- Email: [email protected]
The influence of genetic variants in SORL1 gene on the manifestation of Alzheimer's disease
Poster Presentations: P1 genotype. Statistical analysis was performed on carriers of the rs45455403 genotype, to determine its predictor value for specific cognitive features. Over 80 neurological characteristics were explored for significant differences between the wild-type and the rs45455403 polymorphism carriers. Results: Observed cognitive differences between the wild-type and the rs45455403 polymorphism carriers included processing speed, naming, word-finding difficulties, problems with verbal repetition, reading, judgment, following instructions, sense of direction, anxiety, and insight to memory loss. These were observed even in subjects without overt symptoms of dementia. An additional 100 specimens from African-American individuals with careful cognitive profiling are currently undergoing genetic analysis. Conclusions: Studying the cognitive spectrum of patients carrying the rs45455403 genotype can help identify the earliest manifestations of Alzheimer’s disease in this African-American sub-group. Clinical characterization of this genetic subgroup may assist the identification of at-risk individuals or the clinical biomarkers for early diagnosis and/or progression. Current research suggests that early disease-modifying interventions or prevention strategies are highly effective for at-risk patients. P1-058
THE HSP90 CO-CHAPERONE FKBP51 PRODUCES NEUROTOXIC TAU OLIGOMERS: IMPLICATION FOR AGING AND ALZHEIMER’S DISEASE
Laura Blair1, Bryce Nordhues1, Shannon Hill1, K. Matthew Scaglione2, John O’Leary1, Leonid Breydo1, Zhang Bo1, Pengfei Li1, Lily Wang1, Carl Cotman3, Henry Paulson2, Martin Muschol1, Vladimir Uversky1, Torsten Klengel4, Elisabeth Binder4, Rakez Kayed5, Nicole Berchtold6, Todd Golde7, Chad Dickey1, 1University of South Florida, Tampa, Florida, United States; 2University of Michigan, Ann Arbor, Michigan, United States; 3UCI MIND, Irvine, California, United States; 4Max Planck Institute for Psychiatry, M€ unchen, Germany; 5University of Texas Medical Branch, Galveston, Texas, United States; 6University of California, Irvine, Irvine, California, United States; 7University of Florida, Gainesville, Florida, United States. Contact e-mail: [email protected] Background: FK506 binding protein 51kDa (FKBP51/FKBP5) is a heat shock protein 90 kDa (Hsp90) co-chaperone known to regulated tau biology. Tau is a microtubule associated protein that becomes hyperphosphorylated and forms intracellular neurofibullary tangles in the brain leading to Alzheimer’s disease (AD). Neuronal death in AD exceeds the number of tau tangles, which suggests other toxic tau species are involved in neurodegeneration. Methods: Biochemical analysis of recombinant protein was performed using circular dichroism (CD), dynamic light scattering (DLS), and atomic force microscopy (AFM). Stereotaxic injections were used to deliver FKBP5 tagged AAV particles into the hippocampus of rTg4510 tau mice. Immunohistochemistry was performed on brain tissue to evaluate tau protein levels. Human mRNA and DNA were evaluated for FKBP5 gene expression and methylation respectively. Results: In vitro recombinant tau assays reveal that FKBP51 works in coordination with Hsp90 to preserve tau from proteasomal degradation in an oligomeric state. In vivo, FKBP5 overexpression preserves neurotoxic, oligomeric tau. In the human brain, FKBP5 mRNA expression levels dramatically increase while FKBP5 DNA methylation decreases with age. In AD, FKBP5 mRNA expression levels positively correlate with Braak staging, and inversely correlate with FKBP5 methylation. Conclusions: FKBP51 is able to prevent tau clearance by the proteasome, preserving tau in an oligomeric form that is neurotoxic. FKBP51 is up-regulated with aging and increases further in AD. Overall these data suggest high FKBP51 expression creates a proteostatic imbalance leading to tau toxicity. Therefore, FKBP51 may accelerate AD onset and progression, suggesting that strategies targeting this Hsp90 co-chaperone could be therapeutically relevant for AD and other tauopathies.
