A population-based study of coeliac disease, neurodegenerative and neuroinflammatory diseases

Transcription

1 Alimentary Pharmacology & Therapeutics A population-based study of coeliac disease, neurodegenerative and neuroinflammatory diseases J. F. LUDVIGSSON*,, T.OLSSONà, A.EKBOM, & S. M. MONTGOMERY, *Department of Paediatrics, Örebro University Hospital, Örebro; Clinical Epidemiology Unit, Department of Medicine, Karolinska Institutet, Karolinska University Hospital, Stockholm; àneuroimmunology Unit, Dept. of Clinical Neurosciences, Karolinska University Hospital, Karolinska Institutet, Stockholm, Sweden; Harvard School of Public Health, Boston, MA, USA; Clinical Research Centre, Örebro University Hospital, Örebro, Sweden Correspondence to: Dr J. F. Ludvigsson, Department of Paediatrics, Örebro University Hospital, Örebro, Sweden. jonasludvigsson@yahoo.com Publication data Submitted 16 January 2007 First Decision 19 February 2007 Resubmitted 4 March 2007 Accepted 26 March 2007 SUMMARY Background It has been suggested that coeliac disease (CD) is associated with several neurological diseases. However, the evidence of such an association is inconclusive as earlier research has often been based on small numbers with retrospective data collection. Aim To use Cox regression to examine the risk of neurological disease in individuals with CD. Methods Through Swedish national registers we identified some individuals with a diagnosis of CD ( ) and reference individuals matched for age, sex, calendar year and county. Results Coeliac disease was associated with later polyneuropathy [hazard ratio (HR) = 3.4; 95% CI = ]. We found no statistically significant association between CD and subsequent multiple sclerosis (HR = 0.9; 95% CI = ), Parkinson s disease (HR = 1.2; 95% CI = ), Alzheimer s disease (HR = 1.5; 95% CI = ), hereditary ataxia (HR = 1.3; 95% CI = ), the symptom ataxia (HR = 1.9; 95% CI = ), Huntington s disease (HR = 1.7; 95% CI = ), myasthenia gravis (HR = 0.8; 95% CI = ) or spinal muscular atrophy (HR = 0.5; 95% CI = ). Prior polyneuropathy was associated with subsequent CD (odds ratio = 5.4; 95% CI = ). Conclusions The association between CD and polyneuropathy indicates shared risks. We suggest that individuals with polyneuropathy routinely undergo screening for CD. There is no notable association between CD and other neurological outcomes investigated in this study. Aliment Pharmacol Ther 25, ª 2007 The Authors 1317 doi: /j x

2 1318 J. F. LUDVIGSSON et al. INTRODUCTION Coeliac disease (CD) is an immune-mediated disease that affects up to 1% of the population in the Western world. 1 In individuals with CD, exposure to gluten results in an immunological response characterized by villous atrophy and crypt hyperplasia in the small intestine. It has been suggested that CD is associated with several neurological diseases. 2 However, the evidence of such an association is inconclusive since earlier research has often been based on small numbers with retrospective data collection. It is also important to note that in several studies, neurological disease has been evaluated in relationship to levels of antigliadin antibodies rather than a clinical diagnosis of CD. 3, 4 Both CD and multiple sclerosis (MS) are organ specific inflammatory diseases with strong genetic influences, largely from the human leucocyte antigen (HLA) complex, 5, 6 but also from non-hla genes. 7, 8 Commonalities in genetic influences might result in an increased risk for both diseases in the same individual. Vitamin D deficiency is common in CD, 9, 10 and has been suggested as an aetiological factor in MS. 11 Despite this, we are aware of only two studies of CD and MS (other than case-reports), 12, 13 and although they failed to show an increased risk of CD in individuals with MS, this may have been because of the small study size. Research on CD in other neurodegenerative and neuroinflammatory diseases is also sparse. In 2002, Pellecchia et al. 14 reported that none of the 16 individuals with Parkinson s disease in their study was positive for antigliadin antibodies. This is however not surprising in a study totalling only 16 individuals with CD. 15 Although 2 33 patients with Alzheimer s disease were positive for antigliadin antibodies in the study by Frisoni et al., 16 none of them tested positive for antiendomysium antibodies. Ataxia is probably the only neurological disorder, for which there are numerous reports of an association with CD or increased levels of antigliadin antibodies. 4, These patients often have non-specific ataxia. 18 However, hereditary ataxia has also been associated with the presence of antigliadin-antibodies (although most often without mucosal evidence of clinical CD). 