The Prevalence of Celiac Disease-Specific Auto-Antibodies in Type 1 Diabetes in a Moroccan Population

Objective We aimed to determine the prevalence of specific auto-antibodies to celiac disease (CD) in Moroccan type 1 diabetic (T1D) patients and compare the clinical and biological characteristics of seropositive and seronegative cases. Patients and Methods A cross-sectional study was carried out on 276 T1D patients including 109 adults and 167 pediatric cases. The screening for CD was performed by an Elisa IgA anti-tissue transglutaminase antibody (tTGA) testing, combined with IgA quantification by nephelometry. Positive-IgA-tTGA cases were secondly tested for anti-endomysial antibodies (EMA) using an immunofluorescence technique, and the IgA deficiency cases were screened for IgG-tTGA. Patients with low positive tTGA titers underwent HLA-DQ2/DQ8 typing. Sociodemographic and clinical data of the patients were collected using a hetero-administered questionnaire. The comparison of clinical and biological data between seropositive and seronegative diabetics was done using independent T, Mann–Whitney U, chi-squared, and Fisher tests, which were considered significant if p value <0.05. Results The prevalence of CD-specific auto-antibodies was estimated to be 9.1% (IC = 95%), with 25 positive cases in tTGA and EMA testing. Eight cases displayed low titers of IgA-tTGA, among which 4 were positive for HLA-DQ2, 1 for HLA-DQ8, and 1 for both DQ2 and DQ8. The other 2 cases had a biopsy-proven CD. Compared to seronegative patients, seropositive cases had a higher percentage of associated autoimmune disorders (16% vs. 2.4%, p=0.008), with a significant lower height Z-scores (median: −0.90 (−3.93 to 0.95) vs. −0.51 (−4.54 to 2.18), p=0.029) and a higher HbA1c level (median: 11.30% (7.31 to 16.00) vs. 9.30% (4.40 to17.31), p=0.022). Conclusion The current study gave evidence of a high prevalence of CD specific auto-antibodies in T1D population. The co-existence of these two conditions was associated with a poor glycemic control, a lower height, and other autoimmune diseases. These findings may suggest the necessity of a systematic screening of CD in T1D patients.


Introduction
e association between celiac disease (CD) and type 1 diabetes (T1D) was first reported during the sixties [1]. is association has become nowadays well-known and more common with the development of sensitive and specific related serological tests. e prevalence of CD in T1D patients is reported to be 5 to 7 times greater than in the general population [2]. In a meta-analysis gathering 27 studies with a total of 26,605 T1D patients, Elfström found that the overall prevalence of biopsy-proven CD is almost 6% [3]. Till now, there are few relevant published data about the prevalence of CD in the Moroccan general population and also in common clinical conditions usually associated with CD such as diabetes or autoimmune thyroiditis (AIT) [4,5].
Both diseases share some common genetic background, mainly HLA molecules. In fact, the HLA-DQ2 genotype is found in 90% of celiac patients and in 55% of diabetics [6]. It was also reported that 33% of T1D patients who are homozygous for HLA-DQ2 are seropositive for anti-tTGA (tissue transglutaminase antibodies) versus 2% who are negative for HLA-DQ2/8 [7]. On the other hand, the role of gluten [8] and viral infections [9,10] in the etiopathogenesis of T1D is still a matter of controversy. Viral infections were also investigated in CD [11]. e presence of CD characterizes a subgroup of T1D patients that displays potential high risks of hypoglycemia [12,13], microvascular complications [14,15], and overall mortality [16]. erefore, most of the guidelines recommend a systematic screening for CD in TID [17][18][19].
e main objective of the study was to determine the prevalence of specific auto-antibodies to CD in T1D in a Moroccan adult and pediatric population and to describe the clinical and biological characteristics of seropositive diabetic patients.

