1. Introduction
Celiac disease (CD) is considered a systemic immune-mediated disorder elicited by gluten and related prolamines in genetically susceptible individuals [
1]. The presence in the serum of anti-tissue-transglutaminase 2 (anti-TG2) antibodies is considered a sensitive and specific biomarker for CD [
2].
Anti-TG2 antibodies are produced primarily at an intestinal level by specific B lymphocytes and, using phage display library technology, it has been proven that these antibodies are synthesized in CD patients with preferential use of IGVH5-51 gene family [
3]. Once produced, anti-TG2 antibodies can deposit in the small intestinal mucosa [
4], even before they can be detected in the circulation [
5,
6]. Furthermore, TG2 specific IgA deposition precedes the formation of small intestinal lesion in CD [
6,
7] as it can be detected also in potential CD patients [
6,
8,
9], i.e., patients showing positive serology for CD-specific antibodies, but architecturally normal intestinal mucosa [
10]. The presence of mucosal deposits of anti-TG2 antibodies seems to be a predictive marker of future evolution to villous atrophy, as it was seen more often in those potential patients that progress to overt CD than in those remaining in early phase of disease [
5,
6,
8].
Several investigators have shown that intestinal anti-TG2 IgA deposits are also detectable in subjects with diagnosis other than CD with a prevalence ranging from 5 to 20% [
11,
12,
13]; in our experience, up to 12% of pediatric non-CD patients shows mucosal deposits [
13]. A similar percentage of non-CD patients producing anti-TG2 antibodies only at an intestinal level has been recently shown by us by measuring anti-TG2 antibodies secreted into supernatants from organ culture of small intestinal biopsy [
9]. This phenomenon could be not specific for CD and merely attributable to intestinal inflammation; alternatively, it could be suggestive of a very early condition of gluten reactivity. The subjects initially producing anti-TG2 antibodies only at an intestinal level could become serum positive later and eventually develop small intestinal mucosal damage.
Here we aimed to investigate the intestinal production of anti-TG2 antibodies in patients with normal serum levels of such antibodies and to relate that production to immunohistochemical markers of gluten-dependent enteropathy.
3. Results
3.1. Intestinal Anti-TG2 Antibodies Are Also Produced in Non-CD Patients
Mucosal deposits of IgA anti-TG2 antibodies were detected in 40 of 44 Active CD (90.9%), 72 of 103 Potential CD (69.6%), and 25 of 136 Non-CD patients (18.3%). In CD patients, they presented as thick bands under surface epithelium, around crypts and vessels. In potential CD subjects the bands were less thick with a patchy distribution. In patients with diagnosis other than CD, the staining appeared weak, in some cases deposits were evident only around some crypts.
On the other hand, in all CD patients in the active phase of disease (100%), in 101/103 (98%) patients with potential CD and in 13/136 subjects (10%) with diagnosis other than CD, anti-TG2 antibodies could be measured in culture supernatants. Anti-TG2 antibody titers ranged from 4.21 to 2020.3 U/mL in CD and from 3.3 to 29.3 U/mL in Non-CD patients.
Our data showed that in total 33/136 Non-CD patients (24.3%) had evidence of intestinal production of anti-TG2 antibodies: they were defined below as Positive Non-CD patients. Five of 136 Non-CD subjects were found positive for both assays.
The 33 Positive Non-CD patients had following diagnoses: type 1 diabetes (n = 10), gastroesophageal reflux (n = 9), gastrointestinal functional disorders (n = 6), first degree relatives of CD patients (n = 3), eosinophilic esophagitis (n = 3), HP infection (n = 2). 29/33 patients had Marsh 0 (n = 24) or Marsh 1 (n = 5) duodenal mucosa and the remaining 4 patients had a Marsh 3a mucosa.
All CD patients had HLA-DQ2 and/or DQ8 haplotypes. The celiac disease-type HLA was known in 14/33 Positive Non-CD and in 5/31 Negative Non-CD patients. It was found that 12/13 Positive Non-CD were HLA-DQ2 and/or -DQ8 positive, 2 patients showed DQA1*05-DQB1*03 haplotype, the most frequent HLA haplotype in DQ2/DQ8 negative CD patients [
21]. Three of the five Negative Non-CD patients were HLA-DQ2 positive, the last two were HLA-DQ2/DQ8 negative.
