Association Between AIRE Polymorphisms rs870881(C>T), rs1003854(T>C) and Rheumatoid Arthritis Risk: A Hungarian Case-control Study

Abstract

Background/Aim

Autoimmune regulator (AIRE) is a transcription factor that plays pivotal role in controlling autoimmunity. In the thymus, it supports the presentation of peripheral tissue antigens to developing T cells, where recognition of these self-antigens negatively selects the autoimmune naïve T-cells by central tolerance. Studies demonstrated that single-nucleotide polymorphisms (SNPs) in AIRE alter transcription and propagate clonal survival of autoimmune T cells, therefore increase susceptibility to autoimmune diseases. This study intended to identify SNPs in exon and intron sequences that determine AIRE transcription, where their genotypes are associated with rheumatoid arthritis (RA) risk and clinical parameters.

Patients and Methods

After a thorough in silico research, we enrolled 100 patients with RA and 100 healthy controls to analyze the association of SNP rs870881(C>T) and rs1003854(T>C) in AIRE coding sequence with RA risk by using five different genetic models and selected clinical parameters. Multiplex quantitative polymerase chain reaction was used to determine allelic discrimination of SNPs. RA risk was assessed by odds ratios (ORs) and confidence intervals (CIs).

Results

In a recessive model of rs878081, minor allele TT homozygotes were associated with RA (p=0.032, OR=5.44, 95%CI=1.16-25.52); in a recessive model of rs1003854, minor allele CC homozygotes were associated with RA (p=0.047, OR=4.84, 95%CI=1.02-23.02). Higher C-reactive protein (CRP) levels in patients with RA were significantly associated with minor allele homozygotes in recessive and codominant genetic models (p=0.029 and p=0.043, respectively) of rs1003854.

Conclusion

Genotypes for minor alleles of rs878081 and rs1003854 might be involved in RA pathogenesis and risk prediction.

Rheumatoid arthritis (RA) is a common disease of immune dysfunction characterized by chronic synovitis; the resultant polyarticular arthritis with extra-articular symptoms causes functional impairment. The disease prevalence is 0.5-1% among adults, and the annual incidence in developed countries is 5-50 per 100,000 adults (1).

The immunopathology of RA has not been completely clarified; however, many studies have reported that some autoimmune T cells clonally expand and survive negative selection of the adaptive immune system. These autoimmune T cells then infiltrate the synovium and cause RA (2,3).

Despite the significant association between major histocompatibility complex (MHC) HLA-DRB1 and RA by numerous genome-wide association studies (4-7), the two latest transethnic, high-throughput genome-wide association studies have associated 51 novel risk loci with RA using a combined total of 33,954 cases and 90,649 controls (8,9). One of these risk loci is the coding sequence for AIRE, which is indispensable for the control of autoimmunity.

AIRE is located in the 21q22.3 chromosomal region and encodes the 545 amino acid transcription factor autoimmune regulator (AIRE), which is located in the nucleus of medullary thymic epithelial cells (mTECs) (10). During postnatal development of the immune system, AIRE binds to the DNA of mTECs to ensure the promiscuous, ectopic expression of peripheral tissue self-antigens in the thymic environment, which is the “educational” site of the incoming naïve T cell precursors, thymocytes (2). Self-antigens are presented on MHC molecules of mTECs to the T cell receptors on thymocytes (2). If self-antigens are recognized by the T-cell receptor (TCR) of thymocytes, these naïve T-cells become autoreactive and negative selection clonally deletes them by apoptosis to ensure a normal T cell population without autoreactivity (2,10). This negative selection is the last step in the process of immunological central tolerance (2). AIRE consists of several structural domains: the caspase-recruitment domain controls interactions with binding partners; the SAND domain has DNA-binding and transcription properties and controls peripheral tissue antigen production; the PHD fingers read the histone code; and the LXXLL sequences interact with the ubiquitous transcription activators CREB-binding protein (CBP) and signal transducers and activators of transcription (STAT) protein (10-13). Certain functional mutations in the AIRE protein-coding sequence result in the loss of negative selection, causing development of the systemic autoimmune disease autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED), also known as autoimmune polyendocrine syndrome type 1 (APS-1), which causes multiorgan devastation (10,14).

