A UK study assessing the population prevalence of self-reported gluten sensitivity and referral characteristics to secondary care

Abstract

Background 

Reports suggest that gluten sensitivity (GS) exists in the absence of coeliac disease (CD). This clinical entity has been termed noncoeliac gluten sensitivity (NCGS).

Objectives 

To determine the population prevalence of self-reported GS and referral characteristics to secondary care.

Patients and methods 

A UK population-based questionnaire screened for GS and related symptoms. Diagnostic outcomes were also analyzed in patients referred to secondary care with GS. CD diagnosis entailed a positive coeliac serology (endomysial and/or tissue transglutaminase antibodies) plus Marsh 1–3 on duodenal biopsies. NCGS diagnosis was based on exclusion of CD. Clinical comparisons were made between NCGS and CD.

Results 

A total of 1002 adults in the population (female 55%, mean age 39 years). The self-reported prevalence for GS was 13% (female 79%, mean age 39.5 years, P<0.0001), with 3.7% consuming a gluten-free diet and 0.8% known to have a doctor diagnosis of CD. Individuals with GS had an increased prevalence of fulfilling the Rome III criteria for irritable bowel syndrome, in comparison with those without GS (20 vs. 3.89%, odds ratio 6.23, P<0.0001).

In secondary care 200 GS patients (female 84%, mean age 39.6 years) were investigated, in whom 7% were found to have CD and 93% to have NCGS. All CD patients were human leucocyte antigen DQ2 or DQ8 positive compared with 53% of NCGS cases (P=0.0003). Nutritional deficiencies (P≤0.003), autoimmune disorders (23.1 vs. 9.7%, P=0.0001) and a lower mean BMI (23.7 vs. 25.8, P=0.001) were significantly associated with CD compared with NCGS.

Conclusion 

GS is commonly self-reported with symptoms suggesting an association with irritable bowel syndrome. The majority of patients have NCGS, an entity which demonstrates clinical and immunologic difference to CD.

Introduction

Coeliac disease (CD) is a chronic inflammatory disorder of the small bowel, which affects 1% of the population ^1,2. The condition can be defined as a state of heightened immunological responsiveness to ingested gluten (from wheat, barley or rye) in genetically susceptible individuals ^2,3. The diagnosis of CD is based on the demonstration of histological abnormalities on duodenal biopsies in accordance with the modified Marsh classification ^4,5. Corroborative evidence used to support the diagnosis comes from positive coeliac serology, in the form of endomysial antibody (EMA) and tissue transglutaminase antibody (TTG) ^3,6. The cornerstone of treatment for CD is lifelong adherence to a strict gluten-free diet (GFD), which in the majority leads to an improved clinical outcome, psychological well-being and quality of life ^3,7.

However, the consumption of a GFD seems greatly out of proportion to the projected number of patients with CD. Marketers have estimated that 15–25% of North American consumers want gluten-free foods ^8,9, although recently published data would suggest this to be an overestimation ^10,11. A National Health and Nutrition Examination Survey in the USA, involving 7798 people aged 6 years or older, suggests that 0.63% of the American public consume a GFD, although the majority of these do not have CD ¹⁰. The prevalence of serologically diagnosed CD in this study was found to be 0.71%, yet up to 80% were previously unaware of the diagnosis of CD and not taking a GFD. Elsewhere, work from New Zealand has found that CD affects 1% of children, yet 5% report gluten avoidance ¹¹. Consistent with these findings is the emerging problem encountered in clinical practice of patients complaining of gluten-related symptoms despite the absence of diagnostic markers for CD, such as negative coeliac serology and normal duodenal biopsies. These patients pose a clinical dilemma to healthcare professionals and in the past have been described as belonging to a ‘no man’s land’ due to the diagnostic uncertainty ¹².

There is now growing evidence to suggest that gluten sensitivity (GS) in the absence of CD does exist ^13–18. This has subsequently led to the development of a consensus document with proposal of a new nomenclature and classification of three gluten induced conditions – CD, wheat allergy and noncoeliac gluten sensitivity (NCGS) ¹⁹. Wheat allergy is defined as an adverse immunologic reaction to wheat proteins and can be tested for using immunoglobulin (Ig) E serum assay or a skin prick test to wheat. NCGS is a form of gluten intolerance in whom both CD and wheat allergy have been excluded ¹⁹, although currently no data exist on the epidemiology and magnitude of this condition within the spectrum of gluten-related disorders.

