Optimising delivery of care in coeliac disease – comparison of the benefits of repeat biopsy and serological follow-up

Authors


Summary

Background

The majority of deleterious health consequences of coeliac disease (CD) are most likely to be secondary to intestinal inflammation; hence, mucosal recovery is a desirable goal of therapy. Follow-up in CD is controversial and serological response is often used as a surrogate for histological recovery.

Aims

To inform the clinical management of CD using comparative serological and histological data from a biopsy-driven pathway of care.

Methods

A retrospective analysis of the Cambridge Coeliac Clinic database of 595 patients routinely followed up by biopsy and serology.

Results

Paired biopsy results were available for 391 patients (15% seronegative). Persisting villous atrophy (VA) occurred in 182 patients (47%). The sensitivity of anti-tissue transglutaminase (TTG) antibody for ongoing VA was only 43.6%. Information on dietetic management and further biopsy to assess response was available for 94 initially unresponsive patients, in whom targeted dietetic intervention by removal of identified gluten sources or avoidance of trace amounts of gluten led to resolution of persistent VA in 50%. The effects of institution of a formal care pathway are analysed in 298 patients. Discharge to primary care and clinical management was facilitated by the information derived from repeat biopsy.

Conclusions

Serology appears to be a poor surrogate marker for mucosal recovery on a gluten-free diet; dietary assessment fails to identify a potential gluten source in many patients with ongoing villous atrophy. The benefits of re-biopsy on diet include stratification of patients with coeliac disease suitable for early discharge from secondary care or those requiring more intensive clinical management.

Introduction

Coeliac disease (CD) is a gluten-sensitive enteropathy characterised by histological changes in the small intestine.[1, 2] Although CD-related antibodies have been implicated in organ damage – notably in the skin in dermatitis herpetiformis – it is likely that the majority of clinical symptoms and signs as well as long-term complications and sequelae of CD are a consequence of the inflammatory damage to the intestinal mucosa.[3] Therefore, the optimal assessment of response to diet and prevention of CD-related complications should be the demonstration of mucosal recovery following the institution of a gluten-free diet (GFD).[4] However, determination of histological resolution requires a repeat endoscopy and biopsy, thereby exposing the patient to an additional uncomfortable experience that carries clinical risk, albeit minimal. Furthermore, the healthcare costs associated with a further endoscopy and histopathological examination of the biopsy specimens are not trivial, given the large numbers of patients involved. Routine re-biopsy to assess response to diet in adults is not currently a recommendation of the British Society of Gastroenterology,[5] American Gastroenterological Association,[6] World Gastroenterology Organisation[7] or American College of Gastroenterology[8] guidance as the goal of treatment – whether symptom relief or resolution of histopathological changes – is still undecided. There are currently little data to support a biopsy-driven strategy.

Although few clinics currently offer routine biopsy on diet, follow-up care is generally provided by most to support patients and determine the efficacy of the gluten-free diet.[9] Clinical practice in adult CD follow-up is highly variable, individual and with no markers of quality assurance.[10] A variety of different surrogate markers are used to assess the completeness of dietary gluten exclusion and response to diet. These include simple symptomatic improvement, reduction and/or normalisation of CD-specific antibody titres and structured dietetic interview.[11-13] Whilst all of these methods have their advocates, none has been shown to match the gold standard of histological reassessment[14] – and many of the studies proposing such surrogates do not use histological response as the comparator.

Similarly, there is no uniformity in the organisation of clinical services for the management and support of CD and there is little available guidance on the optimal approach.[4, 15-21] The quality of patient education at diagnosis,[21] availability of expert clinicians and dietitians[15] and regular follow-up in a specialist clinic[18] may all impact adherence to the diet and subsequent outcome. A number of dietitian- or nurse-led clinics have been established in the UK, often separate to follow-up appointments with the gastroenterologist, which may be sporadic and occur in generalist, rather than specialist, clinics.

In 2006, a specialist Coeliac clinic was established in Addenbrooke's Hospital, Cambridge. Patients follow a defined pathway of care (Figure 1) similar to that subsequently proposed by Haines et al. in 2008.[4] The basic principles are: diagnostic certainty with serology and biopsy at presentation, patient education (an introductory group talk given by consultant and dietitian), links with the local Coeliac UK group (present in clinic), appointments with dietitian and/or clinician at every visit, repeat biopsy at 9–12 months with bone density measurement in selected patients and early structured discharge to primary care follow-up using agreed shared care guidelines.

Figure 1.

