new england
journal of medicine
The
established in 1812
january 10, 2013
vol. 368
no. 2
Variant of TREM2 Associated with the Risk of Alzheimer’s Disease
Thorlakur Jonsson, Ph.D., Hreinn Stefansson, Ph.D., Stacy Steinberg Ph.D., Ingileif Jonsdottir, Ph.D.,
Palmi V. Jonsson, M.D., Jon Snaedal, M.D., Sigurbjorn Bjornsson, M.D., Johanna Huttenlocher, B.S.,
Allan I. Levey, M.D., Ph.D., James J. Lah, M.D., Ph.D., Dan Rujescu, M.D., Harald Hampel, M.D.,
Ina Giegling, Ph.D., Ole A. Andreassen, M.D., Ph.D., Knut Engedal, M.D., Ph.D., Ingun Ulstein, M.D., Ph.D.,
Srdjan Djurovic, Ph.D., Carla Ibrahim-Verbaas, M.D., Albert Hofman, M.D., Ph.D., M. Arfan Ikram, M.D., Ph.D.,
Cornelia M van Duijn, Ph.D., Unnur Thorsteinsdottir, Ph.D., Augustine Kong, Ph.D.,
and Kari Stefansson, M.D., Ph.D.
A bs t r ac t
Background
Sequence variants, including the ε4 allele of apolipoprotein E, have been associated
with the risk of the common late-onset form of Alzheimer’s disease. Few rare variants
affecting the risk of late-onset Alzheimer’s disease have been found.
Methods
We obtained the genome sequences of 2261 Icelanders and identified sequence variants that were likely to affect protein function. We imputed these variants into the
genomes of patients with Alzheimer’s disease and control participants and then
tested for an association with Alzheimer’s disease. We performed replication tests
using case–control series from the United States, Norway, the Netherlands, and
Germany. We also tested for a genetic association with cognitive function in a
population of unaffected elderly persons.
Results
A rare missense mutation (rs75932628-T) in the gene encoding the triggering receptor
expressed on myeloid cells 2 (TREM2), which was predicted to result in an R47H
substitution, was found to confer a significant risk of Alzheimer’s disease in Iceland
(odds ratio, 2.92; 95% confidence interval [CI], 2.09 to 4.09; P = 3.42×10−10). The
mutation had a frequency of 0.46% in controls 85 years of age or older. We observed
the association in additional sample sets (odds ratio, 2.90; 95% CI, 2.16 to 3.91;
P = 2.1×10−12 in combined discovery and replication samples). We also found that
carriers of rs75932628-T between the ages of 80 and 100 years without Alzheimer’s
disease had poorer cognitive function than noncarriers (P = 0.003).
From deCODE Genetics (T.J., H.S., S.S.,
I.J., U.T., A.K., K.S.), the University of
Iceland, Faculty of Medicine (I.J., P.V.J.,
U.T., K.S.), and Landspitali University Hospital (P.V.J., J.S., S.B.) — all in Reykjavik,
Iceland; the Department of Medical Genetics, Institute of Human Genetics,
Tübingen (J.H.), Division of Molecular
and Clinical Neurobiology, Department
of Psychiatry, University of Munich, Munich (L.M.U.) and University of Halle,
Halle (D.R., I.G.), and the Department of
Psychiatry, University of Frankfurt am
Main, Frankfurt am Main (H.H.) — all in
Germany; the Department of Neurology,
Alzheimer’s Disease Center, Emory University School of Medicine, Atlanta
(A.I.L., J.J.L.); K.G. Jebsen Center for Psychosis Research, Division of Mental
Health and Addiction (O.A.A., S.D.), and
the Geriatric Department, Norwegian
Center for Aging and Health (K.E., I.U.),
Oslo University Hospital, and the Institute of Clinical Medicine, University of
Oslo (O.A.A., K.E., S.D.) — all in Oslo;
and the Department of Epidemiology,
Erasmus Medical Center, Rotterdam, the
Netherlands, (C.I.-V., A.H., M.A.I.,
C.M.D.). Address reprint requests to Dr.
K. Stefansson at deCODE Genetics, Sturlugata 8, 101 Reykjavik, Iceland, or at
kstefans@decode.is.
Conclusions
Our findings strongly implicate variant TREM2 in the pathogenesis of Alzheimer’s
disease. Given the reported antiinflammatory role of TREM2 in the brain, the R47H
substitution may lead to an increased predisposition to Alzheimer’s disease through
impaired containment of inflammatory processes. (Funded by the National Institute
on Aging and others.)
n engl j med 368;2
nejm.org
This article was published on November 14,
2012, at NEJM.org.
N Engl J Med 2013;368:107-16.
DOI: 10.1056/NEJMoa1211103
Copyright © 2012 Massachusetts Medical Society.
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
107
The
n e w e ng l a n d j o u r na l
A
lzheimer’s disease, the most common form of dementia in the elderly, is a
neurodegenerative disorder that is characterized by a slow but progressive loss of cognitive
function. Extracellular amyloid plaques, intracellular neurofibrillary tangles, and loss of neurons
and synapses resulting in brain atrophy are the
main pathological hallmarks of Alzheimer’s disease.1 Disease onset is usually after the age of
70 years, although the prevalence increases exponentially with age after the age of 65 years and exceeds 25% in those over the age of 90 years.2
The vast majority of variants in the sequence
of the genome that have been shown to markedly
affect the risk of Alzheimer’s disease are rare variants in APP, PSEN1, and PSEN2 (encoding amyloid
precursor protein, presenilin 1, and presenilin 2,
respectively). These variants appear to be fully
penetrant and result in Alzheimer’s disease with
an early onset, in most cases before the age of
60 years.3 However, these variants do not shed
light on the most common, late-onset form of the
disease. Although a number of common, low-risk
variants have been associated with late-onset Alzheimer’s disease,4 the ε4 allele of apolipoprotein E
(ApoE), originally discovered as a risk factor for
Alzheimer’s disease in 1993,5,6 remains by far the
most important sequence variant affecting the
risk of late-onset Alzheimer’s disease because of
its prevalence and the size of its effect on risk,
with reported odds ratios ranging from 3 to 4
(a meta-analysis is available at www.alzgene.org/
meta.asp?geneID = 83).
