Abstract
Objectives
To investigate the time sequence of brain magnetic resonance imaging findings of spontaneous intracranial hypotension.
Methods
We retrospectively reviewed the medical records and brain magnetic resonance imaging findings of consecutive patients with spontaneous intracranial hypotension hospitalized between January 2007 and December 2017. Patients were divided into quartiles based on intervals between initial spontaneous intracranial hypotension symptom onset and brain magnetic resonance imaging scan. Six categorical and five continuous brain magnetic resonance imaging findings were assessed, including venous distension sign, enlarged pituitary gland, diffuse pachymeningeal enhancement, mid-brain pons deformity, subdural fluid collection, flattening of pons, midbrain-pons angle, descent of cerebral aqueduct, mamillopontine distance, distance of suprasellar cistern, and distance of prepontine cistern. In addition, we also calculated the neuroimaging scores with a score ≥5 classified as ‘high probability of spontaneous intracranial hypotension' and a score ≥3 as ‘intermediate-to-high probability.' Then, we analyzed the linkage between the onset-neuroimaging interval and brain magnetic resonance imaging findings, as well as different neuroimaging scores.
Results
A total of 173 patients (57 males and 116 females) were included in the analysis, and the range of onset-neuroimaging interval was 1 to 89 days (median [interquartile range] = 17 [7 to 30 days]). We divided the patients into quartiles based on their onset-neuroimaging interval (the first quartile: 0–6 days; the second quartile: 7–16 days; the third quartile: 17–29 days; the fourth quartile: ≥30 days). Among brain magnetic resonance imaging findings, the incidence of venous distension sign was high (>75%), with no difference among quartiles (p = 0.876). The incidence of diffuse pachymeningeal enhancement (p = 0.001), severe midbrain-pons deformity (p = 0.001), and subdural fluid collection (<0.001) followed a significant stepwise increase from the first quartile to fourth quartile. Patients with shorter onset-neuroimaging intervals were less likely to have neuroimaging scores ≥5 (<17 vs. ≥17 days: 72.9% vs. 86.4%; odds ratio = 2.3 [95% CI 1.1–5.1], p = 0.028), but not neuroimaging scores ≥3 (<17 vs. ≥17 days: 92.9% vs. 92.0%, p = 0.824).
Conclusions
The emergence of brain magnetic resonance imaging findings of spontaneous intracranial hypotension depended on disease duration and appeared sequentially. When using brain magnetic resonance imaging findings or neuroimaging scores for diagnostic purposes, the onset–neuroimaging interval should be considered.
Introduction
Spontaneous intracranial hypotension (SIH) is a disabling headache that is caused by spinal cerebrospinal fluid (CSF) leakage. The classical clinical presentation is acute orthostatic headache, but one study showed orthostatic headache was present in only 77% of SIH patients (1–3). Moreover, accumulating evidence has shown that SIH can present with a wide variety of clinical symptoms, such as parkinsonism, conscious change, or even personality change (2–4). Brain and spinal neuroimaging are important for SIH patients with both diagnostic and outcome prediction values (4). However, brain magnetic resonance imaging (MRI) may be the only and initial neuroimaging performed in some patients with non-classical clinical presentations (i.e., cognitive symptoms or personality change). Therefore, the initial recognition largely depends on brain MRI findings in this situation. Currently, several brain MRI findings have been described in association with SIH (5–8). Additionally, the most updated third edition of the International Classification of Headache Disorders (ICHD-3) includes specific brain or spinal neuroimaging findings as diagnostic criteria of SIH; hence, direct measurement of CSF pressure through lumbar puncture is not considered mandatory in most situations (9,10). However, the emergence of brain MRI findings may differ with different disease durations. For example, although diffuse pachymeningeal enhancement has been considered as a hallmark brain MRI feature of SIH, our previous study found that approximately one-fourth of patients did not have this finding on their initial brain MRI (11). One recent study proposed a three-tier neuroimaging scoring system based on brain MRI findings with a high diagnostic accuracy (12). Nevertheless, whether disease duration could interfere with the diagnostic value of this neuroimaging scoring system remains unknown. The aims of present study were 1). to investigate the time sequence of the brain MRI findings of SIH, and 2). to analyze the accuracy of the neuroimaging scoring system in SIH patients in association with different disease durations (12).
