Field of the invention

The present invention relates to a method for determining the risk of occurrence of Alzheimer's disease.

Technical background

AD is a neurological disorder primarily affecting the elderly that manifests by memory disorders, cognitive decline and loss of autonomy. The incidence of AD is rising rapidly, raising fears that a large fraction of an increasingly elderly population will ultimately be affected. The need for effective preventative and therapeutic measures has motivated extensive research into the disease pathogenesis.

The pathology of AD is characterized by neuronal loss leading to atrophy and to diminution of the cerebral metabolism that initially has limited dispersion but which subsequently spreads to the entire brain. Two principal types of neuropathologic lesions are observed: (i) neurofibrillary degeneration resulting from the intraneuronal accumulation of hyperphosphorylated Tau proteins; (ii) amyloid deposits resulting from the extracellular accumulation of amyloid plaques, principally composed of amyloid β (Aβ) peptide. Currently, the processes leading to formation of these lesions and their dual association with AD are not adequately understood.

Genetic studies have provided significant insights into the molecular basis of AD. Rare hereditary early-onset forms of the disease have been linked to mutations in three different genes: the amyloid precursor protein (APP) gene on chromosome 21 , the presenilin 1 (PS1 ) gene on chromosome 14 and the presenilin 2 (PS2) gene on chromosome 1. In vitro and in vivo studies have shown that pathogenic mutations in APP, PS1 and PS2 favour Aβ peptide production, particularly Aβn-42(43). These latter observations su pport the amyloid cascade hypothesis, namely of overproduction of the Aβ peptides, which would provoke neurofibrillary degeneration and finally neuronal death. These mutations, however, explain less than 1 % of all AD cases whereas the vast majority of AD cases (especially late-onset forms) have other more complex genetic determinants.

Genetic studies have led to the identification of the ε4 al lele of the apolipoprotein E gene (APOE) as one of the susceptibility loci for late-onset AD (Farrer et al. (1997) JAMA 278:1349-1356). Biological investigations have shown the APOE is likely to play a significant role in the fibhllogenesis and in clearance of Aβ peptide. Twin studies suggest that genes may contribute in more than 60% of AD susceptibility whereas APOE may account as much as 50% of this genetic susceptibility (Ashford & Mortimer (2002) J. Alzheimer Dis. 4:169-177).

More than 550 other genes have been proposed as candidates for involvement in of the disease, but thus far none of these has been confirmed to have a role in pathogenesis (Bertram et al. (2007) Nat. Genet. 39:17-23).

Accordingly, it is an object of the present invention to provide alternative genetic risk factors to APOE for Alzheimer's disease.

Description of the invention

The present invention arises from the unexpected finding, by the present inventors, that variants of the apolipoprotein J gene (APOJ) and of the complement component receptor 1 gene (CR1) were risk factor for the onset of Alzheimer's disease.

The present invention thus relates to an in vitro method for determining that an individual at risk of developing Alzheimer's disease, which comprises:

- determining whether the individual harbours at least one variant allele of a susceptibility gene selected from the apolipoprotein J gene (APOJ) and the complement component receptor 1 gene (CR1);

- deducing that if the individual harbours at least one variant allele of the APOJ and/or CR1 gene then the individual is at risk of developing Alzheimer's disease.

In an embodiment, the invention further comprises further determining whether the individual presents at least one other risk factor for Alzheimer's disease and deducing that the individual is at risk of developing Alzheimer's disease if said individual harbours at least one variant allele of the APOJ and/or the CR1 gene and presents said at least one other risk factor.

As intended herein the above-defined individual is preferably a human.

"Alzheimer's disease" is well known to one of skill in the art and is notably characterized by an abnormal accu mu lation of plaq ues resu lting from the aggregation of the β-amyloid peptide.

As intended herein, the expression "the individual is at risk of developing Azheimer's disease" notably means that the individual is predisposed to develop the Alzheimer's disease or that the individual presents an increased risk of developing Alzheimer's disease with respect to a general population of individuals or to a population of individuals which do not harbour the above-defined at least one variant allele of the APOJ and/or CR1 gene.

