The management of refractory coeliac disease

Abstract

A significant proportion of patients with coeliac disease are ‘nonresponsive’ to gluten withdrawal. Most cases of nonresponsive coeliac disease are due to persisting gluten ingestion. Refractory coeliac disease (RCD) is currently defined by persistent symptoms and signs of malabsorption after gluten exclusion for 12 months with ongoing intestinal villous atrophy. Primary (without initial response to diet) and secondary (relapse following response to diet) RCD is recognized. RCD is further classified as type I or type II based on the absence or presence of a population of aberrant intestinal lymphocytes. Quality of dietetic advice and support is fundamental, and lack of objective corroboration of gluten exclusion may result in over-identification of RCD I, particularly in those cases with persisting antibody responses. Over-reliance on lymphocyte clonality similarly may result in over-diagnosis of RCD II which requires careful quantification of aberrant lymphocyte populations. Management of RCD should be undertaken in specialist centres. It requires initial intensive dietary supervision, strict gluten exclusion and subsequent re-evaluation. There is currently insufficient evidence to recommend specific treatments. Steroids are often used in both RCD I and II (albeit with little objective evidence of benefit in RCD II), and azathioprine as steroid-sparing therapy in RCD I. There is growing evidence for the use of cladribine in RCD II with autologous stem cell transplantation in nonresponders, but this requires further multicentre evaluation. There remains considerable controversy regarding the diagnosis, treatment and surveillance of RCD: international consensus in these areas is urgently required to facilitate future therapeutic advances.

Current understanding of coeliac disease

Coeliac disease is a chronic small intestinal immune-mediated enteropathy caused by environmental exposure to gluten in genetically susceptible individuals [Ludvigsson et al. 2012]. Owing to advances in serological testing and the availability of endoscopic intestinal biopsy, the prevalence of coeliac disease is currently thought to be around 0.6–1% in European populations [Biagi et al. 2010; Walker et al. 2010]. Coeliac disease may present with ‘classical’ symptoms of malabsorption: weight loss and diarrhoea or ‘failure to thrive’ in infants. However, an array of diverse symptomatic presentations may lead to the diagnosis including ataxia, headaches, fatigue, muscle weakness, bone fractures, abdominal pain, bloating, nausea, constipation, infertility, treatment refractory hypothyroidism, skin rash, hair loss or nail changes [Dewar and Ciclitra, 2005; Green and Cellier, 2007]. Patients may be subjectively asymptomatic at presentation, with the diagnosis following screening of family members or rejection for blood donation, and it is likely that the majority of patients with coeliac disease remain undiagnosed throughout life [Wahab et al. 2002a].

Peptides derived from gluten, the water insoluble protein fraction of wheat, barley or rye endosperm, trigger the immune response in susceptible individuals. Gluten peptides are relatively indigestible by human endoluminal proteases. The 33-mer [Shan et al. 2002] and 17-mer [Camarca et al. 2009] oligopeptides of Gliadin contain toxic epitopes that are de-amidated by tissue transglutaminase and presented to the mucosal immune system by HLA DQ2 or 8 thereby eliciting an inflammatory cytokine response that results in the hallmark epithelial damage [Green and Cellier, 2007]. Recent studies have highlighted additional peptide sequences that initiate innate immune cytotoxic responses within the epithelium [Maiuri et al. 2003a, 2003b] and increase intestinal permeability via expression of zonulin [Fassano, 2011], facilitating passage of the large peptide fragments to the lamina propria.

The diagnosis of coeliac disease is made by clinicopathological correlation of duodenal biopsy appearances, clinical presentation and typical antibody responses [Sollid and Lundin, 2009]. Standard serological tests, the anti-endomysial antibody (anti-EMA) indirect immunofluorescence test [Chorzelski et al. 1983], and the anti-tissue transglutaminase (anti-TTG) enzyme-linked immunosorbent assay (ELISA) [Dieterich et al. 1997] have a high reported sensitivity for diagnosis [Leffler and Schuppan, 2010]. However, recent studies suggest that the sensitivity may be as low as 85–90% in adults [Dickey et al. 2000a; Hopper et al. 2008] Given the wide range of nonspecific symptomatic presentations (that overlap with common conditions such as irritable bowel syndrome) and the existence of seronegative disease, diagnosis can be challenging in cases with minimal histological change. In such situations, HLA haplotype may help to exclude the diagnosis [Hadithi et al. 2007]. The high specificity of a significantly elevated TTG titre has recently led the European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) to promote diagnosis in children based solely on symptoms and serology without histopathological corroboration [Husby et al. 2012].

