HK1183914A - Susceptibility genes for age-related maculopathy (arm) on chromosome 10q26 - Google Patents

Abstract

Allelic variations in the genes PLEKHA1 and LOC387715 are identified herein as risk factor for Age Related Maculopathy (ARM). A method is therefore provided for identifying a risk of development of ARM in an individual that comprises identification of allelic variations in PLEKHA1 and/or LOC387715. Related apparatus, such as an array, are identified as being useful in implementing those methods.

Description

Age-related macular degeneration (ARM) susceptibility gene on chromosome 10q26

The application is a divisional application with the application number of 200680027015.0 filed on 6/7/2006 and the name of 'age-related macular degeneration (ARM) susceptibility gene on chromosome 10q 26'.

The inventors of the present invention

Michael, B, Gaorui

Johnna Gacobbs Multi Titer

Jevte.P.Koehli

Denier E wilx

Dammar S ma me ferthie

Robert E Farrel

Background

Age-related macular degeneration, known as an age-related macular degeneration, is a major cause of central blindness in the elderly population, and numerous studies indicate that a strong potential gene constitutes this complex disease. A large number of potentially susceptible sites have been identified using genome-wide linked scans of large lineages, diseased sibling pairs, and recently phenotypically discordant sibling pairs. (Klein et al, 1998 Age-related macromolecular expression. clinical expressions in large family and link dictionary 1q. archives of expression vector 116: 1082. loops et al, 2000A functional expression vector for expression vector Genetics 9: 1329. Majewski et al, 2003Age-related macromolecular expression-a expression vector in extended expression vectors. am J. Hum. gene 73: 540. equation 550. scheck et al, 2003A expression vector for expression vector J. expression vector for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression vector expression for expression for expression vector expression for expression for expression for expression for expression for expression for expression for expression, 2004 partition of genomic data in extended polypeptides variants a major locations and oligogenic substoichiometric for-related macromolecular evolution, am J Hum Genet 74: 20-39; kenealy et al 2004 Linkage analysis for image-related volumetric production support a gene on chromosome 10q26.mol Vis 10: 57-61; schmidt et al 2004 Ordered subset linkage analysis subsystem for image-related volumetric generation on chromosome 16p12 BMC Genet 5: 18; weeks et al 2004 Age-related Macuppathy: a genome with connected evidence of susceptibility loci with the1q31, 10q26, and 17q25 regions, am jhoum Gemet 75: 174-189; scantangelo et al 2005A Discordint Sib-Pair Linkage Analysis of Age-Related molecular Generation, optical Genetics 26: 61-68) genome-wide cross-linking scan strongly showed that the 10q26 region contained the age-related macular degeneration (AMD) gene (Weeks et al 2004); this region has been supported by many other studies and is the top-level region in recent Meta analysis (Fisher et al 2005 Meta-analysis of genome scales generation. hum Mol Genet.2005 Aug 1; 14 (15): 2257-64). Recently, three articles on Science (Edwards et al 2005 comparative Factor H Polymorphism and Age-Related macromolecular integration. Science 308, 421-. These findings have been confirmed (Conley et al 2005 Canadiate gene analysis summary a roller for The facial acid biosynthesis and regulation of The compensation system in The environmental invention: hum Mol Gene 14: 1991. 2002; Hageman et al (2005a) From The Cover: A common signature in The compensation regulation gene H (HF1/CFH) compressed synthesis vector recovery. Proc. Natl Act. Sci. USA 102: 7227. 7232; and Zoarrival et al 2005 interaction analysis 402Y 402 in The simulation gene J52. amplification gene J52. circulation gene K52. J. supplement J. K. The invention is a method of using The method of The regulation gene in The environmental regulation gene of The growth gene K. A. The method of The regulation gene K.S. K. The invention is a method of The expression of The regulation gene expression of The molecular gene expression vector K. The invention is a method of The expression gene in The molecular gene K. A. The present application, A, The expression of. CFH has been suspected to function in ARM in Hageman and Anderson previous studies (Hageman and Mullins 1999 molecular location of related to fundamental graphics. molecular video 5: 28; Johnson et al 2000A molecular role for imaging complex in driving format. Experimental Eye Research 70: 441 and 449 comprehensive activity and expression of related processes in driving format and related molecular modeling. Experimental Eye Research 73: 887 and 896; Mullins et al 2000 driven with related graphics and related to specific graphics, molecular graphics 705: 20: 7 and 20: environmental graphics, Research, 2000, and Research, 2: 14, molecular graphics, 2, and 20. environmental, biological graphics, 2, 1, 2. However, CFII remains an independent problem until other genes responsible for ARM have been identified, suggesting that alternative pathways and inflammation are the pathogenesis of ARM, but not the only pathology observed in ophthalmology.

Disclosure of Invention

To this end, allelic variants, including single nucleotide polymorphisms, have been identified on chromosome 10q26, as described above. These allelic variants are shown to be associated with an increased risk of age-related macular degeneration. Allelic variants are located on chromosome 10q26 along with the LOC387715 and/or PLEKHA1 genes. In one embodiment, the allelic variant is associated with LOC 387715.

In one non-limiting embodiment of the invention, a method of identifying an individual at high risk for developing age-related macular degeneration is provided. The method comprises identifying the presence of an allelic variation in chromosome 10q26 that is correlated with risk of developing age-related macular degeneration in a nucleic acid sample derived from a subject. In a non-limiting embodiment, allelic variation occurs in PLEKHA1/LOC387715/PRSS11 on chromosome 10q26. For example and without limitation, an allelic variation is an allelic variation that occurs in one or both of PLEKHA1 and LOC387715, e.g., without limitation, a serine 69 alanine variation at LOC 387715.

In another non-limiting embodiment, the variation is one or more polymorphisms defined as rs4146894, rs10490924, rs1045216, rs1882907, rs760336, rs763720, rs800292, rs1483883, and rs 1853886. Allelic variation may be, without limitation, mutations that result in a non-functional gene product and variant expression of a gene product, such as one or more of a frameshift mutation, a promoter mutation, and a splice mutation.

In one embodiment, the method comprises further defining the presence of a complement factor H allelic variation, e.g., without limitation, a variation corresponding to the single nucleotide polymorphism rs1852883, in a nucleic acid sample derived from a subject.

This method can use any available technology, such as, without limitation: nucleic acid amplification methods, such as PCR, reverse transcription PCR (RT-PCR), isothermal amplification, Nucleic Acid Sequence Based Amplification (NASBA), 5' fluorescent nucleic acid methods (e.g., TAQMAN method), nucleic acid beacon methods, and rolling circle amplification. Such allelic variations may be identified using an array, which typically comprises one or more reagents for identifying the presence of allelic variations comprising two or more single nucleotide polymorphisms, rs4146894, rs10490924, rs1045216, rs1882907, rs760336, rs763720, rs800292, rs1483883 and rs1853886, in a nucleic acid sample derived from a subject.

In another non-limiting embodiment, an array is provided comprising one or more reagent sequences that identifies the presence of an allelic variation in chromosome 10q26 that is associated with a risk of developing age-related macular degeneration in a nucleic acid sample from a subject. Without limitation, allelic variation may occur in PLEKHA1/LOC387715/PRSS11 on chromosome 10q26 and, for example, may correspond to one or more single nucleotide polymorphisms defined as rs4146894, rs1045216, rs10490924, rs1882907, rs760336, rs763720, rs800292, rs1483883 and rs 1853886.

Drawings

FIG. 1: CIDR location and local genotype SNPs relative to the candidate gene. The sites, distances and nucleic acid sites on chromosome 10 were derived from NCBI Entrez-gene and SNP database.

FIG. 2: single-point and multi-point linkage results for chromosome 10. The left panel shows the results when all SNPs were used. The right panel shows the results of the analysis using only H-cluster SNPs. The peak labeled "F" represents a possible false peak due to high SNP-SNP LD, while the peaks labeled "G" and "P" correspond to sites including GRK 5and PLEKHA1, respectively. Horizontal line indicates multiple points S_all(S_all1 maximum) of the support interval.

FIGS. 3A-3D: linkage disequilibrium patterns for chromosome 10 based on 196CIDR SNPs and 179 non-relevant controls. FIG. 3A: false peak of 135cM (see fig. 2), fig. 3B: false peak of 142cM (see fig. 2), fig. 3C: and (4) linking peaks. FIGS. 3D-1 and 3D-2 are enlarged versions of FIG. 3C, originating from line A-A'. With the maximum S in the false peak_allSNPs of (FIG. 3A and 3B) are shown in grey while important SNPs on 5 genes (GRK5/RGS10/PLEKHA1/LOC387715/PRSS11) derived from CCREL (Table 5) in true linkage peaks are shown in grey. The grey shading indicates the significant LD between SNP pairs (dark grey squares without sign indicate pair D ' ═ 1), the white squares indicate evidence of no significant LD, while the grey squares without sign indicate pair D with a value of 1 without statistical significance, LD was determined using D ' and the numbers in the squares give pair LD as D ' × 100.

FIG. 4: the result of multipoint linkage of chromosome 1. The left panel shows the results when all SNPs were used, while the right panel shows the results of the analysis using only H-cluster SNPs. The peak of marker "F" indicates a false peak probably due to high SNP-SNP LD, while the peak of marker "C" corresponds to the CFH gene. Horizontal line indicates multiple points S_all(CFH-1 max S)_all) 1-unit support interval.

Fig. 5A-5C provide: linkage disequilibrium pattern of chromosome 1 based on 679CIDR SNPs and 179 non-relevant controls. FIG. 5A: false peak of 188cM (see fig. 4), fig. 5B: false peak of 202cM (see fig. 4), fig. 5C: linkage peak at CFH site. With the maximum S in the false peak_allSNPs of (B) are shown in grey and are derived from CCREL in the true linkage peak (Table)5) The important SNPs on CFH of (A) are shown in grey. The grey shading indicates the significant LD between SNP pairs (dark grey squares without signs indicate the pair D '═ 1), the white squares indicate evidence of no significant LD, and the grey squares without signs indicate the pair D' with a value of 1 without statistical significance. LD was determined using D 'and the numbers in the squares give the pair LD as D' 100.

Fig. 6A and 6B: FIG. 6A: linkage disequilibrium patterns of GRK5 (region 1), RGS10(SNP6), PLEKHA1 (region 2), LOC387715 (region 3), PRSS11 (region 4). FIG. 6B: linkage disequilibrium pattern of CFH (region 1). The grey shading indicates the significant LD between SNP pairs (dark grey squares without signs indicate the pair D '═ 1), the white squares indicate evidence of no significant LD, and the grey squares without signs indicate the pair D' with a value of 1 without statistical significance. LD was determined using D 'and the numbers in the squares give the pair LD as D' 100. The important S NPs in the CCREL allele trial were the highest in grey (see table 6). Three SNPs (rs6428352, rs12258692 and rs11528141) were excluded due to very low heterozygosity, while one SNPs, rs2736911, was excluded due to its no information. Indicating that these regions clearly show the location of the gene and not represent haplotype regions.

Fig. 7 shows the estimated rough ORs and 95% CIs of CFH, ELOVL4, PLKEHA1, and LOC387715 genes. One or both risk alleles (RR + RN) will be compared to the non-risk allele (NN) bulk homozygotes. The solid line represents 95% CI corresponding to OR (open loop). The dotted vertical line represents a null value with OR of 1. The vertical column gives the contrast value evaluated in the AREDS and CHS queues.

Fig. 8 shows the evaluated ORs and 95% CIs of CFH. A: dominant effect (CT + CC vs. TT) assessment of OR_dom. B: heterozygote (CT vs. TT) risk assessment OR_het. C: assessment of stealth Effect (CC vs. CT + TT)_rec. D: heterozygote (CC vs. TT) risk assessment OR_hom. The dotted vertical line represents a null value with OR of 1.

Fig. 9 shows the evaluated ORs and 95% CIs of ELOVL 4. A: dominant effect (AG + GG vs. AA) assessment of OR_dom. B: heterozygote (AG vs. AA) Risk assessment OR_het. C: assessment of the stealth Effect (GG vs. AG + AA)_rec. D: heterozygote (GG vs. AA) Risk assessment OR_hom. The dotted vertical line represents a null value with OR of 1.

Fig. 10 shows the evaluation ORs and 95% CIs of LOC 387715. A: dominant effect (GT + TT vs. GG) assessment of OR_dom. B: heterozygote (GT vs. GG) risk assessment OR_het. C: assessment of the stealth Effect (TT vs. GT + GG)_rec. D: heterozygote (TT vs. GG) risk assessment OR_hom. The dotted vertical line represents a null value with OR of 1.

Fig. 11 shows the estimated ORs and 95% CIs of PLEKHA 1. A: dominant effect (AG + AA vs. GG) assessment of OR_dom. B: heterozygote (AG vs. GG) risk assessment OR_het. C: AG + GG) assessment of the stealth Effect (AA vs. AG)_rec. D: heterozygote (AA vs. GG) risk assessment OR_hom. The dotted vertical line represents a null value with OR of 1.

FIG. 12 provides estimates ORs and 95% CIs derived from a fusion-analysis data set including Y402H in CFH, and the estimates aggregated from fixed and random effects models. Data display OR at the top_het(OR of CT heterozygotes versus TT) and data at the bottom shows OR_hom(OR of CC heterozygotes versus TT). 'Hage-C' and 'Hage-I' denote Columbia and Iowa cortsD estimates derived from Hageman et al (Hageman, G.S et al (2005) A common hash in the complex regulatory gene factor H (HF1/CFH) predisposes to age-related macromolecular expression. Proc Natl Acadsi USA 2005 May 17; 102 (20): 7227-32.Epub 2005 May 3) 'Jakobs' denotes estimates derived from Jakobsdotir et al (Jakobsdotir, J et al (2005) summary genes for expression-related genes 407-related mammothy on moosomoto 10. Gene J.389 J.407). By "fixed" it is meant that the aggregated estimates from all trials assume that variability between studies is due to chance. By "random" is meant that the aggregated assessment from all experiments is considered in the studyHeterogeneity. "n" is_AMD. "is the total number of ARM cases included in the evaluation value, and" n_con"total number of controls without ARM included in the evaluation value. The dotted vertical line represents the point assessment value ("fixed") of the OR aggregated under homogeneity.

Fig. 13 provides: a: genotype frequencies (%) in the case of non-related ARM in the meta-analyzed array including Y402H in CFH. B: genotype frequency (%) of non-relevant controls without ARM in a study involving meta-analysis of Y402H in CFH. "Hage-C" and "Hage-I" represent the evaluation values from Columbia and Iowa arrays from Hageman et al, and "Jakobs" represents the evaluation values from Jakobsbottir et al, respectively.

Fig. 14 provides estimates ORs and 95% CIs derived from a fusion-analysis data set including S69A in LOC387715, and the estimates aggregated from fixed and random effects models. Data display OR at the top_het(GT heterozygote vs. GG OR) and data at the bottom shows OR_hom(OR of GG heterozygotes versus TT). "Jakobs" indicates the evaluation value derived from the article by Jakobsdottir et al (Jakobsdottir, J et al (2005) characterization genes for age-related genetic on chromosome 10q26.am J Hum Genet.77, 389-407). By "fixed" it is meant that the aggregated estimates from all trials assume that variability between studies is due to chance. By "random" is meant that the pooled estimates from all experiments take into account heterogeneity in the study. "n" is_AMD"is the total number of ARM cases included in the evaluation value, and" n_con"total number of controls without ARM included in the evaluation value. The dotted vertical line represents the point assessment value ("fixed") of the OR aggregated under homogeneity.