The "apolipoprotein J gene (APOJ)", is also known, inter alia, as the clustehn [CLU) gene. APOJ sensu stricto is localized on chromosome 8 in the cytogenetic band 8p21 -p12 and in particular spans bases starting from position 27,455,514 bp from pter to position 27,466,601 bp from pter on the plus strand of chromosome 8 as defined by the build 37 of the human genome released on February 2009. The man skilled in the art can readily define other positions for the APOJ gene depending on the chosen count origin such as the centromere of the chromosome for instance. By way of example, the APOJ coding sequence (mRNA) is represented by the NCBI Reference Sequence NM_001831 (SEQ ID NO: 1 ) and the apolipoprotein J is represented by the NCBI Reference Sequence NP_001822 (SEQ ID NO: 2).

The "complement component receptor 1 gene (CR1)", or complement component (3b/4b) receptor 1 gene, sensu stricto is localized on chromosome 1 in the cytogenetic band 1 q32 and in particular spans bases starting from position 207,669,473 bp from pter to position 207,815,109 bp from pter on the plus strand of chromosome 1 as defined by the build 37 of the human genome released on February 2009. The man skilled in the art can readily define other positions for the CR1 gene depending on the chosen count origin such as the centromere of the chromosome for instance. By way of example, the CR1 coding sequence (mRNA) is represented by the NCBI Reference Sequence NM_000651 (SEQ ID NO: 3) and the CR1 protein is represented by the NCBI Reference Sequence NP_000642 (SEQ ID NO: 4).

As will be clear to one of skill in the art, the expression "determining whether the individual harbours at least one variant allele of a susceptibility gene selected from the apolipoprotein J gene (APOJ) and the complement component receptor 1 gene (CR1)" preferably refers to determining the presence or absence of at least one variation, such as a substitution (in particular a single nucleotide polymorphism (SNP)), a deletion (del) or an insertion (ins), at a site of polymorphism, i.e. a site of the genome which sequence is not identical in all individuals, comprised in said susceptibility gene. It is then deduced that the individual is at risk of developing Alzheimer's disease if the variation is present in the individual. As intended herein the "susceptibility gene selected from the apolipoprotein J gene (APOJ) and the complement component receptor 1 gene (CR1)" also encompasses sequences directly upstream and downstream of the APOJ and CR1 gene sensu stricto, and in particular spans bases starting from position 27,400,000 bp from pter to position 27,520,000 bp from pter on the plus strand of chromosome 8 as defined by the build 37 of the human genome released on February 2009 for

APOJ and bases starting from position 207,615,000 bp from pter to position

207,870,000 bp from pteron the plus strand of chromosome 1 as defined by the build

37 of the human genome released on February 2009 for CR1.

Where the susceptibility gene is APOJ, the invention preferably comprises:

- determining the presence or absence of at least one variation selected from the group consisting of:

- a C for the CVT single nucleotide polymorphism (SNP) rs2279590;

- a C for the CAT SNP rs11136000; and

- a G for the C/G SNP rs9331888;

in the individual;

- deducing that the individual is at risk of developing Alzheimer's disease if said at least one variation is present.

More preferably, it is deduced that the individual is at risk of developing Alzheimer's disease if it is determined that the individual presents a C for the C/T SNP, rs2279590, a C for the C/T SNP rs11136000, and a C or a G, preferably a G, for the C/G SNP rs9331888.

Besides, where the susceptibility gene is CR1, the invention preferably comprises:

- determining the presence or absence of at least one variation selected from the group consisting of:

- an A for the A/G SNP rs6656401 ; and

- an A for the A/G SNP rs3818361 ;

in the individual;

- deducing that the individual is at risk of developing Alzheimer's disease if said at least one variation is present.

More preferably, it is deduced that the individual is at risk of developing Alzheimer's disease if it is determined that the individual presents an A for the A/G SNP rs6656401 , and a G or an A, preferably an A, for the A/G SNP rs3818361. The above-defined variations are well known to one of skill in the art and are notably described in the NCBI database dbSNP (www.ncbi.nlm.nih.gov/SNP/)

As will be clear to one of skill in the art, a C/T, C/G or AZG SNP read on a first nucleic acid strand is respectively identical to a A/G, C/G or C/T SNP read on the complementary strand to said first nucleic acid strand.

