Clinical Phenotype and Mutation Spectrum of Alzheimer's Disease with Causative Genetic Mutation in a Chinese Cohort.

BACKGROUND: Alzheimer's disease with a causative genetic mutation (AD-CGM) is an uncommon form, characterized by a heterogeneous clinical phenotype and variations in the genotype of racial groups affected.
OBJECTIVE: We aimed to systemically describe the phenotype variance and mutation spectrum in the large sample size of the Peking Union Medical College Hospital (PUMCH) cohort, Beijing, China.
METHODS: Next-generation sequencing (NGS) was carried out in 1108 patients diagnosed with dementia. A total of 40 Han Chinese patients with three AD gene mutations were enrolled. A systemic review of all the patients was performed, including clinical history, neurocognitive assessment, brain magnetic resonance imaging, and cerebrospinal fluid (CSF) biomarkers.
RESULTS: We studied the following gene mutation variants: 12 AβPP, 13 PSEN1, and 9 PSEN2, and 23 among them were novel. Most of them were early-onset, but PSEN1 mutation carriers had the youngest onset age. The commonest symptoms were similar to those of AD, including an amnestic syndrome, followed by psychiatric symptoms and movement disorder. On MRI, parietal and posterior temporal atrophy was prominent in PSEN1 and PSEN2 mutation carriers, while AβPP mutation carriers had more vascular changes. The CSF biomarkers profile was indistinguishable from sporadic AD.
CONCLUSION: We identified a small group of AD-CGM subjects representing 3.6% among more than 1000 demented patients in the PUMCH cohort. These subjects usually presented with early-onset dementia and exhibited significant clinical and genetic heterogeneity. Identification required complete screening of genetic mutations using NGS. Although family history was usually present, we found non-familial cases of all three genetic mutations. Copyright© Bentham Science Publishers; For any queries, please email at epub@benthamscience.net.

INTRODUCTION

Alzheimer’s disease with a causative genetic mutation (AD-CGM) represents a rare group of patients with dementia of Alzheimer’s type. Three pathogenic genes have been found to cause AD-CGM, presenilin 1 (PSEN1), presenilin 2 (PSEN2), and amyloid β protein precursor (AβPP) genes [1-3]. The clinical phenotype is heterogeneous and usually differs from typical sporadic late-onset Alzheimer's disease (LOAD) [4]. Most cases of AD-CGM have an early age of onset with rapid cognitive deterioration. Besides memory and cognition impairment, they may have other neurological symptoms, such as movement disorders [4]. Most cases have autosomal dominant Alzheimer’s disease (ADAD), but there are some reports of patients without a family history. To date, over 200 mutations have been identified in AD-CGM, and most are ADAD [5]. The mutations affect a common pathogenic pathway in amyloid precursor protein synthesis and proteolysis, leading to excessive production of beta-amyloid (Aβ) [5].

PSEN1 mutation is the most common genetic cause of AD-CGM. Typical features of patients with PSEN1 mutations include the youngest age at onset and the shorter duration of the disease [6] compared with other AD-CGM cases. Language impairment, psychiatric and behavior problems may be present in addition to the amnestic syndrome [6]. Other neurological symptoms found in patients with PSEN1 mutations include seizures, spastic paraparesis, extrapyramidal signs, and ataxia [6]. PSEN2 and AβPP mutations are less common than PSEN1 mutations. However, the mutation spectrum and clinical phenotypes are different among different ethnic groups. For instance, Asian patients with PSEN1 mutations present more frequently with disorientation and personality change but less frequently with the atypical clinical features mentioned above [7]. Several mutations have been reported previously among Han Chinese with AD-CGM based on pedigrees [8]. However, we have observed that some cases of AD-CGM presented without a family history, and we aimed to systemically describe the phenotype variance and mutation spectrum in the large sample size of the Peking Union Medical College Hospital cohort, Beijing, China.

