Should we be ‘nervous’ about coeliac disease? Brain abnormalities in patients with coeliac disease referred for neurological opinion
- Stuart Currie1,
- Marios Hadjivassiliou2,
- Matthew JR Clark1,
- David S Sanders3,
- Iain D Wilkinson1,
- Paul D Griffiths1,
- Nigel Hoggard1
- 1Academic Unit of Radiology, University of Sheffield, Royal Hallamshire Hospital, Sheffield, UK
- 2Department of Neurology, Royal Hallamshire Hospital, Sheffield, UK
- 3Department of Gastroenterology, Royal Hallamshire Hospital, Sheffield, UK
- Correspondence to Dr M Hadjivassiliou, Department of Neurology, Royal Hallamshire Hospital, Sheffield S10 2JF, UK;
Contributors Study design: SC, MH and NH. Patient recruitment and data collection: SC, MH, DS and NH. Data analysis and interpretation: SC, MC, IW and NH. Manuscript drafting: SC, MH, PDG, DS, IG and NH. Manuscript revision: SC, MH, NH.
- Received 21 May 2012
- Revised 27 July 2012
- Accepted 31 July 2012
- Published Online First 20 August 2012
Objectives To examine the extent of brain abnormality in patients with coeliac disease referred for neurological opinion and evaluate MR imaging sequences as biomarkers for neurological dysfunction, given the lack of readily available serological markers of neurological disease in this cohort.
Methods Retrospective examination of a consecutive cohort of patients (n=33, mean age=44±13 years (range 19–64)) with biopsy proven coeliac disease referred for neurological opinion. Patients were divided into subgroups based on their primary neurological complaint (balance disturbance, headache and sensory loss). 3T MR was used to evaluate differences in brain grey matter density, cerebellar volume, cerebellar neurochemistry and white matter abnormalities (WMAs) between subjects and controls.
Results Cerebellar volume was significantly less in the patient group than in controls (6.9±0.7% vs 7.4±0.9% of total intracranial volume, p<0.05). Significantly less grey matter density was found in multiple brain regions, both above and below the tentorium cerebelli, than in controls (p<0.05). 12 (36%) patients demonstrated WMAs unexpected for the patient's age, with the highest incidence occurring in the headache subgroup. This subgroup averaged almost twice the number of WMAs per MR imaging than the subgroup with balance disturbance and six times more than the subgroup with sensory loss.
Conclusion Patients with established coeliac disease referred for neurological opinion show significant brain abnormality on MR imaging. MR imaging may provide valuable biomarkers of disease in this patient cohort.
- Cerebellar ataxia
- cerebellar degeneration
- b12 deficiency
- cerebellar degeneration
- cerebellar disease
In 1966, Cooke and Smith published the first comprehensive report of neurological manifestations associated with histologically confirmed coeliac disease.1 Detailed post mortem data showed an inflammatory process that primarily, but not exclusively, affected the cerebellum. Since then, numerous publications (mainly single and multiple case studies) have reported on patients with established coeliac disease who then developed neurological dysfunction.
The key findings from these reports were that ataxia (with and without myoclonus) and neuropathy were the most common manifestations.2
Evidence suggests a range of 10–22.5% for the prevalence of neurological dysfunction in patients with established coeliac disease.3 ,4 Given that coeliac disease has a prevalence of 1% of the population,5 the health and economic burden arising as a consequence of neurological dysfunction is potentially substantial. Characterisation of brain abnormality in these patients is fundamental to understanding the pattern and extent of disease. There is the additional need for reproducible biomarkers as there is no readily available specific serological indicator for neurological dysfunction in this cohort.
Single-voxel proton MR spectroscopy provides an insight into the underlying chemical environment of a particular brain region. T2-weighted imaging allows assessment of white matter abnormalities (WMAs) and three-dimensional T1-weighted datasets can be used to evaluate atrophy.
This study evaluated retrospectively the MR features of a consecutive cohort of patients with biopsy proven coeliac disease who were referred for neurological opinion. Brain grey matter volume, cerebellar volume, cerebral WMAs and cerebellar MR spectroscopy data of patients were compared with data of age- and sex-matched controls. The data suggest that patients with established coeliac disease referred for neurological opinion have substantial structural and functional brain deficit.