For the purpose of unambiguously identifying the above variations, the sequences flanking the above-defined variations are notably represented by the following sequences wherein the variation is shown between square brackets:

- rs2279590

CTTCTGATAAGGAAGTCCTCCTGCT[CZT]CTCCAAGGAAACCTAGAGAGCTGACC (SEQ ID NO: 5)

- rs11136000

CACCAAAGCCACACCAGCTATCAAAA[CZT]TCTCTAACGGGCCCTTGCCACTTGA (SEQ ID NO: 6)

- rs9331888

AGAGCAAGAGGACTCATCCTTCCAAA[CZG]GGACTTTCTCTGGGAAGCCTGCTCC (SEQ ID NO: 7)

- rs6656401

ATTTCCTTCTCTGTCTCCATCTTCTC[AZG]TCGCCTTCTCCTCTGTGTGTGTCCT (SEQ ID NO: 8). - rs3818361

AGCCCTCTGGTAAGCATAAGATATA[AZG]CAAAGGAAATTGCCCCATATCTAACA (SEQ ID NO: 9)

Preferably also, in the above-defined method, it is deduced that the individual is at risk of developing Alzheimer's disease as defined above if said individual is homozygous for the at least one variant allele or for the at least one variation.

As intended herein the expression "homozygous" means that the at least one variant allele or the at least one variation is present for the two genomic copies of the gene. As intended herein, the expression "risk factor for Alzheimer's disease" relates to all characteristics of an individual, in particular of a genetic, environmental, or physiological nature, which are known to be predictive of Alzheimer's disease as defined above. Numerous risk factors for Alzheimer's disease are known in the art.

However, it preferred within the frame of the present invention that the at least one other risk factor for Alzheimer's disease is a variant allele of the apolipoprotein E gene.

Variant alleles of the apolipoprotein E gene constituting a risk factor for Alzheimer's disease as defined above are well known to one of skill in the art and notably encompass the ε4 allele of APOE, in particular as described by Farrer et al. (1997) JAMA 278:1349-1356.)

EXAMPLE Methods AD samples

All AD cases were ascertained by neurologists from Bordeaux, Dijon, Lille, Montpellier, Paris, Rouen, and were identified as French Caucasian. Clinical diagnosis of probable AD was established according to the DSM-III-R and NINCDS- ADRDA criteria (McKhann et al. (1984) Neurology 34:939-944). Written informed consent was obtained from study participants or, for those with substantial cognitive impairment, from a caregiver, legal guardian, or other proxy, and the study protocols for all populations were reviewed and approved by the appropriate Institutional review boards. Genomic DNA samples from 2,344 AD cases were available for analysis prior to genotype quality control steps.

Control samples

Controls were selected from the 3C Study (Alperovitch et al. (2003) Neuroepidemiology 22:316-325). This cohort is a population-based, prospective (4- years follow-up) study of the relationship between vascular factors and dementia. It has been carried out in three French cities: Bordeaux (southwest France), Montpellier (southeast France) and Dijon (central eastern France). A sample of non- institutionalised, over-65 subjects was randomly selected from the electoral rolls of each city. Between January 1999 and March 2001 , 9,686 subjects meeting the inclusion criteria agreed to participate. Following recruitment, 392 subjects withdrew from the study. Thus, 9,294 subjects were finally included in the study (2,104 in Bordeaux, 4,931 in Dijon and 2,259 in Montpellier). At 4 years of follow-up, three hundred and fifty individuals were demented, with 143 prevalent cases (91 cases of AD, 39 cases of mixed/vascular dementia, 3 cases of Parkinsonian dementia and 10 cases of other types of dementia) and 207 incident cases (135 cases of AD, 40 cases of mixed/vascular dementia, 15 cases of Parkinsonian dementia and 17 cases of other types of dementia). Prevalent and incident AD cases were included as cases. Patients with other types of dementia, and individuals for whom information on their dementia status during the 4-year follow-up was missing were excluded (n=854; refusal or lost to follow-up and death). At this stage, genomic DNA samples of 7,076 controls were available from the 3C study prior to genotype quality control steps. Supplementary control data on additional anonymised samples were used in some analyses as explained in the text below (Heath et al. (2009) Eur. J. Hum. Genet. 16:1413-29).

Genotvpinq

DNA samples were transferred to the French Centre National de Genotypage (CNG) for genotyping. First stage samples that passed DNA quality control were genotyped with lllumina Human 610-Quad BeadChips. Genotype data were retained in the study for samples that had been successfully genotyped for >98% of the SNP markers. SNPs with call rate <98%, with minor allele frequency (MAF) <1 % or exhibiting departure from the Hardy-Weinberg equilibrium in the control population (p<10-6) were excluded. 134 AD cases and 980 controls samples were removed because they were found to be first- or second-degree relatives of other study participations, or were assessed non-Caucasian based on genetic analysis using methods described in Heath et al. (2009) Eur. J. Hum. Genet. 16:1413-29. 537,029 autosomal SNPs genotyped in 2,032 cases and 5,328 controls were thus retained. In order to avoid any genotyping bias, cases and controls were randomly mixed when genotyping, and laboratory personnel were blinded to case/control status. Genotyping success rate was at least 95%, and no departure from Hardy-Weinberg equilibrium was observed for the markers included in the second stage.