The Cambridge Coeliac Pathway. TTG, Immunoglobulin A anti-tissue transglutaminase antibody; DEXA, dual energy X-ray absorbance; FBC, full blood count; U&E, urea, creatinine and electrolytes; LFT, liver functions tests; B12, Vitamin B12.

This report documents patient follow-up from this clinic comparing serological tests (anti-TTG antibody) with follow-up biopsy being used to direct further dietetic management.

Patients and methods

Patients attending the Adult Coeliac clinic at Addenbrooke's Hospital have been routinely followed up by serology and re-biopsy. In 2006, practice was consolidated in a clinical care pathway. The Cambridge Coeliac Pathway is outlined in Figure 1. This retrospective observational cohort study of 595 cases includes 298 cases diagnosed after the initiation of the pathway in March 2006. Patients were excluded from analysis if they were previously excluding gluten and diagnosed on the basis of a gluten challenge biopsy, or if initial biopsies did not show evidence of villous atrophy, even if subsequently diagnosed with CD.

Serological testing

Immunoglobulin (Ig) A Anti-tissue transglutaminase antibody (Anti-TTG) is measured by ELISA (Phadia) along with total IgA levels. IgA anti-EMA (Endomysial Antibody) indirect immunofluorescence is not tested routinely following an internal audit in 2002 of nearly 1400 samples, showing no additional diagnostic benefit in testing for anti-EMA antibodies as well as TTG. The subsequent cost saving allowed testing of total IgA alongside TTG. Deamidated gliadin antibody tests were not available during the time course of this study and were not therefore tested. Serology is measured prior to every clinic visit, usually within 4 weeks of the visit.

Small Intestinal Biopsy

Standard practice is to perform oesophagogastroduodenoscopy (OGD) and obtain four biopsies from the second part of the duodenum, in different quadrants because of the frequent patchy nature of the disease.[22] Duodenal bulb biopsy had not become a recommendation during the time course of this study and was therefore not routinely practised. Biopsies were analysed by dedicated gastrointestinal pathologists and a descriptive result using the terms used in the modified Marsh criteria[23, 24] was given. Any degree of villous atrophy was considered Marsh 3 or above or Corazza[25] grade B1 (for Marsh grade 3a and 3b, partial or subtotal villous atrophy) or Corazza grade B2 (for Marsh grade 3c, total villous atrophy). Isolated duodenal intraepithelial lymphocytosis was considered Marsh 1 and additional lamina propria infiltrate was considered Marsh 2; both of these correspond to Corazza A. Biopsies were obtained at initial referral to confirm the diagnosis and, according to the Cambridge Coeliac Pathway, after 9–12 months of gluten-free diet to assess histological response. Re-biopsy was not performed on patients who were deliberately non-adherent to the diet, but persisting raised antibody titres to tissue transglutaminase did not prevent or delay repeat biopsy.

Dietetic intervention

Patients are given a group introductory talk by a consultant gastroenterologist and specialist dietitian at diagnosis, and are then given an appointment to see the doctor and dietitian in clinic afterwards. Dietitians give introductory verbal and written information to patients (and their partners or relatives to include those involved in shopping, cooking and preparing their food) on a gluten-free diet, including suitable and unsuitable foods, the Codex standard, current labelling laws and reading food labels, cross-contamination, prescribable and nonprescribable gluten-free substitute products, common pitfalls of the diet and advice for when eating out and travelling abroad. Gluten-free food samples are provided and it is recommended that they join Coeliac UK (represented in clinic by local volunteers). Patients are provided with contact details for further specialist dietetic advice if required. The GP is asked to provide gluten-free foods on prescription. We do not recommend exclusion of pure oats at diagnosis. Initial follow-up dietary review is then scheduled 2 months later with further dietetic appointments scheduled on an individual basis.

Review after follow-up biopsy

Patients are seen in clinic by both a clinician and dietitian after their follow-up biopsy. If histological recovery has occurred (defined as an absence of villous atrophy, Marsh grade 0, 1, 2 or Corazza Grade A changes were acceptable), then no specific changes are recommended. If the biopsy shows persisting villous atrophy, regardless of serology or symptoms, they undergo dietetic review by clinic consultation. At this point, if non-adherence or a potential source of gluten ingestion (above the Codex standard) is identified, patients are given appropriate advice. If there is uncertainty, patients are asked to keep a detailed food diary for 1–2 weeks followed by a further Dietetic review to identify sources of gluten. If no contamination can be identified, a recommendation for a supersensitive diet (SSD) is made. In addition to the strict GFD the patient is already adhering to, this diet excludes foods deemed gluten-free based on Codex Alimentarius Standards.[26] The current standards, defining gluten-free food as containing less than 20 parts per million (20 ppm), were introduced in 2008 and became European law in 2012. Therefore, the majority of cases reported in this study were subject initially to less stringent standards (up to 200 ppm). The SSD thereby excludes Codex wheat starch and barley malt extract and pure oats in view of the controversial possibility of avenin immunotoxicity.