3, 4 Evidence of a positive association between CD 20, and subsequent polyneuropathy is conflicting, although there are several reports of polyneuropathy prior to a diagnosis of CD. 17, Recent case series do however indicate a link between CD and neuropathy 21, 22, 30, 31 and Brannagan et al. 32 have suggested that such an association may be due to sensory ganglionopathy or immune-mediated neuropathy. In addition to the above disorders, there are reports of an association between CD or CD-related antibodies and Huntington s disease, 33, 34 and Myasthenia 35, 36 Gravis. There are also case reports of CD and 23, 37 depressed muscular function or myopathy. For this reason we added spinal muscular atrophy to the list of our outcome measures. Most studies on CD and the above neurological diseases have either been based on screening for CD in individuals with existing neurological diseases, limited by few positive events, or the lack of a comparison group. The aim of the current study was to estimate the risk of neurological disease in a large national cohort of individuals with CD and matched reference individuals, using data from Swedish national registers. METHODS Inclusion exclusion criteria The Swedish National Board of Health and Welfare used the Swedish national inpatient register (IPR) to identify all individuals with a hospital-based discharge diagnosis of CD between CD was defined according to the international classification of disease (ICD) codes: ICD-7: ; ICD-8: , , ICD-9: 579A; ICD-10: K90.0 (in the Swedish ICD-9 system, letters rather than numerals are used to differentiate between various forms of intestinal malabsorption). In clinical practice, a diagnosis of CD is almost always preceded by a small intestinal biopsy. 38 For each individual with CD, Statistics Sweden, the government agency responsible for population statistics, identified up to five reference individuals through the Total Population Register. 39 Reference individuals were matched for age, sex, calendar year and area of residence. We then assessed the risk of neurological disease as defined by the ICD codes (Table 1). Data on diagnoses of diabetes mellitus (DM) and lymphoma were also obtained from the IPR and defined according to relevant ICD-codes. We identified individuals with CD. We then excluded 94 individuals with CD due to data irregularities, such as death recorded as occurring before the date of first recorded diagnosis of CD.

3 COELIAC DISEASE AND NEUROLOGICAL DISEASE 1319 Table 1. International Classification of Disease (ICD) codes Disease ICD-7 ICD-8 ICD-9 ICD-10 Multiple sclerosis G35 Parkinson s disease A G20 Alzheimer s disease B G30 Hereditary ataxia G11 Ataxia (the symptom) D R26.0; T27.0 Polyneuropathy ; ; D; 356E; 357 G60-63 Huntington s disease E G10 Myasthenia Gravis A G7.0 Spinal muscular atrophy ; ; ; G12 Diabetes Mellitus* E10-14 Lymphoma ; ; 201 C81-85 * The Swedish inpatient register does not distinguish between type 1 or type 2 diabetes mellitus. In the analyses of MS, we excluded 12 patients because they had received a diagnosis of MS before the study entry and another four patients as they had received a diagnosis of MS in the first year after study entry (Table 2). Another 1056 individuals were excluded due to a follow-up of one year or less. Individuals with a follow-up of one year or less were excluded to minimize the risk of detection bias due to CD being diagnosed in hospital. Follow-up time ended on the date of first discharge diagnosis of neurological Table 2. Number of participants excluded in relation to date of first discharge diagnosis of coeliac disease (CD), number of participants in the main analyses, age at first neurological diagnosis and duration from first diagnosis of CD until first diagnosis of neurological disease Disease Individuals with CD excluded from study Participants à CD Reference *Age: median and range Duration: median and range Multiple sclerosis ; ; 3 24 Parkinson s disease ; ; 1 27 Alzheimer s disease ; ; 2 24 Hereditary ataxia , 5, 72, 78, 82** 2; 1 10 Polyneuropathy ; ; 1 24 Huntington s disease , 52** 18, 22** Myasthenia gravis , 81** 11, 20** Spinal muscular atrophy ** 3** * In individuals with coeliac disease: age in years at first recorded diagnosis of neurological disease; In individuals with coeliac disease: duration in years from first recorded diagnosis of coeliac disease until first; à Excluded: neurological disease prior to first recorded diagnosis of CD; Excluded: neurological disease 1 year after diagnosis of CD; Excluded: follow-up 1 year after first recorded diagnosis of CD recorded diagnosis of neurological disease. ** Individual values. Median and range are not given. The analysis of CD and the risk of having a diagnosis of the symptom ataxia, was based on individuals with CD and reference individuals. We had then excluded five individuals with the symptom ataxia who had a main diagnosis of trauma or intoxication with alcohol listed as the main diagnosis at time of ataxia.

