Pathogenesis of coeliac disease

CD shares with other autoimmune disorders a close association with HLA-linked genes; >90% of patients with CD are DQ2 positive, with most of the rest carrying DQ8 alleles⁽Reference Sollid and Lie¹⁾. Peptide sequences of dietary gluten, which are resistant to protease activity in the gut, are deamidated by tTG⁽Reference Molberg, Flaete and Jensen²⁾. Deamidated gluten peptides form complexes with HLA-DQ and trigger a T-cell-mediated inflammatory response⁽Reference Tollefsen, Arentz-Hansen and Fleckenstein³⁾. While this response has direct effects on the small bowel mucosa, CD is a multisystem disorder affecting potentially any organ system and there is evidence that autoantibodies directed against tTG as part of the process have consequences elsewhere in the body.

Epidemiology

Failed and delayed diagnosis of CD used to be common, as clinicians failed to recognize the various manifestations of CD and their potential to commence at any age⁽Reference Dickey and McConnell^4–Reference Gasbarrini, Ciccocioppo and De Vitis⁶⁾.

Far from being a rare condition, population-screening studies indicate not only that CD affects approximately 1% of many populations with European ancestry, including those in the Americas and Australasia⁽Reference Mearin, Ivarsson and Dickey⁷⁾, but also that it is as common in the Middle East⁽Reference Shahbazhkhani, Malekzadeh and Sotoudeh^8, Reference Shamir, Lerner and Shinar⁹⁾ and the Indian subcontinent⁽Reference Sood, Midha and Sood¹⁰⁾. Most patients present for the first time with symptoms in adulthood, with peak onset of symptoms in the 40–60 years age-group⁽Reference Dickey and McMillan¹¹⁾. Onset of symptoms and diagnosis are often in later life; in one Italian study 4% of patients were diagnosed after the age of 65 years⁽Reference Gasbarrini, Ciccocioppo and De Vitis⁶⁾.

Clinical presentations

Potential clinical presentations can be categorized into three main groups, although some symptoms have more than one basis:

1. 1. malabsorption, e.g. diarrhoea, abdominal pain, flatus–flatulence, weight loss, anaemia, osteomalacia, osteoporosis;

2. 2. gut dysmotility, e.g. reflux, dysphagia, dyspepsia, constipation;

3. 3. autoimmunity, e.g. dermatitis herpetiformis, neuropathy, ataxia.

BMI and gastrointestinal symptoms

While the patient with CD was traditionally considered to be underweight, with symptoms attributable to malabsorption such as diarrhoea, flatus and abdominal distension, recent studies report that 28–39% of patients are overweight (BMI ⩾25 kg/m²) at presentation⁽Reference Dickey and Kearney^12–Reference Murray, Watson and Clearman¹⁴⁾. In the author's own experience fewer than one-third of patients report diarrhoea as a presenting symptom⁽Reference Dickey and McMillan¹¹⁾. In a US study the percentage of patients diagnosed with CD who presented with diarrhoea fell from 73% pre-1993 to 43% after 1993⁽Reference Lo, Sano and Lebwohl¹⁵⁾. Traditionally, some patients with CD were wrongly diagnosed as having irritable bowel syndrome; new guidelines from the UK National Institute for Health and Clinical Excellence⁽Reference Dalrymple and Bullock¹⁶⁾ recommend that coeliac serology should be checked in all patients suspected of having irritable bowel syndrome. The possibility of dual pathology should be considered in patients with CD who do present with diarrhoea. There is an association between CD and inflammatory bowel disease⁽Reference Yang, Chen and Scherl¹⁷⁾ and the author's own approach is to perform colonoscopy in patients with CD presenting with altered bowel habit after the age of 40 years to exclude coincidental colon neoplasia⁽Reference Dickey¹⁸⁾. Diarrhoea that does not rapidly respond to gluten exclusion should prompt a search for small bowel neoplasia and ulcerative jejunitis complicating CD⁽Reference Brousse and Meijer¹⁹⁾. There are also links with exocrine pancreatic insufficiency, which may be easily detected using faecal elastase assays⁽Reference Leeds, Hopper and Hurlstone²⁰⁾, with small intestinal bacterial overgrowth⁽Reference Tursi, Brandimarte and Giorgetti²¹⁾ and with microscopic colitis⁽Reference Olesen, Eriksson and Bohr²²⁾.

