HLA Association among Thai Patients with Diffuse and Limited Cutaneous Systemic Sclerosis

This study aimed to clarify the association of HLA Class I and II with dcSSc and lcSSc in Thais. HLA typing for 11 gene loci (Class I: HLA-A, B and C, and Class II [HLA-DR, DP and DQ]) was carried out using the Next Generation DNA Sequencing method (three fields) in 92 Thai patients with systemic sclerosis (55 dcSSc, 37 lcSSc) and 135 healthy controls (HCs). The distribution of HLA alleles in patients with dcSSc and lcSSc was compared. When compared with HCs, the AF of A*24:02:01, A*24:07:01, B*27:04:01 and B*27:06 showed an increasing trend in lcSSc patients without statistical significance. DRB1*15:02:01, DRB5*01:02:01, DQA1*01:01:01, DQB1*05:01:24, DPA1*02:01:01 and DPB1*13:01:01 increased significantly in dcSSc patients. DQB1*05:01:24 and DPB1*13:01:01 also increased significantly in lcSSc patients, but less significantly than in dcSSc patients. The association of DPB1*05:01:01 with lcSSc was significantly protective. HLA-A*24:02:01, B*27:06 and C*03:04:01 formed a three-locus haplotype that also constituted an eight-locus haplotype with DRB1*15:02:01, DQA1*01:01:01, DQB1*05:01:24, DPA1*02:01:01 and DPB1*13:01:01. There was a possibility that HLA Class I would play a role in the pathogenesis of lcSSc, while Class II played more of a role in the dcSSc in Thai patients.

Several studies have determined the contribution of HLA Class II molecules to the development of SSc.The charge of amino acid motifs at the 67-74 position on the third hypervariable region of HLA-DRB1 affected their association with SSc [17].The uncharged polar residue tyrosine at 30, and amino acid sequence of 71 TRAELDT 77 on the first domain of the HLA-DQ β chain have been reported in SSc patients, being anti-topoisomerase antibody (ATA) positive [18].Moreover, the amino acid motif 67 FLEDR 71 shared by HLA-DRB1 susceptible alleles was associated more highly with SSc patients who were ATA positive, when compared to the motif of 71 TRAELDT 77 , shared by HLA-DQB1 susceptibility alleles [14].In addition, the DQB1 alleles carrying polar glycine or tyrosine, but not leucine, at position 26 of the DQB1 first domain, also contributed to the pathogenesis of SSc [19].
Associations of HLA among dcSSc and lcSSc patients have never been reported in Thailand.Due to the difference in clinical manifestations between dcSSc and lcSSc patients, this study aimed to compare the three-fields level of HLA risk alleles and haplotypes between these two subsets in Thai patients with SSc.The role of the HLA Class I haplotype in the pathogenesis of Thai patients with SSc was also determined.

Materials and Methods
In this study, the clinical data and HLA typing results of SSc patients and healthy controls (HCs) were from previously published studies that involved HLA in Thai patients with SSc [12].In brief, all SSc patients were followed up at the Rheumatology Clinic of Chiang Mai University Hospital, and the diagnosis of SSc met the 1980 American College of Rheumatology classification criteria [24].Patients overlapping with other connective tissue diseases were excluded.The diagnosis of dcSSc or lcSSc followed the classification of LeRoy et al. [25], and the severity of skin involvement was determined by the modified Rodnan Skin Score [mRSS] [26].The HCs were unrelated to the patients and had no symptoms or signs suggesting SSc or any other connective tissue diseases.Only one healthy control (HC) was allowed from one family.All of the participants were from northern Thailand, and they provided their written informed consent prior to entering the study.
HLA alleles were determined for 11 HLA genes (HLA-A, B, C, DRB1, DRB3, DRB4, DRB5, DQA1, DQB1, DPA1 and DPB1) by the Next Generation DNA Sequencing (NGS) method, as described previously [12,27], and genotyped basically to the three-fields level.The Hardy-Weinberg equilibrium (HWE) model was used to determine the distribution of genotype frequency (GF) in the HC population, which was estimated using the AF and number of subjects carrying each allele.
The HLA haplotypes of the HLA-A, B and C genes and the haplotype frequency (HF) were estimated in SSc patients or HCs by Markov chain-Monte Carlo methods using the PHASE program version 2.1 [28] and at https://stephenslab.uchicago.edu/index.html(assessed on 22 June 2023).The HLA haplotypes and HF among the HLA-DRB1, DQB1 and DPB1 genes, or among alleles from the HLA-A, B, C, DRB1, DQA1, DQB1, DPA1 and DPB1 genes in the overall SSc patient and HC groups, which have been shown in a previous report by the authors [12], are referred to occasionally in the Discussion Section of this study.A family study was not carried out for estimating haplotypes.The linkage disequilibrium (LD) and the relative linkage disequilibrium (RD) were calculated using the HF, which was estimated by the PHASE program and AF of alleles that constituted the haplotype [12].The significance of LD and RD was not determined.
The SPSS statistical program version 16.0 (Chicago, IL, USA) was used for statistical analysis.Continuous variables were expressed as the mean ± standard deviation (SD) and the categorical variables were expressed as frequency or percent.Student's T or Mann-Whitney U tests were used to compare continuous variables.Fisher's exact test was used for comparing categorical variables including allele frequency (AF).P values were corrected (Pc) by Bonferroni's method in order to avoid a type I error from multiple comparisons [12,29].A Pc value of less than 0.05 was considered as a statistically significant difference.The odds ratio (OR) was shown with its 95% confidential interval (95%CI).

