Abstract Rare coding variants in TREM2, PLCG2, and ABI3 were recently associated with the susceptibility to Alzheimer’s disease (AD) in Caucasians. Frequencies and AD-associated effects of variants differ across ethnicities. To start filling the gap on AD genetics in South America and assess the impact of these variants across ethnicity, we studied these variants in Argentinian population in association with ancestry. TREM2 (rs143332484 and rs75932628), PLCG2 (rs72824905), and ABI3 (rs616338) were genotyped in 419 AD cases and 486 controls. Meta-analysis with European population was performed. Ancestry was estimated from genome-wide genotyping results. All variants show similar frequencies and odds ratios to those previously reported. Their association with AD reach statistical significance by meta-analysis. Although the Argentinian population is an admixture, variant carriers presented mainly Caucasian ancestry. Rare coding variants in TREM2, PLCG2, and ABI3 also modulate susceptibility to AD in populations from Argentina, and they may have a European heritage.
Alzheimer’s disease (AD) is the most common form of dementia, and has an estimated genetic component of 60–80%1. Over the last decade, more than 20 loci containing common genetic variants (minor allele frequency (MAF) >5%) have been associated with AD2. The advent of new genetic sequencing technologies has enabled the identification of several rare variants (MAF <1%) with moderate effects on AD susceptibility3. In 2017, the International Genomics of Alzheimer’s Project (IGAP) reported four rare coding variants significantly associated with AD4, all of them being involved in microglial-mediated innate immunity. Two of them are novel nonsynonymous variants: a protective one in PLCG2 (rs72824905) and a risk one in ABI3 (rs616338). The other two were previously reported in the susceptibility gene TREM2 (rs143332484 and rs75932628), and are responsible for p.R62H and p.R47H substitutions, respectively. The findings in TREM2 have been consistently replicated in Caucasian4,5,6 and African-American populations7. However, the association of TREM2 p.R47H with AD could not be found in East Asian population, because its frequency is extremely low8,9. This latter observation suggests ethnic variability fostering investigation of these variants in different ethnic groups. Given the increasing population diversity observed in countries all over the world, understanding population-shared and -specific risk factors of AD will translate into improved and specific prevention and/or treatment for people.
Latin America is a vast territory, with a wide diverse admixture of European, Native American, and African ancestral populations. The genetic architecture of sporadic AD has not been studied in this population beyond APOE-ε4. In this report, we provide the first evidence for an association between TREM2, PLCG2, and ABI3 rare variants and AD in the Argentinian population.
Methods
Subjects Individuals with AD and without cognitive impairment older than 60 years were recruited from outpatient Neurology Departments of the following hospitals: Instituto de Investigaciones Médicas “Alfredo Lanari” and Hospital de Clínicas (Buenos Aires City), Hospital Interzonal General de Agudos “Eva Perón” (General San Martín county), Hospital El Cruce “Dr. Néstor Kirchner” (Florencio Varela county), and several assistance centers located across Jujuy province. Only samples from people born in Argentina or in South America were included in the analysis. This study was approved by the ethical committee “Comité de Bioética Fundación Insituto Leloir (HHS IRB #00007572, IORG # 006295, FWA00020769)” by the approval of protocol CBFIL #22. All participants and/or family members gave their informed consent.
Diagnosis of AD followed diagnostic criteria from the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s disease and Related Disorders Association (NINCDS-ADRDA)10. The diagnosis of AD includes a clinical examination to evaluate functionality and activities of daily living that should be compromised; a complete panel of neurocognitive tests to evaluate memory, attention, language, and executive function, of which one or more should be altered; a computerized tomography and/or magnetic resonance imaging to assess cortical–hippocampal atrophy and vascular events, and a blood test analysis to exclude metabolic or infectious causes of dementia. Individuals were included as controls if neurocognitive and clinical assessments were normal.