Statistical analysis

The case and control differences were evaluated using a logistic regression, which optionally incorporated principal components that were significantly associated with disease status to account for possible population stratification as described in Heath et al. (2009) Eur. J. Hum. Genet. 16:1413-29 and Price et al. (2006) Nat. Genet. 38:904-909. It was hypothesized that the relatively high genomic control without the adjustment with principal components was due to differences in representation of various French regions in the case and control series. Therefore, the robustness of the conclusions were further explored through incorporation of 6,734 anonymised samples from France and other European countries as additional controls (Heath et al. (2009) Eur. J. Hum. Genet. 16:1413-29). With the inclusion of the additional samples, the genomic control parameter was 1.04 without adjustment with principal components, and 1.03 after adjustment. The inclusion of the additional controls did not substantially modify the association statistics for markers in regions showing suggestive evidence of association (p<10-5) after correction for population structure in the primary analysis. Statistical analyses were undertaken under an additive genetic model using logistic regression taking account of age, sex and disease status using SAS software release 9.1 (SAS Institute, Cary, NC). Population controls that were not genotyped specifically for this study were excluded, as were any samples with missing age or gender data. This gave a maximum of 2,025 AD cases and 5,328 controls. Information on age and gender in the cases and controls included in these analyses are shown in Table 1.

Table 1 : population characteristics

Interactions between APOJ or CR1 SNPs and APOE ε4 polymorphism were tested in logistic regression models adjusted for age, gender and centres. The solid spine haplotype block definition in Haploview 4.0 was used to generate linkage disequilibrium blocks of the genomic regions encompassing the APOJ or CR1 genes from imputed SNPs (MAF>5%) (Barrett et al. (2005) Bioinformatics 21 :263-5).

Associations of the APOJ and CR1 haplotypes were estimated using logistic regression or proportional hazards models using Thesias 3.0 which implements a maximum likelihood model and uses a SEM algorithm (Tregouet & Tiret (2004) Eur.

J. Hum. Genet. 12:971 -974). The population attributable risk fraction was estimated using the formula: PAR=F(OR-I )/(F(OR-1 )+ 1 ) where F is the frequency of the deleterious allele in the sample and OR the odds ratio of AD risk associated to the deleterious allele.

URLs

Haploview: http://www.broad.mit.edu/mpg/haploview/index.php

Revman: http://www.cc-ims.net/revman/ Thesias: http://ecgene.net/genecanvas/uploads/THESIAS3.1/

HapMap: http://www.hapmap.org

Results

Study of late-onset AD

A genome-wide association (GWA) analysis of 537,029 SNPs in 2,032 French AD cases and 5,328 French controls was undertaken. Patients with probable AD accord ing to DSM-III-R and NINCDS-ADRDA criteria were ascertained by neurologists. Individuals without symptoms of dementia from French Three-City (3C) prospective population-based cohort (described in Methods) were obtained as controls. The study samples were genotyped with lllumina Human 610-Quad BeadChip, and subjected to standard quality controls procedures (see Methods).

The resulting GWA data were then analyzed for association with a logistic regression taking into sex and age, and using principal components to adjust for possible population stratification as described (see Methods). Prior to adjustment for possible population stratification the genomic control parameter was 1.20, but after adjustment it became 1.04. The observed versus expected χ2 distribution (quantile- quantile plots) did not indicate substantial inflation of the test statistics after adjustment. Additional tests were undertaken to establish the robustness of the statistical results as described in the Methods.

Several APOE-linked SNPs gave strong evidence of disease association (Table 2).

