Table 1 shows the list of aetiologies contemplated in this work. Chronic conditions to consider in the differential diagnosis of enteropathies with VA and negative coeliac serology include the different forms of CD presenting with negative serology, NCEs posing problems of differential diagnosis with seronegative CD (group I) and NCEs not posing problems of differential diagnosis with CD (group II). Conditions with insufficient evidence for causing VA were also identified (group III). Online supplemental table 1 shows the list of enteropathies for which the established threshold for agreement was not reached.27–34 For these conditions, a diagnostic category (as per table 1) could not be assigned.

Table 1

List of enteropathies with villous atrophy and negative coeliac serology evaluated by the consensus group and divided into diagnostic categories.

Diagnostic criteria for non-coeliac enteropathies posing problems of differential diagnosis with seronegative CD

These enteropathies are characterised by a variable degree of duodenal VA unresponsive to a GFD, negative coeliac serology and malabsorption of different severities. For some of these enteropathies, the availability of specific biomarkers may facilitate the differentiation from seronegative CD. This group includes autoimmune enteropathy, enteropathy associated with common variable immunodeficiency, tropical sprue, giardiasis, CD4 + indolent T cell lymphoma and idiopathic VA. Drug-induced enteropathies were initially considered for this section. However, the consensus of the group was that most drug-induced enteropathies can be easily identifiable, and for this reason, this category is discussed under the section ‘non-coeliac enteropathies not posing problems of differential diagnosis with seronegative coeliac disease’.

Autoimmune enteropathy

Autoimmune enteropathy (AE) is a very rare enteropathy described first in children and then in adults.35–40 Based on the votes of our consensus, the following criteria must be satisfied for the diagnosis of autoimmune enteropathy:

Severe malabsorption symptoms (chronic diarrhoea, weight loss, nutritional deficiencies and electrolyte imbalance) unresponsive to any dietary restriction.35 36

Frank VA unresponsive to any dietary restriction.35 36

IgA/IgG positive enterocyte antibodies (indirect immunofluorescence on human/monkey jejunum).

Negative coeliac serology.

Exclusion of other causes of VA.

The following criteria were considered supportive for the diagnosis:

History of associated autoimmune conditions.

Clinical response to immunosuppressive treatments.

Deep crypt lymphocytosis and/or plasma cells infiltration, neutrophilic cryptitis±crypt microabscesses and lack/decrease of Paneth cells on duodenal histology.

Positive serum anti-AIE 75KD antibodies (ELISA) or non-organ specific autoantibodies.

HLA typing is irrelevant for the diagnosis of autoimmune enteropathy. Finally, no consensus was found for the following items: absence of severe immunodeficiencies, diagnostic role of serum antigoblet cells antibodies, involvement of other sites of the GI tract and some duodenal histopathological features. These histopathological aspects, for which also the current literature provides very discordant data, include intraepithelial lymphocytes count, crypt hyperplasia and crypt apoptotic bodies, lack of gamma-delta T cells and depletion of goblet cells.37 38 41–45

Enteropathy associated with common variable immune-deficiency

Common variable immune-deficiency (CVID) is one of the most common forms of primary immune-deficiencies, and the GI tract is frequently involved in these patients.46 47 Although it has been long recognised that CVID can be associated with VA,48 the prevalence of frank VA in CVID and its causes remain poorly understood. Although giardiasis and other GI infections are major causes for CVID enteropathy, intestinal lesions are heterogeneous and can also occur in the absence of any apparent infection. Our consensus focused on these non-infectious forms of VA. The possible association between CVID and CD was also discussed.

The following criteria were considered necessary for the diagnosis of enteropathy associated to CVID:

Presence of GI symptoms regardless from their severity (from sporadic diarrhoea to a frank malabsorption syndrome).

Diagnosis of primary CVID according to European and American societies for immunodeficiency.49

VA.

Exclusion of other causes of VA, including Giardia lamblia and other GI infections.

The following criteria were considered supportive, although not sufficient, for the diagnosis of CVID enteropathy:

Duodenal intraepithelial lymphocytosis.

