Association between HLA alleles and neuropathological outcomes in Lewy body disease

Abstract

Objective:

Lewy Body Disease (LBD) is a complex neurodegenerative disorder characterized by the accumulation of misfolded α-synuclein in the brain. Neuroinflammation has long been implicated in LBD pathogenesis, and recent genetic studies in Parkinson’s disease (a clinical manifestation of LBD) have shown consistent association with the Human Leukocyte Antigen (HLA) gene complex. Herein, we assessed whether variation in HLA alleles influences neuropathologic burden in a path-confirmed series of LBD cases.

Methods:

We conducted a comprehensive analysis of HLA allelic variants in a cohort of 539 autopsy-confirmed LBD cases of European descent from the Mayo Clinic brain bank. High-resolution whole-genome sequencing was employed, and the HLA alleles of each sample were called using the HLA*LA tool. Statistical analyses were conducted to explore associations between common HLA alleles (allele frequency ≥ 1%) and neuropathological outcomes, adjusting for age, sex, and genetic ancestry.

Results:

Our analysis identified associations between certain HLA alleles and specific neuropathological outcomes in LBD, suggesting a potential role for HLA-mediated immune mechanisms in disease progression and subtype differentiation. HLA-DPB1*06:01 was associated with lower Lewy Body counts in the Parahippocampal, Middle frontal, and Inferior parietal gyrus. HLA-DRB1*11:01 correlated with a lower Thal amyloid phase and HLA-B*15:01 with an increased risk of diffuse LBD.

Interpretation:

This study provides a detailed evaluation of the relationship between HLA alleles and LBD pathology, highlighting the importance of immune-related genetic factors in the etiology of LBD.

Introduction

Lewy Body Disease (LBD) is a broad category of neurodegenerative disorders characterized by the accumulation of misfolded α-synuclein protein in Lewy bodies (LBs) and Lewy neurites in various brain regions, accompanied by neuronal cell loss in the central and autonomic nervous systems. Clinically, this manifests as progressive deterioration in movement and multiple cognitive domains, including memory, language, visuospatial abilities, executive functioning, and personality changes, that substantially interfere with daily functioning. ^1,2 LBD can present clinically as several different diseases, predominantly Parkinson’s disease (PD) and dementia with Lewy bodies³, however some patients that present clinically with Alzheimer’s disease (AD) have substantial LB pathology.⁴

LBD requires post-mortem neuropathological examination for definitive diagnosis^3,5,6 and can be classified into three distinct subtypes based on LB distribution: brainstem (BLBD), transitional (TLBD), and diffuse (DLBD). Other important neuropathological features include LB burden, dopaminergic deficits, tau pathology, amyloid plaque distribution, protein clearance impairments, TDP-43 proteinopathy, and cerebrovascular lesions.⁷

While LBD’s etiology remains unclear, evidence suggests a role for neuroinflammatory and immune system dysfunction⁸, with post-mortem studies revealing microglial activation⁹ and altered inflammatory mediators.¹⁰ The Human Leukocyte Antigen (HLA) gene complex, which regulates immune responses, may influence LBD susceptibility through its effects on neuroinflammation and protein aggregate clearance.¹¹

HLA molecules are classified into class I (HLA-A, -B, -C) and class II (HLA-DR, DP, DQ), with varying cellular expression patterns. The HLA nomenclature is described in Fig. 1. According to the standardized nomenclature system, each allele name consists of up to four fields separated by colons; usually, the first two fields, which represent the 2-field resolution, are sufficient for many clinical applications.¹² However, since the antigen recognition domain (ARD) is the most crucial component of the HLA molecule, as it facilitates the binding and presentation of antigenic peptides, the G-group resolution has been established to cluster HLA alleles that share identical nucleotide sequences in the ARD-coding exons, without taking into account the complete gene sequence.^13,14

Fig. 1. The hierarchical structure of the HLA nomenclature system.

