Hereditary Dementias: Familial Alzheimer Disease and Frontotemporal Degeneration

Early-onset autosomal dominant Alzheimer disease (APP, PSEN1, PSEN2), APOE as a risk and modifier gene, and familial frontotemporal dementia within the FTD-ALS spectrum (C9orf72, GRN, MAPT and rare genes: CHMP2B, VCP, TBK1, TARDBP, SQSTM1, FUS, TREM2/TYROBP)

Literature and regulatory information updated: 9 October 2026.

0. Initial Reference Table

The article brings together two entities that share the adult genetics clinic for early-onset cognitive or behavioural decline: early-onset familial Alzheimer disease (AD), with autosomal dominant transmission, and familial frontotemporal dementia (FTD), whose pathological basis is frontotemporal lobar degeneration (FTLD). AD in Down syndrome, the motor component of the FTD-amyotrophic lateral sclerosis (ALS) spectrum, including C9orf72, TARDBP and tofersen, and Huntington disease and its phenocopies are not developed. Prevalence figures measure different things and are discussed in section 9.

1. Introduction

1.1. Dementia, early onset and the place of monogenic forms

The World Health Organization defines dementia as impairment of cognitive function frequently accompanied, and sometimes preceded, by deterioration in emotional control, social behaviour or motivation; it is generally progressive and irreversible, making early identification of treatable causes a priority [36]. Alzheimer disease is the most frequent neurodegenerative disease in older people and the most common cause of dementia from 40 years of age; it is followed by vascular dementia, dementia with Lewy bodies, frontotemporal dementia and Parkinson disease dementia [36]. Dementia beginning before 65 years has conventionally been termed early-onset [36]. In this age group, FTD is the second cause of neurodegenerative dementia after AD, and approximately 35–50% of patients have a family history [37, 38]. The proportion of AD explained by Mendelian genes is small: approximately 95% of all cases begin after 60–65 years and the early-onset familial form accounts for less than 2% of the total [17], whereas approximately one-third of FTD cases are genetic, mainly due to variants in GRN, MAPT and C9orf72 [24].

1.2. Milestones in molecular discovery in Alzheimer disease

The first causal variant identified was a substitution at codon 717 of APP, the “London” variant p.(Val717Ile), described by Goate and colleagues in 1991 as cosegregating with early-onset familial AD, in a setting where late-onset families showed no linkage to chromosome 21, thus demonstrating genetic heterogeneity [39]. In the same year, Crawford and colleagues confirmed that the codon 717 mutation was causal and showed that, in 14 families without it, there were no other variants in the Aβ domain of exons 16 and 17 [40]. In 1995, Sherrington and colleagues cloned the S182 gene, now PSEN1, at the AD3 locus on chromosome 14, responsible for the most aggressive form of familial AD, with five missense variants cosegregating with disease [41]; the same variants were not found in 118 patients with late-onset AD [42]. Also in 1995, Levy-Lahad and colleagues identified the STM2 gene (PSEN2), homologous to S182, on chromosome 1, with the p.(Asn141Ile) variant in families of Volga German origin [43], and Rogaev and colleagues described the same gene (E5-1) with missense variants in other early-onset families [44].

Apolipoprotein E entered AD genetics in 1993. Strittmatter and colleagues showed that ApoE is present in plaques, vascular amyloid and neurofibrillary tangles, binds β-amyloid peptide with high avidity, and that the ε4 allele is more frequent in late-onset familial AD [45]. In 42 families with late-onset AD, Corder and colleagues observed that disease risk rose from 20% to 90% and mean age at onset fell from 84 to 68 years with the number of ε4 alleles; the claim that ε4 homozygosity would be “virtually sufficient” to cause AD by 80 years refers to selected multiplex families and is not a population estimate [46]. In 2006, Rovelet-Lecrux and colleagues demonstrated that APP locus duplication (0.58–6.37 Mb) causes early-onset autosomal dominant AD with cerebral amyloid angiopathy in five families [47], and in 2012 Jonsson and colleagues described, from the genomes of 1795 Icelanders, the APP p.(Ala673Thr) variant, which protects against AD and age-associated cognitive decline [48]. In 2024, Fortea and colleagues proposed that APOE ε4 homozygosity should be considered a distinct genetic form of AD, a proposal that is not a consensus and differs from the revised Alzheimer's Association criteria, in which risk alleles are not included in staging [21, 49].

1.3. Milestones in molecular discovery in frontotemporal dementia

In 1998, Hutton and colleagues identified, in families with “frontotemporal dementia with parkinsonism linked to chromosome 17” (FTDP-17), historically grouped as Pick disease, MAPT missense variants such as p.Gly272Val and p.Pro301Leu (historical numbering in the 2N4R isoform, NM_005910.6; = NM_001377265.1:c.1991G>T p.(Gly664Val) and NM_001377265.1:c.2078C>T p.(Pro693Leu)), and exon 10 splice donor site variants [50]. The observation that MAPT mutations alone suffice to cause neurodegeneration with tau pathology, without amyloid, was incorporated by Hardy and colleagues into a model in which Aβ42 is central in AD and tau can be affected indirectly, by Aβ42, or directly, by MAPT mutations [51]. Several families linked to 17q21 nevertheless had no MAPT mutation and showed tau-negative, ubiquitin-positive inclusions; in 2006, Baker and colleagues and Cruts and colleagues showed that null variants in GRN, which encodes progranulin, explain these families (FTDU-17) [37, 38]. The large Danish family with “familial non-specific dementia” linked to the pericentromeric region of chromosome 3 (FTD3) was explained by the CHMP2B splice variant NM_014043.4:c.532-1G>C, described by Skibinski and colleagues in 2005 [6].

In 2004, Watts and colleagues identified six VCP missense variants in 13 families with inclusion body myopathy, Paget disease of bone and FTD (IBMPFD), present in all 61 affected individuals, with two North American founder haplotypes explaining approximately half the families [52]. In 2011, DeJesus-Hernandez and colleagues and Renton and colleagues discovered that the 9p21 FTD-ALS locus corresponds to a GGGGCC hexanucleotide expansion in a non-coding region of C9orf72 [53, 54]. In the DeJesus-Hernandez series, the expansion was the most common genetic cause of familial FTD (11.7%) and familial ALS (23.5%); in Finland, Renton and colleagues found it in 46.0% of familial ALS and 21.1% of sporadic ALS, with a shared founder haplotype [53, 54]. In 2015, Freischmidt and colleagues, through exome sequencing in 252 patients with familial ALS and 827 controls, and Cirulli and colleagues, through exome sequencing in 2869 patients with ALS and 6405 controls, established TBK1 haploinsufficiency as a cause of familial ALS and FTD and linked it to the autophagy and innate immunity pathways shared with OPTN and SQSTM1 [55, 56].

1.4. Nomenclature and reclassifications

In the now-retired gene-specific GeneReviews chapter, autosomal dominant AD forms were designated AD1 (APP), AD3 (PSEN1) and AD4 (PSEN2), with AD2 reserved for the APOE association; in one neuropathology paragraph, the same text substitutes AD1 for PSEN1, an inconsistency that should not be reproduced [1, 2]. EOFAD and ADAD overlap but are not synonymous: early-onset AD is defined by age (onset before 60–65 years, 1–5% of all AD), whereas autosomal dominant AD is defined by transmission, with disease in at least three people across two or more generations, two being first-degree relatives of the third [11]. Familial AD (at least two affected individuals, third-degree relatives or closer) accounts for 15–25% of cases and is largely non-Mendelian [11].

In FTD, the clinical syndrome (FTD, including the behavioural variant, bvFTD, and primary progressive aphasias, PPA) is distinguished from the neuropathological diagnosis (FTLD), subdivided according to the aggregated protein into FTLD-tau, FTLD-TDP, FTLD-FUS and FTLD-UPS [2, 4, 5]. The term FTDP-17 has been abandoned because MAPT variant carriers reproduce the sporadic FTLD-tau subtypes (Pick disease, corticobasal degeneration, progressive supranuclear palsy and globular glial tauopathy), depending on the variant [8]. The historical names FTDU-17, PGRN-FTD and HDDD (hereditary dysphasic disinhibition dementia) now designate GRN-FTD [4]. The abbreviation IBMPFD has been replaced by multisystem proteinopathy, encompassing VCP, HNRNPA1, HNRNPA2B1 and SQSTM1 [7]. In pedigrees of patients without another affected relative, “simplex” is preferred to “sporadic”, because absence of a family history does not exclude a genetic cause [3].

The 2025 French PNDS (National Diagnosis and Care Protocol) for the behavioural variant uses the term “behavioural variant of frontotemporal degeneration” and avoids “frontotemporal dementia” for the initial syndrome, because the term “dementia” presupposes a loss of autonomy that is absent in the early stages and because “FTD” denotes either the whole group of FTLD or only the behavioural variant, depending on the publication [34]. The same document summarises the sequence of criteria: the first consensus criteria of the Lund and Manchester groups in 1994, the separation between behavioural and language presentations in 1998, and the revised criteria of the International Behavioural Variant FTD Criteria Consortium (FTDC) of 2011, now the most widely used; the term “Pick disease”, which for a long time designated the whole spectrum, became reserved for a neuropathological subtype [34].

1.5. Definition of the entities covered by the article

The first entity is autosomal dominant AD due to a heterozygous pathogenic variant in APP (including locus duplication), PSEN1 or PSEN2, with a clinical and neuropathological phenotype overlapping that of sporadic AD, from which it is distinguished by age at onset and family history or molecular testing [17, 36]. The second is AD with a non-Mendelian genetic component, in which APOE ε4 is the main susceptibility factor, alongside approximately 20 small-effect genes; this article addresses it as context for counselling and pharmacogenetics [17]. The third is AD in trisomy 21, caused by lifelong APP overexpression [17]. On the frontotemporal side, the entities are C9orf72-FTD/ALS due to GGGGCC expansion, GRN-FTD due to progranulin haploinsufficiency, MAPT-FTD as a tauopathy, and rare forms due to CHMP2B, VCP and the other MSP genes, TBK1, TARDBP, FUS and other FTD-ALS spectrum genes [3–7]. Nasu-Hakola disease (biallelic TREM2/TYROBP), CLN11 neuronal ceroid lipofuscinosis (biallelic GRN) and genetic prion disease are addressed as differential diagnoses or allelic forms [2, 19].

2. Molecular Genetics and Pathophysiology

2.1. APP processing, γ-secretase and β-amyloid peptides

Amyloid precursor protein (APP) is a transmembrane protein of 695 to 770 amino acids, encoded by 19 exons, whose transcripts are named according to amino acid number; APP751 and APP770 predominate in peripheral tissues and APP770 corresponds to the reference transcript (MANE Select NM_000484.4) [1, 9]. The long isoforms contain a 57-amino-acid domain homologous to Kunitz-type protease inhibitors [1]. The β-amyloid peptide (Aβ), comprising 38 to 42 amino acids, is encoded by parts of exons 16 and 17 (codons 655–737) and lies partly within the transmembrane domain [1]. Cleavage by α-secretase occurs within Aβ and prevents its formation; sequential cleavage by β-secretase (BACE1) and γ-secretase releases Aβ, and the precise γ-secretase cleavage site determines the proportion of longer isoforms (Aβ42), which are more aggregation-prone and neurotoxic [1, 36].

Presenilin 1 is the catalytic subunit of γ-secretase, which cleaves APP, the Notch receptor and other substrates involved in Wnt signalling, cadherin-mediated adhesion and endoplasmic reticulum calcium homeostasis [2]. It is a 467-amino-acid protein with eight predicted transmembrane domains, homologous to SEL-12 in Caenorhabditis elegans; mice lacking presenilin 1 die in utero with skeletal abnormalities, reflecting its role in the Notch pathway [1]. Presenilin 2-containing γ-secretase preferentially acts on late endosomal and lysosomal substrates and generates the intracellular pool of long Aβ; the holoprotein also functions as an endoplasmic reticulum calcium leak channel [2]. Pathogenic PSEN1 and PSEN2 variants increase the proportion of Aβ42 relative to shorter forms and some reduce overall γ-secretase activity, leading to their interpretation as partial loss of function with a shift in product [1, 36]. In the DIAN (Dominantly Inherited Alzheimer Network) observational study, the degree of reduction in γ-secretase activity attributed to each variant correlated with age at onset and biomarker trajectories [2, 57].

2.2. APP: variants, gene dosage and amyloid angiopathy

Pathogenic APP missense variants cluster near the secretase cleavage sites or within the Aβ sequence and act through two mechanisms: increased production or proportion of Aβ42, when near the cleavage sites, and formation of a more fibrillogenic or vascular-tropic peptide, when within Aβ [2]. The most frequent variant is p.(Val717Ile) (“London”), near the γ-secretase site, and phenylalanine and glycine substitutions also occur at the same codon [1]. The “Swedish” variant, p.(Lys670_Met671delinsAsnLeu), near the β-secretase site, is historically described as c.2010_2011delinsTC and normalised by VariantValidator as a two-nucleotide inversion, c.2010_2011inv [1, 9]. Within Aβ, the “Flemish” variant p.(Ala692Gly) combines cerebral haemorrhage and presenile dementia; the “Arctic” variant NM_000484.4:c.2078A>G p.(Glu693Gly), described in a Swedish family, increases protofibril formation and is associated with marked congophilic angiopathy; and the “Dutch” variant p.(Glu693Gln) causes hereditary cerebral amyloid angiopathy with haemorrhages of the Dutch type, with infrequent dementia and plaques [1, 9]. The retired GeneReviews chapter assigns the designation “Dutch” to p.Glu693Gly, which is the Arctic variant; the correct assignment, confirmed during normalisation, is Dutch = c.2077G>C p.(Glu693Gln) [9].

Codon 673 illustrates how the same position can have opposite effects. The p.(Ala673Thr) variant, identified in Iceland, protects against AD and age-associated cognitive decline [48]. The p.(Ala673Val) variant, at the same codon near the BACE1 cleavage site, has been described as pathogenic only in homozygosity, with recessive inheritance, typical AD neuropathology and cerebellar Aβ [1]. APP-related hereditary cerebral amyloid angiopathy (OMIM 605714) includes the Dutch, Italian, Iowa, Flemish and Arctic variants and manifests as recurrent lobar haemorrhages, stroke, cognitive decline and vascular amyloid deposition [2].

APP locus duplication acts through gene dosage, as does trisomy 21 [17, 47]. In a French series of 43 carriers from 24 families, 90.2% had major neurocognitive disorder, 29.2% symptomatic intracerebral haemorrhage and 51.2% epileptic seizures; 14 of 19 met imaging criteria for cerebral amyloid angiopathy, autopsy showed severe diffuse vascular amyloid, and cerebrospinal fluid (CSF) Aβ42 was abnormal in 18 of 19 [26]. Locus triplication has also been described [2]. Duplication accounts for less than 1% of APP alterations and was not found in 141 Swedish and Finnish patients with early-onset AD, 75 of them familial [1]. Two practical consequences follow: sequencing alone does not detect the duplication, so testing must include copy-number analysis, and an APP finding warrants magnetic resonance imaging (MRI) with sequences sensitive to haemosiderin products even when the presentation is purely cognitive [2]. In trisomy 21, lifelong APP overexpression means that virtually everyone has the neuropathological lesions of AD after 40 years of age, and more than half show cognitive decline when assessed [17, 58].