4 1320 J. F. LUDVIGSSON et al. disease, emigration, death or the end of the study period (31 December 2003), whichever occurred first. Similar exclusion criteria were applied to the reference individuals so that none of the reference individuals had a prior diagnosis of MS at the time of study entry. The main analysis of CD and MS was therefore based on individuals with and individuals without a diagnosis of CD (Table 2). Data for other outcome measures in this study are given in Table 2. Socioeconomic index In a subset of individuals, we had data from Statistics Sweden on socioeconomic index (SEI). In the analyses referring to MS, 8819 individuals with and individuals without CD had such data. Some 6500 of these were children, who had been assigned a socioeconomic code on the basis of the occupation of their mother. Socioeconomic index was based on the occupational classification of SEI. Statistical analyses CD and later neurological disease Cox proportional hazard models were used to estimate the risk of any of the following neurological diseases occurring after study entry and the diagnosis of CD, MS, Parkinson s disease, Alzheimer s disease, hereditary ataxia, polyneuropathy, Huntington s disease, myasthenia gravis and spinal muscular atrophy. In a post hoc analysis we also examined the risk of ataxia (the symptom). The Cox regression was conditioned on risk-set defined by sex, age, year of study entry and county of residence. This internal stratification means that each individual with CD is only compared with his her matched reference individuals. Risk estimates are given as hazard ratios (HRs). Data were stratified by sex and age at first recorded hospital discharge diagnosis of CD ( 15 years; 16 years). In separate analyses we estimated the risk of having at least two hospital discharge diagnoses of each neurological disease after study entry. We did this to increase the specificity of our measures. The proportional hazard assumption was tested through log minus log plots. In a subset of individuals with valid data on SEI, we adjusted for SEI to reduce the risk of confounding by social or material conditions as SEI is associated with a differential use of health services. In separate analyses we also estimated the risk of neurological disease including the first year of follow-up. In individuals with CD and later polyneuropathy (i.e. polyneuropathy occurred after study entry and the CD diagnosis) we performed additional analyses to exclude the possibility that the increased risk of polyneuropathy is due to other CD-associated disorders. We adjusted for DM as it is associated with CD 40 and also estimated HRs when individuals with a diagnosis of DM before end of follow-up were excluded. We estimated the risk of polyneuropathy in individuals without a diagnosis of lymphoma, as lymphoma may cause progressive neuropathy 41 and is associated with CD. 42 In post hoc analyses (Table 5), we estimated the risk of later neurological disease (occurring after diagnosis of CD and study entry) in individuals with CD compared with reference individuals who had been admitted to hospital within less than one year before or after the first diagnosis of CD in the matched individual with CD (the number of inpatient reference individuals in the analysis of polyneuropathy was ). The purpose of these analyses was to minimize the difference in ascertainment bias between individuals with CD and reference individuals (all individuals had been admitted to hospital). To maximize the power we included the first year of follow-up in the post hoc analyses. This was deemed reasonable as all individuals in the post hoc analyses had been admitted to hospital and were therefore at risk of being investigated for neurological disease due to hospital admission. We chose to adjust for sex, age, and calendar period instead of using internal stratification as some strata only consisted of one individual with CD (and no reference individuals), making it impossible for a stratified model to produce a meaningful estimate. Prior neurological disease and CD Conditional logistic regression estimated the risk of CD (the dependent variable) associated with prior neurological disease (where CD was diagnosed after the diagnosis of the neurological disease). The end of follow-up was defined as date of first recorded diagnosis of CD and the same date for the matched individuals without CD. Those with one year or less between the date of the first neurological disease (e.g. MS) diagnosis and diagnosis of CD, and corresponding date in reference individuals, were excluded. Risk estimates are given as odds ratios (OR). 95% CI for HRs not