Further biopsy after second-line intervention

Following any of the above measures (dietary review, food diary, SSD), a third biopsy is considered to assess response.

Discharge from secondary care

An agreed shared care pathway has been developed with regional primary care providers to ensure annual follow-up appointments for patients discharged from the Coeliac clinic. Definitive discharge takes place if the patient is compliant with diet, has no ongoing symptoms or complications and has made an adequate histological response to diet. Normalisation of the biopsy prior to discharge is not a prerequisite if there is further evidence of improved dietary adherence following a biopsy showing only minor villous architectural changes.

Statistical analysis

Data analysis for serological accuracy was performed using standard techniques. Comparison of biopsy results at varying time periods was performed by chi-squared test and comparison of outcomes was performed using Fisher's exact test.

No patient-identifiable data were recorded in the analysis and, as a review of clinical practice, formal ethical permission was not considered necessary.

Results

Patient demographics

At the time of data collection, 632 patients were on the Cambridge Coeliac Disease database, of whom 595 were eligible to be included in the primary analysis. A total of 421 patients (70.8%) were female with an average age of 44 years at diagnosis; 174 patients were male with an average age of 48 years at diagnosis. Across all patients, the mean age was 46 years, range 1–91 years (s.d. 19.5).

Follow-up biopsy

Information on follow-up biopsy was available for 447 patients; 56 of them had no index biopsy, either because it was carried out elsewhere or predated current computerised records. It could be presumed that they were diagnostic (showing some degree of villous atrophy), but we have excluded them from further analysis, given the uncertainty. Of the resulting 391 patients with paired biopsies, 207 (53%) were re-biopsied within 12 months; a further 122 were re-biopsied within 12–24 months. In total, 84% (329/391) were re-biopsied within 24 months. Overall, the median time to second biopsy was 11 months (range 2–268 months); however, there was a notable difference following institution of the pathway:

The mean time to repeat biopsy prior to the institution of the care pathway was 30 months, with a median of 12 months and standard deviation 40.8 months. Following institution of the pathway, the mean time to repeat biopsy was 12 months, median 11 months and standard deviation 7.3 months.

Histological responses to GFD (Figure 2)

One hundred and two patients achieved complete normalisation of their mucosa on follow-up biopsy [normal villous height with no evidence of intraepithelial lymphocytosis (IELs) or lamina propria infiltrate (LPL)]; 54 of them were biopsied within 12 months and 83 were biopsied within 24 months. The noncumulative proportion of those biopsied at each time point, who achieved complete normalisation of duodenal mucosa (Marsh 0), is 26.1% at 12 months, 23.8% at 12–24 months and 30.6% after 24 months (P = 0.75). A total of 224 patients had an absence of villous atrophy on repeat biopsy (Marsh 0,1,2/Corazza A changes); 117 of them were biopsied within 12 months, 191 within 24 months. The noncumulative proportion of those biopsied at each time point showing normal mucosa or only minor changes is 56.5% at 12 months, 60.7% at 12–24 months and 53.2% after 24 months (P = 0.87).

Figure 2.

Timing of follow-up biopsy and histological responses: Percentage of patients with normal mucosa or minor changes on follow-up biopsy, according to the time period in which the biopsy was taken, <12 months, 12–24 months or >24 months since diagnosis.

Data were analysed separately for the 298 adult patients, diagnosed with CD in our institution between May 2006 and January 2012, following the institution of the care pathway.

Of this cohort, 9 patients were found not to have CD on follow-up testing, 2 had refractory disease (one type 1 and one without symptoms) and 10 patients died; therefore, the analysis is based on 277 patients. Seventy (25%) patients did not have follow-up biopsies for a variety of reasons, including moving out of area, being lost to follow-up or refusing. The data were therefore analysed for the 207 patients who had paired biopsy data available. Sixty-two (30%) patients in this cohort demonstrated normalisation of histological appearances on follow-up biopsy, and 127 (61%) demonstrated loss of villous atrophy with either mild or no residual inflammation present.

