expected to improve with anticoagulation.²⁸ On the other hand, anticoagulation is detrimental in cases of SAH due to other causes (such as aneurysmal bleeding) and may lead to a fatal outcome.⁸ Therefore, the association between SAH and CVST needs to be recognised by clinicians and radiologists alike.

The limitations of this study also merit attention. First, this study was a retrospective cross-sectional review of patients at a single tertiary care centre and this may have overestimated the prevalence of SAH in patients with CVST. Second, our study included only Asian patients and we cannot generalise these findings to patients of other ethnicities. Third, conventional cerebral angiography was not performed in all patients, although most patients had undergone either CT angiography or magnetic resonance arteriography to exclude an aneurysmal SAH. In addition, all patients included in our study had undergone contrast-enhanced MRV. In clinical practice, a small (but significant) number of patients have contraindications to gadolinium-based contrast media and can only undergo MRV without contrast. Whether the results of our study can be generalised to those patients remains uncertain. Finally, a single neuroradiologist with relevant expertise in the field was responsible for re-interpreting scans for all patients included in this study. This precluded an accurate assessment of the inter-rater reliability of these findings.

Conclusion

SAH occurs frequently in patients with CVST, especially in patients who develop VI or HVI. However, SAH may occur in conjunction with CVST even in the absence of a VI. The presence of a cerebral convexity SAH with sparing of the basal cisterns may be a useful differentiating sign in such cases.

Ethics

This retrospective study was granted exemption from formal approval by the institutional ethics review committee.

Funding

No funding was obtained or sought for this study.

Conflict of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.