Methods
We performed a retrospective study of patients who were diagnosed with SIH and hospitalized in the neurology ward of Taipei Veterans General Hospital (Taipei VGH), a tertiary medical center in Taiwan between January 2007 and December 2017. The onset-neuroimaging interval was defined as the time interval between the onset date of the first symptom of SIH and the date of brain MRI scanning.
Inclusion and exclusion criteria of SIH patients
The inclusion criteria were as follows: 1) confirmation of ICHD-3 criteria of SIH [7.2.3] (9), and 2) patients received post-contrast brain MRIs at our hospital before epidural blood patching for SIH or burr hole surgery due to complicated subdural hematoma. Patients without post-contrast brain MRIs or exact timing of onset of the first symptoms in their medical records were excluded from the study analyses.
Assessment of the brain MRI
Post-contrast (gadopentetate dimeglumine (Gd) 0.1 mmol/kg b.w.) brain MRIs were performed on 1.5-T or 3-T scanners, and details of the MRI protocol have been described in previous publications (13,14). Brain images were obtained in the transverse plane with T1- and T2-weighted spin-echo sequences. The section thickness was 1 mm, with an acquisition matrix of 256 and field of view (FOV) of 240 mm. The in-plane resolution of the acquired images was 0.937 × 0.937 mm2. All brain MRI were analyzed by board-certified neuroradiologists (STC and JFL) as described in our previous publications (8,13,15). In this study, we recorded the brain MRI findings as categorical or continuous variables. The categorical variables included enlarged pituitary gland (defined as pituitary height >7.2 mm in females and >5.25 mm in males based on previous literatures) (16,17), venous distention sign (defined as a convex bulging margin of the superior and inferior border of the mid-portion of the dominant transverse sinus on sagittal view) (18–20), subdural fluid collection (21), diffuse pachymeningeal enhancement (6,11), and flattening of pons (defined as ventral pons against the clivus with pons deformity caused by downward displacement of the brain structures) (4,5). The midbrain-pons angle was defined as the angle between the line tangential to the anterior margin of the midbrain and the line tangential to the superior margin of the pons on the sagittal midline of the MRI (13). According to our previous publication, midbrain-pons angle <40° predicted a poor treatment outcome; hence, we used midbrain-pons angle <40° as a cut-off value of midbrain-pons deformity (13). The continuous variables included midbrain-pons angle (22), descent of cerebral aqueduct (depicted as iter descent below the incisural line connecting the tuberculum sellae with the confluence of the vein of Galen into the straight sinus, mm) (6,23,24), mamillopontine distance (22), distance of suprasellar cistern (12), and distance of prepontine cistern (3).
Neuroimaging diagnostic scoring system
In this study, we also used a neuroimaging scoring system proposed by the Swiss research group for the diagnosis of SIH, which includes six brain MRI findings (12). Three of the brain MRI findings were weighted as 2 points, including pachymeningeal enhancement, engorgement of the venous sinus, and shortening of the suprasellar cistern's diameter (≤4.0 mm). Another three brain MRI findings were weighted as 1 point, including subdural fluid collection, effacement of the prepontine cistern (≤5.0 mm), and shortening of the mamillopontine distance (≤6.5 mm). According to this scoring system, the total score ranged from 0 to 9 points. Patients with a score ≥5 were classified as high probability and required prompt diagnostic investigations for SIH, and patients with an intermediate probability (score 3-4) required case-by-case discussion.