Coeliac disease is currently managed by lifelong withdrawal of gluten from the diet and can result in complete recovery of the intestinal mucosa [Wahab et al. 2002b; Hutchinson et al. 2010; Rubio-Tapia et al. 2010]. It is likely that there is a dosage effect of the gluten and that the threshold for inflammation varies between individuals and possibly also over time. Therefore, some individuals may be able to tolerate a small amount of gluten in the diet without manifesting mucosal villous atrophy, whilst others may be exquisitely sensitive, even to trace amounts of gluten ingested as a result of contamination during food preparation. Clearly the length of time for mucosal recovery to occur depends on the patient’s learning curve for dietary gluten exclusion. In a study performed in Cambridge, approximately 25% of patients reverted to normal intestinal mucosa within a year, with a further 30% showing minor inflammation without villous atrophy [Corbett et al. 2012]. In those with ongoing villous atrophy, similar proportions respond to more stringent dietary gluten exclusion with mucosal recovery [Sharkey et al. 2012]. Anti-TTG antibody titres decline rapidly in the majority of coeliac patients excluding gluten effectively [Sugai et al. 2010], however levels may be sustained for 1–2 years in a small proportion of patients. A secondary increase in antibody titre is likely to suggest dietary lapse, however serological tests have a very low sensitivity in this setting [Dickey et al. 2000b].

Definitions of refractory coeliac disease

Undoubtedly there exists a small group of patients who do not respond to dietary exclusion of gluten, in whom a diagnosis of ‘refractory’ coeliac disease (RCD) may be entertained. RCD is currently defined as follows:

‘Persistent or recurrent malabsorptive symptoms and signs with villous atrophy despite a strict gluten free diet for more than 12 months’ [Ludvigsson et al. 2012]

The definition embodies the possibility of primarily refractory disease, which does not appear to respond to gluten restriction at all, and secondarily refractory disease, which develops after remission induced by diet. In applying the cutoff of 12 months it should be borne in mind that a substantial proportion (approximately 40%) of coeliac patients without refractory disease still have villous atrophy after 1 year despite apparent adherence to a gluten-free diet. It is notable that of the three axes of diagnosis – symptoms, histology and serology, the above definition of RCD requires only two, omitting the requirement for negative serology.

Symptomatically refractory coeliac disease (‘nonresponsive’ coeliac disease)

Patients diagnosed with coeliac disease may continue to experience symptoms despite being on a gluten-free diet [O’Mahony et al. 1996; Mooney et al. 2012]. The large number of possible causes include the following.

1. Incomplete dietary exclusion of gluten. Gluten is present in many foods and processed meals and even intelligent, well-motivated patients may have great difficulty in excluding it fully from their diet. Elderly patients may be dependent on others for their shopping, and impaired eyesight may limit the ability to read food labels. Commercial food outlets and restaurants are highly variable in their ability to cater for coeliac patients who run the risk of substantial gluten contamination when eating away from home. Similarly, those who travel frequently may have to cope with explaining their dietary requirement in a different language and culture. There is relatively little information available about the amounts of gluten ingested in different cultural settings. Careful dietetic review with patient-completed food records may help to identify potential sources of gluten contamination, but there is currently no reliable objective test to confirm complete gluten exclusion. In the future, it may be that measurement of the digestion-resistant 33-mer peptide in faeces will provide confirmation of dietary lapses [Comino et al. 2012]. Approximately 45% of patients nonresponsive to gluten withdrawal are thought to be noncompliant with the diet, in half of these cases through inadvertent gluten ingestion [Dewar et al. 2012].

2. Differential sensitivity to gluten. Whilst most patients are able to tolerate the trace amounts of gluten (10–20 ppm) present in some wheat-starch-based foods, there does appear to be a range of sensitivity to gluten and others may have to exclude all wheat-containing products as well. The threshold effect may be genetically determined due to the number of HLA susceptibility alleles present [Al Toma et al. 2006a] or known mutations such as in MYO9b [Wolters et al. 2007]. Many of the gene candidates for coeliac disease identified through genome-wide scans are immune modulators and it is likely that they will produce a spectrum of disease severity rather than an ‘all-or-nothing’ effect [Tjon et al. 2010].

3. Symptoms unrelated to gluten. The symptomatic presentation of coeliac disease is broad and overlaps that of other conditions. For instance, bloating and abdominal pain may be due to irritable bowel syndrome [Sanders et al. 2001] and diarrhoea could be due to bile salt malabsorption and be unaffected by gluten withdrawal. Coeliac disease can be entirely asymptomatic and diagnosed only as a result of symptoms due to another condition, for example, colonic cancer may be missed in a patient diagnosed with coeliac disease on the basis of iron deficiency.

