ApoE4 lowers age at onset in patients with frontotemporal dementia and tauopathy independent of amyloid-β copathology

Introduction Apolipoprotein E (ApoE) is the most important genetic risk factor for Alzheimer's disease (AD), with ApoE4 thought to enhance and accelerate amyloid-β (Aβ) pathology. ApoE4 has recently been described to increase neurodegeneration in a mouse model of frontotemporal dementia (FTD), in vitro, and in patients, demonstrating that ApoE4 modifies tauopathy independently of Aβ. This raises the question whether ApoE genotype also modifies the clinical phenotype in patients with FTD with tau pathology. Methods We analyzed 704 patients with FTD, including a genetically and neuropathologically confirmed subset, and 452 healthy elderly controls. We compared ApoE4 genotype frequency and age at onset in tau+ or TDP43+ FTD patients with or without Aβ copathology. Results The ApoE4 genotype lowered age at onset in patients with FTD and tau pathology, particularly once accounting for confounding effects of Aβ pathology. Discussion We conclude that ApoE4 accelerates neurodegeneration in FTD patients with MAPT mutations or FTLD-tau pathology, independent of Aβ.

Apolipoprotein E (ApoE) is the strongest known common genetic risk factor for Alzheimer's disease (AD) [1,2]. Relative to the most common allele (ApoE3), ApoE4 is associated with an increased risk of late-onset AD and an earlier age at onset (AAO); conversely, ApoE2 confers lower risk and later onset [2,3]. Multiple lines of evidence suggest that ApoE4 risk in AD principally relates to enhanced and accelerated cerebral Ab pathology [4]. More precisely, ApoE4 appears to accelerate the early seeding of amyloid pathology, most likely by decreasing Ab clearance and enhancing Ab aggregation [5]. Shi et al. recently described how ApoE4 increases the burden of cerebral tau pathology, neuroinflammation, and brain atrophy in a P301S mouse model of frontotemporal dementia (FTD) and in vitro; they also demonstrated that in patients with a primary tauopathy, ApoE4 was associated with more severe regional neurodegeneration and that ApoE41 AD patients with amyloid-b (Ab) pathology showed faster disease progression [6]. New evidence from Shi et al. demonstrates that ApoE4 modifies tauopathy independent of Ab, raising the question whether ApoE genotype also influences risk or modifies the clinical phenotype in patients with primary tauopathies. Several studies report an association between ApoE and FTD with a protective effect for ApoE2 and an increased risk conferred by ApoE4 [729], whereas another shows no association [10]. A recent publication demonstrated a deleterious effect of ApoE2 in a tau transgenic mouse model [11]. However, interpretations of the associations of ApoE with the clinical syndrome of FTD are complicated by two factors: frontal variant AD can be misdiagnosed for FTD, and the fact that the FTD syndrome comprises tau, TDP-43, or other pathologies. Prompted by Shi et al. [6], we therefore analyzed existing data we held on 704 patients with FTD and 452 healthy elderly controls using SPSS25 to test the hypothesis that ApoE genotype has a modifying effect on clinical phenotype in those with or expected to have tau pathology defined by a highly penetrant MAPT gene mutation or by neuropathological examination. We found that the ApoE4 genotype lowered age at clinical onset in patients with dementia and tau pathology, and was a particularly strong effect once the confounding effects of amyloid b pathology were taken into account.
The patients in this FTD cohort were previously tested for causative genetic mutations [12] and are described in Table 1. Patients with causative mutations in genes not typically associated with FTD were excluded from the present analysis. Patients were analyzed in groups based on genetic and/or neuropathological data, patients in whom no causative mutation had been identified and for whom no neuropathological data were available either were classified as "clinically diagnosed" and not included in analyses of genetic subgroups. Dementia and personality change were the predominant symptoms in our series (54.7% -33.3% "behavioral type FTD", 21.4% "FTD", 1.1% "Dementia"), followed by aphasia (31.8% -16.9% progressive nonfluent aphasia, 13.1% semantic dementia, 1.8% logopenic progressive aphasia) and additional motor or muscular symptoms (11.5% -6.0% corticobasal degeneration, 2.4% progressive supranuclear palsy, 3.7% FTD with motor neuron disease, 0.3% inclusion body myositis with Paget's disease and FTD). As has been shown previously [7], ApoE4 was more common in clinically diagnosed FTD compared with healthy elderly controls; 221 patients (31.4%) and 105 controls (23.2%) were at least heterozygous for ApoE4 (chi-squared test, P 5 .003), carrying either one or two ApoE4 alleles. Based on Shi et al. [6], we hypothesized that ApoE4 would have a modifying effect on AAO in FTD cases with proven or known or expected tau pathology. While AAO is known to be highly correlated among family members with a highly penetrant MAPT gene mutation [13], here, AAO in confirmed FTD patients who had MAPT gene mutations or FTLD-tau neuropathology was significantly lower in the presence of an ApoE4 allele (tau1 ApoE41 compared with tau1 ApoE42 patients; 53.0 years vs. 56.9 years, t-test, P 5 .043, Fig. 1A). In a further exploratory analysis, we tested whether ApoE4 also affected non-tau cases in the same way. However, ApoE4 neither had modifying effect on AAO in FTD patients with FTLD-TDP43 neuropathology or in FTD patients without an ascertained cause, nor did sex affect AAO in the different groups (TDP1 ApoE41 compared with TDP1 ApoE42: 57.6 years vs. 58.1 years; clinical FTD ApoE41 compared with clinical FTD ApoE42: 58.7 years vs. 58.9 years; ANOVA, all P values . 0.05; all Fig. 1A).
We went on to test whether the modifying effect of ApoE in FTD-tau might relate to the presence of Ab copathology without meeting criteria for a formal pathological diagnosis 67.0 (11) 65.6 (10) -NOTE. AAO refers to age at clinical onset, given in years of age; N refers to number of cases. "All FTD" comprises all 704 patients with FTD included in this study; these were all patients from the UCL FTD cohort except patients previously shown to carry mutations not typically associated with FTD [12]. "Clinical FTD" refers to the 449 patients with a clinical diagnosis of FTD but without a genetic diagnosis despite previous testing on a dementia panel [12] and no available neuropathological data.
NOTE. "Confirmed FTD" comprises 255 patients with a genetic or neuropathological confirmation of their diagnosis or both. It includes 59 patients with C9orf72 expansions, 53 patients with MAPT mutations, 44 patients with GRN mutations, five cases with VCP mutations, two cases with TBK1 mutations, as well as one case each with a mutation in CHMP2B, TYROBP, and TARDBP mutation. "Confirmed FTD" also includes 139 patients with neuropathologically confirmed disease: 70 patients with tau pathology and 65 patients with TDP-43 pathology, as well as 4 neuropathologically confirmed FUS cases. 19 patients with neuropathologically confirmed tau pathology also had a genetically confirmed deleterious MAPT mutation, whereas 31 patients with neuropathologically confirmed TDP-43 pathology were found to also carry a deleterious mutation (13 C9orf72 expansion, 17 GRN mutations, 1 TBK1 mutation). Patients with both genetically and neuropathologically confirmed disease were only counted once, explaining disparities of sums in the table. "N Mutation-Tau" refers cases due to causal mutations in the MAPT gene. "N Mutation-TDP" refers to causal mutations in the C9orf72 (59 cases), GRN (44 cases), TARDBP (one case) and TBK1 (2 cases) genes, which are known to cause TBK-43 pathology. "N Mutation-Confirmed FTD" comprises all genetically confirmed cases as described previously. Subcohorts referring to Ab only include cases with available neuropathological data. For 7 patients with tau pathology and 5 patients with TDP pathology, information on Ab status was not available; for one tau patient, information on AAO was not available. AAO is given in years, and in controls refers to age at testing. ApoE41 includes all carriers of at least one ApoE4 allele, both homo-and heterozygous.
