Association of HLA-class II alleles with risk of relapse in myeloperoxidase-antineutrophil cytoplasmic antibody positive vasculitis in the Japanese population

Background Disease relapse remains a major problem in the management of antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV). In European populations, HLA-DPB1*04:01 is associated with both susceptibility and relapse risk in proteinase 3-ANCA positive AAV. In a Japanese population, we previously reported an association between HLA-DRB1*09:01 and DQB1*03:03 with susceptibility to, and DRB1*13:02 with protection from, myeloperoxidase-ANCA positive AAV (MPO-AAV). Subsequently, the association of DQA1*03:02, which is in strong linkage disequilibrium with DRB1*09:01 and DQB1*03:03, with MPO-AAV susceptibility was reported in a Chinese population. However, an association between these alleles and risk of relapse has not yet been reported. Here, we examined whether HLA-class II is associated with the risk of relapse in MPO-AAV. Methods First, the association of HLA-DQA1*03:02 with susceptibility to MPO-AAV and microscopic polyangiitis (MPA) and its relationship with previously reported DRB1*09:01 and DQB1*03:03 were examined in 440 Japanese patients and 779 healthy controls. Next, the association with risk of relapse was analyzed in 199 MPO-ANCA positive, PR3-ANCA negative patients enrolled in previously reported cohort studies on remission induction therapy. Uncorrected P values (Puncorr) were corrected for multiple comparisons in each analysis using the false discovery rate method. Results The association of DQA1*03:02 with susceptibility to MPO-AAV and MPA was confirmed in a Japanese population (MPO-AAV: Puncorr=5.8x10-7, odds ratio [OR] 1.74, 95% confidence interval [CI] 1.40–2.16, MPA: Puncorr=1.1x10-5, OR 1.71, 95%CI 1.34–2.17). DQA1*03:02 was in strong linkage disequilibrium with DRB1*09:01 and DQB1*03:03, and the causal allele could not be determined using conditional logistic regression analysis. Relapse-free survival was shorter with nominal significance in carriers of DRB1*09:01 (Puncorr=0.049, Q=0.42, hazard ratio [HR]:1.87), DQA1*03:02 (Puncorr=0.020, Q=0.22, HR:2.11) and DQB1*03:03 (Puncorr=0.043, Q=0.48, HR:1.91) than in non-carriers in the log-rank test. Conversely, serine carriers at position 13 of HLA-DRβ1 (HLA-DRβ1_13S), including DRB1*13:02 carriers, showed longer relapse-free survival with nominal significance (Puncorr=0.010, Q=0.42, HR:0.31). By combining DQA1*03:02 and HLA-DRβ1_13S, a significant difference was detected between groups with the highest and lowest risk for relapse (Puncorr=0.0055, Q=0.033, HR:4.02). Conclusion HLA-class II is associated not only with susceptibility to MPO-AAV but also with risk of relapse in the Japanese population.

association between these alleles and risk of relapse has not yet been reported. Here, we examined whether HLA-class II is associated with the risk of relapse in MPO-AAV.
Methods: First, the association of HLA-DQA1*03:02 with susceptibility to MPO-AAV and microscopic polyangiitis (MPA) and its relationship with previously reported DRB1*09:01 and DQB1*03:03 were examined in 440 Japanese patients and 779 healthy controls. Next, the association with risk of relapse was analyzed in 199 MPO-ANCA positive, PR3-ANCA negative patients enrolled in previously reported cohort studies on remission induction therapy. Uncorrected P values (P uncorr ) were corrected for multiple comparisons in each analysis using the false discovery rate method.

Introduction
Antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is a group of necrotizing small vessel vasculitides characterized by ANCA production, mainly against proteinase 3 (PR3) or myeloperoxidase (MPO). AAV is classified as microscopic polyangiitis (MPA), granulomatosis with polyangiitis (GPA), or eosinophilic granulomatosis with polyangiitis (EGPA) according to the European Medicines Agency (EMA) algorithm (1). In AAV, epidemiological differences between European and Asian populations are well known. In European populations, GPA and PR3-ANCA positive AAV (PR3-AAV) are predominant, whereas in the Japanese population, MPA and MPO-ANCA-positive AAV (MPO-AAV) account for the majority of AAV cases (2). Such ethnic differences in epidemiology imply that the genetic background of AAV may play a role in its development.
Although patients with AAV achieve remission with immunosuppressive therapy (3), a substantial proportion experience relapse. In European populations, relapse occurs in approximately half of the patients with GPA within 5 years after achieving complete remission (4). Among AAV patients, the risk of relapse has been shown to be higher in those with GPA and PR3-AAV compared with MPA and MPO-AAV (5). In Japanese observational studies conducted to identify risk factors for AAV relapse, the dosage and tapering speed of prednisolone were associated with relapse (6,7).
Although several genes are associated with AAV susceptibility, those associated with relapse have not been well-characterized. In European populations, HLA-DPB1*04:01, the susceptibility allele for GPA, has been shown to be associated with a higher risk of AAV (17) and PR3-AAV relapse (18). However, similar studies have not been conducted in East Asian populations, in which MPA and MPO-AAV account for the majority of AAV cases.
In this study, we focused on MPO-AAV in the Japanese population, and examined whether HLA-DRB1, DQA1, DQB1 and DPB1 alleles are also associated with relapse in MPO-AAV, based on clinical data from Japanese nationwide cohort studies on remission induction therapy in AAV. Data were obtained from "Remission Induction Therapy in Japanese Patients with ANCAassociated Vasculitides" (RemIT-JAV), registered with the University Hospital Medical Information Network Clinical Trials Registry (UMIN000001648) (19) and "Remission Induction Therapy in Japanese Patients with ANCA-associated Vasculitides and Rapidly Progressive Glomerulonephritis" (RemIT-JAV-RPGN) (UMIN000005136) (20)
Genomic DNA samples were obtained from institutes participating in Japan Research Committee of the Ministry of Health, Labour, and Welfare for Intractable Vasculitis (JPVAS) and the Research Committee of Progressive Renal Disease, both organized by the Ministry of Health, Labour, and Welfare of Japan, as well as from research groups organized by Tokyo Medical and Dental University and University of Tsukuba. A total of 264 control samples were obtained from the Health Science Research Resources Bank (Osaka, Japan).
Among patients with MPO-AAV, 88 enrolled in RemIT-JAV (19) and 176 enrolled in RemIT-JAV-RPGN (20) were analyzed for relapse-free survival. The characteristics of these cohorts have been previously described (19,20). Briefly, the enrollment period was from April 2009 to December 2010 (RemIT-JAV) and from April 2011 to March 2014 (RemIT-JAV-RPGN). The enrollment criteria were as follows: 1) receiving a diagnosis of AAV by site investigators, 2) fulfilling the criteria for primary systemic vasculitis as proposed by the EMA algorithm, and 3) starting immunosuppressive treatment

Genotyping
In all patients and controls, HLA-DRB1 alleles were determined at the four-digit level using a WAKFlow HLA typing kit (Wakunaga Pharmaceutical Co., Ltd., Osaka, Japan) based on polymerase chain reaction-sequence-specific oligonucleotide probes (PCR-SSOP). HLA-DQA1, DQB1 and DPB1 alleles were genotyped using this system for the199 patients with MPO-AAV examined for relapse-free survival, In susceptibility analysis, rs11545686C was used as a proxy for HLA-DQA1*03:02. Genotyping of rs11545686 was conducted by Sanger sequencing using 3130xl Genetic Analyzer (Thermo Fisher Scientific, Waltham, MA, USA). For amplification of HLA-DQA1 region surrounding rs11545686, forward (5'-TTTGGTTTGGG TGTCTTCAGATT-3') and reverse (5'-AAAGTTGTTCAGGGA AATTTGAGAATG-3') primers (Primer ID: Hs00412887_CE, Thermo Fisher Scientific) were used and cycle sequencing reaction was conducted using the forward primer and the BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific). Representative Sanger sequencing chromatograms for each genotype are shown in Supplementary Figure S1.

Statistical analysis
The characteristics of MPO-AAV patients with and without relapse were compared using Fisher's exact test in two-by-two tables, excluding age, which was compared using the Mann-Whitney U test.
Relapse-free survival curves were generated using the Kaplan-Meier method and were compared among patients who were classified into GPA, MPA, EGPA and unclassifiable (UC) based on the EMA classification, among the patients treated with glucocorticoid (GC) alone, GC plus immunosuppressants, and immunosuppressants alone, and between patients with and without each HLA class II allele (HLA-DRB1, DQA1, DQB1 and DPB1), and each amino acid encoded by these HLA alleles, using log-rank test. Unadjusted hazard ratios (HRs) and adjusted HRs for E M A c l a s s i fi c a t io n , t r e a t m e n t ( G C a l o n e , G C p l u s immunosuppressants or immunosuppressants alone), age and sex were calculated using Cox proportional hazard model.
The association of HLA-DQA1*03:02 and DRB1*09:01 with susceptibility to MPO-AAV and MPA was tested using logistic regression analysis with an additive model.
Statistical analyses were performed using the R software version 3.5.2. Correction for multiple testing in risk of relapse analysis at each HLA locus and amino acid position was performed by controlling the false discovery rate. Q<0.1 was considered significant. There were 17 DRB1 alleles, 11 DQA1 alleles, 11DQB1 alleles, 8 DPB1 alleles, 79 DRb1 amino acids, 47 DQa1 amino acids, 71 DQb1 amino acids, and 26 DPb1 amino acids. The log-rank test for the pairwise comparison between the combination of presence/ absence of DQA1*03:02 and DRb1_13S was corrected for six comparisons, namely, the combination of 2 out of 4 ( 4 C 2 = 6).

Ethics
This study was reviewed and approved by the Faculty of Medicine Ethics Committee of University of Tsukuba (Approval ID: 122, 123, 180, 227, 268).
This study was also approved by the Ethics Committees of the following institutes that participated in the collaboration and/or recruitment of subjects: Aichi Medical University, Asahikawa
Association of HLA class II alleles with relapse-free survival in MPO-AAV Figure 1 shows a flow chart of the MPO-AAV patients enrolled in the relapse-free survival analysis. Among the 264 AAV patients who enrolled in the cohort studies RemIT-JAV (19) and RemIT-JAV-RPGN (20), more than 80% of patients were MPO-ANCA single-positive. As the frequency of relapse was reported to be higher in AAV patients with PR3-ANCA than in those without it (5), this study focused on 199 MPO-ANCA single-positive patients who achieved remission during the observation period. Patient characteristics are shown in Table 3. During the first 3 months after treatment initiation, 84 (42.2%) patients received GC alone, 112 (56.3%) were treated with GC plus immunosuppressants, and three were treated with immunosuppressants alone (1.5%). In most patients, cyclophosphamide was administered as an immunosuppressant.
The mean observation period from the start of treatment was 676.7 (SD: 147.0) days. Among the 199 MPO-ANCA single positive AAV patients, 39 patients experienced relapse (19.6%). The mean time to relapse after remission was 228.1 (SD: 173.5) days. No significant difference in sex ratio, age, EMA clinical classification, the treatment received, or observation period were observed between patients with and without relapse (Table 3).
Next, we tested whether the EMA classification and the treatment modalities were associated with the risk of relapse using the Kaplan-Meier method for relapse-free survival. GPA patients have previously been shown to be associated with high occurrence of relapse (5). No significant difference in relapse-free survival was observed among MPO-ANCA single positive AAV patients classified by the EMA algorithm (log-rank test uncorrected P [P uncorr ]=0.097) (Figure 2A (Table 4). With respect to DPB1 alleles, no trend toward association was observed among patients with MPO-ANCA single-positive AAV.
Next, we used the Kaplan-Meier method to compare relapsefree survival between the carriers and non-carriers of each HLA allele. As shown in Supplementary Tables S1-S4, a nominal association was observed in DQA1*03:02 (P uncorr =0.020, Q=0.22), DRB1*09:01 (P u n c o r r =0.049, Q=0.42) and DQB1*03:03 When HRs for the associated HLA alleles were conditioned on the EMA classification (MPA, GPA or UC), treatment modality during the initial three months (GC alone, immunosuppressants alone and GC plus immunosuppressants), age and sex using a multivariate Cox proportional hazards model, the associations of DRB1*09:01, DQA1*03:02 and DQB1*03:03 were not affected, whereas the association of DRB1*08:02 was attenuated after conditioning (Table 5). In this analysis, EGPA was not included in the variables of the EMA classification, because no patients with EGPA experienced relapse.
Next, we examined whether specific amino acids in the HLA molecules were associated with relapse. The results of the log-rank test for each amino acid are presented in Supplementary Data 2. The strongest association trend was observed in serine residue at position 13 of HLA-DRb1 (HLA-DRb1_13S: log-rank P uncorr =0.010, Q=0.42). The relapse-free survival period was longer in patients carrying HLA-DRb1_13S (HR: 0.31, 95% CI: 0.12-0.80) ( Figure 3E, Table 5).
When the associations of the HLA alleles nominally associated with the risk of relapse and HLA-DRb1_13S were conditioned on each other, a tendency toward association remained in DRb1_13S (P conditional = 0.057 [conditional on DQA1*03:02] and 0.040 [conditional on DQB1*03:03]) ( Table 6).

Discussion
In this study, we examined whether HLA class II alleles and HLA amino acids were associated with relapse in MPO-AAV, using data from two Japanese nationwide cohort studies on remission induction therapy: RemIT-JAV (19) and RemIT-JAV-RPGN (20). We detected that MPO-ANCA single positive AAV patients carrying HLA-DRB1*09:01, DQA1*03:02 or DQB1*03:03 had nominally higher risk of relapse as compared with those without these alleles, while AAV patients carrying serine residue at position 13 of HLA-DRb1 had a trend toward lower risk of relapse. By combining these two factors, significant differences were observed in relapse-free survival among the patient subgroups. To the best of our knowledge, this is the first study to demonstrate the association between gene and amino acid variations and AAV relapse in an East Asian population.
We previously identified an association between HLA-DRB1*09:01 and DQB1*03:03 and susceptibility to MPA/MPO-AAV in a Japanese population (12)(13)(14)(15). In the present study, we confirmed the association of DQA1*03:02 with susceptibility to MPO-AAV in a Japanese population, which was recently reported in a Chinese population (16). Owing to the tight linkage disequilibrium among these three HLA class II alleles, we could not determine which of the three alleles played a primary causative role.
Currently, the molecular mechanisms underlying the association between HLA class II alleles and MPA/MPO-AAV remain unclear. Among the components of the risk haplotype DRB1*09:01-DQA1*03:02-DQB1*03:03, the polymorphic amino acids encoded by DRB1*09:01 and DQB1*03:03 directly affect antigenic peptide specificity. Because MPO-ANCA has been strongly implicated in MPO-AAV pathogenesis (25), it is possible that specific antigenic peptides derived from human MPO (26) or microbial peptides mimicking human MPO peptides (27) might be Carrier frequency (%) of each HLA allele was compared between the patients with and without relapse by Fisher's exact test. Uncorrected P values (P uncorr ) and false discovery rate (FDR) Q values are shown. *2 Time to relapse after achievement of remission was compared between the patients carrying and not-carrying each HLA allele using log-rank test. P uncorr values for the relapse-free survival rate calculated using the Kaplan-Meier method and FDR Q values are shown. *3 One patient, whose HLA-DQA1 genotype could not be determined, was excluded from the analysis of HLA-DQA1.

FIGURE 2
Relapse-free survival in MPO-ANCA single-positive AAV patients according to the EMA classification and treatment modality. The longitudinal and horizontal axes represent the probability of relapse-free survival and days after achievement of remission, respectively. P value was calculated by log-rank test. preferentially presented by HLA-DRB1*09:01 and/or DQB1*03:03 products, although to our knowledge such peptides have not been identified from the patients for DRB1*09:01 or DQB1*03:03 products, Conversely, the specific amino acid encoded by DQA1*03:02, Asp160, is located in the a2 domain of HLA-DQ molecule, and may be involved in stabilizing HLA class II homodimers, thereby affecting antigenic peptide specificity (16).
Thus, it is possible that multiple molecular mechanisms associated with MHC-peptide complex formation and T cell receptor signaling are involved in the genetic association between the risk haplotype and susceptibility to MPO-AAV.
A previous report showed that HLA-DPB1*04:01, the susceptibility allele for GPA, was also associated with an increased risk of AAV relapse in the Dutch and German patients  Association of HLA alleles and DRb1 position 13 serine with relapse-free survival in MPO-ANCA single positive AAV patients using Kaplan-Meier method. The longitudinal and horizontal axes represent the probability of relapse-free survival and days after achievement of remission, respectively. Uncorrected P values calculated by log-rank test are shown. Q values corrected for multiple comparisons for each locus are shown in the text, Table 4 and Table 5. (17). In this study, GPA and PR3-AAV were predominant. In contrast, this association was not observed in a Danish population (28). Although the reason for this discrepancy is unclear, differences in DPB1*04:01 carrier frequency among patients between these studies (77% (17) and 94% (28), respectively) may play a role. A recent study from the United States reported an association between HLA-DPB1*04:01 and relapse in PR3-AAV, but not in MPO-AAV (18). With respect to MPO-AAV, a small-scale study in a European population did not detect a significant association between relapse and HLA-A, -B, or -DR antigens (29). In our study, HLA-DRB1*09:01, -DQA1*03:02 and -DQB1*03:03, the susceptibility alleles for MPO-AAV in East Asian populations (12)(13)(14)(15)(16) showed a trend toward a higher risk of relapse of MPO-AAV in a Japanese population, possibly suggesting that the susceptibility alleles for AAV may also be associated with relapse in each ethnic group.

HLA-amino acid
Unlike a European study (17), this study did not detect an association between DPB1*04:01 and the risk of AAV relapse. This discrepancy likely occurred because this study focused on MPO-ANCA positive AAV, whereas in the previous European study, 51.7% of the subjects were positive for PR3-ANCA (17). In our MPO-AAV data, no significant association was detected between DPB1*04:01 and relapse (Supplementary Table S4, https://doi.org/ 10.6084/m9.figshare.21876159.v3), which is consistent with a report from the United States (18). This difference may not be explained merely by the low carrier frequency of DPB1*04:01 in the Japanese population, because DPB1*04:01 showed a slight tendency toward decrease in patients with relapse (P=0.70, OR: 0.39). In the Japanese population, the protective allele for MPO-AAV susceptibility, DRB1*13:02, is in linkage disequilibrium with DPB1*04:01 (r 2 = 0.44), and the allele frequency of DPB1*04:01 was also decreased in MPO-AAV (P=2.1x10 -4 , OR: 0.40) due to linkage disequilibrium with DRB1*13:02 (14). Thus, to evaluate the effect of DPB1*04:01 in the Japanese population, PR3-ANCA positive AAV patients with sufficient sample sizes should be investigated in the future.
In contrast to the HLA-DRB1*09:01-DQA1*03:02-DQB1*03:03 haplotype, alleles carrying HLA-DRb1_13S were found to show a trend toward protection from relapse. HLA-DRb1_13S is encoded by DR3, DR11, DR13 and DR14. HLA-DRB1 position 13 amino acid is associated with multiple immune system disorders, and was recently shown to be most strongly associated with individual differences in the T cell receptor complementarity determining region 3 (CDR3) repertoire (30). HRs (95% CI) and log-rank P values for relapse-free survival of alleles encoding HLA-DRb1_13S with minor allele frequency >0.01 in the Japanese population are shown in Supplementary Table S5. These alleles showed a trend toward protection from relapse, although the differences did not reach statistical significance.
As mentioned above, we previously reported that DRB1*13:02, encoding HLA-DRb1_13S, exhibited a protective effect against the MPO-AAV development (14). The data on susceptibility to MPO-AAV with HLA-DRb1_13S, as well as with each allele encoding 13S, are also presented in Supplementary Table S5. HLA-DRb1_13S was protective against the development of MPO-AAV (P uncorr =0.0029, OR 0.71, 95%CI 0.56-.89). Among the HLA-DRb1_13S encoding alleles, the protective effect of DRB1*13:02 (P uncorr =6.9x10 -5 , OR 0.45, 95%CI 0.30-0.66) appeared to be more striking as compared with other HLA-DRb1_13S encoding alleles. On the other hand, in the relapse analysis, HR was not much different between DRB1*13:02 and other DRb1_13S encoding alleles. This could suggest a possibility that the protective effect against relapse may be shared by HLA-DRb1_13S alleles, while protection from disease occurrence may be mainly ascribed to DRB1*13:02. Further studies with larger sample sizes are needed to validate this hypothesis.
When the combined effect of DQA1*03:02 and DRb1_13S was examined, time to relapse was significantly shorter in the highest risk combination (DQA1*03:02 positive and DRb1_13S negative) when compared with the lowest risk combination (DQA1*03:02 negative and DRb1_13S positive), even after correction for multiple testing (Q=0.033). The HR was 4.02, which was higher than that for HLA-DPB1*04:01 in Dutch and American studies (17,18). In view of our results, it could be hypothesized that susceptibility alleles other than HLA may also be associated with the risk of MPO-AAV relapse. To date, no gene other than HLA has been established as a susceptibility gene for MPO-AAV in Asian populations. In a European GWAS, PTPN22 rs2476601 was reported to be associated with GPA/PR3-AAV and MPA/MPO-AAV (10); however, the risk allele of PTPN22 is almost absent in Asian populations. Recently, the association of a BACH2 variant was identified using target sequencing (31). According to the G e n o m e A g g r e g a t i o n D a t a b as e ( g n o m A D ) ( h t t p s : / / gnomad.broadinstitute.org/) (32), the risk allele in the BACH2 gene is not detected in East Asian populations. In a European population, it was reported that -463 A allele (rs2333227) in the MPO gene, which is associated with lower expression of MPO, was also associated with the risk of relapse in patients with MPO-AAV (33), which needs to be replicated in independent studies.
This study had several limitations. AAV is a rare disease, with an annual incidence of 22.6/million people in Japan (2). Because of the limited sample size and lack of independent replication cohort, it should be emphasized that this study is of an exploratory nature. For example, the attenuation of a trend toward relapse in DRB1*08:02 after conditioning (Table 5) could possibly be caused by insufficient detection power (0.589) due to the lower carrier frequencies of DRB1*08:02 compared to other alleles (Table 4). In fact, the HR of DRB1*08:02 remained >2.0 even after conditioning ( T a b l e 5 ) . F u r t h e r s t u d i e s a r e r e q u i r e d t o d r a w a definitive conclusions.
Additionally, because this study was based on observational cohort studies conducted between 2009 and 2016 (19,20), the potential effects of currently available treatments such as rituximab and avacopan could not be addressed. In this study, relapse was defined as the recurrence or new onset of clinical signs and symptoms attributable to active vasculitis, similar to previous studies on Japanese patients (6,23). Although some previous studies employed strict definition, including a requirement for treatment escalation (17), we did not use this definition, because we thought that whether treatment is reinforced or not partly depends on each physician's decision, and might possibly cause a bias. Finally, the classification of AAV is based on the EMA algorithm (1), and the compatibility of our findings with the 2022 American College of Rheumatology/European Alliance of Associations for Rheumatology Classification Criteria (34)(35)(36) needs to be validated. These limitations should be addressed in future studies.

Conclusion
In the present study, we demonstrated that DRB1*09:01, DQA1*03:02 and DQB1*03:03, susceptibility alleles for MPO-AAV, are also nominally associated with the risk of relapse in MPO-AAV in the Japanese population. In addition, we found that the carriers of HLA-DRb1 allele with serine residue at position 13 are nominally associated with lower risk of relapse. By combining these two factors, a statistically significant difference in the time to relapse was observed. These findings may be relevant in detecting individuals at high risk for AAV relapse of after remission, and adjusting treatment.

Ethics statement
The studies involving human participants were reviewed and approved by The Faculty of Medicine ethics committee, University of Tsukuba, as well as by all the institutes involved in the recruitment of subjects, as listed in the Materials and Methods. The patients/participants provided their written informed consent to participate in this study.
Committee of the Ministry of Health, Labour, and Welfare for Intractable Vasculitis (JPVAS) and Research Committee of Intractable Renal Disease of the Ministry of Health, Labour, and Welfare of Japan for recruiting the patients and collecting clinical information. We would like to thank Editage (www.editage.com) for English language editing. A preprint of the manuscript is a v a i l a b l e a t t h e m e d R x i v p r e p r i n t s e r v e r ( h t t p s : / / www.medrxiv.org/content/10.1101/2022.12.28.22283983v2) (37).