4. Secondary causes of symptoms. Coeliac disease is also associated with secondary lactose intolerance that may persist despite villous recovery [Leffler et al. 2007]. Other conditions that can cause gastrointestinal disturbance as a result of coeliac disease include lymphocytic colitis [Dewar et al. 2012] (which may respond slowly or poorly to gluten withdrawal) and pancreatic exocrine deficiency [Leeds et al. 2007], as a result of impaired enteroendocrine responses to luminal nutrients.

5. Associated conditions. Patients with coeliac disease are also prone to other autoimmune conditions which will not respond to gluten withdrawal; examples include Addison’s disease that may present with fatigue, weight loss and diarrhoea [Elfstrom et al. 2007] or autoimmune gastritis resulting in iron and vitamin B12 deficiency [Dickey, 2002].

6. Incorrect diagnosis of coeliac disease. The diagnosis of coeliac disease is not always straightforward and may be made erroneously in a substantial number of cases [Biagi et al. 2009]. Lymphocytic duodenosis and mild grades of villous atrophy can occur in a variety of conditions including hypersensitivity, infection, inflammatory bowel disease and drug reactions, most notably with nonsteroidal anti-inflammatory drugs (NSAIDs) [Kakar et al. 2003; Memeo et al. 2005].

The majority of nonresponsive coeliac disease is therefore not RCD.

Histological RCD

Complete recovery of the intestinal mucosa occurs in the minority of coeliac patients on a gluten-free diet [Wahab et al. 2002b; Rubio-Tapia et al. 2010]. As the initial diagnosis of coeliac disease relies on the presence of villous atrophy rather than lymphocytosis of the epithelium or lamina propria, lesser degrees of persisting histological damage on follow up are usually discounted. Given that complete recovery is possible with gluten exclusion, it is likely that persisting minor grades of inflammation are due to low-level or intermittent gluten exposure. Intensive dietetic input for patients with persisting villous atrophy can result in normalization of the histological appearances [Sharkey et al. 2012], however a small number of patients fail to respond. If these patients also experience symptoms and signs of malabsorption, then they would be considered to fulfil the definition of RCD.

However, coeliac disease may present with no apparent symptoms at all. There is no correlation of the severity of symptoms experienced with the degree of histological damage and it is not uncommon for asymptomatic patients with no signs of malabsorption to manifest almost total villous atrophy on duodenal biopsy at presentation. Therefore, a subset of patients with coeliac disease may have ongoing villous atrophy, but without symptoms or signs of malabsorption and thereby not meet the diagnostic criteria for RCD. Unfortunately the definition [Ludvigsson et al. 2012] does not state the symptoms or signs required for the diagnosis of RCD to be made and this remains an area of disparity between reported series. There is evidence to suggest that asymptomatic histologically refractory disease is associated with poor outcomes [Rubio-Tapia et al. 2010; Kaukinen et al. 2007].

Histologically refractory disease is currently under-recognized as follow-up biopsies are not routinely performed in all centres.

Serological RCD

Titres of anti-TTG antibodies fall rapidly on institution of a gluten-free diet and normalize in the majority of individuals, despite persisting histological changes in a large proportion [Sugai et al. 2010]. The sensitivity of the TTG titre for detection of ongoing villous atrophy on follow up is poor. A small proportion of patients show persisting elevated TTG titres. In the majority of such cases, the antibody titre and associated villous atrophy are due to ongoing gluten ingestion and expert dietetic input can resolve both. However, a small number of patients demonstrate persistently elevated antibody titres despite normalization of villous architecture, and therefore similarly, a proportion may have ongoing antibody responses with refractory histology despite gluten exclusion. In these cases it is currently impossible to know whether the patient is fully excluding gluten except by subjective reporting. There is significant disparity between centres regarding the proportion of RCD patients with persisting seropositivity, from 0% to 59%, and it is highly likely that in the latter case, patients with ongoing gluten ingestion were misclassified [Roshan et al. 2011].