Quantitatively inappropriate presentation of self-antigens may decrease the effectiveness of central and peripheral tolerance, causing the survival and maturation of autoreactive thymocytes which leads to various autoimmune diseases, including RA. This effect has been confirmed for the glucose-6-phosphate isomerase self-antigen in the K/BxN mouse model of RA (15). Numerous studies describe the fundamental contribution to RA development of defective selection of autoimmune T cells and the resultant synovial infiltration of CD4⁺ type 1 T helper (Th) 1, Th2, Th17, and regulatory T cells (3,16-18). We endorse the rising scientific attention on thymic T cell selection and the necessity to analyze the role of gene variants of its key player, AIRE. The role of allelic polymorphisms in AIRE transcription was first described in an in vivo study that detected two distinct AIRE single-nucleotide polymorphisms (SNPs), AIRE-230Y and AIRE-655G. The haplotype AIRE-230T transcriptionally modified AIRE gene expression, which elevated the risk of autoimmunity through the alteration of negative selection (19). Although AIRE SNPs attract scientific attention, only few case-control studies have analyzed the association between allelic polymorphisms of AIRE and autoimmune diseases, such as vitiligo (20), alopecia areata (21), Addison’s disease (22), and systemic sclerosis (23). The majority of these molecular epidemiological studies analyzed the association of AIRE SNPs with RA. Similarly, we previously conducted a meta-analysis to verify the association of AIRE allelic polymorphisms with RA in Asian populations (24). So far, the association in European populations is poorly understood. To date, one study has analyzed AIRE and RA in the Spanish population (25). According to our knowledge, our report is the first European study to examine the association of genotypes of AIRE allelic polymorphisms and RA risk using five different genetic models (allelic, dominant, recessive, codominant heterozygous, codominant homozygous, and overdominant) to determine risk genotypes.

Patients and Methods

Preliminary in silico research. We performed preliminary in silico research to identify SNPs that could be relevant to RA risk. We used the Ensembl automatic annotation of the human genome sequence (GRCh38.p13) and the UCSC Genome Browser (GRCh38 assembly) as a human genome database (26,27). Our inclusion criteria for SNPs to be tested were global minor allele frequency (MAF) of 0.01-0.5 and presence of the variant in at least one of the following databases: 1000 Genomes Project Phase III (including the HapMap project), NCBI dbSNP, NHLBI GO Exome Sequencing Project (human only), Exome Aggregation Consortium (human only), Genome Aggregation Database (human only), and NHLBI Trans-Omics for Precision Medicine (human only) (28-32).

Hits were excluded as they had a low probability of detection in our study population. Therefore, we established a criterion to exclude a genetic variant if it existed in the 1000 Genomes Project database with a global MAF of 0.01-0.5 but the MAF was less than 0.01 in the EUR or CEU subpopulations.

Study subjects. Patients with RA and healthy control subjects were consecutively recruited by voluntary participation from individuals receiving outpatient or inpatient care at the Harkány Thermal Rehabilitation Centre, Hungary. We collected blood samples from the recruited subjects from May 2018 to September 2022; the subjects gave informed consent and received a study information sheet.

Blood samples were obtained and collected in citrate tubes (one 4 ml tube per subject). Whole blood was stored at −75˚C until analysis. Our inclusion criteria for the patients with RA were that the subjects aged more than 18 years, had signed a consent statement, and had been clinically diagnosed with RA, based on the American College of Rheumatology classification (33). Our inclusion criteria for the healthy control subjects were that they aged more than 18 years, had signed a consent statement, and did not have RA, based on the American College of Rheumatology classification and had blood values of C-reactive protein (CRP) ≤20 mg/dl and rheumatoid factor (Rf) ≤25 IU/ml (34). Ultimately, we analyzed data from 100 patients with RA and 100 healthy control subjects.

Laboratory assessment. The blood samples were analyzed for complete blood count, cholesterol profile, blood sugar profile, erythrocyte sedimentation rate (ESR), CRP, sodium, potassium, urea, creatinine, glutamate-oxaloacetate transaminase, glutamate-pyruvate transaminase, gamma-glutamyl transferase, alkaline phosphatase, lactate dehydrogenase, creatine kinase, protein albumin, magnesium, phosphorus, and serum bilirubin. We measured rheumatoid factor (Rf), anti-cyclic citrullinated peptide (aCCP), and antinuclear antibodies to confirm seropositivity in patients with RA and seronegativity in control subjects (35,36). Finally, we calculated the functional status of patients with RA by the Functional Independence Measure and disease activity score with 28-joint counts (DAS28) (37). ESR was quantified using the Westergren method (38). CRP and Rf were quantified at the Harkány Thermal Rehabilitation Centre using Konelab Arena 20XT Clinical Chemistry Analyzer (Thermo Fisher Scientific, Waltham, MA, USA) in accordance with the manufacturer’s recommendations. Levels of aCCP were analyzed using an Immunoscan CCPlus enzyme-linked immunosorbent assay kit (Svar Life Sciences, Malmö, Sweden) at the Department of Immunology and Biotechnology, Medical School, University of Pécs, Hungary, in accordance with the manufacturer’s protocols.