The naming convention begins with the gene name, followed by a series of numbers separated by colons that provide increasing levels of specificity about the allele. The first set of digits represents the allele group (1-field resolution), indicating the broad antigen specificity. The second set denotes specific protein variations within the allele group (2-field resolution). Subsequent sets of digits provide further distinctions at the nucleotide sequence level, including synonymous substitutions and intronic variations. This structured nomenclature allows for precise identification and communication of HLA alleles in clinical and research settings.

Several studies have explored the potential association between HLA alleles and LBDs, with most studies involving clinical PD. Several haplotypes have been implicated in conferring an increased risk for PD development. For instance, the HLA-DRB1*15:01 allele has been associated with a stronger binding affinity to the α -synuclein Y39 epitope and carriers of this allele were more likely to exhibit a response to this epitope in PD patients.^15,16 Additionally, the DRB1*03:01 allele was found to be more prevalent in sporadic PD cases compared to controls.¹⁷ Certain haplotypes have also been implicated as potential risk factors for PD development.¹⁸ On the other hand, several HLA alleles and haplotypes have been identified as protective against PD. The DRB1*04 allele group has been consistently reported to exhibit a strong protective effect against PD development.^17–22 Similarly, the HLA-DQB1*03:02 allele^21,22 and the DRB1*04~DQA1*03:01~DQB1*03:02 haplotype¹⁵ have been associated with a protective effect against PD.

In our previous research, we examined two specific HLA variants, rs9275326, and rs13201101, in relation to neuropathological outcomes in a series of LBD cases, but we were not able to identify any significant associations.⁷ Using the same cohort, the current study aims to build upon these findings by conducting a more comprehensive analysis of HLA alleles and their potential role in modifying the neuropathological presentation of LBD. Unlike the previous study, which was limited to two HLA variants, this research employs high-resolution whole-genome sequencing (WGS) and next-generation sequencing (NGS) techniques to evaluate a broader range of specific HLA allelic variants. This approach not only provides a more detailed and expansive analysis but also has the potential to uncover additional HLA variants that could be significant in understanding the relationship between HLA alleles and LBD pathology, offering a more robust and nuanced understanding than previously possible.

Materials and methods

Case Material

The study includes neuropathologically confirmed LBD cases from the Mayo Clinic, Jacksonville, brain bank. We excluded cases lacking information on any of the neuropathological outcome measures examined, those with significant coexisting non-AD neurodegenerative conditions (e.g., multiple system atrophy, amyotrophic lateral sclerosis, progressive supranuclear palsy, corticobasal degeneration, etc.), and those with amygdala-predominant LBs in the setting of advanced AD. After applying these exclusion criteria, a total of 539 cases of European descent were included in the study. For these LBD cases, demographic information regarding age at death and sex was collected for each case, while information regarding disease duration and age at disease onset was available for only a subset of 185 LBD cases.

Neuropathological assessment

Formalin-fixed brains were sampled in a standardized and systematic fashion for neuropathologic assessment. Histopathologic evaluation covered six neocortical regions, two hippocampal levels, a basal forebrain section (including amygdala, lentiform nucleus, and hypothalamus), anterior corpus striatum, thalamus (at the subthalamic nucleus level), midbrain, pons, medulla, and two cerebellar sections with one containing deep cerebellar nuclei. Paraffin-embedded sections, 5 μm thick, were mounted on glass slides and stained with hematoxylin and eosin (H&E) and thioflavin S (Sigma-Aldrich, St. Louis, MO). Immunostaining with anti-α-synuclein antibody (NACP; rabbit polyclonal; 1:3000; gift from Dr. Petrucelli, Mayo Clinic, Jacksonville, Florida; formic acid pretreatment) was conducted on cortex, hippocampus, basal forebrain, and brainstem sections to confirm a neuropathological diagnosis of LBD.²³