Classification of RCD

The diagnosis of RCD clearly remains controversial and the majority of cases thought to have RCD probably have either ongoing gluten ingestion or a very high sensitivity to trace amounts of gluten that are difficult or impossible to exclude from the diet [Dewar et al. 2012]. However, analysis of the cell populations in duodenal mucosa in patients with RCD led to the discovery of an aberrant lymphocyte immunophenotype in a subset of patients that has helped to define the condition [Badgi et al. 1999; Cellier et al. 1998, 2000]. This atypical population bears remarkable similarity to cells that have transformed into enteropathy-associated T-cell lymphoma (EATL), lacking surface expression of CD3 and CD8, but retaining intracellular CD3 [Alfsen et al. 1989]. Remarkably these cells also demonstrate clonality of T-cell receptor (TCR) rearrangement and therefore resemble a neoplastic or preneoplastic population. The absence or presence of these aberrant lymphocytes within the epithelium is the basis of the classification of RCD into type I and type II, respectively. The classification into type I and type II RCD is not always straightforward. The same aberrant intra-epithelial lymphocyte (IEL) phenotype is present in patients at diagnosis and also those with poor dietary compliance and ongoing villous atrophy. Transient TCR clonality can be detected in patients at diagnosis and with poor compliance [Prisco et al. 1997; Ubiali et al. 2007; Liu et al. 2010]. The background IEL TCR repertoire is oligoclonal and therefore if only small amounts of DNA are isolated from biopsies, it may be possible for a dominant clone to give the false appearance of monoclonality in this setting. On the other hand, detection of a clonal population requires at least 10% of cells to express the same TCR, which may be overlooked in the oligoclonal environment.

Controversy remains over the optimal technique for detecting the aberrant IEL phenotype with some groups strongly proposing flow cytometry of cell preparations from fresh biopsies [Verbeek et al. 2008; Leon, 2011], whilst others rely on immunohistochemical staining for CD3 and CD8 on paraffin embedded sections [Patey-Mariaud de Serre et al. 2000; Liu et al. 2010]. The benefit of flow cytometry is that it allows for easy identification and quantification of cells lacking surface expression of CD3 and CD8 with intracellular CD3 [Sanchez-Munoz et al. 2008]. However, methodological difficulties and contamination with lamina propria lymphocytes may confound the proportional quantification of the aberrant cells. Two-colour (CD3, CD8) immunohistochemistry allows localization of cells to the epithelium but does not discriminate between intracellular and surface CD3 expression. Using flow cytometry, a proportion of aberrant IELs greater than 20% is thought diagnostic of RCD type II [Verbeek et al. 2008], whilst more than 40% [Liu et al. 2010] is the cutoff using immunohistochemistry.

Epidemiology and natural history of RCD

Given the diagnostic difficulties and the reliance on inexact techniques for assessing dietary compliance, it is not surprising that the prevalence of RCD varies widely between studies – from 0.6% to 4% [West, 2009; Roshan et al. 2011]. These studies are based on relatively small numbers of patients and are often skewed by reporting from established referral centres. The true prevalence is likely to be extremely low, and only represent about 10% of referrals for suspected RCD [Dewar et al. 2012]. The ratio of females to males diagnosed with RCD is around 3:1 [Al Toma et al. 2007a; Malamut et al. 2009; Rubio-Tapia et al. 2009], similar to the ratio in responsive coeliac disease, however the ratio is reversed in patients with EATL [Rubio-Tapia et al. 2009]. Childhood cases of RCD have been reported [Mubarak et al. 2011], but the peak age of diagnosis is from the sixth decade onwards [Rubio-Tapia and Murray, 2010].

Retrospective analysis of archived biopsies in patients later diagnosed with RCD II or EATL has shown a progressive increase in aberrant IELs and development of clonality suggesting that progression of RCD I to RCD II occurs in a proportion of cases [Liu et al. 2010].

Both RCD I and RCD II are associated with worse outcomes than responsive coeliac disease [Al Toma et al. 2007a; Malamut et al. 2009; Rubio-Tapia et al. 2009; Maurino et al. 2006]. Progression to EATL has been reported rarely in RCD I, but occurs in 32–67% of those with RCD II (thereby lending credence to considering this condition to be preneoplastic), and it accounts for the 5-year survival of 44–58% following this diagnosis [Rubio-Tapia and Murray, 2010] as EATL carries such a poor prognosis following presentation [Gale et al. 2000].

