Abstract
Background
Persistent post-traumatic headache (PTH) is frequent, and intracranial hypotension may be an important cause of PTH. The present study aimed to examine whether magnetic resonance imaging (MRI) signs of intracranial hypotension are more frequent in people with persistent PTH than in healthy controls (HCs).
Methods
Adults with persistent PTH attributed to mild traumatic brain injury and age- and sex-matched HCs attended a single non-contrast, brain MRI 3T session. Fluid attenuated inversion recovery T1- and T2-weighted sequences were acquired to assign a modified Bern score. The score ranges from 0 to 9 points, with higher scores indicating a greater probability of cerebrospinal fluid (CSF) leakage leading to intracranial hypotension. The primary outcome was the difference in modified Bern score between participants with persistent PTH and HCs. All images were examined by a certified neuroradiologist who was blinded to the group status.
Results
Imaging data from 97 participants with persistent PTH and 96 age- and sex-matched HCs were eligible for analyses. A modified Bern score of ≤2 was present in 90 (93%) participants with persistent PTH and 85 (89%) HCs, indicating a low probability of CSF leak. None of the persistent PTH participants or the HCs had a score of >4. There were no significant differences in modified Bern scores between participants with persistent PTH and HCs.
Conclusions
There is a low prevalence of typical MRI Bern score signs of intracranial hypotension in PTH or HCs. Thus, intracranial hypotension is unlikely to be an underlying factor in persistent PTH attributed to mild traumatic brain injury.
Trial Registration
The study was registered on ClinicalTrials.gov (identifier: NCT03791515). Date of registration 2018-12-29.
This is a visual representation of the abstract.
Introduction
Post-traumatic headache (PTH) is a disabling neurologic disorder which is often caused by mild traumatic brain injury (TBI) and affects 69 million people per year worldwide (1). In most people, PTH remits spontaneously, but up to one in three will experience persistent PTH that may last years and resist all varieties of treatment (2,3). Spontaneous intracranial hypotension (SIH) is most often secondary to a spinal cerebrospinal fluid (CSF) leak and can present with varying symptoms, such as postural headache, neck pain and dizziness (4). In its most classic presentation, especially during the acute phase of CSF leak, SIH is characterized by orthostatic headache and brain imaging signs of intracranial hypotension such as pachymeningeal enhancement. However, accumulating data indicate that the longer a CSF leak persists: (i) the less likely the headache will show orthostatic features (5); (ii) the less likely pachymeningeal enhancement will persist to suggest SIH on brain
Magnetic resonance imaging (MRI) (6); and (iii) the less likely opening pressure will be low (5). Current estimates are that opening pressure is normal in 66–95% of CSF leaks (7,8). Spine MRI can be equally misleading, with up to half of all spinal CSF leaks arising from CSF venous fistulas that will not be visible on spine MRI (9,10). Even when the CSF leak and SIH are new at least 20% of brain MRIs will be normal in SIH (4). All of these features contribute to difficulty in arriving at the correct diagnosis of SIH, and recent data suggest that the mean duration of symptoms before correct diagnoses may be up to eight years in some cohorts (11). Furthermore, with the conversion of orthostatic headache to non-orthostatic headache over time, the primary clinical feature differentiating PTH as a result of TBI and a post-traumatic CSF leak headache is lost. Additionally, both syndromes are characterized by high levels of similar comorbid symptoms such as dizziness, tinnitus, nausea, fatigue and neurocognitive impairments. By contrast to persistent PTH, SIH can be effectively treated if diagnosed correctly and if the leak is identified (4,12).
The etiology of intracranial hypotension may be spontaneous, iatrogenic or, similar to PTH, may follow trauma (13). A head trauma may lead to an indirect transmission of forces to the spine, which potentially could lead to a spinal CSF leak (14). A CSF leak would then likely cause persistent headache and other symptoms. In support, SIH has been reported to occur after even mild head trauma (15–17). The underlying mechanisms behind this are not fully known but may involve direct or indirect trauma to the spine.