Study Design.
A multicenter cross-sectional study was conducted in 276 adult and pediatric diabetic patients who were enrolled from five centers, including 2 pediatric centers, 2 adult endocrinology centers, and 1 specialized primary health care center. Patients were randomly recruited over a period of 23 months. e sample size of the study was calculated with an estimated precision of the prevalence equal to 3% with a confidence interval of 95%.
All patients were interviewed by the same investigator using a hetero-administered questionnaire focusing on most relevant symptoms of CD such as abdominal pain, bloating, diarrhea, constipation, vomiting, chronic fatigue, anorexia, and pallor, history of bone fractures and recent weight loss, ketoacidosis episodes, and hypoglycemic events that had occurred over a period of 6 months before the enrollment.

Laboratory Testing Methods
(i) Anti-tTG IgA and IgG testing was performed using commercial ELISA kits (tTG ELISA Orgentec, Diagnostika GmbH, threshold: 10 IU/ml). (ii) EMA testing was done by an indirect immunofluorescence assay (Euroimmun AG, Lübeck, threshold: 1 : 10), with a semiquantitative assessment. (iii) IgA class dosage was evaluated using a nephelometry system (BN Prospec, Siemens). (iv) HLA DQ2 and DQ8 typing were performed on the basis of PCR-SSP (polymerase chain reaction-sequence specific primers) technique (One Lambda, USA) realized on DNA extracted from patient samples. Positive tTGA and EMA testings or irresolute cases following the serology and HLA typing were recommended for intestinal biopsy to confirm or to rule out the diagnosis of CD.

Statistical
Analysis. e data were recorded and then analyzed using SPSS 16.0 statistical software. Descriptive statistics (mean, standard deviations, median, range, and proportions) were used to represent the trend and distribution of the results. e comparison between the 2 subgroups (celiac and diabetics vs. diabetics) was made using the independent T test and the Mann-Whitney U test for continuous outcomes and by chi-squared test and Fisher's exact test for proportions. Results were considered statistically significant when p value is less than 0.05.

Ethical Considerations.
According to the Declaration of Helsinki (1964), patients have been informed about the objectives of the study. An informed consent was obtained from patients or their tutors if minors, prior to enrollment.

Results
e mean age of patients was 14.1 ± 8 years, ranging from 1 to 42 years, with 144 females versus 132 males (sex-ratio M/ F � 0.91). e mean duration of diabetes was 5.4 ± 5.3 years. e overall seroprevalence of CD markers was 9.1% (IC � 95%) including 24 positive cases of IgA-tTGA and 1 positive case of IgG-tTGA that displayed an IgA deficiency (seronegative for IgA-tTGA). Among all positive cases, 17 of them had high titers of tTG antibodies (>100 IU/ml) and 8 others had titers ranging between 13.23 and 68 IU/ml. All the 24 positive cases on IgA-tTGA were also positive on EMA testing. e other one had IgA deficiency and had not been tested for IgA-EMA.
Among 8 cases of low IgA-tTGA titers, 4 were positive for HLADQ2, 1 was positive for HLADQ8, and 1 was positive for both HLADQ2 and DQ8. e two other cases, including the IgA deficiency case, had undergone an intestinal biopsy, confirming CD and went on GFD (Table 1). e mean age of the pediatric population was 9 ± 3.7 years, of which, 17 were seropositive for CD specific markers. In contrast, the adult population had a mean age of 21 ± 6 years, and 8 of them were seropositive as well. e other demographic, clinical, and biological characteristics of the two populations are shown in Table 2.
Among the associated T1D-tested auto-antibodies, 79.7% of cases had at least one positive auto-Ab, 42.4% had positive anti-GAD65 Ab, 36.2% had positive anti-ZnT8 Ab, 31.9% had positive anti-Insulin Ab, and 28% had anti-IA2 Ab. e frequency of auto-Ab categories is given in Tables 3  and 4, according to the population groups and CD-specific auto-Ab status, respectively.
According to the age at the onset of diabetes, the autoimmunity status of our patients (positive versus negative) was as follows: 63.2% vs. 67.9%, 29.5% vs. 21.4%, 6.4% vs. 10.7%, and 0.9% vs. 0% in less than 10 years, between 10 and 20 years, between 20 and 30 years, and superior to 30 years, respectively.

