Characterization of frontotemporal dementia and/or amyotrophic lateral sclerosis associated with the GGGGCC repeat expansion in C9ORF72

Clinical cohorts

One proband was recruited and followed in the Alzheimer's Disease Patient Registry programme, which focused on subjects residing in Olmsted County, Minnesota, and once the FTD phenotype and familial nature of the disorder was appreciated, other relatives were recruited and followed longitudinally. The Mayo Clinic Rochester clinical cohort includes all cases with an FTD or primary progressive aphasia diagnosis regardless of family history, who were enrolled in the Mayo Clinic Alzheimer's Disease Research Centre or Alzheimer's Disease Patient Registry programmes from 1986 to 2011; since recruitment of familial FTD has been a focus of interest for many years, this cohort was expected to have a higher frequency of familial cases. The Mayo Clinic Florida Alzheimer's Disease Research Centre clinical cohort includes all cases with an FTD or primary progressive aphasia diagnosis regardless of family history who were enrolled from 1990 to 2011. The Mayo Clinic Florida ALS cohort includes all cases with ALS, regardless of family history, who were enrolled from December 2007 to August 2011.

The available information on each case with a diagnosis of FTD or primary progressive aphasia was reassessed to determine how each would be classified using the updated behavioural variant FTD (Rascovsky et al., 2011) and primary progressive aphasia (Gorno-Tempini et al., 2011) subtype criteria [almost all cases had been previously classified by the Neary et al. (1998) criteria]. Each case previously classified as FTD indeed met criteria for the behavioural variant FTD diagnosis, and DNA on all available cases were genotyped. Among the reclassified cases with primary progressive aphasia, only cases with the non-fluent/agrammatic and semantic subtypes were included in the DNA screen; since most with the logopenic subtype likely have underlying Alzheimer's disease, they were excluded from this screen. Subjects classified with the primary diagnoses of mild cognitive impairment, Alzheimer's disease, dementia with Lewy bodies, corticobasal syndrome, progressive supranuclear palsy and posterior cortical atrophy were also excluded from the DNA screen for purposes of this initial analysis focused on cases with FTD, primary progressive aphasia and/or ALS.

Subjects were classified as being familial (the presence of a positive family history) or sporadic based on specific sets of criteria. For subjects with a dementia syndrome as the major phenotype, a positive family history was defined as a first- or second-degree relative with FTD and/or ALS or a first-degree relative with memory problems, behavioural changes, parkinsonism or another suspected neurodegenerative disorder. A positive family history in the Mayo Clinic Florida ALS series was defined as a first- or second-degree relative with ALS (Byrne et al., 2011). Since the published criteria for determining familial ALS does not consider a family history of dementia or parkinsonism (Byrne et al., 2011), a looser definition of positive family history was also used, in which the presence of one or more first- or second-degree relatives with dementia, parkinsonism or ALS was also used in frequency calculations, with the strict definition of sporadic disease being the absence of any first or second-degree relatives with dementia, parkinsonism or ALS.

Clinical features and diagnoses

Age of onset for dementia was the age at which the subject first demonstrated behavioural/personality, cognitive ± parkinsonism as noted by themselves, their relatives or their clinicians. Age of onset for ALS was the age at which any symptom reflecting upper motor neuron and/or lower motor neuron dysfunction was noted by themselves, their relatives, or their clinicians. Survival (in years) was based on the number of years from onset of any neurological symptoms to death. All neurological clinical data (Members of the Department of Neurology, 1998) were reviewed. The presence or absence of the following clinical features were recorded: personality/behaviour changes (self explanatory), executive dysfunction (defined as impairment in sustained attention, multi-tasking, decision-making, problem solving, etc.), memory impairment (defined as forgetfulness for details of recent events or upcoming plans and poor delayed word list recall), aphasia (defined as difficulties with object or person naming, or receptive or expressive language functioning), parkinsonism (defined as some combination of masked facies, stooped posture, shuffling gait, rest tremor, bradykinesia, rigidity, postural instability), frontal release signs (defined as grasp, snout or palmomental reflex present), and pseudobulbar affect (defined as the tendency to cry or laugh inappropriately). Features of upper motor neuron and lower motor neuron dysfunction were considered present when specifically recorded in the clinical record, and were subclassified based on the level of involvement. The upper motor neuron features were characterized as follows: bulbar—pseudobulbar affect, snout or palmomental reflex, exaggerated jaw jerk, spastic dysarthria; cervical—hyper-reflexia and/or spasticity in the upper limbs, Hoffmann sign present; thoracic—absent superficial abdominal reflex; lumbar—hyper-reflexia and/or spasticity in the lower limbs, crossed adductor reflex, Babinski sign present. The lower motor neuron features were characterized as follows: bulbar—facial or tongue fasciculations, muscle wasting, and weakness, flaccid dysarthria; cervical—upper limb fasciculations, muscle wasting and weakness; thoracic—trunk or abdominal fasciculations, muscle wasting and weakness; lumbar—lower limb fasciculations, muscle wasting and weakness.

Clinical diagnoses were based on published criteria for FTD, progressive non-fluent aphasia, semantic dementia, ALS, etc. (Neary et al., 1998; Brooks et al., 2000), and as noted above, each case was reclassified using the updated behavioural variant FTD and primary progressive aphasia criteria (Gorno-Tempini et al., 2011; Rascovsky et al., 2011). Patients with ALS were classified as ‘clinically definite’, ‘probable’, ‘probable lab-supported’, ‘possible’ or ‘suspected’ ALS according to the El Escorial criteria, allowing for concomitant presence of signs of FTD in patients with FTD and ALS (Brooks et al., 2000). The site of onset—bulbar, upper and lower spinal level—was also noted based on available historical and clinical data for each case. Those who underwent electromyography were noted as such. The diagnosis of combined FTD/ALS was based on the presence of both FTD and ALS features. Those that had some combination of masked facies, stooped posture, shuffling gait, rest tremor, bradykinesia, rigidity, postural instability, etc. were characterized as having parkinsonism based on examination or informant report; the clinical diagnosis of Parkinson's disease was reserved for those who were formally examined and had the cardinal features based on published criteria (Hughes et al., 1992). Those that had a dementing illness characterized by an insidious onset and progressive course, but were either examined in the late stage of the illness or whose characterization was based on informant report alone, were labelled as having dementia. The designation of ‘clinically definite’ behavioural variant FTD, ALS, parkinsonism, or some combination of these, was reserved for only those subjects who were personally examined by one or more of the co-authors.

Since recent data suggest that psychosis and altered eating behaviour are associated with FTD/ALS and/or TAR DNA binding protein molecular weight 43 (TDP-43) pathology (Lillo et al., 2010; Burrell et al., 2011; Piguet et al., 2011), the presence or absence of these features was also sought by reviewing the available clinical records.

Laboratory analyses

Genomic DNA was extracted from peripheral blood samples using standard procedures. For each subject, genomic DNA was screened for the presence of the expanded hexanucleotide repeat in C9ORF72 using the repeat primed polymerase chain reaction method as previously described (DeJesus-Hernandez et al., 2011). In short, polymerase chain reaction was performed with 100 ng/µl of genomic DNA in the presence of 1 M betaine (Sigma), 5% dimethyl sulphoxide (Sigma), 5 mM of dCTP, dATP, dTTP (Promega) and 7-deaza-2-deoxy GTP (Roche) in substitution for dGTP. The cycling programme included an initial denaturation at 98°C for 10 min followed by 10 cycles of denaturation at 97°C for 35 s, annealing at 64°C for 2 min and extension at 68°C for 8 min, followed by 25 cycles of denaturation at 97°C for 35 s, annealing at 64°C for 2 min and extension at 68°C for 8 min plus an additional 20 s for each cycle.

Neuropsychological examinations

A battery of neuropsychological tests was available for 27 subjects. Four additional subjects were deemed too impaired to complete a full battery and their evaluation was limited to one or more cognitive screening exams, including the Mini-Mental State Examination (Folstein et al., 1975), Dementia Rating Scale-2 (Mattis, 1988) or ALS-Cognitive Behavioural Screen (Woolley et al., 2010). Some subjects who completed the neuropsychological battery also completed one or more screening exams.

The neuropsychological battery included tests of attention and executive functioning [Digit Span (n = 14) (Wechsler, 1981, 1997a), Trail Making Test (n = 20) (Reitan, 1958) and Stroop Colour-Word Test (n = 12) (Golden, 1978)]; language [Boston Naming Test (n = 25) (Kaplan et al., 1983), Controlled Oral Word Association Test (n = 21) (Benton et al., 1994), and semantic/category fluency for animals, fruits and vegetables (n = 21)]; learning and memory [delayed recall from Logical Memory (n = 22) (Wechsler, 1987, 1997b), Visual Reproduction (n = 19) (Wechsler, 1987, 1997b) and Auditory Verbal Learning Test (n = 20) (Rey, 1964) or California Verbal Learning Test-II (n = 4) (Delis et al., 2000)], visuospatial skills [Block Design (n = 13) (Wechsler, 1981, 1997a) and Judgement of Line Orientation (n = 11) (Benton et al., 1994)], and estimated premorbid functioning [Reading subtest from the Wide Range Achievement Test, 3rd edition (n = 17) (Wilkinson, 1993)]. Age-corrected normative data were obtained from the Mayo Older American Normative Studies (Ivnik et al., 1992, 1996, 1997; Lucas et al., 1998a, b) and published test manuals. To minimize variability due to normative reference sources, the youngest Mayo Older American Normative Studies age bracket was extended downward to include subjects aged 50–55 years. Normative data for adults aged <50 years were derived separately for the Trail Making Test (n = 2) (Bornstein, 1985), Boston Naming Test (n = 4) (Tombaugh and Hubley, 1997), and Judgement of Line Orientation (n = 1) (Benton, 1994). Cognitive impairment was defined as test scores 1.5 standard deviations below the respective normative group mean.

Neuroimaging examinations

Demographic and clinical findings

The demographic and clinical findings of pertinent members of the 43 kindreds are presented in Supplementary Table 1. A summary of the key demographic findings is presented in Table 4. Details regarding our index kindred have been previously reported (Boxer et al., 2011).

Neuropsychological findings

Neuropsychological battery

As shown in Fig. 1, subjects demonstrated predominant impairment across timed tasks, with a high percentage showing specific deficits on measures of complex attention/executive functioning and word fluency. There were no obvious differences in the neuropsychological profile of impairment between the behavioural variant FTD, FTD/ALS and ALS phenotypes. Fewer subjects were impaired on tests of basic attention, memory and visual naming. No subject was impaired on a test of single word reading, suggesting relative sparing of this ability and highlighting that test's utility as an index of estimated premorbid cognitive function in FTD/ALS. The discrepancy in impairment between two tests involving visuospatial functioning (Block Design and Judgement of Line Orientation) is worth noting. Block Design is a timed task that requires motor manipulation of blocks to match target designs and relies on processing speed, visuospatial skills and an executive problem-solving component. In contrast, Judgement of Line Orientation is a non-timed task that requires visual discrimination of angular relationships. Eleven of the 13 subjects who completed Block Design obtained an impaired score (84.6%), compared with 1 of 11 (9.1%) subjects who completed the Judgement of Line Orientation task. Of the five subjects who completed both tasks, all had lower scores on Block Design and their mean score was ∼1 standard deviation below their mean score on Judgement of Line Orientation. These results, although from a limited number of subjects, suggest that the high rate of impairment obtained on Block Design may be due to slowed processing speed and executive dysfunction, not visuospatial deficits.

Open in a separate window

Figure 1

Per cent of subjects with impaired neuropsychological test scores. Cognitive impairment defined as test scores less than −1.5 SDs from respective normative sample. AVLT = Auditory Verbal Learning Test; B.Design = Block Design; BNT = Boston Naming Test; Cat. Fluency = Category Fluency; COWAT = Controlled Oral Word Association Test; CVLT-II = California Verbal Learning Test-II; D.Span = Digit Span; JLO = Judgement of Line Orientation; L. Mem = Logical Memory; Stroop C/W = Stroop colour-word interference trial; TMT-B = Trail Making Test part B; V. Rep = Visual Reproduction; WRAT-3 = Reading subtest from the Wide Range Achievement Test-3. Asterisk represents no subject impaired in WRAT-3 Reading test.

Neuroimaging findings

Magnetic resonance imaging

MRI was available for review in 18 patients (14 with behavioural variant FTD, three with FTD/ALS and one with ALS), and representative images highlighting the relatively consistent pattern of symmetric bilateral frontal ± temporal cortical atrophy are presented in Fig. 2. Significant atrophy was determined in individual subject MRI scans using a threshold of Z-score less than −2.5. Atrophy was identified in 15 patients, with frontotemporal predominance in 12. One patient with behavioural variant FTD, one with FTD/ALS and one with ALS had no detectable atrophy. Parietal cortical atrophy was present in five patients and insular atrophy was found in eight patients. Subcortical white matter signal changes were absent in all cases. Longitudinal MRI scans spanning ≥2 years were available in eight patients, which all showed progression of atrophy specifically in the frontal lobes and ventricular enlargement.

Open in a separate window

Figure 2

Atrophy patterns in individual subjects. MRI scans of individual subjects overlaid with regions of severe atrophy in yellow using MRI STAND-maps (Z-score less than −2.5 relative to normals was used as a threshold). The brightness of the overlaid colour indicates the degree of atrophy. Numbers represent age of each subject in years. Kind = kindred; P = proband; R = relative.

SPECT and positron emission tomography

SPECT scans were available for analysis in four subjects (Fig. 3). The findings indicate statistically significant anterior and middle cingulate gyri hypoperfusion compared with controls (P = 0.037) (Supplementary Fig. 2). One subject (Kindred 16, Case IV.2) who had normal findings on his initial MRI (Fig. 1) also underwent both SPECT and PET imaging, and each showed anterior cingulate changes. However, the frontal cortical changes were more prominent on the PET when the subject was also 2 years older. Five subjects underwent FDG-PET. In four of five FDG-PET scans (Fig. 4), anterior cingulate Z-scores were significant (average right 2.2; average left 1.9) and showed consistent hypometabolism in these subjects. Posterior cingulate metabolism was normal in these subjects (average right 0.98; average left 0.98). Frontal cortical hypometabolism in these four subjects ranged from mild to severe. One of the five subjects had a parietal/precuneus pattern of hypometabolism with sparing of frontal cortical regions. These SPECT and PET findings have regional abnormalities that show localization similarities with the salience network (Supplementary Fig. 3).

Open in a separate window

Figure 3

Individual subject relative SPECT intensity maps. SPECT maps shown for a 54-year-old healthy control subject (A), 51-year-old with behavioural variant FTD (B), 51-year-old with behavioural variant FTD (C), 56-year-old with FTDP/ALS (D), and 63-year-old with behavioural variant FTD (E). The relative SPECT intensity for each subject's normalized SPECT scans is overlaid on a template brain for anatomical reference. The healthy control subject is displayed at the top (blue box), and images for subjects B to E are arranged in increasing order of severity of anterior cingulate reduction in SPECT intensity. Colour bar encodes SPECT intensity relative to pontine region of interest.

Open in a separate window

Figure 4

Fluorodeoxyglucose-PET Z-score 3D-SSP images. PET images shown for a 68-year-old with behavioural variant FTD (A), 69-year-old with behavioural variant FTD (B), 64-year-old with behavioural variant FTD (C), 53-year-old with behavioural variant FTD (D), and 57-year-old with behavioural variant FTD (E). The images from left to right are right lateral, left lateral, right medial, and left medial Z-score projection maps. The upper two rows show mild frontal cortical and cingulate hypometabolism while the lower two show moderate to severe dorsolateral frontal, anterior temporal cortical and cingulate hypometabolism. The middle subject images show a parietal/precuneus pattern of hypometabolism. Subject D in this figure is the same person as Subject B in Fig. 3.

Demographic considerations

Among all subjects with dementia and/or ALS, and among only those who were examined by one of the co-authors, there was a slight male predominance (52–54%). Considering these three cohorts together, males with the primary diagnosis of behavioural variant FTD were over-represented (63%), and females with ALS ± FTD were slightly over-represented (18/30 or 60%), particularly in the ALS phenotype (61%), but whether these slight male–female differences reflect a sampling bias or have biological relevance in terms of sex factors impacting the topography of degeneration and hence the clinical phenotype will require further study.

Genetic/inheritance considerations

Among the three cohorts of subjects, the frequency of the mutation was highest in the Mayo Clinic Rochester FTD cohort (11.2% of all FTD ± ALS cases) and lowest in the Mayo Clinic Florida FTD cohort (3.6%), which likely reflects the strong focus of recruiting familial FTD kindreds in the Mayo Clinic Rochester centre and likely higher frequency of northern European ancestry in the Midwest (which is where most Mayo Clinic Rochester participants reside). Among those with a positive family history, the frequency of the expansion was highest in the Mayo Clinic Florida ALS cohort (at least 23%), but also reasonably high in the Mayo Clinic Rochester FTD cohort (19.1%). There are likely two factors that explain this. The newly identified C9ORF72 mutation explains a high proportion of familial ALS based on the available data (DeJesus-Hernandez et al., 2011; Renton et al., 2011). Furthermore, other data suggest that a common founder from Scandinavia might underlie the disease (Mok et al., 2012). More work is needed to determine if the mutation stems from a common founder, represents a ‘weak point’ in the genome which allows an increased frequency of expansions to randomly occur, or a combination of these and other factors.

As shown in Table 2, the frequency of the C9ORF72 mutation is highest in the FTD/ALS phenotype compared with the behavioural variant FTD and ALS phenotypes regardless of whether they are considered among all cases or familial cases only. Furthermore, no mutations in MAPT or PGRN were detected in the cases with sporadic or familial FTD/ALS and ALS. Also, at least in this Mayo Clinic series, the frequency of familial behavioural variant FTD being explained by the C9ORF72 mutation is higher than that explained by MAPT and PGRN despite the strong interest in recruiting familial FTD families over the past 20 years and contributions to the original identification of mutations in MAPT and PGRN causing familial FTD. These data therefore indicate that a sizable proportion of cases with familial FTD and/or ALS is explained by the C9ORF72 mutation. Importantly, however, an even greater proportion is not explained by mutations in these three or other known genes (>50% for each phenotype), indicating that other yet-to-be-identified genes are associated with FTD and/or ALS.

More than 80% of our probands were familial, of which more than half had both the behavioural variant FTD and ALS phenotypes represented in the kindreds, a third had only dementia or behavioural variant FTD, and the remaining minority had only ALS represented in the families. Therefore, the repeat expansion should be considered in any kindred with familial FTD and/or ALS. The other 16% of our probands did not meet predefined criteria for a positive family history and hence were considered sporadic.

The observation that the GGGGCC hexanucleotide repeat expansion in C9ORF72 may be associated with FTD, FTD/ALS and ALS in the same pedigree supports recommendations included in recently proposed criteria for the diagnosis of familial ALS (Byrne et al., 2011). In the Mayo Clinic Florida ALS cohort we classified eight patients who carried the C9ORF72 repeat mutation and had no known family history of ALS in a first- or second-degree relative, as sporadic ALS. However, available information suggests that five of these eight patients might be reasonably classified as having a familial disorder given a family history of dementia in a first-degree relative in four and in a second-degree relative in one patient. The latter classification is in keeping with criteria we applied in designating patients with FTD and FTD/ALS as familial or sporadic. This leaves a small number of patients with FTD or ALS in whom family history suggests sporadic disease, but evaluation of parental DNA will be important in future investigation to determine whether these pedigrees represent non-penetrance or spontaneous expansion of the GGGGCC hexanucleotide repeat from a non-pathogenic parental form. Adoption of stringent criteria for the diagnosis of familial ALS along the lines of those recently proposed by Byrne et al. (2011) seems especially important in giving increasing recognition that expanded phenotypes such as extrapyramidal disease may be associated with genes linked to FTD and ALS.

Whether the sporadic cases represent incomplete penetrance or de novo mutations is being investigated. Two subjects appeared to be obligate carriers but died while neurologically asymptomatic—one at age 34 from an accident, and another at age 72 from a myocardial infarction. The younger subject might have exhibited features had he survived decades more, and the 72-year old may have done likewise if he had survived a few more years. However, considering that 72 was the maximal age of onset of any clinically examined subject in this series, and the subject with the Alzheimer's disease phenotype whose onset was at the age of 74 and autopsy revealed FTLD-TDP-43 pathology (the repeat expansion has not been confirmed in this case but is presumed), this asymptomatic case who died at age 72 suggests that incomplete penetrance might occur even among aged individuals.

A curious observation is the apparent earlier age of onset (>10 years) in the youngest generation compared with the previous generation in 11 (26% of all probands) of our families with c9FTD/ALS (Supplementary Fig. 1). Considering that the mechanism of this genetic defect involves a hexanucleotide repeat expansion, and the concept of anticipation (i.e. an earlier age of onset in successive generations occurs with increasing repeat length) is well established in many of the nucleotide repeat disorders (Lindblad and Schalling, 1999), it is tempting to hypothesize that a similar mechanism may be at play in these families. However, this will be challenging to study. First, the repeat expansion is unstable and variable across cells, and variability in single subjects has already been seen (DeJesus-Hernandez et al., 2011). Furthermore, the current methodology for quantifying repeat length is imperfect, particularly when several hundreds of repeats are present. Investigating this issue further is clearly worthy to support or refute an anticipation mechanism associated with this hexanucleotide repeat. Repeat length may also impact the topography of neurodegeneration and hence the clinical phenotype expressed.

Clinical considerations

As noted above, the core phenotypes associated with this repeat expansion were behavioural variant FTD, ALS or combined FTD/ALS, with one-third of total cases and nearly half with behavioural variant FTD or FTD/ALS also having some degree of parkinsonism, which was typically an akinetic-rigid syndrome without rest tremor. When parkinsonism was present, it tended to evolve over the initial 2 years of the disease course. When both behavioural variant FTD and ALS occurred in the same individual, the features almost always evolved within 2 years. Even among subjects with the primary behavioural variant FTD syndrome, almost 40% in this series had some degree of upper and/or lower motor neuron dysfunction but not of sufficient breadth to warrant the FTD/ALS diagnosis. A comprehensive neurological examination in any patient with dementia will permit identification of these clinical findings. Almost 20% of the subjects with a primary diagnosis of ALS had some degree of behavioural variant FTD features but either did not fully meet criteria for the FTD/ALS diagnosis or could not be formally evaluated owing to motor impairment. Features of behavioural variant FTD may be subtle and not characteristic of fully developed dementia in ALS resulting in potential misattribution of signs of cognitive dysfunction to situational factors or depression. Weakness and fatigue may limit the practical assessment of FTD in subjects with ALS, and eliciting the historical and clinical features of behavioural variant FTD may not be routinely performed in clinics focused on standard elements of ALS patient care. Thus, the 20% frequency of behavioural variant FTD features in this cohort of subjects with ALS may be an underestimate. Yet the overlapping features of behavioural variant FTD and ALS in many subjects with the repeat expansion in C9ORF72 underscores several key points: (i) this genetic alteration can impact cognition, behaviour and motor functioning via degeneration in cerebral and/or spinal cord neurons; (ii) clinicians will often identify ALS features in subjects with behavioural variant FTD and behavioural variant FTD features in subjects with ALS if detailed questioning, clinical examinations and electromyograms are performed (Lomen-Hoerth et al., 2002, 2004); and (iii) assessments over time permit the identification of more widespread findings as the disease process evolves.

Neuropsychiatric morbidity was frequent in the subjects in whom this information was collected. The caregivers of subjects endorsed the presence of changes in appetite and eating behaviour in all who were queried, and almost half endorsed the presence of delusions or hallucinations. These findings are consistent with other recent publications on the neuropsychiatric features in FTD/ALS and/or those with TDP-43 positive pathology (Lillo et al., 2010; Burrell et al., 2011; Piguet et al., 2011).

