Comparison of the Reliability of 17 Celiac Disease Associated Bio-Markers to Reflect Intestinal Damage

Copyright: © 2017 Lerner A, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Comparison of the Reliability of 17 Celiac Disease Associated Bio-Markers to Reflect Intestinal Damage Aaron Lerner1,2*, Patricia Jeremias2, Sandra Neidhöfer2 and Torsten Matthias2 1Rappaport School of Medicine, Technion-Israel Institute of Technology, Haifa, Israel 2ESKU.KIPP Institute, Wendelsheim, Germany


Introduction
Celiac disease (CD) is a small intestine inflammatory autoimmune disorder in genetically susceptible individuals, triggered by digestion of prolamins contained in wheat, barley, rye and partially in oat. The accepted incidence in Western countries is 1-1.5%. The majority of patients are still under-diagnosed, but increasing awareness and improved serological performances, raise the diagnostic yield. In contrast, in high-risk populations the average risk of CD can reach 5-10% [1]. In the last decades CD frequency is increasing, joining the surge in autoimmune disease incidence and prevalence worldwide [2,3].
CD is a life-long multi-faced condition with an increased risk of complications. Hematological and gastrointestinal malignancies, osteoporosis/penia and other extraintestinal manifestations, decreased height, malnutrition and nutritional deficiencies, fertility impairment, stillbirth, dismaturity, hypercoagulability, psychosocial compromise, impairment of quality of life and-if left untreated-increased mortality and additional autoimmune conditions, to name a few [4]. The epidemiology and phenotype of CD are constantly changing, early diagnosis and subsequent adherence to a gluten-free diet is highly recommended. It has been shown that the classic intestinal clinical picture of malnutrition, chronic diarrhea and nutritional deficiencies are disappearing and extraintestinal presentations are emerging. Nowadays, we are witnessing an epidemiological shift in the disease phenotype toward a more advanced age, and increased prevalence of latent, hyposymptomatic or asymptomatic presentations [5]. All these changes make reliance on symptomatology more remote and therefore diagnosis of the disease more difficult [6]. Recent, major improvements in serological markers performance, with sensitivities and specificities above 90%, are an additional reason why serological screening of CD has achieved prime importance [1].
In the beginning of 2012 the European Society of Pediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) released a new set of guidelines for the diagnosis of CD, based on both: the significant progress in the development of specific antibody tests and the understanding of the high prevalence of specific HLA haplotypes [7]. These advances were the base of updated guidelines for CD diagnosis in symptomatic and asymptomatic children, where, in certain circumstances, small bowel biopsy can be omitted. Contrasting, multiple adult guidelines still advocate serological screening followed by obligatory intestinal biopsies for the diagnosis.
Since only the tip of the CD iceberg is above the waterline and a much larger portion remains under water undetected, it can be expected that prevalence of the disease will increase continuously. Therefore, presenting symptoms will continue to change towards a-/ hyposymptomatic, supporting the need for improved serological screening methods. There are multiple serological tests on the market: IgA anti-endomysial antibody (EMA), IgA and/or IgG tissue transglutaminase (tTg), IgA and/or IgG deamidated gliadin peptide (DGP), whereby IgA-tTg is the most frequently used and ESPGHAN's recommended one. Several combination tests are available too: DGP IgA+G, tTg IgA+G, Gliadin IgA+G (check), CeliCheck IgA, IgG or IgA+IgG (autoantibodies against the cross-linked complex of tTg and gliadin peptides (tTg neo-epitope) and the recently described IgG and/ or IgA or IgG+IgA combined mTg neo-epitope and mTg (AESKULISA® mTg neo-epitope and mTg, RUO) [1]. Being a new serological biomarker of CD mTg neo-epitope deserves some background information. Microbial transglutaminase (mTg) is capable of crosslinking numerous molecules (including gliadin), thereby creating an mTg-gliadin neo-complex. This post-translational modification of gliadin imitates its family member (human tTg) involved in CD.
For the last 26 years, numerous studies evaluated only one serological marker in relation to intestinal pathology, mainly EMA and later on, tTg [8,9]. Fewer studies checked two antibodies, mainly IgA-tTg and IgA-EMA, the latest being in 2012-13 [10][11][12][13]. Only in two Israeli studies, 5 different antibodies' levels were correlated to the mucosal damage [14,15]. Parizade M. et al. found that high antibody concentrations were predictive of villous atrophies for IgA+IgG-tTg, IgA+IgG-DGP and IgA-EMA [14], where EMA and IgA-tTg stood out to be the best. The second multi-marker study evaluated IgA-tTg, IgA+G-DGP, and IgA+G-EMA were the last one had the best correlation [15]. The present study extended the serological repertoire and checked, for the first time, 17 sub-types of the most frequently used antibodies to screen in order to diagnose CD.
In view of the increasing importance of serological biomarkers for screening and diagnosis of CD, their differential performance, and the lack of back to back comparison, we undertook the task to evaluate the reliability of those individual or combined antibodies (ABs) to reflect the intestinal histological injury in CD children.