Many patients with CD present with symptoms more typically associated with the upper gastrointestinal tract, including reflux, nausea, dysphagia and epigastric pain. These symptoms are related in part to gut dysmotility, which is well documented in CD, although the mechanism is uncertain⁽Reference Usai, Bassotti and Usai Satta²³⁾. The constipation that is paradoxically seen in some patients with CD is probably explained on the same basis. Patients with these symptoms are more likely to undergo upper gastrointestinal endoscopy as a first-line investigation. While duodenal biopsy as a routine at every endoscopy is not cost effective, it should be performed in patients having endoscopy for anaemia who have a family history of CD and considered for those with dyspeptic symptoms that have not responded to conventional acid suppressants. A proportion of patients with VA have endoscopic abnormalities in the duodenum (scalloped folds; fold loss; mosaic, grooved or nodular mucosa; erosions; visible vessels), which should prompt biopsy if seen⁽Reference Dickey²⁴⁾.

Anaemia and B-vitamin deficiency

Liver disease

A substantial minority of patients with CD have raised serum levels of aspartate and alanine transaminase on liver function testing, which settle after gluten exclusion⁽Reference Dickey, McMillan and Collins^34–Reference Novacek, Miehsler and Wrba³⁶⁾. Serological testing for CD should be part of the initial investigation of raised serum transaminases. Where liver biopsies have been performed in this situation they show changes of non-specific hepatitis and specific liver investigation is usually not required. While raised serum transaminases are usually an incidental finding, there have been patients with severe liver disease requiring transplantation in whom CD has been identified and appropriate treatment has reversed liver failure⁽Reference Kaukinen, Halme and Collin³⁷⁾.

However, it is important to differentiate this finding from specific autoimmune liver diseases that have been linked with CD. Patients with primary biliary cirrhosis, primary sclerosing cholangitis and autoimmune hepatitis all have an increased prevalence of CD⁽Reference Ludvigsson, Elfström and Broomé³⁸⁾. Diagnostic confusion may arise from raised serum tTGA levels in primary liver disease with no evidence of CD. tTGA assays may give a false positive reaction in liver fibrosis⁽Reference Vecchi, Folli and Donato³⁹⁾.

Bone and joint disease

Traditionally, CD presenting with overt malabsorption was associated with osteomalacia and rickets, manifest as bone pain and reduced serum Ca and phosphate. This disorder is now seldom seen and the widespread availability of dual-energy X-ray absorptiometry scanning has shown that osteoporosis–osteopenia is by far the commonest bone disorder of CD, affecting 20–50% of patients at diagnosis⁽Reference Meyer, Stravropolous and Daimond^40–Reference Kemppainen, Kroger and Janatuinen⁴²⁾. Aetiological factors include Ca and vitamin D malabsorption and secondary hyperparathyroidism, but serum bone-specific autoantibodies have been described⁽Reference Sugai, Cherñavsky and Pedreira⁴³⁾.

In most cases osteoporosis is asymptomatic and picked up by dual-energy X-ray absorptiometry screening. More important than the prevalence of osteoporosis is its clinical impact, and recent case–control studies suggest that fracture risk remains low in absolute terms, ranging from no risk to only a twofold increase compared with controls⁽Reference Vestergaard and Mosekilde^44–Reference Ludvigsson, Michaelsson and Ekbom⁴⁶⁾. Guidelines produced for the British Society of Gastroenterology recommend that all adults with CD should have dual-energy X-ray absorptiometry at diagnosis⁽Reference Scott, Gaywood and Scot⁴⁷⁾, but the authors of the guidelines have recently questioned this approach⁽Reference Lewis and Scott⁴⁸⁾. The merit of screening patients with osteoporosis for CD is uncertain. There is no increase in CD prevalence among post-menopausal women⁽Reference González, Sugai and Gomez⁴⁹⁾, although the yield in other groups may be higher. Where reduced bone density is identified in CD there is improvement with gluten exclusion⁽Reference Pazianas, Butcher and Subhani^50, Reference Sategna-Guidetti, Grosso and Grosso⁵¹⁾.

Arthritis and joint pain are well described in CD⁽Reference Hernandez and Green⁵²⁾. Of specific rheumatological disorders only Sjogren's syndrome, which shares DQ2 and DQ8 haplotypes, appears to be associated with CD, with a prevalence of 5–15% among patients with Sjogren's syndrome⁽Reference Iltanen, Collin and Korpela^53, Reference Luft, Barr and Martin⁵⁴⁾. The finding of tTGA in patients with other forms of arthritis may be a non-specific finding⁽Reference Picarelli, Di Tola and Sabbatella⁵⁵⁾.