5 COELIAC DISEASE AND NEUROLOGICAL DISEASE 1321 including 1.00 were considered statistically significant. Statistics were calculated using SPSS 11.0 (Chicago, IL, USA, 2002). Power calculations At a significance level of 5%, we had an 80% power to detect an increased risk of subsequent MS in individuals with CD if the HR was 2.0 or above. This study also had the power to detect an increased HR of 1.7 for Alzheimer s disease; 1.5 for Parkinson s disease and 1.7 for polyneuropathy. Ethics This study was approved by the Research Ethics Committee of the Karolinska Institutet. None of the participants was contacted. It was not possible to reveal patients identities using these data. RESULTS A high proportion of study participants were female (Table 3). Although, the median age at study entry of both individuals with CD and reference individuals was low (3 years; range: 0 94), some 5000 individuals with CD in the current study received the diagnosis in adulthood. The median ages at first recorded neurological diagnosis in individuals with prior CD and the duration from first recorded diagnosis of CD until first diagnosis of neurological disease are given in Table 2. CD and later polyneuropathy Individuals with CD were at an increased risk of later polyneuropathy (HR = 3.4; 95% CI = ; P < 0.001; see also Table 4). Risk estimates were similar in males and females (data not shown). Although, risk estimates for later polyneuropathy was lower in individuals diagnosed with CD in childhood (HR = 1.4) than in adulthood (HR = 3.9), a formal interaction test between CD and age found that this difference was not statistically significant (P = 0.653). The increased risk of polyneuropathy remained statistically significant after adjustment for SEI in a subset of individuals with data for this measure (both crude and adjusted HRs = 4.1). Adjustment for DM did not affect the risk estimate (HR = 3.2; 95% CI = ; P<0.001). When individuals with a diagnosis of DM before the end of follow-up were excluded, the HR for Table 3. Characteristics of participants at risk of later multiple sclerosis Characteristics *Reference (no coeliac disease) (%) Coeliac disease (%) Total Age at first recorded diagnosis of coeliac disease (years) (65.2) (34.8) Sex Men (41.1) 5929 (41.3) Women (58.9) 8442 (58.7) Calendar period (3.5) 499 (3.5) (27.5) 3933 (27.4) (43.9) 6295 (43.8) (25.1) 3644 (25.4) Socioeconomic index I 7282 (10.4) 1505 (10.5) II 9282 (13.2) 2157 (15.0) III (28.8) 5157 (35.9) Missing data (47.6) 5552 (38.6) Individuals with at least one year of follow-up after coeliac disease diagnosis or corresponding date in matched individuals (see also text). * For reference individuals we have given the number of individuals who constituted the basis for the Cox regression. We actually had data on socioeconomic index in another 6536 reference individuals but these individuals were not part of the internally stratified calculations due to missing values on socioeconomic index in the matched individual with coeliac disease. Adding the 6536 reference individuals to those presented above, the proportion of missing values was similar among individuals with coeliac disease and reference individuals; Socioeconomic index: I is highest category (see also text). polyneuropathy was 4.5 (95% CI = ; P < 0.001). Exclusion of individuals with a diagnosis of lymphoma before end of follow-up did not affect risk estimates (HR = 3.2; 95% CI = ; P < 0.001). The association between CD and later polyneuropathy remained statistically significant when we included the first year of follow-up (Table 4). Individuals with CD were also at increased risk of having at least two later discharge diagnoses of polyneuropathy (HR = 4.4; 95% CI = ; P < 0.001). Although, the risk of having later polyneuropathy decreased when the reference population was limited to inpatients, the association between CD and polyneuropathy remained statistically significant (HR = 1.7; 95% CI = ; P = 0.036). Among the inpatient reference

6 1322 J. F. LUDVIGSSON et al. Table 4. Coeliac disease and risk of later neurological disease First year excluded First year included Outcome Positive events* HR ; 95% CI P-value Positive events* HR ; 95% CI P-value Multiple sclerosis 5 vs ; vs ; Parkinson s disease 26 vs ; vs ; <0.001 Alzheimer s disease 20 vs ; vs ; Hereditary ataxia 5 vs ; vs ; Polyneuropathy 40 vs ; < vs ; <0.001 Huntington s disease 2 vs ; vs ; Myasthenia gravis 2 vs ; vs ; Spinal muscular atrophy 1 vs ; vs ; * Positive events in individuals with coeliac disease vs. reference individuals (e.g. excluding the first year of follow-up, 40 individuals with coeliac disease had a later diagnosis of polyneuropathy. Sixty-eight reference individuals had a later diagnosis of polyneuropathy); HR, hazard ratio. Estimates derived from Cox s regression. population with later polyneuropathy, DM and cerebral lesions stroke (with or without hypertension) were the most common main diagnoses at the time of study entry. Neurological disease other than polyneuropathy In summary, CD was not statistically significantly associated with any neurological disease in our study other than polyneuropathy (Table 4). Risk estimates did not differ by sex or age at study entry (data not shown). Adjustment for SEI did not affect the risk estimates (data not shown). Two individuals received a diagnosis of Parkinson s disease before the age of 50 years (one individual with CD aged 26 years, one reference individual aged 23 years); excluding these two individuals did not notably influence the risk estimate for Parkinson s disease (data not shown). A post hoc analysis showed a two-fold increased risk for the symptom ataxia among patients with CD, but this association did not attain statistical significance (HR = 1.9; 95% CI = ). When we included the first year of follow-up, CD was positively associated with Parkinson s disease (HR = 2.1; 95% CI = ), Alzheimer s disease (HR = 1.7; 95% CI = ) (Table 4), and with the symptom ataxia (HR = 2.6; ; P = 0.042), but not with other neurological disease. Individuals with CD were not at increased