How to manage adult coeliac disease: perspective from the NHS England Rare Diseases Collaborative Network for Non-Responsive and Refractory Coeliac Disease

Abstract

Adult coeliac disease (CD) affects approximately 1% of the population. Most patients diagnosed will respond to a gluten-free diet; however, up to 30% may have persisting symptoms. Such patients may have ongoing issues associated with adherence, non-responsive CD or refractory CD. This article provides a clinical overview of how to manage this group of patients with persisting symptoms, including an investigational algorithm and details of how to contact the National Health Service England Rare Diseases Collaborative Network for Non-Responsive and Refractory Coeliac Disease. We hope this will be a valuable source of contemporary information for all UK gastroenterologists and internationally.

Significance of this study.

• Historically, in the UK, serology has been used as a marker for adherence, but growing evidence in the published literature does not support this.

• Reported histological remission rates range from 34% to 65% at 2 years after diagnosis, and 66% to 85% at 5 years, with some patients never achieving complete resolution.

• We propose a contemporary treatment algorithm for patients with persisting symptoms, with a starting point of repeat gastroscopy and duodenal biopsy.

• We provide a contemporary perspective of adherence and refractory coeliac disease (RCD1; gluten contamination elimination, elemental diets) and of functional symptoms (alternative diets, low fermentable oligosaccharide disaccharide monosaccharide and polyols).

• In any patient in whom RCD is suspected, we recommend early contact with the National Health Service England’s Rare Diseases Collaborative Network for Non-Responsive and Refractory Coeliac Disease.

Background

Coeliac disease (CD) is a chronic immune-mediated enteropathy precipitated by dietary exposure to gluten in genetically predisposed individuals.¹ CD is common with an estimated prevalence of approximately 1% worldwide,² but currently in the UK about one in four cases are diagnosed.

In the majority of patients, lifelong adherence to a gluten-free diet (GFD) leads to remission both clinically and histologically.¹ Previous investigators have also demonstrated that adherence to a GFD results in normalisation of standardised mortality rates and a reduced risk of long-term-associated health complications.^{3 4} However, in a significant minority of patients, complete clinical response and mucosal recovery do not occur, and symptoms may persist despite strict adherence to a GFD.

Non-responsive CD (NRCD) is defined as persistent symptoms, signs or laboratory abnormalities typical of CD despite between 6 and 12 months of strict adherence to a GFD. However, this time length is arbitrary; therefore, it is important to be guided by the comprehensive clinical picture rather than histology alone. Patients with NRCD may be separated into two main groups: primary (patients who never respond to a GFD) or secondary (patients who respond initially, but in whom symptoms recur after 12 months).⁵ NRCD is common, affecting between 7% and 30% of patients with CD on a GFD.^{6 7}

If NRCD is suspected, the first step is to confirm the original diagnosis was CD. In those with confirmed CD, the ingestion of gluten (either purposeful or inadvertent) is the most common cause of NRCD (35%–50% of cases)^{6 8 9}; however, it is crucial to thoroughly investigate for other pathologies (figure 1).

Figure 1.

National service algorithm for investigating patients with persisting symptoms. Clinicians considering referring to the national centre may choose to undertake any of the investigations (particularly those outlined in the green box) prior to referral. in our experience, the patients who are most commonly referred have already had a gastroscopy that demonstrates persisting villous atrophy and/or presence of a monoclone. *Some clinicians may opt not to check faecal calprotectin and to go directly to colonoscopy. green box: we tend to plan for all of these tests at the initial clinic appointment for NRCD. CD, coeliac disease; GFD, gluten-free diet; HLA, human leucocyte antigen; NRCD, non-responsive CD; RCD, refractory CD; SIBO, small intestinal bacterial overgrowth.

How should we investigate patients with CD who have persisting symptoms?

Confirm diagnosis of CD

In previous studies of patients referred with suspected NRCD, between 3% and 11% of patients had the original diagnosis of CD withdrawn on further investigations.^8–10 Therefore, review of histology and serology is essential (IgA endomysial antibody (EMA), tissue transglutaminase antibody (TTG)^{9 11} and historically gliadin antibodies). Notably, selective IgA deficiency has an association with CD (prevalence 2.6% in patients with CD¹²), with its own unique complications (http://www.piduk.org/static/media/up/SelectiveIgAdeficiencypatientsheet.pdf). For this reason, in the presence of IgA deficiency, an IgG TTG should be requested to ensure there is not a false-negative TTG result. Serologically, negative CD accounts for approximately 3%–5% of all patients with CD,¹³ but caution is required as only around 30% of serology negative villous atrophy (VA) is attributed to CD and there are many other causes for VA in this group.¹⁴ A positive human leucocyte antigen (HLA) DQ2 or DQ8 occurs in 97%–99.7%% of patients with CD,^{15 16} thus negativity for these makes CD unlikely, although there is variability in the laboratory reporting of HLA.¹⁷ HLA DQ2 and DQ8 is only useful as a negative predictive test because it is present in up to 40% of the general population.¹⁸

