An Intronic HCP5 Variant Is Associated With Age of Onset and Susceptibility to Graves Disease in UK and Polish Cohorts

Context: The genetic background of young-onset Graves disease (GD) remains An intronic variant in human leukocyte antigen (HLA) complex P5 (HCP5) has previously been associated with GD susceptibility and age of onset in a cohort of Polish patients. Objective: We aimed to investigate the association of the HCP5 variant rs3094228 with GD susceptibility and age of onset in a UK cohort and conduct a meta-analysis of UK and Polish data. Design and Participants: rs3094228 was genotyped in 469 UK patients with GD using Taqman chemistry. Genotype frequencies were compared with genotypic data available from the Wellcome Trust case-control consortium using logistic regression analysis. To determine whether rs3094228 is independently associated with age of GD onset, the HLA DRB1*0301 tagging variant, rs535777, was also genotyped. Results: The C allele of rs3094228 was overrepresented in the UK GD cohort compared with controls ( P allele =5.08 × 10 –9 , odds ratio 1.76; [95% confidence interval, 1.46-2.13]). This association was more marked in young-onset GD (<30 years) ( P allele =1.70 × 10 –10 vs P allele =0.0008). The meta-analysis of UK and Polish data supported the association of the C allele with GD susceptibility ( P allele =1.79 × 10 –5 ) and age of onset ( P allele =5.63 × 10 –8 ). Haplotype analysis demonstrated that rs3094228 is associated with age of GD onset ( P = 2.39 × 10 -6 ) independent of linkage disequilibrium with HLA DRB1*0301 . Conclusion: The rs3094228 HCP5 polymorphism is independently associated with GD susceptibility and age of onset in a UK GD cohort. Our findings indicate a potential role of long noncoding ribonucleic acids, including HCP5, in GD pathogenesis, particularly in the younger population. ( J Clin Endocrinol Metab 105: e3277–e3284, 2020)


G
raves disease (GD) is the most common cause of hyperthyroidism in young children and adolescents; however, it is still relatively rare, with a reported incidence of between 0.1 and 3 per 100 000 (1) compared with Caucasian European adults where the incidence is reported as 20 to 25 cases per 100 000 (2,3).Longitudinal studies have reported an increasing incidence of hyperthyroidism in both adult and pediatric populations [4][5][6].Similar to other autoimmune conditions, there is a clear female preponderance, with GD affecting up to 3% of women and 0.5% of men, with a peak incidence occurring between 30 and 50 years of age (2).
GD is characterized by the presence of thyroid receptor autoantibodies (TRAbs) that stimulate the cell-surface thyroid-stimulating hormone (TSH) receptor, directly resulting in excessive, autonomous thyroid hormone secretion.The clinical features and prognosis of GD is highly variable depending on age of disease onset, with the remission rate following a course of antithyroid medication being as low as 25% in the pediatric population compared with 50% to 60% in adults (7,8).
GD results from a complex interaction between genetic and environmental factors involving variants in multiple susceptibility genes, each exerting modest individual effects.Family and twin studies over the past 50 years have demonstrated that up to 80% of an individual's predisposition to GD is attributable to genetic factors (9,10).However, only around 20% of this genetic load has been accounted for by the currently associated genomic variants (11).
Genomic polymorphisms associated with susceptibility to GD are primarily found at immune-regulatory loci, such as MHC (12), CTLA-4 (13), PTPN22 (14), and CD40 (15).A stronger genetic association is suspected in the younger population who have had less exposure to environmental factors.Several of the known susceptibility loci are also associated with a younger age of disease onset, including those at CTLA-4 (16), human leukocyte (HLA)-DRB1 (17), and PTPN22 (18), with the most strongly associated variants located at the major histocompatibility complex (MHC) locus (19,20).Determining genetic variants associated with GD can provide mechanistic insight by highlighting pathogenic functional pathways, particularly by studying the younger population where genetics may be the dominant factor (19).
This study aimed to investigate the association of the HLA complex P5 (HCP5) gene in GD susceptibility and age of onset in a UK cohort.The association between HCP5 variants and thyroid autoimmunity was first demonstrated in a multicenter population-based genome-wide association study conducted by Medici et al for serum levels of thyroid peroxidase antibodies (21).The first study showing an association of HCP5 with susceptibility to GD (rs3094228, P = 1.6 × 10 -12 ; odds ratio [OR] = 1.88), was performed in a single center by Kuś et al (22).A subsequent multicenter study with a relatively large pediatric GD cohort demonstrated the HCP5 variant, rs3094228, as a risk locus for young-onset GD (YOGD) (23).
We have studied the same HCP5 polymorphism in a UK GD cohort and performed a meta-analysis of data from the UK and Polish patient cohorts.
The patients providing these samples were of Caucasian European background and had attended outpatient endocrinology at the Royal Victoria Infirmary or the Great North Children's Hospital, Newcastle-upon-Tyne, UK.Each participant with GD was diagnosed by the following criteria: fully suppressed serum TSH with serum free thyroxine and/or free triiodothyronine above the reference range and the existence of detectable TSH receptor antibody (TRAb; ≥1.8 mU/L; Brahms Kryptor).
Genotype data from 5377 control samples from the Wellcome Trust case-control consortium (WTCCC2) database were used for comparison.Informed, written consent was obtained from all participants.This study was carried out with approval of the Leeds East (Ref.05/Q1206/144) and Berkshire Valley ethics committees (Ref.04/12/015).