MRI is a core approach to evaluate patients with SIH. Studies show that eight of 10 patients with SIH will have signs of SIH that can be easily detected using the recently devised MRI based Bern SIH score (18). The score assesses six radiological signs: pachymeningeal enhancement, distention of venous sinus, effacement of the suprasellar cistern of 4.0 mm or less, subdural fluid collection, effacement of the prepontine cistern of 5.0 mm or less, and mamillopontine distance of 6.5 mm or less. The Bern score ranges from 0 to 9 points, with higher scores indicating a greater probability of CSF leakage. These measurements reflect caudal displacement of brain structure that may be subtle and otherwise be missed even by the experienced and vigilant neuroradiologist. Formal utilization of the Bern score has recently been reported to increase the accuracy of neuroradiology reporting signs of CSF leak and SIH (19). It has also been previously reported that when post-contrast pachymeningeal enhancement is not available, a restricted Bern score of cranial MRI features can still be used to predict when a CSF leak will be found on invasive spinal imaging (7).
In the setting of the overlap of a precipitating trauma, clinical symptoms and uncertainty in diagnostic imaging, we hypothesized that there was substantial potential for the diagnosis of PTH in the setting of an undiagnosed post-traumatic CSF leak. We specifically hypothesized that among a cohort of patients with persistent PTH who were initially evaluated and imaged before the publication of the Bern Criteria and before the current understanding of how to detect CSF venous fistulas, re-evaluation of their brain imaging would show elevated levels of subtle measurable brain changes codified in the Bern score compared to control patients.
Methods
Study population
Participants and HCs were included in a retrospective analysis of a previously published study, which investigated white matter hyperintensities and cerebral microbleeds (20). Eligible participants were aged 18 to 65 years and had a diagnosis of PTH following the International Classification of Headache Disorders, third edition (ICHD-3) (21). Another inclusion criterion was that mild TBI had to have occurred at least 12 months before enrollment. Participants were excluded if they had a history of more than one mild TBI, moderate or severe TBI, as well as any history of whiplash injury, medication-overuse headache or a primary headache disorder, except for infrequent episodic tension-type headache (TTH). Contraindications to MRI were also considered reasons for exclusion.
HCs were eligible for inclusion if they were aged 18–65 years and had no history of TBI, whiplash injury, primary headache disorder (except for infrequent episodic TTH), neurologic or psychiatric disorders, and cardiovascular disease. Controls were also excluded if they had first-degree relatives with any primary headache disorder except infrequent episodic TTH or reported daily intake of any medication other than oral contraceptives.
The relevant ethics committee approved the study (H-18011477). All participants provided their written consent after receiving detailed oral and written information. The study was conducted at the Danish Headache Center following the Declaration of Helsinki. The study was registered on ClinicalTrials.gov (identifier: NCT03791515) and approved by the Danish Data Protection Agency.
Study design and procedures
The present study had a cross-sectional design and was performed at a single center. Data were collected using a semi-structured interview, including demographics, clinical characteristics, comorbidities, and medication history. Imaging data were obtained on a 3T MRI Scanner (Achieva; Philips, Amsterdam, The Netherlands) using a 32-channel head coil. Participants and controls underwent a single MRI session with a standardized protocol that included 3D T1-weighted images, susceptibility weighted imaging and T2-weighted fluid attenuated inversion recovery (FLAIR) (20). No contrast was given to obtain the MRI images.