Increased inflammation of the lamina propria.

Crypt apoptotic bodies.

Graft versus host disease-like lesions.

HLA typing was considered irrelevant for the diagnosis of enteropathy associated to CVID. A consensus was not reached on the diagnostic relevance of the following clinical and histopathological features, which reflects the uncertainty reported in the literature. These include association with microscopic colitis and IBD, association with lymphocytic gastritis and atrophic gastritis, mucosal depletion of plasma cells, follicular/nodular lymphoid hyperplasia, Crohn’s-like lesions/granulomas, eosinophilic infiltrate and cryptic abscesses/neutrophilic infiltrate.50–59 The authors of this consensus agreed to evaluate these features on a case-by-case basis.

The clinical dilemma of CD associated with common variable immune deficiency

Although the coexistence of CVID and CD in patients with VA was historically reported in the literature,50–52 this appears to be a very rare event. According to our votes, the major criterion confirming the diagnosis of CD in CVID is the histological and clinical response to a GFD.

Criteria excluding the diagnosis of CD in CVID include:

Lack of response to a gluten-free diet.

Negative HLA-DQ2/DQ8 typing.

The following criteria were considered unable to confirm or exclude the diagnosis of CD in CVID: extension of intestinal lesions to different parts of the small bowel, crypt hyperplasia, intraepithelial lymphocytes (IELs) increase, increased inflammation of the lamina propria, crypt apoptotic bodies, HLA-DQ2 and/or DQ8 positive. Finally, the role of mucosal TTG deposits in CVID has not been substantiated so far.

Tropical sprue

Tropical sprue is a chronic condition, which has been known for many years.60 It should be suspected in patients presenting with malabsorption syndrome, a certain degree of VA (most often mild) with intraepithelial lymphocytosis and a medical history of living in or travelling to tropical countries (particularly regions of South Asia, Southeast Asia, Africa and South America falling within the Tropics) for at least 2 months and in poor hygienic conditions. Exclusion of other causes of VA is mandatory as well as prompt clinical and histological response to a course of antibiotics. Biochemical abnormalities such as low folate responding to supplements, low vitamin B12 with possible megaloblastic anaemia and deficiency of fat-soluble vitamins can be supportive elements to the diagnosis. HLA typing has no diagnostic role.

No consensus was found on whether in tropical sprue ileal involvement is more pronounced than duodenal involvement and whether this can be used as a diagnostic criterion.

Giardiasis

Giardiasis is an infestation due to Giardia lamblia (also known as Giardia duodenalis or intestinalis), a flagellated intestinal protozoan.61 Clinical picture is highly variable ranging from a severe malabsorption syndrome to asymptomatic. In the clinical setting of VA with negative coeliac antibodies and a clinical picture with malabsorption, giardiasis must be considered and thoroughly investigated. Nevertheless, clinical suspicion of giardiasis can be prompted by less severe clinical scenarios such as IBS-like symptoms. In order to confirm the diagnosis, at least one of these tests is necessary:

Positive Giardia specific stool antigens.

Identification of trophozoites on formalin-fixed paraffin-embedded H&E stained duodenal specimens and/or on the duodenal aspirate.

Direct identification of cysts/trophozoites in fresh faeces.

Specific Giardia PCR.

Clinical response to a course of antibiotics further confirms the diagnosis

HLA typing does not have any relevance for the diagnosis, but it might be helpful in patients with borderline tTA to exclude CD. Although it is well known that giardiasis can be found in patients affected by CVID, IgA deficiency and CD, this panel of authors did not reach a consensus on the necessity of ruling out these conditions in patients with VA due to Giardia lamblia. So, the decision on whether or not to investigate other causes of VA is to be taken on a case-by-case basis.