2.3. PSEN1: gene, variant spectrum and correlations

PSEN1 has its coding region in ten exons (3 to 12), with alternative splicing of exon 8 and part of exon 3 [1]. It is the gene most frequently implicated in autosomal dominant AD, accounting for 20–70% of families with early-onset familial AD, depending on the definition of family used [17]. Most variants are missense; they cluster in the transmembrane domains, particularly TM2, and in the cytosolic loop between TM6 and TM7 [1]. The classic exception is in-frame loss of exon 9, shortening the protein by 29 amino acids, caused by splice variants or a 4555 bp genomic deletion including exon 9 that is a founder variant in Finland [1]. The variant lists in the retired chapter, with over 40 variants in over 50 families, are historically dated; current interpretation should rely on maintained databases such as Alzforum, which the donanemab appropriate use recommendations indicate for assessing the association of each variant with amyloid angiopathy [1, 31].

The genotype–phenotype correlations described are useful for recognising atypical presentations but not for predicting the individual course [11]. Exon 9 deletion is associated with early spastic paraparesis and “cotton wool” plaques (cotton wool plaques); NM_000021.4:c.338T>C p.(Leu113Pro) and p.(Val89Leu) (recorded in ClinVar as NM_000021.4:c.265G>T) with a frontotemporal-type behavioural presentation; c.1175T>C p.(Leu392Pro) and c.415A>G p.(Met139Val) with initial psychiatric symptoms; c.697A>G p.(Met233Val) and c.767A>C p.(Tyr256Ser) with very early onset, around 30 years of age, with Lewy body pathology; c.548G>T p.(Gly183Val) and c.436A>C p.(Met146Leu) with Pick pathology; c.509C>T p.(Ser170Phe) with Purkinje cell loss; and c.236C>T p.(Ala79Val) and c.806G>A p.(Arg269His) with later onset, between 50 and 70 years [1]. The p.(Glu280Ala) variant, known as “paisa”, underlies the largest known autosomal dominant AD kindred, in Antioquia, Colombia, with approximately 1200 carriers [59]. The p.(Ala431Glu) variant has been described as a founder variant in Mexico [1, 9]. Somatic mosaicism has been documented: in one family, the variant was present in the cerebral cortex but not in the lymphocytes of a deceased mother, whereas her daughter, with a different clinical presentation, carried it in lymphocytes [1]. Apolipoprotein E may modulate age at onset in carriers of PSEN1 and PSEN2 variants [1].

2.4. PSEN2: a rare gene with an interpretation problem

PSEN2 has 12 exons (ten coding) and encodes a 448-amino-acid protein with eight transmembrane domains [1]. It explains approximately 5% of early-onset familial AD [17]. The best-established founder variant is p.(Asn141Ile), c.422A>T, in families of Volga German ancestry in the United States and also in present-day Germany; Italian and Spanish families have also been described [1, 43]. Variant interpretation is the principal practical problem: much older literature considered variants pathogenic that subsequent evidence did not support, and in ClinVar, at the date of the curation consulted, PSEN2 had only ten single-gene pathogenic or likely pathogenic records (three from old submissions without criteria), 21 records with conflicting classifications and 52 classified as benign [2]. Exome sequencing of a German early-onset AD cohort found a substantial frequency of PSEN2 variants, including p.(Asn141Ile), reinforcing the need for segregation, population allele frequency and current criteria before assigning causality to a missense variant [2, 60]. PSEN2 variants have been described in frontal-variant AD and primary progressive aphasia presentations, so an FTD-like phenotype does not exclude the gene [2].

2.5. Penetrance and age at onset in autosomal dominant AD

Pathogenic PSEN1 and APP variants have been described as completely penetrant, meaning that carriers develop AD if they live a normal lifespan; PSEN2 penetrance has been estimated at approximately 95%, with rare carriers remaining unaffected beyond 80 years [1, 11]. Complete penetrance is not universal: a PSEN1 p.(His163Tyr) carrier followed prospectively for 22 years was asymptomatic, with no CSF or imaging abnormalities, at 65 years, whereas mean onset for that variant is 51 years (standard deviation 7) [61]. Anticipation has not been documented [1]. In a systematic review and meta-analysis of 387 pedigrees (3275 people, 1307 affected individuals with known age at onset, 174 variants, including the Colombian PSEN1 NM_000021.4:c.839A>C p.(Glu280Ala) kindred and the Volga families with PSEN2 NM_000447.3:c.422A>T p.(Asn141Ile)), individual age at onset correlated strongly with parental age at onset and with the mean per variant and per family, even after adjustment for APOE and sex [62]. This finding underpins the concept of “estimated years to onset” (EYO, estimated years to onset), used in DIAN studies to align carriers by predicted disease stage [62]. Nevertheless, age at onset can vary by more than 20 years within the same family [11].

2.6. APOE and other risk genes

ApoE is the principal apolipoprotein in the brain. The three common alleles differ at two positions in the mature protein: ε2 (Cys112/Cys158), ε3 (Cys112/Arg158) and ε4 (Arg112/Arg158), defined by polymorphisms rs429358 and rs7412 [2]. In HGVS (Human Genome Variation Society) nomenclature for the MANE transcript, which numbers the precursor including the signal peptide (an additional 18 amino acids), the ε4 allele corresponds to NM_000041.4:c.388T>C p.(Cys130Arg) and ε2 to c.526C>T p.(Arg176Cys); ε2 corresponds to the T allele of rs7412 on the reference strand, and the curation consulted stating “rs7412 C/C” for ε2 is incorrect [2, 9]. Approximately 20–25% of the general population have at least one ε4 allele and approximately 2% are homozygous; among patients with AD, 10–15% are homozygous [21, 63, 64].

The meta-analysis by Farrer and colleagues (5930 patients and 8607 controls) estimated, in people of European ancestry and compared with ε3/ε3, odds ratios (OR) of approximately 3.2 for ε3/ε4, 14.9 for ε4/ε4 and 0.6 for ε2/ε3; the association was weaker in African Americans and Hispanics and stronger in Japanese people (ε3/ε4 approximately 5.6; ε4/ε4 approximately 33.1), present from 40 to 90 years, attenuated after 70 and different between sexes [65]. These values cannot be transferred between ancestries [2, 65]. The effect mainly concerns age at onset: one ε4 allele advances disease by approximately 5–10 years and two alleles by approximately 10–20 years, leading to the formulation that APOE genotype predicts “when” rather than “whether” [36]. Approximately 42% of patients with AD have no ε4 allele [17]. The ε2 allele appears protective, but is associated with amyloid angiopathy-related haemorrhage and superficial siderosis, and ε2 homozygosity is the genetic basis of type III hyperlipoproteinaemia, which has low penetrance and requires a second factor [2, 31].

Fortea and colleagues analysed 3297 neuropathology donors (NACC, National Alzheimer's Coordinating Center) and 10,039 participants from five biomarker cohorts, including more than 500 ε4 homozygotes [21]. Almost all ε4 homozygotes had intermediate or high AD neuropathological change, compared with approximately half of ε3 homozygotes; biomarkers diverged from 55 years, by 65 almost all had abnormal CSF Aβ42 and 75% had positive amyloid PET (positron emission tomography), and by 80 years 88% were positive for all amyloid and tau biomarkers [21]. The 95% prediction interval for age at onset was 32 years, comparable to that of PSEN1 ADAD (33 years) and Down syndrome (32 years) [21]. The authors propose “semidominant” inheritance, recommend not grouping heterozygotes and homozygotes together, and draw attention to possible underestimation of penetrance through competing mortality and the potentially limited informativeness of family history; they acknowledge that the data predominantly concern European ancestry [21]. The 2024 revised Alzheimer's Association criteria retain APOE ε4 as a risk factor rather than a diagnostic or staging criterion [49].

The APOE3 Christchurch variant, p.(Arg154Ser) in precursor numbering (R136S in the mature protein), was found in homozygosity in a PSEN1 p.(Glu280Ala) carrier who remained cognitively unimpaired until her eighth decade, and is described as a protective modifier in section 10 [9, 59]. Beyond APOE, approximately 20 susceptibility genes (including ABCA7, BIN1, CD33, CLU, PICALM, SORL1 and TREM2) mildly increase risk, are not causal, should not be included in diagnostic panels, and polygenic scores have no known clinical utility [17]. TREM2 p.(Arg47His) (NM_018965.4:c.140G>A), present in heterozygosity in 0.5–1% of the general population, confers an OR of approximately 3.0 for late-onset AD [17, 66]. This heterozygous finding is a susceptibility factor and must not be confused with autosomal recessive Nasu-Hakola disease [2].

2.7. Normalisation of the variants cited

The table brings together HGVS nomenclature on MANE Select transcripts (GRCh38) and the historical designations of variants cited in the article. In APOE and MAPT, historical numbering systematically differs from HGVS: in APOE because of the 18-amino-acid signal peptide, and in MAPT because classical numbering uses the 441-amino-acid 2N4R isoform (NM_005910), whereas MANE corresponds to a 758-amino-acid isoform, with an offset of 392 codons [9, 10].

2.8. Frontotemporal lobar degeneration: aggregated proteins and genes

The neuropathological classification of FTLD is based on the protein forming the inclusions and corresponds directly to the genes. FTLD-tau, with tau deposits in neurones and glia, corresponds to MAPT variants; FTLD-TDP, with inclusions of the 43 kDa TAR DNA-binding protein (TDP-43), corresponds to C9orf72, GRN, VCP, TBK1, TARDBP and other genes; FTLD-FUS corresponds to FUS; and FTLD-UPS, with ubiquitin- and p62-positive inclusions negative for TDP-43, tau and FUS, corresponds to CHMP2B [2, 4, 5]. TDP-43 is a nuclear transcription and splicing factor that shuttles between nucleus and cytoplasm and binds RNA preferentially at UG-rich intronic regions; in FTLD-TDP it forms neuronal cytoplasmic and intranuclear inclusions and dystrophic neurites, with phosphorylation of serines 409/410 [4]. Clinical overlap is considerable, so the phenotype rarely indicates the gene, but neuropathology is discriminatory [4, 5].

2.9. C9orf72: GGGGCC expansion and gain- and loss-of-function mechanisms

C9orf72, at 9p21.2, encodes an endosomal trafficking regulatory protein that interacts with Rab proteins and participates in autophagy and endocytosis in a complex with SMCR8 [2, 3]. The only known pathogenic variant is expansion of the GGGGCC hexanucleotide in the intron between the alternative first exons; there is no evidence that coding sequence variants are pathogenic [3, 53, 54]. GeneReviews classifies alleles as normal (2–24 repeats), of uncertain significance (25–60, rare in the population and typically not segregating with disease) and pathogenic (61 to more than 4000), with reduced age-dependent penetrance [3]. The consensus guideline for ALS genetic testing emphasises that there is no validated threshold between pathogenic and non-pathogenic alleles and that evidence for alleles with 20–29 repeats is contradictory; ClinVar records expansion from more than 24 repeats without defining the intermediate range [10, 27]. The smallest expansion described as cosegregating in a family had 47 repeats in leucocytes, but the brain of the same patient contained a mixture of short and very long expansions (more than 1100), demonstrating somatic mosaicism [3].

The mechanism involves “multiple hits”. Repeat-containing RNA forms nuclear foci that sequester RNA-binding proteins; non-canonical repeat-associated translation (RAN translation), without an ATG codon and in both directions, produces five aggregating dipeptide repeats (poly-GA, poly-GP, poly-GR, poly-PA and poly-PR) that form inclusions; and expression of the expanded allele decreases, causing protein haploinsufficiency [3]. The repeat is 100% GC, highly unstable in tissues and flanked by repetitive sequences; its size in blood does not represent that in nervous tissue, varies between tissues within the same individual and throughout life, and leucocyte size has no prognostic value for disease presence, age at onset or severity [3]. During transmission, germline instability mainly manifests as contractions in paternal transmissions [3]. Anticipation has been described by some groups but not others, and may reflect ascertainment and recall biases and technical sizing difficulties [3, 24].

Penetrance depends on age: approximately 0% at 35 years, 50% at 58 years and almost 100% at 80 years, with patients in their 20s and asymptomatic heterozygotes in their 90s [3]. The ALS consensus guideline considers expansion penetrance much lower than that of some SOD1 and FUS variants and recommends discussing this uncertainty during counselling [27]. A single reported homozygote, born to consanguineous parents and carrying more than 800 repeats on both alleles, developed bvFTD at 43 years with rapid deterioration but within the usual spectrum; a compound heterozygote (approximately 50 and more than 2000 repeats) developed bvFTD with parkinsonism at 58 years [3].

2.10. GRN: progranulin haploinsufficiency

GRN, at 17q21.31, encodes progranulin, a secreted glycoprotein containing seven repeats of a 12-cysteine motif, from which granulin peptides are released by cleavage; it is expressed in rapidly renewing epithelia, the immune system, neurones and microglia [4]. Full-length progranulin is trophic and anti-inflammatory and the cleaved peptides are pro-inflammatory; its lysosomal role, including lysosomal trafficking and glucocerebrosidase activity, is increasingly recognised [4]. Almost all pathogenic variants are nonsense, frameshift or splice variants and lead to nonsense-mediated messenger RNA decay or loss of the allele, reducing circulating and central nervous system progranulin to approximately half; locus deletion produces the same presentation [2, 4]. GRN dosage defines the phenotype: heterozygosity causes GRN-FTD, with FTLD-TDP, and biallelic loss causes CLN11 neuronal ceroid lipofuscinosis, a lysosomal disease beginning in adolescence or young adulthood with retinal dystrophy and rapid visual loss, myoclonic seizures, ataxia with cerebellar atrophy and cognitive decline [2, 4, 67].

The most frequent variant is p.(Arg493Ter), on a haplotype suggesting a founder effect, with 60% of carriers affected by 60 years and more than 95% by 70 [4]. The second most frequently reported, p.(Ala9Asp), has a family history in only 25% of cases, suggesting reduced penetrance or de novo occurrence; the splice variant c.-8+5G>C is a founder variant in a large Belgian pedigree [4]. There are no clear correlations between variant and age at onset, duration or phenotype, and the same variant produces very different presentations [4]. Overall penetrance is approximately 90% by 75 years [4]. In a large French series, 3.2% of simplex FTD cases had a GRN variant [4].

2.11. MAPT: tau isoforms and tauopathy

MAPT encodes microtubule-associated protein tau, which promotes microtubule assembly and stabilisation [5]. In the adult brain, alternative splicing produces six isoforms through inclusion or exclusion of exons 2 and 3 in the N-terminal half and exon 10 in the C-terminal half; exon 10 exclusion generates isoforms with three microtubule-binding repeats (3R) and inclusion generates 4R isoforms, present in similar proportions in normal cortex [5]. Pathogenic variants act by altering the 3R/4R ratio, particularly those in exon 10 and the adjacent intron, such as c.915+16C>T (IVS10+16), affecting microtubule assembly, or by increasing tau aggregation propensity, as with missense variants such as p.Pro301Leu [5, 50]. Depending on the variant, inclusions are predominantly 4R, predominantly 3R or mixed, the latter with paired helical filaments resembling those in AD [5]. In a brain bank series, carriers had onset at 55 ± 4 years, compared with 70 ± 6 years in sporadic tauopathy, and neuropathology reproduced the sporadic subtypes according to variant [8].