Serology at diagnosis and follow-up

Four hundred and nine patients had both initial biopsy and serology available at the time of diagnosis; 8 of this group were IgA-deficient (2%). Of the remaining 401 patients, 58 (15%) of these had negative anti-TTG antibody titres (<7 IU/mL) at the time of diagnosis. HLA haplotyping was available in 14 cases and was positive for HLA DQ2 or 8 in 13 cases. One patient who was HLA DQ2/8-negative had malabsorptive symptoms and villous atrophy, but did not undergo repeat biopsy on diet. Within the group of 343 patients who were seropositive at the time of diagnosis, 237 had results for both repeat biopsy and serology available (50 patients in this group were seronegative at diagnosis). Only 43.6% of those with persisting villous atrophy also had positive serology. Conversely, 25.2% without villous atrophy on follow-up had positive serology. Table 1 shows the sensitivity, specificity and predictive values for serology at follow-up biopsy.

Table 1. Comparison of serology and biopsy at follow-up: the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of anti-tissue transglutaminase for detecting ongoing villous atrophy are shown
 Villous atrophyNo villous atrophy 
Seropositive4136PPV = 66.9% (95% CI: 59–74.1%)
Seronegative53107NPV = 53.3% (95% CI 41.5–64.7%)
 Sensitivity = 43.6% (95% CI: 33.4–54.2%)Specificity = 74.8% (95% CI: 66.9–81.7%) 

The level of TTG at diagnosis also correlated with the histological severity (P = 0.016), but there was no correlation in TTG level and histology at follow-up (Table 2, Figure 3).

Table 2. TTG levels compared with biopsy findings by the modified Marsh criteria at diagnosis and follow-up
  1. TTG, anti-tissue transglutaminase antibody; ULN, upper limit of normal; Marsh 0 = normal mucosa, Marsh I = increased intraepithelial lymphocytes, II = crypt hyperplasia, III = villous atrophy, partial (IIIa), subtotal (IIIb) or total (IIIc).

TTG Level (ULN = 6 IU/mL)TTG<ULNTTGl-5xULNTTG5-10xULNTTG>10xULN
Diagnosis
Marsh IIIa18 (32%)22 (39%)9(16%)7(13%)
Marsh IIIb10(16%)22 (35%)8(13%)23 (36%)
Marsh IIIc22(13%)48 (29%)28(17%)70 (42%)
Follow-up
Marsh 064 (84%)11 (14%)0 (0%)1 (1%)
Marsh I44 (73%)15 (25%)0 (0%)1 (2%)
Marsh II26 (77%)8 (23%)0 (0%)0 (0%)
Marsh IIIa44 (72%)12 (20%)5 (8%)0 (0%)
Marsh IIIb18 (58%)8 (26%)2 (6%)3 (10%)
Marsh IIIc14 (56%)5 (20%)5 (20%)1 (4%)
Figure 3.

Correlation of TTG level at Diagnosis with Histological Severity. TTG, anti-tissue transglutaminase; ULN, upper limit of normal; Marsh IIIa = partial villous atrophy, IIIb = subtotal villous atrophy, IIIc = total villous atrophy.

Outcomes of those with persisting villous atrophy (Figure 4)

Of the patients with persisting villous atrophy on second biopsy, 94 had at least one further duodenal biopsy after intervention. In the majority (65/94, 69%), an overt or inadvertent source of gluten contamination had been identified and excluded prior to the third biopsy. Of these cases, 29 showed histological improvement (14 normal mucosa, 15 Marsh 1 or 2 changes). Twenty-nine patients had no obvious dietary indiscretions and started the SSD as described above; 18 (62%) of these patients were shown to have normal mucosa or minor changes on further biopsy. The 11 patients who still had VA following a SSD were deemed to be ‘histologically refractory’.

Figure 4.

Outcomes of patients with persisting villous atrophy on follow-up biopsy, focusing on those who had second-line intervention followed by a third biopsy to assess response to this. SSD, supersensitive diet, described in text; VA, villous atrophy.

Therefore, focused dietetic intervention may have contributed to improvement in histology in half of patients (47/94, 50%) who had persisting villous atrophy at the time of follow-up biopsy.

Discharge information

Discharge information was analysed for 234 patients diagnosed after institution of the care pathway with follow-up data of more than 18 months post-diagnosis. The reasons for patients without ongoing villous atrophy continuing to be seen in clinic appointments included ongoing gastrointestinal symptoms (n = 12), persistent raised TTG (n = 9), concerning bone density (n = 7), poor dietary adherence (n = 4) and other (n = 11).