Skin disease

Dermatitis herpetiformis, an intensely-itchy blistering rash characteristically affecting the buttocks, elbows and knees is the typical skin lesion of CD and was the first and best-described extraintestinal manifestation of gluten sensitivity. Clinical characteristics of 264 US patients have recently been published⁽Reference Alonso-Llamazares, Gibson and Rogers⁵⁶⁾, showing an excess of male patients in contrast to the female preponderance in most CD series. Patients presenting with dermatitis herpetiformis appear to have a lower risk of complications such as osteoporosis⁽Reference Abuzakouk, Barnes and O'Gorman⁵⁷⁾. Dapsone is highly effective in achieving skin remission but has no effect on other organ involvement⁽Reference Nino, Ciacci and Delfino⁵⁸⁾. Gluten exclusion may take months to control the rash, but will allow Dapsone withdrawal within 18 months in most cases⁽Reference Nino, Ciacci and Delfino⁵⁸⁾.

Infertility and miscarriage

For some years a link between CD and infertility or recurrent miscarriage has been reported⁽Reference Stazi and Mantovani⁵⁹⁾, but studies were small, making interpretation difficult. In a recent large case–control study that used primary-care data for >1500 women with CD and 7700 matched controls it was found that while crude fertility rates are almost identical, age-specific fertility rates show that women with CD have lower fertility when younger but higher fertility when older compared with women without CD⁽Reference Tata, Card and Logan⁶⁰⁾. This age-related increase in relative fertility is independent of treatment for CD. Furthermore, risks of caesarean section (OR 1·33) and miscarriage (rate ratio 1·31) are moderately higher in women with CD, but risks of assisted birth, breech birth, pre-eclampsia, postpartum haemorrhage, ectopic pregnancy, stillbirth and termination are similar. However, the effects of socio-economic or educational advantage rather than a true disease effect could not be excluded. A comparison of the outcomes of children of mothers with CD before and after diagnosis and treatment has shown that rates of intrauterine growth retardation, low birth weight, pre-term birth and caesarean section are higher in the before diagnosis and treatment group, suggesting a beneficial effect of gluten exclusion⁽Reference Ludvigsson, Montgomery and Ekbom⁶¹⁾.

Neurological disease

Neurological manifestations of CD are extensively described in the literature, and again may be the primary presentation with little or no gastrointestinal symptomatology⁽Reference Luostarinen, Pirttila and Collin⁶²⁾. Some, but not all, studies report an excess of epilepsy among patients with CD and vice versa⁽Reference Luostarinen, Dastidar and Collin^63–Reference Ranua, Luoma and Auvinen⁶⁵⁾. A specific syndrome of bilateral occipital cerebral calcification, partial seizures and CD is rare, with <200 patients reported in the literature and geographically localized to Italy, Spain and Argentina⁽Reference Gobbi⁶⁶⁾. Better-established associations are with peripheral neuropathy⁽Reference Ludvigsson, Olsson and Ekbom^67, Reference Hadjivassiliou, Gibson and Davies-Jones⁶⁸⁾ and cerebellar ataxia⁽Reference Hadjivassiliou, Gibson and Davies-Jones^68–Reference Burk, Bosch and Muller⁷⁰⁾. CD has been found in 4% of ninety migraine sufferers compared with 0·4% of 236 controls⁽Reference Gabrielli, Cremonini and Fiore⁷¹⁾. Conversely, the prevalence of headache in 176 patients with CD has been reported to be 46% compared with 29% for controls⁽Reference Cicarelli, Della Rocca and Amboni⁷²⁾. Both studies indicate that gluten exclusion brings about an improvement.

Neurological problems were traditionally assumed to be a result of malabsorption, but nutritional deficiencies have not been consistently demonstrated and neurological symptoms do not respond to supplementation⁽Reference Muller, Donnelly and Smith⁷³⁾. Specific autoantibodies directed at components of the nervous system have been described in patients with CD⁽Reference Alaedini, Green and Sander^74, Reference Tursi, Giorgetti and Iani⁷⁵⁾.

An English study has suggested that neurological disease as a result of gluten sensitivity may exist without enteropathy⁽Reference Hadjivassiliou, Grunewald and Davies-Jones⁷⁶⁾. Patients thus affected may lack tTGA and EmA and have normal duodenal biopsies by conventional criteria. These findings question the reliance on small bowel histology to confirm a diagnosis of CD.