Frequency (Reference allele)

supp.

rs Chr Pos (bp) cases controls controls Analysis 1 Analysis 2 ! Analysis 3

2965109 19 49917185 - - - 1.70E-09 1.60E-07 -

7254776 19 49919582 - - - 2,80E-09 1.50E-07 -

2927488 19 49923318 0,764 0,723 0,741 5,06E-07 5,42E-06 1.78E-05

2965101 19 49929652 0,691 0,634 0,653 6.87E-11 3,84E-09 8,54E-10

17728272 19 49932211 - - - 2.70E-07 3,10E-06 -

8100239 19 49944944 - - - 2,00E-09 3,60E-08 -

8103315 19 49946008 - - - 5.70E-07 5,10E-06 -

4803759 19 50019299 - - - 5,10E-07 2,20E-06 -

10402271 19 50021054 0,626 0,696 0,695 6,44E-16 1.90E-14 <2e-16

7408909 19 50023565 - - - 1.30E-14 7,90E-14 -

7359852 19 50027875 - - - <2e-16 <2e-16 -

2927480 19 50029225 - - - <2e-16 <2e-16 -

1871047 19 50043586 0,64 0,597 0,591 1.59E-06 9,41E-07 1.10E-07

1871046 19 50043777 - - - 1.60E-06 6,80E-07 -

4803763 19 50049131 - - - <2e-16 <2e-16 -

440277 19 50053064 - - - 8.80E-11 2,60E-09 -

12978931 19 50055540 - - - 7,80E-06 5,20E-06 -

6859 19 50073874 0,49 0,42 0,424 1.07E-13 1.01E-12 4,95E-15

11669338 19 50074824 - - - 1.70E-11 1.20E-09 -

11673139 19 50074877 - - - 1.70E-11 1.20E-09 -

3852861 19 50074901 - - - 9,50E-06 9,50E-06 -

3745150 19 50077599 - - - <2e-16 <2e-16 -

283813 19 50081014 - - - 2,90E-07 7,20E-07 -

6857 19 50084094 - - - <2e-16 <2e-16 -

157580 19 50087106 0,288 0,402 0,412 <2e-16 <2e-16 <2e-16

2075650 19 50087459 0,727 0,899 0,884 <2e-16 <2e-16 <2e-16

157583 19 50088513 - - - 4,40E-08 7,20E-06 -

8106922 19 50093506 0,672 0,57 0,598 <2e-16 <2e-16 <2e-16

10119 19 50098513 - - - <2e-16 <2e-16 -

405509 19 50100676 0,542 0,458 0,479 <2e-16 <2e-16 <2e-16

439401 19 50106291 0,28 0,391 0,391 <2e-16 <2e-16 <2e-16

4420638 19 50114786 - - - <2e-16 <2e-16 -

4803770 19 50119193 - - - 2,40E-16 2,60E-15 -

5112 19 50122120 - - - 1,40E-08 9,40E-07 -

12721108 19 50139081 - - - 1.10E-07 2,80E-06 -

Table 3: Association statistics from the GWA at the APOE locus. Markers are shown when p<10 ^"5 in analysis 1. Analysis 1 = GWA cases vs. study controls, logistic regression (age, gender adjusted); Analysis 2 = GWA cases vs. study controls, logistic regression with principal components adjustment; Analysis 3 = GWA cases vs. study and supplementary controls, logistic regression and principal components adjustment. Outside of APOE, one marker, rs11136000, within the apolipoprotein J gene

(APOJ) on chromosome 8p21 -p12 gave P =9.0x10 ^"8 in the association test. This slightly surpassed the criteria for genome-wide significance as evaluated with a conservative Bonferroni correction (P<9.3x10 ^~8 ). Markers in several chromosome regions were observed with suggestive evidence of association (P<10 ^~5 ) as shown in

Table 3. Finally, imputation of genotypes using the HapMap CEU samples to increase the number of SNPs examined in these regions was undertaken. The genotyped and imputed markers from the regions that gave P<10 ^"5 in the original case/control comparison are shown in Table 3.

Frequency (Reference allele)

