C9orf72 Hexanucleotide Repeat Expansions in Clinical Alzheimer Disease

Importance Hexanucleotide repeat expansions in the chromosome 9 open reading frame 72 (C9orf72) gene underlie a significant fraction of frontotemporal dementia and amyotrophic lateral sclerosis.

Objective To investigate the frequency of C9orf72 repeat expansions in clinically diagnosed late-onset Alzheimer disease (AD).

Design, Setting, and Patients This case-control study genotyped the C9orf72 repeat expansion in 872 unrelated familial AD cases and 888 control subjects recruited as part of the National Institute on Aging Late-Onset Alzheimer Disease Family Study cohort, a multisite collaboration studying 1000 families with 2 or more individuals clinically diagnosed as having late-onset AD.

Main Outcomes and Measures We determined the presence or absence of the C9orf72 repeat expansion by repeat-primed polymerase chain reaction, the length of the longest nonexpanded allele, segregation of the genotype with disease, and clinical features of repeat expansion carriers.

Results Three families showed large C9orf72 hexanucleotide repeat expansions. Two additional families carried more than 30 repeats. Segregation with disease could be demonstrated in 3 families. One affected expansion carrier had neuropathology compatible with AD. In the National Institute on Aging Late-Onset Alzheimer Disease Family Study series, the C9orf72 repeat expansions constituted the second most common pathogenic mutation, just behind the PSEN1 A79V mutation, highlighting the heterogeneity of clinical presentations associated with repeat expansions.

Conclusions and Relevance C9orf72 repeat expansions explain a small proportion of patients with a clinical presentation indistinguishable from AD, and they highlight the necessity of screening frontotemporal dementia genes in clinical AD cases with strong family history.

Alzheimer disease (AD) constitutes the most common cause of dementia in clinical practice.¹ Alzheimer disease manifests clinically with progressive cognitive dysfunction with prominent memory loss, while its neuropathology is characterized by the presence of extracellular neuritic plaques and intracellular neurofibrillary tangles.¹ Familial AD can be caused by mutations in amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2).^2-6 Interestingly, mutations in genes typically associated with frontotemporal dementia (FTD), including microtubule-associated protein tau (MAPT) and granulin (GRN), have also been reported in clinical AD cases,^6-13 highlighting the overlapping phenotypes of AD and amnestic FTD.

Recently, an intronic hexanucleotide repeat (GGGGCC) in the chromosome 9 open reading frame 72 (C9Oorf72) gene was identified as a frequent cause of sporadic and familial FTD and amyotrophic lateral sclerosis (ALS).^14,15 Because some individuals with clinical AD carry mutations in FTD genes,^6-13,16,17 2 studies investigated the role of C9orf72 expansions in AD, with conflicting results. The first report identified C9orf72 repeat expansions in 3 of 342 cases with familial AD and 6 of 711 cases with sporadic AD. Interestingly, reanalysis of autopsy material from 2 expansion carriers showed FTD pathology,¹⁸ suggesting that both patients had amnestic presentations of FTD rather than AD. The second study identified no repeat expansions in 568 patients with probable AD by clinical criteria and concluded that C9orf72 repeats are specific for FTD.¹⁹

To clarify the role of C9orf72 hexanucleotide repeat expansions in familial late-onset AD, this study screened 872 unrelated AD cases and 888 unrelated control subjects and investigated whether expanded repeats were associated with age at dementia onset. Furthermore, we addressed whether a larger, but nonexpanded, GGGGCC allele was associated with the risk for AD or age at onset.

Individuals from 872 unrelated families with at least 2 individuals affected by AD and 888 unrelated unaffected control subjects from the National Institute on Aging Late-Onset Alzheimer Disease (NIA-LOAD) Family Study were included. All AD cases had been diagnosed as having dementia of the Alzheimer type using criteria equivalent to the National Institute of Neurological and Communication Disorders and Stroke–Alzheimer Disease and Related Disorders Association for probable AD.^20,21 All patients had a family history of AD but not other types of dementia or other neurodegenerative diseases. Probands were required to have a diagnosis of definite or probable late-onset AD (>60 years) and a sibling with definite, probable, or possible late-onset AD with a similar age at onset. A third biologically related family member (first, second, or third degree) was also required, regardless of affected status. If unaffected, this individual had to be 60 years of age or older, but 50 years of age or older if diagnosed as having late-onset AD or mild cognitive impairment.²² Within each pedigree, we selected a single individual to screen by identifying the youngest affected family member with the most definitive diagnosis (ie, individuals with autopsy confirmation were chosen over those with clinical diagnosis only). A summary of the demographics of all subjects is shown in Table 1. Written informed consent was obtained from all participants, and the study was approved by local institutional review board committees.