To search for sequence variants that influence
the risk of Alzheimer’s disease, we performed a
genomewide association analysis with variants
(found by whole-genome sequencing of samples
from 2261 Icelanders) that were likely to affect
protein function. These variants were imputed in
patients with Alzheimer’s disease and controls
with the use of long-range haplotype phasing and
chip-genotype information. Using this approach,
we have recently reported variants that greatly
influence the risk of the sick sinus syndrome,7
gout,8 gliomas,9 ovarian cancer,10 and Alzheimer’s
disease.11
of
m e dic i n e
Data Protection Authority. Written informed consent was obtained from all participants or their
guardians before blood samples were drawn, and
all sample identifiers were encrypted in accordance with the regulations of the Icelandic Data
Protection Authority.
In 1062 patients, the diagnosis of Alzheimer’s
disease was established according to the criteria
for definite, probable, or possible Alzheimer’s
disease of the National Institute of Neurological
and Communicative Disorders and Stroke and the
Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA).12 In another 2697 patients, the diagnosis was established according
to the criteria for code F00 of the International
Classification of Diseases, 10th Revision (ICD-10). We
assessed cognitive function using data from the
Resident Assessment Instrument (RAI), with which
assessment is performed on an individual basis
and recorded in a Minimum Data Set (MDS 2.0)
form. Data were primarily obtained through RAI
2.0 for Nursing Homes, which is a comprehensive
and standardized instrument originally developed
for residential facilities for the elderly,13 with
additional information provided by the InterRAI
Assessment for Home Care.14 We assessed cognitive function using the MDS Cognitive Performance Scale (CPS), which combines selected MDS
2.0 items expressing different measures of cognitive function on a seven-category scale, ranging
from 0 (intact) to 6 (severe impairment).15 The
CPS is hierarchical and based on an assessment of
several measures of cognitive function; a 1-unit
change is a reflection of distinct and measurable
changes in at least one cognitive domain. A total
of 1236 study participants with a score of 0 on the
CPS scale were used as cognitively intact controls. We selected 110,050 population controls
from among participants in various research
projects at deCODE Genetics, excluding those in
whom Alzheimer’s disease had been diagnosed.
Norwegian Population
The Regional Ethical Committee and the Norwegian Data Protection Agency approved the studies.
The sample of patients with Alzheimer’s disease
consisted of home-dwelling outpatients referred
to three memory clinics in the Southeast Health
Me thods
Region of Norway for suspicious dementia. The
Study Participants
patients underwent a standardized comprehenIcelandic Population
sive assessment, which consisted of taking a medApproval for these studies was obtained from the ical history from the patient as well as a close
National Bioethics Committee and the Icelandic family member, comprehensive neuropsycholog-
108
n engl j med 368;2
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
Variant of TREM2 Associated with Alzheimer’s Disease
ical testing, a physical and psychiatric examination with the use of standardized assessment
scales, blood sample analyses, and brain imaging.16 Diagnoses of Alzheimer’s disease were established in accordance with the ICD-10 criteria
for research. Controls were recruited as a part of
the Thematically Organized Psychosis study (enrolling 700 patients and 291 controls)17 and two
Norwegian studies of attention deficit–hyperactivity disorder (enrolling 626 patients and 898
controls).
pants in whom dementia associated with Alzheimer’s disease was diagnosed fulfilled the
criteria for probable Alzheimer’s disease, according to the NINCDS-ADRDA criteria. The control
group included participants who were randomly
selected from the general population of Munich.
Controls who had a disease of the central nervous system or a psychotic disorder or who had a
first-degree relative with a psychotic disorder
were excluded.
Rotterdam Population
The Rotterdam Study I is a prospective, populationbased cohort study enrolling 7983 residents who
are 55 years of age or older and live in Ommoord,
a suburb of Rotterdam, the Netherlands.18 Baseline examinations took place between 1991 and
1993; follow-up examinations were performed
between 1997 and 1999 and between 2002 and
2006; a final follow-up examination was performed between 2009 and 2011.19
Participants were screened for prevalent dementia with the use of a three-stage process; those
free of dementia remained under surveillance for
incident dementia, a determination that was made
with the use of record linkage and assessment at
three subsequent examinations. We included all
patients in whom Alzheimer’s disease was diagnosed before December 31, 2011; a subset of those
in whom Alzheimer’s disease was not diagnosed
served as controls.
Screening was done with the MMSE and Geriatric Mental Schedule (GMS) for organic (i.e.,
medical or physical) mental illness for all participants. Participants who were deemed to be positive on screening (a score of <26 on the MMSE
or >0 on the GMS organic level) underwent the
Cambridge Mental Disorders of the Elderly Examination (CAMDEX) schedule. Participants in whom
dementia was suspected underwent more extensive neuropsychological testing. When available,
imaging data were used. In addition, all participants were continuously monitored for major
events (including dementia) through automated
linkage of the study database with digitized medical records from general practitioners, the Regional Institute for Outpatient Mental Health Care,
and the municipality.
