Carrasquillo MM, Belbin O, Hunter TA, Ma L, Bisceglio GD, Zou F, Crook JE, Pankratz VS, Dickson DW, Graff-Radford NR, Petersen RC, Morgan K, Younkin SG. Replication of CLU, CR1, and PICALM Associations With Alzheimer Disease. Arch Neurol. 2010;67(8):961-964. doi:10.1001/archneurol.2010.147
To test for replication of the association between variants in the CLU, CR1, and PICALM genes with Alzheimer disease.
Follow-up case-control association study.
The Mayo Clinics at Jacksonville, Florida, and Rochester, Minnesota.
Community-based patients of European descent with late-onset Alzheimer disease (LOAD) and controls without dementia who were seen at the Mayo clinics, and autopsy-confirmed cases and controls whose pathology was evaluated at the Mayo Clinic in Jacksonville. Additional samples were obtained from the National Cell Repository for Alzheimer Disease (NCRAD). A total of 1829 LOAD cases and 2576 controls were analyzed.
The most significant single-nucleotide polymorphisms in CLU (rs11136000), CR1 (rs3818361), and PICALM (rs3851179) were tested for allelic association with LOAD.
Main Outcome Measure
Clinical or pathology-confirmed diagnosis of LOAD.
Odds ratios for CLU, CR1, and PICALM were 0.82, 1.15, and 0.80, respectively, comparable in direction and magnitude with those originally reported. P values were 8.6 × 10−5, .014, and 1.3 × 10−5, respectively; they remain significant even after Bonferroni correction for the 3 single-nucleotide polymorphisms tested.
These results show near-perfect replication and provide the first additional evidence that CLU, CR1, and PICALM are associated with the risk of LOAD.
Late-onset Alzheimer disease (LOAD), a neurodegenerative condition characterized by large numbers of senile plaques and neurofibrillary tangles in the brain, is the most common cause of dementia in elderly persons. Multiple rare mutations in the APP, PSEN1, and PSEN2 genes cause an early-onset, familial form of the disease,1 and twin studies indicate that susceptibility alleles may contribute as much as 80% to the risk of LOAD.2 Until recently, however, APOE ε4 was the only allele reliably associated with increased susceptibility to LOAD.3- 5 A robust technology has emerged that permits genome-wide association studies (GWAS) of large numbers of subjects. This technology has enabled the identification of relatively weak associations that would otherwise go undetected.
Recently Harold et al6 and Lambert et al7 published the 2 largest LOAD GWAS conducted to date and reported genome-wide significant association with 3 novel genes. The study by Harold et al reported the association of single-nucleotide polymorphisms (SNPs) in CLU (OMIM 185430) and PICALM (OMIM 603025). The study by Lambert et al also reported association of CLU with LOAD and additionally reported a novel association with CR1. Here we describe our effort to replicate these findings in an independent LOAD case-control series consisting of 1829 cases and 2576 controls of European descent. Our results show near-perfect replication and provide the first additional evidence of association of these 3 genes with LOAD.
Samples used in this study do not overlap with the samples included in the Harold et al6 or Lambert et al7 publications. The United States case-control series consisted of white subjects from the United States recruited at the Mayo Clinic or through the National Cell Repository for Alzheimer Disease (NCRAD). All subjects assessed at the Mayo Clinics in Jacksonville, Florida, and Rochester, Minnesota, were diagnosed by a Mayo Clinic neurologist. The neurologist confirmed a Clinical Dementia Rating score of 0 for all subjects enrolled as controls; cases had diagnoses of possible or probable AD made according to National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association criteria.8 In the autopsy-confirmed series, all brains were evaluated by Dr Dickson and came from the brain bank maintained at the Mayo Clinic in Jacksonville, Florida. In the autopsy-confirmed series, the diagnosis of definite AD was also made according to National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association criteria. All AD brains analyzed in the study had a Braak score of 4.0 or greater. Brains used as controls had a Braak score of 2.5 or lower but often had brain pathology unrelated to AD and pathological diagnoses that included vascular dementia, frontotemporal dementia, dementia with Lewy bodies, multisystem atrophy, amyotrophic lateral sclerosis, and progressive supranuclear palsy. One AD case from each of the 702 late-onset NCRAD families was analyzed. The NCRAD AD cases were selected based on strength of diagnosis (autopsy-confirmed [32%] > probable [45%] > possible [8%] > family-reported [15%]); the case with the earliest age at diagnosis was taken when several cases had equally strong diagnoses. The 209 NCRAD controls that we included were unrelated white subjects from the United States with a Clinical Dementia Rating of 0, specifically collected for inclusion in case-control series. Written informed consent was obtained for all individuals who participated in this study.