Genotype–phenotype correlations are clearer than in GRN; the MAPT variants cited follow the historical numbering in the 2N4R isoform, NM_005910.6, with the MANE NM_001377265.1 equivalent in parentheses. p.Pro301Leu (c.2078C>T p.(Pro693Leu)) and c.915+16C>T (c.2091+16C>T) usually present as bvFTD, although a considerable minority of c.915+16C>T carriers have parkinsonism; p.Asn279Lys (c.2013T>G p.(Asn671Lys)) usually presents as a parkinsonian syndrome; p.Leu284Arg (c.2027T>G p.(Leu676Arg)), p.Asn296del (c.887_889del; c.2063_2065del p.(Asn688del)), p.Gly303Val (c.2084G>T p.(Gly695Val)), p.Lys317Met (c.2126A>T p.(Lys709Met)) and the splice-altering variants c.853A>C p.(Ser285Arg) (c.2029A>C p.(Ser677Arg)) and c.915T>C p.(Ser305=) (c.2091T>C p.(Ser697=)) may present with parkinsonism; and p.Val337Met (c.2185G>A p.(Val729Met)), p.Gln351Arg (c.2228A>G p.(Gln743Arg)) and p.Arg406Trp (c.2392C>T p.(Arg798Trp)) may begin with episodic memory impairment, with an AD-like phenotype [5]. Penetrance is considered complete, with occasional reduced penetrance in families with p.Leu315Arg (c.2120T>G p.(Leu707Arg)), p.Val363Ile (c.2263G>A p.(Val755Ile)) and p.Gly389Arg (c.2341G>A p.(Gly781Arg); also c.2341G>C in ClinVar) [5]. A partial deletion (exons 6–9) has been described in one patient and gene duplications in ten patients with FTD [5]. The MAPT H1 haplotype is associated with the risk of sporadic progressive supranuclear palsy, without utility as a predictive test [2].

2.12. Rare genes within the FTD-ALS spectrum

CHMP2B (3p11.2) encodes a component of the ESCRT-III complex, which carries out endosomal sorting of transmembrane proteins into multivesicular bodies that subsequently fuse with lysosomes [6]. The mechanism is gain of function: the C-terminally truncated protein accumulates on the endosomal membrane and impairs multivesicular body trafficking, autophagy and endosome–lysosome fusion [6]. In addition to the Danish founder variant c.532-1G>C, NM_014043.4:c.493C>T p.(Gln165Ter) has been described in a Belgian patient, NM_014043.4:c.532-2A>T in a Chinese patient and NM_014043.4:c.35-1G>A in Spanish monozygotic twins; there are no known allelic diseases [6]. The metabolic disease textbook consulted attributes an “FTD with cataracts” to CHMP2B and CHMP4B; FTD is associated with CHMP2B and cataract with CHMP4B, and the two entities must not be merged [68].

VCP (9p13.3) encodes the p97 protein, an ATPase that extracts ubiquitinated proteins and directs them towards degradation, including endoplasmic reticulum-associated degradation with UFD1 and NPLOC4; missense variants cause a proteostasis defect with TDP-43 proteinopathy [2]. MSP can result from heterozygous variants in VCP, HNRNPA1, HNRNPA2B1 or SQSTM1 (sequestosome-1/p62, a selective autophagy receptor), and approximately 5% of cases are de novo [7]. TBK1 (12q14.2) causes FTDALS4, with “definitive” gene–disease validity in ClinGen, through haploinsufficiency (null alleles or CCD2-domain missense variants abolishing optineurin binding); in 252 familial ALS pedigrees, 13 had loss-of-function variants, without enrichment in 1010 patients with sporadic ALS [2, 55]. Biallelic loss of TBK1 causes an autosomal recessive autoinflammatory disease with arthritis and vasculitis [2]. TARDBP (1p36.22) mainly causes ALS and, rarely, pure FTD; FUS, OPTN, UBQLN2 (X-linked), TIA1, TUBA4A, CCNF and CHCHD10 complete the spectrum [5, 18].

Biallelic loss of function of TREM2 or TYROBP causes polycystic lipomembranous osteodysplasia with sclerosing leukoencephalopathy (PLOSL, Nasu-Hakola disease), described in section 6 [2]. In consanguineous families, biallelic TREM2 variants were identified by exome sequencing in patients with an FTD-like syndrome without bone involvement [69]. TMEM106B is associated with FTD and with a hypomyelinating disorder and is described as a modifier; PRKAR1B-related neurodegenerative dementia with intermediate filaments is a very rare autosomal dominant entity (prevalence below 1/1,000,000) to consider in the differential diagnosis [2, 68]. The gene table of the French PNDS (National Diagnosis and Care Protocol) adds to the behavioural variant spectrum CSF1R (variable presentations with bvFTD, parkinsonism, dystonia, spastic paraparesis and epilepsy, in the context of leukoencephalopathy with axonal spheroids), DCTN1 (bvFTD, ALS or FTD-ALS and Perry syndrome), HNRNPA1 and HNRNPA2B1 (with Paget disease of bone and inclusion body myopathy), CHCHD10 (with mitochondrial myopathy) and TREM2 (bvFTD and Nasu-Hakola disease), and places FUS, CHMP2B, UBQLN2, ANG, OPTN, TAF15, MATR3, PFN1 and CCNF mainly in ALS, with occasional bvFTD [34].

2.13. Allelic diseases and associations requiring caution

PSEN1 and PSEN2 variants have been described in families with isolated dilated cardiomyopathy [1]. On 4 March 2026, the ClinGen dilated cardiomyopathy curation group classified the PSEN1 association with dilated cardiomyopathy 1U as “disputed”, and the PSEN2 association with cardiomyopathy 1V has only “limited” evidence; neither should be presented as an established cardiac entity [2]. PSEN1 has also been associated with familial acne inversa, with only four variants in a review of 57 γ-secretase gene variants, compared with 39 in NCSTN [2]. The 17q21.31 microdeletion including MAPT and KANSL1 causes Koolen-de Vries syndrome, attributed to KANSL1 loss rather than MAPT loss, as the curation consulted corrects in relation to the GeneReviews text [2, 5]. VCP is also associated with Charcot-Marie-Tooth disease type 2Y, hereditary spastic paraplegia and isolated familial ALS [2]. Biallelic GRN causes CLN11 [4].

3. Clinical Spectrum and Manifestations by System

3.1. Alzheimer disease: features shared by genetic and sporadic forms

AD usually begins with subtle impairment of memory for recent events, the mild cognitive impairment (MCI) stage, and progresses slowly towards disability, with confusion, loss of judgement, language disturbance, visual complaints, agitation, withdrawal and hallucinations; seizures, parkinsonism, hypertonia, myoclonus, incontinence and mutism occur less frequently [17]. Death generally results from starvation, malnutrition and pneumonia, and the typical clinical duration is 8 to 10 years, with a range of 1 to 25 years [17]. Familial and non-familial forms have the same clinical and neuropathological phenotype and, in an individual, there is no clinical or pathological means of distinguishing genetic from sporadic AD; they are distinguished by age at onset, family history and molecular testing [17, 36]. Autosomal dominant AD may have a long prodrome [17].

3.2. Differences between APP, PSEN1 and PSEN2

PSEN1 variants usually cause disease in the 40s or early 50s, with a range from 30 to the early 60s; onset after 65 years is rare and progression over 6–7 years is common [1, 17]. In the DIAN observational study, median symptomatic onset in 190 PSEN1 carriers was 44.5 years (interquartile range 40.6–51.4) [2, 62]. APP variants usually have onset between 40 and 50 years, occasionally in the 60s (range 30–65), with typical AD dementia and, in some families, neuronal Lewy bodies [1, 17]. PSEN2 variants have the broadest onset range, from 40 to 75 years, and a mean duration of 11 years [17]. Onset before 45 years may indicate faster progression [1].

Non-amnestic manifestations and neurological signs are more frequent than in sporadic AD, but their frequency depends heavily on how patients are ascertained. Tang and colleagues compared 107 symptomatic DIAN-OBS patients (United States, Europe and Australia, February 2008 to July 2014) with 1228 symptomatic patients from 188 publications, 753 of whom had a detailed neurological examination [25].

The authors attribute the higher frequencies in the literature to publication bias favouring atypical phenotypes; in DIAN-OBS, motor signs were more common with earlier onset, and ischaemic stroke was more prevalent with later onset (OR 1.09 per year) [2, 25]. The correlations described in section 2.3 (spastic paraparesis with exon 9 deletion, frontal, psychiatric or Lewy body presentations) apply to specific PSEN1 variants [1]. With APP duplications and intra-Aβ variants, the presentation combines cognitive decline, seizures and lobar haemorrhages due to amyloid angiopathy (section 2.2) [2, 26].

3.3. The presymptomatic phase and biomarker sequence

Studies in carriers have shown that disease begins long before symptoms. In DIAN, biomarker changes emerged 10 to 20 years before predicted onset, with amyloid deposition on PET, low Aβ42 and elevated tau in CSF [1, 70]. In presymptomatic PSEN1 carriers, CSF Aβ42 is low and Pittsburgh compound B (PiB) PET shows early striatal amyloid deposition [1]. In ε4 homozygotes, the sequence resembles ADAD and AD in Down syndrome: CSF Aβ42 diverges in the late 40s, amyloid burden in Centiloids rises before 50 (approximately 15 years before symptoms), and CSF and plasma phosphorylated tau rise in the early 50s (10 to 15 years before) [21]. This knowledge underpins stage 0 of the 2024 criteria, described in section 5.1 [49].

3.4. APOE ε4 homozygosity and Down syndrome

In the neuropathology cohort of Fortea and colleagues, ε4 homozygotes had symptom onset at 65.6 years, MCI at 71.8, dementia at 73.6 and death at 77.2 years, approximately 7 to 10 years earlier than ε3 homozygotes; hippocampal volume was lower at all ages, and at the dementia stage there were no differences in amyloid or tau PET between haplotypes [21]. The lifetime risk of AD dementia in ε4 homozygotes may reach 60% by 85 years, and some studies report 60–80% [21]. Homozygosity is not fully penetrant for clinical dementia [64]. In Down syndrome, virtually everyone has sufficient neuropathological change for AD by the mid-40s, and age at onset and mortality are consistent with complete genetic penetrance [49, 58].

3.5. Frontotemporal dementia: clinical syndromes

FTD affects the frontal and temporal lobes and manifests as changes in behaviour, executive function and language; when the parietal cortex and basal ganglia are affected, parkinsonism, corticobasal syndrome and memory deficits are added [4]. The behavioural variant (bvFTD) is the most common presentation: distractibility, loss of initiative, apathy, lack of interest, neglect of hygiene, social disinhibition, impulsivity and compulsivity, mannerisms and food craving [4]. Loss of judgement and insight leads to disastrous financial decisions, abruptly leaving employment and rudeness towards strangers, or, at the apathetic pole, withdrawal and abandonment of hobbies; initial misdiagnosis as depression, mania or psychosis is common [4]. Primary progressive aphasias are divided into non-fluent/agrammatic (nfvPPA), semantic (svPPA) and logopenic variants, the last being more closely associated with AD [4, 71]. Atypical parkinsonism may meet criteria for corticobasal syndrome or progressive supranuclear palsy (PSP), and motor neurone disease is particularly associated with C9orf72 [3, 5].

The French PNDS (National Diagnosis and Care Protocol), which addresses the behavioural variant without distinguishing sporadic from genetic forms at the outset, places it among the FTLD syndromes (with nfvPPA, the temporal variants, PSP, corticobasal syndrome and the forms with ALS) and attributes about 80–90% of cases to tauopathies and TDP-43 proteinopathies, with about 30% of cases being familial or genetic [34]. The most frequent initial complaints are apathy with emotional blunting without depressed mood, irritability, loss of insight and disinhibition (impulsive acts, excessive spending, reckless driving, over-familiarity with strangers, coarse language, neglect of personal hygiene), followed by repetitive behaviours, unusual over-invested interests, loss of empathy and eating changes with rapid weight gain [34]. Anosognosia is usual and the burden falls mainly on those close to the patient [34]. After 50 years of age and without a psychiatric history, treatment-resistant depression, neglect with hoarding of objects (Diogenes syndrome), excessive fear of illness, unusual addictions and a first manic episode should raise the hypothesis; occupational burnout may be the mode of presentation [34]. The cognitive profile is dominated by impaired social cognition, with executive dysfunction that is usual but sometimes late; episodic memory is impaired in about half of patients, and visuospatial abilities are more often preserved [34].

The neurological examination is usually normal at onset. Parkinsonism, mainly akinetic-rigid and with falls, affects up to 50% of patients with FTLD, particularly in genetic forms, and tremor is rare; limitation of vertical gaze or slowing of saccades point to forms bordering on PSP, and unilateral dystonia, myoclonus, apraxia and the alien hand phenomenon to forms bordering on corticobasal syndrome; motor neurone signs reflect associated ALS, present in 5–15% of bvFTD [34]. The complications to prevent are falls, weight gain (with decompensation or unmasking of diabetes), dysphagia with risk of aspiration pneumonia and of ingestion of non-edible objects, constipation, incontinence and sleep disorders (insomnia, sleep apnoea, daytime sleepiness) [34]. About 10% of patients with bvFTD reportedly have associated bipolar disorder, and C9orf72 carriers have more psychotic, obsessive–compulsive and mood symptoms [34].

3.6. C9orf72

Age at onset ranges from 20 to 91 years, with a mean of 58 years and usually between 50 and 64 [3, 24]. In an international cohort, carriers' clinical diagnoses were distributed as follows: FTD in 34.8% (bvFTD 31.4%, nfvPPA 1.8%, svPPA 0.9%, other clinical tauopathies 0.7%), ALS in 19.3%, FTD with ALS in 11.0%, and atypical presentations mimicking other diseases in 35.0%, such as AD, Parkinson disease, Huntington disease, dementia with Lewy bodies and vascular dementia [3].

The most characteristic feature is neuropsychiatric symptoms: delusions, often bizarre, predominantly visual hallucinations, and psychosis, which may be the initial symptom [3]. Almost all patients have executive dysfunction; common features (more than one-third) are recent memory impairment, language changes, apraxia and dyscalculia, disinhibition, apathy, anxiety, repetitive or compulsive behaviours mimicking obsessive–compulsive disorder, upper and lower motor neurone signs and parkinsonism, whereas preference for sweet foods, dysarthria and dysphagia are less frequent [3]. In C9orf72-associated aphasia, apraxia of speech and grammatical errors predominate [3]. In C9orf72-associated ALS, spinal onset is more frequent than bulbar onset (54% versus 39%) and some early cognitive impairment occurs even in apparently pure ALS [3]. The expansion is present in fewer than 1% of patients with a clinical diagnosis of AD (most with FTLD at autopsy) or Parkinson disease, in 6.5% of depressive pseudodementia diagnoses and in 2% of Huntington disease phenocopies without an HTT expansion [3]. The most frequent extrapyramidal syndrome is akinetic-rigid and symmetric [3].

3.7. GRN

Onset ranges from 35 to 87 years, with a mean of 64.9 ± 11.3 years in GeneReviews series and 61.3 years in the international cohort [4, 24]. Neuropsychology may show frontal or language deficits before overt dementia [4]. bvFTD is the most common initial presentation, followed by progressive aphasia, particularly non-fluent; in one series, 82% of patients eventually developed language changes [4]. Pure semantic dementia is rare and the logopenic variant was not associated with GRN [4]. Parkinsonism predominates in several families, sometimes initially diagnosed as corticobasal syndrome (rigidity, bradykinesia, limb dystonia, apraxia, imbalance, later myoclonus, dysarthria and dysphagia), and most eventually lose ambulation [4]. Some patients present with amnestic MCI, which may be confused with AD [4]. Motor neurone disease is rare or absent [4]. Marked asymmetric atrophy with parietal involvement is a characteristic imaging sign (section 5.4) [4, 5].

