Phenotype variability in progranulin mutation carriers: a clinical, neuropsychological, imaging and genetic study

Abstract

Introduction

Frontotemporal dementia (FTD) is characterized by behavioural and language disorders, and is clinically, pathologically and genetically heterogeneous. Three variants are distinguished by the presenting symptoms and regional pattern of atrophy: frontal variant (fvFTD) with predominant behavioural disorders; semantic dementia (SD) and progressive non-fluent aphasia (PNFA), in which language disorders predominate (Neary et al., 1998; Kertesz et al., 1999, 2005). There are four pathological categories of frontotemporal lobar degenerations (FTLD) with either: hyperphosphorylated tau-positive aggregates (tauopathies); ubiquitin-positive, TDP-43-positive, tau- and synuclein-negative inclusions (FTLD-U); neuronal intermediate filament inclusion disease (NIFID); or no detectable inclusion (Kertesz et al., 2005; Cairns et al., 2007).

Mutations in MAPT (microtubule-associated protein tau) were initially identified in families with FTD-Parkinsonism linked to the chromosome 17 (FTDP-17), associated with tau-positive inclusions (Hutton et al., 1998; Poorkaj et al., 1998; Spillantini et al., 1998). Recently, mutations in the progranulin gene, 1.7 Mb centromeric to MAPT, were found in FTD families linked to the chromosome 17 with ubiquitin-positive, tau-negative inclusions (FTDU-17) (Baker et al., 2006; Cruts et al., 2006). The symbol for the progranulin gene is GRN (HUGO nomenclature) but PGRN is also used. GRN codes for progranulin, the precursor of granulin proteins, a 593 amino acid glycoprotein containing 7.5 cysteine-rich tandem repeats. Progranulin is a secreted growth factor expressed in a variety of tissues that has been implicated in development, wound repair, inflammation and tumourigenesis (Ahmed et al., 2007). It is highly expressed in neurons of the cerebral cortex, the hippocampus and in the cerebellum (Daniel et al., 2000), but its function in the central nervous system is unknown. Ubiquitin-positive inclusions in FTLD-U patients with GRN mutations do not contain progranulin (Baker et al., 2006), consistent with the fact that most GRN mutations are nonsense, frameshift or splice-site mutations producing mutant mRNAs with premature translation termination codons. Mutant mRNA with these codons is degraded by nonsense-mediated decay that prevents expression of truncated proteins (null mutation). As a result, no mutant protein is produced (Baker et al., 2006; Cruts et al., 2006). Neurodegeneration therefore results from a partial loss of progranulin function rather than the aggregation of the mutant protein. How the mutant protein causes ubiquitin-positive inclusions is unknown. However, TDP-43 (TAR-DNA binding protein) was recently identified as a major component of inclusions in FTLD-U with or without GRN mutations and in isolated motoneuron disease (MND) (Neumann et al., 2006).

Patients with GRN mutations have variable ages at onset and phenotypes. FvFTD is most frequent, but primary progressive aphasia (PPA) (Snowden et al., 2006; Mesulam et al., 2007), hereditary dysphasic disinhibition dementia (Mukherjee et al., 2006; Behrens et al., 2007) or corticobasal degeneration syndrome (CBDS) (Benussi et al., 2006; Masellis et al., 2006) have been reported in a few families. The frequency of GRN mutations has not been evaluated, however, in series of patients with PPA and CBDS. A few studies on large numbers of patients have focused on genetic (Gass et al., 2006) or neuropathological aspects of the disease (Davion et al., 2007; Josephs et al., 2007; Mackenzie, 2007). A recent study reported clinical characteristics in a series of 37 patients analysed for a single mutation p.Arg493X (Rademakers et al., 2007). However, the clinical data were collected retrospectively in this study. Neuropsychological characteristics and brain perfusion profiles associated with GRN mutation have not yet been described in large series of mutation carriers. We have analysed GRN in 502 patients including 352 patients with fvFTD, 52 with FTD and MND (FTD-MND), 68 with PPA and 30 with CBDS. We found 18 null mutations in 24 families and report clinical, neuropsychological and brain imaging features of 32 mutation carriers. Some aspects of this study, in particular the neuropsychological characteristics, as well as the brain perfusion characteristics in mutation carriers, have never been described in detail and constitute major new findings.

Methods

Recruitment of patients

Five hundred and two unrelated probands were recruited between 1997 and 2007 through a national network of 15 specialized dementia centres, 350 of which were evaluated prospectively, that is to say after inclusion in the research cohort, with standardized clinical, behavioural, neuropsychological procedures and were included in a follow-up study whenever possible (Le Ber et al., 2006). The clinical evaluation included testing for: parkinsonism, other movement disorders, oculomotor abnormalities, MND, hallucinations, apraxia. The behavioural evaluation was based on a 70-item scale derived from the Frontal Behavioural scale (Lebert et al., 1998), the Frontal Behavioural Inventory (Kertesz et al., 2000) and the Neuropsychiatric Inventory (Cummings et al., 1994), integrating apathy, disinhibition, hyperorality, stereotyped/ritualistic behaviours, emotional/affects and language. Evaluations were collected for each patient, and each symptom was evaluated during follow-up as stable, increased or decreased.