Assessment of dietary adherence to a GFD

Once the original diagnosis of CD has been confirmed, further investigation into dietary adherence is required to exclude inadvertent gluten exposure. Overall reported apparent adherence to a GFD in patients with CD ranges from 42% to 91%,¹⁹ and presently adherence is monitored using a combination of dietetic evaluation, symptom assessment and serological markers. A crucial misconception is that negative follow-up serology equates to healed mucosa (normal villi); however, a recent meta-analysis of serology for the assessment of persisting VA refutes this.²⁰ In all, 11 studies (n=1088) demonstrated that the TTG specificity was 0.83 (95% CI: 0.79 to 0.87) and the sensitivity was 0.5 (95% CI: 0.41 to 0.60). For EMA, the specificity was 0.91 (95% CI: 0.87 to 0.94) and the sensitivity was 0.45 (95% CI: 0.34 to 0.57).

Based on presence or absence of persisting symptoms and biopsy results, patients may be split into four groups (figure 2).

Figure 2.

Outcomes following the diagnosis of adult CD and withdrawal of gluten. Group 1: consider alternate causes for ongoing symptoms. Group 2: consider RCD once ongoing gluten exposure, super-sensitivity and slow response have been excluded. Group 3: discharge to primary care. Group 4: discuss presence of ongoing inflammation (villous atrophy) with the patient, and explain potential risks of long-term complications as a result of this. If the patient agrees, consider further dietetic support. CD, coeliac disease; RCD, refractory coeliac disease.

We recommend a repeat gastroscopy and duodenal biopsy as the starting point in the assessment of patients with persisting symptoms (figure 1). The progression of histological remission with the passage of time is another crucial concept, as this can be used to guide the timing of follow-up biopsy. Observational studies have reported histological remission rates of 34%–65% at 2 years after diagnosis, and 66%–85% at 5 years^{21 22} in individuals with CD who are carefully adherent to a GFD, thus some patients with CD may be ‘slow responders’. A Finnish study (n=263) demonstrated that after 1 year, 68% of patients had histological remission. Of note, those patients with more severe mucosal damage (total/subtotal VA) on index biopsy were least likely to have normal mucosa.²³

It is still unclear how to manage individuals in the long term with persisting VA who are asymptomatic and have no apparent haematinic deficiencies. Other than careful dietary assessment of gluten intake, we are not certain whether steroids or immunosuppressive therapies are appropriate in this group of patients. Of course, this view has to be tempered by the fact that long-term VA is associated with increased risk of osteoporotic fractures and lymphoproliferative malignancy, and may be related to older age at diagnosis, type of clinical presentation and more recently a childhood diagnosis.^{24 25}

Future developments may provide suitable non-invasive markers of VA, with ongoing work on urine or faecal immunogenic peptides currently taking place.^26–29 Ultimately, it is the patient’s choice as to what level of adherence they wish to follow (figure 2, parenthesis), and we must provide the evidence base for them to make an informed decision.

Persisting VA with no evidence of gluten contamination

In patients with NRCD with evidence of persisting VA but no identifiable source of continued gluten ingestion, that is, patients with GFD-adherent NRCD, management becomes more difficult. There is also a subset of patients who are sensitive to smaller quantities of gluten which would normally be tolerated by most patients with CD (less than 20 parts per million),³⁰ and thus have an incomplete response to a GFD with persisting mucosal damage. These patients may be labelled as ‘super-sensitive’.

A gluten contamination elimination diet (GCED) is aimed at eliminating any risk of gluten cross-contamination, and may differentiate between patients reacting to trace amounts of gluten contamination and those who truly have refractory disease, potentially reducing the risk of unnecessary corticosteroids or immunosuppressive therapy in such patients.^{31 32} However, despite its value in discriminating between groups, an elimination diet is highly restrictive for patients to follow and long-term adherence to the GCED has not been reported. Based on the same principle of inadvertent gluten exposure, an elemental diet has also been shown to lead to symptomatic and histological improvement in patients who did not improve initially on a GFD.^{33 34}

If there is no persisting VA, consider other aetiologies for symptoms

The vast majority of patients diagnosed with CD ultimately respond to a GFD, with symptom resolution and histological recovery (figure 2, group 3). However, in some cases, patients do continue to experience gastrointestinal (GI) symptoms, despite adherence to a GFD and mucosal recovery on biopsy (figure 2, group 1). In these patients, other causes for symptoms should be investigated and considered as part of a systematic approach, as several pathologies may be responsible for the persistence of symptoms.

