Clinical phenotype and genetic associations in autosomal dominant familial Alzheimer’s disease: a case series

Background The causes of phenotypic heterogeneity in familial Alzheimer’s disease with autosomal dominant inheritance are not well understood. We aimed to characterise clinical phenotypes and genetic associations with APP and PSEN1 mutations in symptomatic autosomal dominant familial Alzheimer’s disease (ADAD). Methods We retrospectively analysed genotypic and phenotypic data (age at symptom onset, initial cognitive or behavioural symptoms, and presence of myoclonus, seizures, pyramidal signs, extrapyramidal signs, and cerebellar signs) from all individuals with ADAD due to APP or PSEN1 mutations seen at the Dementia Research Centre in London, UK. We examined the frequency of presenting symptoms and additional neurological features, investigated associations with age at symptom onset, APOE genotype, and mutation position, and explored phenotypic diﬀ erences between APP and PSEN1 mutation carriers. The proportion of individuals presenting with various symptoms was analysed with descriptive statistics, stratiﬁ ed by mutation type. Findings and and APP mutations), and 121 mutations and 36 mutations). ed erent mutations six APP mutations novel) . diﬀ ered mutation, a younger onset individuals mutations APP mutations years group, ve PSEN1 presented with amnestic symptoms, making atypical cognitive presentations signiﬁ cantly more common in PSEN1 mutation carriers (n=14; p=0·037). Myoclonus and seizures were the most common additional neurological features; individuals with myoclonus (40 [47%] with PSEN1 mutations and 12 [33%] with APP mutations) were signiﬁ cantly more likely to develop seizures (p=0·001 for PSEN1 ; p=0·036 for APP ), which aﬀ ected around a quarter of the patients in each group (20 [24%] and nine [25%], respectively). A number of patients with PSEN1 mutations had pyramidal (21 [25%]), extrapyramidal (12 [14%]), or cerebellar (three [4%]) signs.


Introduction
Alzheimer's disease is the most common cause of dementia. In fewer than 1% of patients, Alzheimer's disease is caused by autosomal dominant mutations in the presenilin 1 (PSEN1), 1 presenilin 2 (PSEN2), 2 or amyloid precursor protein (APP) genes. 3 Autosomal dominant familial Alzheimer's disease (ADAD) is considered to be clinically similar to sporadic disease (with the exception of younger age at onset) and both are characterised by progressive impairment of episodic memory. Although atypical phenotypes are seen in both familial and sporadic Alzheimer's disease, [4][5][6] relatively little is known about the proportion of individuals with ADAD who present with atypical cognitive symptoms, the prevalence of additional neurological features, or the relationships between genotype, phenotype, and the pathophysiological mechanisms that might underlie them.
Prevention trials for ADAD are underway and have stimulated research into biomarker changes in preclinical Alzheimer's disease. However, these trials also necessitate better understanding of the natural history of Alzheimer's disease in the symptomatic phase and of factors that infl uence age at onset. A recent metaanalysis found that mutation type accounted for a large proportion of the variance in age at onset, but substantial variation was still observed between, and even within, families with the same mutation. 7 Some studies of families with APP, PSEN1, or PSEN2 mutations 8-10 have reported younger age at onset in APOE ε4 carriers, although this association was not evident in a 2014 metaanalysis. 7 Relatively little is known about the factors underlying variability in age at onset for diff erent mutations within a single gene, although PSEN1 mutations beyond codon 200 have been associated with a later onset, more severe amyloid angiopathy, and a greater burden of white matter hyperintensities on MRI than mutations before codon 200. 5,11,12 We aimed to analyse the clinical phenotype (initial cognitive symptoms and the frequency of additional neurological features) of a large cohort of individuals with ADAD; investigate potential associations with age at symptom onset, mutation position, and APOE ε genotype; and report the clinical and neuropathological features of the individuals with novel mutations.