P1-059
WHOLE GENOME SEQUENCING IN AN UNRESOLVED ALZHEIMER’S DISEASE FAMILY LINKED TO 7Q36
Rita Cacace1, Elise Cuyvers1, Karolien Bettens1, Caroline Van Cauwenberghe1, Jasper Van Dongen1, Marc Cruts1, Kristel Sleegers1,
P173
Christine Van Broeckhoven1, 1VIB and Institute Born-Bunge, University of Antwerp, Antwerp, Belgium. Contact e-mail: [email protected]. be Background: We are investigating a 3-generation autosomal dominant family with a clinical diagnosis compatible with Alzheimer disease (AD) (onset 66.867.4 years (N ¼13; age range 47-77 years). In a whole-genome linkage study, we linked the family to a genomic region of 5.44 Mb at 7q36 (Rademakers et al. AJHG, 2005). Exonic sequencing of 29 genes as well as dosage analysis using array-based comparative genomic hybridization (aCGH) identified one synonymous mutation in PAXIP1. However, we could not provide convincing evidence for a functional role of PAXIP1 in disease pathology leaving the possibility that another mutation at this locus might explain the observed linkage. Therefore, we used next generation sequencing (NGS) technology to re-examine the linked candidate region at 7q36. Methods: Four distantly related patients were selected and whole genome sequencing (WGS) was performed by Complete Genomics, Inc. The WGS data was analyzed using an in-house developed software tool GenomeComb. Sequenom MassARRAY Ò and Sanger sequencing were used for validation, for segregation in the family and for screening of control individuals (N¼1029) and AD patients (N¼1263). Results: The sequence of the linked region was extracted from the WGS data and compared between the 4 patients. To select truly called, novel or rare variations (MAF <1%), we applied quality and frequency filters. We identified 93 heterozygous variations that were shared between the 4 patients of which 35 were segregating within the family including the one in PAXIP1. After testing the 35 segregating variations in the control cohort only 4 variations remained for further analyses. Interestingly, is that 3 of these 4 variations are located in the same gene and cluster in the same intron. The remaining one is an intergenic variation. All 4 variations affect sequences of open chromatin and involve putative transcription factor binding sites (ENCODE project data). Conclusions: The WGS approach is a promising tool to identify linked variations in non-coding regions in an unresolved locus identified in an extended AD family. The clustering of the variations is indicative for a putative pathogenic role by influencing the expression and/or regulation of nearby genes. P1-060
THE INFLUENCE OF GENETIC VARIANTS IN SORL1 GENE ON THE MANIFESTATION OF ALZHEIMER’S DISEASE
Eva Louwersheimer1, Carlos Cruchaga2, Pieter Jelle Visser3, Henne Holstege4, Philip Scheltens4, John Van Swieten5, Wiesje Van Der Flier6, 1Alzheimer Center, VU University Medical Center, Amsterdam, Netherlands; 2Washington University School of Medicine, Saint Louis, Missouri, United States; 3VUMC, Amsterdam, Netherlands; 4VU University Medical Center, Amsterdam, Netherlands; 5Erasmus University Medical Center, Rotterdam, Netherlands; 6VU University Medical Center, Amsterdam, Netherlands. Contact e-mail: [email protected] Background: Alzheimer’s disease (AD) is a multigenetic disorder. A