Small bowel indolent CD4+ T-cell lymphoma

Small bowel indolent CD4 +T-cell lymphoma is a rare non-Hodgkin’s lymphoma primarily involving the small bowel.62 This type of lymphoma is quite often mislabelled as type 2 refractory CD given the persistence of VA and malabsorption despite a GFD and the clonal phenotype of intraepithelial lymphocytes.62 Clinical picture prompting the suspicion of indolent CD4+ T-cell lymphoma is characterised by long-lasting malabsorption syndrome with malnutrition unresponsive to a GFD. Duodenal VA is mandatory for diagnosis, after excluding all the other causes of VA.

Diagnosis is based on immunohistochemistry showing diffuse infiltration of the epithelium and/or expansion of the lamina propria by small/medium CD3+CD4+ T cells and presence of monoclonal rearrangement for beta-TCR and/or gamma-TCR on duodenal biopsies. Increased CD3+CD4+ intraepithelial/lamina propria lymphocytes on flow cytometry are also diagnostic.

No consensus was found on the necessity of performing a bone marrow biopsy, further endoscopic/radiological exams to assess involvements of other GI tracts, or molecular diagnostics for STAT3-JAK2 fusions.63 Therefore, the decision whether to perform these investigations should be decided on a case-by-case basis and based on local availability.

Idiopathic villous atrophy

IVA is a very recently recognised and still poorly defined chronic clinical entity characterised by frank VA unresponsive to a GFD, negative coeliac serology and in which all the known causes of VA have been thoroughly excluded.12 17

Duodenal intraepithelial lymphocytosis was voted as supportive for the diagnosis of IVA. HLA typing is helpful only to rule out CD, when negative for coeliac haplotypes. Other aspects of IVA still need to be elucidated, as no consensus was found on the diagnostic relevance of the following clinical and histopathological elements: degree of malabsorption syndrome at presentation, family history for CD, medical history of autoimmunity including dermatitis herpetiformis, possible involvement of other portions of the gastrointestinal tract, role of mucosal deposits of IgA TTG and timing for histological reassessment of duodenal histology.

A classification of different forms of IVA was recently proposed.17 This included type 1 IVA characterised by transient VA resolving spontaneously within 6–12 months; type 2 IVA, characterised by persistent non-clonal VA with excellent long-term prognosis; finally, type 3 IVA is characterised by persistent VA, the finding of aberrant T cell populations or persistent gamma-TCR mono clonality, or a medical history of lymphoproliferative disorders. However, in the present work, a consensus was reached only for type 1 IVA, a chronic enteropathy that should be differentiated from acute and self-limiting forms of enteropathy with variable degree of villous blunting likely due to acute infective gastroenteritis.64 65

Future research directions may consider the possibility of evaluating the clinical applicability of HLA-gluten tetramers and specific biopsy anti-TTG2 deposit for the differential diagnosis between IVA and forms of refractory CD.25 66

Non-coeliac enteropathies not posing problems of differential diagnosis with seronegative CD

These enteropathies are characterised by a variable degree of duodenal VA and a malabsorption syndrome of varying severity. Their diagnosis is usually prompted by a suggestive personal and pharmacological history and typical clinical or histopathological clues, which increase the pretest likelihood of the diagnosis. Particular attention should be deserved to medication-induced enteropathies which, despite being the second most common aetiology for VA with negative coeliac antibodies in adults, can still be overlooked.2 9 67 Patients with a medication-induced enteropathy seen in a coeliac centre have been frequently mislabelled as having seronegative CD unresponsive to a GFD.2 7 9 67 68 While awareness on the issue of medication induced enteropathy has been increasing among coeliac experts, particularly since the discovery of olmesartan-associated enteropathy in 2012,68 there is still a need to improve knowledge for general gastroenterologists and other medical specialists on this topic. Table 2 67–79 shows the major clues guiding the differential diagnosis for each enteropathy included in this group.

Table 2

Clinical clues guiding the diagnosis of enteropathies not posing problems of differential diagnosis with seronegative coeliac disease

Finally, a list of enteropathies for which the established threshold for agreement was not reached is provided in the supplementary section (online supplemental table 1).27–34 So, these conditions could not be assigned to any of the diagnostic categories contemplated in table 1.