The group of patients who had villous atrophy on their last biopsy included 36 patients who did not have a second biopsy performed and 66 who had ongoing villous atrophy on a repeat biopsy. Of the 36 who did not have a follow-up biopsy, 2 were discharged, 17 were lost to follow-up and 17 had ongoing clinic follow appointments. Thus, routine re-biopsy aids stratification of patients for early discharge from secondary care (Table 3).

Table 3. Proportion of patients discharged from secondary care according to last duodenal biopsy appearances
Villous atrophy on last biopsyDischargedLost to follow-upOngoing clinic appointments
  1. a

    P < 0.0001 (Fisher's exact test) compared to no villous atrophy.

No74 (61%)9 (7%)39 (32%)
Yes (either on follow-up biopsy or diagnostic biopsy with no follow-up biopsy taken)15(16%)a21 (22%)58 (62%)
Yes (on follow-up biopsy only)13 (22%)a4 (7%)41 (71%)

Discussion

There is considerable controversy over whether CD follow-up should be based on dietary interview, symptom response, loss of CD-related antibodies, histological recovery or a combination of these modalities.

This series report of 391 patients with CD is necessarily limited by its retrospective nature and potential bias where data are incomplete. However, by demonstrating results of follow-up in a ‘real-life’ setting, it provides useful insights to inform the optimal follow-up management of the condition that may not be applicable from controlled prospective studies.

Symptom relief is not reported in this study. Symptoms at diagnosis of CD in adults rarely present the ‘classical’ picture of diarrhoea and weight loss.[27] A panoply of symptoms of varying severity have been described at the onset of CD, and many patients experience only vague or low-level symptoms that they may not recognise until they are improved by therapy.[28] Worryingly, patients may be asymptomatic for many years or decades before ultimately presenting with a complication of CD;[29] therefore, symptom relief (as experienced early in the pursuit of a gluten-free diet by most patients[30]) may not prevent later complications.[31] Patients who have lived with undiagnosed CD for many years (on the basis of retrospective serological testing) have a reduced life expectancy[32] and also an increased risk of complications, such as osteoporosis[33] and enteropathy-associated T cell lymphoma (EATL).[34, 35] Furthermore, symptoms correlate poorly with villous atrophy. In a controlled study of gluten re-challenge, symptoms were absent in 22% of cases where significant villous atrophy occurred.[36] A large number of patients are asymptomatic or minimally symptomatic at presentation and clearly cannot be followed up using symptom relief as the main determinant of clinical response. In addition, many of the presenting symptoms of CD overlap with other common conditions such as irritable bowel syndrome, which may indeed also be present and not respond to gluten withdrawal. Whilst symptom relief is clearly of immediate importance to patients, the avoidance of long-term complications is of equal relevance. For all of the reasons given above, follow-up on the basis of symptoms alone was considered inadequate in construction of the Cambridge pathway, and symptomatic outcomes are not reported in this study.

Mucosal inflammation is still the ‘gold standard’ for diagnosis of CD and thereby mucosal recovery is the optimal marker also for response to diet. In our practice, only 26.1% of patients experienced complete mucosal recovery within 1 year of the institution of standard dietary intervention, but 56.5% restored villous architecture with only minor degrees of ongoing inflammation.

Reported histological response rates in the literature are variable – 34% of 241 patients in the Mayo series achieved restoration of villous architecture after 2 years on diet,[37] compared to 65% of 158 patients in Holland.[38] Complete histological recovery with no evidence of inflammation was achieved by 35% of 284 patients in Leeds,[39] but only by 8% of a larger series of 465 patients reported from Italy.[40] A recent large retrospective series reported in this journal reported an almost identical rate of persisting villous atrophy on follow-up biopsy.[41] The current reported proportions of patients achieving complete and partial mucosal recovery are therefore in line with previously reported studies.