large number of risk variants have been discovered influencing the risk of sporadic AD, among which APOE is the strongest risk factor. Several studies found a significant association between single nucleotide polymorphisms (SNPs) in the sortilin-related receptor (SORL1) gene on chromosome 11 and AD. SORL1 plays an important role in the intracellular trafficking of APP and is also a low-density lipoprotein receptor. We evaluated how SORL1 SNPs influence manifestation of disease in terms of cognitive profile, hippocampal atrophy and cerebrospinal fluid (CSF) biomarkers in AD patients from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Methods: 83 caucasian patients (non-hispanic and non-latino) with probable AD were included from the ADNI. We selected 8 SNPs from the genotyped data coming from the ADNI database and imputed 11 SNPs. All SNPs were dichotomized based on presence of at least one risk allele. As dependent variables we used measures of cognitive performance (Mini Mental State Exame (MMSE) and Alzheimer’s Disease Assessment Scale (ADAS)), hippocampal volumes and the CSF biomarkers Amyloid-beta142 (abeta), total tau and tau phosporylated at threonine 181 (ptau). First we used univariate linear regression to assess associations between SNPs
P174
Poster Presentations: P1
and dependent variables. Subsequently, to account for multiple testing and to identify the most predictive SNPs, we used stepwise (forward) linear regression models adjusted for age and sex. Results: Within our group of 83 AD patients (age 7568, 45% female, MMSE 2462), 14 SNPs showed a relation with one or more dependent variables (p<0.05). Stepwise linear regression revealed relations between SNP rs4935775, SNP rs11218340, SNP rs1131497, SNP rs1010159 and CSF biomarkers, between SNP rs4935774, SNP rs2298813, SNP rs3824968 and cognitive performance. Finally, SNP rs4935775, SNP rs2276346, SNP rs3824968 were associated with hippocampal volume. Results of the multivariate models are presented in table 1. Conclusions: SORL1 SNPs influence clinical manifestation of AD. We have found that SORL1 SNPs influence all six markers of AD. Most interesting is SNP rs4935775 with relatively high tau, high ptau and low hippocampal volume. These results should be validated in larger cohorts. Table 1 Results stepwise linear regression with 19 SORL1 SNPs and six markers of AD. MMSE
ADAS
Amyloid-beta Tau
Ptau
Hippocampal volume
rs4935774 1.260.4** rs661057 rs11218304 rs560573 rs12364988 rs668387 rs689021 rs641120 rs4935775 33.5613.1** 12.164.2** -1575.26317.9** rs2298813 -6.162.5* rs11600231 rs2276346 753.66303.1* rs11218340 -1.460.6* -33.5615.3* rs2070045 rs3824966 rs3824968 -722.06180.0** rs1010159 -24.068.5** rs1133174 rs1131497 9.063.9* Age 1.560.6** -0.660.2* -49.0611.5** Sex 18.868.5* 23.0611.5* -439.76183.4*
We performed stepwise (forward) linear regression with each of the six dependent variables and the 19 SNPs selected adjusted for age. The results are presented with unstandarized coefficients Beta 6 Standard Error. *p<0.05. **p<0.01. Cognitive tests: MMSE¼Mini Mental State Exam. ADAS¼Alzheimer’s Disease Assesment Scale. CSF biomarkers: Amyloid-beta¼Amyloid-beta1-42. Tau. Ptau¼tau phosporylated at threonine 181. Hippocampal volume¼volume hippocampus left + volume hippocampus right.