Discussion
Our results display a relatively high prevalence of CDspecific auto-antibodies in Moroccan T1D patients. Actually,  International Journal of Endocrinology  International Journal of Endocrinology a similar Moroccan study has shown a higher rate of tTGA positivity (13%), but regarding the very small sample of adult T1D patients studied (31 cases), these data seem to overestimate the real prevalence of associated auto-antibodies to CD [4]. In contrary and in a similar context, Azzi et al. [5] reviewed the records of 720 T1D pediatric patients and concluded a biopsy-proven CD prevalence of 3.05%. However, the authors have considered this rate to be a definite underestimation since most of patients were not systematically screened for CD. On the other hand, the prevalence rate found in our study is comparable to that reported by Barerra et al.  [27][28][29]. In fact, the reasons behind the variability of the prevalence rate of CD in T1D patients between different countries (Table 5) remain mostly unexplained [3]. On the other side, the potential impact of sex and age on the frequency of CD was assessed by Cerutti et al. and concluded in a large cohort of 4322 T1D and CD patients that female sex is an independent predictive risk factor for developing CD [33]. Similar findings have also been reported by many other authors [27,[34][35][36][37][38]. However, Pham-short et al. did not find such associations in their large meta-analysis gathering 26 studies [39]. Our study did not display a significant difference in the mean age between the 2 groups of patients as many similar studies did [27,31,34,40,41]. However, Cerutti et al. reported that the onset of diabetes before 4 years of age is an independent risk factor for developing CD (OR: 3.27 95% IC 2.20-4.85) [33]. Such observation is also supported by many other authors [25,31,34,36,40,42]. e results concerning the duration of diabetes remain, however, more controversial probably due to differences in CD screening strategies as well as the design of the studies.
We observed significant differences regarding the following clinical findings: pallor, chronic constipation, abdominal pain, bloating, and recent weight loss. ese findings are comparable to those of the study of Hansen et al. [31] that displayed significant differences between 33 CD-T1D patients compared to 236 T1D only, concerning abdominal pain, loose and/or frequent stool, bloating, constipation, arthralgia, tiredness, and frequent hypoglycemia,  International Journal of Endocrinology whereas no difference was reported about other symptoms like aphthous ulcers and tooth enamel defects. On another side, the impact of CD on growth in diabetic patients as well as the potential benefit of the gluten-free diet (GFD) on this parameter is still controversial. About that, Hansen et al. [31] found that at the time of CD diagnosis, CD-T1D patients had a smaller mean weight SD than T1D patients. Other authors reported similar results for weight SD [27,30,34,43] and height SD [20,30,34,43]. Our study showed a significant difference between the two groups in only height SD.
It is known that CD is generally associated with an increased risk of having other autoimmune diseases (AID). By the way, Not et al. [44] reported a strong association of CD with the presence of other AID in T1D patients (37.5% vs. 6.3%, p < 0.0001). e most frequent AID in such a population is AIT [31,33]. e overall percentage of AID among seropositive patients in our series is high and statistically significant (16%) but might be underestimated, because there was no systematic screening for AID in both diabetic and celiac patients. Actually, the seroconversion of CD happens in the first years of the T1D onset and almost 85% are diagnosed by 5 years [45].
Furthermore, it is recommended to screen for CD in T1D patients who particularly manifest unexplained bad glycemic control [46]. is recommendation is based on the study of Leeds et al. [13] who observed that CD-T1D patients have a higher HbA1c level in comparison with T1D adult patients (8.2% vs. 7.5%, p � 0.05). Similarly, our study confirms these findings and shows a significantly elevated median of HbA1c level in seropositive cases, in contrary to several series showing no significant difference [23,27,28,31,35,[47][48][49].