Whilst complete mucosal recovery may constitute the optimal outcome, this study demonstrates that this is only achievable in approximately 30% of patients despite intensive dietary intervention guided by the results of repeat mucosal biopsy. Whether normalisation of crypt:villous height with minimal ongoing residual lymphocyte infiltration in the epithelium and/or lamina propria is an acceptable compromise for mucosal recovery is unclear; however, data exist to suggest that even mild forms of enteropathy are associated with adverse outcomes.[42, 43]

A key finding of this report is that resolution of anti-TTG serology is not predictive of mucosal recovery. Only 44% of patients with ongoing villous atrophy on re-biopsy had elevated TTG titres and, conversely, 25% of those who achieved histological recovery remained seropositive. The former proportion is very similar to that found by Dickey et al. in a follow-up series[44] and, interestingly, is also identical to the proportion of patients becoming seropositive on experimental gluten re-challenge.[36] It is perhaps not surprising that the threshold dose response for T- and B cell-mediated effects should differ, and that the B cell response should have little bearing on the mucosal lesion as it is thought not to be implicated in pathogenesis[1, 2] and there appears to be no clinical difference in patients presenting with antibody-negative disease.[45-47] Experimental gluten challenge also demonstrates differing kinetics of the response for tissue damage and antibody titres.[48] Whilst TTG serology is still helpful in the management of follow-up in CD, it can only be so when its validity has been assessed in individuals alongside histological response. In our series, reliance on serology alone for follow-up would have missed the majority of patients with ongoing villous atrophy and, conversely, have led to unnecessary clinical follow-up and dietary intensification of a number of patients with ongoing antibody responses despite normal histology. Furthermore, it is unclear whether the pattern of gluten ingestion – whether it is constant low level, or intermittent, exposure – differentially affects the expression of B- and T-cell responses in CD. Our finding is therefore that serology is not a useful surrogate marker of intestinal recovery and can only be used in follow-up if validated against individual mucosal responses. Such a finding is consistent with a number of studies,[49-51] although serological response is robustly supported in others despite lacking histological correlation.[52, 53] Notably, a high proportion of our cases were seronegative at diagnosis. Although many consider the IgA anti-TTG test to have a sensitivity of about 95%, the true sensitivity is difficult to ascertain due to confounding bias in the population undergoing intestinal biopsy. Our findings of a relatively low initial sensitivity of anti-TTG serology at diagnosis are entirely in keeping with other studies showing a 10–22% false-negative antibody titre.[44-47] Indeed, of 11 publications reported on the sensitivity of IgA anti-TTG antibodies (using human recombinant substrate) that were cited by NICE in their 2009 guidance on the diagnosis of CD in adults,[54] 5 studies (including 590 patients with CD) demonstrated a sensitivity of 90% or less, and 6 studies (that included 259 patients with CD) demonstrated a sensitivity of 91% or greater. The combined results of these studies provide a sensitivity of 86%, which is very similar to the sensitivity demonstrated in our study (85%).

Uniquely, this study reports on the efficacy of further dietary interventions to achieve mucosal recovery in 94 patients. Sixty-five patients (69%) with ongoing villous atrophy on their second biopsy were found to have a defined or probable source of gluten ingestion to account for the mucosal damage and were given further dietary counselling to remove the offending source. In some cases, this was due to non-adherence, contamination or eating out, and in other cases was identified from detailed examination of a food record. Removal of the identified gluten source resulted in histological improvement in a further 50% of cases. Some patients went to extreme lengths to conceal their dietary gluten intake which they only admitted to after demonstration of persisting villous atrophy on biopsy, due to a form of denial of their diagnosis. In 29 patients with ongoing villous atrophy, detailed food records and dietary interview were unable to identify a potential source of gluten ingestion and institution of a ‘supersensitive diet’ by the exclusion of codex wheat starch, oats and barley malt extract resulted in mucosal recovery in 18 (62%) cases.

As far as we aware, this is the only study to report the efficacy of continued dietary intervention in this way, using a third biopsy. A previous study[55] concluded that persisting mucosal changes were not due to trace amounts of gluten on the basis of dietary assessment – and low amounts of gluten ingestion have been deemed safe for the majority of patients.[56-61] In contrast, our study demonstrates that patients with ongoing villous atrophy, but no identifiable source of gluten in the diet, did appear to respond to removal of trace amounts of gluten. This finding suggests that patients differ in their threshold response to gluten, with some exhibiting a high sensitivity. This is entirely in keeping with a biopsy follow-up study of gluten challenge[36] that suggests an individual level of tolerance of gluten[60, 61] and has previously also been demonstrated by symptomatic improvement.[62]

