AIM: To investigate the association between serum antibody levels and a subsequent celiac disease diagnosis in a large series of children and adults.

METHODS: Besides subjects with classical gastrointestinal presentation of celiac disease, the study cohort included a substantial number of individuals with extraintestinal symptoms and those found by screening in at-risk groups. Altogether 405 patients underwent clinical, serological and histological evaluations. After collection of data, the antibody values were further graded as low [endomysial (EmA) 1:5-200, transglutaminase 2 antibodies (TG2-ab) 5.0-30.0 U/L] and high (EmA 1: ≥ 500, TG2-ab ≥ 30.0 U/L), and the serological results were compared with the small intestinal mucosal histology and clinical presentation.

RESULTS: In total, 79% of the subjects with low and 94% of those with high serum EmA titers showed small-bowel mucosal villous atrophy. Furthermore, 96% of the 47 EmA positive subjects who had normal mucosal villi and remained on follow-up either subsequently developed mucosal atrophy while on a gluten-containing diet, or responded positively to a gluten-free diet.

CONCLUSION: Irrespective of the initial serum titers or clinical presentation, EmA positivity as such is a very strong predictor of a subsequent celiac disease diagnosis.

Recent serological screening studies have revealed that up to 1%-2% of the Western population might be affected by celiac disease[1,2]. However, due to its heterogeneous clinical picture the disease remains markedly underdiagnosed. Sensitive serum endomysial (EmA) and transglutaminase 2 antibodies (TG2-ab) are widely used as a method to select subjects for further investigations, but the diagnosis is based on the presence of small-bowel mucosal villous atrophy and crypt hyperplasia[3,4]. Unfortunately, the histological definition of the disease involves several problems. First, invasive studies are needed to acquire the mucosal specimens. In addition, biopsy samples may be of poor quality or wrongly orientated, increasing the risk of false positive or negative results[5]. The mucosal damage may be patchy and missed even if several samples are taken[6,7]. Finally, the histological lesion develops gradually and interpretation of borderline cases can be challenging. Since particularly EmA and high values of TG2-ab seem to predict celiac disease with a high specificity, it has been advocated that in seropositive subjects endoscopic studies might not always be needed to establish the diagnosis[8-15]. However, most studies so far have been carried out in tertiary centers with high-risk patients, and the results might not be applicable in everyday clinical practice.

In our local health-care district active celiac disease case-finding has been carried out since the 1980s. As a result, a substantial part of the patients are detected because of atypical symptoms or by active risk-group screening, and currently about 0.7% of the population have a biopsy-proven diagnosis[16]. Hence, we now sought to establish whether the serum antibodies could predict subsequent celiac disease also in subjects with mild or atypical clinical presentation. Because of the high specificity, EmA has traditionally been considered the gold standard for celiac disease serology, and was thus chosen as the primary inclusion criterion[17,18]. In addition, the results were compared to the widely used serum TG2-ab.

The study cohort comprised consecutive EmA positive children and adults investigated at the Departments of Pediatrics and Gastroenterology and Alimentary Tract Surgery, Tampere University Hospital. Primary care physicians were encouraged to refer individuals with celiac disease suspicion for further investigations applying a low index of suspicion. In addition, subjects who participated in population-based research studies were accepted. In the hospital demographic data, a family history of celiac disease and symptoms leading to the disease suspicion were recorded, and all subjects underwent extensive clinical, serological and histological evaluations. Thereafter, voluntary EmA positive children and adults continued in the trial. Participants who showed small-bowel mucosal villous atrophy and crypt hyperplasia (Marsh III) received a celiac disease diagnosis and were placed on a gluten-free diet. Subjects who had normal villi continued on a gluten-containing diet and were placed on regular serological and histological follow-up. In addition, the possibility to start an experimental trial with a gluten-free diet was offered to EmA positive individuals with normal villous structure (Marsh 0-II). Those who consented were re-evaluated after one year, and if a positive clinical, serological and histological response was observed, celiac disease diagnosis was established. Finally, serum TG2-ab were used for comparison in all from whom they were available.