Results
In total, 135 HCs and 92 SSc patients participated in this study.The proportion of female gender was higher in the SSc patients.Of the 92 SSc patients, 55 and 37 were dcSSc and lcSSc patients, respectively.Details of the demographic data among the patients with dcSSc and lcSSc are shown in Table 1.There was no statistically significant difference in age, gender or disease duration.However, the dcSSc patients had a lesser prevalence of hypertension and diabetes mellitus (p = 0.022 and p = 0.037, respectively).The clinical manifestation between the two groups also was similar, but the dcSSc patients had a significantly higher mRSS score (p = 5.727 × 10 −10 ), and a higher proportion of those had sclerodactyly (p = 2.896 × 10 −6 ).In addition, they also had a numerically higher prevalence of ATA (p = 0.127) and presence of digital pitting scars (p = 0.108), but a lesser prevalence of arthritis (p = 0.206).The AF of the HLA genes studied in HCs and dcSSc and lcSSc patients is shown in Table 2     The AF values of HLA genes between dcSSc and lcSSc patients were compared and are shown in Table 3. Overall, there was no significant difference in the AF of the HLA Class I and II alleles studied.However, the AF of B*40:06:01 and B*46:01:01 was slightly higher without significance, and the AF of DPB1*02:02:01 slightly lower without significance when comparing dcSSc patients with lcSSc ones.The potential HLA Class I and II risk alleles and their associations with clinical features in dcSSc and lcSSc patients are shown in Table 4. Due to the mRSS score and the presence of sclerodactyly being significantly different between the dcSSc and lcSSc patients, only these two clinical parameters were included in the analysis.In using the mRSS score of ≥10, the AF of HLA-A*24:02:01 showed an increasing trend in lcSSc when compared to dcSSc patients (p = 0.0254, Pc = not significant (NS), OR [95%CI] 0.17 [0.04-0.72]).The AF of DPA1*02:01:01 also showed an increasing trend in lcSSc with an mRSS score of ≥10 compared to those with a score <10 (p = 0.021, Pc = NS, OR [95%CI] 5.61 [1.40-22.56]).The AF of this allele also showed a higher trend in dcSSc than that in lcSSc patients, who had an mRSS score of <10 (p = 0.024, Pc = NS, OR [95%CI] 3.57 [1.21-10.57]).The AF of A*24:02:01 in dcSSc patients with sclerodactyly was lower than in those without, and it was also lower than in lcSSc patients with sclerodactyly, but without significance (p = 0.061, Pc = NS, OR [95%CI] 0.29 [0.08-1.04]for both conditions).
The  5.Among the top 50 haplotypes in SSc patients, 20 contained at least one of these alleles.Eleven of these alleles were also contained in the top 50 haplotypes in HCs.Among the alleles showing a higher AF in SSc patients, haplotype analysis showed that HLA-A*24:02:01, B*27:06 and C*03:04:01 constituted a haplotype (ABCHP-1), with the fourth most frequent HF in the top 50 haplotypes of HLA-A, B and C (HF = 0.0377).In HCs, the HF of ABCHP-1 that was ≤0.0039 was not included in the top 50 haplotypes.In the top 50 HLA-A,
In this study, the AF of A*24:02:01, A*24:07:01, B*27:04:01 and B*27:06 showed an increasing trend and the AF of C*03:04:01 was slightly higher in lcSSc patients, when compared with the HCs.On the other hand, a slight increase in the AF of these alleles also was seen in the dcSSc patients, but lower than in the lcSSc patients (Table 2).These results suggest that these alleles contribute more to lcSSc than to dcSSc.Unfortunately, a significant increase in B*08:01 or significant decrease in B*44:03, C*07:02 and C*16:01 among the Europeans, Australians and the Mexican Mestizo [6,21,22] was not observed in either dcSSc or lcSSc in the current study.The authors' previous report found that SSc patients overall tended to have non-significantly higher AF of HLA-A*24  [1.06-15.46])than the HCs [12].The AF of A*24:07:01 was slightly higher, but not significantly, in SSc patients overall than in the HCs.