Aetiology of RCD

The reasons for dietary refractoriness in RCD remain unclear. However, recent insights into innate immune mechanisms in coeliac disease have highlighted the key role of interleukin (IL)-15. This cytokine is produced by enterocytes and mononuclear cells in the intestine, and shares structural similarity with IL-2, signalling via common βγ receptor chains with a specific IL-15Rα receptor to complete the trimer. IL-15 promotes the survival of IELs via an anti-apoptotic pathway [Malamut et al. 2010] and can stimulate proliferation of IELs at high concentration. It also activates direct cytolytic effector functions in IELs via the induction of expression of natural-killer (NK) cell receptors on the IELs and stress proteins on enterocytes. Furthermore, this cytokine appears to bridge innate and adaptive responses by blocking immunosuppressive responses via JNK activation and subsequent Smad3 inhibition, thereby attenuating transforming growth factor (TGF)β signalling and stimulating the production of interferon (IFN)-γ [Meresse et al. 2009]. IL-15 synthesis and release is triggered by a number of factors including innate recognition of double-stranded RNA viruses by the Toll-like receptor 3, upregulation of IFN-α and (via as yet undiscovered mechanisms) a nonimmunostimulatory gliadin peptide fragment (p31-49). The key role of IL-15 in enteropathy is underpinned by the dramatic effects of experimental over-expression in murine intestine [Yokoyama et al. 2009] and the abrogation of IFN-γ release by anti-IL-15 antibodies in ex vivo organ culture of intestine from coeliac patients [Malamut et al. 2010]. Persistent over-expression of IL-15 in the intestine of patients with refractory disease may therefore account for the pathological features, but the trigger remains to be elucidated.

Clinical presentation of RCD

RCD classically presents with new onset of symptoms of diarrhoea, abdominal pain, weight loss and malabsorption following a previous diagnosis of coeliac disease and good response to diet with ongoing strict dietary exclusion of gluten. A primary presentation without prior response to gluten withdrawal is less common. Malabsorption can be severe and life threatening with multiple nutritional deficiencies and associated features. However, it is likely, as with responsive coeliac disease, that a range of symptomatic severity may occur, or that villous atrophy may be manifest by deficiency of single nutrients such as iron or folate. Ideally serological tests (anti-TTG or anti-EMA antibodies) should be negative, having been positive at diagnosis, however many patients with RCD have negative serology at the outset and others have still been considered refractory despite persisting antibody titres.

Imaging may reveal a thickened proximal intestinal wall with mesenteric lymphadenopathy and rarely, lymph node cavitation. Villous atrophy with characteristic visual features of scalloping and notching of duodenal folds may be present on endoscopy, and scattered superficial ulcers indicative of ‘ulcerative jejunoileitis’ may be seen.

In addition to a clonal aberrant IEL population in intestinal biopsies, similar cells may be detectable in the peripheral blood in patients with RCD II, and also in bone marrow leading occasionally to haemophagocytosis (with a corresponding high ferritin level) despite a lack of identifiable intestinal lymphoma.

Investigation of RCD

Persisting or recurrent symptoms of coeliac disease should prompt a full reassessment of the condition with dietary assessment for potential sources of gluten, serology (anti-TTG or anti-EMA antibodies) and duodenal biopsy [Abdulkarim et al. 2002]. The certainty of the initial diagnosis needs to be addressed and in patients with atypical presentations, a check of HLA DQ haplotype may be instructive. Other enteropathies including those due to viral aetiologies (notably norovirus), auto-immune enteropathy and common variable immunodeficiency should be considered. A full nutritional blood screen (including vitamin D, folate, vitamin B12, iron indices, liver blood tests, calcium and magnesium, full blood count and inflammatory markers) should be sent.

If there is no evidence of villous atrophy on the duodenal biopsy, then other causes of similar symptoms should be considered and ruled out. In the presence of persisting villous atrophy, very detailed and careful scrutiny of the diet needs to be undertaken by a dietician expert in managing the gluten-free diet. A food diary may need to be completed. More stringent dietary measures should be instituted such as cutting out oats, barley malt extract and foods permitted by Codex Alimentarius standards with low gluten content (1–20 ppm). Such dietary measures are all the more important if there is persisting positivity of serological tests which usually indicate gluten ingestion, and should be followed by reassessment of the patient prior to confirming a diagnosis of RCD.

Once the possibility of persisting gluten ingestion has been ruled out as far as possible (and bearing in mind the likelihood that gluten may still be the underlying cause of the refractoriness nevertheless), then the biopsies should be subjected to immunohistochemical and molecular analysis to determine whether an aberrant and/or clonal lymphocyte population is present. An alternative would be to take further, fresh biopsies for flow cytometry. A baseline bone density scan should be performed if not carried out recently.

If a diagnosis of RCD II is entertained by the presence of aberrant lymphocytes, then peripheral blood lymphocyte subsets should be analysed by flow cytometry for the presence of the same population. Any abnormalities in a peripheral blood film or a high ferritin level indicative of haemophagocytosis should mandate bone marrow aspirate and trephine. Cross-sectional imaging by CT or MRI for the presence of lymphadenopathy or intestinal tumours should be carried out and capsule or balloon enteroscopy should be performed to attempt to diagnose prevalent EATL. MR enteroclysis may be used to identify mucosal changes in both RCD II and lymphoma [van Weyenberg et al. 2011]. ¹⁸F-deoxyglucose positron emission tomography (PET) should be requested if a high index of suspicion for lymphoma remains [Hadithi et al. 2006].