In addition, physicians’ files from nursing
Munich Population
homes and general practitioners’ records for parPatients with Alzheimer’s disease were recruited ticipants who moved out of the Ommoord disat the memory clinic of the Department of Psy- trict were reviewed twice a year. For suspected
chiatry, University of Munich, Germany. Partici- dementia events, additional information (includEmory Population
All participants underwent a research evaluation
in the Emory Alzheimer’s Disease Research Center in Atlanta. Participants were classified as controls or as having probable Alzheimer’s disease
after a review of the history, physical examination, neuropsychological testing, and available clinical records in a consensus conference among neurologists, neuropsychologists, and other health
care professionals. All controls underwent initial
cognitive screening with a Mini–Mental State Examination (MMSE) and Clock Drawing Test
(CDT), and those who were impaired (z score,
−1.79 or less) after adjustment for age, sex, and
education and all patients with Alzheimer’s disease underwent further neuropsychological testing consisting of a Brief Visuospatial Memory
Test–Revised, Wechsler Memory Scale–Revised
Logical Memory I and II, Wechsler Adult Intelligence Scale–Revised Similarities, Wechsler Adult
Intelligence Scale III Digit Span, Wechsler Adult Intelligence Scale–Revised Digit Symbol, Judgment
of Line Orientation, Trail Making Test A and B,
Category Fluency (Animals, Vegetables), Phonemic Fluency, Boston Naming Test, Consortium to
Establish a Registry for Alzheimer's Disease
(CERAD) Word List Memory, CDT evaluation, the
Beck Depression Inventory (for participants <65
years of age), and the Geriatric Depression Scale
(for participants ≥65 years of age). In interviews,
controls had a negative response to questions about
a personal history of a neurologic disease, and
medical records that included imaging (computed tomography or magnetic resonance imaging),
neuropsychiatric assessment, and ancillary testing were requested and reviewed when available.
n engl j med 368;2
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
109
The
n e w e ng l a n d j o u r na l
ing neuroimaging) was obtained from hospital
records, and research physicians discussed available information with a neurologist experienced
in dementia diagnosis and research to verify all
diagnoses.
Dementia was diagnosed in accordance with
internationally accepted criteria for dementia in
the revised third edition of the Diagnostic and
Statistical Manual of Mental Disorders, and Alzheimer’s
disease was diagnosed on the basis of the
NINCDS-ADRDA criteria for possible, probable,
or definite disease. The criteria of the National
Institute of Neurological Disorders and Stroke–
Association Internationale pour la Recherche et
l’Enseignement en Neurosciences (NINDS-AIREN)
were used to diagnose vascular dementia. The final
diagnosis was determined by a panel consisting
of a neurologist, a neurophysiologist, and a research physician. The diagnoses of Alzheimer’s
disease and vascular dementia were not mutually exclusive.
Data Generation and Analysis
Whole-Genome Sequencing, SNP Calling,
and Imputation
We performed whole-genome sequencing on samples obtained from 2261 Icelandic participants, followed by single-nucleotide-polymorphism (SNP)
calling and genotype imputation, using methods
that were described previously.11 The chip-genotype imputation was based on chip genotypes
from 95,085 persons. Approximately 34 million
markers (SNPs and insertion–deletion polymorphisms), including the 191,777 functional variants
identified through whole-genome sequencing, were
imputed in the Icelandic cases and controls. The
information content for rs75932628 in the imputed data was 0.999 (as compared with 1.0 for
perfect information).
Single-Track Assay SNP Genotyping
We performed single SNP genotyping of
rs75932628 using the Centaurus (Nanogen) platform.20 No mismatches resulted from a comparison of genotypes determined through imputation
and Centaurus genotyping of 964 participants,
including 30 participants who were predicted to
be heterozygous for the rare allele and 2 who
were predicted to be homozygous for the rare allele. Samples from the United States, Germany,
and Norway were also typed with the use of Centaurus assays. Before analysis, we excluded sam-
110
n engl j med 368;2
of
m e dic i n e
ples with a genotype yield of less than 90% and
one member of each pair of duplicate samples. The
genotyping yield was at least 95% in both cases
and controls in samples from all study locations,
and all genotypes were in Hardy–Weinberg equilibrium. Samples from the Netherlands were
genotyped for rs75932628 with Taqman allelic
discrimination Assays-by-Design (Applied Biosystems). All measurements were performed in accordance with the manufacturer’s protocols;
primer and probe sequences are available from
the manufacturer.
Imputation of Genomewide Data
We downloaded genotype and phenotype data
from the Genetic Alzheimer’s Disease Associations (GenADA) study, the National Institute on
Aging Late Onset Alzheimer’s Disease and National Cell Repository for Alzheimer’s Disease
Family Study (NIA-LOAD), and the Electronic
Medical Records and Genomics (eMERGE) genomewide association study of dementia from
the controlled-access portal of the National Institutes of Health Genotype and Phenotype database
(dbGAP, accession number phs000234.v1.p1). Two
small components of NIA-LOAD, phs000168.v1
.p1.c2 (involving 28 participants) and phs000168
.v1.p1.c1 (involving 570 participants), could not be
included because consent from participants was
for nonprofit use only. In addition, rs75932628
could not be successfully imputed on the basis of
the GenADA data (information content associated with the additive test, <0.3), which led to the
exclusion of that study from further analyses.
In the NIA-LOAD and eMERGE studies, patients were classified as having definite, probable,
or possible Alzheimer’s disease, according to
NINCDS–ADRDA criteria. In the eMERGE study,
patients in whom dementia was diagnosed according to electronic-medical-record criteria were
excluded. In both studies, controls were free of
dementia, and participants who were biologic relatives of patients with Alzheimer’s disease were
not included as controls. Imputation was performed with the use of IMPUTE2 with the March
2012 haplotype release of the 1000 Genomes
Project as a reference. Before imputation, participants with a genotyping yield of less than
98% were removed; SNPs with a yield of less
than 98%, a minor allele frequency of less than
1%, or a deviation from Hardy–Weinberg equilibrium (P<1.0×10−5) were also removed. The
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
Variant of TREM2 Associated with Alzheimer’s Disease
information content of the imputed data associ- fixed effects and the individual as a random efated with the additive test was 0.79 and 0.75 for the fect. We used bootstrapping methods to calculate
NIA–LOAD and eMERGE studies, respectively.
standard errors for the analysis of the CPS score
versus age.
Statistical Analysis
For the Icelandic data, we performed case–control
association testing of imputed genotypes using
methods that were described previously.11 Odds
ratios were calculated on the basis of a multiplicative model for the two chromosomes of each
individual. The method of genomic control was
used to correct for relatedness and potential population stratification.
