Diagnostic Dilemmas in Celiac Disease

Michael X Ma; Mina John; Geoffrey M Forbes


Expert Rev Gastroenterol Hepatol. 2013;7(7):643-655. 

In This Article

Diagnostic Dilemmas in Individuals With Possible CD

Screening & Case Finding

Despite the postulated benefits of improved disease detection, studies have not supported population-based screening for CD. Screening for CD must be distinguished from case finding patients with subtle clinical manifestations, where the value of serologic tests as diagnostic aids is widely accepted. A summary of unusual clinical presentations or associations of CD is given in later sections and the presence of these should prompt initiation of testing for CD. Although anti-tTG and anti-EMA assays have high sensitivity and specificity for CD, as disease prevalence is only about 1%, the positive predictive value of these tests when applied to the general population is low.[33] Furthermore, the benefits of screening depend upon a patients' subsequent compliance with therapy. Maintenance of a GFD requires a significant and permanent change in lifestyle that an asymptomatic individual may find difficult to adhere. It is also unclear whether asymptomatic patients with CD benefit from treatment with a GFD.[34] Therefore, current guidelines do not advocate population screening of asymptomatic individuals at average risk for CD.[12,13]

In contrast, testing of certain groups at above average risk for CD has been advocated. Relatives of patients with CD have increased risk of developing CD.[35,36] The risk is highest among monozygotic twins (75%), HLA-identical siblings (40%) and first-degree relatives of families with at least two affected siblings (17%).[35] The risk among first-degree relatives varies from 5 to 11%.[36,37] Thus, case finding of first-degree relatives (particularly siblings), even if asymptomatic, should be encouraged.

Serology Positive, Histology Negative

Some individuals may have positive serology, but normal small intestinal biopsy. Reasons include false-positive serology, false-negative histology, potential CD, latent CD or the recently described and evolving concept of nonceliac gluten sensitivity.

False-positive Serology, False-negative Histology

False-positive tTG antibody results are uncommon (~10%),[19] are typically of low titer and more often occur with older assays that use a nonhuman tTG capture antigen. False-negative histology is rare and may arise due to inadequate sampling in patchy disease, commencement of a GFD prior to testing or missed subtle histologic abnormalities.[27] Review by a specialist gastrointestinal pathologist should be performed if clinical suspicion of CD persists. If a discrepancy in serology and histology remains after excluding these confounders, HLA typing can be performed. In those who carry HLA-DQ2 or -DQ8, repeating serology and duodenal biopsies after taking a diet containing gluten for 6 weeks may improve diagnostic yield.

Potential CD & Latent CD

It is clear that some individuals with genetically determined gluten intolerance do not develop, or develop delayed, small intestinal mucosal changes. Potential CD refers to the finding of positive serology with normal or slightly increased duodenal mucosal IELs in patients ingesting a normal diet, who later develop symptoms and histological changes of CD.[22] Confusingly, the terms latent CD and potential CD have often been used interchangeably. In contrast to potential CD, latent CD is best reserved to describe those patients with known CD and previous villous atrophy, who remain asymptomatic with normal villous architecture following reintroduction of a gluten containing diet; some will relapse clinically and histologically on long-term follow-up.[22,38] Regular clinical monitoring of patients with potential and latent CD is recommended, although data guiding the interval and mode of surveillance for such patients are lacking.

Nonceliac Gluten Sensitivity

More people are estimated to consume a GFD than there are individuals with CD, diagnosed or undiagnosed.[39,40] A likely driver of this demand for gluten-free products is the growing numbers of individuals with symptoms, but who do not fulfill the diagnostic criteria for CD. Interestingly, many also report symptomatic improvement following commencement of a GFD.[39,41]

In an early study of 94 adults reporting abdominal symptoms after cereal ingestion, 63% of participants improved symptomatically from a GFD, despite not having CD nor cereal allergy.[42] A more recent study suggests nonceliac gluten sensitivity (NCGS) is a distinct clinical entity, based upon symptom improvement following a double-blind placebo-controlled dietary challenge to participants with an irritable bowel syndrome (IBS)-like presentation.[43] At present, NCGS remains a collective term and incorporates a range of symptoms (bloating, abdominal pain and diarrhea), which patients may associate with gluten ingestion.[44]

