Differences in Tau and Apolipoprotein E Polymorphism Frequencies in Sporadic Frontotemporal Lobar Degeneration Syndromes

Background  Frontotemporal lobar degeneration (FTLD) has different clinical phenotypes and is associated with several pathologic findings, most commonly dementia lacking distinctive histology or Pick disease. We know that the tau H1 haplotype is associated with some clinical and histologic phenotypes, for example, progressive supranuclear palsy and corticobasal degeneration. Furthermore, the apolipoprotein ϵ4 allele (APOE ϵ4) may be associated with Pick disease.

Objective  To determine if different clinical phenotypes of FTLD are associated with different tau haplotype and APOE allele frequencies.

Patients and Methods  All patients with FTLD with available DNA specimens (n = 63) seen at the Mayo Clinic, Jacksonville, Fla, were retrospectively classified according to the following clinical phenotypes: frontal dementia (FD); progressive, nonfluent aphasia (PA); or fluent, anomic aphasia (AA). DNA specimens were genotyped for APOE allele and tau haplotype frequencies and were compared with cognitively normal patients (n = 338) and patients with Alzheimer disease (AD) (n = 193).

Results  Patients with AA had increased APOE ϵ4 frequency (30.4%) compared with patients with FD (14.8%, P = .04) and cognitively normal patients (11.1%, P<.001). Patients with AA also had increased tau H2 haplotype (37.0%) frequency compared with patients with FD (11.1%,P = .002), patients with AD (21.8%, P = .02), and cognitively normal patients (19.8%, P = .004). The increase in tau H2 haplotype frequency (50.0%) is especially pronounced in patients with AA who are APOE ϵ4 positive compared with patients with FD (18.8%, P = .04), patients with AD (24.8%, P = .005), and cognitively normal patients (15.3%, P<.001). APOE ϵ4 and tau H2 haplotype frequencies are not significantly different in patients with FD and PA compared with healthy patients.

Conclusions  Clinical subtypes of FTLD have different tau and APOE genotype frequencies, suggesting these genes may influence the clinical presentation. Further studies should be performed to confirm this finding and to see if the pathologic phenotypes are also associated with different tau and APOE genotype frequencies.

FRONTOTEMPORAL dementia (FD), progressive, nonfluent aphasia (PA), and semantic dementia (SD) are clinical subtypes of a neurodegenerative disorder designated frontotemporal lobar degeneration (FTLD).¹ These clinical syndromes are determined by the distribution of the pathologic findings of the frontal variants that cause FD or PA and temporal variants that cause SD. The left temporal variants of SD present with a fluent, anomic aphasia (AA), whereas the right temporal variants more likely present with visual agnosia. A rationale for grouping these distinct clinical syndromes under the term FTLD is that these syndromes share common histopathologic findings. The most common finding is a microvacuolar-type condition that has been variously labeled as dementia lacking distinctive histology (DLDH), FD, or frontal lobe dementia.^1-3 The other common pathologic finding is Pick disease (PcD). There seem to be differences in the frequency of these pathologic findings among the clinical subtypes of FTLD.^4-6 However, some consider DLDH and PcD to be part of the same pathologic spectrum and use the term Pick complex.⁷

Frontotemporal dementia with parkinsonism (FTDP-17) is a familial form of FTLD that can present with any of these clinical phenotypes and has been genetically linked to chromosome 17. Most FTDP-17 families have mutations in the gene encoding for the microtubule-associated protein tau,⁸ suggesting a common pathogenic mechanism in these familial forms of FTLD. Nearly all FTDP-17 families also have pathologic tau inclusions on brain autopsy. FTDP-17 can appear clinically and pathologically similar to other neurodegenerative diseases, particularly corticobasal degeneration (CBD) and progressive supranuclear palsy (PSP).^9,10

Polymorphisms in the tau gene are also associated with certain neurodegenerative disorders. These polymorphisms are inherited as 2 classes of haplotypes termed H1 and H2.¹¹ Patients with pathologically confirmed PSP and CBD almost invariably have at least 1 H1 haplotype.¹² One study¹³ found no significant alteration in tau haplotype frequencies in pathologically confirmed PcD, although there was a nonsignificant trend toward an increase in the H2/H2 genotype. In Alzheimer disease (AD), the H1 and H2 haplotypes occur at frequencies equivalent to a control population.¹¹

In addition to FTDP-17, abnormal accumulations of the tau protein occur in other sporadic neurologic diseases, including PcD, CBD, PSP, and AD. Two classes of tau protein, 3 repeat and 4 repeat, result from alternative splicing of exon 10 of the tau gene. In FTDP-17, the amount of 3-repeat to 4-repeat isoform inclusions varies according to the specific genetic mutation in the tau gene. For example, mutations in the exon 10 region lead to an increase in exon 10 splicing and an accumulation of the 4-repeat isoform. The cause of tau accumulation in sporadic diseases is unknown, although some patterns of tau accumulation have emerged. The predominant tau isoform in Pick bodies of PcD is 3-repeat tau, whereas in CBD and PSP tau inclusions are predominantly 4-repeat tau. The neurofibrillary tangles of AD are composed of 3-repeat and 4-repeat tau.¹⁴ Because of these findings and the differences in tau haplotypes frequencies, it has been hypothesized that the tau H1 haplotype found in PSP and CBD may predispose to 4-repeat tau inclusions in these disorders.¹² In addition, DLDH has not been shown to have tau inclusions but does have abnormal ubiquinated protein inclusions.

Another gene implicated in neurodegenerative disorders is apolipoprotein E (APOE). The APOE ϵ4 allele is increased in frequency and modifies the age of onset of AD.¹⁵ The APOE gene has also been implicated in FTLD, although studies^4,16-29 have yielded conflicting results, with some showing an increase in APOE ϵ4 frequency or an age effect and others showing neither. These studies have used various clinical and pathologic classifications of FTLD, and only a few have divided patients with FTLD into groups based on clinical presentation. The association between APOE ϵ4 and pathologically confirmed PcD has been variable but so have the pathologic definitions, with some studies requiring the presence of Pick bodies and Pick cells and others requiring just one of these pathologic findings.^17,23,26,28 APOE polymorphisms are not only associated with AD but also recovery from cardiac arrest,³⁰ head injury,³¹ and risk of dementia after stroke.³² Thus, APOE polymorphisms could plausibly be associated with neurologic damage in other diseases such as FTLD. Therefore, the role of APOE and tau genes in sporadic cases of FTLD still needs to be determined.

The specific aim of this study is to determine the tau haplotype frequencies and APOE allele frequencies in a large series of clinically characterized patients with FTLD grouped according to the clinical presentation. We hypothesize that there will be differences among these clinical groups that may reflect pathologic differences.

Patients with FTLD were seen at the Mayo Clinic Disease Center, Jacksonville, Fla, and were enrolled in the Jacksonville center's patient registry between January 1, 1992, and December 31, 2000. Control groups were patients enrolled in the Mayo Alzheimer's Disease Center registry at the Mayo Clinic, Rochester, Minn, and were diagnosed as having probable AD by National Institute of Neurological and Communicative Disorders and Stroke criteria³³ or were cognitively normal.

Two neurologists (N.R.G.-R.) blinded to genetic data retrospectively classified FTLD patients according to criteria of Neary et al¹ for FD, PA, and SD. Since many patients were seen before this publication, some patients did not have all of these criteria documented. Therefore, the following modified criteria were used. Patients with FD had to have either disinhibited behavior or apathy as the predominant presenting symptom. Each patient also had to have at least 1 of the supportive criteria of Neary and colleagues for FD, such as hyperphagia, perseveration, or utilization behavior. Patients with PA had nonfluent speech with phonemic paraphasic errors or agrammatism as the presenting symptom with other areas of cognition and behavior intact. For patients with SD, we used the subtype of those presenting with AA because we have seen very few patients with initial symptoms of visual agnosia. Patients with AA had fluent speech with a prominent anomia, normal repetition, and normal comprehension for syntactic aspects of language as the predominant early symptom. All had predominant left temporal lobe atrophy. We did not require impairment of word meaning in patients with AA because we have found this may not be present early in pathologically proven PcD affecting the left temporal lobe.³⁴ Furthermore, because we did not require impairment of word meaning, we do not use the term SD but rather AA. All patients with FTLD had relative preservation of orientation, episodic memory, and visual spatial skills. Using these criteria, the 2 reviewers came to a consensus diagnosis of FD, PA, AA, or insufficient clinical data to classify as FTLD. A group of patients could not be fitted into one of these groups because they presented with a frontal and an aphasic syndrome (FD + A) and were classified as such. All patients had structural or functional neuroimaging that corroborated the diagnosis. Abnormalities were predominantly left and right frontal in patients with FD, left frontal in patients with PA, and left temporal in patients with AA.

APOE genotype was determined using basic methods described in the single-day APOE genotyping,³⁵ with the following modifications. The DNA was amplified using a thermal cycler (Hybaid Touchdown Thermal Cycler; Hybaid, Middlesex, England). Polymerase chain reaction (PCR) conditions were an initial denaturation at 94°C for 5 minutes, followed by 35 cycles overall of a 94°C denaturation step for 30 seconds, an annealing step consisting of a 22-cycle 60°C to 50°C touchdown at 0.5°C per cycle for 30 seconds and a 72°C extension step for 45 seconds, then a final extension at 72°C for 10 minutes completing amplification. Genomic DNA was amplified using the following primer sequences: upstream, 5′-TAAGCTTGGCACGGC TGTCCAAGGA-3′; downstream, 5′-ACAGAATTCGC CCCGGCCTGGTACAC-3′.