The presence of the expanded hexanucleotide repeat and the number of repeat units in the longest allele were determined using previously reported methods for repeat-primed polymerase chain reaction (PCR) and fluorescence-based fragment size analysis.¹⁴ Briefly, repeat-primed PCR was performed in a total reaction volume of 28 μL containing 100 ng genomic DNA, 1x FastStart PCR Master Mix (Roche Applied Science), 3.5% dimethyl sulfoxide, 1x Q solution (Quiagen), and 0.18 mM of deaza-dGTP (New England Biolabs). Primer concentrations and sequences (chr9:27563580F and chr9:27563465R) were the same as previously reported.¹⁴ Polymerase chain reaction products were run on an ABI 3130xl Genetic Analyzer (Applied Biosystems) and analyzed using GeneMapper. Consistent with standards used in prior studies, a sample was considered to have a repeat expansion when assay replicates demonstrated more than 30 peaks and a decrementing saw-tooth pattern with 6 base-pair periodicity (eFigure). A normal repeat allele was defined as having 30 or fewer peaks.^14,15,18,19

In all cases, repeat expansions identified with repeat-primed PCR were confirmed by attempting to perform PCR across the GGGGCC repeat using fluorescence-based fragment size analysis (cross-repeat PCR), as reported with minor modifications.¹⁴ Briefly, using previously reported primers with fluorescent labels, PCR was performed on 100 ng genomic DNA with 1x Phusion High-Fidelity DNA Polymerase Master mix (Thermo Fisher Scientific), 5% dimethyl sulfoxide, 1.25 M betaine, and 0.2 mM deaza-dGTP. Products were run on a 2% agarose gel for visual inspection and then analyzed for fragment size determination, as described for repeat-primed PCR. Based on our analysis, cross-repeat PCR can amplify alleles with 35 or fewer GGGGCC repeats but not the 700 to 1600 repeats reported for pathologic expansions.¹⁴ To investigate repeat expansion segregation, we also genotyped all family members (n = 491) from every pedigree with 10 or more hexanucleotide repeats.

Southern blot hybridization analysis was conducted using previously described methods and probe sequences¹⁴ to estimate the number of repeat units in expansion carriers.

We also analyzed all repeat expansion carriers for the 24 single nucleotide polymorphism (SNP) at-risk haplotypes that have been associated with pathologically expanded C9orf72 repeats.²³ All samples were genotyped with the Illumina Human 610 Beadchip, with direct genotyping of all analyzed SNPs. Stringent quality control criteria were applied to remove low-quality SNPs.²² We used the entire NIA-LOAD genomewide association study data set and the HapMap CEU population as reference populations. Mach software²⁴ was used to phase the 24 SNPs.

A linear regression model (SAS Institute Inc) was used to test whether the number of GGGGCC repeats was associated with the risk for AD, including age, sex, and apolipoprotein E (APOE) genotype as covariates. Association with age at onset was carried out using the Kaplan-Meier method and tested for significant differences using a Cox proportional hazards model (proc PHREG, SAS Institute Inc) that included sex and APOE status.

We screened 872 unrelated familial AD cases and 888 unrelated control subjects for expansions of the hexanucleotide repeat in C9orf72. Five white individuals with a clinical diagnosis of AD (0.57%) and 1 normal control subject (age, 73 years) showed an abnormal expansion with repeat-primed PCR, defined as more than 30 repeats with 6 base-pair periodicity and the expected decrementing saw-tooth pattern (eFigure). To better clarify the number of repeats in these patients, we attempted to perform PCR across the GGGGCC repeat. However, in all 6 individuals, only a single peak representing the normal size allele was obtained. Southern blot of AD probands showed expansions of 1200 to 1300 GGGGCC repeats for families 1, 2, and 3, but only 35 to 100 repeat units in families 4 and 5 (data not shown). Regardless of repeat size, all 5 AD cases were found to carry the at-risk haplotype associated with pathologic repeat expansions in all patients with FTD and ALS reported to date (eTable 1).²³ Results from all 5 of these patients were negative for the most common mutations in APP, PSEN1, PSEN2, GRN, and MAPT.⁶