Statistics
All statistical analyses were conducted with IBM SPSS (version 22.0). For the categorical variables, the number and percentage of individuals within each category were presented. For the continuous variables, means were reported with standard deviations, or medians were reported with interquartile ranges. Differences between continuous variables were analyzed by t-test and analysis of variance (ANOVA) for normally distributed variables, and a Mann-Whitney U test was used to compare non-normally distributed continuous variables. The associations between onset-neuroimaging interval and brain MRI findings (or neuroimaging score based on brain MRI findings) were analyzed by Spearman's correlation analyses.
Considering a clear cut-off value was more helpful for clinical applications, we divided our patients into two groups based on the median of onset-neuroimaging interval. The differences of patients having two cut-off values of neuroimaging score (total score ≥5 and ≥3) between two groups of onset-neuroimaging intervals were analyzed by chi-square test. Also, we further divided the patients into quartiles of the onset-neuroimaging interval to observe the stepwise pattern of the emergence of brain MRI findings. The comparisons of patients having each brain MRI finding (or cut-off values of neuroimaging score) between quartiles of onset-neuroimaging intervals were analyzed by the linear-by-linear association. Bonferroni’s corrections for multiple comparisons were applied in the analyses of brain MRI findings, including six categorical variables (corrected for six pairwise comparisons: p = 0.05/6 tests = 0.0083), five continuous variables (corrected for five pairwise comparisons: p = 0.05/5 tests = 0.01), and two cut-off values of neuroimaging scores (corrected for two pairwise comparisons: p = 0.05/2 tests = 0.025).
Ethics
The study protocol was approved by the Institutional Review Board of Taipei Veterans General Hospital (2018-06-015BC).
Results
Patients
A total of 189 patients were admitted to Taipei VGH between January 2007 and December 2017. Among these 189 patients, a total of 16 patients were excluded from this study because of a lack of data about the date of first symptoms onset (n = 6) and no post-contrast brain MRI scan (n = 10). Thus, 173 patients (57 males and 116 females) were included in the analysis. There was no difference in age (mean [standard deviation] of included vs. excluded patients: 39.3 [9.6] vs. 42.3 [11.6], p = 0.251) or sex ratio (female sex [%] of included vs. excluded patients: 67.1% vs. 56.2%, p = 0.385) between the included and excluded patients. The demographics and clinical profiles of the included patients are summarized in Table 1. The median onset-neuroimaging interval was 17 days (interquartile range [IQR] 7 to 30 days), and the mean (standard deviation, SD) onset-neuroimaging interval was 20.8 (17.6) days.
Demographic data and brain MRI findings in SIH patients.
*Differences between normally distributed continuous variables were analyzed by t-test, and the differences between categorical variables were analyzed by chi-square test; p < 0.05 were considered statistically significant.
**Mann-Whitney U test was used to compare non-normally distributed continuous variables, and p < 0.05 were considered statistically significant.
Disease duration and neuroimaging findings
Among continuous variables, the midbrain-pons angle (Spearman r = −0.30, p < 0.001) and mamillopontine distance (Spearman r = −0.28, p < 0.001) were associated with the onset-neuroimaging interval. We divided the patients into quartiles based on their onset-neuroimaging interval. Then, we compared the percentages of patients with different categorical brain MRI findings in each quartile (Table 2) and the mean values of different continuous variables in each quartile (Table 3). Based on linear-by-linear association tests, there were no differences in the incidence of venous distension sign between each quartile, which had an incidence >75% in all quartiles (Table 2). Also, there was a stepwise increment in the frequencies of diffuse pachymeningeal enhancement (Q1 vs. Q2 vs. Q3 vs. Q4: 50.0% vs. 70.5% vs. 72.1% vs. 82.6%, p = 0.001), severe midbrain-pons deformity (Q1 vs. Q2 vs. Q3 vs. Q4: 19.0% vs. 32.6% vs. 42.9% vs. 52.2%, p = 0.001), and subdural fluid collection (Q1 vs. Q2 vs. Q3 vs. Q4: 2.4% vs. 9.3% vs. 26.2% vs. 47.8%, p < 0.001) in quartiles of onset-neuroimaging intervals, as shown in Table 2 and Figure 1.