We used logistic regression to perform association analysis that was based on the NIA-LOAD
and eMERGE data sets, with the first three principal components included as covariates to adjust
for population stratification. Before the analysis,
we removed data for participants with genotyped
sex inconsistent with reported sex, the lower-yield
sample in each pair of duplicates, genetically related older cases and younger controls (to eliminate the inclusion of first- or second-degree relatives), and participants with an estimated fraction
of less than 0.9 European ancestry in analysis
with STRUCTURE software and using as a reference HapMap samples of Utah residents with ancestry from northern and western Europe (CEU),
Han Chinese in Beijing and Japanese in Tokyo
(CHB+JPT), and Yoruba in Ibadan, Nigeria (YRI).
We used Fisher’s exact test to perform other
association analyses. We combined results from
the various replication groups, and from the discovery group and the replication groups, using
inverse-variance–weighted meta-analysis. The relationship between the age at onset and rs75932628T was examined with the use of a linear model
with the age at onset as the response and
rs75932628-T and the ApoE ε4 count as predictors.
We analyzed the effect of age on the CPS score,
using determinations made at several ages for each
participant. The CPS score is based on the Resident Assessment Instrument for Nursing Homes,
which is applied on average three times per year
in Icelandic nursing homes. Since the residency
time in nursing homes in Iceland is on average
3 to 4 years, many determinations of CPS that
are performed at different times are available for
most persons. We assessed the difference in CPS
score between rs75932628-T carriers and noncarriers in the age range from 80 to 100 years using
a mixed model with age and carrier status as
n engl j med 368;2
R e sult s
Association of Variant with Alzheimer’s
Disease
Through whole-genome sequencing of samples
from 2261 Icelanders, we found 191,777 nonsynonymous SNPs, frameshift variants, splicing variants, and stop gain–loss variants and imputed
these variants in patients with Alzheimer’s disease and controls. A total of 3550 patients with
Alzheimer’s disease were included in the analysis. Our control group included persons who had
reached the age of 85 without a diagnosis of Alzheimer’s disease. With the exclusion of the ApoE
locus and the A673T variant in APP,11 only one
marker, rs75932628, showed a genomewide association, on the basis of either the Bonferroniadjusted threshold of P<2.60×10−7 for 191,777
tests or the conventional threshold for genomewide association studies (5×10−8). The T allele of
rs75932628, which encodes a substitution of histidine for arginine at position 47 (R47H) in the gene
encoding the triggering receptor expressed on
myeloid cells 2 (TREM2) on chromosome 6p21.1,
with an allelic frequency of 0.63% in Iceland, was
found to confer a significant risk of Alzheimer’s
disease (odds ratio, 2.92; 95% confidence interval [CI], 2.09 to 4.09; P= 3.42×10−10) (Table 1). No
other variant in TREM2 that is likely to affect protein function showed nominally significant association with Alzheimer’s disease (Table S1 in the
Supplementary Appendix, available with the full
text of this article at NEJM.org).
Risk variants for a late-onset disorder such as
Alzheimer’s disease are expected to be more common in the general population than in elderly
controls without the disease. Thus, the use of
elderly controls without a history of Alzheimer’s
disease would in general be expected to result in
an increased statistical power to detect risk variants for this disease. We therefore investigated the
association of rs75932628-T using samples from
cognitively intact elderly controls, as determined
by their CPS scores. In an analysis of samples from
such controls who were at least 85 years of age, the
odds ratio for the association with rs75932628-T
was 4.66 (95% CI, 2.38 to 9.14; P = 7.39×10−6). By
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
111
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 1. Association between the rs75932628-T Variant and Alzheimer’s Disease in Comparisons with Three Control Groups.
No. of
Participants
Frequency
110,050
Population controls ≥85 yr of age
Cognitively intact controls ≥85 yr of age*
Control Group
Odds Ratio
(95% CI)
P Value
0.63
2.26 (1.71–2.98)
1.13×10−8
8,888
0.46
2.92 (2.09–4.09)
3.42×10−10
1,236
0.31
4.66 (2.38–9.14)
7.39×10−6
%
All population controls
* Intact cognition was defined as a score of 0 on the Cognitive Performance Scale, which ranges from 0 to 6, with higher
scores indicating more severe impairment. CI denotes confidence interval.
contrast, we observed a smaller odds ratio in a
comparison with samples from a general population controls of all ages (odds ratio, 2.26; 95%
CI, 1.71 to 2.98; P = 1.13×10−8) (Table 1). The less
significant P value that was observed for the comparison with cognitively intact controls was due to
the substantially smaller size of this control group
(1236, vs. 8888 elderly controls and 110,050 population controls). Furthermore, we found that the
frequency of rs75932628-T in controls age 85 years
or older without a history of Alzheimer’s disease
(0.46%) was significantly less than in controls
under the age of 85 years (0.64%; P = 0.007). This
observation is expected for alleles associated with
common, late-onset disorders such as Alzheimer’s disease and thus provides further support
for the association between rs75932628-T and
Alzheimer’s disease.
We identified four homozygous carriers of
rs75932628-T in Iceland. Of these homozygotes,
Alzheimer’s disease had been diagnosed in two
but not in two others (ages 51 and 52).
Replication Series
In an attempt to replicate the association between
rs75932628-T and Alzheimer’s disease, we genotyped rs75932628 in cohorts from the United States
(Emory), Germany, the Netherlands (Rotterdam
Study), and Norway. We found that rs75932628-T
conferred a risk of Alzheimer’s disease in all replication cohorts, with a combined odds ratio of 2.83
(95% CI, 1.45 to 5.40; P = 0.002 (Table 2). Combining results from Iceland (using population
controls who were at least 85 years of age) and
the replication cohorts, we found that the overall
association between rs75932628-T and Alzheimer’s disease was highly significant (odds ratio,
2.90; 95% CI, 2.16 to 3.91; P = 2.1×10−12). We also
112
n engl j med 368;2
estimated the effect of rs75932628-T on the risk
of Alzheimer’s disease by imputing the variant in
two publicly available data sets (NIA-LOAD and
eMERGE). Association results that were based on
imputed genotypes for rs75932628-T in these data
sets were found to be consistent with the observed effect of rs75932628-T on disease risk in
the genotyped cohorts (odds ratio, 2.66; 95% CI,
1.46 to 4.84; P = 0.001) (Table S2 in the Supplementary Appendix).