risk of having at least two later diagnoses of neurological disease (Table 5). This study found no association with CD for later neurological disease when reference individuals were restricted to inpatients (Table 5). Prior neurological disease and CD Individuals with polyneuropathy before study entry were at a five-fold increased risk of subsequent CD (OR = 5.4; 95% CI = ; see also Table 6). Individuals with at least two hospital discharge diagnoses of polyneuropathy before study entry had a six-fold increased risk of later CD (CD occurring after the diagnosis of polyneuropathy) (HR = 6.1; 95% CI = ; P < 0.001). Other neurological diseases in this study were not associated with future CD (Table 6). There was no statistically significant association between having the symptom ataxia and later CD (OR = 2.8; 95% CI = ). DISCUSSION This study was based on some individuals with CD and reference individuals matched for age, sex, calendar year and county of residence. It found a statistically significant positive association between CD and polyneuropathy, but not with MS, Parkinson s disease, Alzheimer s disease, hereditary ataxia, the symptom ataxia, Huntington s disease, myasthenia gravis, or spinal muscular atrophy. A potential concern is that the association of CD with subsequent polyneuropathy is due to chance, as several diseases were investigated. This is unlikely as a

7 COELIAC DISEASE AND NEUROLOGICAL DISEASE 1323 Table 5. Coeliac disease and risk of having at least two hospital discharge diagnoses of neurological disease; and risk of neurological disease when comparing with inpatient reference individuals 2 diagnoses of neurological disease* Inpatient reference individuals HRà, 95% CI P-value AHRà, 95% CI P-value Multiple sclerosis 0.7; ; Parkinson s disease 1.0; ; Alzheimer s disease 0.3; ; Hereditary ataxia 0.7; ; Polyneuropathy 4.4; < ; Huntington s disease NC 1.7; Myasthenia gravis 0.7; ; Spinal muscular atrophy 1.2; ; * Estimates represent the risk of having at least two diagnoses of later neurological disease; Estimates represent the risk of having a diagnosis of later neurological disease when reference individuals were restricted to inpatients (see also text). Hazard ratios adjusted for age, sex, and calendar period; à HR, hazard ratio; AHR, adjusted hazard ratio (see text) (estimates derived from Cox s regression); NC, not calculated due to insufficient number of individuals with coeliac disease and later neurological disease. Table 6. Prior neurological disease and risk of coeliac disease Disease OR*, 95% CI P-value Multiple sclerosis 1.8; Parkinson s disease 1.5; Alzheimer s disease 0.4; Hereditary ataxia 1.1; Polyneuropathy 5.4; <0.001 Huntington s disease 5.0; Myasthenia gravis 1.1; Spinal muscular atrophy 1.2; * OR, odds ratio estimated through conditional logistic regression. statistically significant association between CD and prior polyneuropathy was also identified. The risk increase for later polyneuropathy was based on a large number of positive events and did not vary notably by age and sex. We were not able to confirm that the diagnosis of polyneuropathy had been preceded by a positive electromyography, since the IPR does not contain data on diagnostic procedures. We instead validated our diagnosis through estimating the risk of having at least two hospital discharge diagnoses of polyneuropathy. When the specificity for polyneuropathy was enhanced in this way, the HR for the association with CD increased and remained statistically significant (HR = 4.4). The positive association between CD and neuropathy found by this study is consistent with recent data by Hadjivassiliou et al. 22 In their study, 7% of the individuals with chronic axonal neuropathy (9% in the idiopathic group) had an abnormal duodenal biopsy. 22 Unlike their study, our national register-based approach did not allow us to study the immunological characteristics of the participants. 22 Earlier research on CD and later polyneuropathy has been inconsistent. 20, 23 26, 30, 31 Rosenberg et al. 29 have stated that associations between CD and polyneuropathy may be due to disorders other than CD. We carried out separate analyses of CD and polyneuropathy taking into account the potential confounding factors, DM and lymphoma. We also estimated the risk of polyneuropathy in CD after exclusion of individuals with a diagnosis of DM made before or after the date of the first polyneuropathy diagnosis. This is because impaired glucose tolerance, preceding DM, is an independent risk factor for polyneuropathy. 43 Our analyses showed that DM or lymphoma do not mediate the association between CD and polyneuropathy. In DM, another autoimmune disease, there is an increased risk of polyneuropathy; 44 and DM could therefore be used as a model to understand the