Management of RCD

The diagnosis of RCD may be clear when there is evidence of severe malabsorption, villous atrophy, aberrant and clonal IELs on biopsy and there is good supportive evidence for the initial diagnosis of coeliac disease, excellent dietary compliance and subsequent loss of TTG or EMA seropositivity (see Table 1). In all other cases, the diagnosis should be considered provisional as dietary gluten exclusion cannot yet be objectively corroborated.

Table 1.

Characteristics, therapeutic options and outcomes in refractory coeliac disease (RCD).

Condition	IEL phenotype	T-cell receptor clonality	Therapeutic options	Outcomes
Normal	>98% CD3+CD8+	Oligoclonal	N/A	N/A
Untreated coeliac	Increased numbers of γδ+TCR CD3+8+	Oligoclonal, transient clonal population may be present	Gluten-free diet (novel therapies including zonulin antagonists, oral endopeptidases and immunotherapy under trial)	Evidence to suggest overall reduced life expectancy
Type I RCD	As for untreated coeliac disease	As for untreated coeliac disease	Strict gluten restriction; oral steroid; budesonide; azathioprine; nutrition support	5-year survival 93%; progression to EATL 14% at 3 years
Type II RCD	>20% (by flow cytometry) or >40% (by immunohistochemistry) surface CD3-, intracellular CD3+, CD8+	Clonal TCR γ or β rearrangement	Strict gluten restriction; cladribine; ASCT	5-year survival 44–58%; progression to EATL 33–67% at 5 years

ASCT, autologous stem cell transplantation; EATL, enteropathy-associated T-cell lymphoma; IEL, intra-epithelial lymphocyte; TCR, T-cell receptor.

Just as it is the mainstay of responsive coeliac disease, strict dietary compliance is essential in the management of RCD. In many cases of RCD, symptomatic and histological recovery may occur, raising the possibility of latent gluten sources or a high sensitivity to trace amounts of gluten. Some studies have pursued the use of liquid diets or elemental feed to demonstrate responses in RCD as a result of total gluten withdrawal [Mandel and Mayberry, 2001; Olaussen et al. 2005]. There is also evidence for the resolution of clonal IEL populations with strict dietary intervention [Liu et al. 2010]. This suggests that pre-autonomous phases of IEL proliferation and activation may be reversed by antigen exclusion.

Some cases of RCD I and a high proportion of those with symptomatic malabsorption with RCD II require intravenous nutrition support due to intestinal mucosal failure.

More specific therapies for the treatment of RCD are based on anecdotal reports, small uncontrolled open-label cohort studies and case series. There is therefore no robust clinical data on which to recommend treatment strategies and it is advisable for patients to be referred to regional centres with expertise in the management of RCD if further therapeutic intervention is required. A major concern of treating patients with RCD is the possibility of subjecting potentially diet responsive patients to toxic therapy due to misclassification as RCD resulting from inadequate dietetic assessment and involvement.

Steroids

Steroids have been used frequently as first-line immunosuppression in RCD. A series of 47 patients from Amsterdam (including both RCD I and II) were treated with oral prednisolone at a dose of 40 mg/day tapering to 10 mg a day after 6 weeks and thereafter to 0 or 2.5mg after 3 months [Al-Toma et al. 2007a]. A similar-sized French series noted substantial benefits with clinical response in 90% of RCD I and 77% of RCD II patients [Malamut et al. 2009]. However, few patients demonstrated histological recovery (40% of RCD I, 33% of RCD II), and patients became steroid dependent with prolonged treatment (mean 40 ± 32 months for RCD I). Clinical response was loosely defined as a 50% reduction in stool frequency and/or 50% regain of lost weight. Side effects of such prolonged steroid use, including bone density measurements, were not reported. In contrast, a cohort of seven patients from Argentina failed to respond to steroid therapy [Maurino et al. 2002].