The aim of this study was to determine the population prevalence of self-reported GS in adults and to ascertain the diagnostic yield in those patients referred to secondary gastrointestinal care with gluten-related symptoms. Furthermore, we sought to compare demographic, anthropometric and biochemical differences between patients with NCGS against those with CD.

Patients and methods

Population survey

During the period from February to March 2012 a population-based questionnaire was conducted outside large shopping malls and transport stations in Sheffield, UK. Members of the general public, all over the age of 16 years, participated in the study by filling out a modified version of a previously validated written questionnaire ¹, to which there were three sections. The first comprised basic demographic information, including age, sex and ethnicity. The second section screened for symptoms consistent with irritable bowel syndrome (IBS) in accordance with the Rome III criteria and questionnaire ^20,21. In addition, participants were also asked about their past gastrointestinal, psychiatric and allergic history. The final section of the survey enquired for self-reported GS (group 1) and recognized related symptoms, as demonstrated by recent double-blind, randomized, placebo-controlled studies and those of expert opinion ^13–16,19. Participants were also asked for their use of a GFD and if they had seen a healthcare professional for their symptoms. A reported diagnosis of CD in the population group was defined by those who had a doctor diagnosis of CD and were also taking a GFD.

Secondary gastrointestinal care

We also analyzed diagnostic outcomes in all patients referred by general practitioners to a dedicated secondary care clinic at the Royal Hallamshire Hospital, Sheffield, UK, between the years 2006 and 2012 (group 2). The referral criteria were ‘gastrointestinal symptoms attributed to gluten ingestion’.

The gold-standard method of delineating true GS would be dietary elimination followed by double-blind, randomized, placebo-controlled food rechallenges ^14,22. However, this is not performed in our clinical practice. Therefore, we adopted a pragmatic approach to investigate these patients. Participants were openly advised to take a gluten challenge involving 3 g or more of gluten per day for 2 weeks before undergoing coeliac serology (EMA and TTG) and oesophagogastroduodenoscopy in which four distal duodenal biopsies were obtained ^23,24. Historically, reports have suggested that patients should be advised to commence a gluten challenge involving more than 10 g of gluten per day for 6–8 weeks before testing for CD. However, this can be problematic and poorly tolerated especially in a cohort specifically seeking medical attention for GS-related symptoms. The smaller dosage and shorter duration of gluten challenge in our study was based upon data demonstrating that this is sufficient to induce both histological and serological changes in the majority of adults with CD ²³.

During the time of initial investigations, BMI was also recorded, as weight in kg/height in m², and classified in accordance with the WHO criteria as underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9) and obese (≥30) ²⁵. Additional laboratory tests included haemoglobin (normal range: female 11.0–14.7 g/dl, male 13.1–16.6 g/dl), ferritin (15–150 µg/l), folate (4.6–18.7 µg/l), vitamin B₁₂ (191–663 ng/l), albumin (35–50 g/l), Ig levels, IgE wheat serology, and human leucocyte antigen (HLA) DQ2 or DQ8 typing.

The TTG antibodies were assayed using enzyme-linked immunosorbent assay kits (Aesku Diagnostics, Wendelshsheim, Germany). A TTG titre of more than 15 U/ml was considered as positive. EMA was detected by immunofluorescence on primate oesophagus sections (Binding Site, Birmingham, UK).

The duodenal biopsies were fixed in formalin at the time of endoscopies. Specimens were orientated and embedded in paraffin wax with standard 3-µm-thick sections at three levels stained with haematoxylin and eosin. The most severe lesion was noted and graded according to the modified Marsh classification. Although villous atrophy (Marsh 3) is commonly used as the benchmark for diagnosing CD, it is now accepted that CD may present with lesser degrees of histological abnormalities ^26–28. However, there is also evidence to suggest that in cases of NCGS there may be an increase in duodenal intraepithelial lymphocytes (Marsh 1), although the majority of patients will have normal biopsies ¹⁶.

Consequently, a diagnosis of CD was based on a positive coeliac serology (EMA and/or TTG) and histological abnormalities ranging from Marsh 1 to Marsh 3. A diagnosis of NCGS was based on negative EMA and TTG, with normal (Marsh 0) to near normal duodenal biopsies (Marsh 1). Wheat allergy was diagnosed if IgE wheat serology was positive.