MATERIALS AND METHODS

Participants

Next-generation sequencing was carried out in 1108 patients diagnosed with dementia in dementia and leukoencephalopathy clinic of Peking Union Medical College Hospital (PUMCH-cohort), Beijing, China, between Jan 2016 and Jan. 2019. The diagnosis of dementia was defined according to DSM-V [9]. In total, 40 patients (40/1108=3.6%) with Alzheimer’s gene mutation were enrolled. They were from different families and unrelated. They all fulfilled the definition of probable AD dementia in a carrier of a causative AD genetic mutation (AD-CGM) according to 2011 NIA-AA guidelines [10]. The study was done in accord with the ethical standards of the Helsinki Declaration of 1975. All patients or their caregivers were informed of the purpose of the study and the future publication of the article. Written consent was obtained from each patient or their caregiver(s). The ethics committee of Peking Union Medical College Hospital approved the study (No. JS1836). The ethnicity of all these patients was Han Chinese. Detailed clinical information and family histories were provided by reliable informants (27/40=67.5% with family history reports). All patients underwent complete physical and neurological examination, laboratory blood tests, and detail cognitive assessment. The cognitive battery included the following screening tests: Mini-Mental State Examination (MMSE), Montreal Cognitive Assessment-PUMCH edition (MoCA-PUMCH edition) [11], Activities of Daily Living (ADLs), Clinical dementia rating scale (CDR), and Hospital Anxiety and Depression Scale (HAD). After screening, a detailed domain-specific neuropsychological battery test was performed, including story memory, category fluency, Trail Making A and B, digit symbol, digit span, auditory verbal learning test, block design test, and modified Rey figure, among others. All these chosen scales had been validated in the Mandarin language with available normative data for the Chinese population.

Neuroimaging and Cerebrospinal Fluid Analysis

Magnetic resonance imaging (MRI) examinations were performed in participants using a 3T MRI scanner (Avanto, Siemens, Erlangen, Germany). Axial T1-weighted, T2-weighted, T2-Flair sagittal, and/or coronal T1-weighted imaging was performed. Brain atrophy involving the temporal lobe and parietal lobe as well as white matter lesions, were evaluated visually. Cerebral vascular changes, including infarction and lacunae, were recorded. Electroencephalography (EEG) (Nihon Kohden, Japan) studies were performed in a subset of the participants (20 cases). Brain 18F-FDG positron emission tomography, computed tomography (PET/CT) (Biograph, Siemens, Erlangen, Germany) were conducted in some participants (4 cases).

In 8 of the patients, lumbar puncture was performed after informed consent, and cerebrospinal fluid (CSF) biomarkers for AD were measured. The collection of CSF was gravity drip, and low protein absorption tubes were used. All CSF samples were sub-packed and stored at -80°C. Commercial accessible ELISA kits were utilized for measurement of CSF t-tau, p-tau, Amyloid β42 (Aβ42) respectively by INNOTEST Tau ELISA, Phospho-tau, and Aβ42 (Fujirebio, Ghent, Belgium).

Genome Sequencing and Variants Identification

Written informed consent for genetic analysis was obtained from all patients. Genomic DNA was extracted from fresh peripheral blood leukocytes. Whole exome sequencing was carried out in 281 cases, and the rest 827 cases underwent targeted panel sequencing for selected loci, including AβPP, PSEN1, and PSEN2. Next-Generation Sequencing (NGS) technology was performed on Illumina HiSeq (Illumina, USA), which was verified by Sanger sequencing. The results were analyzed according to the standard AβPP (NM_000484.3), PSEN1 (NM_000021.3), and PSEN2 (NM_000447.2) reference sequence. Besides, we analyzed the genotype of the Apolipoprotein E(ApoE) gene.

All the variants detected are listed in Supplementary Table . All variants were searched in ALZFORUM, Mutations Database (https://www.alzforum.org/mutations) and PubMed (https://pubmed.ncbi.nlm.nih.gov), then classified as a novel or reported. They were all classified as pathogenic, likely pathogenic, or uncertain significance according to American College of Medical Genetics and Genomics (ACMG) standards [12]. The including criteria in the analysis were based on bioinformatic analysis and were detailed as follows: the filtering was restricted to causative AD genetic mutations; variants previously reported to be causative AD mutations; non-synonymous variants causing missense, frameshift, or splicing mutation; the frequency of variants in the general population (1000 Genomes project, ExAC database, 1000 healthy Chinese controls) must be less than 0.5%; the variants were predicted to be functional protein damage by three function prediction software analysis (SIFT, PolyPhen, and LRT); benign and like benign variants on ACMG classification were discarded.

Statistical Analysis

Statistical analysis was performed on SPSS 22.0. Measurement data was described as average±standard deviation, and enumeration data was described as a ratio. Measurement data were statistically analyzed with t-test and ANOVA, while enumeration data (frequencies) was analyzed with the χ 2 test. Statistical graphs were produced with GraphPad Prism.

RESULTS

Demographic and Basic Information of the Participants (Table 1)

We reported the results of 40 AD-CGM patients among a total of 1108 subjects with dementia in the PUMCH-cohort (40/1108=3.6%). Participants included the following mutations: 17 AβPP (17/1108=1.5%), 14 PSEN1 (14/1108=1.2%), and 9 PSEN2 (9/1108=0.8%). Variants included were 12 AβPP, 13 PSEN1, and 9 PSEN2. In all the 34 variants detected, 23 were novel, and 11 were reported in the literature (Supplementary Table ). The demographic and basic clinical information of the three genetic groups are summarized in Table . Detailed information of the participants are listed in Supplementary Table .

Phenotype Distribution of the Participants (Table 2)

An amnestic syndrome at onset was the initial manifestation ranging in frequency from 50% to 66.6%, similar to LOAD. Psychiatric symptoms and behavioral problems occurred in 21/40 patients (52.5%). In all 3 mutations, there were patients with aphasia at onset (7 patients, 16.5%) and movement disorders (9/40=22%). In the AβPP and PSEN1 mutation, group 3 patients presented with episodes of cognitive disturbance followed by progressive memory loss, which was uncommon in AD. In the process of disease progression, dementia was characterized by the involvement of all cognitive domains. Psychiatric symptoms and behavioral complications (irritability, delusions, agitation, and depression) were more commonly observed than movement disorder manifestations such as parkinsonism, tremor, spasticity, or pyramidal sign, while no seizures or ataxia were reported. Irritability was significantly more found in PSEN1 mutation carriers, and agitation was more found in PSEN2 mutation carriers. As for movement disorders, spasticity/pyramidal sign was most common in PSEN1 mutation carriers. A total of 39 patients underwent brain MRI evaluation, except for one patient with a metal implant. The MRI of the brain revealed bilateral symmetrical cortical atrophy involving predominantly temporal and parietal lobes in 67% to 70% of the cases. Parietal lobe atrophy was more common in our patients than in sporadic LOAD. White matter hyperintensities were evaluated using the Fazekas scale [13]. Fazekas stages 2 and 3 were used to define severe white matter lesions that were noted in 12/39 (30.8%) patients in our cohort. On the other hand, vascular changes including lacunae and infarction were found in 29.4% of AβPP mutation carriers, while only 1/13 (8%) of PSEN1 and no AβPP mutation carriers had vascular changes. Electroencephalography (EEG) was obtained in 50% of the patients. Abnormal results included an increase of θ or σ slow waves in the bilateral frontal and temporal lobe or diffuse slow waves. No epileptic discharges were found. In 8 patients with CSF biomarkers analysis, the results were in concordance with typical AD manifestations, that included an increase in t-tau (Average 901.8pg/ml, cut-off >286.8pg/ml) and p-tau (Average 79.8pg/ml, cut-off >55.3pg/ml) levels, ratio of t-tau/ Aβ1-42 (Average 1.94, cut-off>0.67), with a decrease in 
Aβ1-42 (Average 546.9pg/ml, cut-off <614.6pg/ml). FDG-PET was obtained in 4 patients and PiB-PET in a single patient; all of them revealed typical AD changes. The bilateral temporal lobe, parietal lobe, and posterior cingulate gyrus hypometabolism were found. These results supported the diagnosis of AD and were indistinguishable from those described in sporadic LOAD.

Phenotype Differences among AβPP, PSEN1, and PSEN2 Mutation Carriers

The demographic and clinical phenotype differences of the three groups are summarized in Tables and (p-value in the last column). Onset age in PSEN1 carriers was significantly younger than AβPP carriers (p-value, 0.026). And that of PSEN2 carriers was in between (Fig. ). PSEN1 and PSEN2 carriers had a higher frequency of positive family history than AβPP carriers. PSEN1 mutation group had a lesser frequency of APOE ε4 carriers than the other two.

The cognition evaluation was indistinguishable among the three groups. Most of the patients were moderately affected. Amnestic onset was the most common in all, and no statistical difference was found among them. As for MRI imaging, parietal atrophy was found more often in the PSEN1 and PSEN2 groups, while temporal lobe atrophy was similar in all three mutations. The AβPP group had significantly more vascular changes than the other two. The CSF biomarkers profile was indistinguishable among them.

Representative Pedigrees Illustration

The pedigree of Proband 1 is shown in Fig. (. The patient is amale, heterozygous mutation carrier of AβPP

Fig. (2)

Pedigrees of the probands. A: proband 1; B: proband 2; C: proband 3. (A higher resolution / colour version of this figure is available in the electronic copy of the article).

(NM_000484.3) Exon16 c.2032G>A p.(Asp678Asn). He presented with memory loss with onset at 64 years of age and was moderately demented on admission. Brain MRI revealed bilateral atrophy of the temporal lobe and parietal lobe (Fig. -). His CSF biomarkers were Aβ42 361.3pg/ml, t-tau 883.3pg/ml, p-tau 66.8pg/ml.

Proband 2 is a woman with an age of onset at 41 years old. Her pedigree is shown in Fig. (. She is a heterozygous mutation carrier of PSEN1 (NM_000021.3) IVS8 c.869-2A>G. She presented with severe memory loss and non-fluent aphasia. Brain MRI revealed bilateral atrophy of the temporal lobe and parietal lobe (Fig. ). Her PiB-PET was positive with amyloid deposition at the frontal lobe and posterior lobes (Fig. ). FDG-PET showed symmetrical hypometabolism at the temporal lobe, parietal lobe, and posterior cingulate gyrus (Fig. ). Her CSF biomarkers were Aβ42 357.0pg/ml, t-tau 1795.1pg/ml, p-tau 148.6pg/ml.

Proband 3 is a woman who presented with memory loss at 47 years of age. She had a positive family history of dementia (Fig. ). She was a heterozygous mutation carrier of PSEN2 (NM_000447.2) Exon8 c.716T>C p.(Met239Thr). Neuroimaging showed symmetrical atrophy of the parietal lobe without temporal lobe atrophy (Fig. -). Her CSF biomarkers were Aβ42 675.1pg/ml, t-tau 892.3pg/ml, p-tau 43.5pg/ml.

DISCUSSION

The current definition of Alzheimer's disease relies heavily on biomarkers in CSF and brain imaging. Although genetic information was not included in the biomarkers’ profile summarized in the 2018 NIA-AA research framework, symptomatic autosomal dominant mutation carriers were considered to have AD neuropathologic change without the use of biomarkers [14]. In fact, it is generally accepted that a causative gene mutation is the strongest “biomarker.” Determination of genetic mutations was important in the clinical diagnosis of AD, especially in early-onset dementia cases. Next-generation sequencing was proved to be a swift, accurate, and cost-effective method for the identification of genetic mutations in clinical diagnosis [15], although the prompt interpretation of results is important. The mutation spectrum reported for AD-CGM showed variations according to different races and areas [16]. The majority of the mutations reported involve PSEN1, but other new gene mutations have been reported in different countries and also in China [8, 16]. The AD-CGM mutation rate in our general dementia cohort was around 3.6%, which was in concordance with previous reports. In all the 34 variants detected in the three genes, 23 variants were novel, classified as uncertain significance, or likely pathogenic according to ACMG guidelines. The mutation profile of AβPP, PSEN1, and PSEN2 in familial AD was studied in several Chinese reports. The detection rate of novel variants in these studies was around 27.5%-50% [8, 17, 18]. We enrolled more early-onset dementia patients without family history in the cohort, which might result in a relatively higher detection rate of novel variants. The frequencies of the variants in the general population were very low. Most of the novel variants were located in the hotspot region and next to pathogenic variants listed in AlzForum. They were worthy for further validation in our near future research.

The age of symptom onset varied from the third or fourth decade of life to the late 70s or 80s in AD-CGM. We confirmed previous reports that mutations in PSEN2 had significantly later onset than mutations in PSEN1 and AβPP, and mutations in PSEN1 had significantly earlier onset than all other groups [19]. We also found PSEN1 mutation carriers had the youngest age onset in our group.

We found that the clinical phenotype of AD-CGM associated with genetic mutations is extremely heterogeneous. Analysis of the psychological characteristics of the dominantly inherited Alzheimer network (DIAN) cohort concluded that the overall cognitive and personality deficits in mild AD-CGM were similar to sporadic AD, which often began with the hallmark deficit in episodic verbal memory as well as difficulties in some aspects of executive function and visuospatial abilities [20]. The patients in our cohort were mostly in the moderate stage, with a wide spectrum of cognitive domain involvement. Psychiatric/behavior syndrome is more common in the AD-CGM patients in our group, occurring in over half of the cases.

The genotype-phenotype relationship occurring in PSEN1 mutation carriers has been widely discussed in the literature [21-23]. A large pedigree of E280A PSEN1 mutation in Columbia showed that the most frequent presentation was memory loss followed by behavior and personality changes and progressive loss of language ability. In the final stages, gait disturbances, seizures, and myoclonus were frequent [24]. Later on, the AD phenotype of PSEN1 has proved to be very broad, including not only cognitive 
decline, behavioral and psychiatric symptoms of dementia but also parkinsonism, myoclonus and epileptic seizures, spastic paraparesis, features suggestive of the phenotype of frontotemporal dementia, progressive aphasia, and cerebellar ataxia [23, 25]. It was reported that atypical symptoms occurred more often in mutations after codon 200 of PSEN1 [4, 6]. Also, PSEN1 mutation carriers were significantly more likely to exhibit myoclonus, corticobulbar deficits, aphasia, and spasticity [26]. In our cohort, a non-amnestic presentation was also foundmore in PSEN1 mutation carriers than others, and language symptoms were most prominent. Also, spasticity/pyramidal sign was most common in PSEN1 mutation carriers in our cohort.

The phenotype of AβPP and PSEN2 mutations was relatively less complicated, with most of them presenting with amnestic symptoms, while behavior symptoms were more common in PSEN2 mutations [6, 26]. AβPP mutation carriers were significantly more likely to present with stroke, both ischemic and hemorrhagic [26]. Literature also proved that several AβPP mutations were associated with amyloidosis and caused significant cerebral amyloid angiopathy-like symptoms in AD-CGM [27]. In our cohort, AβPP mutation carriers had more frequent vascular lesions, which supported the underlying mechanism. PSEN2 mutation carriers experienced more behavior and psychiatric symptoms, especially agitation, than those with AβPP mutation.

The atrophy pattern on MRI was different in AD-CGM and sporadic AD. More neocortical loss, especially in the posterior area, was found in AD-CGM. Among different genes, there was evidence that AβPP subjects had smaller hippocampal volume than PSEN1 subjects; also, AβPP subjects had more medial temporal lobe atrophy, and conversely, PSEN1 subjects showed more neocortical loss [28]. We did not find atrophy pattern difference between AβPP and PSEN1 carriers in our group, probably because of the late disease stage; however, significant parietal lobe atrophy was found. White matter hyperintensities (WMH) were common in AD-CGM. Researchers found increased posterior WMH in the PSEN1 post codon 200 groups, and WMH correlated with the severity of cerebral amyloid angiopathy and cotton wool plaque pathology at post-mortem [29]. WMH was also prominent in our cohort of AD-CGM, but we found no difference among the three gene mutations. Some of our AD-CGM patients with WMH could easily be misdiagnosed as vascular dementia clinically.

The APOE ε4 allele is a well-known genetic risk factor for AD. In some reports, APOE ε4 allele carriers were more likely to develop AD at an earlier age than subjects without the ε4 allele in PSEN1 mutation AD-CGM, suggesting the APOE ε4 allele could also modify the age-onset and disease progression of AD-CGM [30]. However, we found no modifications induced by the APOE ε4 allele in our patients, in agreement with previous research [19]. In contrast, we found more ε4 carriers in the AβPP and PSEN2 groups than in the PSEN1 group.

In French AD-CGM cohort research works, CSF biomarkers profile was consistent with an AD diagnosis (cut-off: Aβ42<500pg/mL, t-tau>350pg/mL, and p-tau>60 pg/mL) in 90% of families carrying mutations on known genes, and all mutation carriers had abnormal CSF biomarkers, at least one of the three indexes (t-tau, p-tau, Aβ42) [31, 32]. The CSF biomarkers profile in our cohort was also consistent with AD, supporting AD pathologic changes and the pathogenicity of genetic variations.

There were some limitations in this study that need future work. Firstly, many of the variations found were of uncertain significance according to ACMG guidelines, and future functional validation would be needed. Secondly, the sample size was still not satisfying, although AD-CGM was rare. Multi-centric research would help in the future. Thirdly, future follow-up studies in our cohort should obtain neuropathology definition of the phenotypes.