Genomic DNA extraction from blood. Genomic DNA was isolated using 1 ml DNAzol™ BD reagent (Thermo Fisher Scientific) mixed with 0.5 ml whole blood in accordance with the manufacturer’s instructions. Genomic DNA was precipitated with 0.4 ml isopropanol for 5 min at room temperature and then centrifuged. After the removal of the supernatant, the pellet was washed with 0.5 ml DNAzol™ BD reagent and then washed with 1 ml 75% ethanol. The genomic DNA was dried, dissolved in 200 μL Tris EDTA buffer (10mM Tris, pH 8.0, and 1mM EDTA), and quantified with a NanoDrop One (Thermo Fisher Scientific). For genotyping by quantitative polymerase chain reaction, 20 ng genomic DNA was used per reaction.

Genotyping AIRE SNPs rs878081 (C>T) and rs1003854 (T>C). We used the predesigned Minor Groove Binder (MGB) TaqMan probe assays (Thermo Fisher Scientific) to detect SNPs in AIRE. The presence of the MGB at the 3’ ends of the probe increases the melting temperature, stabilizes the probe–target hybrids, and enables more accurate discrimination between highly homologous allele sequences. After genomic DNA purification from blood samples, multiplex quantitative polymerase chain reaction was performed. FAM and VIC fluorescence intensity signals were detected using the Rotor-Gene 3000 thermal cycler (Corbett Life Science, Qiagen, Crawley, UK). Samples with signals in either the VIC or FAM channels were considered homozygous for the SNP site of interest, whereas samples with signals in both VIC and FAM channels were considered heterozygous for the SNP site of interest.

Statistical analysis. Our statistical analyses were performed using SPSS software, version 28.0 (IBM Corp. Released 2021, IBM SPSS Statistics for Windows, Version 28.0; IBM Corp, Armonk, NY, USA). Chi-square tests were used to determine the contingency of genotype distributions with Hardy–Weinberg equilibrium. We also used chi-square tests to check significant differences in the genotype frequencies of control subjects. Clinical continuous variables are represented as mean±standard deviation (SD).

The Shapiro-Wilk test was used to determine normality, and Levene’s test was used to analyze the equality of variances. To compare groups for continuous variables, we used independent samples t-tests and analysis of variance. To compare medians of non-normally distributed variables, we used Mann-Whitney U and Kruskal-Wallis H tests with two-tailed significance. Correlation was tested using bivariate correlation with Pearson correlation coefficient and two-tailed significance. Binary logistic regression was used to calculate odds ratios (ORs) with a 95% confidence interval (CI). Statistical significance was determined when p-values were less than 0.05. The power was calculated and indicated if it exceeded 0.8 (39).

Results

Demographic and clinical characteristics of the study subjects. The demographic and clinical data of the study subjects are presented in Table I. The mean±SD age of RA patients was 65.38±9.84 years; 85% were female. According to the literature, the mean age of the disease onset is 65 years (1). To minimize the probability of latent cases among controls, we included participants of age 75 or older. Patients and control individuals were all born in Hungary. The RA group demonstrated significantly higher ESR, CRP, Rf immunoglobulin G (IgG) antibody, aCCP levels (all p<0.001) and had 60% seropositivity compared with 100 control subjects in whom the aforementioned clinical parameters were in the normal range and whose seronegativity was 100% (34,40-43). We also calculated DAS28 for patients with RA, where the mean±SD of DAS28 suggested moderate disease activity (44).

Table I. Demographic and clinical characteristics of study populations.

RA: Rheumatoid arthritis; SD: standard deviation; ESR: erythrocyte sedimentation rate; CRP: C-reactive protein; Rf: rheumatoid factor; IgG: immunoglobulin G; aCCP: anti-cyclic citrullinated peptide; DAS28: disease activity score with 28-joint counts; *p<0.001, compared with control subjects.