3.8. MAPT

In the international cohort, MAPT-FTD began at a mean of 49.5 years (standard deviation 10.0; range 17–82), earlier than C9orf72 and GRN [24]. Most patients have bvFTD and a minority have a parkinsonian syndrome (PSP, corticobasal syndrome or Parkinson disease); fewer than 5% present with aphasia or AD [5]. Parkinsonism may precede or follow behavioural changes; with progression, almost all have both, and most do not respond significantly to levodopa [5]. Episodic memory impairment, unusual in FTD, is frequent in MAPT-FTD, alone or with behavioural changes, and may lead to an initial diagnosis of AD [5]. Semantic impairment often accompanies the behavioural presentation, and initial svPPA or nfvPPA is rare [5]. The disease ends in relatively global dementia, sometimes with mutism, and progression leads to wheelchair dependence or being bedridden [5]. Some patients first present to psychiatry with a schizophrenia-like presentation [5].

3.9. Rare genes

In CHMP2B-FTD, described predominantly in the large Danish family from Jutland, onset is between 46 and 70 years (mean 58), with subtle personality changes, disinhibition or loss of initiative, apathy mimicking depression, stereotyped behaviours, hyperorality with sweet cravings, chain smoking, wandering and ritualistic routines [6, 72]. Non-fluent aphasia progresses to mutism, memory and visuospatial abilities remain relatively preserved until late, and after at least four years an asymmetric akinetic-rigid syndrome emerges with dystonia and pyramidal signs, partly related to neuroleptics [6, 72]. Fasciculations occur without severe motor neurone disease, institutionalisation occurs at a mean of 65 years, death at 68, and carriers lose an average of 13 years of life expectancy [72].

In VCP-associated MSP, the cross-sectional study by Al-Obeidi and colleagues (231 people from 36 families) found myopathy in 90% (mean age at diagnosis 43 years), Paget disease of bone in 42% (mean 41 years) and FTD in 30% (mean 56 years; range 30–86) [73]. Variability is considerable: all three manifestations coexist in 10% of patients, two in 50%, and they occur in isolation as myopathy in 37%, Paget disease in 5% and FTD in 3% [7]. Myopathy is proximal and distal, of limb-girdle type, with normal or slightly elevated creatine kinase (mean 195 U/L; range 40–1145); Paget disease occurs early, with mean alkaline phosphatase of 359 U/L (reference 30–130) [2, 73]. MSP-related FTD begins with dysnomia, dyscalculia, impaired comprehension and paraphasias, with relatively spared episodic memory, and progresses to mutism, alexia and agraphia [7]. In one series, electromyography was abnormal in 17 of 17 patients and there was motor neurone involvement in 11 (65%); approximately 10% had previously been diagnosed with ALS [2]. The spectrum includes dilated cardiomyopathy and levodopa-responsive parkinsonism, and death usually occurs in the sixth or seventh decade from respiratory failure [2, 7].

Pure ALS predominates in TARDBP-ALS-FTD; pure FTD is rare, ALS duration is 3–5 years and FTD duration 1–16 years [18].

3.10. Comparison of natural history in genetic FTD

Moore and colleagues retrospectively assembled 3403 carriers from 1492 families (C9orf72 1433 from 755 families; GRN 1179 from 483 families, with 130 variants; MAPT 791 from 254 families, with 67 variants) from FTD Prevention Initiative centres and the literature [24].

Individual age at onset and at death correlated more strongly with that of relatives in MAPT carriers (death: r=0.58 with parental age and 0.69 with the family mean) than in C9orf72 carriers [24]. The study is retrospective and subject to publication bias, and the data concern disease managed with supportive treatment, without disease-modifying therapies [24]. None of the sources consulted describes a treatment-induced change in the natural history of these forms; the existing differences between treated and untreated disease concern respiratory and nutritional support for the motor component [3].

4. Clinical Recognition and Warning Signs

4.1. When to suspect Alzheimer disease with a genetic cause

The two most important indicators are age at onset and a family history of dementia. There is no fixed threshold: the earlier the onset, the more likely a genetic cause, being more likely below 50 years and unlikely above 70, and having three or more relatives affected by early-onset dementia increases the probability [17]. The joint guideline of the American College of Medical Genetics (ACMG) and National Society of Genetic Counselors (NSGC) defines an autosomal dominant AD family as one with disease in at least three people across two or more generations, two being first-degree relatives of the third, and notes that small families, premature deaths from other causes and, rarely, incomplete penetrance can mask the pattern [11]. Approximately 47% of early-onset AD cases show familial aggregation and approximately 40% are apparently sporadic, some of which may represent concealed dominant transmission [11]. In simplex cases, the probability of finding a variant was approximately 6% in a Spanish series and increases with earlier onset, particularly below 50 years [1].

The genetics history should reconstruct a three-generation pedigree recording affected individuals, ages at onset, rate of progression, and age and cause of death, and investigate gradual (AD) or stepwise (vascular) progression, trauma, infection, alcohol, seizures, movement disorders and ischaemic events suggesting CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy) [36]. The ACMG/NSGC guideline additionally includes age at onset of any neurological or psychiatric symptoms, type of dementia and diagnostic method, and recommends confirming diagnoses through medical records [11]. An individual diagnosed with a psychiatric disorder or Parkinson disease in an earlier generation may represent an unrecognised affected individual [5]. The affected person should undergo assessment by a neurologist or psychiatrist experienced in dementia to identify treatable causes before counselling relatives [36].

4.2. When to suspect genetic frontotemporal dementia

The C9orf72 expansion should be considered in bvFTD with psychosis, bizarre delusions or hallucinations in a young or middle-aged adult, FTD with signs of motor neurone disease, a Huntington disease phenocopy, depressive pseudodementia and a psychiatric patient with progressive decline [3]. A family history of ALS, FTD or other spectrum manifestations across several generations strengthens suspicion, but its absence does not exclude the diagnosis, because C9orf72 is also the most frequent genetic cause in simplex cases [3, 74]. The international FTD neuropsychiatry consortium recommends testing for the C9orf72 expansion in all patients with possible or probable bvFTD and in suspected cases with prominent psychiatric features [75].

GRN-FTD should be considered in bvFTD, non-fluent aphasia, atypical parkinsonism or corticobasal syndrome with marked asymmetric atrophy, in amnestic MCI in a family with FTD and in new-onset psychiatric symptoms in middle age; variability within the same family is the rule [4]. MAPT-FTD should be considered in bvFTD or parkinsonism (corticobasal syndrome or PSP) beginning between 30 and 60 years, an amnestic presentation in a family with FTD, and a dominant family history with different diagnoses among relatives, such as bvFTD in one, parkinsonism in another, and “dementia”, AD or Parkinson disease in earlier generations [5]. The combination of frontal dementia with proximal and distal muscle weakness, bone pain, long-bone deformity or raised alkaline phosphatase suggests VCP-associated MSP [7, 52]. Up to 50% of patients with FTD have a family history of dementia, often labelled AD [4]. The French PNDS (National Diagnosis and Care Protocol) recommends considering a genetic cause in all patients with bvFTD, with or without ALS, and offering them testing in a dedicated clinic after the clinical diagnosis has been disclosed; the three-generation pedigree should look for dementia (including AD), ALS, atypical parkinsonism, psychiatric disease and Paget disease of bone [34]. About 30% of patients have a family history of bvFTD, of another syndrome of the spectrum or of ALS; the C9orf72 expansion and GRN loss-of-function variants explain about 50–70% of familial forms, and the known genes also explain about 10% of apparently sporadic cases [34].

4.3. Warning signs and emergencies

Progression over weeks or a few months, with ataxia and myoclonus, should raise suspicion of prion disease and requires prompt reassessment [19]. Lobar haemorrhages, superficial siderosis or seizures in a patient with early-onset cognitive decline suggest hereditary amyloid angiopathy or APP duplication [2, 26]. In a patient receiving an anti-amyloid antibody, headache, confusion, visual disturbances, dizziness, nausea, gait disturbance, seizures, focal deficits or encephalopathy are signs of amyloid-related imaging abnormalities (ARIA, amyloid-related imaging abnormalities) and require urgent MRI; ARIA with oedema may mimic an ischaemic stroke and should be considered before any thrombolysis [28, 63]. Dysphagia with frequent choking, weight loss and signs of respiratory insufficiency in a C9orf72 or VCP carrier indicate motor neurone or muscle involvement and require nutritional and respiratory assessment [3, 7]. Impaired judgement concerning driving, financial decisions or risk-taking behaviour requires a safety assessment [3, 4]. A subacute behavioural decompensation requires first a search for a non-verbalised somatic, iatrogenic or environmental cause, before any psychotropic drug (section 7.13) [34].

4.4. Newborn and population screening

The sources consulted do not describe newborn or population screening programmes for these adult-onset diseases, and the ACMG/NSGC guideline recommends against testing minors (section 8). Direct-to-consumer APOE testing is not advised [11].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Alzheimer disease: from clinical to biological diagnosis

The historical reference standard is clinicopathological: β-amyloid plaques, intraneuronal tau neurofibrillary tangles and amyloid angiopathy exceeding that expected for age, with frequent concomitant Lewy body pathology in the amygdala and TDP-43 pathology [17]. Earlier clinical criteria (slowly progressive adult-onset dementia, exclusion of other causes and cortical atrophy on neuroimaging) were correct in approximately 80–90% of cases [1, 17]. The revised 2024 Alzheimer's Association criteria define AD biologically: an abnormal “Core 1” biomarker result (amyloid PET, approved CSF biomarkers or accurate plasma biomarkers, particularly tau phosphorylated at threonine 217, p-tau217) is sufficient to establish the diagnosis of AD and support clinical decisions throughout the continuum [49]. The same criteria introduced stage 0, corresponding to genetically determined AD (ADAD or Down syndrome-associated AD) in an asymptomatic person with negative biomarkers, who “has the disease from birth” and progresses to stage 1 when a Core 1 biomarker becomes positive [49]. Risk alleles, including APOE ε4, are not included in staging [49]. This position differs from the proposal by Fortea and colleagues to treat ε4 homozygosity as a genetic form of the disease [21, 49].

The 2024 European intersocietal recommendations, developed by 22 experts from 11 societies using a Delphi method, propose a diagnostic pathway in memory clinics starting from 11 clinical syndromes, defined through history, examination, neuropsychology, laboratory tests, structural imaging and sometimes electroencephalography, and select first- and second-line biomarkers according to the patient's profile; the full text was inaccessible, so specific recommendations on genetic testing are not described here [76].

5.2. Biomarkers and what they measure

Biomarkers of AD pathology (CSF Aβ42/40 ratio, p-tau181 and p-tau217, amyloid PET, fluorodeoxyglucose PET, FDG, and hippocampal volume on MRI) demonstrate the disease, but not its genetic cause: a PSEN1 variant carrier with positive biomarkers has AD, without proving that the variant is causative, and an asymptomatic carrier may have normal biomarkers at ages when disease would be expected [2, 61]. Demonstration of amyloid pathology by PET or CSF is mandatory before starting anti-amyloid antibodies [28, 63]. The 2023 US appropriate use recommendations for lecanemab considered blood biomarkers not yet suitable for selecting patients, although their high negative predictive value helps exclusion; the 2026 UK recommendations likewise do not accept plasma p-tau217 as a substitute for CSF or PET for eligibility [63, 64]. In genetic FTD, structural MRI abnormalities and FDG-PET hypometabolism are detectable approximately ten years before symptoms, CSF or blood neurofilament light chain (NfL) is a non-specific marker of neurodegeneration, and tau PET has no clear value [2, 77]. Serum or CSF NfL may help distinguish bvFTD from a primary psychiatric disorder, and a normal FDG-PET helps exclude bvFTD [75].

For the behavioural variant, the French PNDS (National Diagnosis and Care Protocol) states that there is no validated aetiological biomarker of FTLD, with the exception of blood progranulin for GRN forms [34]. As first line, the laboratory panel of the HAS recommendations for cognitive decline (full blood count, C-reactive protein, sodium, calcium, blood glucose, albumin, creatinine and clearance, TSH, vitamin B12, folate, liver function tests and syphilis serology with treponemal and non-treponemal tests, HIV and Lyme disease depending on the context) is performed more systematically than when AD is suspected, because alcohol, tertiary syphilis and HIV encephalitis can present with behavioural changes [34]. If lumbar puncture is feasible and ethically acceptable, the PNDS recommends CSF AD biomarkers (Aβ1-42 and Aβ1-40, p-tau and total tau, ATN profile): an A−T−N− profile does not suggest AD, and an A+T+N+ profile is compatible with AD but does not exclude co-pathology; plasma p-tau217 is not yet validated for this distinction [34]. Blood or CSF NfL distinguishes bvFTD from primary psychiatric disease with sensitivity and specificity above 80%, is highest in bvFTD with ALS, does not distinguish between neurodegenerative diseases and, in France, is not yet included in the nomenclature or in the HAS recommendations, so its use is limited to specialist centres [34]. Higher serum levels are associated with shorter survival and greater atrophy, and in asymptomatic C9orf72, GRN or MAPT carriers NfL rises during the phase of conversion to symptomatic disease [34]. CSF poly(GP) dipeptides for C9orf72 carriers, CSF phosphorylated TDP-43 and real-time quaking-induced conversion assays (RT-QuIC) for TDP-43 and tau remain under investigation [34].

5.3. Clinical criteria for frontotemporal dementia

There are no specific clinical criteria for genetic forms; syndrome criteria are used [5]. The revised international bvFTD criteria of Rascovsky and colleagues have three levels of certainty and exclusion criteria [78].

In validation involving 137 autopsy-confirmed FTLD cases, 118 (86%) met possible bvFTD criteria and 104 (76%) probable bvFTD criteria, compared with 72 (53%) using the earlier criteria; patients who did not meet the criteria were older and predominantly had atypical presentations with marked memory impairment, which matters in MAPT-FTD and GRN-FTD, where amnestic presentations are not uncommon [5, 78]. A known pathogenic variant therefore advances the diagnosis to “definite FTLD”, giving genetic testing a diagnostic role beyond counselling [78].

Primary progressive aphasia is diagnosed in two steps [4, 71]. First, PPA is confirmed using Mesulam's criteria: language difficulty is the most prominent feature and the principal cause of impaired activities of daily living, aphasia predominates at onset and in the early stages, and the pattern is not explained by other diseases [4]. The variant is then classified according to Gorno-Tempini and colleagues [71]. The non-fluent/agrammatic variant requires at least one core criterion (agrammatism in production; effortful, hesitant speech with inconsistent errors and distortions, i.e. apraxia of speech) and at least two of three supporting criteria (impaired comprehension of syntactically complex sentences; preserved single-word comprehension; preserved object knowledge) [4, 71]. The semantic variant requires both core criteria (impaired confrontation naming; impaired single-word comprehension) and at least three of four additional criteria (impaired object knowledge, particularly for infrequent or familiar items; surface dyslexia or dysgraphia; preserved repetition; preserved speech production) [4, 71]. For atypical parkinsonism, corticobasal syndrome criteria (progressive asymmetric rigidity, apraxia, alien limb phenomenon, cortical sensory loss, focal dystonia, myoclonus and dementia) and Movement Disorder Society PSP criteria are used [4, 5].

The French PNDS adds three clarifications to the Rascovsky criteria. The DSM-5 criteria are almost identical, but accept impaired social cognition in place of executive impairment and allow diagnosis at the stage of mild neurocognitive disorder, which the functional decline criterion of the FTDC criteria makes difficult [34]. The criterion of relatively preserved episodic memory is contested, because about half of patients have significant memory impairment, and its removal is expected in the next revision of the criteria [34]. The diagnosis of bvFTD is aetiological (there is no “bvFTD due to AD” or “bvFTD due to bipolar disorder”), but the exclusions should not be applied rigidly, because bvFTD may reveal itself through a psychiatric disorder that meets DSM criteria and AD co-pathology may render biomarkers positive [34]. Social cognition has also been incorporated into a proposal of criteria for prodromal forms [34].