Diagnoses that may be associated with CD include microscopic colitis, pancreatic insufficiency, irritable bowel syndrome (IBS), small intestinal bacterial overgrowth, inflammatory bowel disease and lactose or fructose intolerance secondary to mucosal damage,6 8–10 35–37 and these conditions should be excluded and treated appropriately.

Functional disorders

Persisting GI symptoms in the presence of a normal duodenal biopsy may be related to IBS-type symptoms or dysmotility superimposed on CD, if no organic cause is found on further tests. There is emerging data supporting the complementary role of a ‘Low fermentable oligosaccharide disaccharide monosaccharide and polyols’ diet under these circumstances, with evidence displaying an improved long-term quality of life and GI symptoms when using this as an adjunct to the GFD.^{38 39} The use of a probiotic mixture in CD patients with persisting IBS-type symptoms has been investigated in a single recent randomised, double-blind, placebo-controlled multicentre trial.⁴⁰

There is also a higher prevalence of reflux symptoms and oesophageal dysmotility in patients with CD compared with the general population, with one study reporting a reflux prevalence of 66% and IBS prevalence of 22% in patients with CD.⁴¹ The same group found that coexisting reflux and IBS are associated with a worse quality of life (p<0.0001). It is therefore worth assessing patients with CD for these conditions, as simple measures such as the adjuvant use of proton pump inhibitor or anti-spasmodics may improve your patient’s symptoms and quality of life.

Refractory coeliac disease

Once patients have been thoroughly investigated, and if no other aetiology for persisting VA is identified, refracory CD (RCD) should be considered. RCD is arbitrarily defined as persistent or recurrent malabsorptive symptoms and signs with VA despite adhering to a strict GFD for more than 12 months.⁴²

Epidemiology

RCD is rare, and its prevalence varies widely between studies, reported to affect between 0.3% and 4% of patients with CD,^43–46 and to be responsible for 8%–23% of NRCD cases in patients investigated in specialist centres.6 8 10 45 However, the true prevalence of RCD is likely to be extremely low, and to only represent about 10% of referrals for suspected RCD.⁹ This is probably as a result of the above definition, meaning it is likely that RCD is overdiagnosed in the population. This is due to the likelihood of ongoing gluten exposure in the majority of cases and the difficulty in differentiating between inadvertent gluten exposure, slow responders, super-sensitive patients and true RCD. RCD is a diagnosis made predominantly in adults aged 50 years or above, although younger cases have also been observed.^{43 47}

Presentation and associations of RCD

RCD is defined by persistent malabsorptive signs and symptoms and, as such, commonly presents with diarrhoea and weight loss, as well as persistent nutrient deficiencies. In patients who present with additional symptoms such as GI bleeding, fever, night sweats and bowel obstruction, conditions associated with RCD should be considered. These include enteropathy-associated T-cell lymphoma (EATL) and ulcerative jejunitis (UJ).⁴⁸

Classification of RCD

RCD is an important diagnosis to consider, as it carries with it significant morbidity and mortality, and early detection and treatment may improve prognosis. There are two types of RCD, types 1 and 2, which differ in presentation, histological diagnosis and mortality, and so differentiation between the two is essential for determining prognosis and management of the condition.

The definition of RCD poses difficulties for clinicians when classifying patients for a number of reasons. First, it is possible (and common) to have severe VA in the absence of malabsorptive, or indeed any, signs and symptoms, yet such patients would be excluded by this definition. Another issue is the possibility of continued gluten ingestion accounting for ongoing symptoms and VA, as a strict GFD is very difficult to follow, and there is currently no reliable method to check for gluten ingestion; this is also true for slow responders and super-sensitive patients who may not fully respond to a GFD. The arbitrary time length of 12 months is also problematic in this definition, as many patients with severe malabsorption due to RCD never respond to a GFD. In such patients, RCD may be suspected long before the 12 months required by definition to diagnose RCD due to the severity of their symptoms, yet this definition does not allow for this.