P1-061
CSF BIOMARKERS FOR ALZHEIMER’S PATHOLOGY AND THE EFFECT SIZE OF APOE-ε4
Ulf Andreasson1, Ronald Lautner2, Jonathan Schott3, Niklas Mattsson4, Oskar Hansson5, Sanna-Kaisa Herukka6, Michael Ewers7, Harald Hampel8, Anders Wallin9, Lennart Minthon10, Nick Fox11, John Hardy11, Kaj Blennow12, Henrik Zetterberg12, 1Clinical Neurochemistry Laboratory, M€ olndal, Sweden; 2Institute of Neuroscience and Physiology, University of Gothenburg, M€ olndal, Sweden; 3Dementia Research Centre, Institute of Neurology, UCL, London, United Kingdom; 4UCSF, CA, USA and University of Gothenburg, Sweden, M€olndal, Sweden; 5Lund University, Malm€ o, Sweden; 6University of Eastern Finland, Kuopio, Finland; 7 University of California, San Francisco, San Francisco, California, United States; 8University of Frankfurt, Frankfurt/ Main, Germany; 9University of Gothenburg, M€ olndal, Sweden; 10Department of Clinical Sciences, Malm€o, Sweden; 11UCL Institute of Neurology, London, United Kingdom;
12 Sahlgrenska Academy, University of Gothenburg, M€olndal, Sweden. Contact e-mail: [email protected]
Background: In a large group of patients with both genetic and cerebrospinal fluid (CSF) biomarker data we aimed to evaluate how biomarker-assisted diagnosis making can alter the effect size of the association between the ε4 variant of the apolipoprotein E (APOE) gene and Alzheimer’s disease (AD). Methods: We included clinically diagnosed patients with either AD dementia (n¼309) or mild cognitive impairment (MCI) due to AD (n¼287), cognitively normal controls (n¼251) and patients with MCI who remained stable over at least 2 years (n¼399) or developed dementias other than AD (n¼99). All had APOE ε2/ε3/ε4 genotypes and results on the CSF biomarkers T-tau, P-tau and Ab42 determined. Results: When pooling AD and MCI-AD patients into one clinical AD group (n¼596) and comparing to all remaining categories (non-AD; n¼749), a positive APOE ε4 carrier status (one or two ε4 alleles) yielded an odds ratio (OR) of 4.45 for a clinical diagnosis of AD at inclusion or follow-up. Disregarding the clinical diagnoses and subgrouping all subjects into amyloid-positive, defined as CSF Ab42 < 546 ng/L (n¼779), and amyloid-negative, defined as CSF Ab42 > 546 ng/L (n¼563) gave an OR for APOE ε4 as high as 6.27. Comparing patients, again disregarding the clinical diagnoses, who had a complete CSF biomarker signature indicative of AD (n¼438) with subjects with a negative CSF biomarker pattern (n¼414), the biomarker diagnosis strengthened the association to APOE ε4 and the OR increased from 4.45 in pure clinical diagnosis to 7.66 in patients classified on the basis of biomarker data alone. In subjects who had both a clinical diagnosis and a concordant complete biomarker profile (n¼324 AD, n¼155 controls), the association of APOE ε4 with AD was even stronger (OR 10.4). Conclusions: APOE ε4 appears as strongly associated with amyloid pathology as clinical AD. Clinical criteria that incorporate biomarker information on Alzheimer’s pathology give a stronger association with APOE ε4 than clinical diagnosis alone. P1-062
EVIDENCE OF RECESSIVE ALZHEIMER’S DISEASE LOCI IN CARIBBEAN HISPANICS: GENOME-WIDE SURVEY OF RUNS OF HOMOZYGOSITY
Mahdi Ghani Kakhki1, Christine Sato1, Joseph Lee2, Christiane Reitz3, Danielle Moreno1, Richard Mayeux3, Peter St. George-Hyslop4, Ekaterina Rogaeva4, 1Tanz Centre for Research in Neurodegenerative Diseases, University of Toronto, Toronto, Ontario, Canada; 2Columbia University College of Physicians and Surgeons, New York, New York, United States; 3Columbia University, New York, New York, United States; 4 University of Toronto, Toronto, Ontario, Canada. Contact e-mail: mahdi. [email protected] Background: The role of recessive inheritance has not been widely investigated for Alzheimer’s disease (AD). However, array technology could help identify rare recessive loci among genomic runs of homozygosity (ROH). Recently, genome-wide measurements of ROH (>1 Mb) were studied in two datasets of North American and