3.2. Phage Display Technology Confirms the Intestinal Production of Anti-TG2 Antibodies in Non-CD Patients
To confirm the intestinal production of anti-TG2 antibodies in positive Non-CD patients, we created IGVH5-51 antibody libraries against TG2 from biopsy specimens of 8/33 Positive Non-CD patients (six positive to detection of mucosal deposits of anti-TG2 antibodies and two who secreted anti-TG2 into culture supernatants); two Non-CD patients negative to both the above mentioned tests served as controls (Negative Non-CD).
Antibody clones recognizing TG2 were isolated from intestinal biopsies of all Positive Non-CD patients. The number of anti-TG2 antibody clones ranged from 2.2% to 13.3%. No difference in percentage of anti-TG2 specific clones was observed between the patients positive only to detection of mucosal deposits and the patients positive only to measurement of anti-TG2 antibodies into culture supernatants. No anti-TG2 antibody clones were isolated from biopsy samples of the two patients double negative to the above mentioned tests (
Table 1). All positive Non-CD patients, with the exception of the two affected by type 1 diabetes, produced anti-TG2 antibodies with a preferential use of VH5 gene family. The type 1 diabetes patients produced anti-TG2 antibodies mainly belonging to the VH3 gene family.
3.3. Positive Non-CD Patients Presented Signs of Activated Cell-Mediated Mucosal Immunity
In order to assess the activation of mucosal immunity and in particular of a gluten-related deranged immune response, the density of mononuclear cells expressing CD25 in
lamina propria and the density of CD3
+ and TCR-γδ
+ cells in epithelial compartment were investigated in 32/33 Positive Non-CD patients and in 31/103 Non-CD patients negative to both above-mentioned tests (Negative Non-CD) that served as control group. Twelve of 32 (37%) Positive Non-CD and 5/31 (16%) Negative Non-CD presented with alteration of two or more of the above-mentioned markers, (
Table 2).
In the Positive Non-CD group, 7 of 32 patients (21.8%) presented increased number of CD3
+ intraepithelial lymphocytes (IELs) (mean ± SD = 28.9 ± 20.5 cells) (
Figure 1A), 11 of 32 patients (34.3%) increased number of TCR-γδ
+ IELs (mean ± SD = 2.6 ± 2.5 cells) (
Figure 1B) and 24 of 32 (75.0%) increased number of
lamina propria CD25
+ mononuclear cells (mean ± SD = 13.2. ± 15.2 cells) (
Figure 1C). On the contrary in the Negative Non-CD group, high number of CD3
+ (26.1 ± 11.0), TCR-γδ
+ (2.0 ± 1.5) and CD25
+ (5.7 ± 4.3) cells were found in 5/31 (16.1%), 6/31 (19.3%) and 14/31 (45.1%) patients, respectively (
Table 2). Only the percentage of subjects with an increased number of CD25
+ mononuclear cells in lamina propria was significantly higher in Positive Non-CD in comparison to Negative Non-CD patients (
p < 0.05) (
Table 2). Likewise the density of CD25
+ cells in lamina propria was significantly higher in Positive Non-CD (13.2 ± 15.12 cells) than that in Negative Non-CD (5.7 ± 4.31 cells) (
p < 0.01) (
Figure 2). Finally, we evaluated TCR-γδ
+/CD3
+ ratio that is considered a more reliable marker of gluten sensitivity than CD3 and TCR-γδ values considered singularly [
20]. 41% (13/32) of Positive Non-CD showed TCR-γδ
+/CD3
+ ratio over cut-off in comparison to 29% (9/31) of Negative Non-CD (
Table 2). The diagnosis of those with high TCR-γδ
+/CD3
+ ratio were: gastroesophageal reflux (
n = 6), type 1 diabetes (
n = 4), eosinophilic esophagitis (
n = 2), gastrointestinal functional disorders (
n = 1).
4. Discussion
In this retrospective study we showed that the great majority of active CD patients produce anti-TG2 antibodies at the level of the intestinal mucosa. In our laboratories, two tests were used to detect intestinal anti-TG2 antibodies: the first is based on the co-localization of antibodies against TG2 and their natural tissue antigen in duodenal cryostat sections by immunofluorescence [
4,
16]; the second one is an ELISA test performed on supernatants from duodenal biopsy culture [
9] and, in this pediatric population, it has a higher sensitivity compared to deposits detection, confirming our previous data [
9,
13]. Moreover, ELISA test detects spontaneous intestinal production of CD-specific autoantibodies even in the very early stage of the disease. Indeed, it results positive in 98% of potential CD, whose intestinal mucosa does not show signs of damage and in whom serum levels of anti-TG2 antibodies are lower than those found in the active phase of disease [
22]. Of note, mucosal deposits are detectable only in about 70% of pediatric patients with potential CD, thus suggesting that even very low antibody production may be detected by ELISA testing in culture supernatants. The absence of mucosal deposits of anti-TG2 antibodies, more often noted in potential CD, could be due to the low affinity displayed by these antibodies to their antigen and/or to their lower concentration.