Comparing noncoeliac gluten sensitivity with coeliac disease

Finally, we retrospectively compared baseline demographic data, BMI, haematological and biochemical parameters for those patients diagnosed with NCGS (in group 2) against a large cohort of patients already under our care with CD based on positive coeliac serology and villous atrophy on duodenal biopsies (n=329, group 3).

Statistics and ethical issues

Statistical analysis was carried out using SPSS version 19.0 software (SPSS Inc., Chicago, Illinois, USA). Categorical variables were summarized by descriptive statistics, including total numbers, percentages, odds ratio, 95% confidence intervals, and associations were analyzed by the two-tailed Fisher exact test. Continuous variables were summarized by mean and SD, with differences between two groups calculated using the Student t-test. A P-value of less than 0.05 was considered statistically significant.

The study has been registered with Sheffield Teaching Hospitals audit and research department and was conducted according to principles of the Declaration of Helsinki and Good Clinical Practice Guidelines.

Results

Population survey

A total of 1002 adults completed the population-based questionnaire, with 55% female and 45% male (mean age 39 years, range 16–93 years).

Within the general population the prevalence of gastrointestinal conditions was as follows – gastro-oesophageal reflux 5%, inflammatory bowel disease 1.4%, gastrointestinal cancers (bowel or stomach) 0.5%, and in the absence of any known organic gastrointestinal disorders the prevalence of individuals fulfilling the Rome III criteria for IBS was 6%, of whom 80% were female (P<0.0001).

The prevalence of self-reported GS in the general population was 13% (129/1002). However, the consumption of a GFD was 3.7% (n=37), with the prevalence of known doctor diagnosis of CD being 0.8% (n=8). The mean age of gluten-sensitive individuals was 39.5 years (range 18–75 years) and 79% were female (P<0.0001) (Table 1). Participants with self-reported GS were more likely to fulfil the Rome III criteria for IBS compared with non-gluten-sensitive individuals, with a prevalence of 20 versus 3.89% (odds ratio 6.23, 95% confidence interval 3.59–10.8, P<0.0001). There was no difference in age, race, gastrointestinal cancers or heartburn between those who were gluten sensitive compared with those who were not gluten sensitive. However, individuals reporting GS had a significantly increased prevalence of anxiety, depression, chronic fatigue syndrome and food allergies/intolerances.

Table 1: Comparison between self-reported gluten-sensitive and non-gluten-sensitive adults in the community (n=1002)

Gluten-sensitive individuals described a combination of intestinal and extraintestinal symptoms in relation to gluten ingestion (Fig. 1). The most common intestinal symptoms described were bloating, abdominal discomfort/pain and altered bowel habit (consistent with IBS), with the most frequent extraintestinal complaints being fatigue and headaches.

Fig. 1: Gluten-sensitive symptoms reported in the adult community.

Secondary gastrointestinal care analysis

A total of 200 adults with GS were investigated. The clinic group comprised 84% females (mean age 39.6 years, range 16–77 years), and were therefore similar in age and sex to those complaining of GS in the community.

CD was diagnosed in 7% (n=14) of the gluten-sensitive cohort. The remaining 93% (n=186) were classified as NCGS. There were no cases of wheat allergy. All patients with CD were HLA positive, compared with 53% NCGS cases (P=0.0003).

Differences between noncoeliac gluten sensitivity and coeliac disease

A total of 329 patients with CD were compared with 186 patients with NCGS. The mean age for patients with CD was 49.8 years in contrast to 39.6 years for patients with NCGS (P<0.0001). Those with NCGS were predominantly female, whereas there was an almost 2 : 1 female to male ratio in CD (P=0.0013). CD patients were significantly more likely to have coexisting autoimmune disorders, anaemia, low levels of ferritin, folate, vitamin B₁₂ and albumin at baseline (Table 2). There was no statistical difference in family history of CD between the two groups.

Table 2: Comparison between coeliac disease and noncoeliac gluten sensitivity at baseline diagnosis

Although the mean BMI of CD patients was within the normal range at 23.7, it was significantly lower by roughly two points in comparison with those with NCGS who had a mean BMI of 25.8 (P=0.001). Categorizing the BMI of CD and NCGS, respectively, according to the WHO classification showed the following subgroups – underweight (12.7 vs. 5.3%), normal weight (57 vs. 43.7%), overweight (20.6 vs. 29.8%) and obese (9.7 vs. 21.2%).

