Background
Coeliac disease (CD) is a common (1:100) autoimmune disorder of the small bowel: CD is triggered by the ingestion of gluten in genetically susceptible subjects carrying the HLA type II DQ2 and/or DQ8 haplotypes [
1‐
3]. CD is characterised by a wide variability of symptoms, ranging from severe malabsorption to subclinical or silent pictures as observed during screening programmes [
4]. CD diagnosis is based on: the detection of specific auto-antibodies (anti-transglutaminase type 2 IgA) and compatible findings at duodenal histology, including surface enterocyte damage, increased intraepithelial lymphocytes (IELs), crypt hyperplasia and villous atrophy being considered the principal hallmark [
4].
Although the aforementioned mucosal alterations are not completely specific to CD, histology is considered the gold standard for its diagnosis in adults: therefore the close collaboration between clinicians and pathologists is required as well as the availability of a standardised classification in order to obtain correct comparable diagnoses [
5]. In 1990 Marsh [
6] proposed a classification describing four types of lesions named: type 1 or infiltrative with an increased number of IELs, type 2 or infiltrative-hyperplastic characterised by an increased number of IELs plus crypt hyperplasia, type 3 or flat/destructive with the above lesions plus villous atrophy, and lastly type 4 or atrophic/hypoplastic with a normal number of IELs and crypts and absence of villi. According to the Marsh criteria revision proposed by Oberhuber et al. [
7] in 1999, the atrophic type 3 lesion was subdivided in 3a, 3b and 3c in the presence of mild, marked and total villous atrophy respectively (Marsh-Oberhuber classification). Another classification classifies CD damage simply as atrophic and non-atrophic [
8].
A further factor, which strongly facilitates CD diagnosis, is the response to a gluten-free dietary regimen (GFD) usually leading to the improvement/disappearance of symptoms together with the normalisation of antibody titres and duodenal histology. Therefore, in the majority of cases, a second histology to confirm the diagnosis is not required when the clinical status is silent and auto-antibodies negative. A great amount of data has been published on the histological criteria for CD diagnosis: less information is available though about the histological modifications during GFD and the ways to compare histological reports before and after GFD. This is a crucial point for both the correct classification of CD and related patient management in view of the recent data showing the persistence of villous atrophy in spite of the complete adherence to GFD [
9,
10].
Therefore, the aim of our study was to highlight the histology of the treated coeliacs, to show how it would vary within a biopsy piece and thus to apply a new proportional method to evaluate the duodenal biopsies from coeliac patients at diagnosis and during follow-up.
Discussion
The present study focused on the evaluation of duodenal biopsies from CD patients before and after GFD. The application of a novel approach (EF method) instead of the classical qualitative one (MO method) led to significant changes in the histological reports, enabling the distinction of patients with different evolution outcomes of the duodenal damage.
Although different systems have been proposed to classify the CD duodenal damage (
i.e. the Marsh score [
13] and its modification by Oberhuber [
7] and the recently introduced Corazza-Villanacci grading [
8]), no approach to follow a CD patient’s evolution has until now been proposed and the dietary state has been rarely considered. The introduction of a new method appeared necessary in case of unchanged responses under the Marsh-Oberhuber classification (the worldwide referral). Previous studies analysed the number of IELs and the villus height/crypt depth ratio [
14‐
16] or the villous area [
17] and their relationship to gluten intake; however, such parameters appear difficult to apply in a clinical setting, directly correlated with the Marsh grade, and thus do not prove useful when this grade is unmodified.