Individuals with NCGS may have increased levels of anti-gliadin IgA/IgG in comparison with healthy controls.[42,43] However, the prevalence of anti-tTG and anti-EMA antibodies is not increased.[41] Furthermore, individuals with NCGS may have normal or minor increases in IELs on duodenal biopsy. Published data suggest that about 60% of individuals with NCGS have a normal intestinal mucosa with the remaining 40% having a mild increase in IELs (Grade I Marsh classification).[43] In CD patients with Grade I Marsh mucosal changes, there is an increase in γδ IELs and deposition of subepithelial anti-tTG IgA when compared to individuals with NCGS.[39,45]

Acknowledging reactions to gluten are not limited to CD, a consensus document based upon expert opinion was developed in 2012 proposing a new nomenclature and classification of gluten-induced conditions – CD, wheat allergy and NCGS.[40] However, a recent study of 37 patients with IBS and previously diagnosed NCGS demonstrated that addition of gluten to a low-fermentable, poorly absorbed, short-chain carbohydrate (FODMAP) diet did not induce gut symptoms, casting doubt over NCGS as a separate clinical entity.[46] At present, there are no specific diagnostic markers to identify NCGS, and it remains a diagnosis of exclusion following negative testing for CD and wheat allergy. Wheat allergy is a type 1 hypersensitivity reaction where patients may present with rhinitis, food allergy (gastrointestinal symptoms, hives, angio-edema or atopic dermatitis) or contact urticaria, and may be diagnosed by serum IgE assay or skin prick test to wheat.

Although further research is required to further elucidate its pathogenesis, epidemiology and natural history, it appears that NCGS does not have a strong hereditary basis, does not cause malabsorption or nutritional deficiencies and is not associated with any increased risk for autoimmune disorders or malignancy.[47] Given these major differences in natural history and outcomes between CD and NCGS, it is important to distinguish between these two disorders, given the implications for ongoing disease monitoring, required duration and strictness of adherence to a GFD and for counselling and testing of family members.

Serology Negative, Histology Positive

Seronegative CD poses a diagnostic dilemma and is identified following integration of consistent clinical, genetic and histopathologic features, as these patients lack serum celiac antibodies.[48] Seronegative CD, although uncommon, may occur in a number of settings. First, selective IgA deficiency will cause negative serology when an IgA antibody assay is used. Second, false-negative serology occurs in about 1% of patients, even if not IgA deficient.[19] Third, serology may be less sensitive than intestinal mucosa to gluten stimulation, and it is possible that a low-gluten diet can cause abnormal histology with negative serology.[49] HLA-DQ2 and -DQ8 typing are particularly useful in patients with suspected seronegative CD, as a negative result virtually excludes the diagnosis. Conversely, clinical and histological improvement on a GFD are important to confirm the diagnosis.

Other Causes of Villous Atrophy

In the setting of negative celiac serology, other etiologies associated with villous atrophy need consideration, and include common variable immunodeficiency, small intestinal bacterial overgrowth, autoimmune enteropathy, infection such as giardiasis, intestinal lymphoma, collagenous colitis, HIV enteropathy, Crohn's disease, malnutrition, Whipple's disease and tropical sprue.[50] Immunosuppressive medications such as methotrexate, mycophenolate mofetil and azathioprine are all associated with villous atrophy.[50,51] A recent report of 22 seronegative patients found reversal of villous atrophy following cessation of the angiotensin receptor antagonist olmesartan.[51] Box 2 represents a proposed list of patient details and initial investigations that may be considered for evaluation of seronegative subjects with villous atrophy.

Testing Individuals Already on a GFD

Individuals may choose to commence a GFD prior to formal diagnosis of CD for a multitude of reasons including perceived health benefits, improved symptoms and in some cases long wait times for medical appointments and access to endoscopy. Testing while on a GFD increases the likelihood of false-negative results and represents a diagnostic challenge as abnormalities in both serology and duodenal histology are reliant on prolonged antigenic stimulation by dietary gluten.