Association with ApoE ε4
We investigated the effect of the ε4 allele of ApoE
on the association between rs75932628-T and Alzheimer’s disease. We found a somewhat higher
odds ratio in ε4 noncarriers (4.03) than in ε4 carriers (2.38) (Table S3 in the Supplementary Appendix). The difference in the frequency of
rs75932628-T in ApoE ε4 carriers, as compared
with noncarriers, had borderline significance
(odds ratio, 0.60; 95% CI, 0.37 to 0.98; P = 0.04).
However, in a logistic-regression model, the interaction between rs75932628-T and ApoE ε4 was
not significant (P = 0.18), and the difference in
frequency according to ApoE ε4 status in cases
did not replicate in the additional data sets (odds
ratio, 1.79; 95% CI, 0.73 to 4.43; P = 0.20) (Table S4
in the Supplementary Appendix).
Although the population frequency of
rs75932628-T was low (0.63% in Iceland), it conferred a risk of Alzheimer’s disease that was
similar to the risk the ApoE ε4 allele, which has
a population frequency of 17.3% in Iceland. (As
compared with controls 85 years of age or older,
the odds ratio for Alzheimer’s disease was 2.92 for
rs75932628-T and 3.08 for the ApoE ε4 allele.) We
also found that in Iceland, each copy of rs75932628T was associated with an age at onset of Alzhei-
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
Variant of TREM2 Associated with Alzheimer’s Disease
mer’s disease that was lower by 3.18 years than in
controls without the variant (P = 0.20). Although
this effect was similar to that of ApoE ε4 (3.22
years; P = 4.1×10−8), it was not significant, owing
to the low frequency of the variant, which resulted in a reduced effective sample size and an
elevated standard error (2.49 for rs75932628-T vs.
0.58 for ApoE ε4). We found a similar result in
the Dutch data (3.65 years per allele, P = 0.13), and
the combined effect was found to be 3.4 years
per allele (P = 0.048).
Association According to Age
No. of
Cases
Group
No. of
Controls
Frequency*
Odds Ratio
(95% CI)
P Value
%
Emory
399
402
0.12
3.03 (0.33–78.35)
0.37
Munich
517
1891
0.19
3.15 (1.06–10.40)
0.04
Rotterdam
944
4950
0.15
2.45 (0.94–6.35)
0.07
Norway
177
2484
0.16
3.52 (0.54–17.21)
0.14
2037
9727
2.83 (1.45–5.40)
0.002
Combined
* The reported frequency is for the presence of the rs75932628-T variant in controls.
Discussion
Inflammation is a well-established histologic feature in the brains of patients with Alzheimer’s disease. Complement factors were identified in amyloid plaques in the 1980s,21,22 followed by reports
of clusters of activated microglia, complementactivation products, and cytokines in and near
amyloid plaques.23-26 There is evidence that inflammation is an early event in the brains of patients with Alzheimer’s disease.27 It has also been
noted that the expression of genes associated
with inflammation in the brain is increased in
aging and that this effect is accentuated in patients with Alzheimer’s disease.28 According to the
amyloid hypothesis, which is the predominant
Carriers
Noncarriers
4
Cognitive Performance Scale
We also investigated how rs75932628-T affects cognitive function in elderly controls in whom Alzheimer’s disease had not been diagnosed. Cognitive function declined steadily with age in elderly
persons between the ages of 80 and 100 (Fig. 1).
We found that carriers of rs75932628-T showed
worse cognition (a mean increase of 0.87 units on
the CPS) than did noncarriers (P = 0.003). Clinical
determination of Alzheimer’s disease is partially
based on progressive loss of cognitive function,
in particular memory, with time. Thus, the decline in cognitive function that we observed in
rs75932628-T carriers may be due to early cognitive deficits that ultimately result in Alzheimer’s
disease. Alternatively, the decline may at least partially be due to a loss of cognitive function in old
age that is not associated with Alzheimer’s disease.
The latter explanation is in keeping with the hypothesis that Alzheimer’s disease may be the extreme of the cognitive decline of the elderly and
caused by the same biochemical mechanism.11
n engl j med 368;2
Table 2. Replication Analysis of the Association between the rs75932628-T
Variant and Alzheimer’s Disease.
3
2
0
80
85
90
95
100
Age (yr)
Figure 1. Cognition as a Function of Age in Controls
Who Were Carriers or Noncarriers of the rs75932628-T
Variant Associated with the Risk of Alzheimer’s Disease.
Shown are scores on the Cognitive Performance Scale
(CPS) for carriers and noncarriers of the rs75932628-T
variant associated with Alzheimer’s disease, according
to age. Scores on the CPS range from 0 to 6, with higher
scores indicating more severe impairment. Values are
shown in 2-year bins (i.e., the data point for 81 years of
age contains data for ages 80 and 81), except for the last
bin, which represents ages of 98, 99, and 100 years. No
CPS data were available for carriers in the last age bin.
Each data point represents the average CPS score for
participants in the respective age bin. The I bars represent standard errors. The graph is based on 307 measurements from 53 carriers and 24,152 measurements
from 3699 noncarriers. Patients in whom Alzheimer’s
disease had been diagnosed were not included in the
analysis.
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
113
The
n e w e ng l a n d j o u r na l
theory about the pathogenesis of this disease, inflammation is a downstream effect of amyloidogenesis, which provides a trigger for the inflammatory response.
Genomewide association studies have also provided evidence of the importance of inflammation
in Alzheimer’s disease. Thus, low-risk variants
have been found in CR1,29 which belongs to the
complement factor family of genes; in MS4A6A and
MS4A4E,30 which are members of a cell-surface
gene family expressed in lymphoid tissue; and in
CD33,31 which encodes a myeloid cell-surface receptor.