Patient populations
Three groups of patients were investigated: Group 1: 95 pediatric CD patients (CD), mean age 8.3 ± 4.4 years, F/M (1/0.9). The CD group was divided according to the degree of intestinal injury, using Marsh criteria, to 6 groups M0, MI, MII, MIII a/b/c. With M0 representing a normal intestinal biopsy and MIII c total villous atrophy [16]. Those CD sub-groups contained MI=7, MII=13, MIII a=41, MIII b=27, MIII c=7 children, respectively. The CD and the AP groups underwent esophago-gastroduodenoscopy using GIF-xp 20 endoscope (pentax, Tokyo, Japan). At least 5 biopsies were obtained: 4 from the second part of the duodenum and 1 from the antrum, for the diagnosis or exclusion of CD. The biopsies were immediately fixed in buffered formalin and embedded on edge in paraffin. Sections were stained with hematoxylin-eosin and Giemsa, analyzed by the pathologist and graded according to Marsh criteria, as previously described [16].
Celiac disease was diagnosed according to the revised criteria of the European Society for Pediatric Gastroenterology and nutrition, based on specific serology (anti-tissue transglutaminase antibodies, by ELISA) and duodenal biopsies [17]. All participants were on gluten containing diet and had physical examination, laboratory work-up and celiac serology. On the day of endoscopy, 5 ml of peripheral blood was withdrawn, centrifuged at 5000 c/sec for 10 minutes and the serum was kept in -80° Celsius until used in serology assays.
Group 3: 99 normal children, recruited from a normal school, mean age 8.5 ± 4.2 years, F\M 1:0.7, served as a normal control group (NC).
The study was approved by the Carmel Medical Center Helsinki committee and participants or legal guardians signed an informed consent.

Antibody determination by IFA
AESKUSLIDES® EMA (endomysium antibodies) were used for indirect fluorescent IgA and IgG determination of human antiendomysium antibodies, according to the manufacturer instruction.
The endomysium is the supporting, conective structure that surrounds the smooth fibers, located in the middle third of the esophagus. A representative positive EMA is shown in Figure 1.

Statistics
Statistical analyses were performed using the Software MedCalc® (V 15.6.1). Normally distributed values were expressed as mean ± standard deviation (SD) and were compared by the Student`s t test. Correlations were assessed by the Pearson correlation test. p<0.05 was considered as significant.
Scatter diagrams and regression analysis comparing the 17 antibodies' OD activities to the degrees of the intestinal damage were correlated. Antibody results were compared to the degree of intestinal injury, using revised Marsh criteria.