European origin. However, global burden analyses of ROH did not reveal any significant difference between cases and controls, likely due to the outbred nature of the investigated datasets. Methods: We studied a dataset of Caribbean Hispanics with a high prevalence of AD and evidence of inbreeding, which consisted of 547 unrelated cases (49% familial AD, 71% females) and 542 normal controls (68% females). Samples were genotyped on the Illumina Human Hap 650Y array. ROH (>1 Mb) were identified by PLINK using a sliding window of 50 SNPs across the genome. The genome-wide total/average length of the ROH per sample and the total number of ROH per sample were analysed with a one-tailed test (10,000 permutations). Results: We identified 17,137 autosomal regions with ROH (8,912 in 542 controls and 8,225 in 547 cases), including 4,041 ROH in 267 FAD cases. The global burden analysis revealed significant association of AD with a larger average ROH size per person: 2,133 Kb in cases vs. 1,934 Kb in controls (p¼0.0039); and this association was stronger with familial AD (p¼0.0005). Total distance spanned by the ROH per person was marginally longer in familial AD cases (40 Mb) vs. controls (34 Mb) (p¼0.05). Association analysis of genes intersecting ROH is in process and will be presented. The analyses will be further
THE HSP90 CO-CHAPERONE FKBP51 PRODUCES NEUROTOXIC TAU OLIGOMERS: IMPLICATION FOR AGING AND ALZHEIMER’S DISEASE
Laura Blair1, Bryce Nordhues1, Shannon Hill1, K. Matthew Scaglione2, John O’Leary1, Leonid Breydo1, Zhang Bo1, Pengfei Li1, Lily Wang1, Carl Cotman3, Henry Paulson2, Martin Muschol1, Vladimir Uversky1, Torsten Klengel4, Elisabeth Binder4, Rakez Kayed5, Nicole Berchtold6, Todd Golde7, Chad Dickey1, 1University of South Florida, Tampa, Florida, United States; 2University of Michigan, Ann Arbor, Michigan, United States; 3UCI MIND, Irvine, California, United States; 4Max Planck Institute for Psychiatry, M€ unchen, Germany; 5University of Texas Medical Branch, Galveston, Texas, United States; 6University of California, Irvine, Irvine, California, United States; 7University of Florida, Gainesville, Florida, United States. Contact e-mail: [email protected] Background: FK506 binding protein 51kDa (FKBP51/FKBP5) is a heat shock protein 90 kDa (Hsp90) co-chaperone known to regulated tau biology. Tau is a microtubule associated protein that becomes hyperphosphorylated and forms intracellular neurofibullary tangles in the brain leading to Alzheimer’s disease (AD). Neuronal death in AD exceeds the number of tau tangles, which suggests other toxic tau species are involved in neurodegeneration. Methods: Biochemical analysis of recombinant protein was performed using circular dichroism (CD), dynamic light scattering (DLS), and atomic force microscopy (AFM). Stereotaxic injections were used to deliver FKBP5 tagged AAV particles into the hippocampus of rTg4510 tau mice. Immunohistochemistry was performed on brain tissue to evaluate tau protein levels. Human mRNA and DNA were evaluated for FKBP5 gene expression and methylation respectively. Results: In vitro recombinant tau assays reveal that FKBP51 works in coordination with Hsp90 to preserve tau from proteasomal degradation in an oligomeric state. In vivo, FKBP5 overexpression preserves neurotoxic, oligomeric tau. In the human brain, FKBP5 mRNA expression levels dramatically increase while FKBP5 DNA methylation decreases with age. In AD, FKBP5 mRNA expression levels positively correlate with Braak staging, and inversely correlate with FKBP5 methylation. Conclusions: FKBP51 is able to prevent tau clearance by the proteasome, preserving tau in an oligomeric form that is neurotoxic. FKBP51 is up-regulated with aging and increases further in AD. Overall these data suggest high FKBP51 expression creates a proteostatic imbalance leading to tau toxicity. Therefore, FKBP51 may accelerate AD onset and progression, suggesting that strategies targeting this Hsp90 co-chaperone could be therapeutically relevant for AD and other tauopathies.