After amplification, 10 units of CfoI (Promega Corp, Madison, Wis) enzyme and its buffer were added directly to 20 µL of PCR product. The mixture was incubated at 37°C for 5 hours, giving main fragment sizes of 91, 83, 72, and 48 base pairs. The digest was run on a 4.5% agarose gel with ×1 Tris-boric EDTA buffer.

Tau mutations and haplotypes were determined by amplifying tau exons 7 and 9 through 13 from genomic DNA with primers designed to flanking intronic sequence. A total of 25 ng of DNA was used in a 50-µL reaction mixture containing 20pM of each primer, 0.2mM dNTPs (deoxyribonucleoside triphosphate solution), 1 U of Taq with ×10 buffer (QIAGEN Inc, Valencia, Calif), and 1.5mM magnesium chloride. Amplifications were performed oil-free in thermal cyclers (Hybaid). Conditions were 35 cycles of 94°C for 30 seconds, 60°C to 50°C touchdown for 30 seconds, and 72°C for 45 seconds, with a final extension of 72°C for 10 minutes. All products were purified using the QIAquick PCR Purification Kit (QIAGEN Inc). For each exon, 100 ng of product was sequenced in both directions using ABI Big Dye (Applied Biosystems Inc, Foster City, Calif) with relevant PCR primers. Sequencing was performed on an ABI377 automated DNA sequencer (Applied Biosystems Inc). Tau haplotypes were determined from polymorphism analysis of sequence data.

A χ² test was used to assess whether differences in APOE ϵ4 allele and tau haplotype frequencies existed with the FTLD subgroups and also by pairwise comparison with cognitively normal and AD control groups. Analysis was performed also for tau haplotype frequencies by stratifying for the presence or absence of APOE ϵ4.

Sixty-three patients met criteria for 1 of the FTLD diagnoses, and demographic data are presented in Table 1. We found no tau mutations in the patients with FTLD. Table 2 gives the individual APOE ϵ4 and tau H2 haplotype frequencies and the tau H2 frequency when APOE ϵ4 is either present or absent. The FTLD subgroups have significant overall differences in tau H2 haplotype frequency (P = .007) and have differences that approach significance in APOE ϵ4 frequency (P = .07) and tau H2 haplotype frequency in APOE ϵ4–positive patients (P = .07), justifying individual subgroup comparisons. Table 3 gives each P value for χ² tests of individual subgroup comparisons. The significant results are summarized as follows. The APOE ϵ4 allele is more common in patients with AA compared with FD (P = .047) and cognitively normal (P<.001) patients. The tau H2 haplotype is more common in patients with AA compared with FD (P = .002), AD (P = .02), and cognitively normal (P = .004) patients. When tau haplotype frequencies are stratified by the presence or absence of APOE ϵ4, we find the tau H2 haplotype more common in APOE ϵ4–positive patients with AA compared with FD (P = .04), AD (P = .005), and cognitively normal (P<.001) patients. The tau H1 haplotype is more common (ie, H2 haplotype is less common) in patients with FD without an APOE ϵ4 allele compared with cognitively normal patients (P = .048). The FD + A group (n = 9) was not included in the statistical analysis because of the small numbers, but the APOE ϵ4 allele frequency (16.7%) is comparable to cognitively normal controls, and the H2 haplotype frequency is low (5.6%). There are very few APOE ϵ2 alleles in the FTLD groups (overall APOE ϵ2 frequency equaled 3.5%). There is no significant effect of the presence or absence of the APOE ϵ4 allele or tau H2 haplotype on the age of onset.

We did not find any pathogenic tau mutations in patients with FTLD, which is consistent with a previous report³⁶ that found tau mutations to be a rare cause of sporadic FTLD. Also, we found that the APOE ϵ4 frequency is not increased in the FD and PA subtypes of FTLD, which is consistent with most previous studies of APOE ϵ4 in FTLD. These previous studies are summarized in Table 4. Since we have somewhat small numbers of FD and PA patients, our study could be underpowered to detect an increase in APOE ϵ4 in these patients. However, the APOE ϵ4 frequencies in FD and PA patients are very close to frequencies in cognitively normal patients, and we doubt that increased numbers would detect a difference. Nonetheless, some studies do show a significant increase in APOE ϵ4 in FTLD populations, and we believe this is because most previous studies did not separate patients based on clinical presentation. Our study shows that there are genetic differences within the clinical subtypes of FTLD. An explanation for these differences could be the differences in pathologic features among different FTLD clinical subtypes. As mentioned, most cases of FTLD will have either DLDH or PcD on histopathologic examination. In one of the first pathologic series³ of patients presenting with frontal lobe dementia, 16 patients had pathologic findings equivalent to DLDH and 4 patients had PcD. At least 50% of patients with PA will also have DLDH, whereas 20% will have PcD.⁴ However, in a summary⁵ of the 12 pathologic reports of SD (all had AA), 7 had PcD and 5 had DLDH, suggesting that PcD may be a more common finding in these patients.