For the Jacksonville and Rochester samples, DNA was isolated from whole blood using an AutoGen instrument (AutoGen Inc, Holliston, Massachusetts). The DNA from autopsy-confirmed samples was extracted from the cerebellum using Wizard Genomic DNA Purification Kits (Promega Corporation, Madison, Wisconsin). The DNA from the Rochester and autopsy-confirmed series was scarce, so samples from the 2 series were subjected to whole-genome amplification using the Illustra GenomiPhi V2 DNA Amplification Kit (GE Healthcare Bio-Sciences Corp, Piscataway, New Jersey).
All genotyping was performed at the Mayo Clinic in Jacksonville using TaqMan SNP Genotyping Assays in an ABI PRISM 7900HT Sequence Detection System with 384-Well Block Module from (Applied Biosystems, Foster City, California). The genotype data was analyzed using the SDS software version 2.2.3 (Applied Biosystems).
For the analysis of the Mayo series, an allelic dosage model using logistic regression adjusted by sex, age, and APOE ε4 status was used to generate P values and odds ratios (ORs) for the association of LOAD with the minor allele. The results reported in the 2 original GWAS articles by Harold et al6 and Lambert et al7 were also generated using logistic regression analyses, assuming an additive model to test for association with the minor allele. However, the covariates used differed between the 3 studies. Harold et al did not adjust for age or sex, only for geographical region and genotyping array used, while Lambert et al adjusted for age, sex, and center. The results from the Mayo series stay essentially the same when no covariates are used (ie, when no adjustment is made for sex, age, and APOE ε4 status). Because genotype counts were not available for the Harold et al follow-up series or for either the GWAS or follow-up series from the Lambert et al study, we used a Fisher combined test to combine P values across series.
We genotyped the most significant SNP in each of the 3 genes reported by Harold et al6 and Lambert et al7 to determine if the associations could be detected in our follow-up series. Remarkably, the direction and magnitude of each association replicated well in our series, and addition of our follow-up data to the results previously reported increased the strength of evidence (Fisher combined summary statistic) of each of the associations (Table).
The Harold et al GWAS showed significant association of the CLU SNP, rs11136000, with reduced risk of LOAD (OR, 0.86; 95% CI, 0.82-0.90; P = 8.5 × 10−10) in their combined series (GWAS + “extension series,” 5964 LOAD cases and 10 188 controls). This association was replicated in the Lambert et al study (OR, 0.86; 95% CI, 0.81-0.90; P = 7.5 × 10−9) in their combined series (5791 LOAD cases and 8420 controls). In our LOAD case-control follow-up series of 1819 LOAD cases and 2565 controls, we observed an odds ratio of the same magnitude and direction (OR, 0.82; 95% CI, 0.75-0.91; P = 8.6 × 10−5).
In their GWAS, Lambert et al reported a significant association of the CR1 SNP, rs3818361, with increased risk of LOAD (OR, 1.19; 95% CI, 1.11-1.26; P = 8.9 × 10−8), an effect that was replicated by Harold et al in their GWAS series of 3939 LOAD cases and 7848 controls (OR, 1.17; 95% CI, 1.09-1.25; P = 9.2 × 10−6). Again, we observed an odds ratio of the same magnitude and direction (OR, 1.15; 95% CI, 1.03-1.29; P = 1.4 × 10−2).
Finally, Harold et al reported significant association of the PICALM SNP, rs3851179, with reduced risk of LOAD (OR, 0.86; 95% CI, 0.82-0.90; P = 1.3 × 10−9). This association was not reported by Lambert et al but we observed the same association as Harold et al in our follow-up study (OR, 0.80; 95% CI, 0.73-0.89; P = 1.3 × 10−5).