Among the clinically examined subjects, most with ALS features fulfilled El Escorial criteria for definite, probable or possible ALS, and the site of onset was approximately equal across the bulbar, upper and lower spinal cord segments. The rare variants included atypical primary lateral sclerosis (no DNA available), monomelic ALS (mutation identified), and progressive muscular atrophy (mutation identified). Although most had ALS-like findings, neuropathological studies also showed patients with motor neuron degeneration consistent with progressive muscular atrophy. Hence, this genetic alteration should be considered in any ALS variant, particularly with a positive family history of FTD and/or ALS.

Importantly, none of our cases with FTD/ALS with early and profound aphasia, nor any of our non-fluent/agrammatic and semantic subtypes of primary progressive aphasia, were found to have a hexanucleotide repeat expansion. This observation not only has diagnostic relevance, but also suggests the repeat expansion on C9ORF72 affects cerebral networks in a more bilateral, patchy or diffuse manner as opposed to a focal/asymmetric manner which is what the primary progressive aphasia syndromes reflect. If this repeat expansion is also found to be rare or non-existent in other focal/asymmetric cortical degeneration syndromes such as the corticobasal syndrome, associative agnosia/prosopagnosia, posterior cortical atrophy, etc., this concept of clinical features reflecting more symmetric, patchy or relatively diffuse cerebral neurodegeneration will need to be explained based on the mechanism of how the repeat expansion causes cerebral dysfunction.

Neuropsychological considerations

Consistent with the phenotypic presentation of FTD/ALS, the predominant cognitive profile in subjects with the C9ORF72 hexanucleotide repeat expansion is associated with slowed processing speed, complex attention/executive dysfunction and impairment in rapid word retrieval. This pattern of dysfunction implicates frontal lobe structures and subcortical pathways involving dorsolateral prefrontal cortex and anterior cingulate. In contrast, naming, episodic memory, word reading and gross visuoperceptual/spatial skills are relatively spared, suggesting infrequent or minimal involvement of temporal and parietal lobe structures.

Approximately one-third of our subjects did not demonstrate impairment on cognitive screening measures. This finding may reflect the known poor sensitivity of screening measures to executive dysfunction or the presence of cases with behavioural variant FTD that have yet to develop cognitive deficits. It also bears noting that neuropsychological tests in common practice may not capture the entire spectrum of cognitive impairment possible in behavioural variant FTD or FTD/ALS (Torralva et al., 2009). Thus, the need for broad-based and more sensitive measures of cognitive assessment will become paramount when evaluating subjects in the presymptomatic and early symptomatic stages of the disease.

Neuroimaging considerations

The syndrome of behavioural variant FTD is typically associated with mild to marked frontal and/or temporal lobe abnormalities, particularly in the non-dominant cerebral hemisphere (usually right), on structural and functional neuroimaging studies (Miller et al., 1991; Rosen et al., 2002; Foster, 2003; Whitwell et al., 2004, 2005, 2009; Whitwell and Jack, 2007; Rohrer, 2011). On MRI, the degree of cerebral cortical atrophy in patients with C9ORF72 mutation tended to be rather mild, and was most apparent in the dorsolateral prefrontal and insular cortex. The temporal lobes were normal to only mildly affected. Some degree of parietal cortical atrophy was evident in some cases. And, the pattern of atrophy was never in a strikingly focal or asymmetric manner in any of our cases. Signal changes on T₂-weighted and fluid attenuation inversion recovery images in the cortex or subcortical white matter were also minimal or non-existent.

The findings on SPECT and PET were similar to those on MRI, with the dorsolateral prefrontal cortex, and anterior ± medial cingulate cortex involved. This topography of atrophy, hypoperfusion and hypometabolism correlates well with other features such as the prominent apathy and inertia as part of the behavioural variant FTD phenotype. The neuroimaging findings are also correlated with poor performance on neuropsychological measures that assess psychomotor speed, word fluency and sustained attention. Similar to the MRI findings in some subjects, the degree of hypoperfusion on SPECT or hypometabolism on PET was quite mild in some despite the presence of an obvious behavioural variant FTD syndrome. Also, a few cases had atrophy on MRI or hypometabolism on PET that was apparent in the biparietal/precuneus regions.

Therefore, abnormalities on neuroimaging studies that maximally involve the dorsolateral prefrontal cortex and anteromedial cingulate cortex were the most consistent findings. Parietal and/or temporal cortex abnormalities were present in some. Findings were mild in many, particularly early in the course, and relatively symmetric cerebral cortical involvement was the rule.