Image analysis
All images were transferred to a centralized server, where they were examined by a certified neuroradiologist (ES), who was blinded to the group status of the participants and controls. Images were interpreted for the presence and number of the following findings: distention of venous sinus, dural thickening, suprasellar cistern effacement, subdural fluid collection, effaced suprasellar cistern ≤4 mm, effacement of the prepontine cistern ≤5 mm and decrease mamillopontine distance ≤6.5 mm. Based on the findings, a modified Bern score was calculated (Table 1). The imaging analysis in this study differed from the original validated Bern score study (18) because contrast images were not performed. Dural thickening was assessed on FLAIR sequences, using adjacent normal-appearing brain parenchyma as a reference area to define dural thickening as previously described (22). Venous distension was identified based on 3D sagittal T1 weighted images. The most dominant of the two transverse sinuses was selected. The middle third of this dominant sinus was approximated to ensure that contour evaluations were not influenced by proximity of the torcula herophili or the sigmoid sinus (23).
Modified Bern score (18)
Probability of having cerebrospinal fluid leakage: 0–2 points: low, 3–4 points: intermediate, and ≥5 points: high.
Outcomes
The primary outcome was the difference in modified Bern scores between participants with persistent PTH and HCs. The secondary outcome was the difference in the number of participants with PTH and HCs who had modified Bern scores greater than two points.
Statistical analysis
Descriptive statistics were used to summarize demographic and clinical characteristics. Continuous data were presented as the mean ± SD or with median and interquartile range (IQR) when not normally distributed. Categorical data were summarized as frequencies and percentages. The Mann–Whitney U test was used to compare modified Bern scores between participants with persistent PTH and HCs. Fisher's exact test compared the number of participants with persistent PTH and HCs with a Bern score greater than two points. p < 0.05 (two-sided) was considered statistically significant.
Results
In total, 97 participants with persistent PTH and 96 age- and gender-matched HCs were enrolled and provided for the analyses of the modified Bern score. The mean ± SD age of the participants was 35.8 ± 11.6 years and 80 (82.5%) were women. Age, gender, height and weight were comparable between the two groups (Table 2). PTH participants reported a mean headache frequency of 25.5 ± 7.1 days per month (Table 2), and most participants had a migraine-like phenotype (88 of 97). The median duration of PTH to the MR scan was 37 months (IQR = 23–66).
Demographic and clinical characteristics of the participants with persistent post-traumatic headache and healthy controls
BMI = body mass index; IQR = interquartile range; NA= not applicable; TTH = tension-type headache.
The mean ± SD modified Bern score for the PTH group was 0.84 ± 1.09, whereas the mean ± SD score for the HC group was 0.85 ± 1.17 (p = 0.994). A modified Bern score of 0 was noted in 53 (55%) of 97 participants with persistent PTH and 50 (52%) of 96 HCs (Table 3). Furthermore, 27 (28%) participants with persistent PTH and 19 (20%) HCs had a modified Bern score of ≥2 points. There were no differences in modified Bern scores ≥2 points for participants with persistent PTH and HCs (p = 0.334). None of the participants had a modified Bern score of ≥4 points (Table 3).
Total modified Bern score findings in study populations
PTH= post-traumatic headache.
None of the participants with persistent PTH or HCs with a modified Bern score of ≥1 exhibited distension of venous sinus, dural thickening or subdural fluid collection. Effacement of the suprasellar cistern was observed in 18 (19%) of 97 participants with PTH and 13 (14%) of 96 HCs. Furthermore, effacement of the prepontine cistern was noted in 25 (26%) of 97 participants with PTH and 39 (41%) of 96 HCs. A decreased mamillopontine distance was found in 20 (21%) of 97 participants with PTH and 20 (21%) of 96 HCs (Figure 1 and Table 4). There were no signs of brain sagging, tonsillar descent, midbrain descent or superficial siderosis in any participants with PTH or HCs.
Sagittal magnetic resonance images of effacement of the suprasellar cistern in a person with post-traumatic headache, effacement if the prepontine cistern is in a healthy control and decreased mamillopontine distance in a healthy control
Detailed modified Bern score in participants with persistent post-traumatic headache and healthy controls
PTH= post-traumatic headache.