Discussion

To our knowledge, this is the first study of its kind to assess the prevalence of self-reported GS in the community and analyze diagnostic outcomes in those referred to secondary gastrointestinal care. We have shown that GS is self-reported by 13% of the population, with 3.7% consuming a GFD, despite only 0.8% being aware that they have a formal diagnosis of CD. Gluten-sensitive individuals are predominantly female, report an association with IBS, and experience both intestinal and extraintestinal symptoms on gluten ingestion. Of those patients presenting to the gastroenterology department, the majority do not have CD but NCGS. In contrast to CD, we have established that NCGS patients have a higher mean BMI by two points and are significantly less likely to suffer complications such as nutritional deficiencies and coexisting autoimmune disorders.

Previous reports on food and gluten intolerance support our findings. Perceived food hypersensitivity is common, particularly in women, with a reported prevalence of 20.4% in the UK community ²². A wide range of systemic symptoms may be experienced related to consumption of the intolerant food ²². In addition, patients demonstrate considerably more generalized subjective health complaints in comparison with healthy controls ²⁹. A recent case–control study has noted systemic complaints of IBS, musculoskeletal pains and fatigue to be extremely prevalent comorbidities in those who also report food hypersensitivity ³⁰. Frequently, in food hypersensitivity more than one dietary trigger is identified, with wheat being a commonly recognized culprit ³¹. Double-blind, randomized, placebo-controlled studies have shown both wheat and gluten per se to induce intestinal and extraintestinal symptoms, most commonly abdominal discomfort, bloating, stool dissatisfaction and fatigue ^13,14.

Anxiety and depression have also been noted to be a common feature among patients with food hypersensitivity ³². However, psychological factors do not seem to predict or explain the severity of systemic symptoms experienced thus, suggesting a common association rather than a causation ³³. There is currently one study on gluten-sensitive individuals, which, in contrast to those on food hypersensitivity, suggests there to be a low level of somatisation and no elevation in anxiety and depression, although this was restricted in patient number and all were on a GFD at the time of baseline questionnaires, which may possibly explain the negative findings ³⁴. This suggestion would be supported by studies in other patient groups, wherein, for example, cure of duodenal ulcers leads to normalization of anxiety and neuroticism ³⁵.

In secondary gastrointestinal care, we found that 93% of patients investigated for GS had NCGS, with the remaining 7% fulfilling the criteria for CD. Similar to reports from other investigators, all patients with CD were HLA DQ2 or DQ8 positive in comparison with 53% with NCGS ^16,36. HLA DQ2 or DQ8 are present in 25% of the healthy general population ³⁷, and the difference seen in NCGS support the notion that this is a distinct clinical entity and that these haplotypes may be involved in the pathogenesis. Supportive evidence for this comes from controlled trials wherein a positive HLA DQ2 or DQ8 has been shown to be predictive of a response to a GFD in non-CD IBS patients naive to prior exclusion of gluten ^38,39. Gluten exposure has also been shown to alter small bowel barrier functions in this select group of patients, which may explain the symptoms experienced through induction of inflammation and hypersensitivity ^39,40. Furthermore, current interest lies in underpinning the role of the innate and adaptive immune in NCGS, although as yet this remains complex and incompletely understood ⁴¹. Nevertheless, due to the changes in permeability and the immune response generated, this may contribute to the local and systemic symptoms experienced similar to the manifestations of postinfectious IBS ^42,43.

Finally, we have established clinical differences between CD and NCGS. Previous studies have shown CD to be commonly diagnosed between the fourth and six decades, with a 2 : 1 female to male preponderance ^44,45. Our findings show similar results with regard to CD patients and also demonstrate that in comparison NCGS patients are predominantly women who on average tend to be 10 years younger at the time of diagnosis. This difference may be explained by the specific presentation of NCGS patients, who on reporting gluten-related symptoms are directly investigated for associated disorders, thus potentially allowing a prompt diagnosis. However, in CD the presentation can be nonspecific and heterogeneous, which may lead to multiple investigations, incorrect initial diagnoses (i.e. IBS), and in some cases consultations with numerous physicians including gastroenterologists before the eventual diagnosis of CD is reached. As a result, reports suggest that there may be a delay in diagnosis of CD by around a decade ^45–48.