Neuropathological considerations

All cases had TDP-43 pathology, and more than half also had evidence of motor neuron degeneration and substantia nigra degeneration. Overt clinical evidence of MND or parkinsonism was not always associated with motor neuron or substantia nigra degeneration, respectively. Cortical neuronal loss, gliosis and spongiosis was most prominent in frontal lobe, with less in parietal and temporal lobes. Hippocampal sclerosis, which is common in FTLD-TDP (Josephs and Dickson, 2007) was detected in less than half the cases with c9FTD/ALS, possibly owing to the greater frequency of FTLD-MND in this series, where it has been previously shown that hippocampal sclerosis is less frequent (Josephs and Dickson, 2007). In an unselected autopsy series of c9FTD/ALS, MND was less common, while hippocampal sclerosis was more common than in this series (Murray et al., 2011). The paucity of prominent amnestic syndromes in the present series may be accounted for partly by the low frequency of hippocampal sclerosis.

The TDP-43 pathology was variable, with the majority being type A. Type A (Mackenzie type 1) is characterized by many cortical neuronal cytoplasmic inclusions and dystrophic neurites often with intranuclear inclusions and widespread involvement of subcortical areas, while type B (Mackenzie type 3) have many neuronal cytoplasmic inclusions, but few dystrophic neurites and more limited subcortical pathology (Josephs et al., 2009). These findings contrast to what was expected from previously suggested pathological correlates of c9FTD/ALS (Boxer et al., 2011; Mackenzie et al., 2011), which was suggested to be associated with type B (Mackenzie type 3) TDP-43 pathology. The explanation for the heterogeneity in TDP-43 pathology remains to be determined. It is of note that none of the cases had type C (Mackenzie type 2) pathology, which is associated with prominent cortical neuritic pathology with few cortical neuronal inclusions and clinically by the semantic subtype of primary progressive aphasia, was also not a clinical syndrome found in any patient in this series. The present findings contrast with detection of at least a few cases with type C in an unselected autopsy series of cases with the C9ORF72 mutation (Murray et al., 2011).

It has been suggested that ubiquitin-positive (P62-positive) inclusions in cerebellar granule neurons are characteristic of c9FTD/ALS (Al-Sarraj et al., 2011; Boxer et al., 2011; Murray et al., 2011), and we were able to confirm this except for three cases in which such inclusions could not be detected. It has also been suggested that in the hippocampus, cases with c9FTD/ALS have neuronal inclusions that are disproportionately positive for ubiquitin compared with TDP-43 (Al-Sarraj et al., 2011). We were able to confirm this finding. These are promising pathological features that might one day prove to lead to a specific and sensitive biomarker for c9FTD/ALS when their protein composition is eventually determined.

Comparison of ante-mortem and post-mortem features and findings in subjects with mutations in C9ORF72, MAPT and PGRN

Comparing and contrasting the clinical, neuropsychological, neuroimaging and neuropathological data between C9ORF72, MAPT and PGRN features will provide insights for clinicians to better target genetic testing (particularly for those with the behavioural variant FTD phenotype), and also enlighten our understanding of neurodegeneration associated with different mutations. A summary of the known features associated with these three mutations is shown in Table 6. Future analyses will likely include comparisons such as those shown in Supplementary Figs 5 and 6.

We thank Drs Ian Mackenzie, Adam Boxer and Bruce Miller for collaborating on the VSM-20 kindred, and many colleagues for referring subjects to our centre. We also extend our appreciation to the late Dr Emre Kokmen who personally examined many of the affected relatives in Kindred 4, Drs Glenn E. Smith PhD and Robert J. Ivnik PhD for their expertise on neuropsychological aspects of these subjects, and to the staff of the ADRC, ADPR and ALS Centre for participating in the evaluation and care of the patients and families. The assistance of Pamela Desaro, Amelia Johnston and Thomas Kryston is acknowledged for the collection of ALS DNA samples and clinical data in the MCF ALS Centre. We particularly thank the patients and their families for participating in ageing and neurodegenerative disease research.