In an attempt to reduce the potential toxicity of long-term systemic steroid use, budesonide has been used in a number of patients. This is a potent nonhalogenated corticosteroid with high first-pass metabolism resulting in systemic bioavailability of only 12% after oral ingestion in controls [Edsbacker et al. 2003]. It is formulated to deliver topical steroid to the distal ileum and colon, however approximately 30% is released and absorbed in the upper intestine [Edsbacker et al. 2002, 2003] and therefore exhibits suboptimal delivery for treating the proximal enteropathy of coeliac disease. Nevertheless, budesonide therapy has been reported to benefit patients with both RCD I and II [Brar et al. 2007] or just RCD I [Daum et al. 2006], based on subjective measures and bowel frequency. However, again there was no improvement in histological features noted, and three out of four treated patients subsequently received azathioprine. Worryingly, osteoporosis was detected in five of six patients tested [Daum et al. 2006] prior to steroid use. Furthermore, the loose criteria used for diagnosis of RCD suggests that many of these patients were probably slow to respond to diet rather than having RCD I. In addition, a higher systemic bioavailability is seen in Crohn’s disease than in controls (21% versus 12%) [Edsbacker et al. 2003] and has not been tested in coeliac disease: a systemic corticosteroid effect of budesonide cannot therefore be excluded.

Purine analogue antimetabolites

Azathioprine has been used for the treatment of RCD I and II without the complications associated with long-term steroids, but with high levels of withdrawal due to side effects [Enns et al. 1997; Vaidya et al. 1999]. Eight out of 10 patients with RCD I treated with induction steroids and subsequent azathioprine at 2 mg/kg in one series showed histological improvement and symptomatic benefit, however a high proportion of those with RCD II progressed to EATL leading to concern about the possibility of a detrimental effect [Goerres et al. 2003]. Similarly, a small study from Latin America of five patients (1 RCD I and 4 RCD II) demonstrated histological and symptomatic improvement [Maurino et al. 2002]. This was not however the case in one other series (six patients; one RCD I and five RCD II) where three patients experiences clinical improvement but there was no evident improvement in histology [Malamut et al. 2009].

In view of the variable metabolism of azathioprine, the active metabolite, thioguanine, has been tested in 13 RCD I patients with good tolerability, even in those with poor prior tolerability of azathioprine [Tack et al. 2012]. All patients improved subjectively although there was no significant difference in biochemical markers, and the majority showed a histological response.

Cladribine (2chlorodeoxyadenosine) is an adenosine nucleoside analogue that has been used in the treatment of hairy cell leukaemia [Jehn et al. 2004]. By inhibiting RNA and DNA synthesis it has an effect on proliferating cells, but also induces lymphocyte apoptosis and may have long lasting effects on T-lymphocyte cell numbers. An oral preparation has recently been withdrawn and studies of its use in RCD II use an intravenous infusion of 0.1 mg/kg/day over 5 consecutive days [Wahab et al. 2000; Al Toma et al. 2006b; Tack et al. 2011a].

The largest reported series of 32 patients [Tack et al. 2011] unfortunately changed the criteria for diagnosis of RCD II during the study by including patients without a clonal IEL population, using flow-cytometry-based immunophenotype alone as the differentiator between RCD subtypes [Verbeek et al. 2008; Al Toma et al. 2006]. In addition, some of the patients were previously treated with immunosuppression whilst others were naïve to such therapy. Clinical remission was reported in 81% of patients, with histological recovery in 47% and a reduction in aberrant IELs of more than 20% in 41%. However, these patients still met diagnostic criteria for RCD II on the basis of the high remaining proportion of aberrant lymphocytes after treatment. Clinical responders in this study had a significantly increased survival rate and lower progression to lymphoma without any other identifiable factor on multivariate analysis.

Other immunosuppressant agents

Small numbers of patients have been reported in individual cases to have been treated partially successfully with other therapies including Campath (anti CD-52 monoclonal antibody) [Vivas et al. 2006; Lundin et al. 2006; Malamut et al. 2009], infliximab (anti-TNFα antibody) [Turner et al. 2005; Chaudhary and Ghosh, 2005; Gillett et al. 2002; Constantino et al. 2008; Schmidt et al. 2009; Malamut et al. 2009], cyclosporin A [Wahab et al. 2000; Rolny et al. 1999; Malamut et al. 2009] and methotrexate [Malamut et al. 2009]. In general, symptomatic improvement has been demonstrated without significant resolution of histological changes. A pilot study of 10 patients treated with recombinant IL-10 [Mulder et al. 2001] failed to demonstrate any significant benefit in terms of histological or clinical response.