TREM2 was originally identified as a DAP12associated receptor that was expressed on macrophages and dendritic cells32 and was later shown
to be expressed on osteoclasts and microglia.33
TREM2 is a transmembrane glycoprotein, consisting of an extracellular immunoglobulin-like
domain, a transmembrane domain, and a cytoplasmic tail, which associates with DAP12 for its
signaling function.32,34 TREM2 has both exogenous ligands on pathogens and endogenous ligands that remain largely unknown, although a
recent study has shown that Hsp60 is an agonist
of TREM2 in neuroblastoma cells and astrocytes.35 In addition, an endogenous ligand on
dendritic cells has been found.36
In brain cells, TREM2 is primarily expressed
on microglia, the resident histiocytes of the central nervous system.37 Activation of microglia may
lead to phagocytosis of cell debris and amyloid,
but microglia can also be activated to promote the
production of proinflammatory cytokines, or they
may differentiate into antigen-presenting cells.38
A recent study showed that TREM2 expression is
induced concomitantly with the formation of
amyloid plaques in APP transgenic mice expressing the Swedish mutation (K670N/M671L) in
APP,39 and this expression was found to correlate
positively with amyloid phagocytosis by unactivated microglia.
The expression of TREM2 also correlated positively with the ability of microglia to stimulate
the proliferation of CD4+ T cells, as well as the
secretion of tumor necrosis factor and CCL2, but
not interferon-γ, into the extracellular milieu.
This led the authors to speculate that TREM2positive microglia on plaques capture and present self-antigens to lymphocytes infiltrating the
central nervous system without promoting proinflammatory responses.39 Furthermore, knock-
114
n engl j med 368;2
of
m e dic i n e
down of TREM2 or DAP12 in microglia resulted
in reduced phagocytosis of apoptotic neurons,
whereas the overexpression of TREM2 increased
such phagocytosis,40 suggesting that microglia
recognize and phagocytose apoptotic neurons
through TREM2 ligation. TREM2 has an antiinflammatory function; it inhibits macrophage response to ligation of toll-like receptor (TLR),41 and
it negatively regulates TLR-mediated maturation
of dendritic cells, type I interferon responses,
and the induction of antigen-specific T-cell proliferation.36 Furthermore, TREM2 stimulation of
dendritic cells induces partial activation without
any production of proinflammatory cytokines.34
Polycystic lipomembranous osteodysplasia with
sclerosing leukoencephalopathy, which produces
increased signals from the deep white matter of
the brain on T2-weighted magnetic resonance imaging, is called Nasu–Hakola disease. It is a rare
recessively inherited disease that is characterized
by painful bone cysts in wrists and ankles, psychotic symptoms, and progressive presenile dementia with onset in the fourth decade of life,
usually leading to death in the fifth decade of
life.42-44 Loss-of-function mutations in DAP12 and
TREM2 were originally found in patients with
Nasu–Hakola disease about a decade ago,45,46 suggesting that the TREM2–DAP12–mediated pathway may be important for human brain and bone
tissue. Nasu–Hakola disease and Alzheimer’s disease are distinct from each other, and the clinical
symptoms of Nasu–Hakola disease (early onset,
painful bone cysts, fractures of bones of the limbs,
and sclerosing leukoencephalopathy) are incompatible with the diagnosis of Alzheimer’s disease.
Bearing in mind that it is possible that rare mutations accounting for a small proportion of cases
of common diseases may define a clinical subgroup, we looked for but did not find clinical features (e.g., sex distribution, radiographic features,
and rate of disease progression) that clearly separate carriers of the R47H mutation from noncarriers with Alzheimer’s disease, although the age
at disease onset was on average 3.18 years earlier in the carriers than in the noncarriers.
A homozygous mutation in the 5′ consensus
donor splice site in intron 1 of TREM2 in a Lebanese family, leading to early-onset dementia without bone cysts, has been reported.47 Furthermore,
three homozygous mutations in TREM2 have recently been reported in three Turkish probands
with frontotemporal dementia-like disease in the
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
Variant of TREM2 Associated with Alzheimer’s Disease
absence of bone cysts,48 and there is also a report
of memory deficits in heterozygous carriers of a
loss-of-function mutation in TREM2 in an Italian
family.49 These findings suggest that TREM2 may
be crucial for the integrity of cognitive function.
The R47H substitution encoded by rs75932628T is located within the extracellular immunoglobulin-like domain of TREM2. The amino acid
substitution may result in decreased affinity of
TREM2 for its natural ligands and affect its signaling. It has recently been proposed that
TREM2 may represent a proteolytic substrate for
γ-secretase, although the exact cleavage site was
not identified.50 If this proteolytic activity is confirmed, processing of TREM2 may be affected by
the R47H substitution.
In conclusion, we have found a new risk variant, rs75932628-T, for Alzheimer’s disease. Although this variant occurs with less frequency
than the ApoE ε4 allele, it confers a risk of Alzheimer’s disease with an effect size that is similar to that of ApoE ε4. Given the involvement of
TREM2 in the phagocytic role of microglia on
amyloid plaques, it is possible that reduced TREM2
activity caused by the R47H substitution may lead
to brain damage through the inability of the
brain to clear these toxic products.
References
1. Castellani RJ, Rolston RK, Smith MA.
Alzheimer disease. Dis Mon 2010;56:484546.
2. Qiu C, Kivipelto M, von Strauss E.
Epidemiology of Alzheimer’s disease: occurrence, determinants, and strategies
toward intervention. Dialogues Clin Neurosci 2009;11:111-28.
3. Cruts M, Theuns J, Van Broeckhoven
C. Locus-specific mutation databases for
neurodegenerative brain diseases. Hum
Mutat 2012;33:1340-4.
4. Bertram L, Lill CM, Tanzi RE. The genetics of Alzheimer disease: back to the
future. Neuron 2010;68:270-81.
5. Saunders AM, Strittmatter WJ, Schmechel D, et al. Association of apolipoprotein E allele epsilon 4 with late-onset familial and sporadic Alzheimer’s disease.
Neurology 1993;43:1467-72.
6. Strittmatter WJ, Saunders AM,
Schmechel D, et al. Apolipoprotein E:
high-avidity binding to beta-amyloid and
increased frequency of type 4 allele in
late-onset familial Alzheimer disease.