A total of 14 different neuropathological outcomes were assessed. Specifically, the subtype of LBD was classified as brainstem, transitional, or diffuse, following criteria described by McKeith et al.²⁴ The likelihood of clinical Lewy body dementia (with Braak tangle stage 0–II and transitional/diffuse LBD or Braak tangle stage III–IV and diffuse LBD) was determined using criteria from the fourth report of the Dementia with Lewy bodies consortium.²⁵ Lewy bodies (LBs) were manually counted in the middle frontal (MF), superior temporal (ST), inferior parietal (IP), cingulate (CG), and parahippocampal (PH) gyri in regions with highest observed density at x200 magnification. Tyrosine hydroxylase-immunoreactivity (TH-ir) in the dorsolateral and ventromedial putamen was evaluated using a rabbit TH polyclonal antibody (1:600; Affinity Bioreagents, Golden, Colorado), as previously reported;²⁶ a lower TH-ir score indicates greater putaminal dopaminergic degeneration. Neuronal loss in the ventrolateral substantia nigra (SN) was semi-quantitatively scored on H&E-stained sections as follows: 0 = none, 0.5 = minimal, 1 = mild, 1.5 = mild-to-moderate, 2 = moderate, 2.5 = moderate-to-severe, 3 = severe.

Braak tangle stage and Thal amyloid phase were assessed with thioflavin S fluorescence microscopy, a method validated in prior studies and endorsed by the National Institute on Aging-Alzheimer’s Association guidelines for its consistency with tau immunohistochemistry, especially in large-scale studies.^27–29 Mitophagy alterations were evaluated in the hippocampus using immunostained slides with the mitophagy marker phosphorylated ubiquitin at serine 65 (pS65-Ub, in-house rabbit polyclonal antibody, 1:650). pS65-Ub positive cell density was quantified via a digital pathology algorithm (Aperio ImageScope v12).^30–34 TDP-43 pathology, including neuronal and glial cytoplasmic inclusions, dystrophic neurites, neuronal intranuclear inclusions, spheroids, or perivascular inclusions, was screened in the amygdala using phosphoTDP-43 immunostained slides (pS409/410, mouse monoclonal, 1:5000, Cosmo Bio, Tokyo, Japan).³⁴ Vascular disease (VaD) pathology, including lacunar infarcts, microscopic infarcts, hemorrhages, cerebral amyloid angiopathy, and leukoencephalopathy, was assessed using H&E-staining or thioflavin S fluorescent microscopy.³⁵ All immunohistochemistry procedures were conducted with the IHC Autostainer 480S (Thermo Fisher Scientific Inc., Waltham, MA) and DAKO EnVision^™+ reagents (Dako, Carpinteria, CA).

There was a small to moderate amount of missing data for LB counts (n=105 to 110 missing), dorsolateral and ventromedial putaminal TH-ir (n=252 missing), ventrolateral SN neuronal loss score (n=121 missing), Thal amyloid phase (n=99 missing), pS65-Ub level (n=114 missing), TDP-43 pathology (n=102 missing), and VaD (n=1 missing). The characteristics of the LBD cases are summarized in Table 1.

Table 1.

Characteristics and outcomes of LBD cases

Variable	N	Median (minimum, maximum) or No. (%) of cases
Age at death (years)	539	79 (50, 103)
Sex (Male)	539	345 (64.0%)
Disease duration (years)	185	7 (0, 20)
LBD subtype	539
Brainstem		2 (0.4%)
Transitional		162 (30.1%)
Diffuse		375 (69.6%)
Lewy body counts
Middle frontal gyrus	434	5 (0, 30)
Superior temporal gyrus	434	10 (0, 50)
Inferior parietal gyrus	434	3.50 (0, 25)
Cingulate gyrus	429	12 (2, 45)
Parahippocampal gyrus	433	16 (1, 45)
Putaminal TH-ir
Dorsolateral	287	3.10 (0.26, 42.18)
Ventromedial	287	8.79 (0.26, 33.54)
Ventrolateral substantia nigra neuronal loss score	418
0=none		1 (0.2%)
0.5=none/mild		12 (2.9%)
1=mild		70 (16.7%)
1.5=mild/moderate		41 (9.8%)
2=moderate		137 (32.8%)
2.5=moderate/severe		62 (14.8%)
3=severe		95 (22.7%)
Braak NFT stage	539
0		9 (7.1%)
I		17 (3.2%)
II		120 (22.3%)
III		168 (31.2%)
IV		112 (20.8%)
V		64 (11.9%)
VI		49 (9.1%)
Thal amyloid phase	440
0		53 (12.0%)
1		40 (9.1%)
2		32 (7.3%)
3		120 (27.3%)
4		48 (10.9%)
5		147 (33.4%)
pS65-Ub	425	2.56 (0.12, 27.35)
TDP-43 pathology	437	243 (55.6%)
VaD	538	125 (23.2%)