The frequencies of brain neuroimaging findings in each quartile: categorical variables.
*The differences of these categorical variables between different quartiles were analyzed by linear-by-linear association.
**The differences were considered significant at Bonferroni-corrected threshold (corrected for six pairwise comparisons) of p = 0.05/6 tests= 0.0083.
***The differences were considered significant at Bonferroni-corrected threshold (corrected for two pairwise comparisons) of p = 0.05/2 tests = 0.025.
Brain neuroimaging findings (continuous variables) in each quartile.
*The differences of these continuous variables between different quartiles were analyzed by ANOVA
**The differences were considered significant at Bonferroni-corrected threshold (corrected for five pairwise comparisons) of p = 0.05/5 tests = 0.01.
The association between disease duration and brain MRI findings.
Application of the neuroimaging scoring system in patients with different onset-neuroimaging intervals
Based on the neuroimaging scoring system, a total score ≥5 was classified as ‘high probability of SIH' and a score ≥3 was classified as ‘intermediate-to-high risk' (12). In our SIH population, the onset-neuroimaging interval was correlated with the neuroimaging score (Spearman r = 0.27, p = 0.004). We divided the patients into quartiles based on their onset-neuroimaging interval and compared the percentage of patients with different cut-off value of neuroimaging score in each quartile. Based on linear-by-linear association tests, the percentage of patients with a neuroimaging score ≥5 was higher in those patients with longer onset-neuroimaging intervals, following a stepwise pattern (Q1 vs. Q2 vs. Q3 vs. Q4: 69.0% vs. 76.7% vs. 85.7% vs. 87.0%, p = 0.022). (Table 2) Patients with a shorter onset-neuroimaging interval (<17 days) had a lower probability of being classified as high risk (score ≥5) by neuroimaging score (<17 vs. ≥17 days: 72.9% vs. 86.4%, odds ratio = 2.3 [95% CI 1.1–5.1], p = 0.028). On the other hand, there was no difference in the probability of having a neuroimaging score ≥3 between onset-neuroimaging intervals <17 days and ≥17 days (<17 vs. ≥17 days: 92.9% vs. 92.0%, p = 0.824).
Discussion
Based on a sizable patient cohort with SIH, we provided a thorough analysis of the incidence of each brain MRI finding in patients with different disease durations. Venous distension sign has a >75% incidence regardless of disease duration and is common in patients with already <7 days of onset-neuroimaging intervals. From short to longer disease duration, the frequencies of other brain MRI findings have a stepwise increment as follows: diffuse pachymeningeal enhancement, midbrain-pons deformity, and subdural fluid collection.
Our previous publication proposed two compensatory mechanisms underlying the cerebral neuroimaging findings: cerebral venous dilation and brain deformity. Additionally, we hypothesized that brain deformity is dominant after the cerebral venous dilation reaches its elastic limit (13). Among brain MRI findings, only the prevalence of diffuse pachymeningeal enhancement was proven to be associated with the timing of brain MRI (11,25). Our previous study found that diffuse pachymeningeal enhancement may not be present in patients with short onset-neuroimaging intervals, which is compatible with the present study (11). Another study found that the prevalence of diffuse pachymeningeal enhancement is lower in the chronic stage of SIH (25). However, the latter study was based on a patient cohort with longer disease duration (20.5 months), different from the short disease duration population in our study (3 weeks on average) (25). In this study, we further explored the association between each brain MRI finding and disease duration. We found cerebral venous dilation-related brain MRI findings (i.e., enlarged pituitary gland, venous distension sign, and diffuse pachymeningeal enhancement) generally developed earlier than brain deformity-related brain MRI findings (i.e., flattening of pons and midbrain-pons deformity), which was consistent with our hypothesis that the compliance was better for cerebral veins than brain structures during spinal CSF leaks (13).