The above observations suggest that the increasing trend or increase in frequency of the A*24:02:01, B*27:06 or C*03:04:01 alleles observed in lcSSc patients was possibly due to the significant increase in the DRB1*15:02:01, DQB1*05:01:24 or DPB1*13:01:01 alleles.However, the increasing trend in A*24:02:01, B*27:06 and C*03:04:01 was observed mainly in lcSSc patients, despite the small sample size.Although A*24:02:01, B*27:06 and C*03:04:01 did not increase significantly in SSc patients in this study, they possibly played an important role in the pathogenesis of Thai SSc, especially in lcSSc.According to these observations, the possibility of A*24:02:01, B*27:06 and C*03:04:01 contributing genetically to lcSSc, but not dcSSc patients, could not be denied.In order to clarify the contribution of HLA-A, B or C alleles to SSc, further investigation using a large sample size would be necessary.
In addition, this study suggested that certain HLA alleles might be associated with clinical manifestations among SSc subtypes (Table 4).In those with an mRSS score ≥10, the AF of A*24:02:01 showed a higher trend in lcSSc than in dcSSc.These observations suggested the possibility of A*24:02:01 contributing not only to the pathogenesis of lc-SSc (Table 2) but also to a higher mRSS in lcSSc (Table 4).The AF of this allele was lower in dcSSc with sclerodactyly than in dcSSc without, and also lower than in lcSSc with sclerodactyly.These observations suggested the possibility of this allele contributing protectively to sclerodactyly in dcSSc.Similarly, the AF of DPA1*02:01:01 showed a trend to be higher in lcSSc patients with an mRSS score ≥10 than in those with an mRSS score <10.Considering the significant association of this allele with dcSSc, in comparison with the AF of HCs (Table 2), these findings suggested the possibility that DPA1*02:01:01 was not only contributing to the pathogenesis of dcSSc, but also to a higher mRSS in lcSSc (Table 4).Unfortunately, the number of patients in these subgroups was too small to draw conclusions.
There are some limitations in this study.It was performed from only one study center, with a rather small sample size and no family study.Further studies with a large number of samples are needed to clarify these findings.
Lastly, the non-significant changes in the HWE indicated that the distribution of genotype frequency (GF) in the HCs of this study remained unchanged.In addition, the distribution of AF in most alleles of the HCs in this study and the Thai population from the Allele Frequency Net Database [31] and that from northern Thailand [32] were the same or nearly the same, although there were some alleles that showed a significant difference without correction.Considering the relatively small number of samples, the limited area (near Chiang Mai) where the sampling was carried out in the HCs, the NGS method used for typing and the AFs that were reported in three fields compared to the other two databases which reported in two fields [31,32], the HC data in this study were compatible with the other two databases.

Table 1 .
Clinical characteristics of Thai patients studied with SSc and their subtypes.

Table 2 .
Comparison of HLA Class I and II alleles between HCs and dcSSc and lcSSc patients.

Table 3 .
Comparison of HLA Class I and II alleles between dcSSc and lcSSc patients.

Table 4 .
Comparison of the AF of HLA risk alleles with clinical data in dcSSc and lcSSc patients.

Table 5 .
HLA-A, B and C haplotypes with alleles showing an increasing trend in dcSSc or lcSSc patients.