Fig. 15 provides: a: genotype frequencies (%) in the case of non-related ARM in the meta-analyzed array comprising S69A in LOC 387715. B: genotype frequency (%) of non-relevant controls without ARM in a study involving meta-analysis of S69A in LOC 387715. "Jakobs" refers to an evaluation value derived from the article by Jakobsbottir et al.

FIG. 16 provides the amino acid (SEQ ID NO: 19) and nucleic acid sequence (SEQ ID NO: 20) of LOC 387715.

Detailed Description

Allelic variants, including single nucleotide polymorphisms, have been identified on chromosome 10q26, as described below. The allelic variants shown here are associated with a high risk of developing age-related macular degeneration. Example 1 identified PLEKHA1 and/or LOC387715 as being ARM associated allelic variant sites. Further studies, as shown in example 2, confirm and further support the correlation of LOC387715 variants as ARM markers. The association of LOC387715 variants as ARM markers has not been ruled out, but recent evidence from genetic studies has shown more strongly that the variants occur in the LOC387715 gene.

Thus, methods for identifying individuals at high risk for developing age-related macular degeneration are provided. The method comprises identifying P LEKHA1 and/or LOC387715, in one non-limiting embodiment LOC387715, in a sample of nucleic acid from a subject, allelic variations or the occurrence of a particular haplotype (including several allelic variations). Specific Single Nucleotide Polymorphisms (SNPs) have been identified at these sites, including, without limitation, these SNPs identified as rs4146894, rs1045216, rs10490924, rs1882907, rs760336, and rs 763720. The method further comprises identifying an allelic variant in complement factor h (cfh), e.g., a non-limiting allelic variant as defined as rs 1853883.

As used herein, "allelic variant" refers to a variation in the original amino acid sequence of one or more alleles of a nucleic acid, typically a subject, e.g., a patient. Allelic variations include single or multiple nucleotide and amino acid substitutions, additions or deletions with any effect on protein expression, including but not limited to: promoter activity to regulate transcription, frameshifting, early protein termination, protein misfolding, selective protein processing, disruption (or enhancement) of the active site or binding site of a protein, missplicing of mRNA or any other characteristic of a nucleic acid or protein that affects the expression and/or function of the final gene product. Variations in an amino acid and nucleic acid sequence may be non-silent, i.e., non-phenotypic effects, such as disease threats, may be associated with sequence variations. Allelic variation, on the other hand, may be a variation of a putative "wild-type" nucleic acid or amino acid sequence that is believed to be implicated, correlated or associated with disease risk, such as ARM, by the exemplary and non-limiting statistical methods described herein. Thus, the LOC387715 single nucleotide polymorphism rs10490924(Ser69Ala) is an allelic variation.

A number of methods, including high throughput methods, can be used to determine SNPs and/or other allelic variations, such as, without limitation, PCR and restriction fragment length polymorphism methods as described in the following examples. In one embodiment, the SNP or small allelic variation is identified by sequencing (resequencing) the DNA in the sample by any method. Many methods of resequencing are known in the art, including high-throughput methods. Amplification-based methods can also be used to identify allelic variations, such as SNPs, including, without limitation: PCR, reverse transcription PCR (RT-PCR), isothermal amplification, Nucleic Acid Sequence Based Amplification (NASBA), 5' fluorescent nuclease methods (e.g., TAQMAN methods), molecular beacon methods, and rolling circle amplification. Other methods, such as RFLP, which are suitable and effective for identifying restriction fragment length polymorphisms of different alleles, can also be used. The method may be multiplexed, meaning that two or more reactions may be performed simultaneously at the same physical location, such as at the same location in the same tube or array-as long as the reaction products of the multiplexed reactions are distinguishable. As a non-limiting example, the TAQMAN or molecular beacon approach can be compounded by using and monitoring the aggregation or depletion of different fluorescent fuels corresponding to two different sequence-specific probes. In many cases, the appropriate method is dictated by personal choice and experience, existing equipment and reagents, the need for high throughput and/or complex methods, expense, accuracy of the method, and the skill level of the technician operating the method. The design and implementation of these techniques are widely-known and within the average ability of those skilled in the art.

In the practice of the methods provided herein, an array may be used. Arrays are particularly useful when performing high-throughput methods. The array typically includes one or more reagents, such as, without limitation, nucleic acid primers and/or probes, to identify the occurrence of allelic variation in a human subject nucleic acid sample corresponding to one or more single nucleotide polymorphisms at LOC387715 and/or PLEKHA1, e.g., without limitation, SNPs identified below: rs4146894, rs1045216, rs10490924, rs1882907, rs760336, rs763720, rs800292, rs1483883, rs1853883 and rs 1853886. The array can simultaneously test and identify one or more allelic variations on LOC387715 and/or PLEKHA1, e.g., without limitation, SNPs identified as follows: rs4146894, rs1045216, rs10490924, rs1882907, rs760336, rs763720, rs800292, rs1483883, rs1853883 and rs1853886, and simultaneously determining allelic variation on CFH, other genes/loci and control genes/loci/nucleic acids.

The term "array" as used herein refers to reagents that facilitate the identification of allelic variations in a gene that are localized at two or more identified locations. In one embodiment, the array is a device having two or more separate, defined reaction chambers, e.g., without limitation, 96-well plates, in which reactions involving defined components are performed. In one exemplary embodiment, two or more nucleic acid primers or probes are immobilized on a substrate in a spatially addressable manner such that each individual primer or probe is immobilized at a different and (addressable) defined location on the substrate. Substrates include, without limitation, multi-well plates, silicon wafers and silicon beads. In one embodiment, the array comprises two or more sets of beads, each set of beads having a defined label, such as a quantum dot or fluorescent label, such that the beads can be individually identified using, for example, without limitation, a flow cytometer. In one embodiment, within the scope of the presently described, the array may be a reaction chamber comprising two or more reaction chambers with primers that amplify DNA to identify SNPs or probes that bind specific sequences. Thus, reagents, such as probes and primers, can be bound or deposited at specific locations on the array. The agent may be in any suitable form, including, without limitation: solution, drying, freeze-drying or vitrification.

Useful array technologies include, for example and without limitation, Affymetrix GeThe Array may be, for example,resequencing arrays (commercially available from California, Santa Clara, Affymetrix Inc.) and similar technologies. Analysis of array data and/or information identifying genetic risk factors from data obtained from patient samples and/or statistical software or other computer-implemented programs are known in the art.

As used herein, "an agent for identifying an allelic variation occurring in a nucleic acid sample of a subject," wherein the allele is specifically identified or identified in a gene or site, refers to an agent for identifying a particular allelic variation by any suitable method, such as, without limitation, by PCR, resequencing 5' exonuclease (TaqMan) method and/or array or high-throughput method. Non-limiting examples of such reagents include sequence-specific primers, primer pairs and probes used in any useful method system. The primers and probes can be in any useful form, but are typically nucleic acids, but can be nucleic acid analogs, such as, without limitation, phosphorothioate.

In example 1, a family-based cross-linking study and a disease-control related study were performed using high-density SNP panels at two cross-linked regions of 1q31 and 10q26. SNP cross-linking and correlation results for chromosome 1q31 confirmed that the cross-linking peak and the strongest ARM association are located on the CFH gene. The family and disease-control data for chromosome 10q26 were analyzed to identify the next major ARM susceptibility-related gene.

Example 2 describes the following studies in which subjects recovering from cardiovascular studies (CHS), a population-based cohort that did not use age-related macular degeneration (ARM) status as a study factor, and an age-related eye disease study (AREDS), a population-based cohort that used ARM status as a study factor, were subjected to nested disease-control design. These cohorts were used to study CFH, PLEKHA1, LOC387715 and ELOVL4 genes in ARM susceptibility in two cohorts with different study strategies. In addition, these two cohorts included an additional 1 and 4 disease-control studies in the meta-analysis of OC387715, respectively. CFH in both cohorts was significantly correlated with ARM case (p < 0.00001) and meta-analysis confirmed that risk alleles in both heterozygous and homozygous cases (OR, 2.4 and 6.2; 95% CI (confidence interval), 2.2-2.7 and 5.4-7.2, respectively) provided this susceptibility. LOC387715 in both cohorts was significantly correlated with the ARM case (p < 0.00001) and meta-analysis confirmed that the risk allele in both heterozygous and homozygous cases (OR, 2.5 and 7.3; 95% CI, 2.2-2.9 and 5.7-9.4, respectively) provided this susceptibility. PLEKHA1, which is closely related to LOC387715, is significantly related to the ARM case in the AREDS queue rather than the CHS queue, while ELOVL4 is significantly unrelated to ARM in both queues. This study provides further support for CFH and LOC387715 genes in ARM susceptibility by evaluating arrays with different research strategies for ARM status and further support for meta-analysis.

Example 1

Materials and methods

Family and disease-control arrays

A total of 612AMD families and 184 unrelated controls were sent to the genetic disease research Center (CIDR) for genotyping. Our data is limited to analysis of caucasian groups only due to the possibility of ethnic base. The caucasian group had 594 AMD families, including 1443 genotyped individuals, and 179 unrelated controls. The caucasian family includes 430 genotype-related consanguinity pairs, 38 genotype-related primary father pairs, and 52 genotype-related first generation cousin pairs.

A total of 323 caucasian families, 117 peak unrelated controls, and 196 unrelated diseases also partially genotyped additional SNPs. The partial groups included 824 genotypic individuals, 298 genotypically associated Primary kindred pairs, 23 genotypically associated Primary father pairs and 38 genotypically associated first generation cousin pairs. PdeStats in Merlin package (Abecasis et al, 2000) can be conveniently used to profile family data.

Disease condition model

Three classical models (A, B and C) are used to define the severity of the ARM case (Weeks et al, 2004). In the simplest case, this is limited to the "type a" disease that is the most severe and conservative diagnosis of interest. Only non-relevant controls that were unaffected in all three diagnostic models were used. Unaffected are those individuals who had eye care records and/or fundus photographs showing no changes in any macula (including drusen) or a small number (less than 10) of hard drusen (50 microns diameter or less) without any other RPE changes. Individuals with a high number of extramacular drusen do not remain unaffected when there is available information.

To account for the specific ARM subtype, only those cases that are terminal to the disease, with choroidal neovascular membranes (CNV) demonstrated in either eye, or regional atrophy (GA) in either eye, were of selective interest. A large number of individuals are reported to have regional atrophy and CNV, however it will be problematic because in these cases it is difficult to know whether regional atrophy is secondary to damage to the CNV or from treatments that limit CNV growth (e.g., laser, surgical, or photodynamic treatments). Patients with both diseases are included in the CNV group because it is difficult to discern in the photograph or record whether the patient's eye GA precedes the occurrence of CNV. However, only a subset of the overlapping groups was included in the regional atrophy group, particularly where regional atrophy was reported in one eye without evidence of CNV. Table 1 shows the number of individuals in the three case groups. These studies may rule out a small fraction of individuals in the regional atrophy group who had asymmetric regional atrophy prior to CNV development in the same eye or symmetric regional atrophy already when CNV developed in both eyes.

Table 1 subtype distribution of advanced ARM patients. The numbers in parentheses refer to individuals who both have CNV and GA and are included in the GA group (see selection criteria) for probability and attribution risk assessment and related trials.

Pedigree and typing errors and data processing

The Pendeheck program (O' Connell and Weeks 1998 PedCheck: the program for analyzing genotype incompatibility in cross-linking analysis Am J Hum Genet 63: 259-266) was used to verify Mendelian contradiction. It is very difficult to determine which genotype is wrong in a small family (Mukhopadhyay et al, 2004 synthetic stuck methods for genetic error determination. hum-Hered 58: 175-189), all genotypes being set in each deletion problem marker in the family containing Mendelian contradictions. Mega2(Mukhopadhyay et al,http://watson.hgen.pitt.edu/register) Used to build files for cross-linking analysis and to assess allele frequencies by gene-counting.

Allele frequency and Hardy Weinberg equilibrium

Allele frequencies in the cross-linking assay were assessed by direct counting from non-relevant and non-affected controls. All controls were unaffected in all three disease condition models. The genotype spouse of the absent child, or a child not part of the study, was incorporated into the study controls. Accurate Hardy Weinberg equilibrium was performed as Mega2 (Mukhopadlyay et al 2005 Mega 2: data-handling for malfunction genetic linkage and association analysis systems. bioinformatics) for our SNPs.

Mendel Version 5(Lange et al, 2001 MENDEL Version 4.0: A complete package for the exact genetic analysis of the variant tracks in the genetic and position data sets. am J of Hum Genet 69 (Supplement): A1886) was also used directly to assess the allelic frequency of the family data, Mendel completely elucidating the relevance of subjects in assessing allelic frequency. Since members of the genotype family are mostly diseased, these estimates are quite close to those obtained using non-related diseases.

Genetic map

Rutgers combined cross-linking-physical mapping (version 2.0) (Kong et al, 2004A combined linkage-physical map of the human genome. AmJ Hum Genet 75: 1143-. Because the distribution of our SNPs in the desired region is very dense, the estimated recombination between several SNPs is zero; the reset in these cases is set to 0.000001. Physical sites of all our SNPs were obtained from the NCBI dbSNP database (human construct 35).

Cross-linked unbalanced structure

Disregarding high cross-linking imbalances (LD) between SNPs when performing cross-linking assays can lead to false positive results (Schaid et al, 2002 fraction on molecular biology with a software that is associated with the linkage equality. am J Hum Gent 71: 992. 995; Huang et al, 2004 Igaring linkage disequilibrium amplitude linkage markers-positive observation of linkage for infected site pair analysis. am JH Gent 75: 1106. 1112). Efforts to consider high SNP-SNP LDs include:

1. the LD structure in the non-relevant control was studied using the H-cluster method (Rinaldo et al, 2005 Characterization of multiple linkage disequilibrium. genetic Epidemiology 28: 193- "206), which operates with R (RDelevation Core Team 2004R: A language and environment for statistical calculation. R Foundation;" Rn for Statistical Computing，Vienna，Austria.ISBN 3-900051-07-0，R statistical software website，http://www.r-roject.org/). The objective was to determine the unimodal-labeled SNPs (htSNPs) in the cross-linking assay. The method uses hierarchical clustering to pool highly correlated SNPs. After clustering, selecting htSNP in each cluster by using an H-cluster method; htSNP is the SNP in the cluster that is most relevant to all other SNPs. The SNPs are selected such that each SNP has at least one htSNP correlation coefficient (r)²) Greater than 0.5;

2. the program, Haplo View (Barrett et al 2005, Haplovie of LD and haplotype maps. bioinformatics 21: 263-265.) was used to obtain SNP-SNP LD graphical displays on chromosomes 1 and 10; and

3. haplotype-based correlation analysis was performed using two and three-SNP moving windows (see below).

Cross-linking analysis

1. Single point analysis. As in our previous study (Weeks et al, 2004), LOD scores were single (disease allele frequency 0.0001 and penetrance vector 0.010.900.90 ═]). Due to ARM overt and late onset, only two diseases overt were used: "influence in model A" and "unknown". Calculating LOD values (HLOD) of variables under heterogeneity while using linear S_allStatistical calculation model-free LOD values. Both values were calculated using Allegro (Gudbjartsson et al, 2000A llegro, a new computer program for multipoint linkage analysis. NatGenet 25: 12-13.).