HCP5 genotyping
The HCP5 variant rs3094228 was genotyped in genomic deoxyribonucleic acid extracted from venous blood using TaqMan chemistry as per the manufacturer's instructions (assay C_2995657_10) and run on the QuantStudio 7 Flex Real-Time PCR (polymerase chain reaction) System (Applied Biosystems).Twenty percent of the samples were genotyped in duplicate to ensure assay fidelity.The overall genotyping call rate was 99.8%.
In the Polish cohort, the HLA-DRB1 polymorphism was genotyped using the low-resolution single specific primerpolymerase chain reaction (SSP-PCR) method with use of the Dynal All Set SSP DR Kit or the HLA-Ready Gene DR Kit, as previously described (22).

Statistical analysis
Statistical association analysis was performed using PLINK (24) and SPSS version 25 (25).All the control sample genotypes were in Hardy-Weinberg equilibrium (P > 0.4).Study data were compared with WTCCC2 control data using logistic regression with sex as a covariate.A subgroup association analysis was performed comparing young-onset (aged <30 years) GD to older-onset (aged ≥ 30 years) GD.A metaanalysis, using the Review Manager (RevMan) Version 5.0 program (Nordic Cochrane Centre, Copenhagen, Denmark ( 26)) was then undertaken, using a random effects model to calculate ORs, 95% confidence intervals (CI) and 2-sided P values.The impact of heterogeneity between the cohorts was estimated using an I 2 index.Kaplan-Meier plots and log-rank tests were applied to determine whether genotype was significantly associated with age of GD onset.Logistic regression and haplotype analysis (UNPHASED 3.1.7( 27)) was performed to determine the independent association of rs3094228 with age of GD onset.

GD age of onset
Although an increased frequency of the C allele was present in both the YOGD (99/236; 42%) and LOGD groups (204/702; 29%) compared with controls (2118/10 754; 20%), the difference was more significant in the YOGD group (P = 1.70 × 10  2).This suggests that the C allele and CC genotype have a stronger association with susceptibility to GD at a younger age.
Using the combined UK and Polish cohorts, Kaplan-Meier plots and log-rank tests demonstrated a significant association with genotype and age of GD onset, confirming the association of the CC genotype with earlier-onset GD (P = 0.003) (Fig. 1B).In addition, a recessive model demonstrated an increased risk of earlieronset GD in homozygotes for the minor allele compared with carriers for the common allele (P = 0.001).The median age of GD onset in those with the CC genotype was 34 years compared with 40 and 43 years with the CT and TT genotypes, respectively.