A highly sensitive test able to demonstrate the intestinal production of anti-TG2 antibodies could be relevant in clinical practice for its possible ability to predict evolution towards mucosal damage. Moreover, the ability to detect a very early production of intestinal anti-TG2 antibodies could be helpful to identify a condition of gluten sensitivity in patients with absence of serum CD-associated autoantibodies [
23,
24].
On the other hand, we showed that the above-mentioned tests have a high specificity even though they do not reach 100%. In this particular population, approximately 25% of Non-CD patients produce anti-TG2 antibodies only at intestinal level. Our data showed that some of those patients without circulating anti-TG2 antibodies exhibit mucosal deposits of anti-TG2 antibodies with a pattern similar to that observed in potential CD and/or spontaneously secrete the same antibodies in supernatants after 24 hours of small intestine culture. The results obtained using the phage display technologies definitely support the above-mentioned data. Indeed, the construction of antibody libraries from biopsy samples showed a production of an anti-TG2 specific antibody repertoire dominated by the usage of VH5-51 gene segment (in 75% of patients), the same used by active CD patients [
3,
25], and by the VH3 (in 25%) gene segment. Of interest, the only two first degree relatives of CD patients included in the phage display library analysis of this study showed a VH5-51 gene usage, confirming a previous report by Not et al. [
24]. Furthermore, two of the patients studied by phage display library have type 1 diabetes and mainly produce VH3 anti-TG2 antibodies, as we already previously published [
26]. With the same technology we were not able to isolate TG2-specific antibody clones from Non-CD patients who did not present any sign of intestinal anti-TG2 antibody production. Altogether, our data showed an intestinal production of anti-TG2 antibodies in a subgroup of Non-CD patients.
The demonstration of intestinal production of anti-TG2 antibodies in non-CD patients, poses the question of whether this finding is related to a deranged mucosal immune response to gluten, as shown in other settings, such as in first-degree relatives of CD patients [
24] or in patients with genetic gluten intolerance [
24,
27]. Notably, individuals producing intestinal anti-TG2 antibodies, taken as a group, more often (75% of them) show inflammatory signs, demonstrated by the increased density of CD25
+ cells, in the small intestinal mucosa as compared to Non-CD subjects lacking evidence of local anti-TG2 antibodies production. Furthermore, they tend to present increased density of markers considered specific for gluten-sensitive enteropathy, such as TCR-γδ
+ intraepithelial lymphocytes [
28] and increased value of TCR-γδ
+/CD3
+ ratio [
20]. On the other hand, it is known that the increase of intraepithelial lymphocytes, peculiar of Marsh 1 duodenal lesion, and also the higher intraepithelial density of TCR-γδ
+ lymphocytes have no absolute specificity for CD [
29,
30].
Our findings are not surprising for subjects with type 1 diabetes and first-degree relatives of CD patients included in our study group, since they belong to groups at risk to develop CD [
28,
31], but are intriguing for individuals unrelated to CD. However, the association between gastrointestinal disorders such as eosinophilic oesophagitis, gastro-oesophageal reflux disease, or
Helicobacter pylori infection and intestinal production of anti-TG2 antibodies is to date unknown as yet little explored.
HLA typing could be of help as the absence of HLA-DQ2/-DQ8 alleles in these patients would suggest the hypothesis that the production of anti-TG2 antibodies at intestinal level is not exclusive of CD. Although we have typed for HLA only 42% of intestinal anti-TG2 positive Non-CD patients, all cases but two turned out to be HLA-DQ2/-DQ8 positive and the two negative cases have DQA1*05-DQB1*03 haplotype, the most frequent HLA haplotype in DQ2/DQ8 negative CD patients [
21].
Both possibilities remain then open: intestinal anti-TG2 IgA production could be due to a general state of inflammation or could indicate a very early stage of a gluten-dependent pathology. The effect of a gluten free diet and a more prolonged clinical follow-up could help to establish whether our finding is expression of a gluten-dependent mucosal response and if those subjects belong to the spectrum of gluten-related disorders.