HLA Haplotype Association with Celiac Disease in Albanian Pediatric Patients from Kosovo

Genetic predisposition to celiac disease (CD) is strongly associated with the presence of HLA alleles in the individual genotype encoding HLA-DQ2 and/or HLA-DQ8 heterodimers. The main aim of this study was to analyze the HLA-A, -B, -DRB1, and -DQ allele and five-locus haplotype frequencies in 60 Albanian pediatric CD patients and 124 non-CD children from Kosovo. The most prevalent haplotype in patients was the ancestral AH 8.1 haplotype present in 22.5% of the cases compared to 2.8% of the controls (P < 0.0001). Additionally, two other haplotypes were also overrepresented in patients (HLA-A∗02~B∗50~DRB1∗07~DQA1∗02:01~DQB1∗02:02 and HLA-A∗68~B∗44~DRB1∗07~DQA1∗02:01~DQB1∗02:02). Analysis showed that 95.0% of CD patients and 43.3% of controls were carriers of HLA-DQ2 and/or HLA-DQ8 heterodimers. The most frequent CD-predisposing HLA-DQ haplotypes in patients were HLA-DQ2.5 (46.7%) and HLA-DQ2.2 (11.6%), while the most prevalent genotypes were HLA-DQ2.5/DQX (58.3%) and HLA-DQ2.5/DQ2.2 (20.0%). The frequency of the HLA-DQ8 heterodimer among CD patients (4.2%) compared to the control group (8.1%) was without statistical significance. The given data demonstrate differences in the distribution of HLA haplotypes among Albanian CD patients from Kosovo in comparison to other European and non-European populations, as well as provide additional population data to supplement the thus far undisputed importance of the role of HLA-DQ2 and HLA-DQ8 heterodimers in the development of CD.


Introduction
Celiac disease (CD) is an autoimmune systemic disorder characterized by chronic inflammation of the small intestinal mucosa triggered by gluten and related prolamine uptake that occurs in genetically susceptible individuals [1]. Europe is historically considered a geographical area with a high incidence of CD, with a prevalence of 1%, which may be higher in Northern European countries [2,3]. Some studies from Scandinavian countries and the United Kingdom population tended to show a higher prevalence of CD of approximately 1.0-1.5% [4]. Although it was thought that some countries, including the United States, were exempt from this disease, it has recently been shown that it has a similar prevalence as in Europe, 0.5-1.0% of the general population [5].
Susceptibility to CD and its activation and perpetuation involve a combination of environmental and genetic factors through some immunological mechanisms [6]. The involvement of Human Leukocyte Antigens (HLA) in CD pathogenesis was first described in the 1970s [7]. Over the subsequent years, the specific alleles that underlie the described associations became clear: CD is strongly associated with HLA-DQ, which encodes HLA-DQ2 and HLA-DQ8 heterodimers.
Most of the studies exploring the linkage of HLA genes with CD are focused on HLA class II genes encoding HLA-DQ heterodimers, while very few studies investigate extended HLA haplotypes [11]. In contrast to a relatively conserved HLA-DR3 haplotype, the linkage between HLA loci forming the HLA-DR7 and HLA-DR11 haplotypes are not so strong, and since they can form the HLA-DQ2.5trans heterodimer, we wanted to explore if there is a difference in HLA haplotype distribution among CD patients in comparison to healthy subjects.
The importance of genetic factors in the pathogenesis of CD is well documented by many studies. It is clear that CD rarely develops in the absence of HLA-DQ2 and/or HLA-DQ8 heterodimers, and that the predisposing HLA-DQ2 and HLA-DQ8 subtypes are necessary, but not sufficient for causing the disease [12]. Anderson et al. suggested that a combination of HLA typing and confirmatory serology could reduce the number of unnecessary endoscopies as well as the number of false negatives and/or positive diagnoses [13]. In 2012, the role of HLA in the diagnosis of CD was firmly recognized, which resulted in important changes in diagnostic criteria and the inclusion of HLA typing in CD diagnostic guidelines [1].
As far as we know, no previous study has focused on the frequency of CD-predisposing HLA genotypes in affected and nonaffected individuals in Kosovar Albanian children. Thus, the main aim of this research was to analyze the HLA-A~B~DRB1~DQA1~DQB1 haplotype distribution as well as the frequency of CD-predisposing HLA-DQ genotypes in Albanian pediatric CD and non-CD subjects from Kosovo.