Pedigrees for the 5 families with abnormal (>30) C9orf72 repeats are shown in Figure 1 and clinically summarized in Table 2. All additional affected individuals had been diagnosed as having probable AD based on the National Institute of Neurological and Communication Disorders and Stroke–Alzheimer Disease and Related Disorders Association criteria. DNA for additional family members was available for all but family 1. Family 2 showed complete segregation of the large repeat expansion and risk haplotype with disease status (Figure 1 and eTable 2). Family 3, in which all the affected individuals were APOE 3/3 homozygous, showed segregation of the expanded repeat and risk haplotype except for a single individual who developed dementia with only 7 repeat units. Additionally, this individual did not carry the risk haplotype and had dementia onset at a later age than relatives with repeat expansions. In families 4 and 5 (repeat expansions between 35 and 100 repeats), the characteristic repeat-primed chromatogram pattern and the risk haplotype segregated perfectly with disease status. Overall, in these 5 families, 10 of 11 affected individuals for whom DNA was available carried abnormal GGGGCC expansions and the risk haplotype. It is important to note that all the individuals with more than 30 repeats also carried the risk haplotype.

The age at onset for individuals with more than 30 repeats was earlier than cases with normal repeat alleles (mean [SD], 65.6 [5.5] vs 71.4 [6.8] years; P = .04; Figure 2). This association retained statistical significance even after APOE genotypes were included in the model (P = .03).

The proband of family 1 (1200-1300 repeats), an APOE 2/3 carrier, was diagnosed as having dementia, Alzheimer type, at age 73 years. Formal neurocognitive testing was not available, but prominent memory loss with repetitive questioning and wandering behavior had developed by age 78 years. On death 12 years after diagnosis, a brain autopsy was carried out by the Brain Bank at McLean Hospital in Belmont, Massachusetts, in 1999. Microscopic examination found extensive plaque and tangle pathology. Neurofibrillary tangles and neuritic plaques (>40 plaques per 100X field) were present throughout the neocortex, hippocampus, amygdala, and nucleus basalis of Meynert. Scattered neocortical Lewy bodies were seen. Severe neuronal loss in the substantia nigra pars compacta was accompanied by a few Lewy bodies. This case would be diagnosed as having NIA-Reagan high likelihood criteria for AD and coexisting neocortical-predominant Lewy body dementia using current criteria.^25,26 Tissue was no longer available to carry out additional relevant staining (eg, tau, ubiquitin, TDP-43, or p62), preventing an assessment for possible FTD pathology.

Aside from age at symptom onset and meeting clinical criteria for probable AD, limited clinical information was available for the remaining index cases. The proband of family 2 (1200-1300 repeats) was diagnosed clinically as having probable AD at the age of 71 years, 3 years after disorientation and memory loss began. By the time of study inclusion at age 74 years, the patient was nonverbal and too impaired to participate in cognitive testing. The proband of family 3 (1200-1300 repeats) was diagnosed as having probable AD at the age of 60 years. Three years later, the Clinical Dementia Rating score was 0.5, which progressed over 3 years to 1.0. Review of limited caregiver reports suggested that difficulties with disinhibition and anxiety were out of proportion to memory impairment. No additional clinical information was available for members of families 4 and 5 (35-100 repeats in each).

Previous studies in ALS and FTD have clearly demonstrated that C9orf72 repeat expansions are causative for disease.¹⁴ However, the minimum repeat number required for disease has not been established. Furthermore, it is not known whether higher repeat numbers (but still within the normal range) are associated with the risk for ALS or FTD. To address this question in AD, we compared the longest nonexpanded allele in cases and control subjects. The average longest allele was not statistically different from control subjects (mean [SD], 6.5 [4.1] vs 4.48 [3.7] repeat units; P = .10), and the distribution of longest allele lengths were similar. Thus within the normal range, higher repeat numbers did not appear to be a risk factor for AD in this population. Furthermore, we found no association between the length of the longest nonexpanded allele and age at onset (P = .52), nor evidence for an interaction with APOE genotype.

In a previous study, we sequenced APP, PSEN1, PSEN2, MAPT, and GRN genes in a discovery series comprising 439 cases included in this study.⁶ The most common pathogenic mutation identified by sequencing in the discovery series (PSEN1 A79V) was then genotyped in the entire cohort (follow-up series). Overall, the A79V mutation was found in 4 of the 872 cases (0.46%)⁶ compared with the 5 pedigrees where abnormal C9orf72 repeat expansions were found. Furthermore, we analyzed the overall frequency of AD gene mutations (APP, PSEN1, and PSEN2) vs FTD gene mutations (MAPT, GRN, and C9orf72) and found that 1.82% of probands carried a pathogenic, or very likely pathogenic, mutation in APP, PSEN1, and PSEN2, while a slightly larger number (1.94%) had mutations in MAPT, GRN, or C9orf72.

This study assessed C9orf72 hexanucleotide expansions in familial late-onset AD cases and normal control subjects, identifying 5 AD families carrying abnormal C9orf72 hexanucleotide repeat expansions. This frequency is very similar to that found in an independent AD series,¹⁵ but significantly lower than in FTD or ALS.

Three families with clinical AD (0.34%) were found to have repeat expansions in the range reported for FTD and ALS (>1 000). In a previous report documenting expansions in clinically diagnosed AD, reevaluation of autopsy material demonstrated FTD pathology and suggested that AD cases with C9orf72 repeat expansions represent amnestic variants of FTD.^{9,14,15,27,28} We were unable to perform an equivalent analysis because autopsies had not been performed or tissue was no longer available. Therefore, even in the proband from family 1, where the neuropathology showed coexisting AD and Lewy body dementia pathology, we cannot rule out the possibility that the families identified in this study also represented amnestic presentations of FTD rather than AD. It is notable that of the many C9orf72 FTD cases reported, several showed concurrent AD pathology^{9,14,15,27,28} and an additional expansion carrier with FTD+ALS demonstrated enough plaque and tangle pathology to meet diagnostic criteria for AD.²⁷ These cases suggest that some individuals with C9orf72 repeat expansions could present with clinical symptoms of AD and have a high enough burden of AD neuropathology that biomarker analysis (cerebrospinal tau or β-amyloid, and/or Pittsburgh Compound B–positron emission tomography neuroimaging) would also support an AD diagnosis. This hypothesis is supported by 3 recently reported individuals with early-onset AD, cerebrospinal fluid profiles typical of AD,²⁹ who were found to carry C9orf72 repeat expansions. In this setting, the correct diagnosis (amnestic FTD) would presumably only be reached by neuropathologic studies or genetic testing. Our cases and previous studies reinforce the heterogeneous clinical and neuropathologic presentations of C9orf72 repeat expansions (ALS, frontotemporal lobar degeneration, frontotemporal lobar degeneration–ALS, and clinical AD).

We also identified 2 families carrying smaller, but abnormal repeats (>35, but < 100 units). Despite segregating with disease status, it remains unclear whether these smaller repeat expansions cause disease, increase the risk for dementia, or are incidental. Future studies correlating quantified repeat sizes with disease status will be required to answer this question.

Although the frequency of large C9orf72 repeat expansions was low in our cohort, it was the second most common pathogenic mutation (3 of 872), just behind PSEN1 A79V (4 of 872). In addition, mutations in FTD genes were as common as mutations in AD genes (1.94% vs 1.87%). Our results confirm that the clinical phenotype of mutations in FTD genes, including GRN, MAPT, and C9orf72, can be clinically indistinguishable from typical AD.^30-34 This fact has important implications for clinicians, who should consider both FTD and AD genes when evaluating families with strong histories of AD.

Correspondence: Carlos Cruchaga, PhD, Department of Psychiatry, Washington University School of Medicine, 660 S Euclid Ave, B8134, St Louis, MO 63110 (cruchagc@psychiatry.wustl.edu).

Accepted for Publication: August 27, 2012.