Participants
Between July 1, 1987, and Oct 31, 2015, families with histories suggestive of ADAD were referred to the Dementia Research Centre (DRC) at University College London's Institute of Neurology (London, UK) from clinical and research centres across the UK and Ireland. We used clinical and genetic data from these families (11 with APP mutations, 55 with PSEN1 mutations) in this study. Five families with APP duplications have also been identifi ed, but are not included in the analyses presented here because data have been reported elsewhere. 13 We did not include individuals with sequence variants of questionable pathogenicity in this study.
Ethical approval for the study was provided by The National Hospital for Neurology and Neurosurgery and Institute of Neurology Joint Research Ethics Committee (subsequently the National Research Ethics Service Committee, London Queen Square). Written informed consent was obtained from all participants or from their guardian if cognitive impairment prohibited written informed consent.

Procedures
NSR evaluated contemporaneous records to determine age at symptom onset-defi ned as the age at which progressive symptoms of cognitive, behavioural, or motor changes were fi rst noticed by someone who knew the patient well-the initial cognitive or behavioural symptoms, and the presence of the following neurological features: myoclonus, seizures, pyramidal signs (such as

Research in context
Evidence before this study We searched PubMed for reports on the clinical phenotype of autosomal dominant familial Alzheimer's disease (ADAD) up to April 23, 2016, using the following search terms: "familial Alzheimer's disease", "autosomal dominant Alzheimer disease", "presenilin", "PSEN1", "PSEN2", and "APP", with no language restrictions. We identifi ed 200 publications reporting clinical information on individuals with ADAD, mostly from single pedigrees or small case series. We found 22 reviews of this literature, although the results of such reviews could potentially be subject to the publication bias caused by reporting atypical phenotypes more frequently than typical presentations. Therefore, while it is clear from the literature that atypical phenotypes occur in ADAD, less is known about the frequency of their occurrence, correlations between genotype and phenotype, and the pathophysiological mechanisms that might underlie them.

Added value of this study
We investigated the clinical phenotypes of ADAD in a large UK case series, including patient data collected since identifi cation of the fi rst mutation over 25 years ago. We ascertained the frequency of presenting cognitive symptoms and additional neurological features, and investigated their associations with age at symptom onset, APOE ε4 genotype, and mutation position. 44 diff erent mutations in the PSEN1 or APP genes were present in the cohort, including fi ve novel variants that are reported here for the fi rst time. We found clinically important phenotypic diff erences between patients with APP mutations and those with PSEN1 mutations. In addition to their younger age at symptom onset, PSEN1 mutation carriers more frequently presented with atypical cognitive symptoms and additional neurological features. Exploration of heterogeneity of clinical presentations between diff erent PSEN1 mutations suggested that mutation position might infl uence phenotype. Atypical cognitive presentations and spastic paraparesis were associated with PSEN1 mutations beyond codon 200, particularly involving exon 8. Conversely, particularly early ages at onset were observed for a cluster of mutations before codon 200 involving the fi rst hydrophilic loop of PSEN1.

Implications of all the available evidence
In describing the wide clinical spectrum of ADAD presentation, we highlight the importance for clinicians of considering genetic testing in young patients with dementia and additional neurological features, particularly when there is a family history of Alzheimer's disease or when the family history is not available. Appreciation of atypical ADAD phenotypes is important from a diagnostic perspective and might also off er insights into the mechanisms by which diff erent mutations cause disease. In view of the phenotypic heterogeneity that exists within ADAD, particularly between APP and PSEN1 mutation carriers, it could be informative to examine diff erent mutation types separately in observational studies and clinical trials of patients with ADAD. spastic paraparesis), extrapyramidal signs (such as rigidity), and cerebellar signs (such as ataxia). We classifi ed neurological features as early (≤5 years from symptom onset) or late (>5 years from onset). APOE ε4 status was determined by the Medical Research Council (MRC) Prion Unit (London, UK) using minor groove binding probe genotyping assays (TaqMan, Applied Biosystems).
Mutation analysis was carried out as described previously. 14 The likely pathogenicity of novel variants was predicted using a previously published algorithm, 15 and the tools PolyPhen (version 1.1.3) and PROVEAN (version 2). We assessed individuals with novel variants in PSEN1 or APP for the presence of additional mutations in other dementia-related genes using the MRC Dementia Gene Panel (appendix). 16 Where possible, when a novel sequence variant was found in the proband, other aff ected family members were genotyped by sequencing the relevant exon to demonstrate cosegregation between the mutation and disease.
Two individuals with novel variants underwent postmortem brain donation to the Queen Square Brain Bank at the UCL Institute of Neurology. We assessed amyloid β-positive plaque pathology using the Consortium to Establish a Registry for Alzheimer's Disease recommendations 17 and neurofi brillary tangle pathology with Braak staging. 18

Statistical analysis
We investigated diff erences in age at symptom onset between the APP and PSEN1 mutation groups, and between APOE ε4 carriers and non-carriers within each genetic group, using two-sample t tests. We analysed associations between age at onset and PSEN1 mutation using a linear mixed eff ects model with random eff ects for mutation and family. The intraclass correlation coeffi cient (ICC) was used to quantify the proportion of variance in age at onset explained by mutation, and by mutation and family. We analysed groups of individuals with APP and PSEN1 mutations separately to calculate the proportion of individuals presenting with amnestic symptoms or with atypical symptoms of behavioural change, language impairment, dyscalculia, or executive impairment; and the proportions with myoclonus, seizures, and pyramidal, extrapyramidal or cerebellar signs. We used two-sample t tests to investigate whether age at onset diff ered between individuals with typical Sex-specifi c information was not recorded during evaluation of patient medical histories. *The exon 9 deletion (NM_000021.3:c.869-1G→T; p.Ser290Cys; Thr291_Ser319del) is commonly referred to as ΔE9. †One patient had both Thr291Ala on exon 9 and Ala434Thr on exon 12. presentations and those with atypical presentations and between individuals with and without each additional neurological feature. Fisher's exact tests were used to investigate associations between atypical cognitive presentations or additional neurological features and APOE ε4 status, PSEN1 exon, and PSEN1 mutation location (compared with codon 200). We used a p value of less than 0·05 as our measure of statistical signifi cance. We used Stata version 12 for all analyses.

Role of the funding source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
Age at symptom onset was available for 213 individuals (168 with PSEN1 mutations and 45 with APP mutations; In patients with a PSEN1 mutation, age at onset was found to be infl uenced by the specifi c mutation, with 72% of the variance in age at onset explained by mutation (ICC 0·72). Mutation and family membership together explained 82% of the variance in age at onset (ICC 0·82). Individuals with mutations located before codon 200 had, on average, a younger age at onset (41·3 years [SD 7·2]) than did those with mutations beyond codon 200 (45·8 years [6·4], p<0·0001), which appeared to be driven by a younger age at onset for mutations involving exon 4 and 5 (fi gures 1, 2). Age at onset for a patient with two PSEN1 substitutions (p.Thr291Ala and p.Ala434Thr) was excluded from our analyses because it was unclear whether pathogenicity was due to one or both of these aminoacid substitutions. The intron 4 (NM_000021.3:c.338+1delG) mutation was classifi ed as involving exon 4 because it is located just outside this exon (appendix). 20 Detailed contemporaneous records documenting medical history and neurological examination fi ndings were available for 121 of 213 individuals (85 with PSEN1 mutations and 36 with APP mutations), and APOE ε4 genotype could be established for 101 of these individuals (71 PSEN1 and 30 APP). 35 of the 36 individuals with APP mutations presented with typical amnestic symptoms; the other patient presented with dyscalculia but developed memory problems soon after (table 2). Of the 85 individuals with PSEN1 mutations, 71 (84%) presented with amnestic symptoms and 14 (16%) with atypical cognitive presentations, which were more frequently associated with PSEN1 than APP mutations (p=0·037). Of the 14 PSEN1 mutation carriers with atypical initial cognitive features, seven (8%) presented with behavioural change, three (4%) with language impairment, two (2%) with dyscalculia, and two (2%) with a dysexecutive syndrome (table 2). The PSEN1 subgroup with atypical cognitive presentations had, on average, a somewhat older age at onset than those with typical amnestic symptoms (46·2 years [SD 5·9] vs 42·0 years [7·4], p=0·046). Prevalence of atypical cognitive presentations diff ered markedly between exons, occurring in ten (45%) of 22 individuals with exon 8 mutations, and fewer than 20% of individuals with mutations involving other exons (appendix). As a result, atypical presentations were signifi cantly more common in individuals whose mutation was located after codon 200 (p=0·006). There was no association between atypical cognitive symptoms and APOE ε4 status (data not shown).
In the APP group, myoclonus and seizures were the only additional neurological features observed, and the frequency of myoclonus and seizures did not diff er signifi cantly between the PSEN1 and APP groups. In the APP group, 12 (33%) carriers had myoclonus and nine (25%) developed seizures. Of the 12 individuals with myoclonus, onset of myoclonus was early (≤5 years from onset) in fi ve (42%), late in two (17%), and uncertain in fi ve (42%). Of the nine individuals with seizures, onset was early in three (33%), late in three (33%), and uncertain in three (33%). In the PSEN1 group, 40 (47%)  Of the 40 individuals with myoclonus, onset of myoclonus was early in 28 (70%), late in nine (23%), and uncertain in three (7%). Of the 20 individuals with seizures, onset was early in six (30%), late in 10 (50%), and uncertain in four (20%). Individuals with myoclonus were signifi cantly more likely to develop seizures: 16 (40%) developed seizures in the PSEN1 group (p=0·001) and six (50%) in the APP group (p=0·036). There was no association between myoclonus or seizures and age at onset or APOE ε4 status in either the APP or PSEN1 groups. There was no association between seizures or myoclonus and APOE ε4 status, exon or mutation location with respect to codon 200 in the PSEN1 group (appendix).
Pyramidal, extrapyramidal and cerebellar signs were only seen in patients with PSEN1 mutations (table 2, appendix). Pyramidal signs were observed in 21 (25%) of the 85 PSEN1 carriers. All of these individuals had spastic paraparesis, and 18 also had upper limb pyramidal signs. Of the 21 patients with pyramidal signs, 15 (71%) developed them early, although none were reported to have these signs before onset of cognitive symptoms. The remaining six (29%) patients developed them late, with an absence of pyramidal signs at earlier assessments. There were no associations between pyramidal signs and age at onset in the PSEN1 cohort as a whole, and insuffi cient numbers to investigate such associations at the level of individual families or mutations. Pyramidal signs were, however, observed more frequently in association with PSEN1 mutations after codon 200 than before codon 200 (p=0·024), with particularly high frequency (50%) in patients with mutations on exon 8 (appendix).
Extrapyramidal signs were observed in 12 (14%) of the 85 PSEN1 mutation carriers, occurring early in eight (67%), late in three (25%), and of uncertain onset in one (8%). No associations were found with age at onset, APOE ε4 status, or exon or PSEN1 mutation location (compared with codon 200). One of the patients with early extrapyramidal signs (PSEN1 p.Tyr115His) had markedly asymmetrical features consistent with a corticobasal syndrome (appendix).
Cerebellar signs were observed in three (4%) of the 85 PSEN1 mutation carriers, occurring early in two, and late in one. No associations were found with age at onset, APOE ε4 status, or exon or PSEN1 mutation location (compared with codon 200).
We identifi ed four novel mutations in PSEN1 and one novel mutation in APP. The novel variants in APP     16 All novel sequence variants identifi ed were absent from 100 healthy, white control participants who were unrelated to these patients. With the exception of p.Ser132Ala, which was seen in one individual of European ancestry, none of the variants were found in the ExAC dataset. We discuss the likely pathogenicity of the novel variants and provide further clinical and neuropathological details in the appendix.

Discussion
Clinically signifi cant diff erences are present between PSEN1 and APP mutation carriers with ADAD, emphasising the potential importance of examining these groups separately in observational research and clinical trials. In addition to the younger age at symptom onset for individuals with PSEN1 mutations than in those with APP mutations, PSEN1 mutation carriers more frequently presented with atypical cognitive symptoms and additional neurological features. Behavioural, language, and dysexecutive presentations, spastic paraparesis, and other pyramidal, extrapyramidal, and cerebellar signs were only seen in the individuals with PSEN1 mutations. By contrast, myoclonus and seizures aff ected a similar proportion of patients with APP and PSEN1 mutations. In both genetic groups, individuals with myoclonus were more likely to develop seizures than were those without myoclonus. These fi ndings highlight the need for clinicians to be vigilant of symptoms of seizure activity when myoclonus is present.
Limitations of our study are that some atypical phenotypes, such as movement disorder presentations or much older onset, might not have been seen in our case series due to our centre being more likely to get referrals for younger patients with cognitive symptoms. Also, not all patients were followed to advanced stages of illness, so late neurological features might be more frequent than we describe here. With enrolment of individuals over a long period of time, there is the potential for families to have greater awareness and therefore earlier recognition of symptom onset with successive generations. The relative non-diversity of individuals seen in a single country might also limit generalisability of the fi ndings. However, the mean age at onset in our cohort was very similar to that in a French case series 28 and in recent systematic reviews of ADAD. 5,7,28 As in our case series, spastic paraparesis and extrapyramidal and cerebellar signs were seen in French PSEN1 mutation carriers, but not APP mutation carriers, usually manifesting within 5 years of symptom onset. The proportion of French patients with PSEN1 mutations presenting with frontal symptoms (11%) was also similar to the combined proportion of individuals in our series whose initial cognitive symptoms were behavioural (8%) or dysexecutive (2%). 28 Finally, a non-amnestic presentation has been reported in 16% of individuals with ADAD worldwide; 5 which is the same proportion as in individuals with PSEN1 mutation carriers in our cohort.
While some of the PSEN1 mutation carriers in our study presented with non-amnestic cognitive symptoms, all but one of the APP mutation carriers had initial memory symptoms. These phenotypic diff erences have some support from neuroimaging studies: we have previously reported that APP mutation carriers have greater hippocampal atrophy than PSEN1 mutation  carriers of similar disease severity, whereas PSEN1 mutation carriers show more extensive neocortical atrophy and white matter involvement; the latter could underlie some of the atypical features observed in the PSEN1 group. 29 PSEN1 forms the catalytic subunit of gamma-secretase, which processes APP, but also a large number of other substrates involved in various physiological functions, including myelin repair and vascular and immune function. Gamma-secretase carries out an initial endopeptidase cleavage of its substrates, followed by successive carboxypeptidase-like cleavages. PSEN1 mutations all appear to decrease the effi ciency of this carboxypeptidase-like activity, resulting in the release of longer amyloid β peptides, which are more prone to aggregation. Most PSEN1 mutations also aff ect the endopeptidase activity, but to various degrees, potentially aff ecting the processing of other substrates in addition to APP. 30,31 We speculate that altered processing of other substrates could contribute to the atypical phenotypes witnessed in association with some PSEN1 mutations. Supporting this notion, atypical cognitive presentations and pyramidal signs in participants of this study were seen more frequently in association with PSEN1 mutations involving exon 8. The residues encoded by exon 8 lie within the hydrophilic sequence between transmembrane domains six and seven of PSEN1, which is where the cleavage site processed by autocatalytic activity resides. 32 Furthermore, patients with two atypical phenotypes-corticobasal syndrome (p.Tyr115His) or dementia with Lewy bodies (p.Ser132Ala) presentationshad mutations involving hydrophilic loop 1, which has been proposed to form the initial substrate binding site in PSEN1, with Ser132 playing a crucial role. 33 Indeed, the p.Tyr115His mutation has been found to reduce endopeptidase effi ciency due to substantially decreased affi nity for the Notch substrate, while the affi nity for APP is aff ected to a lesser extent. 30 Certain mutations might therefore diff erentially aff ect the substrate specifi city of the gamma-secretase complex and investigating whether this mechanism contributes to atypical clinical phenotypes is an important direction for future work. It was notable that the patient with the PSEN1 p.Ser132Ala mutation, who presented with a dementia with Lewy bodies phenotype, had severe neocortical Lewy body pathology. However, concomitant Lewy body pathology is a frequent fi nding in ADAD. 34 Large cohort studies will be important to further investigate clinical phenotype and clinicopathologic correlations in ADAD, ideally with unaff ected family members acting as controls.