Patients and Control
Group. Sixty pediatric patients with CD (40 females and 20 males) at the University Clinical Centre of Kosovo (UCCK) and 124 unrelated age-and gender-matched children (64 females and 60 males) without a history of autoimmune diseases were included in this study. The patients' age at time of diagnosis ranged from 17 months to 18 years, with a mean age of 5.5 years (SD ± 3 31). The control group age at the time of sample collection ranged from 1 to 18 years with the mean age of 8.7 years (SD ± 5 8). The CD diagnosis was achieved according to the criteria of the European Society of Pediatric Gastroenterology and Nutrition [14] and the revised guidelines of the European Society of Pediatric Gastroenterology, Hepatology, and Nutrition [1]. The study was approved by the Ethics Committee of the UCCK (approved February 12th, 2013) and by the Ethics Committee of the University Hospital Centre (UHC) Zagreb (approved March 28th, 2013) and is in accordance with the Helsinki Declaration. Written informed consent was obtained from parents or caregivers of all children prior to their enrollment in the study.
2.2. HLA Typing. Three milliliters of EDTA blood samples were collected from each child included in the study. Blood samples were stored frozen at -30°C until all samples were collected. DNA isolation and HLA typing were performed in the Tissue Typing Centre, Department of Transfusion Medicine and Transplantation Biology, UHC Zagreb. Genomic DNA was isolated from whole blood with the MagNA Pure Compact Instrument using the corresponding commercially available MagNA Pure Compact Nucleic Acid Isolation Kit I (Roche Diagnostics GmbH, Mannheim, Germany). The final DNA concentration was adjusted to 50 ng/μl. All CD patients and controls were typed for HLA-A, -B, -DRB1, -DQA1, and -DQB1 applying the standard polymerase chain reaction-sequence-specific oligonucleotide probing (PCR-SSOP) method using the commercially available Immucor LIFECODES HLA-SSO typing kit (Immucor Transplant Diagnostics Inc., Stamford, USA), based on the hybridization of a labeled single-stranded PCR product to SSO probes [15]. After hybridization, the samples were processed and quantified on the Luminex 200 flow analyzer (Luminex Corporation, Austin, USA) and the resulting data were analyzed with Match It! DNA Software, version 1.2.4. The HLA-DQA1 and -DQB1 alleles determined at medium resolution level were assigned to the most common corresponding fourdigit HLA allele which was additionally confirmed when necessary by the standard polymerase chain reactionsequence-specific primer (PCR-SSP) high-resolution protocol (CareDx, Olerup SSP AB, Sweden).

Results
3.1. HLA Allele Frequencies. The observed HLA allele frequencies of HLA-A, -B, -DRB1, -DQA1, and -DQB1 loci with a statistically significant difference between the Albanian pediatric CD patients and the control group from Kosovo are shown in Table 1 and Table 2. A total of 15 HLA-A alleles were found in the CD patient group, among which the most frequent were HLA-A * 01 (26.7%), -A * 02 (21.7%), and -A * 03 (10.0%). A statistically significant difference was observed for HLA-A * 01 (P < 0 0001), present with a higher frequency among CD patients compared to the control group, while the HLA-A * 02 allele had a lower frequency among CD patients (P = 0 0312).

Frequencies of CD-Predisposing HLA-DQ Haplotypes and
Genotypes. The presence of the HLA-DQ2 and/or HLA-DQ8 heterodimers was detected in 57/60 (95.0%) CD patients and in 50/124 (40.3%) control subjects. The detailed CDpredisposing HLA-DQ haplotype and genotype distribution among CD patients and a comparison with the frequency of those haplotypes/genotypes in the control group is presented in Table 4. The most frequent CD-predisposing HLA-DQ haplotypes found in patients were HLA-DQ2.5 (cis and trans conformation) and HLA-DQ2.2 with a frequency of 46.7% and 11.6%, respectively. Consequently, the most prevalent genotype among CD patients was HLA-DQ2.5/DQX with a frequency of 58.3%, followed by the HLA-DQ2.5/DQ2.2 genotype with a frequency of 20.0%. When compared with the control group, both genotypes show a statistically significantly higher frequency among CD patients (P < 0 0001 and P = 0 0005, respectively). Three patients were homozygous for HLA-DQ2.5 (having a genotype positive for two copies of the HLA-DQ2.5 heterodimer), with no presence of this genotype in controls. The frequency of the HLA-DQ8 haplotype among CD patients is low (4.2%), and there is no significant difference compared to the control group (