2. Multipoint analysis of cross-linking imbalance was omitted. Since the internal-labeling distance is usually very small, the LD between SNPs can be very high and violates the LD-free assumption obtained by many cross-linking analysis procedures. Allegro was used for multiple point analysis to ignore LD (Gudbjartson et al, 2000). HLODs and S_allStatistics are all calculated.

3. Multipoint analysis using htSNPs. When LOD values were calculated using only htSNPs, the number of SNPs on chromosome 1 and chromosome 10 was reduced to 533 and 159. Multi-point LOD analysis was performed as above (Weeks et al, 2004). Neglected SNPs fit well with the SNP-SNP LD structure evaluated by Haplo View (Barrett et al, 2005).

Correlation analysis

In order to allow all cases in the family to be used, a new CCREL program (Brown et al, 2005, Case-Control single-marker and regenerative analysis of genetic identity 28: 110-122) was employed, which tested for relevance by using both relevant and non-relevant controls. CCREL was used to analyze SNPs at the cross-linking peaks of chromosomes 1 and 10 to determine correlation. The CCREL assay focuses on bio-related subjects by calculating the effective number of cases and controls. For these analyses, non-relevant controls were given a "normal" phenotype, while family members not affected by "type a" ARM were given an "unknown" phenotype (CCREL approach has not been able to use both relevant cases and relevant controls). An effective number of controls for each SNP used in a correlation test is the number of controls that genotype define the SNP. One allele test, one unimodal test using two SNP sliding windows, one unimodal test using three SNP sliding windows and one genotype test were performed. Analysis was performed according to the authors' use of the CCREL R package (Browning et al, 2005).

GIST analysis

To investigate the most critical alleles/SNPs for cross-linking signal, a genotype-IBD share test (GIST) was performed between chromosome 1 and 10 cross-linking peaks using the local genotype SNPs and important SNPs derived from the CCREL test. The GIST assay determines whether an allele or LD allele is part of the observed crosslinking signal (Li et al, 2004). Weights for each affected relationship were calculated for three different disease models (implicit, explicit, additive) -these weights were unbiased under the null hypothesis of no disease-marker correlation. The basis of the experimental statistics is the correlation between the family weight variables and the nonparametric linkage (NPL) values. Because GIST testing is currently only applied to affected sibling family members, family members are classified into their member core families prior to calculating NPL values. Since the following disease models are unknown, we performed the assay under three different disease models (recessive, dominant, additive), applying p-values appropriate for multiplex assays to the three models, and taking mixed values.

Triple analysis

The analysis was performed in three consecutive steps. First, a set of data that has been genotyped with CIDR is analyzed. Second, 8 additional SNPs were locally genotyped in chromosome 10PLEKHA1/LOC387715/PRSS11 region and the local-genotype dataset was analyzed. Note that all known non-synonymous SNPs in PLEKHA1 on all PRSS11 areas were studied. The two sets of data differ in size and composition and are easily analyzed separately (table 2). Allele frequency assessment, CCREL correlation test, and GIST test were performed on these (overlapping) data sets as described above. Third, we tested the interaction of the chromosome 1 and chromosome 10 regions and determined whether the difference in risk was a factor of geographic atrophy or choroidal neovascular membranes.

TABLE 2 statistical analysis per section and sample size summary

Part 1: analysis of CIDR SNPs

CIDR SNP genotyping

To identify the response gene on chromosome 10q26, the genetic disease research Center (CIDR) performed high-density custom SNP genotyping for the 612AMD family and the 184 non-relevant controls, including 199SNPs at 13.4Mbp (26.7cM) on the desired region. 196SNPs were used for analysis: three of these were ignored due to the lack of polymorphism in the controls (examined in family data, missing alleles were very rare and only present in heterobinders). 684SNPs comprising 45.7Mbp (47.1cM) on chromosome 1q31 were also genotyped; the five SNPs were either absent or very rare and only present in heterobinders in the family data due to the absence of polymorphisms in the controls that were ignored-missing alleles. See Table 3 for correlations between marker alleles and actual alleles, and non-synonymous SNPs, amino acid changes.

Table 3 allelic markers. For each study marker, allele signature, amino acid changes in non-synonymous SNPs, allele frequency for CIDR control (179) and local control allele frequency (117 overlapping CIDR control) and HWE p-value for the actual trial.

SNP	Alleles	Label (R)	Amino acids	CIDR control	Local contrast	HWEP value
							rs6658788	A	1		0.511	0.483	0.58
	G	2		0.489	0.517

[0081]

SNP	Alleles	Label (R)	Amino acids	CIDR control	Local contrast	HWEP value
							rs1538687	A	1		0.693	0.658	0.41
	G	2		0.307	0.342
							rs1416962	T	1		0.648	0.607	0.44
	C	2		0.352	0.393
							rs946755	T	1		0.656	0.620	0.70
	C	2		0.344	0.380
							rs6428352	T	1		0.997	0.996	1.00
	C	2		0.003	0.004
							rs800292	A	1＝Ile	Ile62Val	0.232	0.269	0.82
	G	2＝Val		0.768	0.731
							rs1061170	T	1＝Tyr	Tyr402His		0.690	0.26
	C	2＝His			0.310
							rs10922093	G	1			0.295	0.66
	A	2			0.705
							rs70620	T	1		0.173	0.150	0.28
	C	2		0.827	0.850
							rs1853883	G	1		0.511	0.568	0.45
	C	2		0.489	0.432
							rs1360558	A	1		0.397	0.389	0.70
	G	2		0.603	0.611
							rs955927	T	1		0.609	0.615	0.85
	A	2		0.391	0.385
							rs4350226	A	1		0.905	0.897	0.34
	G	2		0.095	0.103
rs4752266	A	1		0.777	0.774	0.18
								G	2		0.223	0.226
rs915394	T	1		0.813	0.791	1.00
								A	2		0.187	0.209
rs1268947	G	1		0.883	0.885	0.65
								C	2		0.117	0.115
rs1537576	G	1		0.567	0.581	0.35
								C	2		0.433	0.419
rs2039488	T	1		0.885	0.885	0.01
								C	2		0.115	0.115
rs1467813	T	1		0.293	0.295	0.66
								C	2		0.707	0.705
							rs927427	A	1		0.464	0.487	0.10
	G	2		0.536	0.513
							rs4146894	A	1		0.466	0.474	1.00
	G	2		0.534	0.526
							rs12258692	C	1＝Pro	Pro233Arg		1.000	-
	G	2＝Arg			0.000
							rs4405249	T	1			0.158	1.00
	C	2			0.842
							rs1045216	G	1＝Ala	Ala320Thr		0.573	0.46

[0082]

SNP	Alleles	Label (R)	Amino acids	CIDR control	Local contrast	HWEP value
								A	2＝Thr			0.427
rs1882907	A	1		0.813	0.816	0.76
								G	2		0.187	0.184
rs10490923	G	1＝Arg	His3Arg		0.859	0.39
								A	2＝His			0.141
rs2736911	C	1＝Arg	Arg38Ter		0.881	1.00
								T	2＝Ter			0.119
rs10490924	G	1＝Ala	Ser69Ala		0.807	0.21
								T	2＝Ser			0.193
rs11538141	A	1＝Glu	Gly54Glu		0.995	1.00
								G	2＝Gly			0.005
rs760336	T	1		0.520	0.526	0.58
								C	2		0.480	0.474
rs763720	A	1		0.212	0.226	0.79
								G	2		0.788	0.774
rs1803403	T	1＝Cys	Cys384Gly		0.030	1.00
								G	2＝Gly			0.970

Part II: analysis of locally typed SNPs

The eight additional SNPs on chromosome 10, including the three susceptibility genes, PLEKHA1(rs12258692, rs440524 and rs1045216), LOC387715(rs10490923, rs2736911, rs10490924) and PRSS11(rs1153715, rs1803403), were also genotyped. This genotyping included all non-synonymous SNPs that had been reported in the NCBI database (see FIG. 1). As part of another study (Conley et al, 2005candidate gene analysis provided the fatty acid biosynthesis rules and the adjustment of the complement system in the etiology of age-related macular degeneration), two CFH variants (rs10922093 and rs1061170) were genotyped and used here. Genotyping of additional SNPs at the site GRK5/RGS10 is ongoing. Genotyping data for rs12258692, rs1803403 and the newly defined SNP, 3' one base of rs4405249, rs12258692, collected by sequencing (Rexage n Corporation, Seattle, WA) and analyzed with Sequencher software (GeneCodes Corporation, Ann Arbor, MI). Genotyping data for rs11538141, rs2736911, rs10490923 and rs10490924 was collected using RFLP. Suitable primers, amplification conditions and restriction enzymes are shown in Table 4, where SNPs are genotyped by sequencing or RFLP.

Table 4 genotyping by sequencing or RFLP, primers for gene data collection, annealing conditions and restriction enzymes. No application of NA

The genotyping data for rs1045216 was analyzed using a 5' exonuclease Assay-on-Demand TaqMan Assay (Applied Biosystems Incorporated, La Jolla, Calif.). Amplification and genotyping were performed using ABI7000 and SDS2.0 software (Applied Biosystems). Two non-related CEPH samples were genotyped for each variable and included each gel and each TaqMan plate to determine the internal stability of the genotype calls. In addition, double-blind genotyping was performed for each variable, and comparison was performed using either raw data or re-genotyping to locate each differential point. See table 3, where correlations between the allele signatures and actual alleles are provided, as well as, for non-synonymous SNPs, amino acid changes.

Part III: interaction and Odds Ratio (OR) analysis

Non-related cases-CIDR did not genotype non-related cases, but 196 non-related cases were genotyped for locally additional SNPs. To calculate the odds ratio and cross-analyses (see below), a "type A" group of affected patients from each family was selected to construct a non-relevant disease group. The 321 local-genotype family has at least one "type A" affected patient. If a family has more than one "A" affected patient, it will be selected from a population that is genotyped primarily at rs800292(CFH), rs1061170(CFH), rs1537576(GRK5) and rs4146894(PLEKHA 1); if there is no significant difference in the number of genotyped SNPs between the two individuals, young disease-producing individuals will be selected, otherwise the "A" affected disease will be randomly selected from the population that is most genotyped and occurs earliest. 577CIDR family has at least one "A" affected patient, 321 of these families are also genotyped locally, and the selected "A" affected patients are the same as in the local group. The remaining 256 families, based on the same criteria described above, only used rs800292(CFH), rs1537576(GRK5) and rs4146894(PLEKHA1) to find the most fully genotyped patients.

Analyzing CFH interactivity

The interactivity of CFH with genes on chromosome 1 and chromosome 10 can be assessed by GIST whether SNPs on CFH contribute to the crosslinking signal of chromosome 10 and SNPs on chromosome 10 contribute to the crosslinking signal of chromosome 1. By using the weights of SNPs derived from one chromosome and family-based NPLs on the other.

According to North et al (2005) logarithmic degradation application of disease-control related studies involving two causal sites, Hum Hered 59: 79-87, log degradation can also be used to evaluate the tested interactivity of the different interaction models. In this application, a number of interactive models of different possibilities, taking into account both the additive and the main effects of the two sites, will compare the relevant possibilities of the different models in order to pick the most similar and sparing model. As previously described (North et al, 2005), suitable models include the MEAN model, which evaluates only the average case, the ADD1, ADD2 and ADD models, which evaluate the additional effects at one or the other or two sites, the DOM1, DOM2 and DOM models, which additionally integrate the main effects, and two models, ADDINT and DOMINT, which consider the interaction effects (see North et al (2005) for more details). Since some models are paired non-nested, the Akaike Information Criteria (AIC) comparison will be used; in this study, the model with the lowest AIC was considered the most suitable and economical. These analyses were performed using the program provided by North and its collaborators. After improvement of some procedural defects that have already been found; the results were double-checked with our own R program. To minimize sample size, CIDR SNPs within high LD with highly important non-synonymous SNPs within each gene were selected. CIDR SNP rs800292 was chosen to represent rs10611710 (Y402H variant of CFH), CIDR SNP rs4146894 represents rs1045216 within PLEKHA 1. Similarly, representative CIDR SNPs within GRK5, RGS10, PRSS11 were selected.

Magnitude of correlation

Crude odds ratios were calculated in each gene and the ascribed risks for SNPs were evaluated. Alleles that are frequently present, at least in controls, are considered risk alleles. The attributable risk was assessed using the equation AF ═ 100 × P (OR-1)/(1+ P (OR-1), where OR is the odds ratio and P is the frequency of risk alleles in the population assessed from the control,. The "type A" influencers compared to the control, CNV subjects compared to the control, and GA subjects compared to the control were used.

Multiple assay protocol

To explore strong evidence of ARM-susceptible sites on the chromosome 10q26 region in previous studies, the analysis performed here focused more on assessing the sites of susceptible genes, except for hypothesis testing. Multiple assay protocols are crucial and relevant to the hypothesis test. In the evaluation, the focus of these studies is simple when the signal is very strong. In any case, any correction of the multiple tests did not rank the results of the steel whip. Bonferroni correction, which since LD does not care about any inter-assay correlation, 196 assays at the 0.05 level will result in an important limit of 0.05/196 to 0.00026; larger margins will result due to the LD dependence.

Results

The analysis was carried out in three sequential steps. First, a set of data genotyped at CIDR was analyzed. Second, an additional 8 SNPs from the PLEKHA1/LOC387715/PRSS11 region on chromosome 10 were locally genotyped, and we subsequently typed the local-genotyping data set. Allele frequency assessment, Hardy-Weinberg equilibrium (Table 3), CCREL correlation test, and GIST test were performed on these (overlapping) data sets as described above. Third, the interactivity of the chromosome 1 and chromosome 10 regions was analyzed and the difference whether the risk of geographic atrophy or choroidal neovascularization was functional was tested.

Part I: analysis of CIDR SNPs

CIDR Cross-linking results

As our previous studies used CIDR SNPs and applied the same cross-linking analysis (Week et al, 2004), the Sall multipoint curve peaks on chromosome 10 indicated the GRK5 region ("G" in FIG. 2; rs1537576 in GRK5 had a single-point Sall of 1.87 and a maximum single-point Sall value of 3.86 in rs555938, and GRK5 centromere of 206kb), but some evaluated the two-point non-parametric Sall LOD score and our highest xenogenous LOD score (HLOD), which focused on the PLEKHA1/LOC387715/P RSS11 region ("P" in FIG. 2). In this region, SNP rs4146894 within PLEKHA1 had a two-point Sall of 3.34 and the highest two-point HLOD of 2.66, while SNPs rs760336 and rs763720 within PRSS11 had two-point Sall of 2.69 and 2.23, respectively. However, the support spacing was large (10.06cM, fig. 2), and the localization derived from cross-linking analysis alone was not tight.

Comparison of the multi-point scores of all SNPs (FIG. 2, left panel) with those calculated with htSNPs alone did not take into account the effect of SNP-SNP LD. Two of the peaks in H-tandem SNPs disappeared almost completely when only H-tandem SNPs were used (referring to the wrong peak, "F" s, fig. 2, left part); interestingly, these two peaks were present in the haplotype region (FIGS. 3A and 3B) and the LD around the highest multi-and two-point LOD scores was lower (FIGS. 3C, 3D-1 and 3D-2), indicating the importance of focusing on LD when performing the cross-linking analysis.