RCD cases may be subdivided into primary, where symptoms never improve despite starting and maintaining a strict GFD, or secondary, where there has been a relapse of symptoms after apparent response to GFD for at least 1 year, although the clinical and pathophysiological significance of this is uncertain. These patients can then be classified as RCD1 or 2, depending on the presence or absence, respectively, of an aberrant population of intraepithelial lymphocytes (IELs). Patients with RCD2 generally have a poorer response to treatment, with a higher risk of progression to EATL and a poorer prognosis. Five-year survival in RCD2 is around 50% compared with 90%–100% in patients with RCD1.43 44 49 50 Mortality in RCD2 may occur either as a direct result of malabsorption or from development of associated complications including UJ and EATL, with progression to EATL reported to occur in 33%–67% of RCD2 cases, which has a 5-year survival of only around 10%.43 44 49 50

Investigations and diagnosis of RCD1 and 2

The first step for patients with suspected RCD is a small bowel biopsy to confirm the presence of persistent VA, and to make the distinction between RCD1 and 2.^{45 49} The biopsies should be analysed to determine whether an aberrant and/or clonal population of IELs is present. Detailed analysis of cell populations in the duodenal mucosa is essential to distinguish between RCD1 and 2, using either immunohistochemistry or flow cytometric analysis to categorise these patients.

RCD2 is characterised by the expansion of a population of phenotypically aberrant IELs, which lack the usual expression of surface CD3 and CD8, but do express intracellular CD3.⁴⁸ It is important to note that the phenotype of this IEL subset is present in normal individuals, uncomplicated CD and RCD1, but at lower frequencies to that found in RCD2.^{51 52} Clonality alone is therefore not an adequate indicator of RCD2, which requires quantification of the aberrant IEL subset within the small intestine.

Management of RCD

Once a diagnosis of RCD has been confirmed, initial investigations should be carried out to assess for mucosal lesions and evidence of underlying EATL. These include cross-sectional imaging, fluorodeoxyglucose-positron emission tomography, enteroscopy and capsule endoscopy (figure 1).

Patients may present with severe malabsorption, malnutrition and weight loss, in which case trace elements such as zinc and copper should be checked and corrected, as well as routine investigations including haematinics, iron studies, full blood count, urea and electrolytes, albumin, magnesium and calcium.⁴⁴ Parenteral nutrition (with the risks of refeeding syndrome^{53 54}) and management of other complications related to malabsorption such as osteoporosis may be required. Management of RCD is challenging, and we recommend patients are referred to the national centre for assessment and ongoing follow-up.

Treatment strategies in RCD 1

Management of RCD1 is aimed at improving symptoms and promoting histological recovery, with an emphasis on ongoing follow-up and surveillance for features of RCD2 or malignancy.

The use of steroids in RCD1 has been shown to be largely successful, with clinical improvement reported in up to 90% of patients.^{43 44 55} Prednisone, budesonide or a combination of prednisone and azathioprine are clinically effective to induce clinical remission and mucosal recovery in most patients with RCD1 (48).⁴⁸ Treatment should be initiated with oral prednisolone (0.5–1 mg/kg/day) with prophylaxis to maintain bone mineral density as required. Following a response to steroids, azathioprine can be added (2–2.5 mg/kg/day). When considering any preparation of budesonide (Entocort or Budenofalk), these are only bioavailable in the small bowel if the tablet is ground down or the capsule is opened. We therefore ask the patient to grind the medication with their teeth prior to ingestion to optimise small bowel efficacy.⁵⁶

Combination therapy of azathioprine and prednisolone might exert better histological restoration, although complete normalisation of villi is only seen in 50% of patients.^{43 57} There is some concern that thiopurines might enhance the risk for development of lymphoma, but this was not observed in a group of 43 patients with RCD1 with a mean follow-up of 72 months.⁴⁹

Finally, there are case reports/series describing the use of infliximab,^{58 59} thioguanine⁶⁰ and small intestinal release mesalamine⁶¹ in RCD1.

In patients who respond to treatment, we arbitrarily recommend annual gastroscopy with duodenal biopsies is performed to check for aberrancy of IELs and TCR clonality to ensure no progression to RCD2. It is recommended that clinicians consider withdrawing azathioprine after 2–3 years of complete response to confirm the diagnosis of RCD1 rather than a slow response to gluten withdrawal.