Haplotype analysis
To determine whether the effect of rs3094228 on age of GD onset is independent of the common GD susceptibility locus, HLA DRB1*0301, a gender-adjusted logistic regression analysis on the Polish data was undertaken, including 439 patients with GD with available data on HLA DRB1*03, rs3094228, and age of GD onset.This demonstrated that the observed effect of rs3094228 on age of GD onset is independent from HLA DRB1*03 (P = 0.006).
This was further studied in the UK GD cohort using a known tagging single nucleotide polymorphism for HLA DRB1*0301 in the Northern and Western European (CEU) population (rs535777; r 2 =0.87,D'=0.99) (28).Consistent with the Polish data, a gender-adjusted   3).Haplotype analysis with HCP5 as the test marker and HLA DRB1*0301 as the conditioning marker demonstrated that HCP5 is independently associated with age of GD onset (P = 2.39 × 10 -6 ).

Discussion
This meta-analysis includes more than 1600 participants with GD, including more than 460 early-onset cases, from UK and Polish populations.Given the relative rarity of GD in the pediatric population, combining genotype data sets is essential to improve study power.Despite the established strong hereditary component of GD, particularly in the development of GD at a younger age, there is limited data documenting genetic risk variants specifically accounting for YOGD (19,29,30).This study finds a robust association between susceptibility to GD and an earlier age of disease onset with an intronic HCP5 polymorphism in a UK cohort, replicating previous findings in a Polish population (23).The HCP5 ribonucleic acid (RNA) gene is located within the MHC class I region, centromeric of the HLA-B gene between the MICA and MICB genes, and it encodes a long noncoding RNA (lncRNA) (31).MHC genes encode cell-surface antigen-presenting proteins that are essential to mount an autoimmune response.Genome-wide association studies have identified polymorphisms at the MHC locus as risk variants for many autoimmune and inflammatory diseases, including GD in both Asian and European populations (32,33).Furthermore, Immunochip genetic analysis has demonstrated that variants within the MHC have the strongest association with YOGD (aged <30 years) (19).
The role of the MHC class II region is well established in GD, where antigen presentation of thyroidstimulating hormone receptor to CD4 + T cells is crucial to drive the B cells to produce the pathogenic TRAb autoantibodies (34).However, genetic variants in the MHC I region (HLA-B and HLA-C) have also previously been independently associated with GD (35).Indeed, messenger RNAseq analysis of GD thyroid tissue found that HLA-C in the MHC I region was the most overexpressed gene compared with controls (36).The primary function of MHC I molecules is to present nonself antigens derived from intracellular sources, such as viruses, to CD8 + cytotoxic T cells, which then mount a cytotoxic response against the presented antigen.It has been proposed that viruses may trigger an autoimmune response through molecular mimicry where viral antigens are structurally similar to self-antigens, or bystander activation where the viral infection triggers a nonspecific activation of autoreactive cells (37).Various HCP5 variants are demonstrated to promote susceptibility to adverse immune-related cutaneous drug reactions, such as Stevens-Johnson syndrome, as well as being associated with disease progression and viral load in untreated patients with HIV, suggesting a possible specific role for HCP5 in modifying the immune response to medications or viral infections (38,39).Certain genetic variants within the MHC I region, such as HCP5 polymorphisms, may predispose to GD by modifying an individual's response to an infectious agent, resulting in an excessive immune response with the potential to initiate an autoimmune reaction in genetically susceptible individuals.
The HCP5 polymorphism rs3094228 investigated in this study has previously been associated with susceptibility to GD in a Polish cohort (22), and an association has been demonstrated between the C allele and thyroid peroxidase antibody levels in autoimmune thyroid disease (21).Other HCP5 variants have been associated with susceptibility and autoantibody production in various autoimmune disorders, including systemic lupus erythematous (40), Sjögren syndrome (41), psoriasis, and psoriatic arthritis (42).The MHC gene region is highly polymorphic and characterized by extended linkage disequilibrium (LD), making it challenging to determine functional variants from tagging single nucleotide polymorphisms, which highlights the importance of rigorous casecontrol matching in genetic association studies.LD analysis indicates that the known GD risk locus, HLA DRB1*03, and rs3099844 studied in systemic lupus erythematous and Sjögren syndrome, are in partial LD with rs3094228 (HLA DRB1*03: r 2 =0.45,D' =0.86 in the Polish population (22), rs3099844: r 2 =0.69,D' =0.93 in the British population (28)).However, our haplotype analysis with HLA DRB1*03 demonstrates that HCP5 is independently associated with age of GD onset.
Interestingly, there is accumulating evidence suggesting that lncRNA, such as HCP5, has a crucial role in the development of autoimmunity by altering the adaptive and innate immune response through transcriptional and epigenetic regulation (32,(43)(44)(45).Studies have demonstrated that lncRNAs are associated specifically with autoimmune thyroid disease (AITD), including GD.Indeed, it has been proposed that rs3094228 alters the expression of thyrocyte ligands (MICA, MICB and HLA-C) of immunoreceptors on natural killer cells in those with AITD, promoting antibody-dependent natural killer cell-mediated cytotoxicity of the thyrocyte (46), making it a potentially functional variant.Furthermore, a study by Shirasawa et al demonstrated that a polymorphism in the promoter region of a B-cell-specific antisense RNA transcript, SAS-ZFAT, is associated with susceptibility to AITD (44), and the lncRNA transcript, Heg, has demonstrated negative correlation with TRAb concentrations in untreated patients with GD (47).Although the precise mechanism remains to be determined, this suggests that both variation and dysregulation of lncRNAs are implicated in AITD, including GD.
HCP5 is expressed at high levels in cells of the immune system such as the spleen, blood, and thymus (48).The aging immune system may contribute to the phenotypic differences observed between YOGD and LOGD, where involution of the thymus and reduced B-and T-lymphocyte production may alter the mechanisms driving the autoimmune response in the older population (49).Phenotypic differences may also be explained by genetic variation including those in the MHC region (HLA subtypes DB1*02, DQA1*05, and DRB1*03), which have been associated with a higher risk of relapse in GD (50).
As this study was performed in a Caucasian population, further studies are required to investigate whether a similar effect is also detected in other (non-Caucasian) populations.Further functional studies should also aim to elucidate the underlying mechanism behind the observed association.