The instruments recommended by the PNDS differ according to the context [34]:

5.4. Gene-specific neuroimaging in frontotemporal dementia

In C9orf72-FTD, MRI shows symmetric, widespread atrophy (frontal, temporal, insular, cingulate and posterior cortex), with subcortical and cerebellar involvement; FDG-PET shows frontotemporal hypometabolism before MRI, and CSF may show non-specific AD biomarker changes [3]. In GRN-FTD, atrophy is focal and frequently asymmetric, frontal, temporal and parietal, and the global brain atrophy rate (3.5% per year) is higher than in MAPT-FTD; SPECT and PET show frontal, temporal and parietal hypoperfusion and hypometabolism, and hypometabolism appears in the presymptomatic phase before structural changes [4]. In MAPT-FTD, frontal and temporal atrophy is frequently symmetric, with two described patterns: “temporal” (hippocampus, amygdala, anteromedial temporal cortex and insula) and “frontotemporal” (frontal and lateral temporal) [5]. Amyloid PET helps distinguish AD, and FDG-PET non-genetic causes with frontal involvement [5]. The French PNDS recommends non-contrast brain MRI in all patients, with 3D gradient-echo T1, FLAIR, susceptibility and diffusion sequences, or non-contrast computed tomography if MRI is contraindicated; the request should be justified to direct the radiologist towards frontotemporal atrophy, and a normal MRI does not exclude the diagnosis [34]. In case of doubt, a follow-up MRI after 1–2 years allows atrophy to be compared [34]. FDG-PET (or, failing that, perfusion SPECT) is second line after a non-contributory MRI and is not recommended in presymptomatic diagnosis; a normal FDG-PET makes neurodegenerative disease less likely but does not exclude it, particularly in genetic forms, and some psychiatric diseases show reversible prefrontal hypometabolism, so a follow-up at one year may be useful [34]. In GRN-FTD hypometabolism is often asymmetric and involves the parietal cortex, and in C9orf72-FTD metabolism may be almost normal or show thalamic and cerebellar hypometabolism with variable neocortical changes; in these forms and in advanced disease, AD biomarkers in CSF, in blood or by amyloid PET are more useful, because the metabolic pattern no longer spares the posterior regions [34]. Dopamine transporter scintigraphy (¹²³I-ioflupane) or FDOPA PET help separate PSP and dementia with Lewy bodies, electromyography confirms motor neurone involvement and oculography supports forms bordering on PSP; the PNDS recommends discussion at a multidisciplinary meeting [34].

5.5. Plasma progranulin

Plasma or serum progranulin is markedly reduced in carriers of GRN loss-of-function variants and predicts mutation status in symptomatic patients and in asymptomatic relatives [79]. It is useful for screening, particularly to detect whole-gene deletions that sequencing may miss, but does not establish the diagnosis, which requires identification of the variant [2, 79]. The sources consulted do not provide a universal threshold, which depends on the assay and the laboratory. In France, plasma progranulin is measured before any molecular analysis, based on ten years of French experience and a systematic review of more than 7000 people cited by the PNDS (National Diagnosis and Care Protocol); a low value leads to GRN analysis, the only test that confirms the variant [34].

5.6. Who should be tested

The ACMG/NSGC guideline, published in 2011 and reaffirmed in 2018, recommends testing PSEN1, PSEN2 and APP in a symptomatic patient with early-onset AD and a family history of dementia or unknown family history (for example, through adoption), in a family with autosomal dominant dementia and at least one early-onset case, and in anyone with a relative carrying a known variant [11, 80]. Since 2016 (NSGC) and 2018 (ACMG), this document has been classified as a practice resource rather than an evidence-based guideline, and the addendum replaces “mutation” with “pathogenic variant” [80]. The Oxford clinical genetics manual suggests considering a panel in an affected person with onset before 61 years or a suggestive family history, such as affected individuals in three generations with at least one early-onset case [36]. The likelihood of finding a variant decreases with fewer affected relatives and later ages at onset, which should be explained before testing [11]. The detection rate in autosomal dominant families ranges from 40% to 80%, depending on the family definition used [11].

In the FTD-ALS spectrum, the consensus guideline for genetic testing in ALS, developed using a modified Delphi method with GRADE grading, recommends offering testing to all people with ALS (strong recommendation, level A), using a panel including at least C9orf72, SOD1, FUS and TARDBP, and additional genes with strong or definitive ClinGen validity [27]. Because of genetic overlap, the same laboratory rules apply to C9orf72-FTD [27]. The review by Chambers and colleagues argues that genetic testing should form part of routine neurological care across the ALS-FTD spectrum, that absence of a family history does not exclude a genetic aetiology, and that neurologists should know when to refer for genetic counselling [74]. The international FTD neuropsychiatry consortium recommends testing for the C9orf72 expansion in all patients with possible or probable bvFTD [75].

The French PNDS (National Diagnosis and Care Protocol) recommends offering genetic testing to all patients with bvFTD, with or without ALS, respecting the right not to know, with written informed consent from the patient or legal representative and with prior information on the legal obligation to inform at-risk relatives; variant classification follows ACMG standards [34]. In ambiguous situations in which a psychiatric disease cannot be distinguished from bvFTD, the PNDS allows testing for the C9orf72 expansion only exceptionally: when a first-degree relative with bvFTD or ALS has the expansion, the psychiatric symptoms are late-onset, psychotic and disabling, and the indication has been discussed at a multidisciplinary meeting of a specialist centre [34].

5.7. Testing strategy and methods

Single-gene testing is rarely useful and is not recommended, because presentations and ages at onset overlap; a multigene panel offers the highest yield and the fewest variants of uncertain significance (VUS) [5, 17]. For AD, the panel should contain APP, PSEN1 and PSEN2 and, if the laboratory provides it, APOE for ε4 detection and interpretation; susceptibility genes should not be included [17]. “Neurodegenerative” panels that also include genes for FTD, ALS, Parkinson disease and prion disease are reasonable, because the clinical diagnosis is usually only possible or probable [17]. If the panel is negative, exome or genome sequencing follows and, if the exome is non-diagnostic, multiexon deletion and duplication analysis [4, 17]. The C9orf72 expansion requires a dedicated parallel test [3, 27].

These values are proportions of known variants detected by each method (detection frequency), rather than clinical sensitivity: a negative panel in a patient with early-onset AD does not exclude a genetic cause, because 20–40% of EOFAD families have no variant in the three known genes [1, 17].

5.8. Testing the C9orf72 expansion

Conventional PCR detects normal alleles and up to approximately 80 repeats; apparent homozygosity for normal alleles does not exclude an expansion and requires repeat-primed PCR (RP-PCR) or Southern blot [3]. RP-PCR detects the expansion through a “sawtooth” pattern but does not size it, and variants adjacent to the repeat may cause false negatives, which bidirectional RP-PCR reduces [3, 27]. Southern blot is the gold standard and the only commercial method that sizes large expansions, although it may not distinguish intermediate alleles; dual-mode PCR covers 2 to approximately 950 repeats; and ExpansionHunter applied to PCR-free genome sequencing reliably indicates presence or absence, but size is an estimate requiring validation by an orthogonal method [27]. In a blinded 2014 study, only 5 of 14 laboratories produced PCR results fully concordant with Southern blot [27]. Blood DNA is sufficient for diagnosis despite somatic instability [27]. The guideline recommends reporting the sizes of unexpanded alleles, the maximum number of repeats detectable by the method and, for alleles with 20–29 repeats, uncertainty of interpretation [27]. Exact sizing has no demonstrated clinical utility [3].

5.9. Testing limitations

Short-read panel, exome or genome sequencing does not detect the C9orf72 expansion without dedicated methods, and APP duplication and GRN deletions escape sequencing without copy-number analysis [1, 3, 4]. Leucocyte DNA may not detect brain-restricted somatic mosaicism, as described for PSEN1 and the C9orf72 expansion [1, 3]. Regions poorly covered by the method, such as CHCHD10 exons 1–2, should be stated in the report [27]. A negative result in a patient without a known familial variant is ambiguous and does not exclude a genetic form; periodic variant reassessment is recommended [11, 27]. PSEN2 missense variants and variants in genes with “limited” or “refuted” ClinGen validity require particular caution [2, 27].

5.10. Report content and variant classification

The report should state the gene, transcript and version (for example, NM_000021.4 for PSEN1 or NM_001377265.1 for MAPT, accompanied by historical numbering), HGVS c. and p. nomenclature, zygosity and classification under ACMG/AMP standards (American College of Medical Genetics and Genomics and Association for Molecular Pathology), in which “pathogenic” and “likely pathogenic” are equivalent for clinical decision-making and a VUS neither establishes nor excludes the diagnosis [4, 5, 9]. Before communicating results, the ACMG/NSGC guideline recommends consulting variant databases and not using the specific genotype to predict phenotype [11]. For a VUS, segregation analysis in affected and unaffected relatives, population allele frequency and, in AD, pathological biomarkers help reclassification [2]. For the C9orf72 expansion, the report includes normal allele sizes, the method's detection limit and interpretation of intermediate alleles [27].

5.11. The three uses of the APOE genotype

The APOE genotype has three distinct uses, with different rules. It has an adjunctive role in supporting diagnosis in symptomatic patients, because it is neither sensitive nor specific [17]. Predictive testing in asymptomatic people is not clinically recommended because of limited utility and poor predictive value; if the person persists after counselling, it may be considered at the clinician's discretion following the Huntington disease predictive testing model [11]. Pharmacogenetic testing before anti-amyloid antibodies informs ARIA risk; it is recommended in the United States and mandatory in the European Union and United Kingdom (section 7.3) [28, 29, 64]. This last use is distinct from predictive testing, has implications for all first-degree relatives, and should not occur as a by-product without specific counselling [2, 63].

5.12. Practical algorithm

5.13. Prenatal diagnosis and preimplantation genetic testing

Prenatal diagnosis and preimplantation genetic testing (PGT) are technically possible after identification of the familial variant for APP, PSEN1, PSEN2, C9orf72, GRN, MAPT and CHMP2B [1, 3–6]. Ethical issues and the ACMG/NSGC guideline recommendation on prenatal testing are discussed in section 8.6 [11].

6. Differential Diagnosis

6.1. Treatable and acquired causes of cognitive decline

Treatable causes of cognitive decline should be excluded before considering any genetic diagnosis: depression, chronic drug intoxication, chronic central nervous system infection, thyroid disease, vitamin B12 and thiamine deficiencies, central nervous system angiitis and normal-pressure hydrocephalus; computed tomography or MRI identifies neoplasms, hydrocephalus and vascular disease [17]. The Oxford manual additionally lists human immunodeficiency virus infection, drug toxicity in older people, nutritional deficiency, hypothyroidism, alcohol, tumours, vasculitis, head injury and transmitted Creutzfeldt-Jakob disease [36]. In adults, inborn errors of metabolism may present with atypical psychosis, depression, ataxia, spastic paraparesis, movement disorders and dementia; treatable examples or those with specific treatment include cerebrotendinous xanthomatosis (tendon xanthomas, ataxia, spastic paraparesis, dementia and psychiatric signs), Niemann-Pick disease C (psychosis, cognitive decline, ataxia, vertical supranuclear gaze palsy, dystonia and splenomegaly), the psychiatric form of metachromatic leukodystrophy, cystathionine β-synthase deficiency and untreated phenylketonuria; catatonia, visual hallucinations, fluctuating symptoms and deterioration with psychotropic drugs are psychiatric warning signs [81]. Adult-onset X-linked adrenoleukodystrophy also enters the differential [36].

6.2. Comparative table of early-onset genetic dementias

6.3. Distinguishing Alzheimer disease from frontotemporal dementia

The distinction is not always clinical. AD may have a “frontal variant”, and PSEN1 variants may present with a prominent bvFTD-like behavioural syndrome, leading to confusion with MAPT-FTD [3, 5]. Conversely, MAPT-FTD with p.Val337Met or p.Arg406Trp and GRN-FTD may begin with episodic memory impairment and be diagnosed as AD [4, 5]. Extended neurodegenerative panels are therefore preferable to phenotype-directed testing, and amyloid and tau biomarkers help distinguish the two pathologies [5, 17]. The donanemab use recommendations emphasise that, in a syndrome suggestive of another aetiology (dementia with Lewy bodies, bvFTD, nfvPPA or svPPA), amyloid positivity more likely indicates copathology than AD as the cause, whereas atypical AD presentations (posterior cortical atrophy, logopenic PPA, dysexecutive AD) share AD pathophysiology [31].

6.4. Differential diagnosis with primary psychiatric disease

bvFTD, particularly C9orf72-FTD and MAPT-FTD, is frequently mistaken for depression, obsessive-compulsive disorder, bipolar disorder or schizophrenia, particularly in young adults with prominent behavioural manifestations [3, 5]. Longitudinal follow-up distinguishes most cases, but the two conditions may coexist [3]. The international consortium recommends testing these patients for the C9orf72 expansion and using NfL, MRI with visual rating scales and FDG-PET as adjuncts [75]. Schizophrenia typically begins between adolescence and 30 years, whereas MAPT-FTD with a psychotic presentation begins between 30 and 60 [5].

6.5. Differential diagnosis with movement disorders and Huntington disease

MAPT-FTD with parkinsonism is symmetric, without resting tremor and without response to levodopa, with early disinhibition, which separates it from Parkinson disease [5, 82]. Dementia with Lewy bodies, one of the three most common causes of dementia in older people, is characterised by cognitive fluctuation, visuospatial dysfunction, marked attentional impairment, complex visual hallucinations and mild extrapyramidal signs [36]. Chorea with behavioural changes in a C9orf72 carrier may be mistaken for Huntington disease; 2% of Huntington phenocopies without an HTT expansion have the C9orf72 expansion [3].

6.6. Differential diagnosis of motor involvement

With isolated upper motor neurone signs, the differential includes compressive cervical myelopathy, hereditary spastic paraplegia, adrenomyeloneuropathy and cerebrotendinous xanthomatosis; with lower motor neurone signs, it includes plexopathies, chronic inflammatory demyelinating polyneuropathy, multifocal motor neuropathy, toxic, infectious or metabolic neuropathies and myopathies, including inclusion body myositis in MSP [3]. A rare prion disease variant with ALS and FTD exists [3].

6.7. Differential diagnosis of the behavioural variant in the French PNDS

The most frequent differential diagnoses in specialist centres are primary psychiatric disorder, cerebrovascular disease, AD and the sequelae of brain injury of various causes [34]. The PNDS (French National Diagnosis and Care Protocol) groups the remainder into psychiatric disorders (schizophrenia, depression, bipolar disorder, personality and neurodevelopmental disorders), neurodegenerative diseases (AD, dementia with Lewy bodies, Creutzfeldt-Jakob disease, Huntington disease), vascular diseases (strategic infarcts, vasculitides, CADASIL and CARASIL), inflammatory and autoimmune diseases (multiple sclerosis, temporal arteritis, neurosarcoidosis, lupus, autoimmune and paraneoplastic encephalitides), infectious diseases (syphilis, herpes simplex, HIV, tuberculoma), traumatic, neoplastic and metabolic causes (thyroid disease, vitamin B12 deficiency), toxic causes (alcohol with Wernicke-Korsakoff syndrome, drugs, heavy metals, radiotherapy), neurometabolic and neurogenetic diseases (adult metachromatic leukodystrophy, CSF1R-related leukoencephalopathy, mitochondrial DNA diseases such as MELAS, spinocerebellar ataxias, fragile X syndrome, Wilson disease) and CSF circulation disorders (normal pressure hydrocephalus, spontaneous intracranial hypotension) [34].