P1-059
WHOLE GENOME SEQUENCING IN AN UNRESOLVED ALZHEIMER’S DISEASE FAMILY LINKED TO 7Q36
Rita Cacace1, Elise Cuyvers1, Karolien Bettens1, Caroline Van Cauwenberghe1, Jasper Van Dongen1, Marc Cruts1, Kristel Sleegers1,
P173
Christine Van Broeckhoven1, 1VIB and Institute Born-Bunge, University of Antwerp, Antwerp, Belgium. Contact e-mail: [email protected]. be Background: We are investigating a 3-generation autosomal dominant family with a clinical diagnosis compatible with Alzheimer disease (AD) (onset 66.867.4 years (N ¼13; age range 47-77 years). In a whole-genome linkage study, we linked the family to a genomic region of 5.44 Mb at 7q36 (Rademakers et al. AJHG, 2005). Exonic sequencing of 29 genes as well as dosage analysis using array-based comparative genomic hybridization (aCGH) identified one synonymous mutation in PAXIP1. However, we could not provide convincing evidence for a functional role of PAXIP1 in disease pathology leaving the possibility that another mutation at this locus might explain the observed linkage. Therefore, we used next generation sequencing (NGS) technology to re-examine the linked candidate region at 7q36. Methods: Four distantly related patients were selected and whole genome sequencing (WGS) was performed by Complete Genomics, Inc. The WGS data was analyzed using an in-house developed software tool GenomeComb. Sequenom MassARRAY Ò and Sanger sequencing were used for validation, for segregation in the family and for screening of control individuals (N¼1029) and AD patients (N¼1263). Results: The sequence of the linked region was extracted from the WGS data and compared between the 4 patients. To select truly called, novel or rare variations (MAF <1%), we applied quality and frequency filters. We identified 93 heterozygous variations that were shared between the 4 patients of which 35 were segregating within the family including the one in PAXIP1. After testing the 35 segregating variations in the control cohort only 4 variations remained for further analyses. Interestingly, is that 3 of these 4 variations are located in the same gene and cluster in the same intron. The remaining one is an intergenic variation. All 4 variations affect sequences of open chromatin and involve putative transcription factor binding sites (ENCODE project data). Conclusions: The WGS approach is a promising tool to identify linked variations in non-coding regions in an unresolved locus identified in an extended AD family. The clustering of the variations is indicative for a putative pathogenic role by influencing the expression and/or regulation of nearby genes. P1-060
THE INFLUENCE OF GENETIC VARIANTS IN SORL1 GENE ON THE MANIFESTATION OF ALZHEIMER’S DISEASE
Eva Louwersheimer1, Carlos Cruchaga2, Pieter Jelle Visser3, Henne Holstege4, Philip Scheltens4, John Van Swieten5, Wiesje Van Der Flier6, 1Alzheimer Center, VU University Medical Center, Amsterdam, Netherlands; 2Washington University School of Medicine, Saint Louis, Missouri, United States; 3VUMC, Amsterdam, Netherlands; 4VU University Medical Center, Amsterdam, Netherlands; 5Erasmus University Medical Center, Rotterdam, Netherlands; 6VU University Medical Center, Amsterdam, Netherlands. Contact e-mail: [email protected] Background: Alzheimer’s disease (AD) is a multigenetic disorder. A large number of risk variants have been discovered influencing the risk of sporadic AD, among which APOE is the strongest risk factor. Several studies found a significant association between single nucleotide polymorphisms (SNPs) in the sortilin-related receptor (SORL1) gene on chromosome 11 and AD. SORL1 plays an important role in the intracellular trafficking of APP and is also a low-density lipoprotein receptor. We evaluated how SORL1 SNPs influence manifestation of disease in terms of cognitive profile, hippocampal atrophy and cerebrospinal fluid (CSF) biomarkers in AD patients from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Methods: 83 caucasian patients (non-hispanic and non-latino) with probable AD were included from the ADNI. We selected 8 SNPs from the genotyped data coming from the ADNI database and imputed 11 SNPs. All SNPs were dichotomized based on presence of at least one risk allele. As dependent variables we used measures of cognitive performance (Mini Mental State Exame (MMSE) and Alzheimer’s Disease Assessment Scale (ADAS)), hippocampal volumes and the CSF biomarkers Amyloid-beta142 (abeta), total tau and tau phosporylated at threonine 181 (ptau). First we used univariate linear regression to assess associations between SNPs