Chromosome 1 cross-linking resulted in three peaks with Sall greater than 2, only one peak was obtained when the analysis was restricted to htSNPs (fig. 4). The retained peaks span the complement factor h (cfh) gene and include two SNPs with very high two-point Sall and HLOD values; rs800292, a non-synonymous SNP on CFH with Sall of 1.53 and HLOD of 2.11, SNP rs1853883, 165kb telomere of CFH with Sall of 4.06 and HOLD of 3.49. These results strongly support the discovery of CFH's lesions in early ARM (Conley et al, 2005; Edwards et al, 2005; Hageman et al, 2005b A common signatures in the complex regulatory gene factor H (HF1/CFH) pre-dispersed derivatives to related cellular generation. Proc Natl Acad Sci USA; Haines et al, 2005; Klein et al, 2005; Zaarersi et al, 2005 a). The missing peak ("F" s in fig. 4, left part) was localized to the strong haplotype region (fig. 5A and 5B) and the LD under the CFH peak was relatively low (fig. 5C) as can be observed in the cross-linking analysis using all of our SNPs.

CIDR correlation results

To obtain a finer localization than cross-linking, correlation analysis was performed (which was successfully used to study CFH on chromosome 1). Here, the CCREL method was used for correlation analysis, which can be performed by appropriately adjusting the correlation of the disease while using our non-correlated control and all related families. In the CIDR sample on chromosome 10, within our cross-linking peak, we found that the cross-linking of 3 genes: four additional SNPs with very small p-values (rs4146894, rs1882907, rs760336 and rs763720) for PLEKHA1, LOC387715 and PRSS 11. The strongest CCREL result on chromosome 10 is within the PLEKHA1 including SNP rs4146894 (table 5). A moving window haplotype analysis ("haplotype 3") using three SNPs simultaneously achieved very small p-values across the entire PLEKHA1 to PRSS11 region (table 5). The correlation test produced some moderately small p-values in the region of GRK5, which is the strongest evidence of cross-linking.

CCREL was performed on 56 SNPs spanning the cross-linking peak on chromosome 1, and two highly important SNPs (rs800292 and rs1853883) on CFH were found (table 5). The discovery of a strong association of early ARM with CFH was further supported by the very low p-value across the entire CFH gene obtained by a moving window haplotype analysis using both two ("haplotype 2") and three ("haplotype 3") SNPs (table 5).

CIDR GIST results

In GIST experiments on the CIDR data set, the two smallest p-values (0.006, 0.004) occurred in the GRK5/RGS10 region on chromosome 10q26, while the third smallest p-value (0.008) occurred in PLEKHA1 (Table 5). All four SNPs on GRK5 gene have small GIST p-values. GIST results indicate that GRK 5and PLEKHA1 are critical for the cross-linking signal on chromosome 10, and that CFH is critical for the cross-linking signal on chromosome 1. Both SNPs on PRSS11 are independent of the crosslinking signal on chromosome 10. There is no evidence that these two signals on chromosome 10 correlate with the cross-linking signal observed for chromosome 1.

Part II: local genotype SNP analysis

Local analysis result

After local typing of additional SNPs, both allelic and genotypic tests produced very small p-values in all three genes PLEKHA1/LOC387715/PRSS11 (Table 6). A moving window haplotype analysis ("haplotype 3") using three SNPs simultaneously achieved very small p-values across the entire PLEKHA1/LOC387715/PRSS11 region (Table 6). Thus, when the correlation suggests the LEKHA1/LOC387715/PRSS11 region, it is among these genes and is for differentiation.

Table 5 CCREL, GIST, and allele frequency assessments were performed at CIDR for typing family (594) and control (179). The frequency of the smaller allele in the control was reported in both the control (assessed by calculation) and the family (assessed by Mendel version 5), with the allele frequencies shown in bold when the allele frequencies between the control and the family differ by more than 0.1. The P-values for the allele test, haplotype 2SNP moving window test, haplotype 3SNP moving window test and the genotyping test for CCREL are shown in bold at ≦ 0.05 and in bold and underlined at ≦ 0.001. GIST P-values derived from chromosomes 1 and 10 evaluated using NPL are reported and indicated in bold when less than 0.05, and shown in bold and underlined at ≦ 0.001.

Table 6 CCREL, GIST, and allele frequency assessment of the locally-typed family (323) and control (117). The frequency of the smaller allele in the control was reported in both the control (assessed by calculation) and the family (assessed by Mendel version 5), with the allele frequencies shown in bold when the allele frequencies between the control and the family differ by more than 0.1. The P-values for the allele test, haplotype 2SNP moving window test, haplotype 3SNP moving window test and the genotyping test for CCREL are shown in bold at ≦ 0.05 and in bold and underlined at ≦ 0.001. GIST P-values derived from chromosomes 1 and 10 evaluated using NPL are reported and indicated in bold when less than 0.05, and shown in bold and underlined at ≦ 0.001. The italicized SNPs are locally typed SNPs.

Local GIST results

Of the three genes PLEKHA1/LOC387715/PRSS11, GIST suggested PLEKHA1 more strongly (Table 6). LOC387715 also produced a small p-value (rs10490924), but this SNP had a high LD with the PLEKHA1 SNPs (see FIG. 6A). Similar to the non-significant results obtained with the large CIDR samples above, GIST did not produce any significant results in PRSS11 when using the local typing dataset. Indicating that PLEKHA1 (or sites with strong LD) is most associated with AMD, while LOC387715 is possible.

To fairly assess which SNPs are involved in cross-linking signals within a region, NPLs were calculated using only the local-genotyping family. A direct comparison of the PLEKHA1/LOC387715/PRSS11 results and GRK5/RGS10 results in Table 6 is provided. The GRK5 GIST results were also expected in the local typing data set, with suitably small p-values of the same order as those obtained using GISTs in CFH (Table 6). However, the p-value is not as small as that seen when analyzing the CIDR data set. Since the SNPs of all GRK5 regions are CIDR SNPs, the difference is a separate function of sample size, and the local-typing dataset is smaller than the CIDR dataset (see table 2).

Part III: interaction and Odds Ratio (OR) analysis

GIST results

No strong evidence of interaction between the chromosomal 1 and 10 regions was observed in the GIST assay. When the CIDR dataset was used to test whether SNPs on chromosome 10 contribute to the cross-linking signal for chromosome 1 (table 5 "GIST (NPL 1)"), only rs763720 on PRSS11 gave a p-value of less than 0.05, whereas rs763720 did not contribute significantly to the cross-linking signal for chromosome 10, making this p-value less convincing. When only local data sets are used, one GRK5 variable (rs1537576), which is unimportant in larger CIDR data sets, is given a p-value of less than 0.05. Similarly, there is no evidence that SNPs in CFH contribute to the cross-linking signal on chromosome 10, only one SNP (rs955927) gives a p-value of less than 0.05, however this SNP is not within CFH and has no strong LD with any SNPs in the CFH gene (see fig. 6B).

Results of logistic regression

Logistic regression showed that an additional model comprising two variables derived from CFH and PLEKHA1 was the best model to predict disease-controlled conditions; both genes were shown to be important for ARM phenotype. The AIC standard also gives that an additional model including additional interactivity conditions is the next best model (table 7), whereas the interactivity conditions are not important (p-value 0.71). Similar results for the interactivity between CFH and PRSS11 were obtained, where an additional model including two variables appeared to be the best model. The GRK5/RGS10 region, a model with only CFH SNPs, was the most appropriate model, indicating that the addition of GRK5 or RGS10 to the model did not improve the prediction of CFH genotyping disease-controlling events.

Table 7 results from a logistic regression fitting the two-point model. AIC for each model, difference in AIC of the most suitable model. Herein defined for the model.

Odds ratio and risk of cause

Evaluating the magnitude of the correlation by calculating an Odds Ratio (OR) and an Attributable Risk (AR); the significant correlations obtained (table 8) are consistent with the CCREL test results of sections I and II. The two most important SNPs of the PLEKHA1/LOC387715 region are SNPsrs4146894(PLEKHA1) and rs10490924(LOC 387715); these two assays have a high correlation due to the very high LD between these SNPs (D' ═ 0.93) (see fig. 6A). The third most important SNP in the chromosome 10 region (rs1045216) is a non-synonymous SNP within PLEKHA1 and has a high LD between rs4146894(D '═ 97) and rs10490924 (D' ═ 0.91).

We obtained similar results and similar OR and AR values as already reported in the CFH gene (table 8). The three most important SNPs are rs1061170(Y402H variant), rs800292 (in CFH) and rs1853883 (with strong LD with rs1061170, D' ═ 91).

The magnitude of the correlation observed in PLEKHA1/LOC387715 was very similar to the level of correlation observed between CFH and ARM; both sites had very low p-value results (p-value < 0.0001). The OR and AR values are also very similar, with 5.29% dominant OR in CFH (95% CI 3.35-8.35) and 5.03% dominant AR in PLEKHA1/LOC387715 (95% CI 3.2-7.91), 68% and 57% dominant AR in CFH and PLEKHA1/LOC387715, respectively.

Sub-type analysis

We assessed odds ratios and attributed risks for exudative disease versus control, and geographic atrophy versus control (table 9). Odds ratios and corresponding p-values yielded similar findings to the CCREL allele assay (tables 5and 6). We found that the odds ratio in the presence of geographic atrophy or choroidal neovascular membranes was not primarily different.

Our cross-linking studies of the ARM family have established the identification of the chromosome 1q31 and chromosome 10q26 sites, and several other additional sites. Multiple cross-linking studies have repeated this finding and employed focused SNP analysis using the ARM family and unrelated affected individuals and controls. Chromosome 1q31, the gene identified a strong correlation with CFH that has been reported by others (see Conley et al, (2005)), and for the first time showed key SNPs on CHF for cross-linking signals. Interestingly, our minimum GIST p-value (< 0.001) was at rs1853883, which has a high D' value of 0.91 with the Y402H variable, rather than the hypothetical "disease-related" Y402H variable. This raises the possibility that other potential ARM-related variables in the CFH gene are of interest and have a high LD with Y402H.

Our studies of chromosome 10q26 have suggested two potential sites, a very strong-cryptic site, including three highly cross-linked genes, PLEKHA1, LOC384415, and PRSS11, and a weaker-cryptic site, including two genes, GRK 5and RGS10 (FIG. 1). GIST analysis does not support PRSS11 as an ARM-related gene, but does not completely exclude it as a potential candidate. PLEHKA1 had the lowest GIST-derived p-value while LOC387715 had the SNP with the strongest correlated signal and the highest odds ratio. The high linkage imbalance between the SNPs of LOC387715 and PLEKHA1 allowed clear differentiation of these genes from statistical analysis. However, the magnitude of the effect of PLEKHA1/LOC387715 sites on ARM was comparable to that observed in the CFH site. Similar to recent studies in science (Edwards et al, 2005; Haines et al 2005; Klein et al, 2005), we have found that the odds ratio of the CFH allele (heterozygous OR homozygous) in our disease-controlling population is 5.3OR (CI: 3.4-8.4) while the population is due to a risk of 68%. In the same manner, the odds ratio of the high-risk allele at PLEKHA1/LOC387715 locus was 5.0 (CI: 3.2-7.9) and the high-risk allele was 57% when both heterozygous and homozygous individuals were considered. As discovered by Klein et al (2005), odds ratios determined in disease control studies are often over-assessed for balance-related risks needed to determine lifelong risks.

With respect to Complement Factor H (CFH) on chromosome 1, the association data is very compelling for a single gene, even CFH is within the relevant gene region. In addition to the relevant data found in multiple independent groups, there are other biological data including CFH, including single sites of vitreous deposition in ARM patients. Thus, we must also consider the biological relevance of the potential ARM-susceptible genes identified in our study for chromosome 10q26.

As explained above, GIST analysis strongly suggests PLEKHA1, especially when additional non-synonymous SNPs are included in the gene analysis. PLEKHA1(GenBank NM-001001974, NM-021622, NP-001001974 and NP-067635; MIM 607772; UniGene Hs.287830) encodes the protein TAPP1 which has 404 amino acids and carries the phosphatidylinositol 3, 4, 5-triphosphate-binding motif (PPBM) and two Plectstrin Homologies (PH) regions. The last 3C-terminal amino acids have been predicted to interact with one or more 13PDZ regions of MUPP 1 (similar to the PDZ region within PRSS 11). Dowler and coworkers (Dowler et al, 2000 Identification of pleckstrin-homology-domain-conjugation proteins with novel phospho-side-binding specificities. biochem J351: 19-31.) have shown that the complete TAPP1 and C-terminal PH regions act specifically through phosphatidylinositol 3, 4-bisphosphate (PtdIns (3, 4) P2), but not any other phosphoinositides. TAPP1, which has 58% identity with the first 300 amino acids of TAPP2, shows a 5-fold higher affinity for PtdIns (3, 4) than TAPP2, and can eliminate this affinity by mutating the conserved arg212 to leucine within the PPBM region (part of the second PH region). The best-defined role of TAPP1 (and its related species, Bam32 and TAPP2) is lymphocyte activator. PtdIns (3, 4) are preferentially recruited to the cell membrane when lipid phosphatase (SHIP) is activated together with PI3KS (phosphatidylinositol 3-kinase). SHIP dephosphorylates PIP3 to form PtdIns (3, 4) P2. SHIP is a negative regulator of lymphocyte activity, so TAPP1 (and TAPP2) may be important negative regulators and negative regulators of mitogenic signals and PI3K signaling pathways. Thus, consistent with the hypothesis of ARM being closely linked to the inflammatory process, the role of PLEKHA1 and its protein TAPP1 in the eye can be predicted by monitoring local lymphocyte activity.

However, we still need to determine the biological authenticity of the other two candidate genes in the locus, LOC387715 and PRSS 11. The biology of LOC387715(Genbank XM 373477 and XP 373477; UniGene Hs.120359) is rarely known, except that it has been shown to be restricted to expression in the placenta. Our reverse transcription experiments on human retinal RNA confirmed the expression of PLEKHA1 and PRSS11, but i did not determine the expression of LOC387715 in the retina under standard conditions, although we confirmed its expression in placental RNA (data not shown). However, we cannot exclude the possibility that LOC387715 is expressed at very low levels in retinal or retinal pigment epithelial cells, or in non-ocular tissues, such as dendritic cells or migratory macrophages, may be a factor in ARM pathogenesis.

PRSS11(GenBank NM 002775 and NP 002766; MIM 602194 and UniGene Hs.501280) is a gene of the mammalian HtrA (high temperature requirement A) serine protease family with a highly conserved C-terminal PDZ region (Oka et al, 2004 HtrA1 serine protease inhibitors signaling medium by TGFfbeta family protein development 131: 1041- & 1053). These secreted proteases were originally identified for their homology to bacterial types required for survival at high temperatures and for low temperature chaperone activity. ATP-independent serine protease activity is thought to reduce protein misfolding at high temperatures. The mammalian form, HtrA1, unlike HtrA2, has been shown to be selectively stimulated by type III collagen alpha 1C hemiproteins. (Murwandoko et al, 2004 Binding of protein domains regulating the PDZ domains specific activity of HtrA1 serine protease. biochem J381: 895-904). Type III collagen is the major component on bruch's membrane (35-39% of total collagen) and is present in the retinal microvascular basement membrane. Further studies have reported the universal expression of HtrA1, but with temporal and spatial specificity consistent with the regions in which TGF-beta proteins exhibit regulatory effects. (De Luca et al, 2004Pattern of expression of HtrA1 during mouse reduction. J Histochem Cytochem 52: 1609. sup. 1617.) Oka and colleagues (Oka et al, 2004 HtrA1 serine protein inhibiting peptides dimensional proteins development by Tgfbeta family proteins. development 131: 1041. sup. 1053.) have shown that HtrA1 inhibits signals from many TGF-beta family proteins, including Bmp4, Bmp2 and TGF-beta1, possibly by inhibiting the receptor activity required for HtrA1 molecular protease activity. One clue for the potential importance of ARM association stems from the study by Hollborn et al (2004) of the opposing effects of mRNAs to cytokines of cell cycle-and ECM-associated proteins in hRPE cells in vitro. Curr Eye Res 28: 215-223 in vitro assays it was found that the presence of elevated transcription levels of TGF-beta1 and TGF-beta2 and collagen III and collagen IV reduces human RPE cell proliferation. Typically, elevation of collagen III will activate HtrA1 and result in a secondary inhibitory effect of TGF-beta 1. However, if serine protease activity is reduced (due to reduced synthesis or non-functional mutations), this regulatory pathway will be disrupted, resulting in an excessive reduction in RPE cell proliferation potential, possibly leading to RPE atrophy or to further changes in ARM development. The progressive decrease in soluble type III collagen in bruch's membranes was observed with age, partially accounting for the general decrease in HtrA1 activity in aging of individuals.