Genotyping and statistics Genomic DNA was isolated using standard procedures from whole blood or saliva samples. TREM2 (rs143332484 and rs75932628), PLCG2 (rs72824905), and ABI3 (rs616338) variants were genotyped using custom-designed TaqMan assays (Thermo Fisher). Assay accuracy was checked by including positive and negative controls in each experiment. APOE alleles were determined by genotyping rs429358 and rs7412. Association with AD was calculated using Fisher’s exact test with statistical significance of p < 0.05. All variants were in Hardy–Weinberg equilibrium (HWE, p > 0.05). Power calculations were performed using Genetic Power Calculator for discrete traits (http://zzz.bwh.harvard.edu/gpc/cc2.html). Meta-analysis for the effect of rare variants in association with AD was conducted using beta and standard error with Metafor R-package11. Populations from France (n = 8514), Italy (n = 2306), Spain (n = 3966), and Sweden (n = 2286) from the European Alzheimer's Disease Initiative (EADI)4 were included in the meta-analysis. Ancestry of the population European (CEU, n = 85), Yorubas African (AFR, n = 88), and Native American (NAM, n = 46) ancestral populations were obtained from 1000 Genomes (http://www.internationalgenome.org/). Argentinian samples (ARG) were subjected to genome-wide genotyping using the Infinium Global Screening Array (GSA) v.1.0+GSA shared custom content (Illumina). Quality controls (QC) were performed as described before12, using PLINK v1.913 and R v3.4.414. After QC, remaining samples (n = 834) have <5% of missing genotypes and passed sex-check and identity-by-state filters. Remaining single nucleotide polymorphisms (SNPs) have >95% call rate, MAF >1%, are in HWE (p > 10−6), and without differences in call rate between cases and controls (p < 1 × 10−5).
For the ancestry analyses, overall population ancestry was first evaluated for ARG by extracting 446 ancestry informative markers (AIMs), which were specifically selected to estimate ancestry in Latin America15, from the genotyped data. Second, ancestry of chromosomes containing rare variants was evaluated by extracting from the 446 AIMS, the AIMs in chromosome 6 from people carrying TREM2 p.R47H and TREM2 p.R62H, those AIMs in chromosome 16 from people carrying PLCG2 p.P522R, and finally AIMs in chromosome 17 from people carrying ABI3 p.S209F. For each ancestry estimation, the same AIMs were extracted from CEU, AFR, and NAM. Ancestry was predicted using ADMIXTURE v1.3.016.
Results To evaluate the association of the four rare variants recently reported by IGAP4 with AD in an Argentinian population sample (ARG), 905 participants were recruited from different regions of the country. Demographic and clinical information of the 419 AD cases and 486 controls is summarized in Table 1. We first explored the risk effect of APOE-ε4 allele on AD susceptibility confirming thereby previous reports (odds ratio (OR) = 3.14, p < 0.0001)17. For APOE-ε2, we observed the expected protective effect, although it did not reach statistical significance (OR = 0.77, p < 0.33). Next, we genotyped the recently described rare variants4, i.e., TREM2 p.R47H (rs75932628) and p.R62H (rs143332484), PLCG2 p.P522R (rs72824905) and ABI3 p.S209F (rs616338). All of them were detected in ARG with MAFs similar to those reported by IGAP (Table 2)4.
They also showed similar magnitude of association, even though neither of these variants reached statistical significance (Table 2)4. This observation was expected, since our sample had a power of 60% to detect OR = 3 of a variant with MAF = 0.01. Notwithstanding, the fact that all variants showed similar effect sizes and effect directions as in the report of IGAP prompted us to perform a meta-analysis using samples from France, Italy, Spain, and Sweden (EADI)4. Despite larger sample size, statistical power associated with EADI does not allow to reach nominal significant association with AD risk. However, when meta-analyzing both EADI and ARG samples, this gain in power is enough to help in reaching statistical significance for the variants analyzed, in particular the TREM2 p.R47H variant (Table 3, Figure S1). Unfortunately, information for TREM2 p.R62H was not available. This can be explained by the observation that the variant effects in ARG are similar to those detected in the other European populations analyzed (as indicated by heterogeneity statistic (I2), see Table 3). All these results together support the hypothesis that these rare variants are also associated to AD in ARG at a similar level than the one observed in Europe.