The clinical spectrum of CD presenting with negative serology

In the absence of a shared consensus, controversies have surrounded the use of the term seronegative CD, which has been adopted to refer to a wide variety of clinical and histopathological conditions. Uncertainties still exist on whether this term should refer to a single clinical entity, or a spectrum of different forms of CD. In this regard, whether to consider positive coeliac IgG based serology in the context of IgA deficiency as seronegative CD, or instead as a conventional form of CD associated with IgA deficiency has been hugely debated.1–3 9 12–15 23–26 67

The present consensus agreed on the existence of different forms of CD presenting with negative serology. Primarily, seronegative CD, which should be considered separately from CD, associated with selective IgA deficiency. Second, CD with negative serology has been reported in up to 30% of patients with biopsy-proven dermatitis herpetiformis4 26 80 and rarely also in patients affected by CVID (discussed in the section on CVID previously).50–52 Finally, there are two heterogeneous groups of patients, which can present with negative coeliac serology at time of serological testing, which the present consensus agreed to consider as conventional forms of CD rather than seronegative. They include: (1) patients presenting with negative serology if they already are on a GFD or immunosuppressive therapies at time of serological testing. These patients restore their positive serological response if they are challenged with gluten or if immunosuppressants are withdrawn9 26; (2) patients with VA but discrepancies between tTA and EmA results (ie, borderline/low titre positive tTA with negative EmA or vice versa). These last two groups of patients are very commonly encountered scenarios in clinical practice and frequently causes of diagnostic mistakes.9

Seronegative CD and CD associated with IgA deficiency

The following criteria must be satisfied to make a diagnosis of both seronegative CD and CD associated to IgA deficiency:

VA, crypt hyperplasia and an increased intraepithelial lymphocytes count, on correctly oriented duodenal specimens, recovering on a GFD.

Necessity of performing diagnostic investigations before starting the patient on a GFD or immunosuppressive therapy as they may lead to false negative serology.

Exclusion of all the other causes of VA, which means to assess normal levels of immunoglobulins, negative enterocyte antibodies, negative stool parasites/HIV testing/tuberculosis, absence of iatrogenic causes for VA and no history of travelling to/residing in the tropics.

Evidence of HLA typing showing specific coeliac haplotypes, that is, DQ2.5 (DQA1*0501, DQB1*0201), HLA-DQ8 (DQA1*03, DQB1*0302), HLA-DQ2.2 (DQA1*0201, DQB1*0202) or HLA-DQ7.5 (DQA1*05, DQB1*0301).

In equivocal cases, reintroduction of gluten in the diet can be necessary to induce reoccurrence of intestinal lesions and symptoms in order to confirm the diagnosis. Although dosage and duration of diagnostic gluten challenge have not been standardised yet, at least 10 g of gluten/day for 6–8 weeks have been suggested.12 81 HLA typing should always be performed in equivocal cases of VA with negative coeliac serology, as it still has a role in discriminating seronegative CD from NCEs. Although, in Caucasian populations, up to 30%–40% of people carry the HLA-DQ2 or DQ8 haplotypes, a negative HLA typing excludes seronegative CD.1–5 7

A clinical picture with severe malabsorption, associated autoimmune disorders, family history of CD and biopsy-proven dermatitis herpetiformis can be supportive of the diagnosis, but they are not sufficient to make a diagnosis of seronegative CD in the absence of the necessary diagnostic criteria. Similarly, when available, small-bowel mucosal transglutaminase 2-specific IgA deposits can support the diagnosis of seronegative CD and may be helpful to discriminate from other NCEs in patients with normal serum IgA levels.25

Finally, in a patient with negative IgA coeliac antibodies who fulfil these diagnostic criteria (ie, flat duodenal mucosa recovering on a GFD and coeliac HLA), the finding of selective IgA deficiency (total serum IgA level <5–7 mg/dL)±positive IgG coeliac serology will allow differentiation between seronegative CD and CD associated to IgA deficiency. It has been shown that in patients with IgA deficiency sensitivity of IgG tTA and IgG deamidated gliadin peptides antibodies outperform IgG EMA (91% vs 82% vs 76%).82