Results
The performance of the different tests for the 99 pediatric CD, were compared to 45 pathological controls with AP and 99 normal control children, with similar age and sex ratio. All antibodies were detected via ELISA, except for EMA which was checked via immunofluorescence.
A general overview of the different IgA and IgG antibodies' U/ml titers is shown in Figure 2. All CD associated mean antibody's levels were above the cut-off levels. IgA isotype (Figure 2A) levels were higher than of IgG isotypes ( Figure 2B), except for the DGP-IgA and mTg-neo IgA, here IgG dominated. The tTg neo-epitope IgA test had the highest immunoreactivity (P<0.0001) and tTg neo-epitope IgG had a higher immunoreactivity compared to single tTg-IgG. The following Figures  3-6 correlate the various ABs activities to the degree of intestinal damage, as characterized by Marsh criteria.
All assays were able to differentiate between patients of low and high degree intestinal damage. Comparing the different correlations between CD associated IgA and IgG antibodies' isotypes, EMA-IgA, tTg-neo IgA and tTg-neo check, stood out significantly as the best indicators of CD caused intestinal damage. The highest OD values (mean 2.94 ± 1.2, p<0.0001) were achieved using the tTg-neo IgA ELISA with Marsh 3c patients (Figure 3). The correlations, as well as their corresponding statistical significance of each AB isotype, are summarized (in increasing order) in Table 1.
Since EMA testing is performed via immunofluorescence ( Figure  1), which is semi quantitative and operator-depended, it is shown separately in Figure 6. The IgA and not the IgG isotype reflects the intestinal pathology best.
For better comparison of diagnostic reliability, Table 1 lists various ABs' activities of 5 combined (checks) and 12 single isotypes associated with CD. Many of the antibodies had high specificity and low sensitivity. Combining the specificity and sensitivity by analyzing the AUC, tTgneo IgA and EMA-IgA, in the single isotype, and tTg-neo check, in the combined isotype, stood out as best performers. Similarly, tTg-neo check and EMA-IgA reflect best the intestinal damage (p<0.005, Table  1). Of note, mTg-neo IgG and not the IgA isotype, had a high AUC (0.95) and reflected significantly the intestinal histology in the CD group.