Subjects and controls
A retrospective examination of a consecutive cohort of patients with biopsy proven coeliac disease who were referred for a neurological opinion at a neurology outpatients clinic, Royal Hallamshire Hospital, Sheffield, UK and had undergone subsequent MR imaging of the brain as part of their routine clinical care was undertaken. Patients with poor quality cerebellar MR spectra were excluded. Decision about the quality of cerebellar spectroscopy was made by consensus between two of the researchers (NH/SC) and was based on previously published criteria.6 Briefly, this comprised assessment of signal-to-noise ratio, peak shape and separation of choline (Cho) and creatine (Cr) peaks. The clinical notes of all subjects were reviewed and subjects were subdivided into groups according to their referred neurological complaint.
Age- and sex-matched controls were recruited by email dispatched throughout Sheffield Teaching Hospitals NHS Trust. All controls underwent a thorough screening health questionnaire to ensure that they had no present or past medical illness and drug treatment. Particular emphasis was placed on questions attempting to establish whether controls had concurrent cerebellar symptoms. All controls and subjects underwent the same imaging protocol.
The study was approved by the local regional ethics committee and all participants gave their written informed consent before inclusion in compliance with the Code of Ethical Principles for Medical Research Involving Human Subjects of the World Medical Association (Declaration of Helsinki).
Outcome measures comprised: (1) cerebellar volume expressed as a percentage of total intracranial volume (%CV:TV); (2) N-acetyl aspartate to Cr (NAA/Cr) and Cho to Cr (Cho/Cr) ratios in the cerebellar vermis and hemisphere and; (3) the presence of cerebral T2-weighted WMAs. In each case subjects' data were compared with control data. Voxel-based morphometry was also conducted to look for differences in grey matter density between subject and control groups.
Scanning was performed at the Academic Unit of Radiology, University of Sheffield using a 3T system (Philips ACHIEVA 3·0T Best, Netherlands) with an eight-channel receive-only array head coil. The imaging protocol comprised structural and spectroscopic sequences as follows.
Structural MR imaging
High-resolution three-dimensional T1-weighted MR imaging scans were acquired using a magnetisation-prepared rapid gradient echo sequence (TR=11 ms, TE=4.8 ms, flip angle =8°, field of view = 256×205×150 mm, voxel dimension =0·8 mm isotropic, acquisition time =5 min 32 s). Axial T2-weighted images were acquired using a turbo spin echo sequence (TR=3000 ms, TE=80 ms; echo train length =14; 1 NSA, scan time 4 min 12 s).
A point-resolved spectroscopy sequence (TR=2000 ms, TE=144 ms; 128 measurements; 1024 spectral points; spectral bandwidth 2000 Hz) acquired data at two voxel positions. Each voxel comprised a parallelepiped of 2.0×1.0×2.0 cm3 (4 ml) placed over the superior cerebellar vermis and the deep cerebellar white matter of the right cerebellar hemisphere. Care was taken to avoid inclusion of cerebrospinal fluid spaces within the volume of interest. After automatic higher-order shimming, water suppression was achieved using a chemical shift selective imaging pulse technique. Postprocessing of the spectra involved the following steps: zero filling, Gaussian filtering, exponential multiplication, Fourier transform and manual phase correction with baseline subtraction. Results were calculated as the areas under the relevant fits to the resonances, relative to that of Cr, using the MR system manufacturer's proprietary software.
Structural imaging analysis
Whole brain and cerebellar volume analysis
Brain tissue volume, normalised for subject head size, was estimated with structural image evaluation, using normalisation, of an atrophy cross-sectional method.7 Cerebellar volume was calculated using an integrated registration and segmentation tool.8 Both structural image evaluation, using normalisation, of atrophy cross-sectional method and integrated registration and segmentation tool are tools of FSL.9 To allow for variation in head size cerebellar volume was expressed as a percentage of total intracranial volume (%CV:TV).
Structural data were analysed with FSL-VBM, a voxel-based morphometry style analysis10 ,11 carried out with FSL tools.9 First, structural images were brain-extracted using a brain extraction tool (BET).12 Next, tissue-type segmentation was carried out using FAST4.13 The resulting grey matter partial volume images were then aligned to MNI152 standard space using the affine registration tool FLIRT,14 followed optionally by non-linear registration using FNIRT15 ,16 which uses a b-spline representation of the registration warp field.17 The resulting images were averaged to create a study-specific template, to which the native grey matter images were then non-linearly re-registered. The registered partial volume images were then modulated (to correct for local expansion or contraction) by dividing by the Jacobian of the warp field. Finally, the modulated segmented images were then smoothed with an isotropic Gaussian kernel with a sigma of 3 mm.
White matter abnormalities
Images were reviewed by two neuroradiologists (NH/SC). Hyperintense signals on T2-weighted imaging were characterised by number, location and appearance using a modified approach of methods previously published.18 Briefly, location constituted the predominant distribution of the abnormality within white matter—that is, subcortical, periventricular or deep white matter. Appearance was classified on a scale of punctate to early confluent to confluent. The number of WMAs was recorded. We disregarded ventricular capping, as this in our view is a normal variant.
Any difference in mean %CV:TV between subjects and controls was assessed using an independent-samples Mann–Whitney U test (SPSS software (SPSS, Inc.)). This method was also applied to examine any statistical differences in mean NAA/Cr and Cho/Cr ratios of the cerebellar hemisphere and vermis between subjects and controls. To investigate grey matter density changes, we used permutation-based non-parametric inference within the framework of the general linear model (5000 permutations).19 Results were considered significant for p<0.05, fully corrected for multiple comparisons.
Forty consecutive patients with coeliac disease were referred for a neurological opinion from their gastroenterologist between January 2009 and March 2011. Three patients were excluded, as small bowel biopsy diagnosis of coeliac disease could not be confirmed. Four patients were excluded owing to poor MR spectroscopy data, leaving 33 eligible patients (mean age at time of MR imaging 44±13 years (range 19–64), M:F =10:23).
All 33 subjects had duodenal biopsy proven coeliac disease ranging from Marsh 3a to Marsh 3c. All subjects were reviewed by the same experienced consultant neurologist (MH). None of the subjects took medication known to affect cerebellar structure or function. Three (9%) subjects had type I diabetes mellitus. Six (18%) subjects smoked cigarettes; the remaining 27 (82%) were non-smokers. No subject was an ex-smoker. Nineteen (58%) subjects reported alcohol abstinence; the rest were documented as drinking within recommended daily limits.20 Median time from diagnosis of coeliac disease to MR imaging was 2.1 years (range 0.1–13.2). Median time from onset of neurology symptomatology to MR imaging was 1.6 years (range 0.4–19.0).
Patients with coeliac disease had three main types of neurological complaint: (1) balance disturbance; (2) headache and (3) sensory loss. Balance disturbance comprised gait ataxia with patients having difficulty tandem walking and standing on one leg in turn during clinical examination. Five (25%) patients with balance disturbance also demonstrated gazed-evoked nystagmus. No patient in this category had myoclonus, palatal tremor, opsoclonus, chorea or symptoms suggestive of vestibular disturbance. Headaches were episodic and occasionally resembled migraine, and sensory loss constituted mainly lower limb peripheral sensory neuropathy. Subject demographics for the subgroups are shown in table 1.
Controls comprised 33 age- and sex-matched individuals (mean age 44 ± 13 years (range 19–64), M:F = 10:23). None of the controls smoked cigarettes. All controls reported either abstinence from alcohol (n=10) or drinking alcohol within recommended daily limits (n=23).20
Cerebellar volume and voxel-based morphometry
Patients had significantly smaller cerebellar volume than age- and sex-matched controls (6.9±0.7 vs 7.4±0·9, respectively, 95% CI 0.1 to 0.9; p<0·05). When divided into presentation group, only patients in the balance disturbance subgroup had statistically significant smaller cerebellar volume than the age- and sex-matched controls (6.9±0.7 vs 7.4±0.8, respectively, 95% CI 0.1 to 1.0; p<0.05, (%CV:TV headache p>0.1; sensory loss p>0.05).
VBM showed that the patient group had statistically significant reduction in the volume of grey matter in both superior cerebellar hemispheres and multiple supratentorial cortical and subcortical regions in comparison with age- and sex-matched controls (figures 1 and 2).
White matter abnormalities
None of the controls had WMAs. Conversely, 12/33 (36%) subjects had WMAs (figure 3). The prevalence of WMAs was greatest in the headache subgroup. This category also demonstrated the greatest number of WMAs per scan (table 2).
Only one subject with WMAs was a smoker. This was a 60-year-old woman in the headache subgroup who had 14 WMAs and who also had type 1 diabetes mellitus and hypertension. No other subject with WMAs had diabetes. Two subjects of the balance disturbance subgroup with WMAs were recorded as having hypertension (both women, aged 64 and 51, who had 10 and 18 WMAs, respectively). Fifty per cent of patients with WMAs in the balance disturbance and headache subgroups and 100% of the sensory loss subgroup reported abstinence from alcohol. Overall, seven subjects with WMAs (58%) had no history of smoking, diabetes or hypertension (mean age 51±9.0, mean WMAs per scan 9±8). The remaining five subjects had at least one of the three aforementioned vascular risk factors (mean age 57±7, mean number of WMAs per scan 17±11).
No statistically significant differences were found between subjects and controls with respect to the NAA/Cr ratio of the cerebellar vermis (0.97±0.09 vs 0.97±0.08, respectively; 95% CI −0.05 to ±0.06, p>0.05) or right cerebellar hemisphere (1.05±0.14 vs 1.00±0.95, respectively; 95% CI −0.13 to ±0.01, p>0.1). Similarly, no statistically significant differences were found between subjects and controls with respect to the Cho/Cr ratio of the cerebellar vermis (0.79±0.07 vs 0.83±0.10, respectively; 95% CI −0.02±0.10, p>0.1) or right cerebellar hemisphere (0.83±0.17 vs 0.76±0.08, respectively; 95% CI −0·07±0.04, p>0.05). Even when subjects were divided into clinical subgroups and compared with age- and sex-matched controls no statistical significance could be found. Furthermore, no statistically significant difference was demonstrated when compliance with gluten-free diet was taken into account.
Compliance with a gluten-free diet
Twenty-two subjects (67%) were compliant with a gluten-free diet at the time of MR imaging according to negative serology for antigliadin antibodies, endomysial antibodies and tissue transglutaminase 2. Median time from serological antibody testing to MR imaging was 40 days (range 6–671). There was a tendency (at least in those patients presenting with cerebral manifestations (ie, balance disturbance and headaches)) for an increase in the incidence of WMAs in patients who were non-compliant with a gluten-free diet compared with those patients who were compliant (table 3).
Comparing those patients who were compliant with a gluten free-diet with those who were not showed no statistically significant differences in %CV:TV in any of the three subgroups.
This study was primarily designed to examine the brain MR imaging features of patients with established coeliac disease who also have neurological complaints. It also intended to evaluate MR imaging sequences as biomarkers for neurological dysfunction. We particularly targeted the cerebellum as abnormalities in this region have been reported previously in patients with coeliac disease.1
Results from this study show that patients with coeliac disease and neurological complaints have significant cerebral and cerebellar abnormalities in comparison with age- and sex-matched healthy volunteers. Especially, when compared with the control cohort this group of patients with coeliac disease and neurological complaints has significantly smaller cerebellar volume and has significantly reduced grey matter volume in multiple brain regions, including the cerebellum. Patients also demonstrated a high proportion of cerebral WMAs that were not present in the control group.
Our finding of cerebellar atrophy in the patient group concurs with previous autopsy reports that have shown selective loss of Purkinje cells in the cerebellar cortex of patients with coeliac disease and neurological complaints,1 and also shows functional and clinical correlation in that the vast majority of patients complained of balance disturbance. The mechanisms underlying Purkinje cell loss in relation to gluten sensitivity are yet to be fully elucidated. However, current theory tends towards an immune-mediated cellular destruction, which may involve transglutaminase autoantibodies.21
Other areas of the brain including the gyrus rectus and anterior cingulate gyrus also showed significant grey matter loss in the subject group. Abnormalities in these regions have been shown to be associated with executive and psychomotor symptoms,22 and grey matter loss in these areas was found in patients with depression.23 Although this study did not evaluate depression in the patient group, previous studies have shown increased risk of depression in patients with coeliac disease albeit, without assessing grey matter volume loss.24 ,25 Depression has often been attributed to difficulties adjusting to the chronic nature of the disease rather than any structural abnormality.26 Speculatively, gluten-mediated atrophy of the rectus gyrus and anterior cingulate gyrus (akin to cerebellar injury) seems an attractive explanation for depression in patients with coeliac disease, particularly as evidence suggests that patients are only affected if they are non-compliant with the gluten-free diet.27 Future research evaluating a possible causal link between atrophy of these areas and depression in patients with coeliac disease would seem appropriate. Similarly, future neuropsychiatric studies analysing the modulation of conscious processes, internally guided attention, manipulation of mental images and internal self-representation (roles linked to the precuneus and its connections28) may provide important insight into possible subclinical dysfunction in patients with coeliac disease.
This study is in agreement with previous published reports and suggests that WMAs are common among patients with coeliac disease and neurological dysfunction, especially in those who complain of headaches.29
The aetiology of WMAs in coeliac disease is unknown, although previous autopsy reports have suggested the possibility of an immune-mediated vasculitis.30 ,31 The lack of white matter disease in the control group and the presence of WMAs in subjects who appeared to have no known vascular risk factors imply a possible causal link between gluten sensitivity and WMAs rather than the two being an epiphenomenon. The authors appreciate, however, that an exhaustive evaluation of vascular risk factors for the subjects was not performed and the study is liable to recall bias. Nevertheless, the presence of at least one factor (from smoking, diabetes and hypertension) appeared to double the load of WMAs (9±8 to 17±11) in the patient group and thus, it may be of clinical benefit to ensure regular review of modifiable vascular risk factors in patients with coeliac disease.
The consequences of WMAs in patients with coeliac disease are also unclear. It has been shown that the risks of rapid decline in independence for activities of daily living in an elderly population are two- or threefold higher for those with severe WMAs.32 Given the relatively high prevalence of coeliac disease and that the mean age of patients with WMAs in this study was only 54±8 years, any association with increased risk of WMAs may have considerable social and economic impact as those patients age, particularly as evidence suggests that a gluten-free diet does not reverse the white matter changes.29
To our knowledge, there has been only one other published report of cerebellar MR spectroscopy in relation to coeliac disease.33 This comprised a single case report of a 38-year-old man with biopsy proven coeliac disease who developed ataxia, with MR imaging showing several enhancing cerebellar lesions, thought to be due to an unconfirmed inflammatory process. MR spectroscopy of the right hemispheric white matter showed an abnormally low NAA/Cr ratio, which was attributed to neuronal impairment. Our study failed to find statistical significance in MR spectroscopy data between patients with established coeliac disease and healthy volunteers. The patient group had been receiving a gluten-free diet for an average of over 2 years and in some cases for as long as 13 years. A gluten-free diet might have normalised any pre-existing cerebellar neurochemistry, as has been previously reported.34 However, confirmation of such an assumption cannot be made owing to a lack of longitudinal data. Nonetheless, the presence of significant brain abnormalities on MR imaging even in this cohort of patients receiving dietary treatment suggests that permanent neurological dysfunction may be seen even in those patients who adhere to the diet. A similar finding has been observed for the peripheral nervous system, in that 23% of patients with established coeliac disease had evidence of peripheral neuropathy.35
This study is mainly descriptive and has limitations, including a relatively small cohort size and the acquisition of data retrospectively. Patients were taken from a tertiary specialty clinic, introducing the possibility of referral bias. Apart from alcohol and smoking status, vascular risk factors were not analysed in both the control and subject groups. Parameters such as body mass index, blood pressure, exercise tolerance and serum lipid levels would be of particular relevance to further qualify the nature of WMAs in patients with coeliac disease. Although a thorough screening questionnaire was completed for each control, individual clinical examinations were not carried out.
In summary, our findings demonstrate that patients with coeliac disease referred for neurological opinion have significant abnormalities on MR imaging of the brain and especially the cerebellum. The extent and pattern of grey matter loss in patients with coeliac disease receiving a gluten-free diet compared with controls raises important questions about possible subclinical neurological disease in these patients and the need for early diagnosis and treatment with a strict gluten-free diet. Future prospective studies on patients with newly diagnosed coeliac disease may add further information about any neuroprotective affects conferred by a gluten-free diet.
Competing interests None.
Ethics approval Ethics approval was provided by National Research Ethics Service Committee Yorkshire and the Humber, Leeds East.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Unpublished data is kept anonymised and secure within the Academic Unit of Radiology, Royal Hallamshire Hospital, Sheffield, UK. It is available to members of the research team.