The Genetic and Pathological Classification of Familial Frontotemporal Dementia FREE

FRONTOTEMPORAL dementia (FTD) is a clinical diagnosis based on progressive personality change and language impairment related to frontotemporal lobar atrophy. It is a frequent cause of dementia, particularly in the younger age group, accounting for between 12% and 20% of all dementia cases.¹ Pick disease is the archetypal pathological form of FTD. It is characterized pathologically by the presence of swollen α-B-crystallin–positive neurons (Pick cells), and argyrophilic, tau-positive round inclusions (Pick bodies) that are particularly numerous in the granule cells of the hippocampal dentate fascia and the superficial layers of the frontotemporal neocortex. Pick disease is sometimes also used as a clinical term for patients presenting with a progressive frontal syndrome or language disorder, and as a diagnosis for neurological conditions with radiologic frontotemporal lobar atrophy; strictly, however, Pick disease should be reserved for pathologically diagnosed disease.

Although Pick disease is one pathological form of clinically diagnosed FTD, FTD may be caused by a number of other pathological substrates.¹ These include (1) dementia with ubiquitinated inclusions, first associated with motor neuron disease; (2) dementia lacking distinctive histopathologic features; (3) corticobasal degeneration; (4) Alzheimer disease; and (5) familial frontotemporal dementia linked to chromosome 17 (FTDP-17).^{1 - 4} Immunocytochemistry has greatly facilitated the diagnosis of these conditions, in particular with the identification of ubiquitin and/or tau-positive inclusions.

Analysis of the genetic basis of familial diseases has proved to be a powerful and successful approach to the study of neurodegeneration. The identification of pathogenic gene mutations in familial Alzheimer disease has helped our understanding of the pathogenesis of the more common sporadic forms of this disease.⁵ This approach is likely to be more difficult in FTD because of the underlying pathological heterogeneity. Although analysis of a pathological classification of FTD has been previously described, it is worthwhile focusing on the classification of the familial forms of this disease, since each of these families is likely to have an identifiable pathogenic gene mutation.⁶ The description of kindreds with FTDP-17 and the identification of pathogenic mutations in TAU has wider implications for neurodegenerative disease, but it is also the first step toward the genetic classification of the different FTD subtypes.^{7 - 9} In this study, we reviewed the pathological and genetic findings in a series of autosomal dominant families with FTD, in an attempt to define the pathological subtypes and to predict the minimum number of genes that remain to be identified in this condition.

Twenty-two families with autosomal dominant FTD were studied as part of an ongoing London, England–based study of early-onset dementia. Blood for DNA analysis and agreement for autopsy examination were obtained after patients gave informed consent. Exons 9 to 13 of TAU were sequenced after standard polymerase chain reaction and sequencing reactions were performed (Big-Dye terminator sequencing kits; ABI, Foster City, Calif), as previously described. Mutations in TAU were confirmed by both sequence analysis and restriction enzyme digestion.⁷ Sequence products were analyzed on a sequencer (ABI 377) and the results were visualized with software (Sequence Analysis and Auto Assembler; ABI). Autopsy examination was available in 17 of 22 families. From each brain, a comprehensive and standardized set of tissue blocks was taken, processed, sectioned, and stained with a variety of histologic and immunohistochemical stains, including antibodies to tau (AT8, monoclonal mouse, 1:200; Innogenetics N.V., Gent, Belgium) and ubiquitin (ubiquitin, polyclonal rabbit, 1:500; Dako Ltd, Ely, England).

Half (11/22) of the families identified had TAU mutations (Table 1). Nine families had the TAU exon 10 +16 mutation, 1 family had the TAU exon 10 +14 mutation, and 1 family with the exon 10 codon 301 proline-to-serine mutation (P301S) was identified. Two of these cases have been previously reported.¹⁰ The average age at onset for the family with the P301S mutation was 34 years, as compared with an average age at onset for the families with the +16 mutation of 49 years. Three main pathological subtypes were identified (Figure 1, Table 1). The FTD with tau deposition (FTD-tau) involved extensive neuronal and glial tau deposition with coiled oligodendroglial inclusions and tufted astrocytes. The FTD with ubiquitin inclusions (FTD-Ub) involved the occurrence of ubiquitin-positive, tau-negative small tanglelike or dot inclusions and neuropil threads in the frontal and temporal cortices, most prominently in the superficial cortical layers. Ubiquitin inclusions were also consistently seen in the dentate fascia of the hippocampus (Figure 1). The third pathological subtype showed neither tau nor ubiquitin-positive inclusions and was described as FTD with neuronal loss and spongiosis (FTD-NLS) (Figure 1). In family 1, immunocytochemical findings were not available, but in families 2 and 8, immunocytochemical analysis (including examination of the dentate fascia) confirmed the absence of both ubiquitin and tau inclusions. Other pathological findings were similar in all 3 groups. All 3 pathological subtypes generally involved superficial vacuolation and astrocytosis of the frontal and temporal cortex, and flattening and atrophy of the caudate nucleus. Clinically, some affected individuals in each pathological subtype showed evidence of parkinsonism. Only 1 family in this series was identified to have clinical motor neuron involvement (family 21), and this family did not have typical neuronal ubiquitin inclusions, although some granular neuronal ubiquitin immunoreactivity was identified.

Despite the frequent clinical diagnosis of familial Pick disease, no familial pathological Pick disease was identified. Although Pick disease was pathologically diagnosed in 1 autopsy examination, another member of that family had pathologically verified Alzheimer disease, suggesting that this is not a pathologically homogeneous FTD family. All FTD-tau families were identified as having a TAU mutation (Table 1), and the TAU sequence was normal in each of the FTD-Ub and FTD-NLS families.