Proc Natl Acad Sci U S A 1993;90:1977-81.
7. Holm H, Gudbjartsson DF, Sulem P, et
al. A rare variant in MYH6 is associated
with high risk of sick sinus syndrome. Nat
Genet 2011;43:316-20.
Supported by grants from the National Institute on Aging
(P50-AG025688, to Dr. Levey, for samples from the Alzheimer’s
Disease Center at Emory University; U01AG006781 for the Alzheimer’s Disease Patient Registry and Adult Changes in Thought
study; and the Division of Neuroscience for the NIA-LOAD
study); a grant from the Research Council of Norway and SouthEastern Norway Health Authority for samples from Norway; a
gift from 3M to expand the Adult Changes in Thought cohort; a
grant from the National Institutes of Health (U01HG004438) for
genotyping at Johns Hopkins University; cooperative agreements with the National Human Genome Research Institute
(U01HG004610 for genomewide association analyses); by the
eMERGE Administrative Coordinating Center (U01HG004603)
for assistance with phenotype harmonization and genotype data
cleaning; and by the National Center for Biotechnology Information. The Rotterdam Study was funded by Erasmus Medical
Center and Erasmus University, Rotterdam; the Netherlands
Organization for Health Research and Development; the Research Institute for Diseases in the Elderly; the Ministry of Education, Culture and Science; the Ministry for Health, Welfare
and Sports; the European Commission; and the Municipality of
Rotterdam; by a grant (014-93-015; RIDE2) from the Research
Institute for Diseases in the Elderly, Stichting Alzheimer Onderzoek, Hersenstichting Nederland, the Netherlands Genomics
Initiative–Netherlands Organization for Scientific Research
(Center for Medical Systems Biology and the Netherlands Consortium for Healthy Aging), the Seventh Framework Program
(FP7/2007-2013), and the ENGAGE project (grant agreement
HEALTH-F4-2007-201413).
Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
We thank Aaron Isaacs, Ben Oostra, and Jeannette Vergeer for
genotyping and Renee de Bruijn for the clinical workup in the
Rotterdam study; and Pål Zeiner and Heidi Aase for access to
Norwegian control samples.
8. Sulem P, Gudbjartsson DF, Walters
GB, et al. Identification of low-frequency
variants associated with gout and serum
uric acid levels. Nat Genet 2011;43:112730.
9. Stacey SN, Sulem P, Jonasdottir A, et al.
A germline variant in the TP53 polyadenylation signal confers cancer susceptibility. Nat Genet 2011;43:1098-103.
10. Rafnar T, Gudbjartsson DF, Sulem P, et
al. Mutations in BRIP1 confer high risk of
ovarian cancer. Nat Genet 2011;43:1104-7.
11. Jonsson T, Atwal JK, Steinberg S, et al.
A mutation in APP protects against Alzheimer’s disease and age-related cognitive decline. Nature 2012;488:96-9.
12. McKhann G, Drachman D, Folstein
M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group
under the auspices of Department of
Health and Human Services Task Force
on Alzheimer’s Disease. Neurology 1984;
34:939-44.
13. Morris JN, Hawes C, Fries BE, et al.
Designing the national resident assessment instrument for nursing homes. Gerontologist 1990;30:293-307.
14. Morris JN, Fries BE, Steel K, et al.
Comprehensive clinical assessment in com-
n engl j med 368;2
nejm.org
munity setting: applicability of the MDSHC. J Am Geriatr Soc 1997;45:1017-24.
15. Morris JN, Fries BE, Mehr DR, et al.
MDS Cognitive Performance Scale. J Gerontol 1994;49:M174-M182.
16. Brækhus A, Ulstein I, Wyller TB,
Engedal K. The Memory Clinic — outpatient assessment when dementia is suspected. Tidsskr Nor Laegeforen 2011;131:
2254-7.
17. Athanasiu L, Mattingsdal M, Kahler
AK, et al. Gene variants associated with
schizophrenia in a Norwegian genomewide study are replicated in a large
European cohort. J Psychiatr Res 2010;
44:748-53.
18. Hofman A, Breteler MM, van Duijn
CM, et al. The Rotterdam Study: objectives and design update. Eur J Epidemiol
2007;22:819-29.
19. Hofman A, van Duijn CM, Franco OH,
et al. The Rotterdam Study: 2012 objectives and design update. Eur J Epidemiol
2011;26:657-86.
20. Kutyavin IV, Milesi D, Belousov Y, et
al. A novel endonuclease IV post-PCR genotyping system. Nucleic Acids Res 2006;
34(19):e128.
21. Eikelenboom P, Stam FC. Immunoglobulins and complement factors in se-
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.
115
Variant of TREM2 Associated with Alzheimer’s Disease
nile plaques: an immunoperoxidase study.
Acta Neuropathol 1982;57:239-42.
22. Ishii T, Haga S. Immuno-electronmicroscopic localization of complements
in amyloid fibrils of senile plaques. Acta
Neuropathol 1984;63:296-300.
23. Akiyama H, Barger S, Barnum S, et al.
Inflammation and Alzheimer’s disease.
Neurobiol Aging 2000;21:383-421.
24. Griffin WS, Stanley LC, Ling C, et al.
Brain interleukin 1 and S-100 immunoreactivity are elevated in Down syndrome
and Alzheimer disease. Proc Natl Acad Sci
U S A 1989;86:7611-5.
25. Rogers J, Luber-Narod J, Styren SD,
Civin WH. Expression of immune systemassociated antigens by cells of the human
central nervous system: relationship to
the pathology of Alzheimer’s disease.
Neurobiol Aging 1988;9:339-49.
26. Rozemuller JM, Eikelenboom P, Stam
FC. Role of microglia in plaque formation
in senile dementia of the Alzheimer type:
an immunohistochemical study. Virchows
Arch B Cell Pathol Incl Mol Pathol 1986;
51:247-54.