If SD is more likely pathologically to be PcD, it is interesting that we found the APOE ϵ4 allele frequency to be increased in our AA group. The largest pathologically based study²³ of PcD showed an APOE ϵ4 allele frequency of 42%. This study required the presence of Pick bodies and Pick cells for the diagnosis of PcD, whereas other smaller studies^17,26,28 of pathologically confirmed PcD required either Pick bodies or Pick cells and had APOE ϵ4 ranges of 20% to 33%.

Also interesting is the increase in H2 haplotype frequency in AA. This effect was particularly pronounced in those who were APOE ϵ4 positive. The one study¹³ examining tau haplotypes in PcD showed a nonsignificant trend toward an increase in tau H2/H2 genotype compared with controls (15% vs 7%) but no difference in overall H2 frequency (30% vs 26%). This study used the pathologic criteria of Pick bodies and Pick cells for a diagnosis of PcD. If the H2 haplotype does ultimately prove to be increased in PcD, it would support a theory as to the biological effects of tau H1 and H2 haplotypes. As stated previously, PSP and CBD have an overrepresentation of the tau H1 haplotype, which has been hypothesized to predispose to 4-repeat tau inclusions. A converse of this hypothesis could be that the tau H2 haplotype predisposes to 3-repeat tau inclusions as found in PcD.

An alternative explanation for the increase in APOE ϵ4 in our AA group is that many of these patients will have AD-type pathologic findings with amyloid plaques and neurofibrillary tangles. Pathologically confirmed AD may present with AA, but careful assessment of episodic memory and visual spatial functioning reveals deficits in these areas as well.³⁷ Even if some of our patients with AA do have AD, they still seem to be clinically distinct, and the increase in the H2 haplotype frequency compared with the AD controls suggests they are different. An explanation could be that the occurrence of the H2 haplotype is altering the distribution of the AD pathologic features, resulting in an unusual clinical syndrome.

Regardless of the pathologic features, the increase of H2 haplotype frequency in patients with AA with an APOE ϵ4 allele suggests that there may be an interaction between these 2 genes, resulting in the clinical phenotype. A recent article³⁸ also found an interaction between APOE and tau polymorphisms in FTLD. However, their results showed tau H1 haplotype (which they termed A haplotype) in combination with the APOE ϵ4 allele as a risk factor for FD. They divided patients with FTLD into groups by clinical phenotypes, but the SD group was small (n = 8).

We also found the H1 haplotype frequency to be increased in the FD group of patients who did not have an APOE ϵ4 allele, and it also seems increased in the FD + A group. However, this association was relatively weak and will need to be examined in additional case-control series before the significance of this observation can be determined.

In conclusion, we find that there are genetic differences in the distribution of APOE and tau genotypes of patients presenting with various FTLD clinical syndromes. Future studies with pathology series will be important to see if providing genetic information such as the APOE and tau genotypes along with the clinical syndrome helps predict the pathologic findings. This may also ultimately improve our understanding of the biology of these diseases, such as whether a particular tau haplotype predisposes to a particular type of tau inclusion. Therefore, future clinical, pathologic, and genetic correlation studies should increase our diagnostic accuracy and the knowledge of the pathogenesis of FTLD-like disorders.

Accepted for publication November 30, 2001.

Author contributions: Study concept and design (Drs Short, Graff-Radford, and Hutton); acquisition of data (Drs Short, Graff-Radford, and Hutton, Ms Adamson, and Mr Baker); analysis and interpretation of data (Drs Short, Graff-Radford, and Hutton); drafting of the manuscript (Dr Short); critical revision of the manuscript for important intellectual content (Dr Graff-Radford, Ms Adamson, and Mr Baker); statistical expertise (Dr Short); obtained funding (Drs Graff-Radford and Hutton); administrative, technical, and material support (Ms Adamson, Mr Baker, and Dr Hutton); study supervision (Drs Graff-Radford and Hutton).

This study was supported by grant P50 AG16574-02 from the National Institute on Aging, National Institutes of Health, Bethesda, Md.

We thank Zoe Arvanitakis, MD.

Corresponding author and reprints: Neill R. Graff-Radford, MBBCh, FRCP(London), 4500 San Pablo Rd, Mayo Clinic, Jacksonville, FL 32224.