Among the 502 probands, 352 probands had fvFTD according to international clinical criteria (Neary et al., 1998; The Lund and Mancheter groups, 1994), 52 FTD-MND, 68 had PPA (Neary et al., 1998; Mesulam, 2001) and 30 had clinically diagnosed CBDS (Boeve et al., 2003) (Table 1).

Information concerning the family was collected for each proband. Inheritance was considered autosomal dominant when at least one other patient from a different generation than the proband had a history of dementia, PPA or CBDS. The families of the probands with at least one other affected member were extended through the research network. Relatives who had neurodegenerative or neuropsychiatric disorders were examined according to the network procedure whenever possible.

In this article, we use the term proband to designate the index case of each family. Relatives are other members of proband's family, and affected relatives are family members who had dementia, PPA or CBDS. Mutation carriers are all individuals carrying GRN mutations (probands or relative). Phenocopies are relatives that do not carry the mutation although their phenotype is similar to that of mutation carriers. CBDS refers to the patients with clinically diagnosed corticobasal degeneration syndrome but no pathological confirmation.

Mutation screening

DNA was collected with informed consent from the 502 probands. The study was approved by the Paris-Necker ethics committee. MAPT mutations were excluded in fvFTD and FTD-MND patients. GRN comprises 13 exons that were sequenced in the 502 probands and in 179 neurologically healthy control individuals of French origin. Thirteen relatives of nine probands with GRN mutations were also analysed: 11 with dementia, PPA or CBDS, one with schizophrenia and one obligate carrier who was asymptomatic when sampled at age 60.

All GRN exons, including the non-coding exon 0, and at least 50 bp of the flanking intronic regions were sequenced in both directions. Total genomic DNA was prepared from peripheral blood according to standard procedures. Exons and flanking regions were amplified by PCR on genomic DNA (20 ng) using previously described primers (Cruts et al., 2006). Amplification products were purified with 1 U Antarctic phosphatase (New England Biolabs, Ipswich, MA, USA) and 1 U exonuclease I (New England Biolabs) and sequenced in both directions using the BigDye Terminator Cycle Sequencing kit v3.1 (Applied Biosystems, Foster City, CA, USA) on an ABI3730 automated sequencer (Applied Biosystems). Sequences were analysed using Seqscape® software and the Software Package NovoSNP.

We have also looked for deletions of the GRN gene, which might, like most GRN mutations, be responsible for the partial loss of functional progranulin. Two hundred and ten patients with fvFTD or FTD-MND, in whom no mutations were detected by sequencing, were screened for GRN gene copy number variations by quantitative multiplex PCR of short fluorescent fragments (QMPSF), in which multiple short genomic sequences are amplified simultaneously with dye-labelled primers under quantitative conditions (Rovelet-Lecrux et al., 2006). Nine amplicons within or bordering nine exons of GRN were examined (details available upon request).

Phenotype of mutation carriers

Thirty-two affected patients, who carried GRN mutations, including 24 probands and eight affected relatives for whom sufficient clinical data was available, were included in the clinical study. All but three were examined by one of us in a memory disorders clinic. Twenty were assessed prospectively for neurological and behavioural signs using a standardized questionnaire evaluating apathy, disinhibition, emotion/affect, stereotyped/ritualistic behaviours, eating/oral behaviours, attention/executive and language disorders, and were included in a follow-up study by the research network (see the section Recruitment of patients). Clinical data were also taken from the medical records and reports by relatives. Ten index patients with mutations were previously described (Le Ber et al., 2007; van der Zee et al., 2007). Age at onset was when the first symptom appeared according to the principal informant and the patient.

Neuropathological examinations were performed in five mutation carriers. Ubiquitin, tau, β-amyloid and alpha-synuclein immunochemistry was performed in five patients and TDP-43 immunochemistry in four (see Supplementary data).

Neuropsychological evaluation

Twenty-two of the 32 mutation carriers were evaluated with a neuropsychological battery including evaluation of: (i) global efficiency with the mini-mental status examination (MMSE) (Folstein et al., 1975) and the Mattis dementia rating scale (Mattis-DRS) (Mattis, 1988); (ii) verbal episodic memory with the free and cued recall test (FRCT), a specific procedure allowing to distinguish encoding, storage and retrieval deficits (Van der Linden et al., 2004); (iii) executive functions with the frontal assessment battery (FAB) (Dubois et al., 2000) and two verbal fluency tasks (Thuillard and Assal, 1991); (iv) constructional disorders with the Rey–Osterrieth's complex figure (Rey, 1959); (v) search of motor apraxia (motor programmation and bi-manual coordination), ideomotor apraxia (transitive and intransitive gesture on verbal command and imitation), ideational apraxia (use of tools) and (vi) language with tests of verbal fluency and pictures naming (Deloche et al., 1996). Four patients were not able to undergo the full battery. Complementary tests were performed according to the symptoms.

