Compliance with ethical standards
The study was approved by the Bioethical Committee of the Jagiellonian University in Krakow (No. KBET/68/B/2011). All patients were informed about the reasons of the investigation and gave their written informed consent prior to participation in the study.
All participants of the study were of Polish origin. Hashimoto’s thyroiditis was diagnosed by documented primary hypothyroidism requiring thyroid hormone replacement, and the presence of autoantibodies to thyroid peroxidase and/or anti-Tg antibodies, with or without goiter.
Inclusion criteria for the HT group: TSH above 4.5 µIU/ml, the presence of anti-TG and/or anti-TPO antibodies, typical for HT ultrasound of the thyroid.
Inclusion criteria for controls: the absence of any inclusion criterion for HT, negative history for thyroid diseases among participants and their first degree relatives.
Exclusion criteria: other thyroid disorders (thyroid nodular and non-nodular goiter, Graves’ disease, thyroid inflammation), increased anti-TPO antibodies with normal TSH levels (only patients with elevated TSH values were qualified for the study to obtain a homogeneous group for analysis with full disease characteristics), increased antibodies against TSH receptor, lack of consent to participate in the study.
None of the participants were related to each other. 147 HT cases (10.2% men) and 147 controls (13.6% men) were qualified for the analysis. These groups were matched for gender (p = 0.4717), age (p = 0.9923), marital status (p = 0.8148), education (p = 0.1157), monthly income (p = 0.7793), and size of the city they live in (p = 0.8535).
Demographic data and environmental factors
Non-genetic factors which might play a role in thyroid autoimmunity were assessed by an environmental survey. Patients answered questions regarding demographic data including gender, age, marital status, completed education level, their monthly gross income per person and size of city they live in (as a simple measure of air pollution). Questions about environmental factors included tobacco smoking, alcohol consumption, stress, and a dietary questionnaire. Based on the dietary questionnaire, the individual monthly iodine, selenium and fiber intake was calculated.
DNA isolation, quantification and quality control
DNA was isolated from whole blood samples collected on EDTA, with the NucleoSpin Blood kit (Macherey–Nagel), according to the Manufacturer’s protocol. The amount and quality of DNA were assessed on NanoDrop 2000 (Thermo Fisher Scientific).
Our study included genomic regions most commonly mentioned in connection with AITD in the earlier literature, independently on the effect size obtained by the authors. The regions under investigation included the genes PTPN22, CTLA4, HLA-DRB1, IL10, IL6, TPO, IFIH1, TNF, TG, ZFAT, RET, ARID5B, CD40, FOXP3, promoter regions, as well as regions typed out for AITD by GWAS on chromosomes 12 (gene DCN), 13 (ARGLU1, FAM155A), 14 (TTC9, MAP3K9, RGS6, PSEN1), 20 (MAFB, PLCG1, CHD6), and X (TNMD, SYTL4, NOX1) [12, 20]. The gene IFIH1 was typed out by our team as an additional candidate gene based on the literature, although not commonly investigated so far in AITD. Analyzed polymorphisms are listed in supplementary Table 1. Polymorphisms (SNP) in the above-mentioned regions were investigated on Illumina HiScan using Illumina’s Golden Gate custom panel, according to the Manufacturer’s protocol.
Variants in the HLA-DRB1 gene were identified by Sanger sequencing of exon 2 of the gene, which includes most of the gene’s heterogeneity. PCR was performed with FirePol polymerase (Solis BioDyne), according to the Manufacturer’s recommendations, on a BioRad T100 thermocycler. 120 ng DNA were used in each 25 µl reaction. Group-specific primers and PCR conditions were used according to van Dijk et al. . Products were visualized electrophoretically, and the remaining reaction mixture purified with NucleoSpin 96 PCR Clean-up (Macherey–Nagel). The products were eluted with 70 µl nuclease-free water (Ambion), dried for 2 h at 75 °C, and pellets were resuspended in 11 µl nuclease-free water. The 10 µl sequencing PCR mixture was performed with BigDye Terminator v3.1 (ThermoFisher Applied Biosystems) and 5× sequencing buffer, 0.7 µl of the appropriate 10 µM sequencing primer (according to ), and 5.3 µl of the purified PCR product. The sequencing conditions were in agreement with the Manufacturer’s recommendations, with an annealing temperature of 55 °C. Products were purified by ethanol precipitation, and pellets resuspended in 20 µl nuclease-free water for capillary electrophoresis (ABI 3500, Applied Biosystems).
Statistical analyses were performed in Statistica v12.
Demographic data and environmental factors
For categorical data, the Chi-squared test was applied to compare groups. For quantitative and ordinal data, the Mann–Whitney test with correction for incontinuity was used, with an additional control for tied ranks in case of ordinal scale of the data. Results were considered significant at α = 0.05.
Sequencing results were visualized with DNA Baser v4 and aligned to the NCBI dbMHC database with use of the SBT interface. Association analysis of serotypes and alleles was performed with Statistica v12. Alleles were re-coded into amino acid sequences to visualize non-synonymous regions.
The CHI^2 test was used for association analysis of alleles. Associations were considered significant at α = 0.05. Logistic regression was used to reveal a potential pattern of the binding cleft, as has been suggested previously .
Quality control was performed with the Genotyping module of BeadStudio, and based on the GenCall validation. Initial laboratory quality assurance (QA) relied on the GenCall score, a quality metric indicating the reliability of called genotypes that is generated by the BeadStudio software. For initial QA, a sample and SNP GenCall_p10 score threshold of 0.38 and call rate threshold of 85% were set. All analyzed polymorphisms met those criteria. Data were further analyzed with the tool Plink v1.9 . All SNPs and all patients had less than 10% missingness. All polymorphisms under investigation were confirmed to meet the Hardy–Weinberg equilibrium in the control group (α = 0.05). Association of each SNP with HT was determined by Cochran–Armitage trend test (Results are shown in supplementary Table 1). The most probable mode of inheritance for each SNP was examined using the permutation procedure implemented in Plink. All polymorphisms were included in the further stepwise regression analysis, independently of whether they appeared to be significantly associated with HT in the standard association analysis.
Stepwise logistic regression was performed to obtain a model of genetic regions significantly implicated in the predisposition to HT in our sample group. All polymorphisms in study were included in the analysis, considering their mode of inheritance. Also included was phenylalanine at position 67 of the HLA-DRB1 gene, and the intrinsic factors sex and age. Logistic regression has been chosen because genetic as well as non-genetic factors were considered simultaneously. This method also allows to automatically exclude redundant variables, which is a problem in case of linkage disequilibrium. The goodness of fit for the model was assessed by tenfold cross-validation. The explained predisposition to HT based on the obtained model was assessed with the program GCTA , setting the frequency of HT at 1% .