of AD in 139 cases with available neuropathological data. Postmortem cases were neuropathologically staged according to Thal phases and Braak and Braak stages [14,15]. A number of cases (n 5 44) scored at Thal phases 0, 1, 2, or 3, but none reached Thal phase 5 or Braak stage 6, needed for a definitive diagnosis of AD. ApoE4 carrier status increased the likelihood of Ab pathology, both in cases with tau and TDP43 pathology (chi-squared test, P 5 .001 and chi-squared test, P 5 .006, respectively): for the tau1 ApoE41 patients with FTD and available neuropathology data, 15 of 21 (71.4%) had Ab copathology compared with 11/42 (26.2%) ApoE42 tau1 cases (chi-squared test, P 5 .001); for the TDP1 ApoE41 FTD cases, 8/13 (61.5%) had Ab copathology compared with 10/47 (21.3%) of ApoE42 TDP1 cases (chisquared test, P 5 .013). The prevalence of Ab copathology was similar when tau1 cases were compared with TDP1 cases (chi-squared test, P 5.259) and was associated with ApoE4 carrier status, both in cases with tau pathology (chi-squared test, P 5 .001) and without tau pathology (chi-squared test, P 5 .006). Patients who were positive for Ab pathology were significantly older at onset than patients who were found to have no Ab pathology at postmortem (60.9 vs. 57.1 years old, respectively, t-test, P 5 .024). In patients with FTLD-tau, ApoE genotype and Ab copathology had independent associations with age at clinical onset (linear regression, n 5 62, ApoE -10.7 years, P 5 5.2 ! 10 25 ; Ab 17.6 years, P 5 .002). The effects of ApoE41 carrier status and Ab on AAO were independent of each other both in patients with and without tau (univariate ANOVA, P 5 .550 and P 5 .545, respectively).
Our analyses, carried out in a well-characterized cohort of patients with FTD, provide evidence that ApoE4 genotype ac-celerates neurodegeneration in patients with MAPT mutations or FTLD-tau pathology independent of Ab-notably AAO in FTD-tau cases was on average about 3.9 years earlier in those who carried an ApoE4 allele. Previously, Shi et al. proposed that ApoE4 genotype acts through a toxic gain of function mechanism to exacerbate or modify tau pathology, neuroinflammation, autophagy, and reactive astrocyte activation; they provided extensive evidence in a mouse model of FTD and in neuropathological cases of tauopathies, as well as clinical support in an AD cohort [6]. Our data further strengthens this hypothesis providing evidence in human that ApoE4 modifies the clinical phenotype of FTLD-tau, independent of Ab copathology. While ApoE4 status was associated with Ab copathology at postmortem examination in both tau1 and in TDP1 cases of FTD, its effect on AAO was specific to tau1 FTD cases (Fig. 1A). Indeed, ApoE4 and Ab copathology were associated with opposing effects on AAO in our data set in tau1 FTD cases (Fig. 1B). The confounding impact of Ab copathology, along with the heterogeneous nature of FTD pathology and the specificity of the effect of ApoE4 on tau pathology may explain why previous studies of ApoE4 in FTD [7-10] did not detect this influence. Despite the robust statistical finds and the large cohort of patients with FTD, analysis of pathological and gene mutation-defined subgroups inevitably leads to relatively restricted numbers and therefore further replication will be important.

Acknowledgment
The study was funded by the Medical Research Council (UK), the NIHR Queen Square Dementia Biomedical Research Unit, Fig. 1. ApoE41 status is associated with significantly lower age at onset (AAO) in FTD cases with MAPT mutations or FTLD-tau neuropathology, but not in those with TDP pathology or in those without an ascertained genetic or neuropathological diagnosis (clinical FTD) (A). ApoE4 carrier status is associated with lower AAO, whereas Ab copathology was associated with later AAO (B). (B) displays the P values for the regression coefficients ApoE4 and Ab. Each dot corresponds to the AAO of one case, with the mean and 95% confidence interval marked in red. The number of patients in each subcohort is given in line N. For (A), AAO was not available for 3 tau1 ApoE42 patients, 4 TDP1 ApoE42 and 6 TDP1 ApoE41 cases, as well as 31 clinical FD ApoE42 and 9 clinical FTD ApoE41 patients; for Figure 1B, 62 FTD tau1 patients with available ApoE4 status and neuropathological data were included; AAO was not available for 1 ApoE4-Ab2 patient, and the Ab status was unknown for 7 patients. These cases were excluded from the relevant calculations and from the figures. the NIHR Biomedical Research Centre at University College Hospitals NHS Foundation Trust and Alzheimer's Research UK (ARUK). The authors are very grateful to patients and volunteers who have donated blood samples, their time, and their brains to research and have made this study possible. C.K. is supported by a Leonard Wolfson Foundation PhD fellowship. T.L. is supported by an Alzheimer's Research UK senior fellowship and Queen Square Brain Bank is supported by Reta Lila Weston Trust for medical research. J.B.R. is funded by the Wellcome Trust (103838) and the NIHR Cambridge Biomedical Research Center. J.D.R. is supported by an MRC