Results of preliminary in silico research. Using the Ensembl automatic annotation of the human genome sequence and the UCSC Genome Browser as a human genome database, we set up our parameters in the AIRE (HGNC Symbol;Acc:HGNC:360) gene sequence (also known as APECED, APS1, or PGA1), which is located in the 21q22.3 chromosomal region between 44,285,838-44,298,648 on the forward strand (26,27). Without using any restrictions, we identified 23,023 variations (point mutations, indels, and SNPs). To find detectable SNPs, we adjusted the MAF to 0.01-0.5 with at least one representation in each database. We had 68 hits; however, some SNPs had MAFs less than 0.01 in the European (EUR) subpopulation of Utah residents with Northern and Western European ancestry (CEU). We excluded these SNPs by adjusting the MAF to 0.01-0.5 in EUR, because of the low probability of SNP detection in this population. This allowed us to exclude 24 SNPs, and we subsequently shortlisted 44 SNPs in our in silico setting. Each of the shortlisted SNPs is found in the 1000 Genomes Project Phase III data set with MAFs 0.01-0.5 and is present in EUR and CEU subpopulations.

Among the 14 exon regions of AIRE, exon 5 codes for the SAND domain, which is responsible for transcriptional regulation and DNA binding in mTECs during negative selection. Therefore, we analyzed synonymous polymorphisms in the SAND domain-coding sequence that had MAFs of 0.01-0.5 that might alter messenger RNA (mRNA) splicing, mRNA stability, mRNA structure, protein translation, or cotranslational protein folding (10,11,26,45). In the coding sequence of exon 5, we found one synonymous polymorphism, rs878081, and analyzed its strongly linked intron variant rs1003854, for which the pairwise r² value for linkage disequilibrium was 0.912. Domain-coding structural sequences of AIRE and the positions of rs878081 and rs1003854 are illustrated in Figure 1.

Figure 1. Domain-coding sequences of AIRE, and the positions of rs878081 and rs1003854 (image generated in UCSC Genome Browser).

Genotype frequencies and Hardy–Weinberg equilibrium. Genotype frequencies in control groups for rs878081 and rs1003854 were in accordance with Hardy–Weinberg equilibrium. No statistically significant difference was found between the expected and observed genotype frequencies in rs878081 (p=0.316) and rs1003854 (p=0.214); therefore, our sample collection avoided selection bias, nonrandom sampling, and accumulation of relatives.

Association between rs878081 (C>T) allelic polymorphism and the risk of rheumatoid arthritis. Allelic polymorphism in rs878081 was analyzed under allelic, dominant, recessive, codominant heterozygous, codominant homozygous, and overdominant genetic models (46). Under the recessive genetic model, we found statistically significant association with RA, where the genotype frequency of TT homozygotes was significantly higher in patients with RA than in healthy control subjects, supporting RA risk (OR=5.44, 95%CI=1.16-25.52, p=0.032, power >0.8). The genotype frequency of TT homozygotes was also higher among patients with RA in a codominant homozygous setting than among control subjects; however, the association was barely insignificant (OR=4.68, 95% CI=0.98-22.27, p=0.052, power >0.8) but suggested the same direction as the recessive model. No other significant associations were found in the allelic, dominant, codominant heterozygous, or overdominant genetic models. Allele and genotype frequencies of rs878081 in patients with RA and control subjects and risk calculations are presented in Table II. The results of allelic discrimination tests for rs878081 in patients with RA and control subjects are illustrated in Figure 2.

Table II. Distribution of allele and genotype frequencies of AIRE rs878081 in patients with rheumatoid arthritis and control subjects.

SNP: Single-nucleotide polymorphism; OR: odds ratio; CI: confidence interval. abased on logistic regression; *p<0.05, ^†power >0.8.

Figure 2. Result of allelic discrimination tests for rs878081 in patients with rheumatoid arthritis and control subjects.

Association between rs1003854 (T>C) allelic polymorphism and the risk of rheumatoid arthritis. Allelic polymorphism of rs1003854 was analyzed under allelic, dominant, recessive, codominant heterozygous, codominant homozygous, and overdominant genetic models. The genotype frequency in patients with RA was statistically significantly higher in TT homozygotes than in control subjects, supporting RA risk (OR=4.84, 95%CI=1.02-23.02, p=0.047, power >0.8). No other statistically significant association was found under the allelic, dominant, codominant heterozygous, codominant homozygous, or overdominant genetic models. Allele and genotype frequencies of rs1003854 in patients with RA and control subjects and risk calculations are presented in Table III. The results of allelic discrimination tests for rs1003854 in patients with RA and control subjects are illustrated in Figure 3.