The behavioural variant of AD meets the clinical criteria for bvFTD, with more memory impairment and a posterior pattern of atrophy and hypometabolism, and accounts for about 10% of bvFTD presentations, which justifies AD biomarkers whenever possible [34]. Conversely, FTLD with anterograde amnesia in older people often corresponds to limbic-predominant age-related TDP-43 encephalopathy (LATE), which may explain about 25% of AD-like presentations after the age of 85 years [34]. Right temporal variants begin with behavioural changes (logorrhoea and hypergraphia, disinhibition, rituals mimicking obsessive-compulsive disorder, egocentrism, somatic complaints, impaired face recognition) and are mistaken for bvFTD until asymmetric temporopolar atrophy and hypometabolism correct the diagnosis [34]. Up to 50% of patients with bvFTD first receive a psychiatric diagnosis, and a severe psychiatric disorder beginning after the age of 40 years, resistant to or poorly tolerating treatment and with an unfavourable cognitive and functional course should raise the possibility of a neurodegenerative cause; the PNDS adopts the algorithm of the NIC-FTD consortium (Neuropsychiatric International Consortium for FTD) and the PPD list (section 5.3) [34, 75]. The cognitive and behavioural presentations of Huntington disease may meet the criteria for bvFTD, particularly with late onset, family concealment or movements attributed to neuroleptics; encephalitides with anti-IgLON5, anti-LGI1, anti-DPPX and anti-NMDAR antibodies and spontaneous intracranial hypotension syndrome (brain sagging) may mimic bvFTD [34].

A neurometabolic disease should be considered in the presence of early onset, consanguinity or affected siblings, multisystem involvement, decompensation with fever or infection and atypical white matter or basal ganglia lesions on MRI; the PNDS lists the treatable forms and the first-line test [34]:

Source: [34].

Patients who, more than 3 years after onset, continue to meet the clinical criteria without significant decline and without imaging changes are termed phenocopies; the group is heterogeneous and includes indolent bvFTD, some due to C9orf72 expansion, atypical psychiatric causes and neurodevelopmental disorders decompensated by ageing [34].

7. Treatments and Risk Management Strategies

7.1. General principles

None of the entities covered in this article has an approved gene-directed treatment. In symptomatic ADAD, treatment follows that of Alzheimer disease generally, because no clinical differences justify another approach; in genetic FTD, treatment is symptomatic, supportive and multidisciplinary [1, 4, 5, 36]. A person with early-onset dementia should be followed by a neurologist or psychiatrist experienced in dementia, and the plan includes medical and behavioural management of depression, aggression, sleep disturbances, epileptic seizures and hallucinations, physiotherapy and occupational therapy for gait and activities of daily living and, as disease progresses, assisted living or institutional care [1, 36]. Sudden environmental changes and oversedation worsen the condition and should be avoided [1]. The Oxford clinical genetics manual mentions avoiding excessive alcohol consumption in at-risk individuals; the same text's suggestion that folic acid might be protective has no supporting source and does not constitute a recommendation [36].

Onset in the fourth to sixth decades brings problems that late-onset dementias rarely raise with the same intensity: ability to work, driving, financial management, dependent children and advance care planning while the person retains decision-making capacity [3–5]. The C9orf72, GRN and MAPT GeneReviews propose the post-diagnostic assessments summarised in the following table; none of these documents is based on a consensus guideline, because there are no published specific management recommendations for C9orf72-FTD/ALS, GRN-FTD or MAPT-FTD [3–5].

7.2. Symptomatic treatment of Alzheimer disease

Cholinesterase inhibitors (donepezil, rivastigmine and galantamine) partially compensate for cholinergic deficiency, and memantine, a non-competitive antagonist of the glutamate NMDA (N-methyl-D-aspartate) receptor, acts on another neurotransmission system [1]. Their benefit is modest and variable, they temporarily improve function and do not slow progression [1, 63]. They may be continued during anti-amyloid antibody treatment [63]. The Oxford manual describes limited benefit in early-onset dementia [36]. No trials of these drugs have specifically targeted ADAD; their use in this population is based on the general indication for Alzheimer dementia [1, 2]. The table presents dosing from the labels consulted; the European donepezil and galantamine labels and the US immediate-release memantine label were not obtained, so the corresponding rows indicate only the jurisdiction consulted.

The indication differs between jurisdictions: the rivastigmine patch is approved in the United States for severe Alzheimer dementia, whereas the European summary of product characteristics (SmPC) restricts it to mild to moderately severe disease [84, 85]. Memantine is an NMDA receptor antagonist; the description as a “partial agonist” in the GRN GeneReview does not reflect its mechanism [1, 4].

7.3. Anti-amyloid antibodies: indication, selection and APOE genotyping

Lecanemab (Leqembi®, Leqembi Iqlik®) is a humanised immunoglobulin G1 preferentially binding Aβ protofibrils, as well as oligomers and insoluble fibrils [63]. Donanemab (Kisunla®) is an immunoglobulin G1 targeting N-terminally truncated, pyroglutamate-modified Aβ (N3pE-Aβ), which represents approximately 1–3% of Aβ peptides and occurs only in fibrillar plaques and vascular deposits [31]. Both are indicated in the United States for treating Alzheimer disease, to be initiated at the mild cognitive impairment (MCI) or mild dementia stage, after confirming amyloid pathology by amyloid positron emission tomography (PET) or cerebrospinal fluid (CSF) [28, 89]. The labels do not mention PSEN1, PSEN2, APP or ADAD; the indication is not restricted by disease cause, efficacy in ADAD is extrapolated, and neither label supports treating asymptomatic carriers [2].

US appropriate use recommendations (AUR) define candidates as patients with MCI or mild dementia due to Alzheimer disease, with confirmed amyloid and, in trials, Mini-Mental State Examination (MMSE) scores of 22–30 for lecanemab and 20–28 for donanemab, at ages 50–90 and 60–85 years, respectively [31, 63]. Exclusions include uncontrolled immunological diseases or those requiring immunoglobulins, systemic monoclonal antibodies, immunosuppressants or plasma exchange, stroke or transient ischaemic attack, bleeding disorders and recent epileptic seizures (for lecanemab, the AUR recommend excluding any seizure history), pregnancy and breastfeeding, depression with a score above 8 on the Geriatric Depression Scale, body mass index above 35 or below 17 kg/m² and inflammatory amyloid angiopathy [63]. Non-amnestic presentations strongly linked to Alzheimer disease (posterior cortical atrophy, logopenic primary progressive aphasia, dysexecutive variant) may be treated; syndromes suggesting another aetiology (dementia with Lewy bodies, behavioural variant FTD, non-fluent and semantic primary progressive aphasias) are rarely candidates because amyloid positivity in these cases indicates copathology [31].

Baseline non-contrast magnetic resonance imaging (MRI), including FLAIR, sequences sensitive to haemoglobin products (T2*-GRE or SWI) and diffusion, preferably at 3 T, forms part of selection: exclusions include macrohaemorrhage exceeding 10 mm, more than 4 microhaemorrhages, superficial siderosis, vasogenic oedema, extensive white matter hyperintensities (Fazekas 3), multiple lacunar infarcts or territorial infarcts, vascular malformations and other structural lesions; inability to undergo MRI makes the patient ineligible, because computed tomography is inadequate [31, 63]. In 2023, blood biomarkers were not considered suitable for selecting patients, and the 2026 UK recommendations do not accept plasma p-tau217 as a substitute for CSF or PET for eligibility [63, 64].

The APOE genotype informs the decision because the risk of amyloid-related imaging abnormalities (ARIA) depends on the number of ε4 alleles: oedema or effusion (ARIA-E) and haemosiderin deposition, with microhaemorrhages and superficial siderosis (ARIA-H), are more frequent, more severe and more often symptomatic in ε4 homozygotes, who represent approximately 15% of patients with Alzheimer disease [28, 89]. US AUR recommend genotyping all consenting candidates before treatment and counselling on implications for relatives before requesting the test; they accept proteotyping and certified laboratories, since no APOE test is approved by the FDA (Food and Drug Administration) [31, 63]. The US label states in a boxed warning that testing should occur before initiation to inform ARIA risk, but that a patient declining genotyping may be treated, recognising that homozygosity cannot then be excluded [28]. In the European Union, both antibodies are authorised only in ε4 non-carriers and heterozygotes, the genotype must be determined using a CE-marked in vitro diagnostic (IVD) device intended for this purpose or an alternative validated test, and counselling and consent follow national standards [29, 30]. In the United Kingdom, MHRA (Medicines and Healthcare products Regulatory Agency) authorisation is likewise restricted to people with zero or one ε4 copy, testing is mandatory and counselling may be provided by the clinician without mandatory genetics referral [64, 90]. This pharmacogenetic use of the genotype is distinct from predictive testing and has family consequences discussed in section 8.5.

The donanemab AUR recommend additional caution in APOE ε2 carriers: at least two intracerebral haemorrhage cases have been described in ε2 carriers treated with lecanemab or trontinemab, and ε2, although reducing Alzheimer disease risk, is associated with cerebral amyloid angiopathy (CAA)-related haemorrhage and superficial siderosis; the relationship between ε2 and ARIA remains unclear [31].

7.4. Dosing, administration and regulation

For the US subcutaneous route, the patient or caregiver may self-administer lecanemab after at least two consecutive supervised doses, in the abdomen or thigh (or arm, if administered by another person); a missed dose may be given up to 6 days later, and route switching is permitted at defined intervals (first subcutaneous dose 1 week after the last IV dose; from subcutaneous to IV, 2–3 weeks during the initial regimen and 1 week during maintenance) [28]. Subcutaneous efficacy was approved on the basis of pharmacokinetic exposure and amyloid reduction comparable to the IV route, and transition to maintenance is based on pharmacokinetic and pharmacodynamic modelling [28]. After the first IV lecanemab infusion, the AUR recommend 3 h observation (with telephone contact), 2 h after the second and third, and 30 min subsequently if no reactions occur; for donanemab, 1 h after the first four infusions and 30 min thereafter [31, 63].

The four-step donanemab titration regimen (350/700/1050/1400 mg) was approved by the FDA on 1 July 2025 and replaced the 700/700/700/1400 mg regimen used in TRAILBLAZER-ALZ2 because, in TRAILBLAZER-ALZ6 (randomised open-label study, 212 patients on the modified regimen and 207 on the standard regimen, 6 months), ARIA-E fell from 24% to 14% overall and from 57% to 19% in ε4 homozygotes, with similar amyloid reduction [31, 89, 91].

The regulatory chronology is summarised in the following table, with the practical consequence of each difference.

The sources consulted do not include the reimbursement decision in Portugal or ANVISA registration of lecanemab; the NICE appeal decision text was not obtained, and appeal information comes from an Alzheimer Europe news item [94].

7.5. ARIA, efficacy and management of anti-amyloid antibody adverse effects

Efficacy was demonstrated in predominantly sporadic Alzheimer disease populations. In CLARITY AD, 1795 patients with MCI or mild dementia were randomised to IV lecanemab 10 mg/kg every 2 weeks or placebo; at 18 months, the difference in Clinical Dementia Rating – Sum of Boxes (CDR-SB) was −0.45 (27% slowing; p<0.0001), with concordant results on the ADAS-Cog14 cognitive scale (Alzheimer's Disease Assessment Scale–Cognitive) and ADCS-MCI-ADL functional scale (Alzheimer's Disease Cooperative Study–Activities of Daily Living for Mild Cognitive Impairment) [28]. The population had a median age of 72 years (50–90), 69% were ε4 carriers and 77% were white participants; exploratory analysis of ε4 homozygotes (15% of the population) showed no CDR-SB effect, although other scales and biomarkers favoured lecanemab [28]. In TRAILBLAZER-ALZ2, involving 1736 patients, donanemab slowed decline on the iADRS (integrated Alzheimer's Disease Rating Scale) by 35% in patients with low to intermediate tau and by 22% in the combined population at 76 weeks, and on CDR-SB by 36% and 29%, with amyloid reduction of 88 Centiloids versus placebo [31]. The clinical translation of these numbers is modest: the NICE assessment estimated a delay of approximately 4–6 months in progression from mild to moderate disease [90]. The aim is to slow decline rather than improve cognition, and the AUR request that CDR-SB results be explained to patients in understandable terms [63].

The following table presents ARIA by APOE genotype in the two phase 3 programmes. Percentages cannot be compared between drugs because the trials differed in criteria, duration, MRI schedule and titration regimen.

Across CLARITY AD, ARIA occurred in 21% of treated patients and 9% of the placebo group, symptomatic ARIA in 3% (29/898), with severe symptoms in 0.7%, and haemorrhage exceeding 1 cm in 0.7% versus 0.1% [28]. In the subpopulation corresponding to European and UK authorisation (non-carriers and heterozygotes), ARIA-E occurred in 8.9% (67/757) with lecanemab and 1.3% with placebo, symptomatic ARIA-E in 1.6%, and intracerebral haemorrhage in 0.5% versus 0.3%; the most frequent adverse reactions were infusion reactions (26%), ARIA-H (13%), falls (11%), headache (11%) and ARIA-E (9%) [64]. With donanemab, approximately one-quarter of ARIA events were symptomatic, almost all serious events occurred during the first three doses, and independent ARIA-E predictors were baseline microhaemorrhage count, superficial siderosis, mean arterial pressure of at least 93 mmHg and baseline amyloid burden of at least 108 Centiloids [31]. In CLARITY AD, 71% of ARIA-E episodes were detected on the first two MRI scans (weeks 9 and 13) and 81% resolved spontaneously within 4 months; new episodes continued to appear on week 24 and 52 MRI scans in ε4 carriers [63]. ARIA management does not differ by genotype [28, 89].

Three deaths in the CLARITY AD open-label extension were attributed to lecanemab: a non-carrier with cardiovascular disease receiving anticoagulation who developed macrohaemorrhage; an ε4 homozygote with large-vessel occlusion, severe CAA and vasculitis who developed multifocal haemorrhage after tissue plasminogen activator; and an ε4 homozygote with severe ARIA-E and ARIA-H and an inflammatory amyloid angiopathy-like presentation [63]. Patients receiving warfarin or other vitamin K antagonists, direct oral anticoagulants (dabigatran, rivaroxaban, edoxaban, apixaban) or heparin are therefore excluded, and thrombolysis should not be given without considering that ARIA-E may mimic ischaemic stroke [28, 63]. Acetylsalicylic acid up to 325 mg/day and other antiplatelet agents (clopidogrel, prasugrel, ticagrelor) are permitted, with probably greater risk in homozygotes [63]. In CLARITY AD, intracerebral haemorrhage occurred in 2.5% (2/79) of patients receiving anticoagulants, with or without antiplatelets, and in none of the placebo group [28]. If anticoagulation becomes necessary, lecanemab is suspended and may be resumed when anticoagulation is no longer required [63]; in the EU, anticoagulation precludes treatment initiation [29, 30].

In the EU, one-third of patients who experienced ARIA had a recurrence [29]. Symptoms requiring unscheduled MRI are headache, confusion, visual disturbances, dizziness, nausea and gait disturbance and, in severe forms, epileptic seizures, status epilepticus, encephalopathy, stupor and focal deficits [63]. In TRAILBLAZER-ALZ2, 13% of treated patients and 4% of placebo patients stopped because of an adverse reaction, with infusion reactions the main cause (4%) [89]. Minority representation was low: in CLARITY AD, black participants accounted for 2.3% and 2.7% in the two arms [63].

7.6. Anti-amyloid antibodies in ADAD and Down syndrome

No registration trial specifically studied ADAD patients. In TRAILBLAZER-ALZ and ALZ2, APP, PSEN1 or PSEN2 variant carriers were not explicitly excluded, but represented a tiny minority, and efficacy and safety in this population are unknown [31]. US donanemab AUR recommend assessing the specific variant and its degree of association with CAA (using the Alzforum mutation database), excluding patients with APP variants associated with severe CAA, considering carriers of other variants without a known association with severe CAA while acknowledging uncertainty, excluding patients with Down syndrome-associated AD, and strongly referring both groups to dedicated trials (DIAN-TU, Dominantly Inherited Alzheimer Network Trials Unit, and TRC-DS, Trial-Ready Cohort–Down Syndrome) [31]. UK lecanemab recommendations state that rare genetic forms should not be excluded solely for that reason if criteria are met, that there is no evidence for use in Down syndrome-associated AD, a population with more frequent CAA, that the drug should not be used when MRI shows CAA and should be used cautiously with a family history of cerebral haemorrhage [64]. The positions differ in ADAD: the US approach excludes by variant, whereas the UK approach relies on imaging.

This divergence has practical consequences for APP duplications and intra-Aβ variants. In a series of 43 APP duplication carriers from 24 families, 29.2% had symptomatic intracerebral haemorrhage and 14 of 19 met imaging criteria for CAA [26], and Dutch, Italian, Iowa, Flemish and Arctic variants cause hereditary amyloid angiopathy with recurrent lobar haemorrhages [2]. An APP finding requires MRI with haemoglobin-product-sensitive sequences before any decision, and confirmed CAA contraindicates treatment in the EU [2, 29, 30]. With PSEN1 and PSEN2 variants, CAA risk is heterogeneous and variant-dependent [31]. Asymptomatic carriers should access antibodies through trials rather than off-label use [2].

7.7. Symptomatic treatment of frontotemporal dementia

No treatment is approved for genetic or sporadic FTD, and the only FDA label mentioning FTD concerns a diagnostic radiopharmaceutical [2, 4]. Environmental, behavioural and physical interventions, counselling and caregiver support are first-line; caregiver burden is considerable, and information, psychological support and caregiver groups form part of treatment [3–5]. Selective serotonin reuptake inhibitors (SSRIs) are the first pharmacological option for affective disturbances, disinhibition, impulsivity, compulsivity and difficult behaviours; venlafaxine, a serotonin and noradrenaline reuptake inhibitor, is proposed when apathy is prominent [3–6]. Atypical antipsychotics are reserved for severe agitation, aggression or psychosis unresponsive to SSRIs, as a temporary measure, with reassessment at short intervals to discontinue them as soon as possible, because they induce extrapyramidal syndromes; severe aggression may require high doses of antipsychotics or antidepressants [3, 4, 6]. In C9orf72-FTD, frequent psychosis makes this decision more common; in CHMP2B-FTD, part of the late akinetic-rigid syndrome has been associated with neuroleptics [2, 3]. All use of these drugs in FTD is off-label; the GeneReviews consulted do not give doses, and the reference doses for the behavioural variant are those of the French PNDS (National Diagnosis and Care Protocol), presented below [34].

Randomised trial evidence is limited and all findings require confirmation. Trazodone, in a double-blind placebo-controlled crossover trial, may benefit irritability, agitation, depression and eating disturbances; paroxetine showed some benefit in an open-label study, but in a controlled trial of 10 patients worsened associative learning, reversal learning and delayed recognition; galantamine benefited primary progressive aphasia but not the behavioural variant, stabilising language in 36 patients followed for 18 weeks; rivastigmine, in a 12-month open-label study, improved behaviour and caregiver burden without preventing cognitive decline; methylphenidate, in a controlled crossover trial, reduced risk-taking behaviour and worsened spatial working memory; dextroamphetamine improved behaviour in 8 patients; memantine had mixed open-label results and a meta-analysis of small samples showed some benefit [4]. Donepezil was associated with worsening disinhibition and compulsions and should be avoided in the behavioural variant [4]. Anti-amyloid antibodies have no role in genetic FTD [2].

The 2025 French PNDS (National Diagnosis and Care Protocol) starts from a neurochemical profile specific to the behavioural variant, with serotonergic, dopaminergic and glutamatergic deficits and little cholinergic or noradrenergic deficit, unlike AD, which explains prescribing rules that differ from those for AD and for psychiatric disorders [34]. Before any drug, and again in the face of apparent “resistance”, triggering causes are investigated: somatic and iatrogenic (discomfort, pain, confusion, starting or stopping drugs), sensory (deafness), psychological (failures, reproaches, premorbid traits), the human environment (inappropriate responses and exhaustion of those close to the patient), the social and material environment (access to alcohol, driving, money, weapons, noisy places) and neurological [34]. The symptom-based recommendations follow the 2024 French recommendations on psychological and behavioural symptoms in neurocognitive disorders, issued by the French societies of geriatrics, memory centres and psychogeriatrics, and the ERN-RND European consensus on the pharmacotherapy of FTD [34].

• Apathy. Short, playful and rewarding activities, not necessarily linked to previous interests, supported by specialist teams and day centres. Only agomelatine, 50 mg/day, has shown benefit in a trial, with hepatic and headache monitoring; SSRIs and antipsychotics, even at low dose, are not indicated and may worsen apathy [34].
• Disinhibition. Do not blame or reprimand, avoid situations of loss of control, structure access to substances of abuse and warn those close to the patient. Citalopram (10–30 mg/day) or escitalopram (5–15 mg/day) has specific benefit, with an ECG because of the risk of QT prolongation and contraindication of combination with other torsadogenic drugs; class effects include sweating, headache, akathisia in the first weeks, nightmares, bruxism, extrapyramidal signs and hyponatraemia. In sexual disinhibition with medico-legal risk, paroxetine (10–40 mg/day) is the choice; if it fails, the specialist centre may discuss cyproterone acetate (Androcur®, France) or triptorelin (Décapeptyl®, France), despite serious adverse effects [34]. Since the 2020 European re-evaluation, because of the risk of meningioma dependent on cumulative dose, cyproterone at doses of 10 mg/day or more is reserved, outside prostate cancer, for situations in which alternatives have failed, at the lowest effective dose, and is contraindicated in those who have or have had a meningioma [95].
• Eating behaviour changes. Limit access to food stores, low-calorie foods, smaller portions, fixed mealtimes and soft textures that are neither fibrous nor sticky, because covert eating increases the risk of aspiration. Trazodone is the main indication, with monitoring for hypotension at initiation; in France it is available through a compassionate access authorisation, requested by a neurologist, psychiatrist or geriatrician and dispensed by the hospital pharmacy. Paroxetine and fluvoxamine have also been described as useful [34].
• Agitation, aggression and irritability. Avoid constraints, failures and demands, simplify communication, reduce noise, remove access to alcohol, weapons and driving, and diversify the people involved. Trazodone and paroxetine are of interest as background treatment, and the European consensus recommends trazodone for psychomotor restlessness; antipsychotics are not indicated [34].
• Stereotypies. Replace a dangerous or socially difficult stereotypy with another (for example, a foam ball or a lollipop for repetitive shouting, bearing in mind the risk of aspiration). SSRIs are recommended, particularly sertraline and fluvoxamine (myalgia at initiation; higher risk of epileptic seizures with fluvoxamine); for vocalisations resistant to SSRIs, aripiprazole may be tried, with extrapyramidal monitoring [34].
• Psychotic symptoms. Distraction works better than logical argument; symptoms that do not cause distress may not need a drug. This is the only situation with an indication for an antipsychotic, choosing the one with the lowest toxicity, such as aripiprazole [34].
• Poorly evaluated drugs. Benzodiazepines have not been studied and appear to be of little efficacy, with no poor tolerance described; hydroxyzine (Atarax®, France) may be used as a sedative or anxiolytic, without exceeding 4 × 25 mg/day (maximum in the European SmPC for adults since 2015, which falls to 50 mg/day in older people, advises against the drug at this age and contraindicates it in long QT or with risk factors for QT prolongation) [96], because the absence of a cholinergic deficit and the young age reduce the confusional risk of anticholinergics; mood stabilisers are indicated in bvFTD with bipolar disorder, and valproate and lamotrigine are used empirically for disinhibition, agitation and irritability, without evidence [34].
• Cognition. Cognitive rehabilitation in the early stages (memory centres, trained speech and language therapists), approaches centred on activities of daily living and occupational therapy; acetylcholinesterase inhibitors and memantine have no cognitive benefit [34].

The table reproduces the prescribing annex of the PNDS, which applies to adults with FTLD when a psychotropic drug is indicated; all indications are outside the summary of product characteristics [34].

Dose source: [34]. The PNDS maxima are lower than the label maxima for psychiatric indications; the guideline dose is the reference and the label serves only for comparison. In France, off-label prescribing requires informing the patient of the off-label nature, the absence of an alternative, the risks and the reimbursement conditions, writing on the prescription the statement of compassionate access outside the marketing authorisation, and justifying the prescription in the medical record [34].

7.8. Gene-directed management and associated manifestations

In C9orf72-FTD/ALS, the multidisciplinary team includes neurology, specialist nursing, respiratory medicine, speech and language therapy, physiotherapy, occupational therapy, respiratory therapy, nutrition, psychology, social work and genetic counselling [3]. The table summarises the measures proposed in the GeneReview, which provides no doses; management of motor involvement includes riluzole and edaravone [2, 3].

In MAPT-FTD, levodopa is justified only with significant functional impairment, and most patients do not respond; management of behaviour, dysarthria, dysphagia and sialorrhoea follows the preceding table [5]. In CHMP2B-FTD, GeneReview recommendations are based on the Danish centre's experience and additionally include spasmolytics for severe spasticity and cognitive rehabilitation [2, 6]. In multisystem proteinopathy, there is no disease-modifying therapy; bisphosphonates treat pain and disability from Paget disease of bone, extrapolating from Paget disease generally, and management includes weight control, physiotherapy and stretching, assistive devices, foot or scoliosis surgery, respiratory support and home safety; VCP modulators (VCP-lowering oligonucleotide, CB-5083, VCP activators) and arimoclomol are preclinical [2, 7]. In hereditary amyloid angiopathy due to APP variants, management consists of blood pressure control, avoiding antithrombotics when haemorrhagic risk outweighs the indication, seizure control and rehabilitation [2]. In Nasu-Hakola disease, valproate may help frontal symptoms and risperidone is limited by parkinsonism [2]. Tofersen is approved only for SOD1-associated ALS and does not apply to C9orf72, VCP or TBK1 [2].

7.9. Clinical trials in carriers and genetic forms

Trials in ADAD carriers test the hypothesis that removing or preventing amyloid accumulation before symptoms delays disease, and trials in genetic FTD seek to correct the direct consequence of the variant (restore progranulin, reduce tau or expanded RNA). The table presents published or registered results, including negative findings.

Ongoing trials in ADAD carriers include DIAN-TU ART (NCT06384573), an open-label phase IIIb/IV lecanemab study in 40 participants from the gantenerumab extension, with time to further CDR-SB progression compared with external controls and primary completion expected in 2030; DIAN-TU Tau NexGen (NCT05269394), phase II/III, combining anti-tau etalanetug (E2814) with lecanemab in 197 participants, with tau PET as the outcome and completion expected in 2028; DIAN-TU-002 primary prevention (NCT05552157; NCT06647498), phase II/III, comparing subcutaneous remternetug with placebo in approximately 280 asymptomatic carriers aged 18 or older, started in November 2024 with completion expected in 2034; and the new API Colombia trial (NCT06996730), phase 2/3, started on 8 July 2026 in approximately 240 PSEN1 p.(Glu280Ala) carriers aged 25–65, using donanemab for up to 18 months followed by randomised maintenance with RG6289, donanemab, the combination or placebo, with amyloid burden in Centiloids as the outcome [113–117]. The RG6289 registry does not describe the drug's mechanism [117]. In GRN-FTD, ongoing trials include AVB-101 (NCT06064890), an AAV-GRN vector delivered through bilateral thalamic infusion in approximately 18 patients, and PBFT02 (NCT04747431), an AAV1-GRN vector into the cisterna magna in approximately 30 patients with a GRN or C9orf72 variant; VES001 completed phase 1b in 6 asymptomatic GRN carriers in August 2025, without published results in the sources consulted [118–121]. Sponsor-reported interim PBFT02 data (cut-off 15 June 2025) show mean CSF progranulin of 26 ng/mL at 12 months in 4 patients, above the normal range of 3.28–8.15 ng/mL, and related serious adverse events (two venous sinus thromboses and one liver enzyme elevation); NfL comparison with external controls in 4 patients does not permit efficacy inference [122].

Trials cannot be compared indirectly: they differ in populations (asymptomatic carriers of a single variant, as in API, or mixed variants and stages, as in DIAN-TU), outcomes, duration and regimen. A relative CDR-SB reduction of 27% corresponds to 0.45 points on an 18-point scale and a delay measured in months [28, 90]. The three prevention or early-treatment trials in ADAD carriers with results (DIAN-TU-001, gantenerumab extension and API Colombia) did not demonstrate clinical benefit on the primary outcome, although gantenerumab removed amyloid, showing that a biomarker effect is insufficient [102, 104, 105]. INFRONT-3 conveyed the same lesson in GRN-FTD: plasma progranulin increased 165% without any clinical effect [108]. In the asymptomatic DIAN-TU-001 groups, the placebo group showed no decline during the trial, limiting statistical power [102].

7.10. Developments in the last 12 months

Between October 2025 and October 2026, the most consequential developments were as follows, with evidence level indicated. In a 21 October 2025 press release, Alector announced that latozinemab failed to meet the INFRONT-3 coprimary clinical outcome and that the open-label extension and continuation study were discontinued (sponsor press release); results were subsequently presented in an abstract at the 2026 American Academy of Neurology meeting (conference abstract, without full editorial review) [107, 108]. The negative API ADAD Colombia crenezumab trial result was published in 2026 (published randomised trial result) [105]. On 13 July 2026, the FDA approved subcutaneous initiation of lecanemab treatment (regulatory authorisation), after approving the maintenance autoinjector on 29 August 2025 [28, 92]. The European Commission authorised donanemab on 24 September 2025, restricted to ε4 non-carriers and heterozygotes, and adopted variation decisions for the two antibodies on 26 February and 11 June 2026, whose content is not detailed in the sources consulted (regulatory authorisation) [29, 30]. In England, appeals against NICE's negative recommendations were upheld in January 2026 and assessment returned to committee (ongoing health technology assessment process) [94]. UK lecanemab use recommendations were published in 2026 (expert recommendation) [64]. The new API Colombia trial with donanemab and RG6289 began on 8 July 2026 (registered trial) [117]. Exploratory BIIB080 analyses suggesting less decline were published (human study, exploratory analysis) [112], and interim PBFT02 data were released (sponsor data, not peer-reviewed) [122]. A search for clinical results of C9orf72-specific ASOs in FTD within this period found only preclinical work, which does not establish the absence of other studies.