At present, the evaluation of duodenal histology does not differ whether CD patients are or are not on a GFD regimen, as the analysis is mainly focused on spotting the presence of an architectural damage (atrophy). These parameters have been summarised in the Marsh-Oberhuber classification and strongly influence the specialists in their clinical/therapeutic decisions, especially when the histological pattern is unmodified [
18]. However, the effects of GFD can be interpreted for further diagnostic clues. Our findings suggest that in adult CD patients, even those on a strict dietary regimen, the histological response can be slow (or absent) and that mucosal atrophy does not always turn into complete/partial normalisation [
9,
10]. This result has been reported by different studies [
19] and corroborated by data from videocapsule endoscopy studies, consistent with the persistence of macroscopic signs of intestinal atrophy with a skip fashion [
20]. Again, during a gluten-free regimen, serological tests (
i.e. anti tissue transglutaminase IgA) are associated with clinical responses but do not correlate with the histological improvement of the duodenal mucosa (mucosal healing) [
21]. The gap between serological and endoscopic/histological findings represents a potential problem for patient management in case of persistent mucosal atrophy. In their study Kaukinen et al. [
21] underlined that the occasional ingestion of gluten is not associated with any increase in anti-transglutaminase IgA titers. Interestingly, another series reported only minimal mucosal changes after the controlled ingestion of gluten [
22]. Moreover, there is evidence that different types of food commonly believed gluten-free, in fact contain gluten traces [
23] and that an accurate diet analysis often reveals the incomplete adherence to GFD thus explaining the persistent mucosal atrophy or removing the suspicion of a refractory sprue [
24,
25]. Autoptic studies [
26] showed a progressive decrease of mucosal injury from the proximal to the distal tract of the small intestine; thus, the clinical picture can reflect the longitudinal involvement of the small bowel rather than the type of mucosal damage [
27]. Additionally, the histological damage in naïve patients is often patchy [
28,
29]. For this reason, in order to minimise the risk of under-estimation, the evaluation of at least 4 oriented duodenal endoscopic biopsies is strongly recommended.
The reason underpinning the persistence of villous atrophy in treated CD is not easy to understand, although a miRNA signature might be involved [
20]. In our tertiary referral Centre, a significant proportion of CD patients were referred to undergo complete endoscopic investigations, highlighting how much the need to compare duodenal histology before and after GFD is particularly felt, in order to proceed with a watchful follow-up in selected cases. The above data led to a relevant question, i.e. does the incomplete normalisation of the intestinal mucosa represent a risk factor for CD complications? If yes, the current data and international guidelines on CD follow-up may suggest the need for further biopsies after diagnosis, at least until some mucosal healing is present. Therefore, the opportunity of quantifying the damage and evaluating its trend would be important.
Since some symptoms or complications of CD are correlated with mucosal damage (
e.g., osteoporosis) [
30], and other ones are caused by autoantibodies (
e.g., infertility) [
31], mucosal biopsy and serological tests should be considered complementary, and the quantitative evaluation of mucosal atrophy proves helpful in the evaluation of clinical findings and patient management.
When applying the EF method to patients with persistent duodenal atrophy, 91 % of cases changed the histological assessment, with a regression in 72 % or even a progression in 18 %.
Whilst the simple detection of atrophy in the duodenal mucosa of naïve patients is enough to make a correct diagnosis (as in the Corazza-Villanacci grading), the context dramatically changes when dealing with patients on GFD, for whom the main question is about the presence or not of any histological improvement. The answer to this question requires a more specific scoring system. A possible problem is represented by the loss of inter-observer agreement when the number of grades increases. In our study the k test showed good agreement levels but we suggest that this type of analysis should be used in tertiary settings by expert pathologists.
Moreover, the possible availability of drug-based therapies for CD in the next future would add to the interest in systems, such as the EF method, aimed at comparing duodenal histology before and after GFD, In such a case, the judgment about the improvement of duodenal atrophy will be pivotal in the therapeutic decision-making process [
2].
However, some weak points of the study need to be discussed. The study has a retrospective design and thus prospective series are necessary to evaluate the long term clinical impact and prognosis of patients with an unchanged MO classification but showing a different finding with the EF one; in particular, if unchanged or worsened duodenal histology with the EF method has an increased risk of complications. Interpretation of duodenal histology could be challenging due to the patchy nature of CD damage in some circumstances and the need of a strict collaboration between clinical gastroenterologist and pathologist is essential. Moreover, the EF comparison between duodenal histology at diagnosis and during follow up could be performed only in case of correct orientation of the samples (not always possible). In our study the time between diagnostic and follow-up biopsy is not uniform and a dietetic questionnaire was impossible to ascertain the correctness of the GFD.
Noteworthy, this retrospective study was conducted in a tertiary referral centre where complicated or challenging cases are managed; thus, a different result could be present in other scenarios where dedicated and/or expert personnel are absent. Another point about the EF method is the cost analysis in the light of a possible clinical benefit; in fact, the EF method is time consuming and imperfectly oriented biopsies could introduce the necessity of an increased number of samples per patient.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
LE, EZ, MTB and SF planned the study. EZ, SF, SB and LR performed histological evaluation and preparation of the samples. LR and CT prepared the database and data extraction. MTB, DC and SB supervised the work and prepared the manuscript. All the authors approved the final draft submitted.