Treatment strategies in RCD 2

RCD2 remains rare, and there is a paucity of evidence for management of the condition, which is limited to a number of case series. Initial treatment for RCD2 relies on prednisolone or budesonide.^{44 56} The use of the adenosine nucleoside analogue cladribine has been shown to be well tolerated and can induce clinical and histological improvement, although it does not appear to prevent EATL development.^{62 63} Other immunosuppressive medications have been reported, including infliximab,58 59 64 65 campath (anti-CD52),^{44 66} methotrexate,⁴⁴ ciclosporin,^{44 67} interleukin 10 (IL-10)⁶⁸ and recently anti-IL-15 monoclonal antibody.^{69 70}

In patients who do not respond to any other treatments, autologous haemopoietic stem cell transplantation (aHSCT) has been used with uncertain success in a small number of cases.^{71 72}

Outcomes in RCD

Outcomes and prognosis in RCD2 are extremely disappointing, and results from previous studies are summarised in table 1.43–45 49 50 73

Table 1.

Outcomes in RCD by country

Author	Country	RCD patients (N)	Evolution into EATL	Mortality	5-year survival rate
Malamut G, Gastroenterology 200944	France	Total=57 RCD1=14 RCD2=43	Total=18 (32%) RCD1=2 (14%) RCD2=16 (37%)	Total=29 (51%) RCD1=3 (21%) RCD2=26 (60%)	RCD1=93% RCD2=44%
Al-Toma A, Gut 200749	The Netherlands	Total=93 RCD1=43 RCD2=50	Total=26 (28%) RCD1=0 RCD2=26 (52%)	Total=54 (58%) RCD1=3 (7%) RCD2=28 (56%) RCD2 +EATL=23 (88%)	RCD1=96% RCD2=58% RCD2 +EATL=8%
Rubio-Tapia A, Gastroenterology 200943	USA	Total=57 RCD1=42 RCD2=15	Total=10 RCD1=0 RCD2=10 (67%)	Total=15 (26%) RCD1=8 (19%) RCD2=7 (46%)	RCD1=80% RCD2=45%
Daum S, Eur J Gastr Hepatol 200950	Germany	Total=32 RCD1=23 RCD2=9	Total=4 (12%) RCD1=0 RCD2=4 (44%)	Total=8 (25%) RCD1=4 (17%) RCD2=4 (44%)	RCD1=90% RCD2=53%
Roshan B, Am J Gastroenterol 201145	USA	Total=29 RCD1=24 RCD2=5	Total=2 RCD1=0 RCD2=2	Total=2 (EATL) RCD1=0 RCD2=2	RCD1=100%* RCD2=60%* (*2-year mortality)
Nasr I, Nutrients 2015*73	UK	Total=30 RCD1=0 RCD2=30	None	None	All alive

*The only previous UK study assessing RCD2 reported a 5-year survival of 100%⁷³; however, the diagnosis of RCD2 was made solely on TCR clonality which is unreliable. Given the disparity of outcome in this UK study, and by comparison to the international literature, it is likely that the patients in this cohort may have been predominantly RCD1.

EATL, enteropathy-associated T-cell lymphoma; RCD, refractory coeliac disease.

What is the National Health Service England’s Rare Diseases Collaborative Network for Non-Responsive and Refractory Coeliac Disease?

Sheffield (lead centre) and Cambridge have been approved as National Health Service (NHS) England’s Rare Diseases Collaborative Network (RDCN) for Non-Responsive and Refractory Coeliac Disease. Clinicians who have any cases of suspected RCD (type 1 or 2) can contact us on david.sanders@sth.nhs.uk or jeremy.woodward@addenbrookes.nhs.uk. We will then provide clinical support at the level that the referring clinician would wish. This can be by reviewing notes, histological review, telephone consultation (with patient or clinician) or by seeing the patient, undertaking small bowel biopsy (for monoclonal lymphocyte population) and any other investigations (figure 1). Therapeutic intervention can be provided by the RDCN or the referring site. Our aim is to ensure that patients across England have standardised care and that we can support our colleagues in a collaborative manner. We have created a National Register to provide prospective outcome and treatment data on such patients analogous to Intestinal Transplant Registry,⁷⁴ which will allow us to consider novel treatments such as IL-15 or aHSCT for these patients.

In conclusion, RCD is a rare condition and it is both difficult and crucial to discriminate between RCD1 and RCD2. We would respectfully suggest that clinicians seeing any cases of RCD should make early contact with their National Centre to try and coordinate and optimise the best management options for their patients. This will also give these patients access to novel therapies which may in time reduce the high mortality of this disease.

Licence for publication:

The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive licence (or non-exclusive for government employees) on a worldwide basis to the BMJ Publishing Group Ltd to permit this article (if accepted) to be published in FG and any other BMJPGL products and sublicences such use and exploit all subsidiary rights, as set out in our licence (http://group.bmj.com/products/journals/instructions-for-authors/licence-forms).