Autologous stem cell transplantation

The dire prognosis of RCD II has resulted in the consideration of autologous stem cell transplantation (ASCT) [Al Toma et al. 2007b]. In the latest update from Amsterdam (where this treatment is being pioneered for this condition), 13 patients with RCD II unresponsive to cladribine therapy have undergone ASCT [Tack et al. 2011b] with three deaths: one transplant related, and the others due to encephalitis and EATL respectively. A 4-year survival of 66% in this cohort compares with the death of all of the other five patients with unresponsive RCD II who were unable to undergo ASCT within 1 year. Of these, only two were scheduled for therapy as the others developed EATL before therapy could commence. Clinical improvement and histological recovery occurred, but only in two out of six patients with villous atrophy prior to transplantation after 1 year, and the treatment did not eradicate the aberrant IEL population which appeared to increase. The effect of treatment on clonality was not reported, and one patient who underwent ASCT still developed EATL. The preneoplastic aberrant IEL population therefore appears highly resistant to ablation by the conditioning regimen of fludarabine and melphalan, and consideration of other agents may be required for this purpose.

Surveillance in RCD

Owing to the significant potential for progression to EATL, patients with RCD should undergo careful monitoring and surveillance if no lymphoma is found at presentation. The appropriate tests for EATL surveillance are not currently established but should include regular interval endoscopy and intestinal biopsy [Liu et al. 2010] to monitor villous morphology and clonality and quantification of aberrant IELs, as RCD I may progress to RCD II and patients with RCD I may develop EATL [Malamut et al. 2009]. The sensitivity of capsule endoscopy [Collins et al. 2012] or cross-sectional imaging for the detection of EATL in RCD II is unclear, as is the appropriate interval between tests. Furthermore, there is as yet insufficient data to suggest that early detection and treatment of EATL by surveillance improves the outcome when compared with symptomatic presentation. Clinical supervision in RCD should not be limited to EATL surveillance but should also include monitoring for nutritional deficiencies and bone density measurements.

Future prospects for the treatment of RCD

An anti-IL-15 monoclonal antibody (AMGEN 714) might be of considerable benefit in the management of RCD given the importance of this cytokine in driving the epithelial response [Malamut et al. 2010]. Unfortunately at the time of writing, this agent has been withdrawn from clinical trials as a result of lack of benefit in other clinical areas. Downstream signalling from the IL-15 receptor may be inhibited by using JAK3 blocking agents such as tofacitinib [Sandborn et al. 2012] (currently being trialled in rheumatoid disease and psoriasis), or stat-5 inhibitors currently under development.

New treatments for coeliac disease such as luminal endopeptidases [Bethune and Khosla, 2012], zonulin antagonists [Paterson et al. 2007] and peptide vaccination [Camarca et al. 2009] are yet to be evaluated but may provide benefit for patients with RCD, particularly those labelled as type I with high gluten sensitivity. Further understanding of the stimulation of the IL-15-driven response may permit additional measures to block the innate immune activation that results in epithelial damage in RCD II. Techniques, such as faecal identification of the proteolysis-resistant gliadin 33-mer [Comino et al. 2012], to identify continued gluten ingestion rather than relying on dietary records and recall would substantially clarify the diagnosis of RCD I and enable appropriate management.

However, a prerequisite of all future studies in this area is a greater degree of international consensus regarding the definition, investigation and classification of RCD, and collaboration in larger-scale trials of therapy.

Conclusions

A significant proportion of patients with coeliac disease are ‘nonresponsive’ to gluten withdrawal. Nonresponsive coeliac disease is due to persisting gluten ingestion rather than RCD in the majority of cases.

RCD is currently defined as the presence of persistent symptoms and signs of malabsorption after gluten exclusion for 12 months with ongoing intestinal villous atrophy. Primary (without initial response to diet) and secondary (relapse following response to diet) RCD is recognized. RCD is further classified as type I or type II based on the absence or presence of a population of aberrant lymphocytes in the intestinal epithelium.

The quality of dietetic advice and support is fundamental to the diagnosis of RCD, and lack of objective corroboration of gluten exclusion is likely to result in over-identification of RCD I, particularly in those cases with persisting antibody responses. Over-reliance on lymphocyte clonality similarly may result in over-diagnosis of RCD II which requires careful quantification of aberrant lymphocyte populations.

Management of RCD should be undertaken in centres specializing in RCD and requires initial intensive dietary supervision with strict gluten exclusion and subsequent re-evaluation. There is currently insufficient evidence to recommend specific therapeutic interventions. Steroids are often used as first-line therapy in both RCD I and II (but with little objective evidence of benefit in RCD II), and Azathioprine is used as steroid-sparing therapy in RCD I. There is growing evidence from one centre for the use of cladribine in RCD II with ASCT in nonresponders. However, formal multicentre collaborative evaluation is required before such interventions can be universally recommended.

There remains considerable controversy regarding the diagnosis, treatment and surveillance of patients with RCD and international consensus in these areas is urgently required in order to facilitate future therapeutic advances.