High‐resolution HLA genotyping identifies risk alleles in both class I and II for primary autoimmune neutropenia in early childhood in a Danish cohort

HLA studies in patients with autoimmune neutropenia (AIN) have shown very consistent results for the association with HLA class II alleles at low resolution. This study aimed to examine the association of both HLA class I and class II at high resolution to clarify the contribution of risk alleles to the disease. A total of 107 AIN patients were genotyped for six loci of HLA class I (HLA‐A, ‐B and ‐C) and class II (HLA‐DRB1, ‐DQB1, and ‐DPB1) genes by a high‐resolution (3‐field, 6‐digit) analysis and compared with HLA typing of 1000 healthy controls. Compared with the controls, the allele frequencies were significantly higher in AIN patients for A*02:17:01G, C*01:02:01G, DRB1*10:01:01G, DRB1*14:01:01G, DRB1*16:01:01G, DQB1*05:02:01G, and DQB1*05:03:01G but lower significant for C*03:04:01G, DRB1*04:01:01G, DRB1*13:02:01G, DQB1*03:02:01G, and DQB1*06:04:01G. Frequently associated two‐locus haplotypes were found to be DRB1*10:01:01G‐DQB1*05:01:01G and DRB1*16:01:01G‐DQB1*05:02:01G, while the S2 (Q‐ or D‐KRAA) shared epitope (SE) was associated with lower risk. A unique association with HLA alleles was observed between patients with specific anti‐HNA‐1a antibodies and broad‐reacting anti‐FcγRIIIb. Anti‐HNA‐1a antibody‐positive patients were associated with C*01:02:01G, DRB1*01:01:01G, DRB1*16:01:01G, DQB1*05:01:01G, DQB1*05:02:01G, DQB1*06:04:01G, and DPB1*10:01:01G; the two‐locus haplotypes DRB1*01:01:01G‐DQB1*05:01:01G and DRB1*16:01:01G‐DQB1*05:02:01G; and the S3P (Q‐ or R‐RRAA) SE. Anti‐FcγRIIIb antibody‐positive patients were associated with the alleles A*02:17:01G, DRB1*10:01:01G, and DQB1*05:02:01G; the haplotypes DRB1*10:01:01G‐DQB1*05:01:01G and DRB1*11:01:02G‐DQB1*05:02:01G; and the S3D (DRRAA) SE. The different associations regarding FcγRIIIb antibody specificities could indicate disease heterogeneity.