Serum immunoglobulin A (IgA)-class EmA were measured by an indirect immunofluorescence method using human umbilical cord as antigen[19]. A dilution of 1:5 was considered positive, and positive sera were further diluted 1:50, 1:100, 1:200, 1:500, 1:1000, 1:2000 and 1:4000. The antibody titers were further graded as low (1:5-1:200) and high (1:500-1:4000). Serum IgA-class TG2-ab were measured by enzyme-linked immunosorbent assay (ELISA) (Celikey, Phadia GmbH, Freiburg, Germany) according to the manufacturers’ instructions. Serum TG2-ab values ≥ 5.0 U were considered positive, and the values were further graded as low (5.0-29.9 U/L) and high (≥ 30 U/L)[13]. Total IgA values were tested in all subjects negative to the IgA class serological tests. In case of IgA deficiency the corresponding antibodies were measured in immunoglobulin G (IgG) class.

Upon upper gastrointestinal endoscopy a minimum of three forceps specimens were taken from the distal duodenum, and small-bowel mucosal morphology was determined from several well-oriented biopsy sections as previously described[17]. The degree of mucosal damage was further graded according to the Marsh-Oberhuber classification, where Marsh 0 represents normal mucosa, MarshI-II represents increased intraepithelial lymphocytosis without (I) or with (II) hyperplastic crypts and Marsh III partial (a), subtotal (b) or total (c) villous atrophy[20,21]. A patchy mucosal lesion was graded according to the most severe histological damage.

Genotyping of the participants for celiac disease-associated human leukocyte antigen (HLA)-DQB1*02 and DQB1*0302 alleles (DQ2 and DQ8) was performed using the DELFIA^® Celiac Disease Hybridization Assay (PerkinElmer Life and Analytical Sciences, Wallac Oy, Turku, Finland) or the SSP^TM DQB1 low resolution kit (Olerup SSP AB, Saltsjöbaden, Sweden) according to the manufacturer’s instructions.

χ² with cross-tabulation was used for statistical analysis. A P value less than 0.05 was considered statistically significant.

The study protocol was approved by the Ethics Committee of Tampere University Hospital. All subjects or their parents gave written informed consent.

In total, 405 EmA positive children and adults participated in the study. In 10 subjects the quality of the small-bowel biopsies was insufficient, in 14 EmA was determined as positive (1:5) without further dilution and in three subjects the clinical data were ambiguous. These cases were excluded from further statistical analyses. One patient had selective IgA deficiency and the corresponding antibodies were measured in the IgG class. Gastrointestinal symptoms remained the primary reason for celiac disease suspicion, but almost half of the patients were detected on the basis of extraintestinal symptoms or by screening of at-risk groups and the population (Table 1).

By definition, all participants were positive for EmA. Serum TG2-ab were measured in 316 EmA positive subjects and proved positive in 286 (91%) of them. Altogether 41% of the participants had high EmA and 54% high TG2-ab value defined at baseline. There was a significant association between serum TG2-ab level and clinical presentation, low antibody values being more common in the screen- than symptom-detected subjects (Table 2). A similar trend was observed with EmA, but the results were not statistically significant (P = 0.061).

Small-bowel mucosal villous atrophy and crypt hyperplasia (Marsh III) were found in altogether 85% of the EmA-positive subjects. There was a significant association between high antibody values and more severe small-bowel mucosal deterioration; in total 94% of those with high EmA titer evinced villous atrophy (Table 3). There was in this respect no significant difference between children and adults. The percentage of subjects evincing severe small-bowel mucosal damage increased progressively with higher EmA titers, but only the highest titer 1:4000 was 100% predictive of subsequent villous atrophy and crypt hyperplasia (Table 4).

In total, 40 patients had low and 17 high serum antibody values without simultaneous villous atrophy (Table 5). Irrespective of the baseline titers, 45 (79%) of these subjects (96% of those who remained on follow-up) either subsequently developed villous atrophy while on a gluten-containing diet, or experienced a positive clinical and serological response and disappearance of early mucosal changes on a gluten-free diet (Table 5). The presence of the celiac disease-associated HLA-DQ2 or DQ8 genotype was assessed in 299 EmA positive subjects and was found in all of them.