8 1324 J. F. LUDVIGSSON et al. pathogenesis of polyneuropathy in CD. CD shares several traits with DM, including HLA type 45 and associations with early infant feeding pattern. 46, 47 We, and others, 30, 32 suggest that immune-mediated mechanisms, common to DM and CD, may underlie the increased risk of neuropathy. One such mechanism could be oxidative stress, known to occur in both CD 48 (suggested to occur in conjunction with white-matter lesions 19 ) and DM, 49 as well as in other autoimmune disease. 50 Oxidative stress occurs in low-grade inflammation, which may be seen in diagnosed CD even after several years on a gluten-free diet. 51 Another mechanism may involve CD-related antibodies directed against the nervous system. 20, 52 If antibodies are involved in the pathogenesis of CD, 53 they may also play a role in the occurrence of associated neurological complications. It is also possible that the association between CD and polyneuropathy may be nutritional. Vitamin B-12 deficiency is a risk factor for peripheral neuropathy in older patients. 54 Vitamin B-12 deficiency is common in patients with CD both before and after diagnosis of CD. 10 Animal studies have shown that vitamin B-12 regulates a number of cytokines and growth factors in the nervous system and potentially has a neurotrophic effect, 55 which could affect nerve conduction. 56 We found no increased risk of MS in individuals with CD. On the basis of earlier research, 11, 57 we speculated that low vitamin D-levels (common both before and after diagnosis among individuals with CD) 9, 10 may contribute to a positive association between CD and later MS. Interestingly, the HLA haplotype that predisposes for CD (DR3DQ2), also modestly increases the risk for MS. 5 The lack of positive association between CD and MS indicates these diseases are unlikely to share important genetic or environmental risk factors. CD was not associated with more common disorders such as Parkinson s disease and Alzheimer s disease or less common ones such as spinal muscular atrophy and hereditary ataxia. The lack of a relationship between CD and hereditary ataxia or the symptom ataxia in this study was unexpected and could be due to several factors including low specificity for ataxia, lack of power, or lack of a true association. Results on the association of CD and ataxia with greater diagnostic specificity are inconsistent. 3, 18 It is also noteworthy that we studied ataxia and CD, not ataxia and CD-related autoantibodies. Earlier research often studied ataxia in autoantibody positive individuals, 3, 4 some of these individuals had normal duodenal mucosa. Although undetected substantial associations between CD and most of our outcome measures are unlikely, we cannot fully exclude the possibility of an association between CD and diagnoses such as Alzheimer s disease and Parkinson s or the symptom ataxia as the risk estimate for these three conditions were notably increased when we included diagnoses that were made during the first year of follow-up. In the main analyses, we excluded the first year of follow-up to minimize the risk of ascertainment bias. A number of patients with neurological disease may not have been identified since the attained age at end of follow-up was low in individuals entering the study in childhood and this may also have selected a somewhat atypical subset of patients with CD. The Swedish IPR data may not allow identification of all individuals with neurological disease because a proportion of such patients only receive outpatient care. The Swedish IPR has otherwise often been used to identify neurological disease (MS, 58, 59 Parkinson s disease, 60 Alzheimer s disease, 61 polyneuropathy neuropathy, 62 and myasthenia gravis 59 ). The IPR has a high specificity for many chronic diseases. 63 Although, we cannot exclude the possibility of misclassification of neurological disease in the current study, misclassification should only affect our risk estimates systematically if misclassification differed between individuals with CD and reference individuals, which is unlikely. Smedby et al. 64 have shown that the specificity of CD, in individuals with CD and concomitant lymphoma, exceeds 85% in the IPR. Small bowel biopsy is considered mandatory before diagnosis of CD in Sweden. 65 Unfortunately, the IPR does not contain any information on this investigative procedure. Although we may not have identified all individuals with CD, our study has considerable power to detect associations and was based on a larger number of positive outcomes than earlier research in this area. There is a risk that inpatients with CD have more severe disease than the average patient with CD. However, in the first part of the study period, hospital admission was common as part of the investigative procedures including small-bowel biopsy, and is still common for small children undergoing diagnostic endoscopy. Although ascertainment bias through the use of hospital-based registries is likely to have affected our risk estimates, it cannot explain why this study found a 3- to 4-fold increased risk of polyneuropathy in CD but no increased risk of any other neurological disease. Ascertainment bias due to hospital admission