Table III. Distribution of allele and genotype frequencies of AIRE rs1003854 in patients with rheumatoid arthritis and control subjects.

SNP: Single-nucleotide polymorphism; OR: odds ratio; CI: confidence interval; ^abased on logistic regression; *p<0.05, ^†power >0.8.

Figure 3. Result of allelic discrimination tests for rs1003854 in patients with rheumatoid arthritis and control subjects.

Comparison of clinical parameters under the genetic models of rs878081 and rs1003854 in patients with rheumatoid arthritis and control subjects. Our study aimed to compare ESR, CRP, Rf, aCCP, and DAS28 levels between the genotype subgroups of genetic models of rs878081 and rs1003854 in patients with RA and control subjects. The first stage was designed to find significant differences by comparing group means, and the second stage was designed to reveal correlations between clinical parameters and genotypes.

In patients with RA, ESRs were significantly higher in the CC subgroup compared to TT homozygotes under the codominant homozygous model of rs1003854 (p=0.048). The CC subgroup demonstrated significantly higher CRP levels compared to TT homozygotes in the codominant homozygous model and the TT+TC subgroup in the recessive model (p=0.043 and p=0.029, respectively). Furthermore, the CC homozygote patients with RA were significantly correlated with CRP under the codominant homozygous and recessive models (p=0.006 and p=0.002, respectively; r=0.350 and r=0.338, respectively). No other significant differences were found in other clinical variables and genetic models of rs878081 and rs1003854. These results are illustrated in Table IV.

Table IV. C-reactive protein levels among the genotypic subgroups of genetic models of rs1003854 within patients with rheumatoid arthritis and control subjects.

SNP: Single-nucleotide polymorphism; RA: rheumatoid arthritis; CRP: C-reactive protein; ^agenotype subgroup comparison of CRP levels in patients with RA in each genetic model; ^bPearson correlation of CRP with genotype subgroups of patients with RA in each genetic model; ^cgenotype subgroup comparison of CRP levels in control subjects in each genetic model; ^dPearson correlation of CRP with genotype subgroups of control subjects in each genetic model; *p<0.05, **p<0.01; Clinical parameters are reported as mean±SD.

Discussion

RA is a complex, multifactorial disorder in which genetic and environmental causes contribute to disease initiation (1-3). In vitro and in vivo studies have demonstrated that presentation of peripheral tissue self-antigens is quantitatively determined by AIRE transcription, and AIRE SNPs may cause clonal survival (19,47). Autoreactive CD4⁺ T cells are recruited to the synovium where they initiate an inflammatory environment by the production of interferon-gamma, tumor necrosis factor-alpha, interleukins (ILs), proinflammatory cytokines, and chemokines, causing tissue destruction (48-50). The inflammatory environment activates fibroblast-like synoviocytes (FLSs) to produce chemokines in order to attract more autoreactive CD4+ T cells, creating a vicious circle (50,51). This was demonstrated by an in vitro study by Bergström et al. in which IL-1β and tumor necrosis factor-stimulated FLSs from patients with RA and in which there was 26-fold statistically significantly higher CXCL10 and CCL8 production compared with cells with AIRE expression silenced by sense siRNA (51). Nonstimulated FLSs did not produce chemokines (51). Based on our current understanding, AIRE SNPs are important in two ways: they allow the escape of autoreactive CD4⁺ T cells during central tolerance, and they cause FLSs to produce chemokines to attract increasingly more self-reactive CD4⁺ cells, thereby maintaining the hyperplastic inflammatory environment (19,51). Hence, synonymous SNPs and their strongly related intron variants can be associated with disease activity. In Asian populations, a strong association was found between rs2075776, rs760426, and RA in different genetic models (24,52-55). Furthermore, Terao et al. demonstrated genotype-specific AIRE expression in lymphoblastoid cells using the Gene Expression Omnibus database and raised the idea that different genotypes may alter disease activity (56). Allelic polymorphisms of rs878081 have been analyzed in Chinese, Egyptian, and Spanish populations (25,56,57). Yang et al. measured AIRE expression in blood lymphocytes and found significant differences between rs878081 TT and CC homozygotes in Asian populations (56).