Brain imaging

Brain MRIs were available for 21/32 mutation carriers (Supplementary Table 1), and CT scans for two. These exams were visually analysed.

Brain perfusion SPECT was performed after intravenous injection of 99mTc-ECD in 10/32 mutations carriers (six fvFTD, two CBDS, one PPA, one Alzheimer's disease) in four centres using the same ECD protocol. Perfusion profiles were compared to 28 healthy age-matched controls, and also to 31 fvFTD patients with no GRN and no MAPT mutations, matched for gender, age at onset, disease duration and behavioural profile at onset (disinhibition, apathy, mixed). Details on each group are available in Supplementary Table 2.

A voxel-by-voxel intergroup comparison was performed with SPM2. The centre effect was minimized by filtering and masking as previously described (Le Ber et al., 2006; Guedj et al., 2007). Significant hypoperfusion in the group of mutations carriers was looked for first in comparison to healthy subjects, then to non-carriers. SPM{T} maps of the mutation carriers were thus compared to those of healthy subjects at a height threshold of P = 0.05, corrected for multiple comparisons for the cluster. Then, areas with significant hypoperfusion in mutation carriers were compared to the same regions in non-carrier patients, using a height threshold corrected for multiple comparisons for the voxel (False Discovery Rate method, FDR). Interhemispheric asymmetry was then evaluated visually in the two groups of patients.

The same statistical threshold was used for additional voxel-based analyses, including, (i) subjects of the same centre (Pitié-Salpêtrière) with SPECT performed with the same gamma-camera (five GRN mutation carriers; seven fvFTD patients with no mutations matched for gender, age at onset, disease duration, behavioural profile, and 16 healthy age-matched controls); (ii) six GRN mutation carriers with fvFTD (excluding patients with CBDS, PPA, Alzheimer's disease) in comparison to 20 fvFTD patients with no mutations, matched for gender, age at onset, disease duration, behavioural profile and to 28 healthy age-matched controls.

Results

Mutations in GRN

We identified 18 GRN null mutations, seven of which were novel (Table 2), in 24 independently ascertained probands and in eight affected relatives (32 mutation carriers). In addition, we found mutations in two relatives of F716/001, one asymptomatic at age 60 and one with schizophrenia.

Nineteen probands had autosomal dominant inheritance. Five (F001/001, F057/007, F263/001, F583/001, F587/006) had no family history of the disease. Their parents were deceased, without neurological disorders, after age 65 (except for one). DNA was not available, so we could not distinguish between reduced penetrance in the transmitting parent and de novo mutations.

Ten small deletions/insertions causing frameshifts, three nonsense mutations and three splice site mutations (Table 2), all leading to premature termination codons, were predicted to result in haploinsufficiency (Baker et al., 2006). We also found a mutation in the initiation codon (c.1A>G, p.Met1?) that segregated with the disease in family F716, and a missense mutation (c.19T>C, p.Trp7Arg) in the signal peptide of the protein. None of these mutations were found in 179 French controls. Segregation with the disease was confirmed in eight families. The most frequent mutation, p.Glu498fsX11, was detected in three families, and four mutations were identified in two families each (Table 2). There were no GRN gene copy number variations.

No mutations were found in 3/11 relatives with dementia from three families (F124, F171, F741). They were also negative for MAPT mutations. Notably, a branch of family F171 had autosomal dominant FTD-MND (Supplementary Fig. 1), but no GRN mutations. No mutations were found either in a sister of F741/007 who had dementia with frontal symptoms and a sister of the proband in family F124 who had dementia at age 64.

Fig. 1

Brain MRI T2 (A–E) and T1 (F) sequences and CT scan (G, H) in GRN mutation carriers. (A–D) Moderate to marked white matter lesions (arrows). (E–H) Clear asymmetric and posterior atrophy.

Open in new tabDownload slide

Frequency of GRN mutations in probands with fvFTD, PPA, CBDS and FTD-MND

The frequency of mutations in fvFTD, PPA, CBDS and FTD-MND was calculated using the number of probands with identified mutations in each disease category (Table 1). The frequency of GRN mutations was 5.7% (20/352) in fvFTD (17.9% in those with autosomal dominant inheritance, 19/106), 4.4% (3/68) in PPA and 3.3% (1/30) in CBDS. No mutations were found in the 52 probands with FTD-MND.

Phenotype in mutation carriers

Clinical data were collected for 32 mutation carriers (24 probands, eight relatives). There were 13 males and 19 females (Table 3).

Disease onset

The mean age at onset was 59.1 ± 7.7 years (49–79). Age at onset was >65 years in 7/32 (22%), and was variable among and within families.

At onset, 24/32 (75%) had marked behavioural disorders, 7/32 (22%) had language disorders and 3/32 (9%) memory disorders (Table 3). The mean disease duration at the first consultation in a memory disorders clinic was 3.0 ± 2.1 years.

Description of the mutation carriers according to the phenotype

Patients with fvFTD (n = 22)

Twenty-two mutation carriers had fvFTD (20 probands, two relatives). Diagnosis of fvFTD was made at onset in 18 probands and two relatives (20/32, 63%). Two (F171/002, F263/001) had aphasia with behavioural disorders at onset. One proband (F524/001) with fvFTD initially had a psychiatric presentation, with visual and auditory hallucinations and severe paranoid delusions.