For more than 15 years, APOE alleles were the only genetic variants that showed replicable association with altered susceptibility to LOAD. More than 500 additional candidate genes were investigated in more than 1200 studies9 but little progress was made until the last several years when large LOAD case-control series and meta-analyses were used to gain the power necessary to detect associations much weaker than those of the APOE alleles. Except for the 3 genes examined in this article, the most significant evidence of a novel LOAD gene comes from Carrasquillo et al,10 who found impressive association of PCDH11X (rs2573905) with LOAD (P = 5.4 × 10−13) in their 2-stage GWAS of 7 case-control series with a combined total of 5010 subjects. The AlzGene Web site, maintained by Bertram and Tanzi, summarizes genetic association studies of LOAD and currently lists 35 loci (as of November 2009) with 1 or several variants that are nominally significant when tested for allelic association in random-effects meta-analyses.11 Some of these were identified using a candidate gene approach, others in the 11 LOAD GWAS (eg, CLU, CR1, PICALM, and GAB2) that have been performed to date.9 The SORL1 gene, which directs β-amyloid away from the β-amyloid–generating pathway and was discovered using a candidate gene approach,12 shows particularly impressive association with several variants that associate with LOAD in multiple, large, case-control series. The association for many of the other candidate genes is based on relatively few subjects, shows substantial heterogeneity from series, or is weak. Thus, many of the nominally significant associations for candidate genes are tenuous and require additional replication.
Genome-wide association studies often give inflated ORs that are substantially reduced in follow-up series. In this first independent follow-up analysis of CLU, CR1, and PICALM, this was not the case, as we obtained ORs that were essentially identical to those observed in the initial studies. Thus, our findings provide strong additional evidence that all 3 genes are novel LOAD genes. CLU (also known as APOJ) encodes clusterin, which interacts with β-amyloid13- 16 and appears to influence the aggregation and toxicity of this important AD-related peptide.17- 20CR1 encodes the major receptor of C3b, a protein involved in complement activation, and could mediate complement-driven phagocytosis that fosters β-amyloid clearance.7PICALM (also known as CALM) encodes phosphatidylinositol-binding clathrin assembly protein, which is involved in clathrin-mediated endocytosis,21,22 a process that could alter risk for AD through an effect on synaptic transmission or by altering endocytosis of the β-amyloid protein precursor. Thus, all 3 new genes afford good opportunities for pursuit in biological experiments aimed at identifying novel approaches to the treatment of AD.
Correspondence: Steven G. Younkin, MD, PhD, Department of Neuroscience, Mayo Clinic, 4500 San Pablo Rd, Birdsall Building, Jacksonville, FL 32224 (firstname.lastname@example.org).
Accepted for Publication: January 22, 2010.
Published Online: June 14, 2010. doi:10.1001/archneurol.2010.147
Author Contributions: Dr Younkin had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Carrasquillo and Younkin. Acquisition of data: Bisceglio, Carrasquillo, Dickson, Graff-Radford, Hunter, Ma, and Petersen. Analysis and interpretation of data: Belbin, Carrasquillo, Crook, Morgan, Pankratz, and Younkin. Drafting of the manuscript: Belbin, Carrasquillo, Morgan, and Younkin. Critical revision of the manuscript for important intellectual content: Belbin, Carrasquillo, Crook, Dickson, Graff-Radford, Morgan, Pankratz, Petersen, Younkin, and Zou. Statistical analysis: Carrasquillo, Crook, Morgan, Pankratz, and Younkin. Obtained funding: Dickson, Graff-Radford, Morgan, Petersen, and Younkin. Administrative, technical, and material support: Carrasquillo, Dickson, Graff-Radford, Petersen, and Younkin. Study supervision: Carrasquillo, Morgan, and Younkin.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant R01 AG18023 from the US National Institutes of Health, National Institute on Aging (Drs Graff-Radford and Younkin); grant P50 AG16574 from the Mayo Alzheimer's Disease Research Center (Drs Petersen, Dickson, Graff-Radford, and Younkin); grant U01 AG06576 from the Mayo Alzheimer's Disease Patient Registry (Dr Petersen); grants AG25711, AG17216, and AG03949 from the National Institute on Aging (Dr Dickson); cooperative agreement grant U24 AG21886 from the National Institute on Aging (NCRAD); the Robert and Clarice Smith Postdoctoral Fellowship (Dr Carrasquillo); Robert and Clarice Smith and Abigail Van Buren Alzheimer's Disease Research Program (Drs Petersen, Dickson, Graff-Radford, and Younkin); the Palumbo Professorship in Alzheimer's Disease Research (Dr Younkin); the Alzheimer's Research Trust and the Big Lottery Fund (Dr Morgan); and a travel grant from the Alzheimer's Research Trust (Dr Belbin).
Role of the Sponsor: Samples from the National Cell Repository for Alzheimer's Disease were used in this study. The funding organizations had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Additional Contributions: We thank contributors, including the Alzheimer disease centers that collected samples used in this study, as well as subjects and their families, whose help and participation made this work possible.