27. Eikelenboom P, van Exel E, Hoozemans JJ, Veerhuis R, Rozemuller AJ, van
Gool WA. Neuroinflammation — an early
event in both the history and pathogenesis of Alzheimer’s disease. Neurodegener
Dis 2010;7:38-41.
28. Blalock EM, Chen KC, Stromberg AJ,
et al. Harnessing the power of gene microarrays for the study of brain aging and
Alzheimer’s disease: statistical reliability
and functional correlation. Ageing Res
Rev 2005;4:481-512.
29. Lambert JC, Heath S, Even G, et al.
Genome-wide association study identifies
variants at CLU and CR1 associated with
Alzheimer’s disease. Nat Genet 2009;41:
1094-9.
30. Hollingworth P, Harold D, Sims R, et
al. Common variants at ABCA7, MS4A6A/
MS4A4E, EPHA1, CD33 and CD2AP are
associated with Alzheimer’s disease. Nat
Genet 2011;43:429-35.
31. Bertram L, Lange C, Mullin K, et al.
Genome-wide association analysis reveals
putative Alzheimer’s disease susceptibility loci in addition to APOE. Am J Hum
Genet 2008;83:623-32.
32. Bouchon A, Dietrich J, Colonna M.
Cutting edge: inflammatory responses
can be triggered by TREM-1, a novel receptor expressed on neutrophils and
monocytes. J Immunol 2000;164:4991-5.
33. Colonna M. TREMs in the immune
system and beyond. Nat Rev Immunol
2003;3:445-53.
34. Bouchon A, Hernández-Munain C,
Cella M, Colonna MA. DAP12-mediated
pathway regulates expression of CC chemokine receptor 7 and maturation of human dendritic cells. J Exp Med 2001;
194:1111-22.
35. Stefano L, Racchetti G, Bianco F, et al.
The surface-exposed chaperone, Hsp60,
is an agonist of the microglial TREM2 receptor. J Neurochem 2009;110:284-94.
36. Ito H, Hamerman JA. TREM-2, triggering receptor expressed on myeloid cell2, negatively regulates TLR responses in
dendritic cells. Eur J Immunol 2012;42:
176-85.
37. Sessa G, Podini P, Mariani M, et al.
Distribution and signaling of TREM2/
DAP12, the receptor system mutated in
human polycystic lipomembraneous osteodysplasia with sclerosing leukoencephalopathy dementia. Eur J Neurosci 2004;20:
2617-28.
38. Carson MJ, Doose JM, Melchior B,
Schmid CD, Ploix CC. CNS immune privilege: hiding in plain sight. Immunol Rev
2006;213:48-65.
39. Melchior B, Garcia AE, Hsiung BK, et
al. Dual induction of TREM2 and tolerance-related transcript, Tmem176b, in
amyloid transgenic mice: implications for
vaccine-based therapies for Alzheimer’s
disease. ASN Neuro 2010;2(3):e00037.
40. Takahashi K, Rochford CD, Neumann
H. Clearance of apoptotic neurons without inflammation by microglial triggering receptor expressed on myeloid cells-2.
J Exp Med 2005;201:647-57.
41. Hamerman JA, Jarjoura JR, Humphrey
MB, Nakamura MC, Seaman WE, Lanier
LL. Cutting edge: inhibition of TLR and
FcR responses in macrophages by triggering receptor expressed on myeloid cells
(TREM)-2 and DAP12. J Immunol 2006;
177:2051-5.
42. Hakola HP. Neuropsychiatric and genetic aspects of a new hereditary disease
characterized by progressive dementia
and lipomembranous polycystic osteodysplasia. Acta Psychiatr Scand Suppl 1972;
232:1-173.
43. Madry H, Prudlo J, Grgic A, Freyschmidt J. Nasu-Hakola disease (PLOSL):
report of five cases and review of the literature. Clin Orthop Relat Res 2007;
454:262-9.
44. Nasu T, Tsukahara Y, Terayama K. A
lipid metabolic disease — “membranous
lipodystrophy” — an autopsy case demonstrating numerous peculiar membranestructures composed of compound lipid
in bone and bone marrow and various
adipose tissues. Acta Pathol Jpn 1973;
23:539-58.
45. Paloneva J, Kestilä M, Wu J, et al. Lossof-function mutations in TYROBP (DAP12)
result in a presenile dementia with bone
cysts. Nat Genet 2000;25:357-61.
46. Paloneva J, Manninen T, Christman
G, et al. Mutations in two genes encoding
different subunits of a receptor signaling
complex result in an identical disease
phenotype. Am J Hum Genet 2002;71:65662. [Erratum, Am J Hum Genet 2003;72:
225.]
47. Chouery E, Delague V, Bergougnoux
A, Koussa S, Serre JL, Mégarbané A. Mutations in TREM2 lead to pure early-onset
dementia without bone cysts. Hum Mutat
2008;29(9):E194-E204.
48. Guerreiro RJ, Lohmann E, Brás JM, et
al. Using exome sequencing to reveal mutations in TREM2 presenting as a frontotemporal dementia-like syndrome without bone involvement. Arch Neurol 2012
October 8 (Epub ahead of print).
49. Montalbetti L, Ratti MT, Greco B,
Aprile C, Moglia A, Soragna D. Neuropsychological tests and functional nuclear
neuroimaging provide evidence of subclinical impairment in Nasu-Hakola disease heterozygotes. Funct Neurol 2005;20:
71-5.
50. Wunderlich P. γ-Secretase mediated
proteolytic processing of the triggering
receptor expressed on myeloid cells-2 —
functional implications for intracellular
signalling. Bonn, Germany: University of
Bonn, 2010.
Copyright © 2012 Massachusetts Medical Society.
RECEIVE THE JOURNAL’S TABLE OF CONTENTS EACH WEEK BY E-MAIL
To receive the table of contents of the Journal by e-mail
every Wednesday evening, sign up at NEJM.org.
116
n engl j med 368;2
nejm.org
january 10, 2013
The New England Journal of Medicine
Downloaded from nejm.org on February 18, 2022. For personal use only. No other uses without permission.
Copyright © 2013 Massachusetts Medical Society. All rights reserved.