Discussion
A strong association between the CD and HLA-DQ allelic groups encoding for HLA-DQ2 and HLA-DQ8 heterodimers is well-described, and these specific HLA heterodimers are usually seen in more than 90.0% of the patients with CD [17]. However, due to the huge polymorphism of the HLA system and the different HLA risk levels of CD among populations [18], it is important to perform the analysis in each specific population separately. This is the first analysis of the HLA-A, -B, -DRB1, -DQA1, and -DQB1 allele frequencies and the HLA-DQ2 and HLA-DQ8 heterodimer frequencies in 60 Albanian children from Kosovo diagnosed with CD. The limitation of the study is the small sample size, although the post hoc sample size calculation showed that the number of 60 CD patients was enough for the obtained results with a confidence interval of 85%. The analysis of the HLA-A, -B, -DRB1, and -DQ frequencies among CD patients and compared to the control group revealed a pronounced increase of the HLA-A * 01, -B * 08, -B * 50, -DRB1 * 03, -DRB1 * 07, -DQA1 * 02:01, -DQA1 * 05:01, -DQB1 * 02:01, and DQB1 * 02:02 alleles. On the other hand, HLA-A * 02, -B * 35, -DRB1 * 11, -DRB1 * 13, -DRB1 * 16, -DQA1 * 01, DQB1 * 05, and -DQB1 * 03:01 were significantly less present among CD patients than in controls. The higher frequency of these HLA alleles among CD patients is also reflected in the haplotype frequency. The top-ranked haplotype observed with a high frequency of 22.5% was the ancestral AH 8.1 haplotype (HLA-A * 01~B * 08~DRB1 * 03~DQA1 * 05:01~DQB1 * 02:01), which was by contrast present in 2.8% of controls. These results are in concordance with  results showing that DQ2 is an absolute requirement for the development of CD, but the presence of the wellknown ancestral haplotype AH 8.1 and additional genetic factors in the HLA class I region induce an increased risk of CD [19][20][21][22]. What is also interesting are the two HLA alleles with a surprisingly high frequency in the CD patient group: HLA-B * 50 and HLA-DRB1 * 07, with frequencies that were more than four times higher and three times higher in comparison to controls, respectively. Consequently, the second most frequent five-locus haplotype observed in our CD patients was HLA-A * 02~B * 50~DRB1 * 07~DQA1 * 02:01~DQB1 * 02:02. This haplotype has not been previously reported as a haplotype associated with CD, and it is mostly associated with a probable Euro-Asiatic origin, having been reported in Mongolians (HF: 3.2%), in the Chaouya population from Morocco (HF: 2.9%), Turks and Kurds (HF: 1.3%), Spaniards (HF: 1.2%), and Italians (HF: 0.5%) [23][24][25]. The second uncommon and unexpectedly frequent haplotype, ranked fourth among our CD patients, was HLA-A * 68~B * 44~DRB1 * 07~DQA1 * 02:01~DQB1 * 02:02. This haplotype was not observed in the control group, and this five-locus haplotype has not been reported in any European population so far, but only in the Sri Lanka Colombo population with a frequency of 7.0% [26].
These results raise new questions: what is the distribution of this five-locus HLA haplotype in other neighboring populations as well as in other populations of European origin? Furthermore, is the presence of this haplotype specific just for Albanian CD patients from Kosovo, or is this haplotype present among CD patients from other countries?.
The results of this research revealed that 95.0% of the Albanian pediatric CD patients from Kosovo were HLA-DQ2-and/or HLA-DQ8-positive compared to the 40.3% of positive individuals in the control group. The given results are similar to those reported from other studies, and at the Table 4: The frequency of celiac disease-predisposing HLA-DQ haplotypes and genotypes detected among Albanian pediatric celiac disease patients from Kosovo (N = 60) and the control group (N = 124).