CONCLUSION

In conclusion, we described the genotype and phenotype heterogeneity observed in the PUMCH-cohort. Early-onset was a common feature. PSEN1 mutation carriers had the youngest age of onset, followed by PSEN2 and AβPP mutation carriers. The most common symptoms were amnestic syndrome, followed by psychiatric symptoms and movement disorder, similar to sporadic AD. Parietal and posterior atrophy was more prominent in AD-CGM and especially in PSEN1 and PSEN2 mutation carriers, while AβPP mutation carriers had more vascular changes. The CSF biomarkers profile was indistinguishable from sporadic AD. Our recommendation is to screen for genetic mutations using NGS in atypical early-onset dementia cases.

Table 1

Demographic and basic information of APP, PSEN1, and PSEN2 mutation carriers.

-	APP	PSEN1	PSEN2	p
No. of patients	17	14	9	-
No. of variants	12	13	9	-
Sex (male: female)	9:8	7:7	4:5	0.919
Age of onset (+ S)Age range (yrs)	65.7 + 12.6(41-86)	55.4 + 11.4(41-74)	59.4 + 7.9(47-70)	0.051
Age of onset ≤65 yr	10/17 (58.8%)	11/14(78.6%)	7/9 (77.8%)	0.415
Time from onset to examination (yrs) (+ S)	2.76 + 1.39	2.64 + 2.24	3.78 + 1.86	0.311
Positive family history	8/17 (47.1%)	12/14 (85.7%)	7/9 (77.8%)	0.055
APOE ε4 carrier	7/17 (41.2%)	2/14 (14.3%)	5/9 (55.6%)	0.100
MMSE (+ S)	16.2 + 7.6	19.3 + 8.6	13.0 + 7.3	0.283
MoCA (+ S)	15.7 + 6.8	19.3 + 7.5	14.3 + 8.0	0.459
ADL (+ S)	37.8 + 11.7	30.8 + 7.6	35.7 + 11.1	0.240
CDR (+ S)	1.7 + 0.85	1.5 + 1.1	1.9 + 1.1	0.701

Table 2

Clinical phenotype and neuroimaging of APP, PSEN1, and PSEN2 mutation carriera.

-	Total	APP	PSEN1†	PSEN2	p
Symptoms:	-	-	-	-	-
Amnestic onset	24/40(60%)	11/17(64.7%)	7/14(50.0%)	6/9(66.7%)	0.635
Psychiatric/behavior	21/40(52.5%)	6/17(35.3%)	9/14(64.3%)	6/9(66.7%)	0.172
Irritability	7/40 (17.5%)	1/17 (5.9%)	6/14 (42.9%)	0/9 (0%)	0.008
Delusions	10/40(25.0%)	4/17 (23.5%)	2/14 (14.3%)	4/9 (44.4%)	0.260
Agitation	4/40 (10.0%)	1/17 (5.9%)	0/14 (0%)	3/9 (33.3%)	0.026
Depression	2/40 (5.0%)	0/17 (0%)	2/14 (14.3%)	0/9 (0%)	0.142
Language disorder	7/40(17.5%)	3/17(17.6%)	3/14(21.4%)	1/9(11.1%)	0.817
Movement disorder	9/40(22%)	4/17(23.5%)	3/14(21.4%)	2/9(22.2%)	0.990
Parkinsonism	4/40 (10.0%)	3/17 (17.6%)	0/14 (0%)	1/9 (11.1%)	0.263
Tremor	2/40 (5.0%)	1/17 (5.9%)	0/14 (0%)	1/9 (11.1%)	0.479
Spasticity/pyramidal sign	3/40 (7.5%)	0/17 (0%)	3/14 (21.4%)	0/9 (0%)	0.049
MRI Imaging:	-	-	-	-	-
Bilateral temporal atrophy	27/39(69.2%)	12/17(70.6%)	10/13(76.9%)	5/9(55.6%)	0.558
Bilateral parietal atrophy	28/39(71.8%)	10/17(58.8%)	10/13(76.9%)	8/9(88.9%)	0.237
Severe white matter hyperintensities	12/39(30.8%)	5/17(29.4%)	4/13(30.8%)	3/9(33.3%)	0.979
Vascular changes	6/39(15.4%)	5/17(29.4%)	1/13(7.8%)	0/9(0.0%)	0.091
Fazekas score (+ S)	-	1.1 + 1.0	0.8 + 1.2	1.1 + 1.0	0.822
Abnormal EEG‡	10/20(50.0%)	4/8(50.0%)	2/7(28.6%)	4/5(80.0%)	0.214

Note: †1 patient with PSEN1 mutation had no MRI imaging due to a metal implant.

‡ Some patients had EEG evaluation.