Discussion
The main findings of the present study were that MR signs based on the Bern score of SIH are not more frequent in people with PTH compared to age- and sex-matched HCs. To our knowledge, the present study is the first to systematically assess the associations of persistent PTH with signs of SIH. The study did not detect imaging evidence to support the hypothesis that a cohort of PTH participants might be enriched with patients with undiagnosed SIH. This is reassuring because treatment of SIH could have led to a marked improvement in this cohort of severely affected PTH participants.
Interestingly, the study demonstrated that each of the three SIH signs, distention of the venous sinus, dural thickening and subdural fluid collection, were not seen in any participants with PTH nor in any HCs. By contrast, individually, the effacement of the prepontine cistern, mamillopontine distance and suprasellar cistern was frequent: being seen in 41%, 20% and 21%, respectively. Collectively, however, these subtle measurements are less frequently seen in healthy control patients. Our data provide the first direct data on the frequency of individual Bern scores among healthy patients. Only 11% of healthy controls had a Bern score of ≥3, and only 6% had a Bern score of 4. No healthy controls had a Bern score of ≥5. These data have direct implications for the management of patients with orthostatic headache lacking the classic pachymeningeal enhancement. When a patient with a daily refractory headache presents with a brain MRI that fails to show pachymeningeal enhancement but does show subtle reductions in suprasellar distance, mamillopontine distance and prepontine distance (Bern score of 4), The decision to pursue further spinal imaging can be informed by knowledge that the pre-test probability of seeing this constellation of findings under the null-hypothesis of no-leak was 6%. The healthy control data are consistent with the original Bern score paper, which had a smaller control group of 60 patients. The Bern group found 37% with prepontine cistern <5 mm, 28% with mamillopontine distance <6.5 mm, 13% with suprasellar cistern <4 mm, 2% with pachymeningeal enhancement, 2% with subdural fluid collection and 0/60 with venous engorgement (18). Consistent with our data, in their cohort only 7% of healthy controls had a score of ≥3. Based on these findings obtained in more than 150 healthy controls from both cohorts, subdural fluid collection, pachymeningeal enhancement and distension of veins are each not normal findings, and any one of them should warrant further work-up in a patient with a headache. Berns scores of ≥4 should also be understood within the context that 94% of healthy patients will not show this set of findings. Furthermore, it is crucial to note that the original study validating the development of the Bern score established that a cutoff of ≥5 yielded a sensitivity of 79% and specificity of 98%, whereas a cutoff of ≥3 resulted in a sensitivity and specificity of 93% (7). This indicates that CSF leak remains a possibility in individuals with orthostatic headache and a Bern score of ≤4.
In PTH, several non-headache symptoms have been reported, such as nausea, visual disturbances, dizziness and/or vertigo, gait and/or postural imbalance, and impaired memory and/or concentration (24). A broad variety of symptoms reported in spontaneous SIH are nausea/vomiting, neck pain/stiffness, tinnitus, dizziness, hearing disturbances, photophobia, other visual symptoms, diplopia, vertigo, back pain, cognitive symptoms, reduced level of consciousness and even movement disorders (4). Thus, there is a significant overlap in the plethora of symptoms in PTH and SIH. The most important difference is the orthostatic headache, which has not been reported in PTH. A pitfall is that the orthostatic headache may change over time. This was demonstrated in a study of 137 SIH patients, in which 93% of the patients with <10 weeks of symptoms displayed typical orthostatic headache, but only 62.5% with >10 weeks of symptoms (25). In the literature, there are five cases of SIH after various types of trauma mechanisms. In the first case, orthostatic headache progressed over 10 days, and there was reported a mild head trauma 4 months prior as a result of a minor ski accident, which makes it uncertain that the headache was linked to the case (15). In the second case, there was a bicycle accident leading to paralysis of the right arm as a result of a suspected brachial plexus lesion. Then the patient had orthostatic headache and diplopia upon mobilization 13 days after the trauma. This indicates a direct trauma to the spine, which likely caused a leak at the lower cervical level (15). In the third case, the patient was in a car accident, collision with a deer while the patient sat in the front seat, without any direct trauma to the head (16). The headache was markedly orthostatic. In the fourth case, an athlete landed on his back, leading to a hyperextension of his cervical spine and concussion symptoms, including a headache that was not described as orthostatic (17). In the fifth case, a football player was hit on the chin, forcing his neck to be extended and he had concussion symptoms, including a headache that was not described as orthostatic (17). Overall, the cases are mostly without the report of a significant direct trauma to the head indicating that these rare SIH cases more likely are caused by direct trauma to the upper part of the spine.
