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

Setting the Scene: Current Gold Standard Diagnosis

The current CD diagnostic model is based on the integration of clinical suspicion, positive serological tests, diagnostic duodenal biopsy and presence of the HLA-DQ2 or -DQ8 haplotype. Guidelines appropriately state that individuals should maintain a gluten containing diet prior to testing to minimize the risk of false-negative serology or histology results.[12,13,101,102] However, there is little evidence defining the quantity and duration of dietary gluten intake required. A minimum of 6 weeks exposure to gluten containing foods in at least one meal per day is recommended in the 2009 National Institute for Health and Care Excellence guidelines.[103] Similarly, the Celiac Society of Australia suggests a 6 weeks gluten 'challenge' comprising at least four slices of bread (~6 g of gluten) per day prior to testing for adult patients.[102] Nonetheless, as regular intake of even 50 mg of gluten per day causes mucosal injury,[14] a diet comprising daily consumption of foods such as bread, chapattis, pasta, biscuits or cakes will usually provide adequate gluten to enable reliable diagnostic testing.

Celiac Serology

Celiac serology is recommended as the first-line test for patients with suspected CD.[15,103] In CD, deamidation of gluten by tissue transglutaminase (tTG) in the small-bowel lamina propria promotes presentation of gluten peptides (gliadin in wheat, secalin in rye and hordein in barley) by HLA-DQ2 or HLA-DQ8 dendritic cells to pathogenic local CD4+ T cells.[16] Available serological tests quantify either autoantibodies [anti-tTG and anti-endomysial (EMA) antibodies] or antibodies against the antigenic gliadin component of dietary gluten [anti-gliadin (AGA) and anti-deamidated gliadin peptide (DGP) antibodies]. The most widely used serological test is anti-tTG IgA as it is generally regarded as having a higher sensitivity for detecting CD than other available assays,[15] although anti-DGP assays, which use synthetic gliadin peptides that mimic tTG-modified gliadin sequences to bind serum IgA or IgG against DGP, may have similar diagnostic accuracy.[17,18] In contrast, traditional AGA antibody tests have lower diagnostic accuracy with frequent false-positive results and are no longer recommended for initial diagnostic evaluation. The sensitivities and specificities of available serological tests used in CD diagnosis are described in Table 1.

Interlaboratory and interregional variation in individual test performance complicate interpretation of serology. For example, for anti-tTG, different generation assays exist, which have different dose–response curves and performance in external audit and quality assurance testing.[19] As there is currently no international standard to ensure comparability between assays, antibody results are unable to be expressed in absolute immunoglobulin concentrations. The majority of kits use a calibration curve with antibody dilutions that provide numerical values proportional to antibody concentrations in relative units.[12] The measurement of complex mixtures of antibodies introduces variability in the positive/negative/equivocal cutoffs of different immunoassays, which affects test sensitivity and specificity. Developing local population thresholds and laboratory participation in external quality assurance programs are therefore important.[19]

A concurrent serum IgA level should be measured when an individual undergoes celiac serology testing, although some kits can quantitate both. Selective IgA deficiency is increased 10- to 16-fold in those with CD, and these individuals will therefore test negative to IgA-based assays.[16] In such cases, testing for DGP IgG antibodies, tTG IgG antibodies and/or EMA IgG antibodies should be considered.[20,21] Seronegative CD may also be due to false-negative antibody testing or the taking of a gluten-free diet (GFD) prior to testing, and an approach to diagnosis in these settings is discussed in later sections.

Duodenal Histology

Duodenal mucosal biopsy remains an important component of diagnostic testing in individuals with suspected CD. Multiple biopsies should be obtained from the first, second and third parts of duodenum. The optimal number of biopsies remains controversial, but current literature supports obtaining four to six.[22,23] Notably, the diagnostic accuracy reduces to 90–95% if only two or three biopsy specimens are obtained.[23] There is growing evidence that routine duodenal bulb biopsy, in addition to distal biopsies, increases the detection of villous atrophy by 9–13%.[24–26] Given the patchy nature of villous atrophy, high-resolution magnification endoscopy has also been used to improve detection of mucosal abnormalities.[27]

Abnormalities of CD on biopsy are classified by the Marsh–Oberhuber system, which grades disease severity to five morphologic levels (Grades 0–IV) according to degrees of villous atrophy, crypt hyperplasia and intraepithelial lymphocytosis.[28] A simplified grading classification based on three morphologies (A: nonatrophic; B1: atrophic, villous-crypt ratio <3: 1; B2: atrophic, villi no longer detectable) and an intraepithelial lymphocyte (IEL) count of >25/100 enterocytes has been proposed, and reported to have a higher interobserver reproducibility than the Marsh–Oberhuber classification (mean κ 0.55 versus 0.35, respectively).[29]

HLA Typing

Testing for HLA-DQ2 and -DQ8 is valuable in suspected CD, as virtually all patients with CD possess a specific HLA background: 95% of patients carry the DQ2 haplotypes HLA-DQ2.5:DQA1*05:01/DQB1*02:01 or HLA-DQ2.2:DQA1*02:01/DQB1*02:02, and almost all of the remaining 5% carry the DQ8 haplotype DQA1*03:01/DQB1*03:02.[20,30] Accordingly, the absence of these HLA DQ2/DQ8 genotypes effectively excludes the diagnosis of CD and is particularly useful when serology and duodenal histology are equivocal.[31] Unlike serology and histology, the presence of CD HLA genotypes is not influenced by dietary gluten. Conversely, the high frequency of these predisposing genes in the overall population (40%) limits the utility of HLA typing in CD case finding, and their presence is insufficient to confirm a diagnosis of CD.[32]

Diagnostic Algorithm–Combining Serology, Histology and HLA Typing

Whether to proceed to duodenal biopsy is determined both by the clinical suspicion of CD and results of initial serology tests. Individuals with positive serology are recommended to have duodenal biopsy for histological confirmation. In those with low risk of CD, negative serology reduces the post-test probability to sufficiently low levels that duodenal biopsy is not recommended. If the individual has typical symptoms or when the clinical suspicion for CD is high (Box 1), negative serology does not exclude the condition and duodenal biopsy is necessary. Absence of HLA-DQ2 or -DQ8 virtually excludes CD.[31] A suggested diagnostic algorithm is detailed in Figure 1. Diagnostic difficulties arising within this algorithm are elaborated upon in subsequent sections.

Figure 1.

Diagnostic algorithm for suspected celiac disease.
CD: Celiac disease.