| INTRODUCTION
Genetic variations in HLA may be the result of evolutionary adaptations in response to environmental stress, such as climate and the prevalence of infectious diseases. 1The molecular mechanisms connecting HLAs and autoimmune diseases are unclear, but antigen presentation and T-cell activation seem to be triggering in most autoimmune diseases. 2,3Primary autoimmune neutropenia (AIN) in early childhood is a disease characterized by the presence of autoantibodies that recognize antigens of neutrophils (human neutrophil antigens [HNAs]), mostly located on immunoglobulin G (IgG) Fc gamma receptor type 3b (FcγRIIIb [CD16b]), causing their peripheral destruction. 4Two main types of autoantibodies have been observed in AIN, antibodies directed against a specific HNA variant (HNA-1a), and broad reactive antibodies against FcγRIIIb.The clinical significance of autoantibody specificities remains unknown, but in several studies, we have reported different genetic backgrounds for the two groups. 5,6he cause of AIN is still unknown, and there is not enough data to propose a genetic or environmental cause.Viral infections have been suggested as a possible trigger especially in secondary AIN, 7 initiating the activation of autoreactive B cells and CD4+ T cells, but there are also indications of a general deficiency in peripheral selftolerance mediated by an alteration in either the function or number of CD4+ Treg cells. 8Activation of regulatory T cells (Tregs) is controlled by HLA, and certain genotypes protect or increase the risk of exposure to autoimmune diseases through Tregs.A connection between AIN and circulating Tregs has been shown by Nakamura et al., 9 which supports HLA association and gives rise to further investigation.
A background of genetic susceptibility, especially to class II HLA, has been confirmed for multiple ethnicities.
The first association between AIN and HLA was described in 1991 in a German cohort (n = 26), which showed an association with serologically determined DR2 (DRB1*15 and *16) and DQ1 (DQB1*05 and *06). 10ater, the DQB1*05:03 genotype was suggested for AIN in a small study from Taiwan (n = 31) and confirmed by our group in a Danish cohort (n = 80), where we also found an association with DRB1*14. 11,12In another study, we explored the relationship between DRB1 and DQB1 alleles and AIN, where we expanded the cohort to 160 patients and compared it with a control group of 1000 healthy Danish individuals. 5In doing so, we found a higher risk associated with DRB1*10, DRB1*14, DRB1*16, and DQB1*05 and a lower risk associated with DRB1*04, DRB1*13, and DQB1*03.We also found that the associations with DRB1 and DQB1 alleles differed between patients positive for anti-HNA-1a-specific antibodies and patients positive for broadly reactive anti-FcγRIIIb antibodies.DRB1*01, DRB1*04, and DQB1*03 were only associated with anti-HNA-1a positivity, and DRB1*10 was restricted to broadly reactive anti-FcγRIIIb positivity.
Class I and II HLA alleles are important genetic risk factors for a variety of autoimmune diseases, 13 but prior to this study, only HLA class II alleles were investigated in regard to AIN.Study results for DRB1 and DQB1 have been concordant but with limited genotyping resolutions at the 1-field level.A study with higher resolution will not only provide more knowledge about subtypes and haplotypes but also make it possible to investigate the shared epitope (SE) concept in relation to AIN.SE's are based on amino acid motifs.5][16] The SE hypothesis predicts that RA-associated DRB1 molecules bind the same peptide(s) and thus facilitate the development of autoreactive T cells involved in the pathogenesis of RA. 17 HLA-B SE are called Bw4 and Bw6 and are determined by amino acid residues 77 and 80-83, respectively, on the α-1 helix. 18,19his study examined the association of six loci at the high resolution (3-field, 6 digit), HLA-A, -B, -C, -DRB1, -DQB1, and -DPB1, using third-generation sequencing technology to clarify their genetic contribution to AIN.The association between risk alleles and the specificity of autoantibodies, as well as two-locus haplotypes and SE's, was determined.The HLA gene pairs with the strongest linkage disequilibrium (LD) values were B-C and DRB1-DQB1, which are expected due to their chromosomal proximity. 20Because of the high disease association with the DRB1 and DQB1 alleles, we chose to study this two-locus haplotype.

| Study cohort
The Department of Clinical Immunology, Aalborg University Hospital, is the national center for diagnostic AIN testing in Denmark and was the center for sample collection in this study.A total of 107 patients were included, all diagnosed with AIN and tested positive for FcγRIIIb autoantibodies between 2004 and 2023.The inclusion criteria were the presence of neutropenia, an absolute neutrophil count not above 1.5 Â 10 9 cell/L in two repeated tests, age under 5 years at the time of diagnosis, and the presence of antineutrophil antibodies in the flow cytometric indirect granulocyte immunofluorescence test (Flow-GIFT) as previously described. 12Patients with initial negative antibody screening underwent repeated tests as suggested by Bux. 21atients with congenital neutropenia, neutropenia related to inborn syndromes, postinfection neutropenia, or hematological malignancies were excluded.The 107 included patients belong to the same cohort as previously published studies investigating HLA association with AIN in Danish patients at low resolution. 5,12The control group consisted of 1000 randomly selected, anonymous, healthy adult Danish blood donors from the Aalborg University Hospital blood bank, Aalborg, Denmark.A high number of controls was used to provide certainty of the frequency of rare alleles.Both, patients and controls, consisted primarily of White individuals.
All procedures performed in the study were in accordance with the ethical standard of the institution and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.It was approved by the North Denmark Region Committee on Health Research Ethics (Approval number: N-20170026).Oral and written consent from the participants or their guardians was obtained according to the Danish Healthcare Act.

| DNA preparation
DNA was extracted from EDTA-stabilized whole blood using the Maxwell RSC Whole Blood DNA Kit on the Maxwell RSC instrument (Promega, USA).

| Shared epitopes
HLA-B alleles were divided into two groups, Bw4 and Bw6, according to the IPD-IMGT/HLA database. 22This classification of DRB1 alleles, observed in our samples, was in accordance with the amino acid sequence at positions 70-74, as described by du Montcel et al. 15 DRB1 alleles were divided into two groups (S and X alleles) according to the presence or absence of the RAA sequence at positions 72-74.The S alleles were subdivided into three categories depending on the amino acid at position 71.The S1 group had an alanine (A) or a glutamic acid (E) at position 71 (A-or E-RAA).S2 had a lysine (K) at position 71 (K-RAA), and S3 had an arginine (R) in the sequence (R-RAA).S2 had either Q or D at position 70 (Q-or D-KRAA).S3 was subdivided according to the amino acid at position 70 into S3D (D-RRAA) and S3P (Q-or R-RRAA).

| Statistics
The HLA allele frequencies and two-locus haplotype frequencies were characterized by direct counting of the respective number of HLA alleles among healthy controls and AIN patients.Statistical analyses of the differences among patients and controls were determined by Fisher's exact test using Stata v.17 (StataCorp LLC, College Station, TX, USA).Two-tailed p values, odds ratios (ORs), and 95% CIs were obtained.4][25] Bonferroni correction was used to adjust p values in the case of multiple statistical testing; however, no adjustment for multiple comparisons was applied to the SE, as this was considered an exploratory study.

| Demographic and clinical data
We included 107 patients diagnosed with AIN with a median age at diagnosis of 14.2 months (range, 3-54 months).All AIN patients were positive for anti-FcγRIIIb antibodies, and of these, 50.5% had anti-HNA-1aspecific antibodies.The patients were investigated both as a combined group and as two individual groups divided according to their antibody specificity.The two groups consisted of 54 patients who were anti-HNA-1a antibody positive and 53 patients who were anti-FcγRIIIb antibody positive.The control group consisted of 1000 healthy randomly selected and unrelated Danish blood donors.The control cohort was not in HWE for A, B, and DPB1 but was in HWE for C, DRB1, and DQB1 (Table S11).In this study, we present HLA alleles and two-locus haplotype frequencies, not as a population study but as a comparison to a disease cohort of the same origin.The control group consisted of randomly selected healthy individuals, but association with alleles not in HWE should be taken with caution.

| HLA typing and linkage analysis
High-resolution analysis of HLA alleles in 1000 healthy controls and 107 AIN patients was performed and analyzed at G-group resolution.Associated class I and II alleles are reported in Table 1, and all detected HLA alleles are reported in Tables S1-S6.Linkage analysis revealed that the pairs B-C (D 0 = 0.88) and DRB1-DQB1 (D 0 = 0.93) had the highest LD values (Figure S1).Two-locus haplotype frequencies of DRB1-DQB1 were determined in cases and controls.Associated two-locus haplotypes are reported in Table 2, and all haplotypes with frequencies above 1% are reported in Tables S7-S9.SE's in HLA-B and -DRB1 proteins were determined based on frequencies of the alleles in the two loci, and comparisons between the healthy controls and AIN patients can be seen in Tables 3 and S10.

| Protective HLA alleles and HLA class II two-locus haplotypes in AIN patients
Protection against AIN was not observed for any HLA-A or -B alleles but was observed for C*03:04:01G (p = 0.0003, OR = 0.35 [0.17-0.65])(Table 1).A protective association of HLA class II alleles for AIN was found   (Table 1).No association was observed for DPB1 alleles or any DRB1-DQB1 haplotypes (Tables 1 and 2).The S2 (Q-or D-KRAA) SE for the DRB1 protein was found to be protective against AIN ( p = 0.0016, OR = 0.38 [0.18-0.73])(Table 3).Analysis of HLA-Bw epitopes showed no significant association with AIN (Table S10).An overview of the protective-associated alleles, haplotypes, and SE's is presented in Table 4.  3).An overview of the associated alleles, two-locus haplotypes, and SE's for the two antibody-specific groups is presented in Table 4.

| DISCUSSION
In this study of primary AIN in early childhood among Danish patients, we analyzed the frequency of class I and II HLA alleles with G-group resolution in 107 patients and 1000 healthy controls and determined the associated two-locus haplotypes and SE's.As expected, the associated HLA class II alleles in DRB1 and DQB1 were similar to those that have been previously described for AIN, 5,[10][11][12] but we are the first to report associations with class I alleles.This is also the first study to investigate HLA association with AIN at the high resolution (3-field, 6-digit), as well as SE's.Interestingly, we found associations with alleles for HLA-A and -C.The A*02:17:01G allele is rare, and while it was not observed in the control group, we did observe it in the patient group.Thus far, there is no previous disease association with this allele in the literature, and investigation of the patients carrying it did not reveal any ethnic explanation.However, the strongest HLA-A binders with virus peptides have been found to belong to the A*02 lineage. 26For HLA-C, both a risk allele and a protective allele were identified.C*01:02:01G was more frequent in AIN patients than in healthy controls.The proposed function of HLA-C protein in autoimmune and inflammatory diseases is to present antigens to T cells and to drive innate immunity through binding activating or inhibitory receptors on natural killer (NK) cells. 27The proposed interaction among viral infection, C*01, is through the killer cell immunoglobulin-like receptors of NK cells. 28In this study, protection against AIN was observed for the allele C*03:04:01G, and the C*03:04 allele has previously been linked to sarcoidosis in Korean patients. 29or HLA class II, we replicated all our previously published low-resolution findings of associated DRB1 alleles in this cohort of Danish AIN patients 5,12 and, at a G-group resolution, found the risk alleles to be DRB1*10:01:01G, DRB1*14:01:01G, DRB1*16:01:01G and the protective alleles to be DRB1*04:01:01G and DRB1*13:02:01G.For DQB1, we similarly replicated our previous findings for DQB1*03 and DQB1*05, and we found an association with DQB1*03:02:01G, DQB1*05:02:01G, and DQB1*05:03:01G.In addition, we also observed a protective effect of DQB1*06:04:01G.
1][12] The study from Taiwan is the only study that has reported more than 1 digit, and they reported an association with DQB1*05:03. 11Our results support that DQB1*05:03:01G was indeed associated with a higher risk of AIN, but also DQB1*05:01:01G and DQB1*05:02:01G were found to be associated.
Other of the associated alleles have previously been associated to autoimmune diseases.DRB1*14 has previously been found to be strongly associated with a higher risk of the rare autoimmune disease pemphigus vulgaris in White Europeans. 30,31In pemphigus vulgaris, IgG autoantibodies bind to the protein desmoglein 3 (dsg3), which is found in desmosomes in keratinocytes near the bottom of the epidermis. 30,31Pemphigus vulgaris has also been associated with DRB1*04, DQB1*03:02, DQB1*05:01, and DQB1*05:03. 30,31e DRB1*10:01 allele has been found to be a strong binder of virus peptides. 26It has also been associated with a relatively new disorder caused by neurodegenerative and autoimmune mechanisms named anti-IgLON5 disease.The main symptoms of the disease are sleeping disturbances, bulbar symptoms, and gait abnormalities.Similar to AIN, anti-IgLON5 disease is associated with a highly specific antibody and particular HLA alleles.The frequency of the DRB1*10:01 allele is very low and, when present, strongly segregates with DQB1*05:01, which has also been found to be associated with IgLON5. 32The twolocus haplotype of DRB1*16:01:01-DQB1*05:02:01 has been associated with drug-induced liver injury. 33his differentiation between HLA associations and HNA-1a antibody specificities in AIN patients has been observed previously for both DRB1 and DQB1 5 but also in regards to genetic variation in low-to-medium Fcγ receptors. 6In this study, we observed differences regarding HLA class I and II alleles.DRB1*01, which is unique to anti-HNA-1a positivity, has been reported to be protective for sarcoidosis [34][35][36][37] and a predisposing factor in diseases such as recurrent lymphocytic meningitis and juvenile idiopathic arthritis in Hungarian patients, 38,39 exhibiting contrary effects depending on the disease.HLA-DRB1*01:01 has been shown to be a strong binder for viruses and is overrepresented in individuals who develop infectious mononucleosis compared with individuals with asymptomatic Epstein Barr virus (EBV) infection. 26,40A more significant reduction in EBV copy number has been observed in individuals with HLA-DRB1*01:01, suggesting that immune control of viral replication is more effective in these individuals. 41PB1 has not been studied in AIN to date, and we found that DPB1*10:01:01 was a risk factor for patients with anti-HNA-1a antibodies.DPB1*10 has been found to be associated with a higher risk of severe aplastic anemia.42 The etiology of acquired severe aplastic anemia is Abbreviations: AIN, autoimmune neutropenia; CI, confidence interval; HNA, human neutrophil antigens; OR, odds ratio, SE, shared epitope.not understood but is likely related to abnormal immune responses and environmental exposures.
We investigated SE's for both HLA-B and -DRB1 and did not find an association with HLA-B SEs.AIN has been associated with several DRB1 alleles, and we found associations with several DRB1 SE groups.Both the combined group of AIN patients and the two antibody subgroups had protective associations with the S2 (Q-or D-KRAA) allele group.This corresponds with S2 containing DRB1*04 and DRB1*13, which are both found to be protective.Anti-HNA-1a positivity had a risk association with S3P (Q-or R-RRAA), corresponding with the unique signal to DRB1*01 for this patient group.For anti-FcγRIIIb-positive patients, the S3D (DRRAA) group was associated with higher disease risk, which corresponds with the DRB1*16 association.In the case of SE's, we see an indication that all patients share the same protective epitope but that the risk epitope is connected to antibody specificity.
In this study, we found several associations for different HLA alleles, although we applied a strict correction factor.As stated in the beginning of the discussion, we did not observe any difference from our previous findings of associated DRB1 and DQB1 alleles in this cohort, which is why we feel confident that the cohort is representative of Danish AIN patients.Nevertheless, our study has limitations, and due to the small number of patients, some of the observed associations, especially in comparisons with antibody specificity, could arise by chance and therefore should not be overinterpreted before Note: NA = not applicable (only observed in one group).Abbreviations: AIN, autoimmune neutropenia; HNA, human neutrophil antigens; OR, odds ratio.
independent replication in another cohort.Our patients were selected according to strict criteria, and our findings can only be expected to represent this restricted portion of patients suffering from neutropenia.We did not obtain HWE in the control group for HLA-A, -B and -DPB1, which is why the findings in these loci should be interpreted with caution.The controls and patients are genotyped with two different methods and due to the high level of ambiguities in next generation sequencing data, the comparison was only possible at G-group resolution.This study presents a high number of alleles associated with AIN.An explanation for this could be the wide range of infections that is observed among this group of patients.4][45][46][47] Additionally, among Danish AIN patients, a wide range of different infections is observed, including upper respiratory tract infections, otitis media, dermal infections/abscesses, gastroenteritis, oral infection, conjunctivitis, and pneumonia (unpublished data).Our results suggest that specific HLA alleles and haplotypes might play a role in susceptibility to and protection against AIN.We also observed different associations regarding FcγRIIIb antibody specificities, which could indicate disease heterogeneity and may be related to different infections.
Frequencies of DRB1-shared epitopes in AIN patients and controls.