Testing for CD in individuals already on a GFD should be clinically and patient influenced. In those wishing to proceed with testing, celiac serology should be the initial investigation. Individuals with positive serology are referred for small-bowel biopsy. The specific serologic and histologic features of CD do not normalize immediately upon commencement of a GFD and if the duration of GFD prior to testing is short (<1 month), abnormalities often remain and may be used to diagnose CD.[47]

In the context of an individual being on a GFD, negative serology does not exclude CD and HLA DQ2/DQ8 genotyping is the next investigation of choice. A negative result will eliminate the need for further work-up. However, the presence of HLA-DQ2 or -DQ8 is insufficient to confirm a diagnosis of CD. Gluten challenge remains the gold standard for CD diagnosis in HLA-DQ2 or -DQ8 positive patients who have normal serological findings when tested on a GFD.[47] Gluten challenge is the process, whereby an individual on a GFD with suspected but unproven CD reverts to a normal, gluten-rich diet, under medical supervision, for diagnostic testing.[47,52] Those who develop severe symptoms following gluten ingestion are not suitable candidates for gluten challenge.

In individuals positive for HLA-DQ2 or -DQ8, an initial gluten challenge comprising 3 g dietary gluten (or two slices bread) per day for 2 weeks may be initiated.[52]

In one study of subjects receiving a 2 week gluten challenge, Marsh III histology was seen in 68%, positive celiac serology (anti-tTG or anti-DGP) in 50% and either Marsh III histology or positive serology in 84%.[52] If the initial challenge is tolerated, a full gluten challenge of 3 g gluten per day for a further 6 weeks (total 8 weeks) can be performed, followed by duodenal biopsy.[47] However, the added diagnostic sensitivity of extending the challenge to 8 weeks is unknown. Celiac serology should also be checked at the end of the gluten challenge and, if negative, repeated in 2–6 weeks. The reason for measuring delayed antibody levels is that celiac antibody concentrations continue to rise following completion of a 2 week gluten challenge.[52] It may be argued that if an individual has experienced symptomatic improvement on a GFD and intends to remain gluten free in the longer term, investigating for CD may not be necessary. However, it is important to differentiate CD from other conditions, especially NCGS, given the significant and varied complications of CD and the implications for long-term management. As such, patients should be informed of the rationale for formal diagnosis of CD and made aware of the availability of definitive testing should they so desire.

Atypical Presentations of CD

CD is increasingly recognized as exhibiting wide clinical presentations, presenting at any age, involving multiple organ systems and with variable symptom severity. As such, terms such as typical, atypical and silent CD have been used to categorize its possible presentations.[16] Use of the term silent CD has been criticized, however, as subtle abnormalities are often found following detailed history, examination and laboratory investigations in these patients.[53]

Patients with atypical CD exhibit predominantly extraintestinal manifestations, often with only minor gastrointestinal complaints. However, most will have duodenal mucosal damage and elevated celiac antibodies. As with classical CD, the diagnosis is established by serologic testing, abnormal duodenal histology and improvement of symptoms following a GFD.

A cutaneous example of CD is dermatitis herpetiformis, an intensely pruritic rash on the extensor surfaces of the extremities with pathognomonic cutaneous IgA deposition.[20] Iron-deficiency anemia is common in CD and may be the only presenting feature. Other presentations include unexplained short stature, delayed puberty, infertility, recurrent fetal loss, vitamin deficiencies, unexplained fatigue, protein calorie malnutrition, elevated transaminases, osteoporosis and dental enamel hypoplasia.[1,13] In addition, a variety of neuropsychiatric conditions such as dysthymia, peripheral neuropathy, ataxia and migraine headaches have been reported.[1,16] Testing for CD should be considered if no other explanation is found for these clinical presentations.

CD may also be associated with autoimmune disorders such as thyroiditis, type 1 diabetes, autoimmune myocarditis, idiopathic dilated cardiomyopathy, Sjögren's syndrome, systemic lupus erythematosus, autoimmune hepatitis, autoimmune cholangitis, primary biliary cirrhosis, inflammatory bowel disease (IBD) and systemic and cutaneous vasculitis.[16,47] In the presence of these conditions, symptomatic patients should also be referred for CD testing.

Expert Commentary: Modifications to Current Diagnostic Criteria – Is Small-bowel Biopsy Required?