This study indicates the genetic and pathological heterogeneity of familial FTD; in this series, 11 (50%) of 22 familial FTD cases had TAU mutations. Three other groups have studied the prevalence of TAU mutations in familial FTD.^{11 - 13} Rizzu and colleagues¹² report that 47% of patients with FTD and a positive family history had mutations in TAU, whereas Houlden and colleagues¹¹ suggested that only 11% (6/54) of FTD families had identifiable TAU mutations. However, neither of these studies provided a pathological analysis of the families studied. More recently, Poorkaj and colleagues¹³ studied a large series of familial FTD cases and identified TAU mutations in 10.5% of the familial cases and only 33% of the familial cases with tau pathological findings.¹³

In contrast, our study suggests that the presence of tau pathological findings in a familial FTD case strongly predicts the presence of a TAU mutation, whereas the presence of FTD-NLS or FTD-Ub pathological findings effectively excludes a mutation in TAU. Possible reasons for the discrepancies between these studies include the age at onset of families studied and the strength of evidence of a concordant family history. Our results correlate well with analysis of pathological findings in the families with unequivocally chromosome 17q21–linked FTD, in which practically all affected members had significant tau deposition and concomitant mutations in TAU.² Although neurodegeneration in familial tau deposition without TAU mutations can occur in families with clinically diagnosed progressive supranuclear palsy,¹⁴ this has not been commonly described in families with pathologically diagnosed tau-deposition FTD with multiple affected members. One exception may be the family with hereditary dysphasic disinhibition (HDD2), linked to the TAU region with a maximum lod score of 3.68. Tau immunocytochemical analysis has identified variable tau deposition in this family, although a recent report described depletion of tau on Western blot analysis.^{15 - 16} A mutation in TAU has not been reported in the family with HDD2.

Our series suggests that there are no clinical features that can reliably distinguish these 3 familial FTD subtypes, and clinical features of motor neuron disease with FTD were uncommon. We identified FTD-Ub pathological features in 3 (18%) of 17 families with pathologically diagnosed FTD. These are similar to the family with ubiquitin inclusion described by Kertesz and colleagues,⁴ the sporadic cases with semantic dementia identified by Rossor and colleagues,¹⁷ and the sporadic and familial cases identified by Jackson and colleagues¹⁸ and described as motor neuron disease inclusion dementia. These pathological reports confirm that superficial neocortical cell loss with vacuolation and ubiquitin inclusions in the dentate gyrus are core features of this disease.

We have also identified 4 (24%) of 17 families with FTD-NLS. The relationship between the FTD-NLS reported herein and other reported FTD subtypes is harder to establish given the overall pathological similarities between all of these diseases and the absence of tau and/or ubiquitin immunohistochemical analysis in some reports. The diseases described as "dementia lacking distinctive histopathology," "dementia lacking distinctive histological features," "familial dementia of adult onset with pathological features of a nonspecific nature," and "dementia with microvacuolar pathology and laminar spongiosis" may all correspond to either FTD-Ub or FTD-NLS, depending on the results of ubiquitin immunohistochemical analysis.^{3 ,6 ,19 - 20} Genetic linkage to chromosome 3 has been reported in a Danish kindred originating in Jutland, apparently without distinctive pathological features (OMIM No. 600795).²¹ A recent study using ubiquitin immunohistochemical analysis did not show intraneuronal ubiquitinated inclusions in the chromosome 3–linked kindred, and, conversely, analysis of a large family with ubiquitin inclusion dementia excluded linkage to chromosome 3.^{4 ,22} This suggests that chromosome 3–linked dementia may correspond to FTD-NLS described in this series. Assuming continuing pathological-genetic correlation across FTD, this series suggests that there are at least 2 additional genes to be identified that may be responsible for familial FTD.

In conclusion, we have demonstrated 3 pathological subtypes of familial FTD and a close correlation between the presence of a TAU mutation and the presence of tau pathological findings. Despite earlier reports suggesting that Pick disease was a common familial dementia, this was not supported by this study. There is genetic and pathological heterogeneity in familial FTD and, undoubtedly, additional genes will be identified that are responsible for these disorders.

Accepted for publication July 23, 2001.

Dr Morris is a Medical Research Council clinical training fellow. The Dementia Research Group and MRC Brain Bank are supported by the Medical Research Council (London). This work was also supported by the Progressive Supranuclear Palsy (Europe) Association (Wappenham, England) (Dr Morris), the Guarantors of Brain (London) (Dr Morris), the Bogue Research Fellowship, University College London (Dr Morris), the National Institutes of Health/National Institute on Aging (Bethesda, Md) program grant on tau (Drs Hardy and Hutton), the Mayo Clinic (Jacksonville, Fla) (Drs Hardy and Hutton), and the Smith Fellowship, Mayo Clinic (Dr Hutton). Dr Morris is now with the Department of Neurology, St Thomas' Hospital, London.

We are grateful to J. V. Clark, FRCPath, and S. Huson, MD, for allowing us to study their patients. We acknowledge the skillful technical assistance of Heidi Barnes of the MRC Brain Bank.

Corresponding author and reprints: Martin N. Rossor, MD, Dementia Research Group, Institute of Neurology, Queen Square, London WC1N 3BG, England (e-mail: m.rossor@dementia.ion.ucl.ac.uk).