Conclusion
This study has confirmed a significant association of the HCP5 polymorphism, rs3094228, with GD susceptibility and age of disease onset in a UK cohort and replicates the findings from a study of patients with GD in Poland.Adult-onset and young-onset GD share multiple common genetic risk variants, many of which remain unknown.Our findings indicate a potential role for HCP5 as a contributor to GD susceptibility, particularly in the younger population.Further research to determine the role of lncRNAs, including HCP5, in the pathogenesis of early-onset GD is now warranted.

Figure 1 .
Figure 1.Age-associated cumulative incidence of Graves disease (GD) onset associated with the rs3094228 polymorphism.(A) Forest plot of meta-analysis results in GD susceptibility and age of onset in UK and Polish cohorts: meta-analysis using a random effects model of rs3094228 in the HCP5 gene examining GD susceptibility compared with controls, and YOGD compared with LOGD.(B) Kaplan-Meier analysis and log-rank test.The CC genotype was associated with earlier-onset GD.CI, confidence interval; LOGD, later-onset Graves disease (aged ≥30 years); YOGD, youngonset Graves disease (aged <30 years).Abbreviation: M-H, Mantel-Haenszel test.Downloaded from https://academic.oup.com/jcem/article/105/9/e3277/5851730 by guest on 07 March 2022