The present study has clinical implications. First, patients may present to clinicians with different disease durations in real-world practice. Our study provides the time sequence of brain MRI findings of SIH, which helps improve the accuracy of neuroimaging diagnosis in patients with different disease durations. In SIH, the venous distension sign has an incidence ≥75% in all quartiles of disease duration, indicating that it is an early and reliable sign for diagnosis (Figure 1). Our finding is in line with Canadian research, which demonstrated high sensitivity and specificity (>90%) of the venous distension sign (19). Second, the timing of MRI may interfere with the reliability of the diagnostic value of the three-tier neuroimaging score system. The original setting of the three-tier neuroimaging score system can classify the patients into high (score ≥5), intermediate (score 3–4), and low risk (score ≤2) (12). Patients classified as high risk should consider spinal neuroimaging or other investigations for diagnosing SIH promptly, and those with an intermediate probability require case-by-case discussion (12). In our cohort, SIH patients (confirmed by spinal neuroimaging) with onset-neuroimaging intervals <17 days had a less than 80% probability of being classified as high risk (score ≥5), which is not satisfactory for diagnostic purposes. The present study found that more than 92% of SIH patients had a neuroimaging score ≥3 regardless of the disease duration, and the original Swiss study showed a low false-positive rate (6.7%) with this cut-off value (12). Therefore, in the application of neuroimaging score for diagnostic purposes, prompt investigations for SIH should be considered in individuals to be classified as intermediate-to-high risk (score ≥3), especially in patients with disease duration <17 days.
Our study has limitations. First, SIH patients are heterogeneous, and the time sequences of different brain MRI findings may be different in patients with different CSF leakage patterns or severities. For example, patients with severe and intractable spinal CSF leaks may come from other hospitals after unsuccessful treatment. Therefore, brain MRI findings linked to longer disease duration may represent the disease severity, not disease duration. The best way to observe the time sequence of neuroimaging change is by performing serial brain MRI prospectively without interventions. However, this is impossible because of the intolerable symptoms and the risk of complications of SIH. Second, as our institution is a tertiary medical center, the patients with SIH may represent a relatively severe subgroup. On the other hand, our healthcare system allows patients to visit specialized headache clinics without a referral, which could help to reduce the potential patient spectrum bias in the present study. Third, our patients received brain MRI with two different scanners in this study. However, since all brain MRI findings were measurable and commonly used in daily practice. Hence, different MRI scanners might not seem to confound our results significantly.
Conclusions
Our study demonstrated that brain MRI findings were associated with the disease duration in patients with SIH. The presence of cerebral venous dilation-related MRI findings generally occurs earlier than brain descent-related signs. In the application of the neuroimaging scoring system and brain MRI findings for SIH diagnosis, the interval between symptom onset and brain neuroimaging should be considered.
Article highlights
The study findings showed that brain MRI findings were associated with the disease duration in patients with SIH. The presence of cerebral venous dilation-related MRI findings generally occurs earlier than brain descent-related findings. The interval between symptom onset and brain neuroimaging should be considered an important variable when using the neuroimaging scoring system on brain MRI findings for diagnosing SIH.
Footnotes
Acknowledgment
We wish to thank the participants of the American Academy of Neurology (AAN) 2019 Annual Meeting in Philadelphia for inspiring discussions during the meeting.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported in part by grants from Taipei Veterans General Hospital [V108C-105 and V110B-009], Ministry of Science and Technology of Taiwan
[MOST109-2314-B-075-002, MOST110-2314-B-075 -035 -MY2 and MOST 110-2314-B-075-081], Vivian W. Yen Neurological Foundation, and Brain Research Center, National Yang Ming Chiao Tung University from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan.