Diagnosis of coeliac disease

The most important step in diagnosis is to think of the possibility, which requires awareness of the many possible presentations of CD. In addition, suspicion should be higher where there is a family history of CD and in patients with a number of autoimmune conditions that are associated through HLA with an increased prevalence of CD; as well as liver disease and Sjogren's syndrome already mentioned, these conditions include type 1 diabetes mellitus⁽Reference Mahmud, Murray and Kudva⁷⁷⁾, autoimmune thyroid disease⁽Reference Ch'ng, Biswas and Benton⁷⁸⁾ and Addison's disease⁽Reference Betterle, Lazzarotto and Spadaccino⁷⁹⁾. There is an increased prevalence of CD in Down syndrome⁽Reference Bonamico, Mariani and Danesi⁸⁰⁾. Upper gastrointestinal endoscopy, for whatever indication, in these patients should include duodenal biopsies as a matter of routine, as well as in patients with anaemia and with visible stigmata of VA in the duodenum⁽Reference Dickey²⁴⁾. It is important that an adequate gluten intake is maintained until serology and biopsies are obtained. Empirical reduction in gluten intake, on the advice of family, friends, alternative practitioners and even physicians may result not only in false negative serology but also normalization of duodenal histology.

Serological tests

Serological testing, using EmA and tTGA, has revolutionized the diagnosis of CD by allowing investigation by non-gastroenterologists who are more likely to observe its manifestations beyond the gut. As a result, the majority of cases of CD in the author's practice are identified initially by general practitioners⁽Reference Dickey and McMillan¹¹⁾. Detection of EmA requires an operator with a microscope to check for antibody fluorescence, making the test labour-intensive and subjective and providing at best a semi-quantitative result. In contrast, tTGA assays typically use ELISA, which can provide a quantitative and non-operator-dependent result by automated spectrophotometry. Recently, kits have become commercially available that allow rapid near-patient testing for tTGA. These products include the Biocard (Ani Biotech Oy, Vantaa, Finland), which measures IgA human tTGA on a finger-prick blood droplet, and the Stick CD1 (Operon SA, Saragoza, Spain), which uses serum to test for IgG and IgA human tTGA. Their performance is similar to lab-based assays and the ease of use of the Biocard offers the opportunity for testing in the community by paramedical staff such as district nurses⁽Reference Korponay-Szabo, Szabados and Pusztai⁸¹⁾.

Duodenal histology in coeliac disease

The demonstration of characteristic inflammatory and atrophic abnormalities of the proximal small intestine in duodenal biopsies remains central to the diagnosis of CD, both to support the diagnosis where serology is positive and to confirm the diagnosis in patients with characteristics suggesting CD despite negative EmA or tTGA. Traditionally, VA was required to make a diagnosis of CD, but in 1992 a spectrum of changes consistent with gluten enteropathy were described by Marsh⁽Reference Marsh⁸³⁾. In the Marsh I lesion the villi are intact but infiltrated by intraepithelial lymphocytes (>30/100 enterocytes), in Marsh II this lymphocytosis is accompanied by hyperplasia of the crypts, while the ‘classic’ CD lesion, showing these changes plus atrophy of the villi, is termed Marsh III⁽¹⁰⁰⁾. Milder degrees of gluten enteropathy are clinically significant. Relatives of patients with CD who have Marsh I lesions identified through family screening are more likely to have symptoms, anaemia and reduced bone density than those with normal histology⁽Reference Esteve, Rosinach and Fernández-Bañares¹⁰¹⁾. However, the characteristic finding of Marsh I enteropathy, intraepithelial lymphocytosis, poses a greater problem in relation to specificity for gluten enteropathy, as it is seen in 1–2% of all duodenal biopsy samples⁽Reference Kakar, Nehra and Murray^102, Reference Mahadeva, Wyatt and Howdle¹⁰³⁾, and may be a result of other gut pathologies, including other dietary protein intolerance, parasites, Helicobacter pylori gastritis, bacterial overgrowth and inflammatory bowel disease⁽Reference Brown, Mino-Kenudson and Deshpande¹⁰⁴⁾. Support for gluten sensitivity is provided by family history of CD, by positive serology in a minority of cases⁽Reference Dickey, Hughes and McMillan^94, Reference Mohamed, Feighery and Coates⁹⁵⁾ and by determination of HLA-DQ2 and -DQ8 status. In the absence of these factors and with no alternative diagnosis a trial of gluten exclusion may be warranted.

The recent description of ‘microenteropathy’, with changes associated with gluten sensitivity visible only under electron microscopy⁽Reference Sbarbati, Valletta and Bertini⁹⁷⁾, suggests that the spectrum of gluten-sensitive enteropathy extends even further than Marsh I.