On the basis of the potential impact of CD on insulin requirements, some authors recommend a screening for CD in patients with unusual low doses of insulin. In fact, this may be explained essentially by malabsorption. Linked to that, a case-control study conducted by Mohn et al. noticed the reduction of insulin doses 12 months prior to CD diagnosis (0.6 UI/Kg vs. 0.9 UI/Kg, p � 0.05). e same observation was reported by Abid et al. and Poulain et al. [36,50]. e authors reported a rise in insulin requirements after the startup of a GFD [12,51]. Regarding our cases, we observed a tendency towards higher insulin doses in seropositive patients, which may reflect the differences in glycemic control for this group of patients.
T1D is associated with a 2-to 4-fold increase of fractures' global risk [52]. To the same extent, Heikkilä et al. has found in their meta-analysis that CD is associated with an increase in the global risk of fractures with an OR of 1.30 [53]. Furthermore, Lunt et al. demonstrated that BMD (bone mineral density) evaluated by DXA (dual-energy X-ray absorptiometry) in the lumbar region of nontreated adult celiac and type 1 diabetic women is inferior in comparison with type 1 diabetics only [54]. In our patient series, we observed a higher proportion of positive bone fractures' history in T1D patients compared to seronegative diabetics (20% vs. 8.8%, p � 0.081). is observation requires further investigation especially in the adult population.
On another side, the influence of CD status on the course of T1D remains controversial, especially regarding degenerative complications [55]. However, a Swedish study pinpointed the fact that it is the duration of CD that conditions the impact of CD on diabetic retinopathy [15]. e authors reported a beneficial effect the first 5 years (HR 0.57 95% CI: 0.36-0.91), no effect during the following 5 years (HR 1.03 95% CI: 0.68-1.57) and an aggravation after 10 years (HR of 2.83 after 10 years). Otherwise, CD is associated with a higher risk of end-stage renal disease in T1D patients, with a HR of 2.03 after 10 years [55]. Leed et al. have reported similar results concerning diabetic nephropathy, with a positive effect of GFD on renal markers after 1 year [13]. e effect of CD status on the severity of T1D onset has been studied by Rami et al. [49] and did not report any association. Our study came to the same conclusions. However, a clinical form of CD occurring before T1D has been reported in the literature, characterized by severe diabetes onset and a higher prevalence of other AID [56]. Similarly, our only patient for whom the diagnosis of CD was made before T1D onset corresponded to a case of IgA deficiency associated with epilepsy, with Addison's disease and asthma. Moreover, CD does not seem to be associated with a higher risk of ketoacidosis and diabetic coma during the course of T1D [30,49,57,58].
To the best of our knowledge, our study is the first relevant multicenter cross-sectional study conducted in Morocco. At the end, our results give evidence of a high CDspecific auto-antibodies positivity in T1D but remain relatively limited to well assess the impact of such associations on the course of T1D. In fact, the small number of seropositive patients and the necessity of a further intestinal biopsy as well as a follow-up are the factors to be considered in the conclusions to be drawn from these data. With the same perspectives in mind, we have communicated the findings of our CD laboratory testing to the physicians for further investigation (biopsy) and GFD if required, with an active surveillance.

Conclusion
e current study concluded a high prevalence of CDspecific antibodies in Moroccan T1D patients. In comparison with the diabetic group, seropositive celiac and diabetic patients displayed significant association with poor glycemic control, lower height SD, and an increased frequency of other AIDs. ese findings emphasize the benefit of a systematic screening for CD in diabetic patients, which can hopefully minimize the potential risk of complications due to undiagnosed CD. Furthermore, controlled longitudinal studies in a larger sample with an assessment of the potential benefit of a GFD, especially in asymptomatic patients, are needed.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest.