HLA‐A*11:01 Identifies Patients at Risk for Persistent Viraemia and Advanced Fibrosis in HDV Infection Authors: Maria Francesca Cortese, Adriana Palom, Ariadna Rando Segura, Francesc Rudilla, Emma Enrich Rande, Arnau Puig Lopez, David Tabernero Caellas, Judit Vico Romero, Juan Carlos Ruiz‐Cobo, Mar Riveiro‐Barciela, Maria José Herrero, Maria Buti Published: 2026 ABSTRACT ABSTRACTBackgroundHepatitis delta virus (HDV) infection causes the most severe form of viral hepatitis, with highly heterogeneous outcomes ranging from viral control to persistent viraemia and progressive liver disease. The Human Leukocyte Antigen (HLA) system plays a central role in antiviral immunity, yet its contribution to HDV disease remains poorly defined. We aimed to identify HLA genetic polymorphisms associated with virological control and disease severity.MethodsHigh‐resolution genotyping of classical (A, B, C, DPB1, DQB1, DRB1) and non‐classical (E, G) HLA was performed in 72 anti‐HDV‐positive patients. A group of healthy donors (N = 150) was used as a reference. Patients were stratified by longitudinal HDV RNA status into viraemic (persistent viraemia, N = 36) and non‐viraemic (undetectable viraemia, N = 36) groups. Clinical data, including liver fibrosis stage, were collected. T cell receptor (TCR) β‐chain sequencing was performed in isolated CD3+ T cells from viraemic patients and compared to a group of healthy donors (HD; N = 44).ResultsThe HLA‐A*11:01 allele was exclusively observed in viraemic patients (p‐value < 0.001, FDR = 0.02; allele frequency = 0.15). Multivariate analysis confirmed its independent association with persistent viraemia and advanced fibrosis. Notably, its frequency did not differ between our cohort and the overall reference population. Furthermore, viraemic HLA‐A*11:01 carriers exhibited significantly higher TCRβ chain clonality, consistent with a restricted and potentially ineffective antiviral T‐cell response.Conclusion HLA‐A*11:01 identifies a subgroup of HDV patients at high risk of persistent viral replication and advanced fibrosis. The integration of HLA genotyping into risk stratification algorithms may help to early identify patients requiring intensified monitoring and timely therapeutic intervention. Introduction Hepatitis delta virus (HDV) infection represents the most severe form of chronic viral hepatitis, significantly accelerating liver disease progression compared to Hepatitis B virus (HBV) monoinfection [1]. At diagnosis, a significant proportion of patients already present advanced liver fibrosis, placing them at high risk for liver disease complications including liver decompensation, hepatocellular carcinoma, and liver transplantation [2]. Among the prognostic factors, persistent HDV viraemia has consistently emerged as a key determinant of adverse clinical outcomes, being strongly associated with disease progression and an increased risk of liver‐related events [3, 4]. Despite its clinical severity, chronic hepatitis D is characterised by a highly heterogeneous disease course. Both the natural history of infection and the response to currently available therapies vary widely across patients. Treatment with pegylated interferon‐α (pegIFN‐α) achieves sustained virological response in approximately 30% of patients [5], while bulevertide achieves HDV RNA suppression in 20%–50% of patients after 144 weeks of therapy, with sustained off‐treatment responses of approximately 20% of patients [6]. Interestingly, a proportion of untreated individuals can spontaneously achieve long‐term viral control [1]. Together, these observations underscore the profound interindividual variability in HDV infection and highlight a critical gap in our understanding of the mechanisms driving divergent clinical trajectories. Identifying predictive factors of disease progression and treatment response therefore remains a major unmet clinical need, essential for improving risk stratification and enabling personalised management strategies. Host adaptive immune response plays a crucial role in controlling chronic viral infections. Its activation is the result of the interplay between antigen‐presenting cells that expose the viral epitopes through the Human Leukocyte Antigen (HLA) system, and their recognition by the T cell receptor (TCR) on T cells. Notably, HLA genes are highly polymorphic. Each allele presents a specific set of epitopes determined by its anchor motifs and residue preferences. This molecular diversity may be a primary driver of the varied immune responses seen among patients, establishing the HLA as a critical genetic determinant of disease susceptibility, progression, and viral control, especially in the context of chronic viral infections [7, 8]. Compelling evidence from other viral diseases supports this concept. In HIV infection, specific HLA Class I alleles such as HLA‐B*27 and HLA‐B*57 are strongly associated with effective viral control [9]. In chronic HBV infection, distinct HLA Class II variants have been linked to disease progression and the development of HCC, potentially through altered antigen presentation and suboptimal immune activation [10, 11]. In contrast, non‐classical HLA molecules, including HLA‐E and HLA‐G, display limited polymorphism and primarily exert immunomodulatory and tolerogenic functions. Variations within these loci have also been associated with differential outcomes in HBV infection, further highlighting the complexity of HLA‐mediated immune regulation [12, 13]. Beyond antigen presentation, the diversity and functionality of the TCR repertoire represent another critical layer of antiviral immunity. TCR diversity is generated through somatic recombination and is dynamically shaped by antigen exposure. Effective immune control requires a broad and adaptable TCR repertoire capable of recognising diverse viral epitopes. Conversely, chronic antigen stimulation may lead to clonal expansion, repertoire skewing, and T‐cell exhaustion, ultimately impairing viral clearance. In HBV infection, reduced TCR diversity has been associated with more severe disease, while successful antiviral therapy has been linked to the emergence of new TCR clonotypes, suggesting that restoration of repertoire diversity may contribute to viral control [14, 15]. Despite the central role of adaptive immunity, the contribution of HLA polymorphisms and TCR repertoire dynamics to HDV infection remains poorly defined. Given the marked heterogeneity in clinical outcomes and treatment responses, a better understanding of host immunogenetic determinants could provide critical insights into disease pathogenesis and inform clinical decision‐making. In this proof‐of‐concept study, we aimed to investigate the association between HLA genetic variation and disease course in chronic hepatitis D, and to explore the functional effects on immune response through the analysis of the TCRβ repertoire. By integrating host genetic and virological data, we sought to identify biomarkers of viral persistence and disease progression with direct potential clinical and translational relevance. Patients and Methods Patient Population This was a retrospective–prospective cohort studyentation using the NGSgo kipositive hepatitis B surface antigen (HBsAg) and anti‐HDV antibodies who attended the outpatient clinic at Vall d'Hebron Hospital between August 2017 and February 2024. Inclusion criteria were: Age ≥ 18 years, chronic hepatitis D‐defined as the presence of anti‐HDV antibodies for more than 6 months, ability to read, understand and provide informed consent. Patients with active Human immunodeficiency virus (HIV) and/or Hepatitis C virus (HCV) co‐infection or other known immune‐mediated disorders were excluded. Participants were classified into two groups based on HDV‐RNA status: Viraemic patients, defined as having detectable HDV‐RNA in more than 3 consecutive samples collected 6 months apart, and non‐viraemic patients, defined as having undetectable HDV‐RNA in at least three consecutive samples. Demographic, laboratory, and clinical data were collected from medical records. Demographic variables included sex, age, and ethnicity. Laboratory parameters included alanine aminotransferase (ALT), serological markers, and virological parameters. Serological testing included HBsAg (Elecsys HBsAg II, Roche Diagnostics, IN, USA), HBeAg (Elecsys HBeAg), anti‐HBe (Elecsys Anti‐HBe), anti‐HCV (Elecsys Anti‐HCV II), HIV Ag/Ab combination assay (Elecsys HIV Duo), and anti‐HDV (LIAISON XL MUREX Anti‐HDV assay, Diasorin, Saluggia, Italy). Virological parameters included quantitative HBV‐DNA (cobas HBV, Roche Diagnostics, Mannheim, Germany) and HDV‐RNA, quantified using an in‐house RT‐qPCR assay calibrated against the WHO International Standard for HDV‐RNA with a limit of detection of 57.5 IU/mL and a lower limit of quantification of 575 IU/mL [16]. Liver cirrhosis was defined by histology (METAVIR F4), non‐invasive assessment (liver elastography > 12.5 kPa [17]), or by clinical criteria, including nodular liver surface and signs of clinically significant portal hypertension (collateral circulation, splenomegaly, and thrombocytopenia). Advanced fibrosis was defined by histology (METAVIR F3 and/or bridging fibrosis) or liver elastography > 10.2 kPa [18]. Prior treatment for chronic hepatitis D was also recorded (including pegylated interferon alpha, pegylated interferon lambda, and lonafarnib). Patients provided informed consent for the use of their clinical records and for the extraction of an additional blood sample for HLA and TCR profiling. This study was approved by the Vall d'Hebron Ethics Committee (PR(AG)627/2023; 15 July 2024). To ensure that the HLA allele distribution reflected the background population, a reference group (N = 150) was randomly selected from a larger cohort of healthy bone marrow donors with prior high‐resolution HLA typing, following approval by the Catalan Transplant Organisation (OCATT). High‐Resolution HLA Genotyping by Next‐Generation Sequencing Briefly, genomic DNA was extracted from peripheral blood using the QIAsymphony DNA Mini Kit (Qiagen, Hilden, Germany). DNA amplification of HLA‐A,‐B,‐C,‐DRB1,‐DQB1,‐DPB1,‐E and‐G was carried out using an in‐house multiplex PCR, and library preparation was performed by fragmentation using the NGSgo kit (GenDX, Utrecht, Netherlands). Paired‐end sequencing was carried out on MiSeq and NextSeq 1000 instruments (Illumina, San Diego, CA, USA), using a MiSeq Reagent Kit v2 (300 cycles) and NextSeq 1000 Reagent Kit v2.5 (300 cycles) (Illumina). Sequence data were analysed using NGSengine software (GenDx) and the updated IPD‐IMGT/HLA Database. HLA‐G 3′ UTR variants rs66554220, rs1063320, rs9380142, and rs1610696 were analysed using the QIAGEN CLC Genomics Workbench (QIAGEN). All procedures were performed as previously described by Gómez EM et al. [19]. TCR Repertoire Analysis CD3+ T cells were isolated from frozen peripheral PBMCs through immunomagnetic positive selection following the manufacturer's protocol (EasySep, Stemcell Technologies, Vancouver, Canada). Genomic DNA was isolated using the QIAamp DNA Blood Mini Kit (Qiagen, Düsseldorf, Germany). Amplification and sequencing of the TCR beta chain (TCRβ) variable domain were performed as previously described [20]. Briefly, multiplex PCR was used to amplify variable (TRBV), diversity (TRBD), and joining (TRBJ) gene rearrangements of the TCRβ locus. Around 100,000 T cells were amplified per sample, which were extrapolated based on the DNA concentration and the weight of a diploid genome (~6.6 pg). Libraries were generated in a second round of PCR, purified, and sequenced on a MiSeq system using a 500‐cycle MiSeq Reagent Kit v2 (Illumina). Sequence data were processed using MiXCR and VDJtools, and only productive clonotypes were included in downstream analyses. TCRβ repertoire diversity was evaluated using observed diversity, the D50 index, and the Chao1 estimator [21]. Observed diversity represents the number of unique clonotypes based on their nucleotide sequence. The D50 index shows the number of unique clonotypes accounting for 50% of the repertoire (lower values indicate higher clonality). In contrast, the Chao1 estimator reflects the richness of the repertoire by showing the contribution of rare clonotypes in overall TCR diversity (lower values indicate fewer rare, non‐expanded clonotypes). To ensure the comparison across samples, diversity metrics were normalised based on the number of reads obtained in each sample. The results from viraemic patients were compared to another cohort of 44 healthy donors (HD) with a previously sequenced TCR profile [20]. Statistical Analysis Allele frequencies were determined by calculating the proportion of the occurrence of each allele relative to the total alleles present in each patient or reference group. The allele frequencies between the groups were compared by applying a Fisher's exact test with FDR correction. For alleles showing different frequencies (unadjusted p‐value < 0.1), the percentage of carriers was determined. This was defined as the percentage of individuals from each group carrying at least one copy of the specific allele. Groups were compared using Fisher's exact test. The association between specific HLA alleles and HDV infection outcomes (persistent viraemia or high degree of fibrosis) was assessed by applying a Penalised Maximum Likelihood (PML) model from the logistf package (R version 4.2.3). Only the alleles showing a different proportion of carriers (p < 0.05 by Fisher's exact test) were included in univariate and multivariate analyses. Since not all viraemic patients progress to advanced liver fibrosis, univariate and multivariate analyses were also performed by grouping patients according to the co‐occurrence of persistent viraemia and advanced fibrosis (F3–F4). To perform internal validation, we used 1000 bootstrap resamples of the multivariate model to calculate AUC, optimism‐corrected AUC, the calibration slope, and calibration intercept. The diversity of TCR repertoire was compared through the Mann–Whitney test or Kruskal‐Wallis test followed by post hoc Dunn test for multiple comparisons (PMCMRplus package). Results Study Population Seventy‐two anti‐HDV positive patients were enrolled in the study. The baseline characteristics are shown in Table 1, stratified according to HDV‐RNA status (viraemics vs. non‐viraemics). Overall, more than half (58%) were male, with a median age of 50 years. Most patients had Caucasian (61%) or African origin (29%). TABLE 1Demographic and clinical features of the anti‐HDV‐positive patients.Total (N = 72)Viraemics (N = 36)Non‐viraemics (N = 36) p Age, years 50 (39–56)46 (36–54)51 (42–58)0.072 Male Sex 42 (58%)20 (55%)22 (61%)0.531 Ethnicity Caucasian44 (61%)25 (69%)19 (53%)0.390Asian6 (8%)3 (8%)3 (8%)African21 (29%)8 (22%)13 (36%)Middle East1 (1%)01 (3%) ALT (U/L) 38 (25–77)63 (44–97)26 (18–38)0.018 ALT>UNL 34 (47%)26 (72%)8 (22%)< 0.001 HBeAg‐positive 8 (11%)7 (19%)1 (3%)0.033 HBsAg (log IU/mL) Median (IQR) 3.6 (3.1–4.1)4.0 (3.5–4.3)3.3 (2.5–3.6)< 0.001 HBV‐DNA detectable 13 (18%)1 (3%)12 (33%)< 0.001 HDV‐RNA (log IU/ml) —4.5 (3.7–5.4)—— Advanced fibrosis (F3–F4) 34 (47%)24 (67%)10 (28%)< 0.001 Liver cirrhosis 25 (35%)21 (58%)4 (11%)< 0.001 HBV NUC treatment 44 (61%)32 (89%)12 (33%)< 0.001 Duration of infection, years Median (IQR) 3.6 (0.2–9.5)6.4 (0.2–10.3)3.1 (0.1–7.8)0.198 Prior treatment N (%)29 (40%)21 (58%)8 (22%) N Response6 (8%)n/a6 (17%)0.004 Note: Numeric variables are represented as median (IQR), whereas categorical variables are presented as the number of patients (N) and the percentage related to the group. ALT UNL (upper normal limit) of 35 U/L for women and 50 U/L for men.Abbreviations: ALT, alanine aminotransferase; HBeAg, HBV e antigen; HDV‐RNA in log IU/mL; NUC, nucleos(t)ide analogues. Demographic and clinical features of the anti‐HDV‐positive patients. HBsAg (log IU/mL) Median (IQR) HDV‐RNA (log IU/ml) Duration of infection, years Median (IQR) Note: Numeric variables are represented as median (IQR), whereas categorical variables are presented as the number of patients (N) and the percentage related to the group. ALT UNL (upper normal limit) of 35 U/L for women and 50 U/L for men. Abbreviations: ALT, alanine aminotransferase; HBeAg, HBV e antigen; HDV‐RNA in log IU/mL; NUC, nucleos(t)ide analogues. Around half of non‐viraemics (12/36) presented at least one previous documented detectable HDV viraemia sample. Notably, in all non‐viraemic patient the HDV RNA consistently remained undetectable during follow‐up. Viraemic patients had significantly higher ALT and HBsAg levels, were more often HBeAg‐positive, presented more advanced fibrosis and liver cirrhosis, and were more frequently treated with nucleos(t)ide analogues. Twenty‐one of the viraemic patients received prior treatment but failed to achieve a virological response. Conversely, 8 non‐viraemic patients received previous treatment with pegIFN: Two of them were treated years before the study and became undetectable during off‐treatment follow‐up, whereas 6 presented an on‐treatment response. Notably, HBV and HDV genotyping was not performed in this study. However, based on our previous studies, genotype 1 for HDV and genotypes D and A are dominant in our patients' population [22, 23, 24]. HLA Typing in HDV Patients Compared to the Reference Group To assess potential population stratification, HLA allele frequencies in the overall anti‐HDV positive cohort (including both viraemic and non‐viraemic patients) were compared with those observed in a reference group of healthy donors (N = 150, Figure S1). We observed 35 alleles with significantly different allele frequencies (AF) between patients and the reference group (p < 0.05, Figure 1). Of these, 24 were mapped to Class I loci, with 10 alleles for the HLA‐A locus, 8 for HLA‐B, and 6 for HLA‐C. Differences in Class II loci were mainly observed at DQB1 (n = 4) and DRB1 (n = 5), while two involved the DPB1 locus. HLA alleles showing differential frequencies between HDV patients and the reference group. Allele frequencies of Class I (A) and Class II HLA (B) in all the HDV patients (N = 72) and the reference group (N = 150). Panel (C) shows the allele frequencies of the differentially prevalent alleles considering only HDV patients of Caucasian ethnicity (N = 44) and the reference group (N = 150). p‐values and FDR‐adjusted p‐values obtained using Fisher's exact test are reported. A more detailed analysis of Class I loci (Figure 1 panel A) showed that the reference group had higher frequencies of the A*02:01 allele (AF: 0.25, OR: 0.49, 95% CI: 0.27–0.86, p = 0.001, FDR = 0.04) and the A*29:02 allele (AF: 0.08, OR: 0.16, 95% CI: 0.02–0.64, p < 0.001, FDR = 0.03). In contrast, HDV patients exhibited several alleles at the HLA‐A locus, with some of them exclusively observed in these patients at low frequency (A*23:01, A*03:01, A*30:01, A*68:02, A*30:02, A*33:01, A*74:01, A*34:02, with frequencies ranging from 0.02 to 0.09; Table S1). Anti‐HDV positive patients were also enriched for the following HLA‐B alleles: B*53:01 (OR: 19.81, 95% CI: 2.70–872.68, p < 0.001, FDR = 0.01), B*37:01 (OR: 10.69, 95% CI: 1.18–508.99, p = 0.02, FDR = 0.19), B*52:01 (OR: 5.04, 95% CI: 1.13–30.65, p = 0.02, FDR = 0.19), and B*35:03 (OR: 2.9, 95% CI: 1.10–8.09, p = 0.02, FDR = 0.19) with frequencies ranging from 0.01 to 0.08 (Figure 1, panel A; Table S1), as well as C*06:02 (OR: 2,66, 95% CI: 1.32–5.41, p < 0.01, FDR = 0.08), and C*04:01 (OR: 1.91, 95% CI: 1.04–3.49, p = 0.03, FDR = 0.19). Four additional HLA‐B (B*42:01, B*13:01, B*15:03, and B*39:10) and two HLA‐C alleles (C*02:10, C*07:06) were detected only in anti‐HDV positive patients at very low frequencies (between 0.01 and 0.02, Table S1). For Class II loci (Figure 1, panel B), few alleles were exclusively observed in anti‐HDV positive patients. In this latest, higher frequencies related to the reference were seen for the alleles DPB1*01:01 (OR: 18.74, 95% CI: 5.54–99.02, p < 0.001, FDR < 0.001; AF: 0.16 in HDV and 0.01 in reference), DQB1*05:01 (OR: 2.25, 95% CI: 1.23–4.12, p = 0.01, FDR = 0.03; AF: 0.19 and 0.1 respectively), DQB1*03:19 (OR: 8.50, 95% CI: 0.83–421.11, p = 0.04, FDR = 0.17, AF: 0.03 and 0.003), DRB1*10:01 (OR: 13.47, 95% CI: 2.94–125.72, p < 0.001, FDR < 0.001; AF: 0.08 and 0.01, respectively), DRB1*15:03 (OR: 12.93, 95% CI: 1.55–598.05, p = 0.01, FDR = 0.08, AF: 0.04 and 0.003), and DRB1*01:02 (OR: 3.77, 95% CI: 0.94–17.85, p = 0.04, FDR = 0.38, AF: 0.05 and 0.01). Being the reference group enriched for individuals with Caucasian ancestry, the analysis was repeated by restricting the anti‐HDV positive cohort to Caucasian patients (N = 44, Figure 1 panel C). After ethnicity adjustment, a few low‐frequency patient‐exclusive alleles were observed (A*03:01, A*23:01, and A*30:02 with frequencies of 0.09, 0.03, and 0.03, respectively). Importantly, the enrichment of B*35:03 (OR: 4.6, 95% CI: 1.66–13.04, p < 0.01, FDR = 0.06), C*04:01 (OR: 2.23, 95% CI: 1.10–4.39, p = 0.02, FDR = 0.25), and DPB1*01:01 (OR: 8.49, 95% CI: 1.89–52.03, p < 0.01, FDR = 0.02) remained evident, thus suggesting that the observed enrichment may be not strictly dependent on population stratification. No relevant differences were observed in non‐classical HLA loci. The allele frequencies of HLA‐G and HLA‐E were similar between patients and the reference group (Figure S1), and no differences were detected in the analysed HLA‐G 3′UTR variants (Figure S2). HLA Alleles Associated With Persistent Viraemia To identify HLA alleles potentially associated with persistent HDV viraemia, allele frequencies were compared between viraemics and non‐viraemics. Three alleles were detected in both sub‐groups and showed statistically different frequencies (p < 0.05, Figure 2, panel A): C*07:01 (OR: 3.49, 95% CI: 1.01–15.48, p = 0.04, FDR = 0.64), DRB1*01:01 (OR: 7.13, 95% CI: 0.88–328.99, p = 0.6, FDR = 1) and G*01:01 (OR 2.35, 95% CI: 1.13–4.97, p = 0.02, FDR = 0.17). HLA alleles in HDV non‐viraemic and viraemic patients. (A‐C) Forest plot showing the odds ratio and 95% CI when comparing the allele frequencies of HLA‐C (panel A), HLA‐DRB1 (panel B) and Non‐classical HLA (HLA‐G and HLA‐E; panel C) between HDV patients who control viral replication (non‐viraemic, N = 36) and those with stable detectable HDV‐RNA (viraemic, N = 36). The dot shows the odds ratio value. The 95% CI is reported with a line (black for those alleles with a Fisher p‐value < 0.1). (D) Allele frequencies of the alleles showing more difference in anti‐HDV positive patient subgroups. The Fisher‐derived p‐value (with or without FDR correction) is reported. (D) Percentage of HDV patients presenting the alleles with different frequencies in HDV viraemic and non‐viraemic groups. The p‐value obtained using Fisher's exact test is also reported. In addition, some HLA alleles (Figure 2, panel B) were observed only in viraemic patients. This was the case for A*11:01 (AF: 0.15, p‐value < 0.001, FDR = 0.02) and C*12:02 (AF: 0.08, p = 0.03, FDR = 0.64). Conversely, the DRB1*13:03 was exclusively observed in non‐viraemic patients (AF: 0.06, p = 0.05, FDR = 1). To further evaluate these differences, the proportion of HDV patients carrying these alleles was also assessed (Figure 2, panel C). C*07:01 and DRB1*01:01 were no longer statistically significant as they were observed in 29.7% (n = 11) and 13.5% (n = 5) of viraemic patients, respectively, compared with 17.1% and 2.8% of non‐viraemic patients (p > 0.1). Notably, the A*11:01 and C*12:02 were exclusively observed in viraemic patients, respectively, in 27% (10/36, p < 0.001) and 14% (5/36; p = 0.05) of them. To determine whether the presence of these two alleles was independently associated with persistent viraemia in anti‐HDV positive patients, univariate and multivariate analyses were performed, including clinical and demographic variables known to be potentially related to HDV replication, such as ethnicity, advanced fibrosis, elevated ALT levels, and prior treatment. As expected, elevated ALT levels, advanced fibrosis (F3–F4) and prior treatment were associated with persistent viraemia (Table 2). Importantly, the presence of A*11:01 or C*12:02 was also related to persistent viraemia (p = 0.003 and p = 0.02, respectively). After adjusting for all relevant variables, however, the A*11:01 remained independently associated with the outcome (p = 0.03). TABLE 2Univariate and multivariate analysis of variables associated with persistent viraemia (detectable HDV‐RNA).Univariate (persistent viraemia)Multivariate (persistent viraemia)OR95% CI p OR95% CI p Sex Male 0.80.3–2.00.6 Ethnicity Caucasian Asian African 1.9 1 0.5 0.8–5.2 0.2–5.0 0.2–1.4 0.1 1.0 0.2 Prior treatment Yes 4.61.7–13.3 0.002 2.90.8–10.50.07 ALT level >ULN 8.53.1–25.4 < 0.001 5.81.9–19.7 0.003 Fibrosis F3 or F4 4.91.9–13.8 0.002 1.90.6–6.40.4 C*12:02 Yes 12.71.3–1697 0.02 3.50.2–4890.4 A*11:01 Yes 28.93.4–3780 < 0.001 14.11.1–2072 0.03 Note: Results were obtained using a Firth‐penalised logistic regression model. Statistically significant p‐values are reported in bold.Abbreviations: 95% confidence interval; OR, Odds ratio; 95% CI, p, p‐value. Univariate and multivariate analysis of variables associated with persistent viraemia (detectable HDV‐RNA). Sex Male Ethnicity Caucasian Asian African 1.9 1 0.5 0.8–5.2 0.2–5.0 0.2–1.4 0.1 1.0 0.2 Prior treatment Yes ALT level >ULN Fibrosis F3 or F4 C*12:02 Yes A*11:01 Yes Note: Results were obtained using a Firth‐penalised logistic regression model. Statistically significant p‐values are reported in bold. Abbreviations: 95% confidence interval; OR, Odds ratio; 95% CI, p, p‐value. As expected, persistent viraemia was independently associated with elevated ALT levels, consistent with chronic inflammation driven by viral replication. The model was also retested by using 1000 bootstrap resamples. The model demonstrated excellent predictive performance, with an AUC of 0.89 and an optimism‐corrected AUC of 0.86, accompanied by a calibration slope of 1.0 and a calibration intercept of 0.13. Additionally, a sub‐analysis comparing HLA alleles between treatment responders and non‐responders was also performed, showing no significant differences (data not shown). HLA Alleles and Liver Fibrosis Because chronic inflammation is a major driver of liver fibrosis and since advanced liver fibrosis may be observed in both viraemic and non‐viraemic groups, HDV patients were stratified by fibrosis stage into advanced fibrosis (F3–F4, N = 34) and mild fibrosis (F1–F2, N = 38). Patients with advanced liver fibrosis showed higher frequencies of A*11:01 (OR: 3.68, 95% CI: 0.87–22.07, p = 0.07, FDR = 1), A*02:01 (OR: 2.3, 95% CI: 0.79–7.35, p‐value = 0.09, FDR = 0.88) and DPB1*17:01 (OR 4.88, 95% CI: 0.93–48.89, p = 0.05, FDR = 0.42), whereas DPB1*01:01 predominated in the mild fibrosis group (OR: 0.18, 95% CI: 0.04–0.57, p < 0.001, FDR = 0.02); Figure 3 panels A, B and C. HLA alleles in HDV patients with severe (F3 or F4) or mild fibrosis (F0 to F2). (A and B) Forest plot showing the odds ratio and 95% CI when comparing the allele frequencies of HLA‐A (panel A) and HLA‐DPB1 (panel B) genes between HDV patients stratified according to fibrosis status. The dot shows the odds ratio value. The 95% CI is reported by a line (black for alleles with a Fisher p‐value < 0.1). (C) Allele frequencies of the alleles showing more differences in anti‐HDV‐positive patient subgroups. The Fisher‐derived p‐value is reported. (D) Percentage of anti‐HDV‐positive patients presenting the alleles with the greatest difference in HDV patients stratified according to their fibrosis state. The p‐value obtained by Fisher's exact test is also reported. These results were consistent when analysing the proportion of allele carriers (Figure 3, panel D). Among patients with advanced fibrosis, 23.5% carried A*11:01 (8/34), vs. 5.3% in the mild fibrosis group (p = 0.04). Similarly, 20.6% of patients with advanced fibrosis were carriers of DPB1*17:01 vs. 5.3% in the mild fibrosis group (p = 0.07). In contrast, the DPB1*01:01 was more frequent in the mild fibrotic group, with 34.2% of carriers (13/38) vs. 11.7% in the advanced fibrosis group (p = 0.03). To determine whether the presence of these alleles may be associated with fibrosis severity, univariate and multivariate analyses were performed, including variables that could potentially influence liver disease progression. Analysis was restricted to HLA alleles showing statistically significant differences in prevalence among patients. Notably, several variables were associated with advanced fibrosis in univariate analysis, such as Caucasian ethnicity, prior treatment, detectable HDV‐RNA, and high ALT (OR > 1, p < 0.05; Table 3). In contrast, African ethnicity was negatively associated with this outcome (OR = 0.3). Regarding HLA alleles, the presence of A*11:01 was associated with advanced fibrosis (OR:4.7, p = 0.03), while the presence of DPB1*01:01 showed a negative relation (OR: 0.3, p = 0.03), being primarily observed in patients with mild fibrosis. TABLE 3Univariate and multivariate analysis of the variables associated with advanced fibrosis (F3–F4).Univariate (Advanced Fibrosis) p Multivariate (Advanced Fibrosis) p OR95% CIOR95% CI Sex Male 1.50.6–3.80.4 Ethnicity Caucasian Asian African 2.7 1.1 0.3 1.0–7.3 0.2–5.7 0.1–0.9 0.04 0.9 0.04 1.0 0.6 0.2–6.2 0.1–4.7 1 0.6 Prior treatment Yes 5.72.1–16.6 < 0.001 2.80.9–9.10.08 HDV‐RNA detectable 4.91.9–13.8 < 0.001 1.90.5–7.00.3 ALT >ULN 3.81.5–10.3 0.05 2.20.7–7.50.2 A*11:01 Yes 4.71.2–26.3 0.03 1.80.3–12.10.5 DPB1*01:01 Yes 0.30.08–0.8 0.03 0.50.1–1.80.3 Note: Results were obtained through a Firth‐penalised logistic regression model. Statistically significant p‐values are reported in bold.Abbreviations: 95% CI, 95% confidence interval; OR, Odds ratio; p, p‐value. Univariate and multivariate analysis of the variables associated with advanced fibrosis (F3–F4). Sex Male Ethnicity Caucasian Asian African 2.7 1.1 0.3 1.0–7.3 0.2–5.7 0.1–0.9 0.04 0.9 0.04 1.0 0.6 0.2–6.2 0.1–4.7 1 0.6 Prior treatment Yes HDV‐RNA detectable ALT >ULN A*11:01 Yes DPB1*01:01 Yes Note: Results were obtained through a Firth‐penalised logistic regression model. Statistically significant p‐values are reported in bold. Abbreviations: 95% CI, 95% confidence interval; OR, Odds ratio; p, p‐value. After adjusting for all predictors, none of the variables were independently associated with advanced fibrosis. Although prior treatment showed a trend toward association, it did not reach statistical significance (OR: 2.8, 95% CI: 0.9–9.1, p = 0.08). Notably, although A*11:01 was associated in univariate analysis with fibrosis, it did not retain statistical significance after multivariate adjustment, potentially masked by the presence of persistent detectable HDV‐RNA (Pearson's Chi‐squared = 0.002). As a proof, this allele was independently associated with advanced fibrosis when HDV‐RNA and ALT states were not considered (OR: 7.2, 95% CI: 1.5–55.7, p = 0.01). HLA‐A*11:01 and Advanced Fibrosis in Viraemic Patients Patients were also re‐stratified according to the co‐occurrence of advanced fibrosis with persistent HDV‐RNA as the outcome (detectable HDV‐RNA with F3–F4 fibrosis, N = 24 vs. the other 48 HDV patients). Notably, all the HLA alleles previously associated in univariate or multivariate analyses with persistent HDV‐RNA or fibrosis were included. Thirty‐three per cent of the patients with both persistent viraemia and advanced fibrosis carried A*11:01 (8/24, Fisher's exact test, p = 0.002). In contrast, no significant differences were observed when considering C*12:02 and DPB1*01:01 alleles (p‐value > 0.05; Figure 4). Percentage of patients carrying the most relevant alleles based on both their HDV‐RNA and fibrosis status. p‐values from Fisher's exact test are reported. The number of patients carrying each allele is also reported. To assess whether an association between A*11:01 and the outcome exists, a univariate and multivariate analysis was performed. Again, all variables potentially associated with persistent viraemia or advanced fibrosis were included in the analysis (Table 4). TABLE 4Univariate and multivariate analysis of all the variables and their association with advanced fibrosis in persistent viraemic patients.Univariate (F3–F4 in viraemic)Multivariate (F3–F4 in viraemic)OR95% CI p OR95% CI p Sex Male 1.60.6–4.60.3 Ethnicity Caucasian Asian African 3.3 0.5 0.4 1.1–10.6 0.05–2.7 0.1–1.2 0.03 0.4 0.1 2.30.7–8.20.2 Prior treatment Yes 5.11.9–15.0 0.001 3.10.9–10.20.06 ALT >ULN 5.61.9–17.4 < 0.001 3.91.2–13.7 0.02 A*11:01 Yes 9.62.3–54.8 0.001 7.71.5–55.4 0.01 Note: Results were obtained through a Firth‐penalised logistic regression model. Statistically significant p‐values are reported in bold.Abbreviations: 95% CI, 95% confidence interval; OR, Odds ratio; p, p‐value. Univariate and multivariate analysis of all the variables and their association with advanced fibrosis in persistent viraemic patients. Sex Male Ethnicity Caucasian Asian African 3.3 0.5 0.4 1.1–10.6 0.05–2.7 0.1–1.2 0.03 0.4 0.1 Prior treatment Yes ALT >ULN A*11:01 Yes Note: Results were obtained through a Firth‐penalised logistic regression model. Statistically significant p‐values are reported in bold. Abbreviations: 95% CI, 95% confidence interval; OR, Odds ratio; p, p‐value. In univariate analysis, Caucasian ethnicity, prior treatment, and high ALT levels were associated with the outcome (p < 0.05), along with the presence of A*11:01 (OR: 9.6, p = 0.001). After multivariate adjustment, however, most of the predictors lost significance. Although prior treatment showed a trend toward association with the outcome, only high ALT (OR: 3.9, 95% CI: 1.3–12.4, p = 0.01) and the presence of A*11:01 (OR: 5.1, 95% CI: 1.1–27.7, p = 0.03) remained independently associated with advanced fibrosis among HDV viraemic patients. Consistently, 33% (8/24) of the patients with persistent viraemia and advanced fibrosis carried the A*11:01 allele, and 76% of them (19/25 patients) presented elevated ALT levels. Internal validation of the multivariate model showed a good apparent AUC of 0.85 and a limited drop in discrimination after optimism correction (optimism‐corrected AUC = 0.82), with a calibration slope of 0.9 and a calibration intercept of −0.06. These results suggest a strong agreement between predicted and observed outcomes, with limited overfitting. Higher Clonality of the TCR Repertoire in HLA‐A*11:01 Carriers To explore how HLA‐A*11:01 may influence the immune response in viraemic patients, the β chain of the TCR repertoire was analysed by next‐generation sequencing. TCRβ sequences were obtained from 25 of the 37 viraemic patients and compared with a group of HD (N = 44) used as a control. When comparing the diversity of the TCRβ repertoire between viraemic patients and the HD group, the first exhibited lower diversity (1.4‐fold, p = 0.007, Figure 5, panel A). This trend was also confirmed when using the D50 and Chao1 indices, which were 1.6‐fold and 1.3‐fold lower, respectively, in viraemic patients compared with the HD group (p < 0.05, Figure 5, panels B and C). These findings were consistent with the presence of viral replication and the consequent clonal expansion of T cell clones. TCRβ repertoire in anti‐HDV viraemic patients. Diversity (A), D50 (B), and Chao1 estimate (C) in anti‐HDV viraemic patients (N = 25) and the HD group (N = 44). p‐value was obtained through the Wilcoxon test. Diversity (D), D50 (E), and Chao1 estimate (F) were compared between anti‐HDV viraemic patients grouped according to their HLA‐A allele (8 vs. 17). p‐values (Holm‐adjusted, using the Kruskal–Wallis test with Dunn's multiple‐comparison test or the two‐sample Wilcoxon test) are reported. The bar shows the median value, whereas the line shows the IQR. Notably, when viraemic patients were stratified according to the presence of A*11:01 (9 vs. 16), the carriers of this allele displayed a significantly reduced diversity in relation to the HD (p‐value and Holm adjusted‐p‐value < 0.05). This reduced diversity was also detected when A*11:01‐positive viraemic patients were compared with those carrying other HLA‐A alleles, although the difference did not reach statistical significance (p‐value and adjusted p‐value > 0.1, Figure 4, panel D). A similar pattern was observed for the D50 index. Although not supported by strong statistical significance, A*11:01 viraemic carriers presented the lowest value (2.3‐fold and 1.4‐fold reductions compared with HD and viraemic patients carrying other HLA alleles, respectively; Figure 4, panel E). This observation was further reinforced by the evidence of a reduced Chao1 index in A*11:01 carriers compared with other viraemic patients. Specifically, A*11:01 carriers were characterised by the lowest Chao1 value (p = 0.002 and p = 0.04 when compared respectively to HD and non‐A*11:01 carriers), although this statistical relevance was partially lost when adjusted for multiple comparisons using Holm correction (adjusted p‐values of 0.006 and 0.09, respectively; Figure 4, panel F). Overall, these data suggest that A*11:01 carriers exhibit a more restricted TCRβ repertoire characterised by greater expansion of selected clonotypes (low D50), potentially reflecting a narrower T cell response. Discussion In this study, HLA‐A*11:01 emerged as a potential genetic factor associated with persistent HDV viraemia. Approximately 27% of viraemic patients carried this allele, and its association remained independent of ethnicity, prior treatment, ALT elevation, or fibrosis stage. Persistent viraemia was strongly associated with elevated ALT and advanced fibrosis, underscoring liver injury driven by active ongoing viral replication. Although A*11:01 was initially associated with advanced fibrosis in univariate analysis, this association was lost in multivariate models. However, A*11:01 remained an independent predictor of the composite outcome (persistent viraemia plus advanced fibrosis), suggesting its contribution to fibrosis is largely mediated through sustained viral replication. The identification of host genetic factors that predict disease course could be very useful for early patient risk stratification and for enabling a more personalised approach to clinical management. Even so, very few data are available related to HDV infection, including regarding HLA polymorphisms. While previous studies identified HDV epitopes restricted by HLA‐A*02:01 or HLA‐B*27 [25, 26], the HLA‐A*11:01 has been poorly studied. Previous studies reported this allele at a frequency of 0.1 in HDV patients [27], whereas its frequency increased up to 0.12–0.14 when considering the A*11 allele group [28, 29]. Nevertheless, no relation between this allele and HDV clinical outcome has yet been depicted. In this study, the HLA‐A*11:01 observed in viraemic patients was similar to that reported in previous studies. It is worth noting that none of them were specifically designed to compare patients with persistent viraemia to those who were non‐viraemic, which limits the comparison between allele frequencies from different studies [27]. Notably, HLA‐A*11:01 is generally more frequent in Asian individuals (AF: 0.14) than in Caucasians (AF: 0.05), according to data from http://www.allelefrequencies.net/. This relation with ethnicity was not observed in our study, since only two viraemic anti‐HDV positive patients carrying this allele were of Asian origin. While HLA‐A*11 has been associated with persistent HCV infection in non‐Asian populations [30, 31], no relationship between this allele and different outcomes of chronic viral hepatitis has been reported in Asian cohorts. Specifically in our cohort, the frequency of HLA‐A*11:01 was similar between patients and the reference group (respectively 0.08 and 0.07), aligning closely with the frequency already reported in Catalan bone marrow donors (frequency of 0.06) [19]. In addition, HLA‐A*11:01 also emerged as a potential factor associated with advanced fibrosis. While significant in univariate analysis, this relationship was not maintained in multivariate analysis, likely due to other confounding variables, such as persistent HDV‐RNA detectability. Patients were re‐stratified using a composite outcome of persistent viraemia and advanced liver fibrosis. A significantly higher proportion of viraemic patients with advanced fibrosis carried the HLA‐A*11:01 allele, and this association remained significant in multivariate analysis. These findings suggest that the relation between HLA‐A*11:01 and liver fibrosis may be mediated, at least in part, by persistent viral replication. Notably, beyond viral hepatitis, HLA‐A*11 alleles were also associated with autoimmune diseases such as Type 1 diabetes or autoimmune hepatitis [32, 33], and severe COVID‐19 in Asian populations [34]. This association may also be dependent on its ability to present viral epitopes that show similarity with self‐antigens, as observed for SARS‐CoV‐2 infection [35]. Mechanistically, TCRβ repertoire analysis revealed that viraemic HLA‐A*11:01 carriers exhibited reduced diversity and higher clonality, with expansion of a narrow set of T‐cell clones. Similar patterns have been associated with expanded and activated T‐cell responses, as shown in COVID‐19 patients [36]. This restricted repertoire may limit effective recognition of HDV epitopes, impairing viral clearance and promoting chronic inflammation, a pattern observed in other viral infections such as chronic hepatitis C virus infection [37]. Alternatively, reduced diversity may result from less efficient viral peptide presentation, as suggested in severe COVID‐19 cases associated with A*11:01 [38]. In this context, impaired epitope presentation could promote the selection of T‐cell clones that elicit a less effective antiviral response, or even self‐activated clones, contributing to inflammation without effectively controlling viral replication. This mechanism may be particularly relevant considering the exceptionally high genetic variability of HDV [39, 40]. Secondary findings included an initial association of HLA‐C*12:02 with persistent viraemia, which was not significant after multivariate adjustment and a potential protective effect of DPB1*01:01, which was more frequent in patients with mild fibrosis. HLA‐C*12:02 has been mainly studied in the context of HIV infection [41] and severe COVID‐19 [34]. DPB1*01:01, instead, is among the five most common HLA‐DPB1 molecules worldwide [42] and has been associated with hepatitis B virus persistence [43]. However, no association with persistent HDV viraemia was observed in our cohort. It is important to highlight that this allele is particularly enriched in our anti‐HDV population, even when considering just Caucasian individuals. This enrichment may potentially bias the observed association with mild fibrosis. We also saw that viraemic patients presented a higher allele frequency of the HLA‐G*01:01. This allele group is notably less polymorphic, and no direct association between specific alleles and viral infection outcome has been reported. A different titre of the soluble form of HLA‐G had been related to different clinical stages of HBV infection [44], and single nucleotide polymorphisms (SNPs) in the 3′ untranslated region had been linked to susceptibility to HBV‐related HCC [45]. Of note, the amplicon used in this study for HLA‐G profiling includes part of the 3′UTR region. Despite finding some SNPs, no difference in their frequency was observed when comparing different patient groups (data not shown). Another minor finding is the enrichment of HLA‐A*02:01 in anti‐HDV patients with advanced fibrosis. Notably, it is one of the most common HLA‐A alleles in the European population, and some restricted epitopes have been described for HDV [25, 46] and HBV [47, 48]. This allele has also been associated with a higher risk of drug‐induced liver injury due to amoxicillin‐clavulanate/clavulanic acid administration [49]. In this study, despite showing an association with advanced fibrosis, the relation was lost in multivariate analysis, thus suggesting that other factors may be involved. It is also worth noting that several alleles were significantly enriched in anti‐HDV patients compared with the reference group. However, these results did not imply a difference in HDV infection susceptibility but rather reflect a higher or lower frequency of certain alleles between the two tested groups. In this study, the reference group was used exclusively to ensure that our primary findings were not confounded by the disproportionate presence of specific alleles. While insightful, these findings are limited by a modest sample size and a single‐centre scope. This may result in an ethnic distribution that does not fully reflect the general population, thereby constraining the generalisability of our results. Further studies, including cross‐validation in larger cohorts from different geographic and ethnic backgrounds, with more viral genotypes, along with functional in vitro modelling, are required to confirm these findings and further elucidate the HLA‐TCR‐HDV axis and its role in HDV disease progression. In conclusion, this study shows that HLA‐A*11:01 may be associated with persistent HDV replication, advanced fibrosis, and a restricted TCR repertoire, highlighting its potential role in disease pathogenesis. Further studies and cross‐validation in a larger independent cohort are required to confirm these results. Author Contributions Adriana Palom: investigation, formal analysis, data curation, writing – original draft, validation. Maria Francesca Cortese: conceptualization, funding acquisition, writing – original draft, investigation, data curation, visualization, methodology, supervision, project administration, formal analysis. Emma Enrich Rande: methodology, writing – original draft, formal analysis, investigation. Francesc Rudilla: methodology, writing – original draft, validation, formal analysis. Arnau Puig Lopez: methodology, investigation, formal analysis. Juan Carlos Ruiz‐Cobo: investigation, methodology. Maria José Herrero: validation, writing – review and editing. Mar Riveiro‐Barciela: investigation, writing – review and editing, methodology. Maria Buti: conceptualization, funding acquisition, investigation, writing – original draft, supervision, project administration. Judit Vico Romero: investigation, methodology. Ariadna Rando Segura: methodology, writing – review and editing, resources. David Tabernero Caellas: methodology, investigation. Funding Research reported in this publication was funded by the Instituto de Salud Carlos III (ISCIII) under the projects PI23/01065 and PI20/01692 and co‐funded by the European Union. Ethics Statement The study was conducted in accordance with the principles of the Declaration of Helsinki and Spanish regulations on biomedical research and personal data protection. The protocol was approved by the Vall d'Hebron Ethics Committee (PR(AG)627/2023; 15 July 2024). Consent Patients provided informed consent for the use of their clinical records and for the extraction of an additional blood sample for HLA and TCR profiling. Conflicts of Interest Cortese MF has served as a speaker and consultant, and has received research funding from Roche Diagnostics. Rando Segura A reports receiving research funding from Gilead and Roche Diagnostics. Riveiro Barciela M has served as a speaker and has received research funding and travel support from Gilead, AbbVie, Novartis, BMS, and Giffols. Buti M has served as a speaker and consultant, and has received research funding from AbbVie, Gilead, and Janssen. Supporting information Figure S1: Allele frequencies in anti‐HDV patients and the reference group. The heatmap shows the frequency of each allele between anti‐HDV patients and the reference healthy donor group. The colour indicates the frequency: Darker colour refers to a higher frequency. Figure S2: Polymorphisms in the HLA‐G promoter between anti‐HDV and reference. Percentage of anti‐HDV patients and reference individuals showing polymorphisms in the HLA‐G promoter. Table S1: Allele frequency (AF) between anti‐HDV‐positive patients and the reference group (N = 150). Odds ratio, 95% CI (lower and upper limits) and p‐value (both unadjusted and FDR‐adjusted) after applying Fisher's exact test are reported. Table S2: Allele frequencies (AF) of the unique alleles, observed only in anti‐HDV patients or the reference group (N = 150). n/a: not applicable.