Both PRSS11 and PLEKHA1 were expressed in the retina, SAGE encapsulation of the central and peripheral retina (GEO expression data), indicating high levels of transcription of both genes in the central macula (PLEKHA1 was more abundant than PRSS 11). Multiple studies (GEO signature reports) have shown that the expression of PLEKHA1 is significantly reduced in various cell types when exposed to specific validation factors. P RSS11 has been studied as part of a microarray expression analysis comparing oxidation-threatened dermal fibroblasts in normal and ARM patients. In this study, half of the ARM samples (9/18) had lower levels of Htra1 expression than any normal samples. The low level of Htra1 in the non-ocular tissues of ARM patients indicates that it is a substantial difference in patients compared to normal individuals, and not a resultant or degenerative change in the eye.

Some evidence supports the GRK5/RGS10 site. Our S_allThe multipoint curves directly span GRK5, our largest single-point S_all3.86(rs555938) is only 206kb from the GRK5 centromere. The p-values of the GIST analysis of GRK5/RGS10 CIDR data were 0.004 and 0.006, which is less than the SNP p-value (0.005) in PLEKHA 1. Using our local-genotyping sample, the GIST p-value at GRK5 site was 0.012, comparable to the p-value of the Y402H variant found in CFH (p ═ 0.011). However, the CCREL analysis is not very important for GRK5 SNPs and the odds ratio is least important.

Based on biological evidence, GRK5(Gen Bank NM 005308he NP 005299; UniGeneHs.524625; MIM 600870; and PharmGKB PA180) is a possible ARM candidate gene with the function of regulating chemotactic neutral response and interaction with Toll4 receptor (Haribabu and Snayderman 1993 identity protocols of human G-protein-coordinated receptor enzyme multiplex activity. Proc. adaptor USA 90: 9398. 9402; Fan and Malik 2003 Toll-like receptor-4(TLR4) signaling influences expression-induced gene hybridization synthesis cell expression of genetic expression. native. model 9: 78. 55. related to ARM 5955. see No. 5. 3. the related application No. 2. 3. 2. 4. the related to human tissue receptor, Zibambient expression of genetic expression. 4. see. 3. the related to. 2. 4. the related to. environmental expression of human expression of interest. 3. 2. 4. the related to. environmental expression of human receptor, 2. the same, 2. 3. the related to. the related application No. 3. 2. the related to the related application No. 3. the related to the invention, 2. 3. 2. the related to the. Retinal or RPE expression of GRK5 is not a causal specific association as it may be expression and function of GRK5 in migrating lymphocytes and macrophages, which is crucial in the role of the ARM pathogenic immune/inflammatory pathway. The strongest GIST result occurs at rs2039488, which is actually localized between GRK 5and RGS10, at the 3' end of both genes. Some other SNPs on the GRK5 gene also had small GIST p-values, while the RGS10 SNP had non-significant GIST p-values. However, we cannot completely rule out the possibility of having SNPs in RGS10 that are strongly cross-linked imbalanced with rs 2039488.

RGS10(GenBank NM 001005339, NM 002925, NP 001005339 and NP 002916; UniGene Hs.501200; and MIM 602856) is a member of the G protein-coupled receptor family, which has been implicated in chemoattractant-linked lymphocyte oratz et al, 2004 Regulation of chemokine-induced lymphocyte pathology by TGS proteins enzymes Enzymol 389: 15-32.) and its expression in dendritic cells (which have been identified in an ARM-related vitreous bed) is regulated by a Toll-like signaling pathway (Shi et al, 2004 Toll-like signaling indicators of the expression fo regulator of G protein signaling proteins in dendritic cells: j Immunol 172: 5175-5184). Expression of TGS10 and GRK5 in both AMD and control subject oxidation-compression fibroblast microarray studies showed minor fluctuations in the samples, but no significant differences between the control and affected cases. This unnecessarily reduces the potential of these genes in ARM, as dermal fibroblasts lack cell populations with TGS 10-and/or GTK 5-related protein regulation.

We have focused on the potential interactivity between the high-risk allele within PLEKHA1/LOC387715 and the GRK5/RGS10 site of CFH on chromosome 1. This may be the first report using GIST to determine these interactions, and we found no evidence that SNP data on chromosome 10 could occupy NPL data of chromosome 1. In contrast, we found no correlation between the NPL data of chromosome 10 and the SNP data of the CFH allele. Logistic regression analysis also did not identify interactivity and showed that simple additional risk models were the simplest. We have performed some initial logistic analysis, including exposure to smoking. These analyses were performed due to previous suggestions on the relationship between smoking and complement factor H biology (Esparza-gordillo et al, 2004 Genetic and environmental factors inducing the human factor H plasma level. immunology 56: 77-82.) and our previous studies found the relationship between smoking and chromosome 10q26 locus (Weeks et al, 2004). From the data, we did not find a connection between western and CFH or PLEKHA1/LOC387715 sites, but we still investigated possible interactions and different model strategies for GRK5/RGS10 sites. We also determined the association of ARM subtypes with SNPs on chromosomes 1 and 10 (Table 9). We found that there was no major difference in the odds ratio in the presence of geographic atrophy or choroidal neovascular membranes, suggesting that these ARM sites may be responsible for the usual pathological pathways that trigger either end-stage form of the disease. This does not exclude the possibility of it being a genetic locus for other, as undescribed, geographic atrophy or a defined specific risk of CNV development.

Summarizing, these SNP-based cross-linking and correlation studies show the ability and limitations of these approaches to identify causative alleles and potential ARM susceptibility genes. These genetic studies let us determine that the gene responsible for the disease and its variants, whether or not they are tissue-specifically expressed by high density SNP genotyping, we have narrowed down candidate genes in the cross-linked peak found on chromosome 10q26 from hundreds of original GRK 5and PLEKHA1, but we cannot completely rule out the possibility of RGS10 and/or PRSS11 and LOC387715 effects. Additional genotyping of non-synonymous 3' SNPs within the GRK5 gene may help to further distinguish GRK5 from RGS10, but it does not establish a causally determined form. Repeats of other studies (such as the CFH case) may focus on a single gene, but there is still a clear possibility that we cannot obtain further decisions from the correlation studies, or clearly determine whether there are more than 2 ARM susceptibility genes on chromosome 10q26. However, molecular biologists can study the potential role of each candidate gene in the mouse ARM model and localize the causal role of the disease pathology.

Example 2 investigation of example 1

This example provides additional data that supports and identifies the conclusions and findings provided in example 1, where allelic variants within PLEKHA1 and the putative LOC387715 gene were defined as risk factors for age-related macular degeneration.

The etiology of ARM is complex, with both environmental and genetic susceptibility contributing. Correlation-based analysis is generally more sensitive to small genetic effects than cross-linking-based analysis and is more valuable for mapping of disease-related genes. (Cordell et al (2005) Genetic association students Lance. 366, 1121- "1131.) the study of disease-control associations using unrelated individuals is advantageous over family-based studies, particularly the contemplated use of a multiple-location Genetic model (Howson et al, (2005) the Comparison of position and family-based methods for Genetic association analysis in the presentation of interaction logic. Gene expression.29, 51-67.Risch et al (2001) the analyses of multiple incidence of gene-analysis students. ORPol. Bill.60, 215-" 220.), however these studies have potential sensitivity to the study of disease and control groups. For this reason, it is valuable to evaluate candidate genes in a population with different research programs. This example investigates the complement factor h (cfh) gene, the extended chain fatty acid-like 4(ELOVL4) gene, the PLEKHA1 gene, and the hypothetical LOC387715 gene in two discriminatory cohorts.

The correlation of the CFH gene with ARM susceptibility has been established in samples of European and American descent (Edwards et al, (2005), Haines et al, (2005), klein et al, (2005), Hageman et al, (2005), Conley et al, (2005), Zareprsi et al, (2005) and samples from the United kingdom-Sepp, T. et al (2006) complement factor H variant Y402H is the major risk impairment of geographic atrophy and choroidal neovascularization in smokers and smokers Invest Ophthalmol Vis Sci.47, 536-smokers 540, Germany-Rivera et al (2005) the hypothesis that LOC387715 is the second major susceptibility gene for age-related macular degeneration, independently of the risk of disease providing complement factors H.hum Mol Genet.14, 3227-3236, France-Souied et al (2005) the age-related complement factor Y402H, France-related macular degeneration in the population (1140) Morwear-related to exudative age-related macular degeneration, France-Val H1135, 1140, France-Rivers-1135, Iceland-Magnusson et al (2006) CFH Y402H confirm a similar risk of soft drusen and two severe forms of AMD. PloS med.3, e5. and japanese-Okamoto et al (2006) age-related macular degeneration complement factor H polymorphisms in the japanese population. Mo vis.12, 156-158.

Three studies supported that the PLEKHA1/LOC387715 site on chromosome 10q26 (Rivera et al (2005), Jakobsdottir Jr et al (2005) and Schmidt et al (2006) smoking strongly altered the LOC387715 and the age-related macular degeneration association Am J Hum Genet.78, 852-864, Jakobsdottir et al (2005) age-related macular degeneration susceptibility gene on chromosome 10q26 AmJ Hum Genet.77, 389-407 reported that PLEKHA1/LOC387715 site was significantly associated with ARM status, whereas the strong and inexpensive imbalance between PLEKHA1 and LOC 715 in the independent family-based and disease-control population used in the studies meant that one gene would not be determinable across another gene (Jakobsdottir, et al, (2005) since Jakobstir et al had applied the gene, it was hypothesized to be more susceptible to Rivera et al (Rivera et al) via evidence that Rivera et al had issued similar to ARM conditions, (2005) and Schmidt et al, using family basis and disease-control. All three studies indicate that the association of the region on chromosome 10q26 with the ARM status is independent of the association of CFH, which has been previously reported in all three populations (Haines et al (2005, condey et al, (2005), river et al, (2005)). furthermore, based on the studies of Schmidt et al, it was shown that the effect of the LOC387715 site can be altered by the smoking history, Schmidt et al (2006).

Two studies have evaluated the potential role of ELOVL4 in human ARM. Ayyagari et al, (2001) evaluate age-related macular degeneration patients for the ELOVL4 gene. Opthalmic Genet.22, 233-. However, Conley et al found that ELOVL4 has a significant link to ARM status in our familial and sporadic disease-control analyses Conley et al (2005). Differences between these studies may be related to the characteristics of exudative ARM disease in each population, as Conley et al found that ELOVL4 was specifically associated with the exudative subtype. These results indicate that additional studies are required to establish or disprove the relationship between ELOVL4 and ARM.

The two cohorts used in this study were Cardiovascular Health Studies (CHS), a population over 65 years old that did not factor in the role of ARM status studies-basal cohort, Fried et al (1991). Cardiovascular health study: design and basic principles. Ann epidemic.1, 263-276 and the age-related eye disease study (AREDS) in a cohort of 55 to 80 year old individuals in a randomized controlled clinical trial of anti-oxidative and zinc interference, with ARM status as a factor in the study. Age-related eye disease research group (1999). Age-related eye disease study (AREDS): and (5) design description. AREDS reports number one. And controlling the clinical test. 20, 573-600. These cohorts have been previously described (Klein, R, et al (2003) Heart blood health study early age-related macular degeneration, Ophthalmology.110, 25-33 and age-related eye study exploration group (2000) Risk factors for age-related macular degeneration disease-control studies in the age-related eye study, age-related eye study report No. 3, Ophthalmology.107, 2224-.

The study design evaluated cfh. elovl4, PLEKHA1, and LOC387715 genes in two independent cohorts with a different study strategy than ARM status and integrated the findings into meta-analysis. The correlation of the ARM susceptibility gene without consideration of the study plan will further improve the evidence that the correlation is true and increase the likelihood that the assessment of the gene will be truly defined in at risk individuals. Abbreviations: ARM-age-related macular degeneration; geographic atrophy of GA; CNV ═ choroidal neovascular membranes; OR ═ odds ratio; PAR-population due to risk; OR (OR)_domDominant effect odds ratio; OR (OR)_recRecessive effect odds ratio; OR (OR)_hetOdds ratio for risk allele heterozygotes; OR (OR)_homOdds ratio of risk allele homozygote.

Materials and methods

Cardiovascular Health Study (CHS) participants-sampling and visualization

CHS is a population-based, longitudinal study designed to identify factors associated with cardiovascular disease for the 65-year-old and older. Retinal studies were performed on visitors and survivors in the cohort that had been evaluated for 18 years at year 8. In Forsyth county of NC; CA sacrmento county; MD Washington county; and PA Pittsburgh for community-based summons. Health care fund-managed medical insurance eligibility charts are used to identify individuals 65 years and older. Members of the list family that are 65 years old or older are also suitable. Inclusion criteria were minimal and included non-institutional, except that for at least three years of retention, informed consent, no wheelchair-restriction, no admission to housing care, no tumor radiotherapy or chemotherapy was given. Fried et al (1991). DNA samples of CHS were used in this study.

CHS subjects the optional ocular retina was photographed and graded by dr. gorin using the same grading model previously described, Weeks et al (2004) age-related macular degeneration: contiguous evidence genomic scans for susceptibility genes within the1q31, 10q26, and 17q25 regions. Am J Jum Genet.75, 174-189. The analysis included only caucasian individuals, and the sample size of the other ARM groups was too small for the study results: 182 black controls but only 3 cases, and 5 controls of other ethnicities. All CHS cases used for analysis (n-126) were "type a", which was subjected to the clinically graded highest standard model Weeks et al, (2004). Individuals of this class are clearly affected by ARM based on widespread and/or combined drusen, pigment changes (including pigment epithelial cell detachment), and/or the appearance of end-stage disease [ Geographic Atrophy (GA) and/or Choroidal Neovascularization (CNV) membranes ]. Very few cases of CHS have end-stage ARM, GA or CNV (Table 10); analysis of specific subtypes of ARM has not been performed. All CHS controls (n ═ 1, 051) were AREDS grade 1. A few potential controls (n-22) had unclear GA or CNV signals and were excluded from the analysis.

Table 10 characteristics of the study population. The average age and phenotypic grading in the AREDS cohort was based on the age of the last fundus photograph. The numbers within the circular registry indicate the severity of the disease according to the AREDS rating method. The average age in the CHS cohort was based on the baseline age of the visit, but retinal assessments were performed at the 8 th year of the visit.