Our results suggest that multiple factors could contribute to phenotypic heterogeneity in ADAD. There was sometimes considerable variability in the clinical features of individuals with the same mutation, even within a single family. Even so, we found that mutations before codon 200 were associated with younger age at onset, whereas mutations beyond codon 200 were more frequently associated with later ages at onset, atypical cognitive presentations, and pyramidal signs. Given the relatively small numbers of patients manifesting each atypical feature, and the numbers of associations (although not independent), it is important to be cautious about nominally signifi cant associations. Nonetheless, we felt it important to report them to allow replication in other cohorts. Indeed, a 2015 systematic review 5 also reported that PSEN1 mutations before codon 200 had younger ages at onset and were more frequently associated with seizures and myoclonus, whereas mutations beyond codon 200 were more frequently associated with spastic paraparesis. Codon 200 is an arbitrary cut-off , and mapping the mean ages at onset for diff erent mutations to the structure of PSEN1 (fi gure 2) suggests that there might be certain areas of the protein where mutations cause particularly early-onset disease, such as the fi rst hydrophilic loop encoded by exons four and fi ve. This extracellular loop contributes to a key allosteric core that changes amyloid β profi les through carboxypeptidase-like activity without aff ecting the endopeptidase function of gamma-secretase. 35,36 As qualitative changes in amyloid β profi les appear to underlie the pathogenicity of PSEN1 mutations, 30,31 one could speculate that these might be more dramatically altered by mutations involving this allosteric core, resulting in a more aggressive phenotype.
We have demonstrated that a subset of patients with ADAD do not have typical amnestic presentations. Because atypical presentations also occur in sporadic Alzheimer's disease, we do not think that our fi ndings challenge the idea that familial cases represent a paradigm for Alzheimer's disease, but rather highlight the importance of distinguishing and investigating atypical phenotypes to understand the complex underlying mechanisms that might contribute to disease. The clinical features of PSEN1-associated ADAD could erroneously suggest a diagnosis of frontotemporal or vascular dementia, corticobasal degeneration, or dementia with Lewy bodies. We suggest that it is important to consider ADAD in the diff erential diagnosis of patients with early-onset dementia with additional neurological features. ADAD detection rates have, at least historically, been lower than expected based on genetic epidemiology data, and have shown considerable variability across diff erent regions of the UK. 37 Failure to identify a mutation in these families might deprive the aff ected patient of a correct diagnosis and appropriate symptomatic treatment, and also has implications for the individual's family. Individuals at risk of ADAD should, if they wish, be given access to genetic counselling so that they can discuss their choices in a variety of areas, including predictive genetic testing and reproductive options such as preimplantation genetic diagnosis. They might benefi t from the peer support of connecting with other families aff ected by ADAD, 38 and from opportunities to participate in research, including preclinical treatment trials aiming to delay or prevent the onset of symptoms.
Contributors NSR, MNR, and NCF conceived of the study. NSR, PSJW, YL, MNR, and NCF contributed to recruitment and clinical assessment. JMN contributed to statistical analysis; TL, TR, and JH analysed the neuropathological data. RG, GA, JK, JB, LC-G, BdS, and SM contributed to genetic analysis or interpretation of the genotype and phenotype data. NSR drafted the initial version of the report, all authors contributed to revision and editing of the report.

Declaration of interests
NCF reports fees (all paid to University College London) for consultancy from Novartis, Sanofi , Roche/Genentech, and GlaxoSmithKline for contracted image analyses from Janssen Alzheimer's Immunotherapy, and for serving on a data monitoring committee from Aducanumab/ Biogen. MNR reports fees (paid to University College London) for serving on a data monitoring committee for Servier. BdS reports grants and consultancy fees from Janssen Pharmaceutica, consultancy fees from FORUM Pharmaceutica and reMYND. Additionally, BdS has a patent pending for presenilin defi cient multipotent cell lines and screening methods for g-secretase activities and modulators of g-secretase activities using these lines (EP 00200671.6), a patent pending for binding domains between presenilins and their substrates as targets for drug screening (EP 01201015.3), and a patent pending for peptides inhibiting specifi c cleaving activities of presenilins (EP 02078915.2). All other authors declare no competing interests.