LBD=Lewy body disease; TH-ir=tyrosine hydroxylase immunoreactivity; NFT=neurofibrillary tangle; VaD=vascular disease.

NGS and Sequence alignment

NGS and alignment of reads on the reference genome were performed as described in detail in Chia et al.³⁶ Briefly, genomic DNA was quantified using PicoGreen assay, and PCR-free paired-end libraries were constructed using Illumina TruSeq chemistry. Sequencing was carried out on an Illumina HiSeq XTen platform with 150-bp paired-end cycles. The resulting data were processed using the Centers for Common Disease Genomics (CCDG) pipeline standard.³⁷ The GRCh38DH reference genome was used for alignment, as specified in the CCDG standard. Alignment was performed against the GRCh38DH reference genome using the Broad Institute’s implementation of the functional equivalence standardized pipeline, which incorporates GATK (2016) Best Practices.³⁸

HLA typing

The process of identifying HLA allelic variants in a given individual is known as HLA typing. In the context of our study on the association between HLA alleles and neuropathological outcomes in LBD, accurate and high-resolution HLA typing is of great importance. To achieve this, we employed the HLA typing tool HLA*LA,³⁹ to infer the HLA haplotypes of our LBD in G-group resolution. The G-group resolution clusters together HLA alleles that share identical nucleotide sequences in the ARD coding exons, which are responsible for antigen binding and presentation to T cells, thereby capturing the most functionally relevant variations. To validate our primary HLA calls, we also utilized SpecHLA⁴⁰ as a secondary tool. The results from both tools showed high similarity, confirming the robustness of our HLA typing approach. Consequently, we proceeded with our analysis using the primary HLA*LA calls, confident in their accuracy. The SpecHLA results are not presented.

Genetic analysis

HLA-A, HLA-B, HLA-C, HLA-DQA1, HLA-DQB1, HLA-DRB1, HLA-DPA1, HLA-DPB1, HLA-DRB3, HLA-DRB4, HLA-E, HLA-F, and HLA-G were studied. There was no missing data for any HLA types. A summary of common (allele frequency ≥ 1%) HLA alleles is provided in Supplemental Table 1 for each HLA type; in total there were 119 common HLA alleles.

Statistical analysis

In our primary analysis, separately for each HLA type, associations of common HLA alleles with neuropathological outcomes were evaluated using multivariable regression models appropriate for the nature of the given outcome (continuous, binary, ordinal, or count). These multivariable regression models were adjusted for age, sex, and the top five principal components (PCs) of genome-wide genetic (GWAS) data (see Supplementary GWAS data). Specifically, proportional odds logistic regression models were used for ordinal outcomes (ventrolateral SN neuronal loss score, Braak stage, Thal phase), linear regression models were used for continuous outcomes (dorsolateral and ventromedial putaminal TH-ir, pS65-Ub level), binary logistic regression models were used for binary outcomes (LBD subtype [diffuse vs. transitional/brainstem], TDP-43 pathology, VaD), and negative binomial regression models were used for count outcomes (LB counts). Due to their skewed distributions, in linear regression analysis pS65-Ub and dorsolateral putaminal TH-ir were examined after natural logarithm scale transformation, while ventromedial putaminal TH-ir was examined on the square root scale. LBD subtype was dichotomized as diffuse vs. brainstem/transitional owing to the very small number of brainstem LBD cases. Each HLA allele was examined under a dominant model (i.e., presence vs. absence of the given allele).

For proportional odds logistic regression models, odds ratios (ORs) and 95% confidence intervals (CIs) are interpreted as the multiplicative increase in the odds of a more severe category for cases with the presence of the given allele for the given HLA type. For linear regression models, regression coefficients (denoted as β) and 95% CIs are interpreted as the additive increase in the mean value of the given outcome measure (on the natural logarithm scale or square root scale) for cases with the presence of the given allele for the given HLA type. For binary logistic regression models, ORs and 95% CIs are interpreted as the multiplicative increase in the odds of occurrence of the given outcome for cases with the presence of the given allele for the given HLA type. For negative binomial regression models, multiplicative effects and 95% CIs are interpreted as the multiplicative increase in the mean outcome measure for cases with the presence of the given allele for the given HLA type. To account for multiple testing in these primary analyses while accounting for correlation between the 119 common HLA alleles, and also for correlation between the 14 different neuropathological outcomes that were assessed, we utilized a SNPSpD correction⁴¹ when applied to all 1,666 statistical tests that were performed. This SNPSpD correction indicated that there are 12.379 independent neuropathological outcomes, and 117.666 independent common HLA alleles, and correspondingly the Bonferroni-corrected significance level is 3.43 × 10⁻⁵ (i.e., 0.05 / [12.379 * 117.666]). However, although applying such a correction for multiple testing limits the probability of a type I error (i.e., false-positive finding), it also correspondingly increases the likelihood of a type II error (i.e., a false-negative finding). For this reason, we considered p-values < 0.001 as displaying “suggestive” evidence of an association in our primary analysis.

In secondary analysis, for common HLA alleles that displayed a significant or suggestive association with a given neuropathological outcome, we assessed the presence of interactions of that HLA allele with both age at death and sex using the aforementioned multivariable regression models, with the addition of an interaction term between the given HLA allele and age at death, or the given HLA allele and sex. Notably, age at death was dichotomized based on the sample median in these analyses only for ease of presentation only (i.e., in stratified analyses which were used to illustrate presence or absence of any interactions), and was analyzed as a continuous variable in all regression models. In further secondary analysis, we evaluated associations of age at death and sex with the 14 different neuropathological outcomes using the aforementioned multivariable regression models. P-values <0.05 were considered as statistically significant in all secondary analyses. All statistical tests were two-sided. Statistical analysis was performed using R Statistical Software (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria).

Results

In our analysis, we identified one significant (P < 3.43 × 10⁻⁵) and four suggestive (P < 0.001) associations between HLA alleles and neuropathological outcomes, as illustrated in Table 2. Associations of all HLA alleles for each HLA type with neuropathological outcomes are displayed in Supplemental Tables 2 – 31.

Table 2.

Significant and “suggestive” associations between common HLA alleles and neuropathological outcomes.

HLA allele	No. (%) of carriers of the allele	Outcome	Association measure	Estimate (95% CI)	P-value
Significant (P<3.43E-05) associations
HLA-DPB1*06:01	19 (3.5%)	Parahippocampal LB count	Multiplicative effect	0.53 (0.39, 0.72)	3.30E-5
Suggestive (P<0.001) associations
HLA-DRB1*11:01	55 (10.2%)	Thal amyloid phase	Odds ratio	0.34 (0.20, 0.60)	1.56E-4
HLA-DPB1*06:01	19 (3.5%)	Middle frontal LB count	Multiplicative effect	0.40 (0.25, 0.65)	1.80E-4
HLA-DPB1*06:01	19 (3.5%)	Inferior parietal LB count	Multiplicative effect	0.41 (0.25, 0.69)	6.33E-4
HLA-B*15:01	70 (13.0%)	Diffuse LBD	Odds ratio	3.43 (1.74, 7.41)	7.58E-4

LBD=Lewy body disease; LB=Lewy body; CI=confidence interval. Odds ratios and multiplicative effects result from proportional odds logistic regression models (Thal amyloid phase), binary logistic regression models (diffuse LBD subtype), and negative binomial regression models (LB counts) that were adjusted for age at death, sex, and the top 5 principal components of genome-wide genetic data. For the Thal amyloid phase, the odds ratio is interpreted as the multiplicative increase in the odds of a higher Thal phase for cases with the presence of the given HLA allele. For diffuse LBD subtype, the odds ratio is interpreted as the multiplicative increase in the odds of occurrence of diffuse LBD subtype for cases with the presence of the given HLA allele. For LB counts, multiplicative effects are interpreted as the multiplicative increase in the mean LB count for cases with the presence of the given HLA allele. P-values < 3.43E-05 are considered as statistically significant after applying a SNPSpD correction for all 1,666 tests (119 HLA alleles * 14 neuropathological outcomes) that were performed.

Specifically, in analyses that were adjusted for age at death, sex, and the top five PCs of GWAS data, our findings revealed that the HLA-DPB1*06:01 allele is significantly associated with a reduction in parahippocampal LB count (Multiplicative effect=0.53, P=3.30 × 10⁻⁵, Supplementary Fig. 1). Additionally, though not quite statistically significant after correcting for multiple testing, HLA-DPB1*06:01 was associated with a lower middle frontal LB count (Multiplicative effect=0.40, P=1.80 × 10⁻⁴, Supplementary Fig. 1), HLA-DRB1*11:01 was linked to a lower Thal amyloid phase (≥3: 44.7% vs. 74.8%, OR=0.34, P=1.56 × 10⁻⁴), HLA-DPB1*06:01 was associated with a lower inferior parietal LB count (Multiplicative effect=0.41, P=6.33 × 10⁻⁴, Supplementary Fig. 1), and HLA-B*15:01 was linked to an increased risk of diffuse LBD (85.7% vs. 67.2%, OR=3.43, P=7.58 × 10⁻⁴). Of note, we did not adjust the aforementioned regression models for disease duration due to the large amount of missing data for this variable. However, in a sensitivity analysis examining only the 185 LBD cases who did have disease duration information available, results of association analysis for the five aforementioned significant or suggestive associations were very similar when additionally adjusting models for disease duration (data not shown). Table 3 presents other interesting associations with P < 0.01 that have been linked in the past with neurodegeneration.

Table 3.

Interesting HLA alleles that have been associated in the past with neurodegeneration and have a P-value < 0.01.

HLA allele	No. (%) of carriers of the allele	Outcome	Association measure	Estimate (95% CI)	P-value	Reference	Influence
HLA-DRB1*03:01	105 (19.5%)	pS65-Ub	β coefficient	0.40 (0.14, 0.65)	0.002	Sun et al., 2012	Increased PD risk
HLA-DRB1*01:01	108 (20.0%)	pS65-Ub	β coefficient	0.36 (0.11, 0.61)	0.006	Hollenbach et al., 2019	Increased PD risk
						James et al., 2020	Protective for dementia
HLA-DRB1*04:01	95 (17.6%)	Thal amyloid phase	Odds ratio	2.03 (1.28, 3.22)	0.003	Hollenbach et al., 2019	Protective for PD
						Guen et al., 2022
						Yu et al., 2021

pS65-Ub=phospho-ubiquitin; β=regression coefficient. CI=confidence interval. The odds ratio results from a proportional odds logistic regression model. β coefficients result from linear regression models. All models were adjusted for age at death, sex, and the top 5 principal components of genome-wide genetic data. The odds ratio is interpreted as the multiplicative increase in the odds of a higher Thal amyloid phase for cases with presence of the given HLA allele. β coefficients are interpreted as the additive increase in the mean pS65-Ub level (on the natural logarithm scale) for cases with presence of the given HLA allele. P-values < 0.00042 are considered as statistically significant after applying a Bonferroni correction for the 119 HLA alleles that were examined for association with neuropathological outcomes.

We performed two additional analyses of interest. First, for the five aforementioned significant and suggestive associations between HLA alleles and neuropathological outcomes, we examined interactions with age at death and sex. As shown in Supplemental Table 32, we did not identify any significant interactions (all interaction P≥0.11); these significant and suggestive associations were observed in similar magnitude for LBD cases with different ages at death, as well as for males and females. Additionally, though not directly related to the primary aim of our study, we also assessed associations of age at death and sex with neuropathological outcomes (Supplemental Table 33). In analysis that was adjusted for age at death, sex, and the top 5 PCs, a number of significant (P<0.05) associations were observed regarding both age at death and sex. Specifically, the strongest associations with age at death were regarding VaD (OR [per each 10 year increase]: 2.22, P=2.55 × 10⁻⁸), pS65-Ub (β: 0.33, P=5.62 × 10 ⁻⁷), and inferior parietal LB count (Multiplicative effect: 0.81, P=2.22 × 10⁻⁵), while the strongest associations with sex occurred for middle frontal LB count (Multiplicative effect [males vs. females]: 0.69, P=3.67 × 10⁻⁶), inferior parietal LB count (Multiplicative effect: 0.69, P=6.46 × 10⁻⁶), and cingulate LB count (Multiplicative effect: 0.79, P=4.61 × 10⁻⁵). These results were similar when additionally adjusting regression models for all common HLA alleles that were associated with the given neuropathological outcomes with a p-value <0.05 (Supplemental Table 33).

Discussion

Though HLA variation has a well-known role in determining susceptibility to PD, much remains to be understood regarding how HLA alleles may influence the severity of neuropathology in LBD in general, and it is this knowledge gap that we set out to fill in our current study involving 539 LBD cases and 14 different neuropathological outcomes. Our results provide evidence that HLA-DPB1*06:01 is associated with a lower severity of LB pathology in LBD cases as evidenced by lower LB counts in the MF and PH regions as well as the IP region, and also that HLA-DRB1*11:01 is associated with less severe AD pathology, specifically with a lower Thal amyloid phase. Additionally, though not quite significant after adjusting for multiple testing, a weaker but still noteworthy association was observed between HLA-B*15:01 and an increased likelihood of diffuse LBD.

We identified significant or suggestive associations with neuropathological outcomes for two more common HLA alleles in our cohort: HLA-DRB1*11:01 which was observed in 55 cases (10.2%) and HLA-B*15:01 which was noted in 70 cases (13.0%). The protective effect of HLA-DRB1*11:01 indicates that cases carrying this allele are likely to have less severe amyloid pathology in the context of LBD. HLA-DRB1*11:01 has been linked with an increased risk of isolated/idiopathic REM sleep behavior disorder (iRBD),⁴² a synucleinopathy, as well as other non-neurodegenerative disorders and conditions.^43,44 It is important to note that more than 70% of the patients with iRBD will develop PD, Lewy body dementia, or multiple system atrophy (MSA) within 10 to 15 years after being diagnosed, making iRBD a strong predictor for neurodegenerative disease.⁴⁵ The HLA-B*15:01 was associated with a greater than 3-fold increased risk of diffuse LBD in our study, though as previously mentioned, this finding was only suggestive. As HLA-B*15:01 has not been linked to neurodegenerative disease in previous research, validation of this finding will be important.

The strongest associations that we observed in our study involved the HLA-DPB1*06:01 allele, where significant protective associations on LB burden were identified. Specifically, HLA-DPB1*06:01 was associated with reduced MF and PH LB counts, with a weaker association noted for IP LB count. Additionally, as displayed in Supplemental Table 25, this HLA allele was also linked with a lower likelihood of diffuse LBD (OR=0.23, P=0.003), a lower CG LB count (Multiplicative effect=0.61, P=0.002), a lower ST LB count (Multiplicative effect=0.58, P=0.004), and a lower ventromedial putaminal TH-ir (β: −0.93, P=0.006), further underscoring its possible protective role. These findings highlight the potential of HLA-DPB1*06:01 to mitigate the risk of LBD progression.

It is worth briefly highlighting our results for HLA alleles that have a known or proposed involvement in neurodegenerative diseases. Focusing on HLA alleles for which a P < 0.01 association was identified in our study, HLA-DRB1*01:01 has been correlated with increased PD risk,²¹ as well as HLA-DRB1*03:01 has been associated with an increased risk of sporadic PD.¹⁷ For both cases, we were able to find a link between the alleles and an increase of pS65-Ub. Interestingly, pS65-Ub is a marker of mitochondrial dysfunction,^46–48 which is a known feature of neurodegenerative diseases like PD and LBD^49,50 as it is a well-established marker of PINK1-Parkin-mediated mitophagy and reflects mitochondrial quality control responses to cellular stress. Given the relevance of mitochondrial dysfunction in α-synucleinopathies and the involvement of certain HLA alleles in immune and clearance pathways, incorporating pS65-Ub into our analyses would offer valuable insights. In a recent study⁵¹ with a cohort of 176,000 individuals with PD and AD, HLA-DRB1*04:01 showed a protective effect against neurodegenerative diseases. Conversely, we found a weak correlation of HLA-DRB1*04:01 with greater amyloid-beta deposition in the brain (Thal amyloid phase), associated with AD. This weak correlation might suggest that the protective effects of HLA-DRB1*04:01 are context-dependent, potentially mitigating some aspects of neurodegeneration while being less effective, or even counterproductive, in the context of amyloid pathology. Further research is needed to explore the conditions under which HLA-DRB1*04:01 exerts its protective effects and how it may influence amyloid-beta accumulation differently. All previously associated, with LBD, HLA alleles that exist in our cohort are reported in Supplementary Tables 34 and 35 along with our findings for the different neuropathological outcomes.

This study has several limitations. First, although the sample size for this study is relatively large for a study of neuropathologically confirmed LBD cases, the possibility of a type II error (i.e., a false-negative finding) is still possible, particularly after adjusting for multiple testing; we cannot conclude that a true association does not exist simply due to the occurrence of a non-significant p-value in this study. Second, our study only included individuals of European descent, and consequently, assessment of the role of HLA in LBD neuropathology in LBD cases with other ethnic backgrounds will be important. Third, information regarding the clinical diagnoses of our LBD series was not available at sufficient quality to allow for inclusion in our study; this is an inherent limitation of any study involving an autopsy series from a brank bank. However, importantly, examination of our entire LBD series without consideration of clinical subgroups is of primary importance for evaluation of our study’s aims. Finally, as previously referred, our study did not include a replication series, and therefore validation of our significant and suggestive findings in an independent series of LBD cases will be crucial.

In conclusion, our study provides evidence that several different HLA alleles may influence the severity of neuropathology in LBD. By utilizing high-resolution WGS and NGS techniques, our findings indicate that in individuals with LBD, HLA-DPB1*06:01 is linked to lower LB counts, while HLA-DRB1*11:01 correlates with a lower Thal amyloid phase. Additionally, we observed some evidence of an association between HLA-B and increased risk of diffuse LBD. These findings contribute to a deeper understanding of the genetic underpinnings of LBD and reinforce the need for further exploration of HLA-related mechanisms in neurodegeneration.

Abbreviations:

AD

Alzheimer’s disease

ARD

antigen recognition domain

BLBD

brainstem Lewy body disease

CCDG

Centers for Common Disease Genomics

DLBD

diffuse Lewy body disease

GWAS

genome-wide association studies

H&E

hematoxylin and eosin

HLA

Human Leukocyte Antigen

iRBD

isolated/idiopathic REM sleep behavior disorder

LB

Lewy body

LBD

Lewy body disease

MSA

multiple system atrophy

NGS

next-generation sequencing

PD

Parkinson’s disease

SN

substantia nigra

TDP-43

TAR DNA-binding protein 43

TH-ir

tyrosine hydroxylase-immunoreactivity

TLBD

transitional Lewy body disease

VaD

vascular disease

WGS

whole-genome sequencing

Data availability

All sequence data is available through the AMP-PD portal [https://www.amp-pd.org/unified-cohorts/lbd].