In our large series consisting of both children and adults, approximately half of the participants evinced high serum EmA levels, which was indicative of subsequent small-bowel mucosal villous damage in up to 94% of them. The results showed a high antibody titer to be an excellent predictor of villous atrophy and celiac disease also in high disease prevalence areas and in subjects with subtle or atypical symptoms. In the past few decades it has been observed that besides the classical gastrointestinal presentation, celiac disease patients may have a wide range of different extraintestinal symptoms. The patients may suffer for example from arthralgia or arthritis, osteoporosis, infertility and different neurological symptoms. In addition, screen-detected celiac patients may show only minor laboratory abnormalities or have no symptoms at all[3]. It was essential to investigate the performance of the celiac autoantibodies also in these atypical patients, as they are frequently seen in clinical practice, and may in fact represent the most common clinical presentation of celiac disease[16].

In patients with classical gastrointestinal celiac disease, Valdimarsson et al[8] observed a 100% positive predictive value of EmA for celiac disease in 19 adults, and suggested that histological confirmation might not be necessary in all such seropositive patients. Recently, similar results have been obtained with high values of TG2-ab. In a study by Barker et al[10], 48 out of 49 children having TG2-ab more than five times the upper limit of normal (ULN) had diagnostic small bowel mucosal damage. Likewise, Donaldson et al[11] observed a 100% positive predictive value for celiac disease by using the same cut-off level. In adults, Hill et al[13] suggested that TG2-ab levels more than ten times ULN would be exclusively indicative for celiac disease, and some other authors have presented comparable results[12].

Our findings thus largely accord with those in earlier studies carried out in specialized centers with high-risk patients having classical gastrointestinal presentation of celiac disease. Nevertheless, in the present study there was still a subpopulation of individuals in whom the current histological criteria were not fulfilled. Approximately 6 % of the participants with high and up to 21% of those with low antibody values had normal small-bowel mucosal villous structure, and in total this was seen in 15% of the EmA-positive subjects. It could thus be argued that EmA are not sufficiently specific for a definite diagnosis of celiac disease as such. Interestingly, however, there is an increasing body of data showing that EmA positivity is a very strong predictor of forthcoming celiac disease also in subjects with initially normal villi[17,22-26]. In line with this conception, almost all of our EmA positive patients who had no structural villous damage either evinced a positive serological, clinical and histological response to a gluten-free diet, or subsequently developed villous atrophy while on a normal diet. The existence of a celiac-type disorder in these individuals was further supported by the presence of the relevant HLA type in all in whom it was measured. There is thus strong evidence that, irrespective of the initially normal villous morphology, these EmA positive subjects are truly suffering from celiac disease.

Our study is subject to some limitations. First, although a high percentage of the participants had mild or atypical clinical presentation, the number of those found by population-based serological screening was rather low. Consequently, the results cannot be generalized to this patient group, and further studies are needed[27]. Secondly, the mucosal biopsies were taken from the distal duodenum as previously recommended[28]. Judging from recent evidence, however, villous atrophy can occasionally be detected only in the bulb area of the small intestine, and in theory celiac disease might in such cases already have been confirmed at the time of the first biopsy[7]. Nevertheless, interpretation of bulb specimens may be biased on Brunner glands or peptic inflammation, and their role in the diagnostics remains controversial. In addition, a patchy small-bowel mucosal lesion is always possible in celiac disease, which further highlights the importance of serology in the diagnosis.

Although EmA shows excellent specificity for an untreated celiac disease, it has certain limitations. The immunofluerescence method is laborious, time-consuming and always somewhat subjective. Since TG2-abs can be easily measured using a practical ELISA method, it would be tempting to use it instead of EmA. Nevertheless, TG2-ab are measured by commercial tests which use different epitopes of TG2 as antigen, and thus the specificity figures for the method have been somewhat inconsistent. Consequently, the positive predictive value of TG2-ab has sometimes been rather low, particularly in low-risk populations[29]. TG2-ab can also be positive in some conditions such as in liver diseases[30]. For these reasons, we decided to use the more laborious and time-consuming but celiac disease-specific EmA as the primary inclusion criterion in our series. As a consequence, the results should not be applied to TG2-ab positive EmA negative subjects. Finally, since antibody-negative subjects were not included in our study, the overall sensitivity of the serological tests could not be obtained.

To conclude, EmA positivity as such is a very strong predictor of a subsequent celiac disease diagnosis also in patients with low serum antibody titers and subtle or atypical clinical presentation. Judging from the findings here, invasive endoscopic studies might not be obligatory in all such seropositive patients.