P174
Poster Presentations: P1
and dependent variables. Subsequently, to account for multiple testing and to identify the most predictive SNPs, we used stepwise (forward) linear regression models adjusted for age and sex. Results: Within our group of 83 AD patients (age 7568, 45% female, MMSE 2462), 14 SNPs showed a relation with one or more dependent variables (p<0.05). Stepwise linear regression revealed relations between SNP rs4935775, SNP rs11218340, SNP rs1131497, SNP rs1010159 and CSF biomarkers, between SNP rs4935774, SNP rs2298813, SNP rs3824968 and cognitive performance. Finally, SNP rs4935775, SNP rs2276346, SNP rs3824968 were associated with hippocampal volume. Results of the multivariate models are presented in table 1. Conclusions: SORL1 SNPs influence clinical manifestation of AD. We have found that SORL1 SNPs influence all six markers of AD. Most interesting is SNP rs4935775 with relatively high tau, high ptau and low hippocampal volume. These results should be validated in larger cohorts. Table 1 Results stepwise linear regression with 19 SORL1 SNPs and six markers of AD. MMSE
ADAS
Amyloid-beta Tau
Ptau
Hippocampal volume
rs4935774 1.260.4** rs661057 rs11218304 rs560573 rs12364988 rs668387 rs689021 rs641120 rs4935775 33.5613.1** 12.164.2** -1575.26317.9** rs2298813 -6.162.5* rs11600231 rs2276346 753.66303.1* rs11218340 -1.460.6* -33.5615.3* rs2070045 rs3824966 rs3824968 -722.06180.0** rs1010159 -24.068.5** rs1133174 rs1131497 9.063.9* Age 1.560.6** -0.660.2* -49.0611.5** Sex 18.868.5* 23.0611.5* -439.76183.4*
We performed stepwise (forward) linear regression with each of the six dependent variables and the 19 SNPs selected adjusted for age. The results are presented with unstandarized coefficients Beta 6 Standard Error. *p<0.05. **p<0.01. Cognitive tests: MMSE¼Mini Mental State Exam. ADAS¼Alzheimer’s Disease Assesment Scale. CSF biomarkers: Amyloid-beta¼Amyloid-beta1-42. Tau. Ptau¼tau phosporylated at threonine 181. Hippocampal volume¼volume hippocampus left + volume hippocampus right.
P1-061
CSF BIOMARKERS FOR ALZHEIMER’S PATHOLOGY AND THE EFFECT SIZE OF APOE-ε4
Ulf Andreasson1, Ronald Lautner2, Jonathan Schott3, Niklas Mattsson4, Oskar Hansson5, Sanna-Kaisa Herukka6, Michael Ewers7, Harald Hampel8, Anders Wallin9, Lennart Minthon10, Nick Fox11, John Hardy11, Kaj Blennow12, Henrik Zetterberg12, 1Clinical Neurochemistry Laboratory, M€ olndal, Sweden; 2Institute of Neuroscience and Physiology, University of Gothenburg, M€ olndal, Sweden; 3Dementia Research Centre, Institute of Neurology, UCL, London, United Kingdom; 4UCSF, CA, USA and University of Gothenburg, Sweden, M€olndal, Sweden; 5Lund University, Malm€ o, Sweden; 6University of Eastern Finland, Kuopio, Finland; 7 University of California, San Francisco, San Francisco, California, United States; 8University of Frankfurt, Frankfurt/ Main, Germany; 9University of Gothenburg, M€ olndal, Sweden; 10Department of Clinical Sciences, Malm€o, Sweden; 11UCL Institute of Neurology, London, United Kingdom;
12 Sahlgrenska Academy, University of Gothenburg, M€olndal, Sweden. Contact e-mail: [email protected]