Discussion
The present study explored 15 ELISA kits encompassing 5 different families of antigens (gliadin, tTg, DGP and tTg neo-epitope and mTg neo-epitope) for specific IgA, IgG, combined IgA+G reactivities and 2 immunofluorescence EMA kits. The performances of all Kits were assessed with the same blood sample, and then compared against the rigid criterion of the corresponding biopsy-demonstrated duodenal histology, further enhancing the uniformity of the study. The analysis showed that antibody levels were well correlated to the duodenal damage degree in CD children, but there was a hierarchy among the evaluated tests. The tTg neo-epitope IgA Kit occupied the first place in the IgA isotype group, DGP-IgG and tTg neo-epitope IgG shared the first place of IgG isotype Tests. The combined IgA+IgG isotypes had low correlations to intestinal histology, except for the tTg-neo check, which showed better performance than tTg-neo IgG or IgA kits . Concerning the differential performance of Kit groups, IgA Kits surpassed IgG in reflecting intestinal damage, except for the mTg-neo IgG, which showed better performance than the mTg-neo IgA kit.
Our own experience [16,18,19] and that of many others [1] favors combination tests to screen for CD. The main single antigen ELISA kit candidates are anti-tTg IgA and anti-DGP IgG, competing with the new CeliCheck combination of IgA and IgG thus omitting screening for IgA deficiency [16,[18][19][20][21]. However, when reflection of intestinal pathology is concerned, single isotype Test are preferable, mainly the tTg-neo IgA.
The tTg-neo IgA kit uses a neo-epitope formed by complexing tTg and gliadin, the main antigens in CD. The basic idea is that tTg not only able to deamidate gliadin peptides but also to cross-link with a high catalytic rate [22][23][24]. In the latter case, tTg/gliadin complexes are formed, resulting in the formation of new epitopes (neo-epitopes), evidence has been shown in-vitro. More so, formation of the tTg-DGP complex was suggested to involve epitope spreading from gliadin to tTg [21]. The antibodies against neo-epitopes of the tTg-gliadin complex provide a new screening and diagnostic test in CD. Multiple studies have exhibited diagnostic sensitivities of 95% and specificities of 97% or more, when compared with those of traditional antibody assays [25,26]. The neo-tTg/DGP complex is potentially able to drive the development of newly formed epitopes derived from the cross-linkage between the enzyme and the substrate. It is foreseeable that the autoantibodies generated against the neo-epitope may represent the best means for screening populations, diagnosing high-risk groups and identify silent or latent patients. In fact, several studies have shown the superiority of screening for CD using the neo-tTg/DGPs complex strategy in the general population [26,27] or in high-risk groups' subjects [16,18,19,[28][29][30]. The present results go along this serological diagnostic strategy and add the aspect of intestinal damage reflection.
Several debates and disagreements exist in the literature concerning the cross talks between serological markers and reflection of the intestinal pathology in CD.
On the pathological aspect a concern exists that intestinal biopsy is the gold standard for diagnosis. Recently incorrect biopsy interpretation causing under diagnosis [31], lack of uniformity in the use of Marsh-Oberhuber classification [32] and unrecognized, misleading pathological features that are positively associated with more advanced stages of the disease were described [33]. More so, even the cut-off for the intraepithelial lymphocyte count, a hallmark of CD intestinal pathology, is debatable [34][35]. In the serological domain assays uniformity, lack of standardization, plethora of commercial immune fluorescent and ELISA kits with variable cut-off levels are some      of the problematic aspects. On top of it, being on gluten containing diet, CD associated antibodies fluctuates or disappears spontaneously in CD children [36]. Positive celiac serology coexists with normal villous morphology [37]. Normalization of serology does not predict normalization of intestinal histology [38]. A recent study from the UK national external quality assessment service center (UK NEQAS) states that not all commercial available IgA-tTg kits are reliable and that the ESPGHAN guidelines are not readily transferable to use in all centers and should not be used in the UK [39]. Interestingly, even in the most recommended autoantibodies for CD diagnosis, multiple false positive and negative exist [40].
Screening some recent reviews and guidelines, comparing reliability of the celiac specific serological markers, one sees the vast variability in their sensitivity and specificities [1,7,[41][42][43].
The present study has several advantages: 1. Much more CD associated antibodies were evaluated in a backto-back experimental design (12 vs. 5).
2. It is the first study to incorporate the neo-epitopes of tTg/mTg in correlation to the intestinal injury.
3. The group of AP is unique in composing pathological controls with normal intestinal biopsy. 4. It is the only study that compared 17 CD associated serological markers, encompassing IgA, IgG and check isotypes.
5. The addition of the IgA+IgG-AGA is important in reference to less recent historical data. 6. All determinations were done in the same laboratory, by the same person, under same conditions for better uniformity and comparability.
7. The antibodies' activities were correlated to the same intestinal Marsh degree of atrophies.   On the other hand, several limitations should be mentioned: 1. A multicenter study would have better represented the global clinical reality. Moreover, a multicenter-approach could have normalized the potential biases encountered in one single center.
2. Comparing extended study populations would have increased the statistical power and the conclusion's reliability.
3. The present study represents only one aspect (intestinal pathology reflection) in the decisional algorithm of the best serological marker for CD screening and diagnosis.
For transparency, some of the data were published separately: tTg IgA compared to tTg-neo isotypes [24] mTg-neo isotypes compared to other ABs (but not to DGP and EMA isotypes) [44] Since mTg-neo ABs isotypes are the newest published CD biomarkers of the 17 presently compared, in the following, some facts will be summarized. mTg-neo ABs isotypes are not autoimmune ABs, they are directed against neo-epitopes, formed during the conformational changes happening in the formation mTg/ gliadin cross-link complex [22,[44][45][46]. Since mTg is heavily used by the processed food industries, it is consumed daily [44,47]. mTg neoepitopes have, most recently, been shown to be immunogenic in active CD patients [44,46]. Presently, it represents a good reflector of CD intestinal atrophy, although, not the best one. CD is an IgA mediated disease, so it is not astonishing that antibodies against tTg and tTgneo are of IgA isotype. Interestingly, the predominant antibody against mTg-neo is of IgG isotype, reflecting the immune response against an following advantages over the single tTg-IgA ELISAs: • Higher sera reactivity, Better reflection of intestinal pathology.
• Higher sensitivity, though comparable specificity, directed against different/new epitopes compared to the tTg antibody.
• It is suitable for IgA deficient patients, since it includes IgG isotype.
In the hierarchy of the 17 bio-markers, the three autoantibodies best reflecting CD intestinal damage, tTg-neo IgA, tTg-neo check and EMA IgA, won the competition. Therefore, it is suggested that tTg neoepitope antibodies should be preferably used to screen, diagnose and monitor compliance in CD patients. The mTg neo-epitope IgG represent a novel serological biomarker for CD. Its significance, pathogenic role, involvement in other autoimmune diseases or in non-celiac gluten sensitivity, awaits further scientific explorations.