Age-related eye disease study (AREDS) participants-sampling and visualization

AREDS is the natural history of ARM and high dose vitamin and lens implantation clinical study age-related cataract prospective, multicenter study. The enrolled age of the male and female individuals recruited in the AREDS study ranged from 55 to 80 years; these individuals are required to be free of any conditions or diseases that affect long-term follow-up. Inclusion criteria were minimized and included having eye media sufficiently clear to allow fundus photography and either no ARM evidence in either eye or ARM in one eye while the other remained good (20/30 or better) (age-related eye disease research group 1999). DNA samples of the NEI-AREDS gene bank were used in this study.

ARM stage was graded using the AREDS age-related macular degeneration grading system and phenotypic grading based on recent visits. Again, only caucasian individuals were included in the analysis, and the sample sizes of the other sample groups were too small for the study results: only 15 african americans, 2 spain and 3 other ethnic groups. AREDS cases (n 701) are grouped into classes 3, 4 and 5. The AREDS body of level 3 (n ═ 96) has ARM but does not have end-stage ARM, the body of level 4 (n ═ 266) has end-stage ARM in one eye and the body of level 5 (n ═ 339) has end-stage ARM in both eyes. The AREDS control (n 175) had AREDS grade 1 (grade 2 individuals were excluded prior to analysis).

Genotyping

The variant M299V (rs3812153) in ELOVL4, the variant Y402H (rs1061170) in CFH and the variant S69A (rs10490924) within LOC387715 were genotyped using RFLP techniques. The primers, annealing temperature and restriction enzymes for each method were: ELOVL4 was 5'-AGATGCCGATGTTGTTAAAAG-3' (F, SEQ ID NO: 13), 5'-CATCTGGGTATGGTATTAAC-3' (R, SEQ ID NO: 14), 50 ℃ and BspH I; CFH 5'-TCTTTTTGTGCAAACCTTTGTTAG-3' (F, SEQ IN NO: 15), 5'-CCATTGGTAAAACAAGGTGACA-3' (R, SEQ ID NO: 16), 52 ℃ and NlaIII; LOC387715 is 5'-GCACCTTTGTCACCACATTA-3' (F, SEQ ID NO: 17), 5'-GCCTGATCATCTGCATTTCT-3' (R, SEQ ID NO: 18), 54 ℃ and Pvu II.

The A320T variant (rs1045216) within PLEKHA1 was genotyped using the 5' exonuclease Assay-on-Demand TaqMan method (Applied Biosystems). Amplification and genotyping studies were performed using ABI7000 and SDS2.0 software (Applied Biosystems, Inc.). For all genotyping performed in this study, double-blind genotyping was used for each variant, and comparisons and each differential location were performed using either raw data or by heavy-genotyping.

Correlation analysis

Determination of SNP-disease association with allele-and genotyping chi-square assays, p-values were simulated using 100000 replicates; when one is orFisher's exact assay was used when the number of more than one desired cells was less than 5. The correlation strength was assessed by the coarse Odds Ratio (OR) and the Population Attributable Risk (PAR). The general formula is used to calculate PAR: PAR ═ P_r(OR-1)/(1+P_r(OR-1)), wherein P_rIs a prevalence of risk factors in the general population. P_rDerived from the CHS control; this is possible because CHS subjects are not selected based on ARM disease stage and the number of CHS controls is enormous (n 1051). For comparison purposes, adjusted Odds Ratio (OR) of age and gender was assessed_adj). The Logistic regression model was used to calculate the odds ratio of nature and adjustment, using R (38). The least frequent allele in the control group is considered to be a risk allele, the risk alleles (RR) of these homozygotes are compared with the baseline group (normal allele [ NN ] homozygotes), and the risk allele (RN) of the heterozygote is compared with the baseline group, OR and OR are calculated_adj. The comparison of dominant (RR and RN versus NN) and recessive (RR versus RN and NN) was evaluated.

Differentiation between PLEKHA1 and LOC387715

We used a unimodal assay (Valdes, A.M and Thomson, G (1997) Detecting desease-predisposing variant: the halopeptide method. am J Hum Genet.60, 703-716) to identify which of the two sites, A320T in PLEKHA1 or S69A in LOC387715, was followed by the actual disease-inducing variant in the region of probably 10q26. The basis of the unimodal model is simple and multiplicative (for mathematical studies, see Valdes and thomson (1997)). If all predisposing variants are included in the haplotypes, then the neutral variants in a particular disease-predisposing haplotypes appear in the same proportion in cases and controls, although the actual frequency may be different. On the other hand, if all mutagenic entities are identified, an equal rate of non-mutagenic entity unimodal specimen frequency is not desirable.

The expected ratio of A320T-S69A haplotypes is described below, assuming one variant is ARM-induced and the other is a neutral variant. We hypothesized that A320T and S69A are ARM-induced variants of chromosome 10q26 PLEKHA1-LOC387715 unimodal regions. Four possible A320T-S69A are: G-G, a-G, G-T, a-T if a320T is a causal site and S69A is a neutral site we denote:

however, if S69A is a causal site and a320T is a neutral site, we show that:

where f represents the frequency of a particular unimodal specimen in the control or case.

The desired assumptions are:

H_OP: the a329T variant in PLEKHA1 completely contributed to ARM susceptibility in the PLEKHA1-LOC387715 haplotype region.

H_OL: the S69A variant in PLEKHA1 completely contributed to ARM susceptibility to the PLEKHA1-LOC387715 haplotype region.

PLEKHA1-LOC387715 single mode specimen region

Rejection of these hypotheses suggested that the variants tested were not sufficient to result in ARM susceptibility of the PLEKHA1-LOC387715 unimodal specimen region alone. The four 2 x 2 contingency tables can be derived from equations 1a, 1b, 2a and 2 b:

H_OPwhile we looked for homogeneity, H, in contingency tables 1a and 1b_OLWhile we looked for homogeneity in contingency tables 2c and 2 d. From each contingency table, adjusted chi-squared statistics may be computed to produce joint statistics. For H_OPThe statistics are the maximum chi-squared from equations 1a and 1 b. For H_OLThe statistics are the maximum chi-squared from equations 2a and 2 b.

However, the distribution of the joint statistics is unclear due to the dependence of the statistics from the contingency tables of each group. Independent deletions result from (1) the determination of multiple alleles at the disease-inducing sites, and (2) linkage disequilibrium between the disease-inducing and non-disease-inducing sites. Both of these cases are necessary (1) because variants typically have more than one allele, and (2) because, if variants are in complete linkage equilibrium, there is no need to distinguish their dependent correlation signals.

As a result of the dependencies in the data, a subsequent test for replacement of the allele at the predisposing locus (null hypothesis) is required. We grouped the haplotypes according to the predisposing locus allele (every two people). The case-control label was then sequentially changed in each group and joint statistics were calculated for each replicate pair. The replacement process is similar to that proposed by Li H (2001) (A mutation process for the hash message for indication of release-displacing variants. Ann HumGenet.65: 189-196). Phased genotyping is not available and a unimodal specimen must be genotyped from a non-unimodal specimen. Frequency of the unimodal specimen was assessed in control and case, respectively. The program SNPHAP (39) is used to frequency and phase the unimodal specimens in each subject. SNPHAP uses the EM algorithm to calculate the most similar frequency of the assessed haplotypes in the non-phased genotyping data. The late probability of individual haplotypes assigned exceeds 94% of each of A320T and S69A. Table 11 gives the frequency of the unimodal samples evaluated.

Table 11 PLEKHA 1a 320T and LOC387715S69A unimodal specimen frequency (SNPHAP program evaluation). The evaluation derived from the AREDS and CHS cohorts is given.

Interactivity analysis

The interactive analysis is divided into three steps: first, we tested the interactive genetic effect of Y402H within CFH and S69A within LOC387715 in CHS and AREDS samples, then i tested the interactivity of Y402H and S69A with smoking history in CHS and AREDS samples, and finally we calculated the combined ORs of the three risk factors.

We used the model strategy provided by North et al (North, B.V., Curtis, D.and sham, P.C. (2005) Application of localization regression cases-consistent association students in perfusion location. hum Hered.59, 79-87). A series of Logistic regression models were tried on the AREDS and CHS data sets to find the model that best describes the combined effect of CFH and LOC 387715. For each genotyping, models that can perform additional effects (ADD1, ADD2, and ADD-BOTH), and models that integrate dominant effects (DOM1, DOM2, and DOM-BOTH) are suitable. The ADD1 model includes only the condition x for the additive effects of CFH₁Y402H genotyping TT encodes-1, genotyping CT is 0, and genotyping CC is 1. The ADD2 model includes only condition x for the additive effects of LOC387715₂Genotype GG of S69A encodes-1, genotype GT is 0, and genotype TT is 1. ADD-BOTH mimics the combined additive effects of CFH and LOC 387715. DOM1 incorporates the dominant effect of ADD1 and includes x₁And z₁The CT genotype at Y402H was encoded as 0.5, and the TT and CC genotypes were-0.5. Dominant effects of the Integrated ADD2 of DOM2 model consensus and include x₂And z₂Genotyping GT at S69A encodes 0.5, and genotyping GG and TT are-0.5. DOM-BOTH models CFH and LOC387715Combined with a dominant effect. Three further models that model the interactivity between CFH and LOC387715 are suitable: ADD-INT includes the product condition x₁*x₂ADD-DOM comprising x₁*x₂，x₁*z ₂And z₁*x₂DOM-INT comprising x₁*x₂，x₁*z₂，z₁*x₂And z and₁*z₂。

the above model strategy was modified to study the combined effect of CFH and smoking, and the combined effect of LOC387715 and smoking. The modification method was the same as used by Schmidt et al (2006) to test the interactivity between LOC387715 and smoking. The coding design is the same as above except that the never smoker is coded as 0 and the never smoker is 1. The models suitable for CFH and smoking were: ADD1, SMOKE, ADD1-SMOKE, DOM1, ADD1-SMOKE-INT, and DOM1-SMOKE-INT, and models suitable for LOC387715 and smoking were: ADD2, SMOKE, ADD2-SMOKE, DOM2, ADD2-SMOKE-INT, and DOM 2-SMOKE-INT.

All models were compared by Akaike's information standard (AIC). The model with AIC difference < 2 is considered indifferent (North, B.V., Curtis, D. and Sham, P.C. (2005) Application of logical regression to case-control association study. hum Hered.59, 79-87), and the model with fewer parameters was chosen as a simple model. Since age and gender adjustments did not affect the assessment of ORs for Y2402H or S69A (tables 12 and 13), and in order to retain as few parameters as possible, model interactions were not adjusted for these covariates. Based on the above interaction analysis results, joint ORs are calculated.

APOE assay

Previous studies have reported the protective and deleterious effects of the apolipoprotein e (apoe) gene in ARM. The epsilon4 allele may have a protective effect (Klaver, C.C et al (1998) Genetic Association of exogenous protein E with related genomic expression. am J Hum Genet.63, 200. about. 206; Schmidt, S et al (2000) Association of The exogenous protein E gene with related genomic expression. positional expression. 12. about. expression vector expression. about. restriction of protein E with related genomic expression. about. 35. about. 293. about. Schmidt, S et al (2002) A porous-related expression of The exogenous protein E (applied) gene in 2004-related genomic expression. about. 23. about. 209. about. P.N. about. expression vector E with related expression of viral expression E, 1315. about. 12. about. expression vector, and. about. 12. about, 1306-. APOE variants were genotyped by CHS and their association with ARM was evaluated in this study. Individuals were classified by APOE genotyping as APOE-e 3/e 3 genotyping, and APOE-e 2 and APOE-e 4 carriers (indicated as APOE-e 2/, and APOE-e 4/, respectively); individuals with APOE- ε 2/ε 4 genotyping were included in the APOE- ε 2/, and APOE- ε 4/, groups. The Chi-Square assay was used to test the differences in the APOE- ε 3/ε 3 and APOE- ε 2/, and APOE- ε 3/ε 3 and APOE- ε 4/, distributions in the genotyping of controls and cases.

Meta-analysis

We performed meta-analysis methods to assemble the assessment of OR from the previously published reports of CFH and LOC387715 and the two reports presented here. Initial data was analyzed assuming that the relationship-study variation was randomly induced and a mixture-effect model was used. In the mixed-effect model, the maximum likelihood estimate of the aggregate OR is the average of the individual estimates, weighted by the inverse of its variables, the variables of the aggregate OR being estimated by the individual weight sum inverse. Meta-analysis of homozygotes was performed with R (analytical CoreTeam (2005) R: A language and environment for statistical calculation. R Foundation for statistical calculation, Vienna, Austria). The hypothesis of homozygosity was identified by the chi-square test. However, homozygote experiments appear to lack strength, so for comparison we also pooled OR in the random effect group. Meta-analysis of heterozygotes was performed using the restricted maximum similarity (REML) method, as performed by SAS Proc Mixed (SAS software release8.2[ SAS Institute Inc., Cary, NC, USA ]). The pooled RELM estimates are consistent with the DerSimonian-Laird estimates (Dersimonian, R. and Laird, N. (1986) Meta-analysis in clinical trials. control Clin T days.7, 177-. The SAS code of Van Houwelinggen et al was modified for heterozygote analysis.

The Y402H variant in CFH has been found in 11 studies to be strongly related to ARM (Edwards, A.O. et al (2005) comparative factor H polymorphism and age-related relational expression. science.308, 421-424; Haines, J.L. et al (2005) comparative factor H variant information. science.308, 419-421; R.J. et al (2005) comparative factor H variant in molecular expression-related expression. science.308, 419-389; Hageman, G.S. et al (2005) A2005 monomeric in comparative expression. science.308, science.385; Hageman, G.S. et al (2005) A2005 monomeric in comparative expression. scientific expression. 308, scientific 389; Hageman, G.S. et al (2005) A2005 monomeric in comparative expression H1, strain H1, scientific strain H3, scientific 1, strain, 3, scientific strain, III, 3-3, 3. D.S. publication No. (369) patent application No. 3. 9. A2005, scientific expression, 3. 9. environmental expression, 3. 20. A3. environmental expression, 3. A3. expression, 3. A. A.S. 2. expression, 3. expression, 2. A3. expression, 2. expression of expression, 2. A. expression, 2. expression of expression, 2. expression, 2 macromolecular differentiation. am J Hum Genet.77, 149-153; sepp, T, et al (2006) comparative factor H variant Y402H is a major risk decitermin ant for geological characterization and chloro neovascular characterization and nonmokers Invest Ophthalmol Vis Sci.47, 536-; rivera, A. et al (2005) depression LOC387715 is a minor phosphorus refractory gene for an age-related volumetric production, a stabilizing introduction of a minor factor H to a disease risk. hum. ol Gene.14, 3227-3236; souied, E.H. et al (2005) Y402H completed factor H polymeric associated with expressed volumetric generation in the free position. mol Vis.11, 1135-1140; magnusson, K.P. et al (2006) CFH Y402H coefficients similar ridge of soft drive and booth for advanced AMD.PLOS Med.3, e5and Jakobsdottir, J, et al (2005) science genes for age-related breast on chromosome 10q26.am J Hum Gene.77, 389 + 407); 2 of the 11 studies were our, so only the results from the article by Jakobsdotir et al (2005) were used, which were subjected to meta-analysis with all alignments. Klein et al (2005) study used a small panel of AREDS samples, Magnusson et al (2006) reported only ORs-based alleles and no genotyping value. Therefore, these two studies were not included. Results studies by Haines et al (2005) included pooled homozygous-and heterozygous ORs; genotyping value is not useful for assessing dominant and recessive effects of a control. Three studies have reported that smoking cigarettes from a unimodal specimen LOC387715 highly correlated variant, S69A, (river, A. et al (2005); Jakobsdottir et al (2005); Schmidt et al (2006) and Schmidt (2006) strongly affected the association of LOC387715 and age-related macular lesions. All three reports of LOC387715 were included in the meta-analysis. All CFH and LOC387715 study participants were non-hispanic europe and european american descendant white. Tables 14 and 15 describe studies that included CFH and LOC387715 meta-analyses, respectively.