In two other probands (2/32, 6%) (F057/007, F050/001), Lewy body dementia (LBD) was initially suspected because of hallucinations and/or parkinsonism at onset. The diagnosis changed to fvFTD by the time of inclusion, and they were included as such.

Patients with PPA (n = 5/32, 16%)

Five mutation carriers including three probands and two relatives had PPA (Mesulam, 2001). They had severe language disorders at onset and no behavioural disorders during the first 2 years of the disease.

Patients with CBDS (n = 2/32, 6%)

Mutations were found in one proband and her sister who had CBDS according to clinical criteria (Boeve et al., 2006). They had behavioural symptoms at onset, but rapidly developed asymmetric parkinsonian rigidity, left upper limb apraxia and hemineglect; one had dystonia.

Patients with other diagnoses (n = 3/32, 9%)

Three affected relatives (F015/010, F129/002, F160/019) with proven GRN mutations had been diagnosed with Alzheimer's disease in a memory disorders clinic. They had predominant episodic memory disorders as the first symptom, and no or mild behavioural disorders at onset (Table 3). Their neuropsychological profiles (FRCT) were consistent with a hippocampal amnestic syndrome.

Members of family F716 presented with fvFTD (F716/001) or PPA (F716/002). One mutation carrier (F716/005) had schizophrenia but no dementia or language disorders. His cousin also had schizophrenia, but we could not confirm the presence of the mutation. Since the relation between schizophrenia and GRN mutations remains speculative, we did not include patient F716/005 or his cousin in the phenotypic study.

In summary, the clinical diagnosis at onset was fvFTD in 63% (20/32) of the mutation carriers, PPA in 16% (5/32), Alzheimer's disease in 9% (3/32), CBDS and LBD each in 6% (2/32) of the mutation carriers. Illustrative case reports are presented as Supplementary data.

There were no obvious phenotype-genotype correlations. In particular, the phenotype associated with mutation p.Trp7Arg was that of fvFTD. Families with the most frequent mutation (p.Glu498fsX11) presented either CBDS (F540), fvFTD (F741) or PPA (F023) (Table 2). The diagnoses were also variable within families (Table 3) reflecting the great intra-familial phenotype variability.

Main characteristics and progression of the disease

Twenty-eight of the 32 patients (88%) had behavioural disorders during the course of the disease (Table 3): all of the patients with fvFTD and CBDS, but also the three mutation carriers with PPA (F171/004, F258/001, F047/001) with the longest disease durations. Patients diagnosed with Alzheimer's disease may have had mild behavioural symptoms (F160/019), but they were less predominant than the memory disorders. Behavioural changes included apathy (22/32, 69%), disinhibition (10/32, 31%), hyperorality (15/32, 47%) and ritualistic/stereotyped behaviours (9/32, 28%). Twenty of the 32 patients (63%) had reduced spontaneous language, consistent with frontal dysfunction. Mutism developed after a mean disease duration of 4.1 ± 1.4 years.

Parkinsonism developed in 13/32 (41%) but was present at onset in only one patient (F540/002). Parkinsonism was observed in 10/22 patients with fvFTD (41%), 2/2 with CBDS and 1/5 with PPA (20%). Visual hallucinations (animals, people) were present in 8/32 patients (25%). None developed MND.

The mean disease duration when patients became bedridden was 4.9 ± 0.9 years (n = 9). Twenty patients died after mean disease durations of 6.5 ± 1.9 years. The diagnosis of FTLD-U was confirmed in five autopsied patients with fvFTD (F001/001, F160/001, F171/002, F331/001, F741/041). There were characteristic neuronal intranuclear and cytoplasmic ubiquitin- and TDP-43-positive inclusions. There were no abnormal tau, synuclein and β-amyloid deposits (Supplementary data).

Neuropsychological characteristics

The mean scores in 22 mutation carriers are shown in Table 4. The mean disease duration at evaluation was 2.4 ± 1.2 years (Tables 4 and 5). All but two patients presented mild to severe frontal cognitive dysfunction. Episodic memory was impaired in 16/18 patients (89%) evaluated with the FCRT. Most patients (11/18, 61%) had episodic memory disorders characterized by impaired encoding and retrieval that benefited significantly from semantic cueing. This profile is consistent with frontal lobe dysfunction (Pillon et al., 1994). Five (28%) however, had an amnesic syndrome with severely impaired storage consistent with hippocampal dysfunction, but executive functions were less severely or not impaired (Dubois and Albert, 2004).

Language was severely affected in a subset of patients, and five had PPA (Mesulam, 2001). One (F023/004) had progressive non-fluent aphasia with agrammatism (Mesulam, 2001) and one (F171/004) had non-fluent aphasia with impaired repetition, phonemic paraphasias and preserved comprehension, compatible with conduction aphasia (Mendez et al., 2003). Two patients (F258/001, F716/002) had fluent aphasia with impaired comprehension (verbal semantic deficit). Two others (F171/002, F263/001) had characteristics of non-fluent aphasia with agrammatism but also marked behavioural disorders and executive syndromes at onset.

Ten out of 21 patients (48%) had constructional disorders, 24% had ideomotor apraxia (5/21) and/or agraphic apraxia (1/21). Five (45%) of these patients had a biparietal syndrome (dressing apraxia, hemineglect). None had Balint syndrome.

Four distinct neuropsychological and language profiles were therefore discernible in mutation carriers (Table 5): (i) isolated behavioural and frontal cognitive syndrome (frontal type, no sign of parietal or amnestic syndrome); (ii) fluent or non-fluent primary progressive aphasia at onset (aphasic type), only one case having an amnestic syndrome; (iii) parietal dysfunction (ideomotor apraxia, constructional disorders, neglect) associated with a frontal syndrome (parietal/apraxic type) and (iv) hippocampal amnestic syndrome associated with a mild to moderate executive dysfunction (amnestic type). The mean disease duration at evaluation was similar in each group.

Brain imaging (Supplementary Table 1)

Brain MRI characteristics

MRI, performed after a mean disease duration of 2.8 ± 1.3 years showed frontal and/or temporal atrophy in all patients except one evaluated at onset (Fig. 1). Cortical atrophy was markedly asymmetric in 16/21 mutation carriers (76%), with right predominance in 10/21 (48%) and left predominance in 6/21 (29%). Parietal and/or occipital atrophy was present in 10 patients. Small patchy or more extensive white matter lesions were present on T2 sequence images in 4/21 patients (19%), in the absence of vascular risk factors.

Brain perfusion profile

SPECT, performed after a mean disease duration of 2.9 ± 1.3 years, showed significant hypoperfusion bilaterally in the fronto-cingular, the right posterior temporal cortex, including hippocampus, and the inferior parietal cortex, compared to healthy subjects (P < 0.05, corrected for the cluster) (Fig. 2, Supplementary Figs 2 and 3). More significant hypoperfusion compared to fvFTD patients without mutations was seen in the dorsolateral right frontal cortex, the right posterior temporal and inferior parietal cortices, the right hippocampus and the posterior cingulate cortex bilaterally (P < 0.05, corrected for the voxel using FDR). These functional abnormalities were also detected by visual inspection of individual SPECT examinations. There was no area in which patients with no mutations had greater hypoperfusion than GRN mutation carriers.

Fig. 2

Anatomical localization of peaks of hypoperfusion in: (A) GRN mutation carriers compared to healthy subjects, and (B) GRN mutation carriers compared to FTD patients without mutations (right hemisphere is on the right side of the slices). FTD patients with GRN mutations had significantly greater hypoperfusion in the right frontal dorsolateral cortex, right temporoparietal cortex, right hippocampus and, bilaterally, the posterior cingulate cortex, than patients without mutations.

Open in new tabDownload slide

Interhemispheric perfusion asymmetry differed significantly between patients with and without mutations (P = 0.0257, chi-squared test). Hypoperfusion in mutation carriers predominated on the right in six patients (60%), on the left in two (20%), and was symmetric in two (20%). In patients without mutations, hypoperfusion predominated on the right in eight patients (26%), on the left in 12 (39%), and was symmetric in 11 (35%).

An intergroup analysis between centres showed no significant differences at the height threshold corrected by the FDR method used in the present study (Le Ber et al., 2006). Our findings were also validated by including subjects from the same centre. Similar qualitative findings were obtained when (i) only patients investigated at the same centre or (ii) only GRN mutation carriers with fv-FTD phenotype were analysed (Supplementary Fig. 3).

Discussion

We have described in detail a large series of patients with different phenotypes (fvFTD, FTD-MND, PPA, CBDS) analysed for GRN. Previous studies on large series evaluated the genetic contribution of GRN mutations in FTD (Gass et al., 2006), or the neuropathological aspects of this disease (Mackenzie, 2007). Two series, based on pathological examination of FTLD-U cases, included basic clinical data collected retrospectively in smaller numbers of patients (Josephs et al., 2007; Davion et al., 2007). In a study that screened only for the GRN p.Arg493X mutation in patients with FTLD from North-America, 30 families had this mutation (20% of the cohort) due to a founder effect in 27 of the 30 families (Rademakers et al., 2007). The clinical data in the study of Rademakers et al. were obtained retrospectively for 34 patients. This mutation was not found, however, in other Belgian, Dutch or Italian populations, neither in our series of French patients so far. Our study specifically focused on the phenotypic characteristics associated with progranulin mutations, including detailed neuropsychological and brain perfusion profiles not previously reported in a large group of progranulin patients that were not selected according to the mutation. Most patients were evaluated prospectively with standardized procedures, and only patients with proven mutation were included.

Eighteen pathogenic mutations, seven of which were novel, were found in 24 families ascertained independently. The most frequent type of mutations were small deletions/insertions. Nonsense, frameshift and splice site mutations leading to premature termination codons were predicted to result in haploinsufficiency through nonsense-mediated decay (Baker et al., 2006). A missense mutation in the initiation codon (c.1A>G) segregated with the disease in one family. Two other mutations were previously identified in the initiation codon (Baker et al., 2006; Cruts et al., 2006), the first translated codon which has a crucial role in the initiation of protein translation. These mutations probably affect protein translation and cause functional progranulin haploinsufficiency. However, the level of mutant mRNA in a brain sample from a patient with the c.2T>C mutation was reduced, suggesting that the mutant mRNA may be degraded (Baker et al., 2006). We found another missense mutation in the progranulin signal peptide sequence (p.Trp7Arg) that translocates proteins to be secreted into the lumen of the endoplamic reticulum. Mutations in the peptide signal presumably lead to functional haploinsufficiency by a mislocalization of the protein or inefficient trafficking through the secretory pathway (Gass et al., 2006; Mukherjee et al., 2006). Another mutation in the signal peptide (p.Ala9Asp) was previously reported in four families (Gass et al., 2006; Mukherjee et al., 2006; Spina et al., 2007a), including the large HDDD2 family in which the mutation segregated with the disease and a family with CBDS (Spina et al., 2007a). Since the level of mutant mRNA was strongly reduced in the frontal cortex of patients carrying the p.Ala9Asp mutation, it is also possible that mutations in the signal peptide induce a process similar to nonsense-mediated decay (Gass et al., 2006).

Phenocopies were present in 3/24 families in our series. Autosomal dominant FTD-MND was present in a branch of family F171 with no GRN mutations, indicating that another gene is responsible for the disease in this branch. A similar situation was found in a large Calabrian family (Bruni et al., 2007), a branch of which had FTD but no mutations. Caution should be exercised, therefore, when counselling affected members of progranulin families who do not have proven GRN mutations. The occurrence of two rare forms of autosomal dominant FTD in the same family is intriguing. Further studies on extended families are needed to determine whether GRN polymorphisms on the normal allele might enhance the risk of developing other types of FTD.

Sixty-three per cent of mutation carriers had initial clinical diagnoses of fvFTD, whereas a high proportion (37%) had another clinical diagnosis, comprising PPA, Alzheimer's disease, CBDS or LBD. These diagnoses were made in memory centres based on clinical and neuropsychological criteria. There were no correlations between the phenotype and the genotype, as expected, since all the mutations lead to progranulin haploinsufficiency. Furthermore, no obvious modulatory effects of the MAPT haplotype or ApoE genotype on the clinical presentation have been observed (Gass et al., 2006; Bruni et al., 2007). In contrast, Rademakers and collaborators (2007) found an association between ApoEε4 allele with early memory disorders that were present in 30% out 37 patients with the p.Arg493X mutation. However, this study was based on a limited number of patients for which the clinical data were collected retrospectively, which possibly constitutes a bias. The effect of modifying factors on the clinical presentation of the disease should be validated in larger groups of patients in which the clinical symptoms are evaluated prospectively.

The relative frequency of GRN mutations was 5.7% in French patients with fvFTD, up to 17.9% in familial forms. Higher frequencies, reaching 25% of familial forms, were found in North-American and Belgian studies (Baker et al., 2006; Cruts et al., 2006). They were probably overestimated, however, partly due to a founder effect or selection of FTLD-U or 17q-linked families. Our study gives a more accurate evaluation of the frequency in a large unselected population of patients with fvFTD, with separate evaluation of FTD-MND. The frequency of GRN mutations in familial forms in our study is close to that evaluated by Huey and his collaborators (2006), but higher than in two smaller Dutch (4%) and Italian populations (1%) (Bruni et al., 2007; Bronner et al., 2007). These discrepancies may reflect geographical differences or may be due to different criteria for familial disease and variable proportions of patients with FTD-MND, not caused by GRN mutations, included in the different studies.

We evaluated separately the frequency of GRN mutations in patients with FTD-MND. None of the 52 patients had mutations, and none of the mutation carriers developed MND during the course of the disease. Two sequence variants (p.Arg433Trp, p.Ser120Tyr) resulting in an amino-acid substitution but not in a progranulin haploinsufficiency, have been reported in patients with FTD-MND (Schymick et al., 2007; Spina et al., 2007b), but their pathogenicity has not been established. They were not present in our patients with FTD-MND, but p.Arg433Trp was found in one our healthy controls, indicating that it is more probably a benign variant. Our results are in accordance with smaller studies of patients with FTD-MND (Gass et al., 2006; Schymick et al., 2007), and demonstrates that FTDU-17 and FTD-MND are distinct genetic diseases, although both are characterized by ubiquitin-positive inclusions. It is also concordant with the absence of characteristic lesions of MND (Bunina bodies, skein-like lesion) in GRN mutation carriers (Mukherjee et al., 2006; Mackenzie, 2007).

This is the first study evaluating the frequency of GRN mutations in patients with PPA and CBDS. GRN mutations were less frequent in these diseases, 4.4% and 3.3%, respectively, than in fvFTD. Although PPA and CBDS are not usually familial diseases, all these mutation carriers had at least one relative with dementia, PPA or CBDS. In contrast, there were no mutations in patients without family histories of neurodegenerative diseases. It is not surprising that we identified GRN mutations in our patients with PPA. This is consistent with the fact that clinically defined PPA has been shown to be pathologically heterogeneous. The most common characteristics are tau-positive or Alzheimer's disease pathology (Kertesz et al., 2005; Knibb et al., 2006; Josephs et al., 2006). However, a proportion of PPA patients have ubiquitin-positive inclusions (Kertesz et al., 2005; Josephs et al., 2006), as observed in progranulin patients (Cairns et al., 2007; Mackenzie, 2007). Our study shows that some of these patients indeed have mutations in GRN.

This study provides a precise description of the neurological signs associated with GRN mutations and evaluation of their frequency based on the prospective collection of data. Parkinsonian syndromes were frequent in our patients (41%), concordant with the observation of striatal lesions in GRN mutation carriers (Josephs et al., 2007; Mackenzie, 2007). More interesting is the presence of visual hallucinations (25%), which were much more frequent in this series than in fvFTD patients with no mutations (2%) (Le Ber et al., 2006), and appear to have been the first symptom in one of our patients. Hallucinations have previously been reported in a few families with GRN mutations (Benussi et al., 2006; Boeve et al., 2006; Leverenz et al., 2007), but mostly in reports of single families. Hallucinations were not evaluated in two large series of patients (Gass et al., 2006; Mackenzie, 2007). They appear to be much less frequent (8%) in another large series, but their frequency could be underestimated due to retrospective collection of clinical data (Rademakers et al., 2007). The presence of visual hallucinations in 25% of our patients shows that is an important symptom that may help differentiate patients with GRN mutations from other FTD patients. Their presence may sometimes lead to an erroneous diagnosis of Lewy body dementia, in particular when parkinsonian symptoms are associated. However, cognitive symptoms did not fluctuate in GRN mutations carriers as they do in patients with LBD. LBD-pathology was associated to FTLD-U in few GRN cases (Knibb et al., 2006; Josephs et al., 2007; Spina et al., 2007b). One of our five autopsied patients had hallucinations and FTLD-U, but no synuclein pathology. It is not excluded, however, that a subset of our patients with an LBD-like phenotype or hallucinations may prove to have LBD-pathology on autopsy.

A major aim of this study was to provide a description of neuropsychological characteristics associated to GRN mutations in carriers investigated with standardized procedures. Four different neuropsychological/language profiles were distinguished according to the predominance of frontal cognitive syndrome or language disorders at onset, and the presence of apraxic or amnestic syndromes. Importantly, these profiles do not reflect different stages of disease progression since the disease durations were similar. More probably, they reflect different patterns of neurodegeneration, which are concordant with the localization of brain atrophy, which was asymmetric and with more severe parietal involvement in some patients.

Seven mutation carriers had aphasic disorders in our study, five with a diagnosis of PPA. A few patients with PPA due to different GRN mutations were previously reported (Snowden et al., 2006; Mesulam et al., 2007). Non-fluent aphasia was a prominent feature in many patients of a Belgian founder family (Cruts et al., 2006). In contrast, a recent study found that language deficits in GRN mutation carriers were less frequent than FTLD-U patients with no mutations, two out of nine carriers having non-fluent or mixed aphasia (Van Deerlin et al., 2007). Notably, the characteristics of aphasia were not homogeneous in our patients. Aphasia was mostly non-fluent, consistent with PPA with agrammatism (Mesulam, 2001), or with conduction aphasia (Mendez et al., 2003); but fluent aphasia with a verbal semantic deficit, resembling SD (Neary et al., 1998), but without agnosia, was also observed in our patients.

The high frequency of early ideomotor apraxia and constructional disorders, consistent with parietal dysfunction, is of note. Praxis is usually preserved in early stage of FTD (Neary et al., 1998). Apraxia early in the course of the disease may therefore distinguish GRN mutation carriers from other FTD patients. The diagnosis of CBDS was made in two of our mutation carriers, because of severity of asymmetric apraxia associated with left parkinsonian signs, left hemi-neglect and frontal dysfunction. Their clinical characteristics were remarkably similar to the proganulin patients described by Masellis et al. (2006) although their mutations were different. The frequency of apraxia in our study suggests that posterior cortical regions are affected early in GRN mutation carriers. Accordingly, more severe parietal degeneration was found in GRN carriers compared to FTLD-U patients without mutations (Josephs et al., 2007). Remarkably, there was no Balint syndrome in our patients, suggesting that the dorsal parietal network was less involved (Tang-Wai et al., 2004).

The unexpected frequency of episodic memory disorders (89%) determined on the basis of neuropsychological evaluations was also a major observation and may represent a distinctive characteristic of GRN mutation carriers. Most of the patients had encoding and retrieval deficits but a normal storage process. This profile was consistent with frontal executive dysfunction (Pillon et al., 1994). Five patients, however, had hippocampal amnestic syndrome. Three of these patients had clinical diagnoses of Alzheimer's disease made in memory centres based on clinical and neuropsychological criteria. Coexistence of FTLD-U and Alzheimer's disease pathology has occasionally been described in GRN carriers (Snowden et al., 2006; Josephs et al., 2007; Rademakers et al., 2007; Spina et al., 2007b). Neuropathological data was not available for our three patients, two of whom are still alive. Therefore, we cannot formally exclude that they had the pathological lesions of Alzheimer's disease. However, progranulin is highly expressed in the hippocampus (Daniel et al., 2000), in which marked atrophy and neuronal loss may be observed in progranulin patients (Boeve et al., 2006; Mukherjee et al., 2006; Snowden et al., 2006; Mackenzie, 2007). Neuronal degeneration was severe enough in some cases that hippocampal sclerosis was diagnosed (Boeve et al., 2006; Mackenzie, 2007; Rademakers et al., 2007). Accordingly, we can hypothesize that the amnestic syndrome in mutation carriers is related to hippocampal neurodegeration caused by progranulin-related disease.

A particular pattern of hypoperfusion involving the hippocampus, posterior cingulate and parietal lobe, as well as frontotemporal cortices characterizes GRN mutation carriers in our study. However, the small number of patients is a limitation of our study and these results should be confirmed in larger series of patients. The inferior right parietal hypoperfusion was more severe in mutation carriers than in patients without mutations. This may be related to the high frequency of apraxia in our study. A similar pattern of regional atrophy was found using voxel-based morphometry in GRN patients (Whitwell et al., 2007). In this study, grey matter loss in the parietal lobe was a consistent finding and differentiated mutation carriers from non-carriers.

Hypoperfusion in the hippocampus was an unexpected finding, as it is not frequently observed in patients with Alzheimer's disease, a disease with severe hippocampal involvement. It was clearly present on the individual SPECT images of our progranulin patients (Supplementary Fig. 2) and statistically demonstrated by voxel-based analysis in comparison to healthy subjects and non-carrier patients. It may be related to the amnestic syndrome and hippocampal degeneration found in GRN mutation carriers. Perfusion studies in a larger series of patients will be necessary, however, to correlate brain perfusion and performances on neuropsychological tests.

Predominant posterior cingulate hypoperfusion was also an unexpected finding since this is known to be preferentially associated with Alzheimer's disease (Foster et al., 2007). Posterior cingulate hypoperfusion in Alzheimer's disease is presumed to result from deafferentation of the medial temporal cortex (Hayashi et al., 1999; Meguro et al., 1999; Millien et al., 2002). We hypothesize similarly that the hippocampal impairment in our patients with GRN mutations would explain the hypoperfusion in the cingulate cortex.

Our study illustrates the highly variable phenotypes and neuropsychological profiles associated with GRN mutations and provide further evidence that the phenotype is a poor predictor of the underlying histopathology. These variable phenotypes and neuropsychological profiles, as well as the brain perfusion profiles, may reflect different patterns of neurodegeneration in progranulin disease. Since all the mutations led to a progranulin haploinsufficiency, they would not in themselves be responsible for the phenotypic variability. Other modifying factors may therefore play a major role in determining the age at onset, the distribution and progression of brain damage and, thus, the clinical presentation of the disease.

Supplementary material

Supplementary material is available at Brain online.

Acknowledgements

We wish to thank Dr François Sellal (CH Colmar), Dr Véronique Golfier (CH St Brieuc), Dr Catherine Thomas-Antérion (CHU Saint-Etienne), Dr Bernard-François Michel (Hôpital Ste Marguerite, Marseille), Dr Mira Didic (CHU Marseille), Pr William Camu (CHU Montpellier), Dr François Salachas (Hôpital de la Salpêtrière, Paris), Dr Claude Lidy (Hôpital Lariboisière, Paris), Pr Fausto Viader (CHU de Caen), Dr Jean-Michel Thenint, Dr Christine Magne (CHU Nantes), Dr Isabelle Berry (CHU Toulouse), Dr Jean-Albert Lotterie (CH Toulouse), Dr Marielle Decousus (CHU Saint-Etienne), Pr Pierre Véra (CHU Rouen), Pr Annie Laquerrière (CHU de Rouen) and Pr Charles Duyckaerts (Hôpital de la Salpêtrière, Paris). We thank Dr Merle Ruberg for her helpful suggestions on the article (INSERM U679, Hôpital de la Salpêtrière, Paris, France) and Ms. Lydia Guennec, Isabelle Lagroua and Christelle Dussert (DNA and cell bank of IFR70, Hôpital de la Salpêtrière, Paris) for their excellent technical assistance.

This study was supported by CRIC AP-HP 01107 (to A.B.), ANR R6363DS (to A.B.) and France Alzheimer association (to A.B.); and C.V.B. was funded by the Special Research Fund of the University of Antwerp, the Fund for Scientific Research Flanders (FWO-F), the Interuniversity Attraction Poles program P6/43 of the Belgian Science Policy Office, and a Zenith Award of the Alzheimer's Association, USA. I.L.B. had a fellowship from the France Alzheimer association, and JvdZ is holder of PhD fellowship of the Institute for Science and Technology Flanders (IWT-F), Belgium.

References