An alternative explanation for the patients demonstrating histological response after further dietary intervention could be that these were ‘slow responders’. The median time to mucosal recovery has previously been reported as 2–3 years.[37-40, 63, 64] Such data are also gathered from observational studies, as the only way to gain controlled data on the length of time to recovery would be by blinding professionals and patients to the results of biopsy findings at 1 year and repeating the biopsy after 2 years and this has never been reported. The most likely reason for any delay in mucosal recovery is that gluten is difficult to completely exclude from the diet and that there is a learning curve in the achievement of adequate gluten exclusion. It is implausible that there would be any intrinsic biological factor delaying the recovery of the mucosa after removal of the offending agent, and this is clearly demonstrated by the observation of speed of mucosal recovery after discreet insults, such as intestinal transplant rejection, ischaemia, viral infections or chemotherapy. Notably, in our study, some patients demonstrated complete mucosal recovery from severe subtotal villous atrophy in a period as short as 6 months. The rate of mucosal recovery inevitably reflects cultural dietary factors, individual patient characteristics and the intensity and expertise of dietetic intervention and the resources available for patients. Available published estimates of the length of time to mucosal recovery are therefore not transferable between different cultures, healthcare systems or even different clinical support structures. In this study, we were unable to demonstrate any difference in the proportion of patients achieving mucosal recovery between those biopsied at 12 months or those who underwent their second biopsy at any time course subsequently. We therefore believe that the responses demonstrated after further dietary interventions were due to the intervention rather than the passage of time. However, this interpretation is potentially open to bias, as the reasons for delayed biopsy were not recorded. Some patients did not attend for repeat biopsy and whilst in certain cases this was due to moving out of area, it is possible that patients with poor dietary adherence would self-select and be less likely to attend for the second endoscopy. Comparison of the time to second biopsy prior to and following the institution of the formal care pathway shows that the majority of the delayed second biopsies occurred prior to the establishment of the pathway and were therefore likely to be due to organisational delays rather than due to patient factors that would introduce bias.

A re-biopsy strategy provides only a ‘snapshot’, and as well as there being a variability in individual threshold sensitivities to gluten,[65] thresholds may also vary over time. However, our results lead us to suggest that individual gluten sensitivity and patient factors in dietary adherence make dietary assessment of gluten exclusion by itself a poor surrogate marker of histological response – only 69% of patients with proven ongoing villous atrophy had an identifiable gluten source in their diet and dietary intensification by removal of the putative gluten source or institution of measures to remove all source of gluten only resulted in restoration of villous height in 50% of cases. The 50% of patients with ongoing villous atrophy even after the third biopsy were presumably ingesting gluten that could not be detected by themselves or their dietitians despite intensive investigation or had disease that was refractory to gluten withdrawal.

These results are of course not transferable to countries that do not permit the use of wheat-based gluten-free products and fully support the changes in Codex Alimentarius standards that took place in 2008 to lower the permissible gluten content to 20 ppm. Whilst the enactment of these standards in law in 2012 may make the more stringent dietary measures taken in our patients redundant in the future, we are still currently recommending removing codex wheat starch, barley malt extract and oats in our patients with ongoing villous atrophy on biopsy and we will report subsequently on the effects of the changes in Codex standards to our patients following the pathway.

Analysis of this care pathway suggests that further structured dietetic visits prior to the repeat biopsy may improve histological response rates, as putative gluten-containing substances were identified in the diet in 69% of cases of ongoing villous atrophy. This would support the recommendation of 3–6 monthly visits in the first year of diagnosis as proposed in recent guidelines.[7] Whilst further dietitian visits have now been introduced into our revised pathway, it is possible that dietary review prior to re-biopsy would not be as likely to uncover unexpected gluten sources as a post hoc review carried out in the knowledge that there is likely to be significant gluten ingestion due to ongoing histological changes.

In addition to our findings relating to the adequacy of dietary assessment and intervention and serological responses when compared with mucosal recovery, there are additional benefits of following a biopsy-driven strategy in CD follow-up.

Firstly, the diagnosis and classification of refractory coeliac disease (RCD). Current definitions of RCD rely on the persistence or development of new symptoms associated with CD[66] – but without clear specification of the clinical features required. Given the wide range of types and severity of symptoms that can be associated with CD and the lack of correlation of symptoms with histological appearances,[67] this definition is unsatisfactory and, not surprisingly, reported rates of RCD differ significantly between units.