The strengths of the present study are the large number of participants and matched HCs, as well as the neuroradiologist being blinded towards the status of the participants. However, the study also has several limitations. The study lacked detailed insights into the mechanism of head injury, which would have been relevant for accurately interpreting the results. The study did not assess other potential indicators of SIH, such as enlarged epidural veins, empty optic nerve sheath or spinal epidural fluid collections. When evaluating patients with suspected SIH, it is also important to perform spinal MRI with CSF-sensitive sequences (26). Furthermore, CSF leaks can only be identified using dynamic computed tomography (27) or digital subtraction myelography (28). The MRI brain images in this study were performed using FLAIR sequences to assess dural thickening because there was no use of contrast. A recent study compared contrast with FLAIR sequences in SIH patients, showing that FLAIR sequences produced an accuracy and sensitivity of 72% and 72% with a false-negative rate for FLAIR of 28% (11/39) (29). Tosaka et al. (22) demonstrated in a series of eight SIH patients that FLAIR sequences detected dural thickening in individuals who also exhibited enhancement. However, in follow-up scans after treatment, dural enhancement was no longer detectable on FLAIR sequences when only minimal enhancement was present (22). By contrast, O'Cearbhaill et al. (30) investigated 26 SIH patients and showed that dural thickening was correctly identified on the non-contrast FLAIR sequence in all patients. However, we cannot fully exclude a lower estimate of dural thickening compared to an MRI study using contrast. The PTH participants had a median disease duration of 37 (range 23–66) months, and so the MRI changes may have disappeared over time (31).
In conclusion, the MR signs of SIH encompassed by the Bern criteria are not more prevalent in people with persistent PTH compared to age- and sex-matched healthy controls. Some MR signs, such as dural thickening, subdural fluid collection and venous engorgement, do not appear to occur in healthy controls and warrant further work-up. Furthermore, we validate findings from the original Bern score report that 94% of healthy controls have a Bern score of <4, and venous engorgement alone, similar to pachymeningeal enhancement and the presence of subdural collections, is sufficient to prompt a determined investigation for CSF leak. In patients with a Bern score of 4, MRIs are often reported as “no evidence of intracranial hypotension”. The data presented here suggest the radiologic interpretation of such studies would more accurately inform with a statement such as “Bern scores of 4, as seen in this patient, are seen in ≤6% of healthy controls”. Other causes of persistent PTH need to be explored in future studies.
Clinical implications
MR imaging may detect signs of intracranial hypotension
No marked increase of the Bern score was observed in either healthy controls or patients with persistent post-traumatic headache
SIH should be thoroughly investigated in patients with PTH, and MRI findings should be carefully reviewed
Footnotes
Acknowledgements
We thank Hakan Ashina and Afrim Iljazi for their valuable input to the study. We also thank all the participants for participating in the study.
Declaration of conflicting interests
HS reports receiving personal fees from AbbVie, Lundbeck, Pfizer and Teva, and non-restricted research grants from NovoNordisk Foundation outside of the submitted work. RHJ reports receiving personal fees from Lundbeck, AbbVie, Teva, Pfizer and Novartis, and non-restricted research grants from NovoNordisk Foundation and Lundbeck Foundation, all outside the submitted work. HMA reports receiving personal fees from Lundbeck and Pfizer, outside of the submitted work.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.
Ethical statement
All participants gave written consent after receiving detailed oral and written information. The study was conducted at the Danish Headache Center following the Declaration of Helsinki