We have shown that CD patients are significantly more likely to have nutritional deficiencies, anaemia and a lower mean BMI compared with those with NCGS. These findings are consistent with previous reports on CD and biologically plausible due to the reduced absorptive capacity of the proximal small bowel secondary to villous atrophy ^49–52.

There was also a significantly greater association of autoimmune disorders in patients with CD compared with NCGS (23.1 vs. 9.7%). The autoimmune disorders noted in the NCGS group were 14 cases of thyroid disease and one case each of Sjogren’s syndrome, type 1 diabetes mellitus, autoimmune thrombocytopenic purpura and rheumatoid arthritis. Previous studies have shown that roughly 30% of adults with CD may have one or more autoimmune disorders ^53–55. It has been postulated that dysfunction in T-regulatory cells accounts for the loss of immune homoeostasis and the development of autoimmunity as seen in CD and related conditions ⁵⁶.

Both CD and NCGS patients share a similar prevalence of an affected first-degree relative with gluten-sensitive enteropathy (7.3 vs. 12.4%, respectively). The findings related to CD are consistent with the literature, which has suggested that 5–10% have an affected family member ^57,58. There has generally been a paucity of data with regard to family history in NCGS, other than a rectal gluten challenge study in noncoeliac siblings of CD patients showing that up to half demonstrate GS ⁵⁹. More recently, an Italian group found that 12.8% of NCGS patients had a family history of CD, similar to our data ¹⁶.

This study has limitations. First, the population-based study was a self-completed questionnaire for which we were not able to investigate further. Nevertheless, the individuals self-reporting GS in the population appear to have similar characteristics to those referred to secondary care. Second, there can be a discrepancy between perceived food intolerance and the gold-standard method of testing, which is dietary elimination followed by double-blind, randomized, placebo-controlled food challenges ²². This may, therefore, have led to an over representation in our samples as it has recently been shown that of those complaining of GS 30% of patients are truly sensitive to gluten-based products ¹⁴. Finally, there has also been uncertainty as to whether it is gluten withdrawal that benefits patients or another component of wheat. Previous studies have shown that fermentable, poorly absorbed, short-chain carbohydrates [fermentable oligosaccharides, disaccharides, monosaccharides and polyols (FODMAPs)] such as fructans, which are present in wheat, may provoke gastrointestinal symptoms in IBS patients through mechanisms that interplay with gut microbiota, gas production or fermentation ^60–62. In fact, an Australian group, who following a double-blind, randomized, placebo-controlled study were influential in first identifying the concept of NCGS ¹³, have now found that the reduction of FODMAPs in this group is actually the major contributing factor to symptom relief ⁶³. The authors have shown that patients with self-perceived NCGS, on a GFD, gained further benefit after being placed on an additional strict low FODMAP diet. However, following double-blind, randomized, placebo-controlled rechallenges there was no evidence of specific or dose-dependent effects involving gluten. They have therefore challenged the concept of NCGS, although they concluded that the condition may still exist and the differences seen in their two studies could be attributed to different types of gluten used or that gluten may only induce symptoms in the presence of a moderate content of FODMAPs ⁶³. Nonetheless, these reports support the suggestion by some that the clinical term noncoeliac wheat sensitivity is more appropriate and should be used in preference to NCGS ¹⁴.

In conclusion, sensitivity to gluten-based products is a common complaint self-reported by the population, with the use of a GFD outweighing the prevalence of known CD. Affected individuals report a wide range of symptoms related to gluten ingestion and there is a relationship with IBS. Of those investigated by gastroenterology in secondary care, 7% have CD and 93% NCGS. Individuals with NCGS are unlikely to present with the clinical complications associated with CD.

Acknowledgements

I.A. helped design the study, recruit patients, collect, analyze and interpret data and draft the article. N.R.L., M.H., S.N.W., N.R., A.K. and L.N. helped recruit patients, collect, analyze, interpret data and review the final version of the manuscript. D.S.S. conceived and designed the study, helped recruit patients, collect, analyze and interpret data, and review the final version of the manuscript.

Conflicts of interests

D.S.S. has received an unrestricted educational grant from Dr Schär (a gluten-free food manufacturer) to undertake an investigator led research study on GS. For the remaining authors there are no conflicts of interest.