Associations between Alzheimer’s disease polygenic risk scores and hippocampal subfield volumes in 17,161 UK Biobank participants

Highlights

• The hippocampus is not a uniform structure and has a number of subfields.

• Its associations with polygenic risk score for AD (PRS_AD) have been inadequately investigated.

• Higher PRS_AD was associated with lower volumes in certain hippocampal subfields.

• Older individuals showed greater subfield vulnerability to high PRS_AD compared to the younger group.

Abstract

Hippocampal volume is an important biomarker of Alzheimer’s disease (AD), and genetic risk of AD is associated with hippocampal atrophy. However, the hippocampus is not a uniform structure and has a number of subfields, the associations of which with age, sex, and polygenic risk score for AD (PRS_AD) have been inadequately investigated. We examined these associations in 17,161 cognitively normal UK Biobank participants (44–80 years). Age was negatively associated with all the hippocampal subfield volumes and females had smaller volumes than men. Higher PRS_AD was associated with lower volumes in the bilateral whole hippocampus, hippocampal-amygdala-transition-area, and hippocampal tail; right subiculum; left cornu ammonis 1, cornu ammonis 4, molecular layer, and granule cell layer of dentate gyrus. Older individuals (median age 63 years, n = 8984) showed greater subfield vulnerability to high PRS_AD compared to the younger group (n = 8177), but the effect did not differ by sex. The pattern of subfield involvement in relation to the PRS_AD in community dwelling healthy individuals sheds additional light on the pathogenesis of AD.

Introduction

Hippocampal atrophy is one of the most validated and widely used biomarkers of Alzheimer’s disease (AD) (de Flores et al., 2015). However, the hippocampus is not a homogenous structure and differential vulnerability of the hippocampal subfields to neurodegeneration has been reported to affect different cognitive functions (Mueller et al., 2010). Decline in volumes of hippocampal subfields have been associated with age and sex. Studies in healthy adults have shown significant age effects in cornu ammonis (CA)1-2 (Daugherty et al., 2016; Mueller et al., 2007; Pini et al., 2016; Shing et al., 2011; Wisse et al., 2014), dentate gyrus (DG) (Daugherty et al., 2016; Wisse et al., 2014), and CA4 volumes (Wisse et al., 2014). Evidence has been presented for an anterior-posterior gradient in hippocampal volume reduction (Malykhin et al., 2017) with age, which implies that the hippocampal subregions are not uniformly affected along the hippocampal axis. Significant decreases in CA1 and subiculum volumes and shape have been observed in participants with mild cognitive impairment (MCI) and AD (de Flores et al., 2015; Pini et al., 2016). Moreover, sex differences in hippocampal subfield volumes have also been observed—males had larger parasubiculum, fimbria, hippocampal fissure, and presubiculum—whereas females had larger volumes for the hippocampal tail (van Eijk et al., 2020). It has also been shown that total hippocampal volume loss in females was more pronounced than males (Nobis et al., 2019). However, to date, there is a lack of studies examining sex differences of hippocampal subfield volumes in greater detail. Therefore, it is important to further study the impact of normal aging and sex on hippocampal subregions and to determine whether AD genetic risk selectively and differentially impacts distinct subfields.

Although one heritability study using twins by Elman et al. (2019) showed little genetic impact on subfields after accounting for total hippocampus volume, a recent genome-wide association study (GWAS) on hippocampal subfield volumes derived from FreeSurfer reported that genetic determinants varied across different subregions (van der Meer et al., 2018). The authors found that the volumes of all hippocampal subfields were heritable in individuals with a mean age of approximately 47.8 (standard deviation = 17.3) years (h² range: 0.14–0.27). They identified genome-wide significant single nucleotide polymorphisms (SNPs) associated with the whole hippocampus, presubiculum, subiculum, CA1, DG, molecular layer, and hippocampal tail. In addition, in their follow-up analyses on age-stratified subsamples, they also investigated the genetic overlap with AD and showed that 3 of the significant SNPs were associated with hippocampal subfield volumes in the older age group. This study showed that the differences in cytoarchitecture of the hippocampal subfields are partially driven by genetic variation and that AD-related genes may influence the hippocampal volume predominantly later in life.

Previous studies have identified several common risk variants for AD beyond the well-established apolipoprotein E epsilon 4 allele (APOE-ε4) (Li et al., 2018; Xiao et al., 2017). However, these risk alleles have shown small effects on disease risk (Desikan et al., 2015; Lambert et al., 2013). Other studies have used the AD polygenic risk score (PRS_AD) to examine the cumulative genetic risk for AD (Dezhina et al., 2018; Li et al., 2018), which has shown improvements in predicting cognitive decline beyond the APOE locus (Escott-Price et al., 2015).

As hippocampal subfield volumes are differentially affected in the early stages of AD (Zhao et al., 2019), it is possible that this is driven by the genetic risk for AD. Moreover, the PRS_AD may have a differential effect on the subfield volumes in a cohort of non-demented individuals. To date, there has not been any study to investigate the relationship between PRS_AD and hippocampal subregions, which could help understand the relationship among AD genetic risk factors, aging, and the pathobiology of AD.

The overall aim of this study is to investigate hippocampal volume subfields in greater detail in a large cohort using the UK Biobank data of n = 17,161 participants. The study had 3 main objectives: (1) to investigate the relationships among age, sex, and hippocampal subfield volumes; (2) to examine the effects of PRS_AD constructed using genome-wide significant AD risk variants (Lambert et al., 2013) on hippocampal subfield volumes; and (3) to study the interaction of age and sex with the effects of PRS_AD. We hypothesized that age and sex would be associated with hippocampal subfield volumes. We also hypothesized that there would be varying effects of PRS_AD across the different hippocampal subfield volumes, with higher genetic risk of AD being differentially associated with lower hippocampal subfield volumes. We also hypothesized that PRS_AD would be modulated by age and sex on hippocampal subfield volumes.