Patients who are asymptomatic, but with ongoing villous atrophy despite GFD, constitute a subgroup of ‘histologically refractory’ disease that is not recognised by the current definition of RCD, but nevertheless may be associated with poor clinical outcomes and warrants closer clinical supervision.[68] The diagnosis of type 1 RCD depends solely on the presence of persisting (or recurrent) symptoms with ongoing villous atrophy despite adherence to a gluten-free diet. We have demonstrated in this study that a substantial number of patients thought to be following a good GFD still have villous atrophy on biopsy and would be diagnosed as having ‘RCD 1’ in the presence of persisting symptoms. However, further dietary intervention resulted in histological resolution in more than half of these patients, who were clearly not therefore ‘refractory’. In the cohort of patients initiated on the pathway from 2006 to 2010, only two patients were found to be histologically refractory and only one of these showed features of malabsorption. This results in a low prevalence of RCD in this series of 0.4%. The issue of persisting seropositivity is also not addressed in the current definitions of RCD. In some series of RCD, persisting antibody titres are considered a sign of ongoing gluten exposure, which excludes patients from a diagnosis of RCD,[69] whereas continued seropositivity is discounted in others.[70-72] In this study, we have shown that persisting seropositivity can occur in patients regardless of full recovery of intestinal mucosa (hence implying good gluten exclusion) and it would therefore be possible that seropositivity does not simply indicate ongoing gluten ingestion in patients otherwise considered to have RCD. However, the inaccuracy of dietary assessment in fully excluding gluten exposure means that such seropositive cases of RCD must be considered to be questionable unless rigorous dietetic intervention has repeatedly taken place.

Secondly, this study demonstrates that streamlined out-patient management can result in facilitated early discharge to community follow-up. Annual follow-up visits to the hospital serve a purpose in maintaining adherence with GFD and surveillance for associated autoimmune disorders. However, there is no reason why these checks should not take place in primary care as long as there is a structured shared care pathway in place and a mechanism for returning to the secondary care clinic for specialist advice where necessary.[73] Patients frequently request community follow-up with their general practitioner, but it remains to be seen whether this is as effective in maintaining adherence as repeat hospital attendances.

The economic burden of CD has been studied in only a small number of settings.[74-76] Notably, in the UK, the costs per patient are increased after diagnosis as a result of the state carrying a proportion of the additional expense of prescribed gluten-free foods – in other healthcare economies, this is paid for by the patient.[75] Nevertheless, there is a reduction in referral and consultation costs following diagnosis. The Cambridge Coeliac pathway results in the additional cost of a further biopsy and histological examination. However, this would be partially offset by the reduction in hospital appointments and cost savings if improved rates of histological recovery translate in the longer term into clinical benefit and earlier community follow-up. Despite mucosal healing being the logical goal of CD therapy, it has, however, been difficult to demonstrate that this results in better clinical outcomes for the patient.[37] Long-term follow-up studies of patients with undiagnosed CD on the basis of retrospective analysis of stored sera or late presentation show that complications of untreated CD take many years or even decades to become apparent[32, 33, 35] and it is highly unlikely that evidence of benefit from prospective studies over such a timescale will ever be achievable. However, it is notable that, in the Swedish registry, no cases of EATL were apparent in patients achieving complete histological recovery, compared with those with villous atrophy or ongoing mild enteropathy[77] and a very recent study has demonstrated that the risk of lymphoma is related to the presence or absence of villous atrophy on follow-up biopsy.[78]

In conclusion, this follow-up study of the management of CD following a defined pathway of care demonstrates that neither dietary assessment nor serological response can substitute for mucosal healing as a marker of response to diet. Contrary to other reports, there was no increase in mucosal healing rates in our patients biopsied later than 1 year after commencing a gluten-free diet compared with those re-biopsied within 1 year. In those patients who showed ongoing villous atrophy on re-biopsy, dietary modification revealed a proportion of patients without overt gluten ingestion who appeared to recover mucosal architecture on exclusion of trace amounts of gluten, suggesting differential high sensitivity to gluten in this group. A pathway that includes follow-up biopsy on treatment effectively stratifies patients into those who are at low risk and can be discharged early for community follow-up, those who require further dietary modification before discharge, those who require regular specialist dietetic supervision to maintain adherence with GFD and those who are histologically refractory to therapy for closer clinical supervision. This pathway also serves to reassure patients that their diet is effective in controlling their disease. Finally, it provides auditable outcomes regarding the efficacy of dietary intervention and CD follow-up that could lead to standardisation and quality improvement in the management of CD. We believe that such a pathway should become the standard of care in this condition until reliable surrogate markers of intestinal recovery become available.[79]

Authorship

Guarantor of the article: J. M. Woodward.

Author contributions: JMW conceived the study. JMW, LMS, GDC, EC and JL maintain the clinic database. LMS, GDC, NS and JMW collected data. LMS and JMW drafted the paper. All authors read and approved the final manuscript.

Acknowledgement

Declaration of personal and funding interests: None.

Ancillary