Published Online: April 15, 2013. doi:10.1001/2013.jamaneurol.537

Author Contributions: Drs Harms and Cruchaga contributed equally to this work. Study concept and design: Harms, Mayeux, Goate, and Cruchaga. Acquisition of data: Harms, Cairns, B. Cooper, P. Cooper, Mayo, Carrell, Faber, Williamson, Bird, Diaz-Arrastia, Foroud, Graff-Radford, Mayeux, and Chakraverty. Analysis and interpretation of data: Harms, Benitez, Cairns, P. Cooper, Boeve, and Cruchaga. Drafting of the manuscript: Harms, Benitez, and Cruchaga. Critical revision of the manuscript for important intellectual content: Harms, Cairns, B. Cooper, P. Cooper, Mayo, Carrell, Faber, Williamson, Bird, Diaz-Arrastia, Foroud, Boeve, Graff-Radford, Mayeux, Chakraverty, Goate, and Cruchaga. Statistical analysis: Cruchaga. Obtained funding: Mayeux, Goate, and Cruchaga. Administrative, technical, and material support: Harms, Cairns, B. Cooper, P. Cooper, Mayo, Carrell, Faber, Williamson, Diaz-Arrastia, Boeve, Mayeux, and Chakraverty. Study supervision: Williamson, Diaz-Arrastia, and Foroud.

NIA-LOAD/NCRAD Family Study Consortium Investigators: Robert Green, Neil Kowall, Lindsay Farrer, Boston University, Massachusetts; Jennifer Williamson, Vincent Santana, Columbia University, New York, New York; Donald Schmechel, Perry Gaskell, Kathleen Welsh-Bohmer, Margaret Pericak-Vance, Duke University, Durham, North Carolina; Bernardino Ghetti, Martin R. Farlow, Kelly Horner, Indiana University, Bloomington; John H. Growdon, Deborah Blacker, Rudolph E. Tanzi, Bradley T. Hyman, Massachusetts General Hospital, Boston; Bradley Boeve, Karen Kuntz, Lindsay Norgaard, Nathan Larson, Mayo Clinic, Rochester, Minnesota; Dana Kistler, Fracine Parfitt, Jenny Haddwow, Mayo Clinic, Jacksonville, Florida; Jeremy Silverman, Michal Schnaider Beeri, Mary Sano, Joy Wang, Rachel Lally, Mount Sinai School of Medicine, New York, New York; Nancy Johnson, Marcel Mesulum, Sandra Weintraub, Eileen Bigio, Northwestern University, Evanston, Illinois; Jeffery Kaye, Patricia Kramer, Jessica Payne-Murphy, Oregon Health and Science University, Portland; David Bennett, Holli Jacobs, Jeen-Soo Chang, Danielle Arends, Rush University, Chicago, Illinois; Lindy Harrell, University of Alabama at Birmingham; George Bartzokis, Jeffery Cummings, Po H. Lu, Usha Toland, University of California, Los Angeles; William Markesbery, Charles Smith, Alise Brickhouse, University of Kentucky, Lexington; John Trojanowski, Vivianna Van Deerlin, Elisabeth McCarty Wood, University of Pennsylvania, Philadelphia; Steven DeKosky, Robert Sweet, Elise Weamer, University of Pittsburgh, Pennsylvania; I. Helena Chui, Arousiak Varpetian, University of Southern California, Los Angeles; Ramon Diaz-Arrastia, Roger Rosenberg, Barbara Davis, University of Texas Southwestern, Dallas; Thomas Bird, Malia Rumbaugh, Gerard D. Schellenberg, Murray Raskind, University of Washington, Seattle; Alison Goate, John Morris, Joanne Norton, Denise Levitch, Betsy Grant, Mary Coats, Washington University at St Louis, Missouri.

Funding/Support: This study was supported by grants P30-NS069329-01, R01-AG035083, and K08-NS075094 from the National Institutes of Health. Samples from the National Cell Repository for Alzheimer's Disease, which receives government support under cooperative agreement grant U24 AG21886, were used in this study.

Additional Contributions: We thank contributors, including the Alzheimer's Disease Centers and the Washington University Neuromuscular Genetics Project, which collected samples used in this study, as well as patients and their families, whose help and participation made this work possible.