Recognizing that CD is underdiagnosed, some authors have proposed modifications to the current gold standard diagnostic criteria, to allow diagnosis based on serology alone. The 2012 European Society for Pediatric Gastroenterology Hepatology and Nutrition (ESPGHAN) guidelines recommend that duodenal biopsies are not required in symptomatic children and adolescents with anti-tTG titers >10-fold the normal cutoff value, as the likelihood of villous atrophy (Grade III Marsh classification) in such patients is high.[12] For example, in a prospective study of 97 children and 227 adults with biopsy-confirmed CD, the positive predictive value for CD based upon a cutoff value of 30 IU (equivalent to 10 times the upper limit of normal) anti-tTG was 95% in children and 53% in adults.[54] In another study of 148 adult patients, an anti-tTG level >30 IU/ml was absolutely predictive for CD compared against characteristic small-bowel mucosal appearance.[55] The ESPGHAN guidelines suggest that tTG antibody positivity be verified by anti-EMA testing on a separate occasion, to confirm a diagnosis of CD.[12] HLA DQ2/DQ8 typing is also advised in patients with double positive serology without duodenal biopsy, as a positive result aids in reinforcing the diagnosis of CD.

The ESPGHAN 2012 recommendations were evaluated in a subsequent study of 104 pediatric and adult patients with known CD and 537 controls.[56] Serum levels of IgA anti-tTG were quantified using four different commercial assays. Analysis was based upon likelihood ratios and pre- and post-test probabilities. Using an IgA tTG threshold of 10 times the upper limit of normal, the post-test probability of CD was 89–96% when the pretest probability determined by symptoms was 7%. In contrast, the post-test probability of CD decreased to 53–75% when the pretest probability of disease was only 1%. These differences in post-test probabilities highlight the importance of good clinical judgment prior to any testing.

Combination serology testing without small-bowel biopsy has also been investigated as a means to improving CD diagnosis in adults. Most studies have examined the diagnostic accuracy of combining anti-tTG and anti-EMA assays as the former is generally more sensitive and the latter more specific for CD. For example, a recent retrospective study based on 2477 symptomatic adult patients referred for celiac serology testing, found serum anti-tTG titers >118 IU identified CD with a 2% false-positive rate compared with diagnosis based on histology. When anti-tTG titers of 21–118 U were used in combination with an anti-EMA dilution titer of 1:160 or greater, the positive predictive value was 83% for CD in symptomatic patients, in the absence of small-bowel biopsy.[57] The accuracy of using combination serologic testing with anti-TG and anti-EMA for detecting CD was also demonstrated in a separate study of 3,850 general adult volunteers from a US county. In this study, all 31 patients with double positive serology underwent small-bowel biopsy, of whom 29 patients had histological features of CD.[58]

Addition of anti-DGP and combinations of three or four serology tests may provide higher positive and negative predictive values for CD. In a study of 149 CD patients and 119 controls, a combination of four serological tests (anti-DGP IgA, anti-DGP IgG, anti-tTG IgA and anti-EMA IgA), compared with duodenal histology, yielded positive and negative predictive values for CD of 99% and 100%, respectively.[59]

When anti-EMA IgA was omitted from the algorithm, the positive and negative predictive values remained largely unchanged at 99% and 98%, respectively.[59] Although these data are promising, how combination serology testing might replace duodenal biopsy in the diagnosis of CD remains to be determined. This approach is certainly attractive for a pediatric population, where undertaking endoscopy is more difficult than for adults. However, for combination serology to be truly useful as a diagnostic replacement for CD, the negative predictive value of this approach requires further validation and its performance compared with histology tested in larger patient populations. In other words, do negative combination serology tests predict absence of CD as well as normal duodenal histology? As biopsies are often undertaken for purposes of diagnostic confirmation or exclusion of CD in the setting of low-positive or equivocal serology, it is likely that a combination serology approach in these patients will be unable to definitively exclude CD. Furthermore, for asymptomatic patients with low-positive titer antibodies, duodenal histology is almost always required to confirm or exclude CD, due to the risk of false-positive serology. Duodenal histology also provides additional diagnostic information not assessable by serological testing and in particular, the grade of mucosal abnormality. The biopsy taken at diagnosis provides a baseline against which follow-up biopsies can be compared, especially in patients with persistent or recurrent symptoms. Finally, there is also the danger that practitioners will ignore the details of guidelines that are based solely on serology and instead accept any positive serology as justification for commencing a life-long GFD.[60] Therefore, at this time, we suggest that histology remains an important adjunct for diagnosing CD in the vast majority of patients and do not advocate for its replacement by combination serology.