supp. Analysis

rs Chr Pos (bp) cases controls controls Analysis 1 2 Analysis 3

2182912 1 205726967 - - - 5,30E-05 6,19E-05 -

4844597 1 205737892 - - - 4.70E-05 5,79E-05 -

4274065 1 205738099 - - - 4.70E-05 5,90E-05 -

9429945 1 205743391 - - - 8,99E-05 1.70E-04 -

6656401 1 205758672 - - - 6,40E-05 1.10E-04 -

3818361 1 205851591 0,217 0,179 0,199 2,35E-07 2,87E-06 5,96E-06

6701713 1 205852912 0,217 0,179 0,2 2,40E-07 2,89E-06 6,39E-06

2296160 1 205861943 - - - 2,90E-07 3,40E-06 -

1408078 1 205867178 - - - 2,90E-07 3,40E-06 -

4844610 1 205869175 - - - 8,89E-07 9,29E-06 -

1408077 1 205870764 0,212 0,176 0,198 3,68E-07 4.72E-06 9,05E-06

11952762 5 118383450 0,953 0,93 0,945 3,64E-07 6,62E-06 0,000144

11959554 5 118399338 - - - 3,30E-06 4,60E-05 -

12201301 6 31115536 0,969 0,982 0,974 1.46E-06 3,88E-06 4,96E-05

11768400 7 84893471 - - - 3,10E-05 7,00E-07 -

6465004 7 84894842 - - - 3,10E-05 7,10E-07 -

11762648 7 84915141 - - - 5,30E-05 1.70E-06 -

12704129 7 84917284 - - - 2,70E-05 7,91E-07 -

11769386 7 84921805 - - - 2,90E-05 8,20E-07 -

2462051 7 84937706 - - - 6,90E-04 1.20E-05 -

2462049 7 84937834 - - - 5,00E-04 1.10E-05 -

2463670 7 84942229 - - - 3,30E-06 1.80E-07 -

13231722 7 84953822 - - - 6,00E-04 2,40E-05 -

10499889 7 84959000 0,535 0,572 0,55 5,20E-05 4,54E-06 0,000305

6951823 7 84998691 - - - 3,10E-04 1.10E-05 -

10252600 7 85015874 - - - 3,70E-04 1.30E-05 -

12334143 7 85020952 - - - 8,00E-06 8,00E-07 -

12333397 7 85020998 - - - 1.40E-05 1.80E-06 -

3087554 8 27511359 - - - 1.00E-04 2,80E-04 -

2279590 8 27512170 - - - 5,00E-06 4,90E-07 -

11136000 8 27520436 0,346 0,389 0,379 1.32E-06 8,99E-08 1.94E-08

9331888 8 27524779 - - - 1.20E-06 3,10E-06 -

12931878 16 10949695 - - - 1.10E-05 1.70E-04 -

8055533 16 10949740 0,652 0,693 0,679 1.70E-06 1.25E-06 7,06E-06

4781028 16 10966864 - - - 8,59E-06 1.30E-04 -

9302457 16 10967338 - - - 7,50E-07 2,20E-06 -

16957843 16 10968706 - - - 9,71E-05 4,30E-04 -

16957849 16 10972808 - - - 7,40E-06 1.20E-04 -

Table 3: Results from the GWA for regions containing markers with p<10 in analysis 1. All markers with p<10-4 in analysis 1 are shown for these regions. Markers with allele frequencies absent and no results for analysis 3 have been imputed. Analysis 1 = GWA cases vs. study controls, logistic regression (age, gender adjusted) ; Analysis 2 = GWA cases vs. study controls, logistic regression with principal components adjustment; Analysis 3 = GWA cases vs. study and supplementary controls, logistic regression and principal components adjustment. APOJ

At APOJ on 8p21-p12, four SNPs (rs3087554, rs2279590, rs11136000 and rs9331888) were tested that showed evidence of association and were not in complete LD. Three of these markers (rs2279590, rs11136000 and rs9331888) exhibited statistically significant association with AD (Table 4). Strong evidence for association was found in the GWA datasets taking into account the sample origin in the logistic regression, with two markers exceeding the criterion for genome-wide significance (P<9.3x10-8). For the marker showing the strongest evidence of association (rs11136000), the odds ratio for the minor allele was 0.86 (95%CI 0.81- 0.90, P =7.5 x 10-9).

The effect of the APOJ locus taking into account APOE ε4 status in cases and controls was also examined (Table 4).

N MAF HW Association test

Cases Controls Cases Controls OR P value

(95% Cl)

rs2279590 2025 5328 0.36 0.41 3 .1.10 ^" ' 0.83 1 .0.10 °

(0.77-0.90)

rs11136000 2016 5266 0.35 0.39 6 .0.1O ¹ 0.83 1 .5.1 O*

(0.77-0.90)

rs9331888 2025 5328 0.31 0.28 8 .9.1O ¹ 1.19 1 .8.1O ⁵

(1.11-1.30)

Table 4: Association of SNPs at the APOJ locus with AD. P values and odds ratios (OR) with their 95% confidence interval (95% Cl) have been calculated under an additive model using logistic regression models adjusted for age, gender and centres when necessary. MAF = minor allele frequency; HW = P value for the test of Hardy Weinberg Equilibrium in controls.

A statistical interaction between the APOE ε4 status and the APOJ SNPs (ranging from 3.0x10-2 to 5.2x10-2 according to the SNP tested) was detected. For rs11136000, while the association was significant in both ε4 carriers and non- carriers, it was greater in carriers (OR= 0.81 , P = 2.7x10 ^"5 in carriers, OR=O.91 P = 7.0x10 ^"3 in non-carriers). Analysis of the GWA data showed that these markers are within a LD block that encompasses only the APOJ gene. The three APOJ locus markers define three common haplotypes (frequency>2%) that together account for 98.2% of the observations in controls. Compared to the most frequent TTC haplotype, the other two frequent haplotypes were all associated with a statistically significant increased disease risk (Table 5). The odds ratio was highest for the CCG haplotype compared to the TTC haplotype, with the following AD risk and association p-values: OR=1.28, P = 1.5x10 ^"7 .

Haplotypes Cases Controls OR (95% Cl) P values

lie 0 344 0 388 Ref.

CCC 0 334 0 329 1 .14 (1.04-1.24) 3 .0 10 ^"3

CCG 0 302 0 265 1 .28 (1.17-1.41 ) 1 .5 10 ^"7

Table 5: Association results for haplotypes at the APOJ locus. The results have been calculated using the Thesias programme with adjustment for age, gender and centre. The P value for the global association was 1.8.10 ^"6 . The markers as ordered from left to right (5' to 3' ) are rs2279590 , rs 1 1 1 36000 and rs9331888. Minor alleles are underlined. CR1

A second locus of potential interest lies within a LD block that encompasses CR1 on 1 q32. Two SNPs were tested at this locus (rs6656401 and rs3818361 ) which showed evidence of association with disease (Table 6).

N MAF HW Association test

Cases Controls Cases Controls OR P value

(95% Cl)

rs6654601 2025 5324 0.22 0.18 9.9.10 ^" 1.27 1.8.10 ^"

(1.16-1.39)

Rs3818361 2018 5324 0.22 0.18 8.5.10 ^" 1.28 8.5.10 ^" *

(1.17-1.40)

Table 6: Association of SNPs at the CR1 locus with AD. P values and odds rations (OR) with their 95% confidence interval (95% Cl) have been calculated under an additive model using logistic regression models adjusted for age, gender and centres when necessary. MAF = minor allele frequency; HW = P value for the test of Hardy Weinberg Equilibrium in controls.

At this locus, a statistical interaction with APOE ε4 status and risk of disease (P = 9.6.10 ^~3 ) was also detected with significant association in both carrier and non- carriers but stronger in the former. For rs6656401 , OR=1 .38 (95%CI 1 .22-1 .55) in carriers, OR= 1 .13 (95% 1 .04-1.23) in non-carriers was obtained. There was evidence association of the rs3818361 SNP in the APOE ε4 carriers (OR=1 .38 95%CI 1.19-1.60, P = 2.3.10 ^"5 ).

The genotyped markers define two principal haplotypes that account for 97.8% of the observations at the CR1 locus, while a third haplotype has an estimated frequency of 1.2% in the combined control population (Table 7). The odds ratio was highest for the AA haplotype compared to the GG haplotype, with the following AD risk and association p-values: OR=1.28, P = 1.4x10 ^"7 .

Haplotypes Cases Controls OR (95% Cl) P values

GG 0 .772 0 .813 Ref.

GA 0 .01 1 0 .009 1 .25 (0.87-1 .80) 2 .2. 10 ^"1

AA 0 .207 0 .170 1 .28 (1.17-1 .40) 1 .4. io- ⁷

Table 7 Association results for haplotypes at the CR1 locus.

The results have been calculated using the Thesias programme with adjustment for age, gender and centre. The P value for the global association was 7.5.10 ^"7 . The markers as ordered from left to right (5' to 3') are rs6656401 and rs3818361 . Minor alleles are underlined.

The attributable fraction of risk was calculated to be 25.5% for APOE, 8.9% for APOJ and 3.8% for CR1. If the estimate that 60-80% of the AD risk is due genetic factors is correct, additional genetic susceptibility loci still remain to be identified. This situation is similar to that of many other diseases in which loci have been successfully mapped by GWA.

In summary, in addition to the previously known APOE locus, new loci at APOJ and CR1 have been identified that are potentially associated with the risk of late-onset AD.