9 COELIAC DISEASE AND NEUROLOGICAL DISEASE 1325 cannot fully explain the association between CD and later polyneuropathy since this association remained statistically significant positive also when we compared with reference individuals who had been admitted to hospital. It is possible that patients with some specific clinical features associated with CD, such as malabsorption and severe diarrhoea, were more likely to be inpatients. Some disease associations may be influenced by this potential selection effect. The IPR does not contain data on the presenting symptoms of CD, so it is not possible to investigate whether a specific phenotype of CD, whether typical or atypical, is more strongly associated with polyneuropathy. In conclusion, we found an increased risk of polyneuropathy in individuals with CD. This may be due to immune-mediated mechanisms including long-term inflammation in CD. The association between CD and polyneuropathy was independent of DM and strengthened when we increased specificity using two inpatient diagnoses of polyneuropathy as our outcome measure. We failed to show a statistically significant association between CD and MS, Parkinson s disease, Alzheimer s disease, hereditary ataxia, ataxia (the symptom), Huntington s disease, myasthenia gravis and spinal muscular atrophy. ACKNOWLEDGEMENTS Declaration of personal interests: None. Declaration of funding interests: This study was supported by the Swedish Research Council, the Örebro University Hospital, the Swedish Society of Medicine, the Majblomman Foundation, the Juhlin Foundation, the Clas Groschinsky Foundation, the Karolinska Institutet funds and the Swedish Coeliac Society. The funding sources had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. REFERENCES 1 Dube C, Rostom A, Sy R, et al. The prevalence of celiac disease in average-risk and at-risk Western European populations: a systematic review. Gastroenterology 2005; 4: S Bushara KO. Neurologic presentation of celiac disease. Gastroenterology 2005; 4: S Bushara KO, Goebel SU, Shill H, Goldfarb LG, Hallett M. Gluten sensitivity in sporadic and hereditary cerebellar ataxia. Ann Neurol 2001; 49: Hadjivassiliou M, Grunewald R, Sharrack B, et al. Gluten ataxia in perspective: epidemiology, genetic susceptibility and clinical characteristics. Brain 2003; 126: Masterman T, Ligers A, Olsson T, Andersson M, Olerup O, Hillert J. HLA- DR15 is associated with lower age at onset in multiple sclerosis. Ann Neurol 2000; 48: Lincoln MR, Montpetit A, Cader MZ, et al. A predominant role for the HLA class II region in the association of the MHC region with multiple sclerosis. Nat Genet 2005; 37: Sawcer S, Ban M, Maranian M, et al. A high-density screen for linkage in multiple sclerosis. Am J Hum Genet 2005; 77: Monsuur AJ, Bakker PI, Alizadeh BZ, et al. Myosin IXB variant increases the risk of celiac disease and points toward a primary intestinal barrier defect. Nat Genet 2005; 37: Kemppainen T, Kroger H, Janatuinen E, et al. Osteoporosis in adult patients with celiac disease. Bone 1999; 24: Kupper C. Dietary guidelines and implementation for celiac disease. Gastroenterology 2005; 4: S Munger KL, Zhang SM, O Reilly E, et al. Vitamin D intake and incidence of multiple sclerosis. Neurology 2004; 62: Salvatore S, Finazzi S, Ghezzi A, et al. Multiple sclerosis and celiac disease: is there an increased risk? Mult Scler 2004; 10: Pengiran Tengah CD, Lock RJ, Unsworth DJ, Wills AJ. Multiple sclerosis and occult gluten sensitivity. Neurology 2004; 62: Pellecchia MT, Ambrosio G, Salvatore E, Vitale C, De Michele G, Barone P. Possible gluten sensitivity in multiple system atrophy. Neurology 2002; 59: Fasano A, Berti I, Gerarduzzi T, et al. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Arch Intern Med 2003; 163: Frisoni GB, Carabellese N, Longhi M, et al. Is celiac disease associated with Alzheimer s disease? Acta Neurol Scand 1997; 95: Hadjivassiliou M, Gibson A, Davies- Jones GA, Lobo AJ, Stephenson TJ, Milford-Ward A. Does cryptic gluten sensitivity play a part in neurological illness? [see comments]. Lancet 1996; 347: Pellecchia MT, Scala R, Filla A, De Michele G, Ciacci C, Barone P. Idiopathic cerebellar ataxia associated with celiac disease: lack of distinctive neurological features. J Neurol Neurosurg Psychiatry 1999; 66: Kieslich M, Errazuriz G, Posselt HG, Moeller-Hartmann W, Zanella F, Boehles H. Brain white-matter lesions in celiac disease: a prospective study of 75 diettreated patients. Pediatrics 2001; 108: E Volta U, De Giorgio R, Petrolini N, et al. Clinical findings and anti-neuronal antibodies in coeliac disease with neurological disorders. Scand J Gastroenterol 2002; 37: Hu WT, Murray JA, Greenaway MC, Parisi JE, Josephs KA. Cognitive

10 1326 J. F. LUDVIGSSON et al. impairment and celiac disease. Arch Neurol 2006; 63: Hadjivassiliou M, Grunewald RA, Kandler RH, et al. Neuropathy associated with gluten sensitivity. J Neurol Neurosurg Psychiatry 2006; 77: Luostarinen L, Himanen SL, Luostarinen M, Collin P, Pirttila T. Neuromuscular and sensory disturbances in patients with well treated coeliac disease. J Neurol Neurosurg Psychiatry 2003; 74: Pengiran Tengah DS, Wills AJ, Holmes GK. Neurological complications of coeliac disease. Postgrad Med J 2002; 78: Cicarelli G, Della Rocca G, Amboni M, et al. Clinical and neurological abnormalities in adult celiac disease. Neurol Sci 2003; 24: Holmes GK. Neurological and psychiatric complications in coeliac disease. Epilepsy and other Neurological Disorders in Coeliac Disease ( NB. This reference was cited through Rosenberg et al. J Neurol Neurosurg Psychiatry. 