Background: In a large group of patients with both genetic and cerebrospinal fluid (CSF) biomarker data we aimed to evaluate how biomarker-assisted diagnosis making can alter the effect size of the association between the ε4 variant of the apolipoprotein E (APOE) gene and Alzheimer’s disease (AD). Methods: We included clinically diagnosed patients with either AD dementia (n¼309) or mild cognitive impairment (MCI) due to AD (n¼287), cognitively normal controls (n¼251) and patients with MCI who remained stable over at least 2 years (n¼399) or developed dementias other than AD (n¼99). All had APOE ε2/ε3/ε4 genotypes and results on the CSF biomarkers T-tau, P-tau and Ab42 determined. Results: When pooling AD and MCI-AD patients into one clinical AD group (n¼596) and comparing to all remaining categories (non-AD; n¼749), a positive APOE ε4 carrier status (one or two ε4 alleles) yielded an odds ratio (OR) of 4.45 for a clinical diagnosis of AD at inclusion or follow-up. Disregarding the clinical diagnoses and subgrouping all subjects into amyloid-positive, defined as CSF Ab42 < 546 ng/L (n¼779), and amyloid-negative, defined as CSF Ab42 > 546 ng/L (n¼563) gave an OR for APOE ε4 as high as 6.27. Comparing patients, again disregarding the clinical diagnoses, who had a complete CSF biomarker signature indicative of AD (n¼438) with subjects with a negative CSF biomarker pattern (n¼414), the biomarker diagnosis strengthened the association to APOE ε4 and the OR increased from 4.45 in pure clinical diagnosis to 7.66 in patients classified on the basis of biomarker data alone. In subjects who had both a clinical diagnosis and a concordant complete biomarker profile (n¼324 AD, n¼155 controls), the association of APOE ε4 with AD was even stronger (OR 10.4). Conclusions: APOE ε4 appears as strongly associated with amyloid pathology as clinical AD. Clinical criteria that incorporate biomarker information on Alzheimer’s pathology give a stronger association with APOE ε4 than clinical diagnosis alone. P1-062
EVIDENCE OF RECESSIVE ALZHEIMER’S DISEASE LOCI IN CARIBBEAN HISPANICS: GENOME-WIDE SURVEY OF RUNS OF HOMOZYGOSITY
Mahdi Ghani Kakhki1, Christine Sato1, Joseph Lee2, Christiane Reitz3, Danielle Moreno1, Richard Mayeux3, Peter St. George-Hyslop4, Ekaterina Rogaeva4, 1Tanz Centre for Research in Neurodegenerative Diseases, University of Toronto, Toronto, Ontario, Canada; 2Columbia University College of Physicians and Surgeons, New York, New York, United States; 3Columbia University, New York, New York, United States; 4 University of Toronto, Toronto, Ontario, Canada. Contact e-mail: mahdi. [email protected] Background: The role of recessive inheritance has not been widely investigated for Alzheimer’s disease (AD). However, array technology could help identify rare recessive loci among genomic runs of homozygosity (ROH). Recently, genome-wide measurements of ROH (>1 Mb) were studied in two datasets of North American and European origin. However, global burden analyses of ROH did not reveal any significant difference between cases and controls, likely due to the outbred nature of the investigated datasets. Methods: We studied a dataset of Caribbean Hispanics with a high prevalence of AD and evidence of inbreeding, which consisted of 547 unrelated cases (49% familial AD, 71% females) and 542 normal controls (68% females). Samples were genotyped on the Illumina Human Hap 650Y array. ROH (>1 Mb) were identified by PLINK using a sliding window of 50 SNPs across the genome. The genome-wide total/average length of the ROH per sample and the total number of ROH per sample were analysed with a one-tailed test (10,000 permutations). Results: We identified 17,137 autosomal regions with ROH (8,912 in 542 controls and 8,225 in 547 cases), including 4,041 ROH in 267 FAD cases. The global burden analysis revealed significant association of AD with a larger average ROH size per person: 2,133 Kb in cases vs. 1,934 Kb in controls (p¼0.0039); and this association was stronger with familial AD (p¼0.0005). Total distance spanned by the ROH per person was marginally longer in familial AD cases (40 Mb) vs. controls (34 Mb) (p¼0.05). Association analysis of genes intersecting ROH is in process and will be presented. The analyses will be further