Y402H meta-analytical study characteristics in CFH of Table 14

a sample size based on the total genotyping population, wherein the genotyping number is available for the total sample size, excluding missing data.

b mean age and corresponding standard deviation, or other generalized statistics from conventional literature.

c when genotyping numbers are available, p-values derived from the ext test (performed with R Genetics package) are given.

The two data sets of the d Edwards et al article were combined in the meta-analysis with combined control and case HWE p-values of 0.53 and 0.36, respectively.

The results of the e Haines et al article are included in the Ors meta-analysis of heterozygous-and homozygous individuals. The sample size was based on the total number of individuals, excluding missing genotyping data for Y402H in CFH.

f following the conventional literature, the two data sets of the article by Hageman et al were excluded.

The two data sets of the g river et al article were combined in the meta-analysis, with combined control and case HWE p-values of 0.03 and 0.09, respectively.

TABLE 15 meta-analysis of S69A in LOC387715 study characteristics

a sample size based on the total genotyping population, wherein the genotyping number is available for the total sample size, excluding missing data.

b mean age and corresponding standard deviation, or other generalized statistics from conventional literature.

c when genotyping numbers are available, p-values derived from the ext test (performed with R Genetics package) are given.

The two data sets of the d river et al article were combined in the meta-analysis with combined control and case HWE p-values of 0.31 and 0.01, respectively.

Only level 1 subjects in the e meta-analysis were classified as controls (level 2 subjects were discarded). The level 2 individuals in the general study of Schmidt et al are controls. The mean age and% males of the control were from the literature and based on grades 1 and 2.

Results

To further evaluate CFH in ARM, ELOVL4, PLEKHA1, and LOC387715, we genotyped previously reported SNPs for four genes from AREDS and CHS studies. Each data set was analyzed separately using all 701 non-spanish white ARM patients and 175AREDS study controls in the AREDS study and all 126 non-spanish white ARM patients he 1051 controls in the CHS study (see table 10 sample size and other data characteristics, and table 16 genotyping frequency). The disease stage at the time of the subject's last follow-up was the initial endpoint of the assessment AREDS program. The AREDS subjects included grade 1 controls and cases of moderate and severe ARM with monocular or binocular development (grade 3-5). To maintain consistency, the ARM disease stage was assessed by an expert, using a monocular, non-mydriatic fundus photograph, at 8-year follow-up. The major CHS cases had moderate ARM, including multiple drusen with or without pigment epithelial alterations (equivalent to AREDS grade 3), a small fraction of cases with Geographic Atrophy (GA) or choroidal neovascular membrane (CNV) and AREDS grade 1 controls with the removal of cases with severe extramacular drusen.

Table 16 ARM stage genotyping profile in ARMs and CHS cohort. The CEU population (northern and Western European family residents in Utah) evaluation parity for the International HapMap project is shown. AREDS cases were grade 3-5 and AREDS controls were grade 1. The number of genotypes per grade and subtype in table 17 is available. A description of the HapMap CEU population is provided herein.

TABLE 17 genotyping profiles in ARM stage, AREDS and CHS controls

Note-genotyping order: NN-RN-RR, wherein N is the normal allele and R is the risk allele. The risk allele is defined as the least frequent allele in the control. GA is geographic atrophy. CNV ═ choroidal neovascular membranes.

^aLevel 2AREDS subjects were not included in the analysis.

Correlation analysis

The association of each gene, CFH, ELOVL4, PLEKHA1 and LOC387715, with ARM was assessed by chi-square statistics. The level of effect of each gene was assessed by Odds Ratio (ORs) and Population Attributable Risk (PARs). To assess the genes most associated with early and severe ARM, ORs were calculated for each class and subtype (GA and CNV) using AREDS data distribution.

CFH: in the AREDS and CHS cohorts, the Y402H variant in CFH has significant relevance to ARM (P ≦ 0.00001) (Table 18), as verified by our earlier findings (Conley et al (2005) and Jakobsdottir et al (2005)) and others (Edwards et al (2005; Haines et al (2006); Klein et al (2005) and Rivera et al (2005)). ORs evaluation of Y402H in CFH showed that the variant has similar risks for all stages of ARM, and for severe ARM, GA and CNV forms (FIG. 7 and Table 12).

Table 18 results of allelic and genotyping correlation tests. P-values < 0.05 are in bold.^a2-edge p-value of Fisher's sexact test.^bBoth eyes had ARM cases of GA.^cBoth eyes had ARM cases of CNV.

The presence of the allele-dose benefit was shown for the two high risk ARM C alleles compared to the one C allele (Table 12 and FIG. 8). Although there is a higher risk with 2C alleles, the Population Assigned Risk (PAR) is similar to both risk alleles due to the higher frequency of CT genotypes than CC genotypes in the general population. Evaluation of PAR from CHS data set show CT and CC in one line accounting for 27% and 25% ARM in non-spain white, respectively. ELOVL 4: the M299V variant in ELOVL4 was significantly associated with exudative ARM in the AREDS sample (P ═ 0.034) (table 18), consistent with our previous findings (Conley, Y.P et al (2005)). However, none of the ORs were statistically significant at the 95% significance level (fig. 7 and 9 and table 12). These results do not exclude the potential role of ELOVL4 in ARM, but do not strongly support it. A smaller number of exudative ARM individuals do not support subtype analysis in the CHS cohort.

PLEKHA1 and LOC 387715: the important association of the S69A variant of LOC387715 with the presence of ARM (P ≦ 0.00001) in the AREDS and CHS data set (table 18), corroborating our earlier findings (condey, Y.P et al (2005) and Jakobsdottir, J et al (2005) and others (edwards, A.O et al (2005); Haines, j.l et al (2005); Klein R.J et al (2005); Rivera, a et al (2005) and Schmidt, S et al (2006)). the a320T variant in local OLEKHA1, which is located in the same monomode specimen area as LOC387715, the important degree of cross-linking between (P ═ 0.00004) in AREDS specimens but only between the borderline importance (P ═ 0.08) a320 and S69A in CHS specimens (P ═ 0.08) and the important degree of cross-linking imbalance between a320 and S A in the events specimen (70. 15. and s.65. the important gene, which is identified in the AREDS and the hypothesis that is the real mutation of the depresss (14. 7. the depress) or more important gene in the comparison, the depresss, the important gene, the detection method of the depresss, 10, 7, the depresss, the important gene, the depress, the relevant gene, the detection, 703-716). According to the haplotyping method, the allele-related frequency of the neutral variant is expected to be the same as that of the cases and controls in the haplotypes containing all the mutagenic variants. Based on the results obtained using this method, it was shown that S69A in LOC387715, but not A320T in PLEKHA1, is an ARM-induced variant (see "the distinction between PLEKHA1 and LOC 387715" herein). Furthermore, by substitution experiments with null hypotheses: h₀: the S69A variant in LOC387715 completely resulted in ARM susceptibility to PLEKHA1-LOC387715 haplotypes, which was not excluded (P ≦ 0.92 in AREDS data, P ≦ 0.45 in CHS data), whereas a320T was excluded in similar hypotheses (P ≦ 0.0001 in AREDS data, P ≦ 0.0002 in CHS data).

The S69A variant in LOC387715 shows a different risk pattern than Y402H in CFH. In the AREDS data differentiating severe disease, it was shown that the variants substantially increased the risk of severe ARM compared to the risk of moderate ARM (figures 7 and 10 (figure 11 gives the full results of PLEKHA1) and table 12). For example, the AREDS case OR of grade 3 with one OR two T alleles was 3.07 (95% CI 1.92-5.17, whereas the AREDS case OR with one OR two T alleles, with CNV in both eyes was 7.21 (95% CI 4.24-12.27). similar to CFH, S69A showed allele-dose effects with no significant difference in risk of attribution in the population of GT and TT genotyping (table 12 and fig. 10). since only four AREDS controls in S69A were TT homozygotes, point assessments and confidence intervals from conventional logistic regression, recessive and homozygote controls, would be compared to the exact regression assessment (model for SAS software version 8.2[ SAS Institute, Cary, NC, inc, USA)). These quantitative tests did not show a clear difference between the point evaluation (based on the PAR evaluation) and the lower confidence interval limit (based on comparison with ORs), but the upper confidence interval limit was higher (results not shown).

Interaction analysis

We used the logistic regression model to establish the combined effects between CFH and LOC387715, CFH and smoking, and LOC387715 and smoking. A series of models is selected to pick the most similar and simplified models. Models were evaluated using Akaike's Information Criteria (AIC) as described by North et al (2005). When the simplest model was identified we evaluated the joint ORs of risk factors. A separation evaluation value is calculated for each queue. In order to maximize the number of AREDS samples, carrying out subtype-free or sub-grade analysis; the AREDS cases of grade 3-5 were compared to the AREDS control of grade 1.

In the previous literature (Jakobsdottir (2005)) we did not find a combined effect between CFH and PLEKHA1/LOC387715 sites; independent multiplicative effects (in addition to log-measures) best describe the joint effect between two sites. River et al (2005) reported that S69A in LIC387715 rented independently of Y402H in CFH. Schmidt et al) | (2006a) also yielded a similar simplistic model, where again this model is the simplest for the AREDS and CHS data sets (table 19). The joint interaction ORs for risk genotyping in Y402H and S69A were calculated to further understand the joint effect between the two sites (Table 20). Regardless of severity, using all cases, AREDS data showed that individuals with heterozygous risk alleles at one locus and homozygous non-risk alleles at another locus had a higher ARM susceptibility than individuals without risk alleles at both loci (CT-GG combination genotyping, OR2.8, 95% CI 1.6-5.0; TT-GT combination genotyping, OR 3.2, 95% CI 1.7-6.0). The risk of ARM doubles if both sites are heterozygous (CT-GT combination genotyping, OR 7.2, 95% CI 3.8-13.5) and is increased when at least one site is homozygous for a risk allele. The combination ORs of CHS data showed a similar pattern, but with only one risk allele, the risk was substantially increased (CT-GG combination genotyping, OR 1.3, 95% CI 0.6-2.7; TT-GT combination genotyping, OR 1.2, 95% CI 0.5-2.8).

Table 19 results of suitable two-factor models by logistic regression. Detailed model definitions are given in the materials and methods-interaction analysis section. AIC differences were the most suitable ARC differences for the model. The simplest model is shown in bold. The AIC difference of the best fit model (lowest AIC) was 0.

Table 20 combination ORs and 95% of Y402H in CFH and S69A in COL387715

Note that: n is_controlsNumber of controls that completely typed at two sites, n_casesNumber of cases that completely typed at two sites. OR (OR)_Y402HOR, OR of S69A genotyping cross-averaged Y402H_S69AY402H genotyping OR cross-averaged S69A.^aOR individual homozygotes for at least one locus.

A recent study (Schmidt et al (2006a)) reported a strong statistical association between genotype S69A and smoking, based on binary (often vs. never smoked) and continuous comparisons (years of smoking). We did not duplicate this finding in the AREDS and CHS data sets (Table 19). The results of the AREDS samples show that the combined effect between Y402H and smoking is best illustrated by the independent multiplex effect, with no apparent overt or interactive effect. On the other hand, the best model to describe the CHS data only includes the Y402H additive effect. The results of the AREDS samples show that the combined effect between S69A and smoking is best illustrated by the independent multiplex effect, with no apparent overt or interactive effect. The CHA data shows the model with only S69A. When smoking exposure is a continuous variable (years of smoking) and the S69A genotype is encoded in an additive form, the interaction conditions in the CHS data are not important (P ═ 0.40). Years of smoking were not applicable in the AREDS study. To further investigate the combined effect of genes and smoking, the risk genotype of each gene and the combined ORs of smoking were evaluated from the AREDS data (table 21). The results show that when assessing the risk of ARM for any of the risk genotypes (Y402H and S69A) in smokers, both genes have a more profound impact on ARM risk than smoking. Two models and a simple chi-square test (P ═ 0.71) suitable for application showed that the major role of smoking in CHS data was insignificant (binary analysis).

Table 21CFH Y402H and smoking, and LOC387715S69A and smoking-smoking combined ORs and 95% CIs

APOE results: the CHS data was used to determine the major role of the APOE gene in ARM. There was no clear difference in the distribution of APOE-e 4 vector (P ═ 0.41) or APOE-e 2 vector (P ═ 0.42) between cases and controls when compared to APOE-e 3/e 3.

Meta-analysis

CFH Meta-assay: we pooled the evaluated Y402H ORs (including CHS and AREDS queues reported here (Table 14)) from 11 independent data sets using meta-analysis method. Results 5451 cases and 3540 controls of all European and European US races were analyzed. The results confirm that the C allele increased the risk of ARM in non-spain whites (figure 12 and table 22). When homogeneity is assumed in the study, the pooled estimates have narrower CI than any individual study, and non-overlapping CI of heterozygous-and homozygous ORs: OR (OR)_het2.43 (95% CI 2.17-2.72) and OR_hom6.22 (95% CI 5.38-7.19). When analysis is performed under heterogeneity, the point evaluation values are sufficiently similar and the CIs is more extensive. Leave-one-out sensitivity analysis under the fixed effect model showed that no studies had a significant impact on the pool assessment (table 22).

The study of river et al (2005) changed the evaluation value more strongly than any other study; when studies are included, OR_domAnd OR_hetA reduction of about 0.2, and OR_recAnd OR_homThe rise was about 0.2. The study of river et al (2005) was derived from HWE (P ═ 0.03), the only study of genotype distribution in the control group. The allele and genotype distributions in the cases and controls were quite similar in the study. However, the genotype distribution of CHS cases differed from other studies and the frequency of TT risk genotypes was lower compared to other cohorts (fig. 13).

Meta-analysis of LOC 387715: s69 risk-related meta-analysis in ARM of 69A includes 5 independent data sets (including CHS and AREDS cohorts reported here (FIG. 14 and Table 15.) results analyzed 3193 cases and 2405 controls for all European OR European US races LOC387715 studies are more heterogeneous than CFH studies; OR A risk-related meta-analysis in ARM includes 5 independent data sets_domAnd OR_hetClearly different in the study (P < 0.01 and 0.02, respectively). The results support the discovery of the association of the previous T allele with an increased risk of ARM (Table 23). Carrying two T alleles has a substantially higher risk than carrying one T allele; among the linkage-study variables, OR_hetAnd OR_hom2.48 (95% CI 1.67-3.70) and 7.33 (95% CI 4.33-12.42), respectively. At the placeGenotype distribution was similar in the control and all ARM populations, except for the CHS ARM population (fig. 15).

Table 23LOC387715S69A meta-analysis results. ORs (95% CIs) were evaluated from individual studies and all pooled studies. The results of the leave-one-out sensitivity analysis are shown.

^aORs homozygote test P-value in the study

^bCollection Point assessment Difference (Δ) when one study was removed from all study Collection assessments (in the fixed Effect model)

Example 1 after the discovery, the correlation of CFH and PLEKHA1/LOC387715 genes with ARM has been published. Many reports establish a strong correlation of Y402H coding changes in CFH with ARM, and three reports find correlation of S69A coding changes in LOC387715 with ARM, which is similar in magnitude to the correlation of Y402H. The location of these genes on chromosomes, CFH at 1q31 and LOC387715 at 10q26, were consistently identified by family-based cross-linking studies (Seddon, J.M et al (2003); Majewski, J et al (2003); Iyengar, S.K et al (2004); Weeks, D.E et al (2001) Age-related mammothy: an expanded genome-with scan with evaluation of sublattice logic with the same of the1q31 and 17q25 regions. am J Ophthalmol.132, 692; Weeks, D.E et al (2004); Klein, M.L et al (1998); and Kenealy, S.L et al (2004)).

Since the main objective of the Y402H study and the three S69A studies was specifically directed to finding (and finding) genes of ARM complex etiology, the magnitude of the overall risk allele effects of Y402H and S69A might be overestimated. Therefore, independent disease-control cohorts, AREDS and CHS cohorts with minimal inclusion and exclusion criteria based on ARM status were analyzed. The AREDS cohort does not have inclusion and exclusion criteria that include health-related inclusion and exclusion criteria based on ocular condition criteria; however, affected and non-affected individuals were registered (agent-Related Eye Disease Study Group (1999) The agent-Related Eye Disease Study (AREDS): design considerations. AREDS report No.1.control Clean Trials.20, 573-600). The CHS cohort is a population-based cohort that employs 65-year-old and older community-based summons of individuals with minimal inclusion and exclusion criteria (Fried et al (1991)). Retinal assessments were performed during 8 years of follow-up and retinal disease was not a contributing factor to the call. Given the difference in the purpose of the study in the two studies, the duplication of candidate gene association in the two cohorts strongly supports its etiology in ARM pathology.

We evaluated the correlations of four genes, CFH (1Q31), ELOVL4(6Q14), plekah1(10Q26), and LOC387715(10Q26). both CFH and LOC387715 have significant correlations with ARM in the AREDS and CHS cohorts (P ≦ 0.00001). Both genes showed ARM risk allele-dose effects, and the independent multiplex model supporting both genes was the simplest in the AREDS and CHS cohorts. The coding change of a320T in the PLEKHA1 gene was adjacent and in cross-linkage imbalance with LOC387715 on 10a26, and was clearly correlated with ARM in the AREDS cohort (P0.00004) but not the CHS cohort (P0.08). These are based on the results of applying the unimodal approach to the AREDS and CHS cohort, combined with the findings of river et al (2005) which first used a conditional unimodal analysis and determined weak expression of LOC387715 in the retina, and Schmidet et al (2006) which determined only a weak correlation signal on PLEKHA1, strongly indicating that S69A in LOC387715 on 10q26 is the major ARM-induced variant. The results of the unimodal approach showed that PLEKHA1 at 10q26 may not be sufficient to cause ARM-susceptibility; however, a320T and S69A and other unknown variants in PLEKHA1 could not be excluded as causative haplotypes.

Duplication of the CFH and LOC387715 genes in the AREDS and CHS cohort with ARM correlations, where the two cohorts have different study designs, further strongly supports their correlation in ARM. The PLEKHA1 needs to be found differently in the AREDS and CHS queues based on the different considerations of the two queues. In addition to the differences in the study between the case and control populations, the assessment of retinal changes, the literature on retinal findings, and the prevalence of ARM have differences between the two cohorts. Fundus pictures were taken in only one optional eye in the CHS study, and there was no dilated pupil picture, and these limitations would positively affect the sensitivity of determining disease pathology, which is even more likely to affect the determination of early retinal changes. The severe ARM proportion assessed at 8 years follow-up in all CHS cohorts is approximately 1.3% (Klein, r., Klein, b.e., Marino, e.k., Kuller, l.h., Furberg, c., Burke, g.l., and Hubbard, L.D. (2003) early-Related pathological in the cardiac pathological Study, opthalmology.110, 25-33) and approximately 17% in AREDS (Age-Related Eye Disease Study Group (2000)), and differences in the pathological distribution of severe ARM Disease between the two cohorts can lead to differences in findings, particularly when one gene is more likely to affect Disease development. Furthermore, the most important difference between the two queues is the time of the retinal photograph. AREDS participants performed retinal photographs at baseline of follow-up assessment, while CHS participants performed retinal photographs 8 years or more after enrollment, but they may be at least 73 years old. Retinal assessments that may be involved in CHS will deviate from the population requirements of a particular type of study in the world. Also note that in the AREDS cohort, subjects in the cohort, except for the unaffected group, will use vitamins and minerals in clinical trials to assess their role in ARM development. This effect is not yet clear.

As previously explained, a number of studies that have investigated the genetic etiology of ARM have been designed to optimize the genomic mapping of susceptible gene regions in assays for identifying ARM and ARM candidate genes. Using these retrospective studies to assess attribution risk would lead to overassessment. The range of risk of tricks reported was 43% to 68% for the Y42H variant in CFH (Edwards et al (2005); Haines et al (2005); jakobsdotti (2005); and Schmidt et al (2006)), and 36% to 57% for the S69A variant in LOC387715 (Jakobsdotti (2005) and Schmidt et al (2006)). Interestingly, the adjusted population in the CHS population is less attributable to risk (PARs) than previously reported: the Y402H variant in CFH was 38% and the S69A variant in LOC387715 was 25% (table 13). Since the primary CHS cases had an adjusted ARM, the PAR estimates derived from the CHS data did not match exactly the estimates from the previous study, and the proportion of patients with severe ARM was considered higher. However, it matches the AREDS case assessment value using level 3. These estimates are within the previously published PARs range: Y402H in CFH is 49% and S69A in LOC387715 is 46%. These findings would suggest that the ARM risk due to these two susceptible variants may be lower than previously thought, and that the CHS queues based on the ARM case are uncertain. The expected design is needed to more accurately assess the associated risk, which is estimated approximately by ORs from a review case-control design and corresponds to PARs.

We could not repeat the correlation of all ARM with ELOVL4 (Conley et al (2005)). The number of individuals with exudative ARM allowed us to perform subtype analysis in AREDS, not CHS cohort. Subtype analysis was particularly important for ELOVL4, and our previous findings suggest a role for ELOVL4 in exudative ARM; it is (weakly) supported in the AREDS queue. In both cohorts ELOVL4 lacked a strong correlation and a clear ORs in ARM susceptibility, and Ayyagari et al reported the lack of correlation, it was difficult to demonstrate its full role of ELOVL4 in ARM susceptibility. The ability of all ARM to OR 0.6 was determined to be reliable with a type I error rate of 5%, a minor allele frequency of 0.15, a population prevalence of 6%, an ability of-81% in AREDS and-69% in CHS. Under the same conditions, the ability to measure the same effect in exudative ARM was only 53% in the AREDS data. Thus, the potential for its action in all ARM was small for ELOVL4, but it cannot be excluded that it has a slight effect in exudative ARM. QUANTO was used for capacity assessment (Gauderman, W.J and Morrison, J.M (2006) QUANTO 1.1: Acomputer program for power and sample size computers for genetic-epidemic students,http://hydra.usc.edu/gxe)

the AREDS and CHS data support the independent role of Y402H in CFH and S69A in LOC387715 in ARM susceptibility. The multiple risk model for these two variants is the simplest based on AREDS and CHS cohort evaluation; this model is also provided by our previous literature, data demonstrated by jakbsdottir et al (2005) and river et al (2005) and Schmidt et al (2006 a). The risk of ARM showed an increase when the number of alleles for this case increased in Y402H and S69A (Table 20).

Before the discovery of CFH and LOC387715, smoking was one of the most important ARM-related risk factors known. Smoking is generally accepted as a modified risk factor for ARM; van Leeuwen et al provided an ARM epidemiological review and discussed the argument for smoking as an ARM risk factor (Van Leeuwen, r.klaver, c.c., Vingerling, J.R, Hofman, a. and de Jong, P.T. (2003) epidemic of age-related maculopathy: a review. eur J epidemiol.18, 845-854). Schmidt et al (2006) recently reported a statistically significant interactivity of LOC387715 and smoking in ARM. The data show that the association of LOC387715 in ARM was originally derived from genetic influence in moderate smokers. Our own interactivity analyses did not support this finding, while the AREDS data showed that the combined effect of S69A and smoking was multiple.

The role of CFH and LOC387715 in ARM susceptibility is further supported by our meta-analysis results. Meta-analysis, which includes the CHS and AREDS cohorts reported in this chapter, indicates that carrying one or two copies of a risk gene in CFH or LOC387715 will increase the risk of ARM, while carrying two copies is more dangerous. The combined results of all studies and the results in each independent study were very compact (figures 12 and 14). One known limitation of Meta-analysis is susceptibility to publishing bias. Typically, such deviations are the result of unreported negative findings (Normand, S.L. (1999) Meta-analysis: modeling, evaluating, combining, and reporting. Stat med.8, 321-). For CFH and LOC387715, all published studies have reported a strong association with the same trend for ARM, in which CFH the risk allele is the histidine encoding allele and LOC387715 the serine encoding allele. Statistically significant correlation publications that are expected to show random trends are preferred if the significant correlation is a false-positive result. (Lohmueler, K.E., pearce, C.L., pike, M., Lander, E.S., and Hirschhorn, J.N. (2003) Meta-analysis of genetic association participants a restriction of common variants to common variability to common disease. Nat. Gene.33, 177-182). The consistency of the correlation of CFH and LOC387715 with ARM is therefore a result of publishing bias.

Although our statistical analysis results are the major ARM-related gene on 10q26 with LOC387715, it does not prove causal. The possible etiologic role of CFH in the aetiology of ARM has been further supported by the role of its proteins in the location of drusen deposition and in the complement activation pathway in ARM patients. With regard to LOC387715, the biology of this gene is currently less known and how its protein affects ARM susceptibility is not known. Until recently it was shown that expression of LOC387715 was restricted to the placenta, but weak expression in the retina was reported (river et al (2005)), which opened the possibility of a gene tissue-specific effect.

In overview, the results in this example continue to support the role of CHF and LOC387715 in the etiology of ARM, both genes being highly correlated with ARM, regardless of how the patient was identified. Evaluation of PLEKHA1 and ELOVL4 in the AREDS and CHS cohorts showed that these genes were less effective in ARM susceptibility. The CFH and LOC387715 genes function independently in multiple pathways in the pathogenesis of ARM, and the individual homozygotes are at highest risk for risk alleles at either site.

Having thus described the invention, it will be apparent to those of ordinary skill in the art that equivalent operations may be performed within a wide and equivalent range of conditions, forms and other parameters without affecting the scope of the invention or any aspect thereof.

Claims (25)

1. A method for in vitro detection of a nucleic acid sample comprising a polymorphism associated with susceptibility to age-related maculopathy, the method comprising detecting in the sample the nucleotide sequence set forth in SEQ ID NO: 20(rs10490924) wherein the presence of thymine or guanine at base 270 of one or both alleles indicates that the nucleic acid sample comprises a polymorphism which shows increased susceptibility to age-related macular degeneration, and wherein the presence of guanine at both alleles indicates that the nucleic acid sample comprises a polymorphism which shows decreased susceptibility to age-related macular degeneration.

2. A method for detecting in vitro a nucleic acid sample comprising a polymorphism associated with susceptibility to age-related maculopathy, the method comprising testing for one or more allelic variants of PLEKHA1, PRSS11, LOC387715 to detect the presence of a major genomic SEQ ID NO: 20(rs10490924), wherein at least one of the polymorphisms at base 270 of SEQ ID NOs: the presence of thymine at base 270 of 20(rs10490924) shows increased susceptibility to age-related macular degeneration and is found in SEQ ID NO: the presence of guanine at base 270 of 20(rs10490924) shows reduced susceptibility to age-related macular degeneration.

3. The method of claim 1 or2, wherein the age-related macular degeneration is severe age-related macular degeneration.

4. The method of claim 1, wherein the step of detecting comprises detecting allelic variants at PLEKHA1/LOC387715/PRSS11 sites of chromosome 10q26 of the sample.

5. The method according to claim 1 or2, comprising detecting thymine or guanine at rs10490924 of LOC 387715.

6. The method according to claim 1 or2, comprising detecting the base at the single nucleotide polymorphism rs 4146894.

7. The method according to claim 1 or2, comprising detecting a base at the single nucleotide polymorphism rs 1045216.

8. The method according to claim 1 or2, comprising detecting the base on the single nucleotide polymorphism rs 1882907.

9. The method according to claim 1 or2, comprising detecting a base at the single nucleotide polymorphism rs 760336.

10. The method according to claim 1 or2, comprising detecting a base at the single nucleotide polymorphism rs 763720.

11. The method of claim 1 or2, further comprising detecting a polymorphism in complement factor H relative to the single nucleotide polymorphism rs800292 in the sample.

12. The method of claim 1 or2 further comprising detecting a polymorphism in complement factor H relative to the single nucleotide polymorphism rs1853883 in the sample.

13. The method of claim 1 or2 further comprising detecting a polymorphism in complement factor H relative to the single nucleotide polymorphism rs1061170 in the sample.

14. The method of claim 1 or2 comprising detecting an allelic variant in LOC 387715.

15. The method of claim 2, wherein the allelic variant is a mutant that produces a non-functional gene product and alters expression of the gene product.

16. The method of claim 2, wherein the allelic variant is one or more of a frameshift mutation, a promoter mutation and a splicing mutation.

17. The method of claim 1 or2 further comprising identifying complement factor H polymorphisms in the sample that exhibit increased or decreased susceptibility to age-related macular degeneration.

18. The method of claim 17, wherein the identified polymorphism corresponds to one of the single nucleotide polymorphisms rs800292, rs1061170, and rs 1853883.

19. The method of claim 1 or2, comprising defining a detection polymorphism using a nucleic acid amplification method.

20. The method of claim 19, the nucleic acid amplification method comprises one of PCR, reverse transcription PCR (RT-PCR), isothermal amplification, Nucleic Acid Sequence Based Amplification (NASBA), 5' fluorescent nucleic acid method, nucleic acid beacon method, and rolling circle amplification.

21. The method of claim 1 or2, comprising detecting polymorphisms using an array.

22. The method according to claim 21, wherein the array comprises one or more reagents for identifying the presence of allelic variants corresponding to one or more single nucleotide polymorphisms rs4146894, rs1045216, rs4405249, rs1882907, rs10490923, rs760336, rs763720, and rs1803403 in the nucleic acid sample.

23. The method of claim 3, wherein the presence of thymine on one or both alleles indicates increased susceptibility to end stage age-related macular degeneration, and the presence of guanine on both alleles indicates decreased susceptibility to end stage age-related macular degeneration.

24. The method of claim 23, wherein the presence of thymine on one or both alleles indicates increased susceptibility to either or both geographic atrophy or choroidal neovascular membrane, and the presence of guanine on both alleles indicates decreased susceptibility to either or both geographic atrophy or choroidal neovascular membrane.

25. An in vitro method of detecting a protein sample comprising susceptibility to age-related maculopathy, comprising identifying an allelic variant isoform from the gene product of LOC387715, wherein said isoform comprises a serine at position 69 thereof indicating increased susceptibility to age-related maculopathy and an alanine at position 69 thereof indicating decreased susceptibility to age-related maculopathy.