2005: 76: ). London: John Libby, Luostarinen L, Pirttila T, Collin P. Coeliac disease presenting with neurological disorders. Eur Neurol 1999; 42: Chin RL, Sander HW, Brannagan TH, et al. Celiac neuropathy. Neurology 2003; 60: Rosenberg NR, Vermeulen M. Should coeliac disease be considered in the work up of patients with chronic peripheral neuropathy? J Neurol Neurosurg Psychiatry 2005; 76: Chin RL, Tseng VG, Green PH, Sander HW, Brannagan TH 3rd, Latov N. Multifocal axonal polyneuropathy in celiac disease. Neurology 2006; 66: Briani C, Zara G, Toffanin E, et al. Neurological complications of celiac disease and autoimmune mechanisms: preliminary data of a prospective study in adult patients. Ann N Y Acad Sci 2005; 1051: Brannagan TH 3rd, Hays AP, Chin SS, et al. Small-fiber neuropathy neuronopathy associated with celiac disease: skin biopsy findings. Arch Neurol 2005; 62: Bushara KO, Nance M, Gomez CM. Antigliadin antibodies in Huntington s disease. Neurology 2004; 62: Barta Z, Mekkel G, Zeher M. Antigliadin antibodies in Huntington s disease. Neurology 2004; 63: 762; author reply Edwards JH. Letter: gluten and myasthenia gravis. Lancet 1975; 2: Gundlach HJ, Ernst K. Regressive myopathy with myasthenic syndrome in idiopathic gluten enteropathy. Psychiatr Neurol Med Psychol (Leipz) 1975; 27: Hadjivassiliou M, Chattopadhyay AK, Davies-Jones GA, Gibson A, Grunewald RA, Lobo AJ. Neuromuscular disorder as a presenting feature of coeliac disease. J Neurol Neurosurg Psychiatry 1997; 63: Stenhammar L, Hogberg L, Danielsson L, et al. How do Swedish paediatric clinics diagnose coeliac disease? Results of a nationwide questionnaire study. Acta Paediatr 2006; 95: Johannesson I. The Total Population Register of Statistics Sweden. Possibilities and Better Quality. Örebro: Statistiska Centralbyrån, Collin P, Reunala T, Pukkala E, Laippala P, Keyrilainen O, Pasternack A. Coeliac disease-associated disorders and survival. Gut 1994; 35: Kelly JJ, Karcher DS. Lymphoma and peripheral neuropathy: a clinical review. Muscle Nerve 2005; 31: Askling J, Linet M, Gridley G, Halstensen TS, Ekstrom K, Ekbom A. Cancer incidence in a population-based cohort of individuals hospitalized with celiac disease or dermatitis herpetiformis. Gastroenterology 2002; 123: Singleton JR, Smith AG, Russell JW, Feldman EL. Microvascular complications of impaired glucose tolerance. Diabetes 2003; 52: Boulton AJ, Malik RA, Arezzo JC, Sosenko JM. Diabetic somatic neuropathies. Diabetes Care 2004; 27: Bao F, Yu L, Babu S, et al. One third of HLA DQ2 homozygous patients with type 1 diabetes express celiac diseaseassociated transglutaminase autoantibodies. J Autoimmun 1999; 13: Ziegler AG, Schmid S, Huber D, Hummel M, Bonifacio E. Early infant feeding and risk of developing type 1 diabetesassociated autoantibodies. Jama 2003; 290: Norris JM, Barriga K, Hoffenberg EJ, et al. Risk of celiac disease autoimmunity and timing of gluten introduction in the diet of infants at increased risk of disease. Jama 2005; 293: Odetti P, Valentini S, Aragno I, et al. Oxidative stress in subjects affected by celiac disease. Free Radic Res 1998; 29: Rahimi R, Nikfar S, Larijani B, Abdollahi M. A review on the role of antioxidants in the management of diabetes and its complications. Biomed Pharmacother 2005; 59: Sattar N, McCarey DW, Capell H, McInnes IB. Explaining how high-grade systemic inflammation accelerates vascular risk in rheumatoid arthritis. Circulation 2003; 108: Lee SK, Lo W, Memeo L, Rotterdam H, Green PH. Duodenal histology in patients with celiac disease after treatment with a gluten-free diet. Gastrointest Endosc 2003; 57: Hadjivassiliou M, Boscolo S, Davies- Jones GA, et al. The humoral response in the pathogenesis of gluten ataxia. Neurology 2002; 58: Kagnoff MF. Overview and pathogenesis of celiac disease. Gastroenterology 2005; 4: S Mold JW, Vesely SK, Keyl BA, Schenk JB, Roberts M. The prevalence, predictors, and consequences of peripheral sensory neuropathy in older patients. JAm Board Fam Pract 2004; 17: Scalabrino G, Buccellato FR, Veber D, Mutti E. New basis of the neurotrophic action of vitamin B12. Clin Chem Lab Med 2003; 41: Puri V, Chaudhry N, Goel S, Gulati P, Nehru R, Chowdhury D. Vitamin B12 deficiency: a clinical and electrophysiological profile. Electromyogr Clin Neurophysiol 2005; 45: Goldberg P, Fleming MC, Picard EH. Multiple sclerosis: decreased relapse rate through dietary supplementation with calcium, magnesium and vitamin D. Med Hypotheses 1986; 21: Fredrikson S, Cheng Q, Jiang GX, Wasserman D. Elevated suicide risk among patients with multiple sclerosis in Sweden. Neuroepidemiology 2003; 22: Landgren O, Engels EA, Caporaso NE, et al. Patterns of autoimmunity and subsequent chronic lymphocytic leukemia in Nordic countries. Blood 2006; 108: Fored CM, Fryzek JP, Brandt L, et al. Parkinson s disease and other basal ganglia or movement disorders in a large nationwide cohort of Swedish welders. Occup Environ Med 2006; 63: Rosengren A, Skoog I, Gustafson D, Wilhelmsen L. Body mass index, other cardiovascular risk factors, and hospi-

11 COELIAC DISEASE AND NEUROLOGICAL DISEASE 1327 talization for dementia. Arch Intern Med 2005; 165: Miao J, Brismar K, Nyren O, Ugarph- Morawski A, Ye W. Elevated hip fracture risk in type 1 diabetic patients: a population-based cohort study in Sweden. Diabetes Care 2005; 28: Nilsson AC, Spetz CL, Carsjo K, Nightingale R, Smedby B. Reliability of the hospital registry. The diagnostic data are better than their reputation. Lakartidningen 1994; 91: Smedby KE, Akerman M, Hildebrand H, Glimelius B, Ekbom A, Askling J. Malignant lymphomas in coeliac disease: evidence of increased risks for lymphoma types other than enteropathy-type T cell lymphoma. Gut 2005; 54: Danielsson L, Stenhammar L, Ascher H, et al. Proposed criteria for diagnosis of celiac disease in children. Lakartidningen 1998; 95: