Common and rare variants in HFE are not associated with Parkinson's disease in Europeans

Abstract

A recent study suggested that the p.H63D variant in HFE, a gene involved in iron homeostasis, may modify α-synuclein pathology, the pathological hallmark of Parkinson’s disease (PD). If indeed this gene and specific variant are involved in PD, we expect to find differential distribution of HFE variants when comparing PD patients and controls. We analyzed genome-wide association study (GWAS) data from 14,671 PD patients and 17,667 controls and full sequencing data from additional 1,647 PD patients and 1,050 controls, using logistic regression models, and burden and Kernel tests. The HFE p.H63D variant was not associated with PD, nor did all the other common variants in the HFE locus. We did not find association of rare HFE variants with PD as well in all types of burden and Kernel tests. Our results do not support a role for HFE in PD risk.

Introduction

Parkinson’s disease (PD) is a complex neurodegenerative disorder, caused in most cases by multiple genetic and environmental factors, and the normal process of aging. Previous studies have suggested that iron homeostasis may have a role in PD; pathological and imaging studies showed that iron accumulation in the substantia nigra may contribute to PD pathogenesis (Ward et al., 2014). Multiple genes are involved in iron homeostasis, one of the most important of them being HFE, encoding the homeostatic iron regulator protein (also called hereditary hemochromatosis protein). This protein is a membrane protein thought to regulate iron absorption, and bi-allelic HFE mutations may lead to hereditary haemochromatosis, an iron storage disorder.

Although several genetic studies reported lack of association of the HFE p.H63D variant with PD (Biasiotto et al., 2008; Duan et al., 2016; Guerreiro et al., 2006; Rhodes et al., 2014; Xia et al., 2015), a recent study suggested that this variant may be protective in Parkinson’s disease pathogenesis by modifying α-synuclein pathology in cell models (Kim et al., 2020). The same group has previously suggested, using a mouse model with the homolog of the same variant (H67D), that HFE is a disease modifier in PD (Nixon et al., 2018). In order to examine whether this and other HFE variants are associated with PD, we performed a comprehensive analysis of common and rare HFE variants in large case-control cohorts of PD patients and controls of European origin.

Methods

Two cohorts were included in the current study: 1) a total of 14,671 PD patients and 17,667 controls collected through collaborators from the International PD Genomics Consortium (IPDGC), and 2) 1,647 PD patients and 1,050 controls from AMP-PD (https://amp-pd.org/). Demographic details on these cohorts can be found in Supplementary Table S1A and S1B. We performed standard genome-wide association study (GWAS) quality control (QC) on the individual and variant level data as previously described (Iwaki et al., 2019). We performed a similar QC process for the AMP-PD whole genome sequencing data, as detailed in the AMP-PD portal (https://amp-pd.org/whole-genome-data). This QC process also excluded, based on genetic relatedness, any relatives up to third degree of relatedness. Genotype data was extracted from both datasets for all variants within 100kb upstream and downstream of HFE, and was annotated using ANNOVAR (Wang et al., 2010). To examine whether common variants (allele frequency >0.01) in HFE are associated with PD, we performed logistic regression adjusted for age, sex, 10 principal components and sites of cohorts. To determine whether rare HFE variants (allele frequency <0.01) are associated with PD, we performed a set of burden and Kernel association tests included in the RVtests R package, including age and sex as covariates when possible (Zhan et al., 2016). Bonferroni correction for multiple comparisons was applied, and all code used in the current study is available at the IPDGC GitHub at https://github.com/ipdgc/IPDGC-Trainees/blob/master/.

Results

None of the common variants in the HFE region were associated with risk of PD (Figure 1). The specific HFE variant recently reported to be associated with decreased aggregation of α-synuclein, p.H63D, had similar allele frequencies in patients and controls (0.17 and 0.16, respectively), and was not associated with risk of PD (OR=1.02, 95%CI=0.97–1.07, p=0.53). Another HFE variant previously reported to be associated with PD, p.C282Y, had similar frequencies in patients and controls as well (0.052 and 0.059, respectively) and was not associated with PD (OR=0.98, 95%CI=0.91–1.07, p=0.69). Similarly, none of the burden and Kernel association tests we used to examine the association of rare HFE variants with PD yielded statistically significant results (Table 1). These analyses included all rare variants in HFE, and all nonsynonymous variants in HFE. No loss-of-function variants were identified, and only one HFE variant had a Combined Annotation Dependent Depletion (CADD) score of >20, therefore no additional analyses were performed on these categories of variants. The list of all common and rare variants found in HFE is available in Supplementary Table S2 and S3, respectively.

Figure 1.

HFE versus PD risk

Table 1.

Burden tests for HFE in the AMP-PD cohorts

Cohort	Burden test	All variants			Missense variants
		Number of Variants	Number of Analyzed Variants	P-value	Number of Missense Variants	Number of Analyzed Variants	P-value
AMP-PD (PD n=1,647, Controls n=1,050)	Zeggini	54	47	0.5167	10	9	0.3644
	Fp	54	47	0.4510	10	9	0.1134
	CMC	54	47	0.2539	10	9	0.3644
	Madson-Browning	54	47	0.2940	10	9	0.2959
	SKAT	54	47	0.8299	10	9	0.8266
	SKAT-O	54	47	0.6310	10	9	0.3425

Abbreviations: SKAT: Sequence kernel association test, SKAT-O: Sequence kernel association test optimized, CMC: combined multivariate and collapsing

Discussion

In the current study we performed a thorough genetic analysis of HFE in large case-control cohorts of PD patients and controls. Our results do not support a role for common or rare HFE variants in risk of PD. Previous smaller genetic association studies focusing on HFE reported contradicting results, as some studies reported associations of different HFE variants with PD, while other studies reported lack of association (Biasiotto et al., 2008; Borie et al., 2002; Buchanan et al., 2002; Duan et al., 2016; Guerreiro et al., 2006; Halling et al., 2008; Rhodes et al., 2014; Xia et al., 2015). Specifically for p.H63D, all the previous studies, including two meta-analyses (Duan et al., 2016; Xia et al., 2015), reported lack of association. Our study, which is a larger-scale analysis combining both GWAS data, as well as full sequencing data on HFE, analyzed through the same pipeline and including burden analyses of rare variants, also provides no support for the involvement of HFE p.H63D in risk of PD. Furthermore, there is no convincing evidence to suggest an association between other HFE variants and risk of PD. We cannot rule out that a very small effect does exist, or that there is an effect on disease progression but not on risk, but since it was not detected by the recent GWAS with over 50,000 PD patients and proxy-cases (Iwaki et al., 2019), it is likely to be too small to be clinically meaningful, if at all.

Clinical and epidemiological studies do suggest a role for iron homeostasis in PD. Neuropathological studies of the substantia nigra pars compacta demonstrated a correlation between increased iron concentrations and increased severity of PD. Transcranial sonography of the substantia nigra shows hyperechogenicity due to iron accumulation in about 90% of PD patients. In the periphery, increased iron levels are associated with reduced risk of PD. A role for iron in PD is further supported by additional histological, biochemical and pathological data (Ward et al., 2014). While genetic data does not support a role for HFE in risk of PD, other iron-regulating genes should be further studied.

The main limitation of the current study is that it includes only participants of European descent, therefore multi-ethnic studies are further required to study the role of HFE in PD in other populations. In a recent GWAS of Asian population, however, no GWAS-significant association has been reported for HFE (Foo et al., 2020).

Despite the lack of association with PD risk in the current study, the role of iron and iron homeostasis in PD should be further studied, whether by genetic studies or by biochemical, functional, clinical and pathological studies. We recommend that genetic data from the IPDGC and other available studies should be consulted before studying specific genetic variants in different models.