Our investigation is the first attempt to analyze different genotypes and RA risk among Hungarian-born, European individuals in allelic, dominant, recessive, codominant heterozygous, codominant homozygous, and overdominant genetic models combined with clinical parameters, such as ESR and CRP. After our preliminary in silico selection of SNPs, our results showed in a recessive model (TT vs. CC+CT) that rs878081 TT is significantly associated with RA risk (p=0.032). The codominant homozygous model (TT vs. CC) provided the same result; however, only borderline significance was detected (p=0.052). These results made us consider that two copies of the minor alleles may be necessary for a quantitative association with RA risk, as the SNP is synonymous and resides in exon 5. Exon 5 is in the SAND coding sequence of AIRE that is responsible for DNA-binding and cooperation with transcriptional factors, so allelic polymorphisms in this region may functionally determine AIRE production and affect central tolerance (10-13).

Our study analyzed intron variant rs1003854, where significant correlations were found between clinical parameters and genotypes in recessive (CC vs. TT+TC) and homozygous codominant (CC vs. TT) genetic models. The allelic polymorphism rs1003854 resides in intron 7 and may modify the transcription of AIRE by altering alternative-splicing mechanisms with the process of intron-mediated enhancement (58). Interestingly, after we determined RA risk for SNP rs1003854, we concluded that only the recessive and codominant genetic models demonstrated significantly higher levels of CRP (p=0.029 and p=0.043, respectively) and positive, significant correlation with CRP (p=0.002 and p=0.006, respectively) when patients with RA were homozygous for the minor C allele. This finding supports the notion that CRP levels vary in genetic models of AIRE rs1003854 among patients with RA, reflecting a possible association between AIRE genotypes and RA disease activity, where chemokines (such as CXCL10, CCL8, and CCL5) are essential and are strongly correlated with CRP levels (59-61).

To conclude, rs878081(C>T) and rs1003854(T>C) SNPs together with ESR and CRP levels would be useful as potential biomarkers to determine RA susceptibility, follow-up, and prognosis. Before future clinical application, further studies are necessary. To continue contributing to this research area, we intend to carry out studies on larger cohorts of patients with RA and control subjects and to consider multiple loci as well as possible molecular partners of AIRE.

Conclusion

In conclusion, our data demonstrate the effects of gene polymorphisms in rs878081 and rs1003854 on the risk of RA in the European population, where two copies of the minor alleles and the related homozygous genotypes provided susceptibility. Analyses of the clinical parameters ESR and CRP supported our assumption that the homozygous subgroup for the minor allele of patients with RA may have higher ESR and CRP levels, which might reflect disease activity. Therefore, these clinical parameters may be suitable to detect at-risk populations in the future. Based on medical history, a previous family history of RA may be an indication for a risk genotype test, where the detection would prospectively allow a follow-up of the healthy individual and possible preventive physiotherapy and drug therapy.

Conflicts of Interest

The Authors declare no conflicts of interest in relation to this study.

Authors’ Contributions

Bálint Bérczi: conceptualization, methodology, formal analysis, investigation, writing – original draft preparation, writing – review & editing, funding acquisition. Nóra Nusser: methodology, resources, writing—original draft preparation, writing—review & editing. Iván Péter: resources, writing—original draft preparation, writing—review & editing. Balázs Németh: resources, writing—original draft preparation, writing—review & editing. Zoltán Gyöngyi: conceptualization, investigation, methodology, data curation, writing—original draft preparation, writing—review & editing, supervision, funding acquisition. All Authors have read and agreed to the published version of the manuscript.

Acknowledgements

The Authors thank all the patients and control subjects for participation in the study. The Authors would like to also acknowledge the conscientious laboratory assistance by the team of the Harkány Thermal Rehabilitation Centre, Hungary. This research was funded by the ÁOK János Szolcsányi Research Fund supported by the Medical School, University of Pécs, Hungary. The study was conducted in accordance with the guidelines of the Declaration of Helsinki, and the protocol was approved by the Secretary of the Medical Research Council, Scientific and Research Ethics Committee of Hungary (11871-7/2018/EÜIG of 24 April 2018, and 57142-5/2022/EÜIG of 8 November 2022). Informed consent was obtained from all subjects involved in the study.