7.11. Agents to avoid, pregnancy, perioperative care and transition

Donepezil should be avoided in behavioural variant FTD because it worsens disinhibition and compulsions [4]. Anticholinergics used for sialorrhoea or bladder symptoms may worsen cognition, antipsychotics cause extrapyramidal syndromes, levodopa has psychiatric effects, and oversedation worsens the condition [1, 3, 5]. During anti-amyloid antibody treatment, anticoagulants should be avoided, and thrombolysis for a presumed stroke requires considering ARIA first [28, 63]. In APP variant carriers with CAA, antithrombotics should be weighed against haemorrhagic risk [2]. In the behavioural variant, the PNDS (French National Diagnosis and Care Protocol) advises against antipsychotics outside severe psychosis and severe stereotypies resistant to SSRIs at maximum dose: an open-label study of olanzapine (mean 3 mg for 24 months) showed benefit only for delusions and crying, a double-blind study of quetiapine 150 mg showed no improvement at 3 weeks, tolerance is poor (somnolence, cognitive worsening, loss of mobility, antecollis), haloperidol has been described as a cause of death and some groups describe a hypersensitivity similar to that of dementia with Lewy bodies; first-generation antipsychotics have no indication and second-generation antipsychotics require close monitoring [34]. Acetylcholinesterase inhibitors and memantine do not improve behaviour and have worsened perseverative behaviours [34]. In a decompensation, recently introduced psychotropics, step 2 analgesics, anti-Alzheimer drugs and alcohol, even at low doses, should be reviewed as causes [34].

Perioperatively, cholinesterase inhibitors potentiate suxamethonium-type muscle relaxants and may cause bradycardia through vagotonic effects, which should be communicated to the anaesthetic team [83, 86]. There are no adequate pregnancy data for anti-amyloid antibodies, no animal reproduction studies have been performed, and trials excluded pregnant and breastfeeding women; paediatric safety and efficacy are not established [28, 29, 63]. In ADAD, reproductive age coincides with the years preceding onset, and reproductive discussions take place before pregnancy (section 8). The relevant transition in these diseases is from asymptomatic carrier to patient, rather than from paediatric to adult care: under the 2024 Alzheimer's Association criteria, the ADAD carrier remains at stage 0 while biomarkers are negative and progresses to stage 1 when a core biomarker becomes positive, corresponding to the point when participation in prevention trials becomes relevant [49].

7.12. Guidelines and differences between sources

The principal difference between societies concerns ε4 homozygotes (treatable with a warning in the US, excluded in the EU and UK) and ADAD (variant-based exclusion in US donanemab AUR, imaging-based decisions in UK lecanemab recommendations) [28, 29, 31, 64]. A second difference is conceptual: ε4 homozygosity is proposed as a highly penetrant genetic form of AD, whereas the 2024 criteria retain it as a risk factor [21, 49].

The French PNDS (National Diagnosis and Care Protocol) for the behavioural variant diverges from the other sources on several points [34]. For apathy, the GRN GeneReview and others propose venlafaxine, whereas the PNDS recognises only agomelatine and considers that SSRIs and antipsychotics may worsen it [4, 34]. The GeneReviews accept atypical antipsychotics for severe agitation or aggression resistant to SSRIs, whereas the PNDS considers that agitation is not an indication for an antipsychotic and prefers trazodone [3, 4, 34]. The GRN GeneReview describes a behavioural benefit of rivastigmine in an open-label study and mixed results with memantine, whereas the PNDS concludes there is no benefit and that perseverative behaviours worsen [4, 34]. Paroxetine worsened learning tests in a small controlled trial cited by the GeneReview, whereas the PNDS recommends it for sexual disinhibition [4, 34]. The international neuropsychiatry consortium recommends testing for the C9orf72 expansion in all possible or probable bvFTD and in suspected cases with psychiatric features, whereas the PNDS, in agreement regarding bvFTD, reserves testing in ambiguous psychiatric presentations for cases with a carrier first-degree relative, disabling late-onset psychosis and a decision in a multidisciplinary meeting [34, 75]. Plasma progranulin is a parallel screen in the algorithm of section 5.12 and a mandatory prior step in France [34, 79]. The PNDS describes almost complete penetrance after the age of 80 years for dominant forms in general, whereas the ALS panel considers that of the C9orf72 expansion to be much lower [27, 34]. A normal FDG-PET helps to exclude bvFTD for the international consortium, but the PNDS points out that it does not exclude some genetic forms, particularly C9orf72 [34, 75]. The GeneReviews set no follow-up intervals in FTD without ALS, whereas the PNDS proposes a specialist consultation every 6 months and a general practice consultation at least every 3 months [3, 4, 34]. Finally, the PNDS maxima (citalopram 30 mg, escitalopram 15 mg, sertraline 100 mg, aripiprazole 15 mg) are below the maxima of the US labels for psychiatric indications, and the reference dose is that of the PNDS, except for the QT safety limits in older people (citalopram 20 mg/day above 60 years of age; escitalopram 10 mg/day in most older people), which the PNDS does not specify and which prevail [34, 97–100]. The 4-year survival that the PNDS attributes to bvFTD with motor signs corresponds, in the study cited, to the PSP and corticobasal syndrome group [123].

7.13. Disclosure, patient protection, crises and social support in the behavioural variant

Disclosure of bvFTD is a process lasting several months, often a revision of the diagnosis after years of diagnostic wandering with psychiatric labels; uncertainty may last for years in possible forms without imaging changes, and the clinician should be cautious, given the social and family consequences of a diagnosis made during working age [34]. Psychological support should be offered, and therapeutic education has minority uptake among patients because of anosognosia, but it is one of the most useful interventions for caregivers, whose burden is greater than in AD, particularly when the caregiver is a partner of working age or a child of school age [34].

The PNDS treats bvFTD as a medico-social and medico-legal emergency from diagnosis: uncontrolled spending, gambling, vulnerability to scams, shoplifting, altercations and aggressive driving may have legal consequences, and uncontrollable offending behaviours may justify early institutionalisation, particularly in people living alone [34]. In France, the order of 28 March 2022 makes neuro-evolutive diseases incompatible with driving as soon as there is diagnostic doubt, until specialist assessment, and permanently from stage 3 of the Reisberg scale; in the early form still compatible with driving, reassessment takes place every 6 months, and doubt is resolved in favour of public safety [34]. Measures include restricting access to money, weapons and vehicles, legal protection (guardianship, curatorship, family authorisation), early involvement of the occupational physician, because symptoms often appear before retirement age, sick leave before a dismissal for misconduct, the invalidity pension, ALD no. 15 (full reimbursement), MDPH or APA applications according to age and retirement status and, since 19 February 2025, derogatory granting of the disability compensation benefit after the age of 60 years in severe rapidly progressive diseases; the family doctor may involve social services and the coordination support scheme in complex cases [34].

When behaviour worsens, the investigation looks for non-verbalised pain (ALGOPLUS scale), recent treatment changes, toxic substances including small amounts of alcohol, psychological distress and environmental discomfort (noise, cold), with full blood count, C-reactive protein, electrolytes, urea, creatinine, urine culture and plain abdominal radiograph for faecal impaction or pica [34]. Several French centres have semi-urgent referral pathways, which the caregiver or family doctor can activate, to avoid the emergency department [34]. Hospital admission, as short as possible, is justified in an acute crisis exceeding the capacity of the setting (with investigation for somatic or iatrogenic confusion) or in chronic disabling behaviour after failure of home interventions; cognitive-behavioural units are the main resource, psychiatry is used only for crises with defined goals and duration, and palliative care and advance directives should be rediscussed throughout the disease [34].

8. Reproductive Options and Family Counselling

8.1. Framework and reference documents

The 2011 joint document of the American College of Medical Genetics (ACMG) and National Society of Genetic Counselors (NSGC) remains the most frequently cited reference for AD counselling, but NSGC reclassified it as a practice resource in 2016 and ACMG did likewise when reaffirming it on 25 June 2018, with an addendum replacing “mutation” with “pathogenic variant”, consistent with the 2015 ACMG/AMP standards [11, 80]. Its recommendations have no formal evidence grades and should be cited as practice guidance. In the FTD-ALS spectrum, the 2023 ALS genetic testing expert panel, with graded recommendations, applies directly to C9orf72-FTD and TBK1- and VCP-related forms, and a 2023 neurology review recommends incorporating testing into routine neurological care because absence of a family history does not exclude a genetic cause [27, 74].

The ACMG/NSGC guideline states that genetic testing occurs only within genetic counselling, in person or by videoconference, supported by an experienced professional; that the pedigree covers at least three generations, including onset age of neurological or psychiatric symptoms, dementia type and diagnostic method, current age or age at death and causes of death, confirming diagnoses through medical records; and that risk assessment classifies the family as early- or late-onset and autosomal dominant, familial or sporadic [11]. The ALS panel adds that the pedigree should investigate FTD, other dementias, movement disorders and psychiatric diseases because they express the same spectrum [27]. Before disclosing a result, the guideline recommends consulting a variant database and not using genotype to predict phenotype [11].

8.2. Recurrence risks by scenario

The correlation between a carrier's onset age and relatives' onset ages is useful but not deterministic. In ADAD, individual onset age correlates with parental onset age and the variant and family means, even after adjustment for APOE and sex, and may vary by more than 20 years within a family [11, 62]. In genetic FTD, the correlation is stronger for MAPT than C9orf72 [24]. High penetrance is not absolute: a PSEN1 p.(His163Tyr) carrier followed for 22 years had no symptoms or CSF or imaging abnormalities at 65 years, when mean onset for that variant is 51 years [61].

8.3. Apparently isolated cases, de novo variants and mosaicism

A negative family history may result from early parental death, unrecognised diagnoses (a psychiatric disorder not recognised as FTD, for example), reduced penetrance or misattributed paternity or maternity [1, 3, 5]. For an apparently isolated proband with a C9orf72 expansion or GRN or MAPT variant, parental testing is recommended; if neither has the variant, explanations include a de novo variant or germline mosaicism, which leucocyte DNA does not detect, and sibling risk remains slightly above population risk [3–5]. In MAPT, two affected siblings with neither parent carrying the variant have been described, attributed to germline mosaicism [5]. Somatic mosaicism has been described in PSEN1 [1]. For the C9orf72 expansion, negative family history is considered confirmed only after negative parental testing, and the smallest expansion cosegregating in a family measured 47 repeats in leucocytes, with large expansions in the same individual's brain [3].

8.4. Cascade and predictive testing

The ACMG/NSGC guideline recommends testing APP, PSEN1 and PSEN2 in a symptomatic person with early-onset AD and a family history of dementia or unknown history (for example, through adoption), in a family with an autosomal dominant pattern and at least one early-onset case, and in a relative of a known variant carrier [11]. An affected relative should be tested first; if unavailable, an asymptomatic relative may be tested after being informed of the low probability of an informative result, and DNA banking from the affected individual should be discussed [1, 11]. A symptomatic patient expressing any reluctance should not be tested, and family disagreement is resolved through a family meeting [11]. For C9orf72, the ALS panel recommends informing relatives and discussing presymptomatic testing in adults, with counselling [27].

Predictive testing of asymptomatic adults follows the protocol developed for Huntington disease: two pre-test counselling sessions, at least one face-to-face session for results, a support person, baseline neurological assessment and psychological assessment, with psychology or psychiatry consultation when indicated; untreated depression or recent bereavement justify postponing testing [11]. Pre-test discussion addresses the person's motives and knowledge, the impact of a positive, negative or uncertain result, where results will be recorded, health, life and disability insurance, employment discrimination and legislative limitations, implications for other relatives, confidentiality and the post-result follow-up plan [1, 11]. Testing someone with equivocal symptoms is effectively predictive testing, and a positive result does not prove that symptoms are attributable to the variant [1, 17]. In a study of 21 people at risk of ADAD or MAPT-FTD, most coped effectively with presymptomatic testing, but long-term effects are unknown [1]. In France, presymptomatic testing for FTLD is available only to adults from families with an identified genetic cause, in a multidisciplinary clinic accredited by the Agence de la Biomédecine and following the Huntington disease protocol; the law requires carriers to inform at-risk relatives, directly or through the doctor, and the results consultation should preferably include a psychologist [34].

Testing asymptomatic minors is considered inappropriate because it provides no therapeutic benefit, removes autonomy over a future decision and may affect family dynamics and cause discrimination and anxiety; symptomatic people may be tested at any age [3–5, 11, 17].

8.5. The APOE genotype in counselling

The APOE genotype provides a risk estimate rather than a diagnosis, and absence of ε4 does not eliminate risk in someone with a first-degree relative with AD, whose risk remains increased 2–4-fold [2, 11]. Predictive APOE testing in asymptomatic people is not recommended, and direct-to-consumer testing is discouraged [11]. In the REVEAL trials, APOE genotype disclosure to adult children of AD patients did not cause significant short-term psychological harm, but approximately half of those correctly recalling the risk believed their risk differed from that disclosed, and ε4 carriers more often changed long-term care insurance; cohorts were self-selected, highly educated and tested free of charge outside clinical care [11]. When the laboratory determines APOE, whether other implications, such as cardiovascular implications, will be reported should be decided beforehand [11]. Discussion with an ε4 homozygote differs from discussion with a heterozygote because the risk magnitude and proposed genetic-form classification apply only to homozygosity [2, 21].

Genotyping before anti-amyloid antibodies inevitably generates predictive information for patients and relatives. US AUR include genetic expertise among necessary resources and request counselling on family implications before testing; UK recommendations allow clinician-delivered counselling and provide support and referral for homozygotes [31, 63, 64]. These UK recommendations state that homozygotes should be informed that each child has a 50% probability of having at least one ε4 allele [64]. This contradicts Mendelian transmission: an ε4/ε4 parent transmits an ε4 allele to every child, so all children have at least one ε4 allele; the probability of a child also being homozygous is zero if the other parent lacks ε4, 50% if heterozygous and certain if homozygous. A heterozygote transmits ε4 to each child with a 50% probability. Approximately one-quarter of the general population has at least one ε4 allele [63, 64].

8.6. Prenatal diagnosis and preimplantation genetic testing

Once the familial variant is identified, prenatal diagnosis (the Oxford manual mentions chorionic villus sampling) and PGT are technically possible for APP, PSEN1, PSEN2, C9orf72, GRN, MAPT and CHMP2B [1, 3–6, 36]. The ACMG/NSGC guideline advises against prenatal testing when the person intends to continue pregnancy with a carrier fetus [11]. Views on prenatal diagnosis of adult-onset diseases for termination of pregnancy differ among professionals and families, and the decision is personal [1, 3, 5]. PGT was described in an asymptomatic 30-year-old woman carrying an APP variant who had a child without the variant, and the case prompted published ethical debate [1]. A C9orf72 expansion in an embryo or fetus does not predict phenotype or course [3]. Family planning should be discussed before pregnancy, and counselling should be offered to affected or at-risk young adults [1, 3, 5]. The French PNDS (National Diagnosis and Care Protocol) provides that requests for prenatal diagnosis, and more rarely for PGT, in genetic FTLD are examined case by case by a specialist multidisciplinary team, according to the variant identified [34].

8.7. Participation in studies and trials

For asymptomatic carriers, natural history cohorts and trials provide access to structured follow-up and experimental therapies: DIAN and its expanded registry for ADAD families, GENFI (Genetic Frontotemporal Dementia Initiative) in Europe and Canada, and ALLFTD (ARTFL LEFFTDS Longitudinal Frontotemporal Lobar Degeneration) in North America [5, 124–126]. Participation does not require knowing one's genotype: in the API Colombia trial, non-carriers from the same family received placebo to maintain masking of genetic status [105]. US AUR recommend referring patients with ADAD and with Down syndrome-associated AD to DIAN-TU and TRC-DS [31]. The French PNDS (National Diagnosis and Care Protocol) adds that several centres take part in GENFI, which offers annual follow-up to carriers of pathogenic variants and their relatives, whatever their genetic status, and recommends raising brain and spinal cord donation for research early, through the Neuro-CEB network [34].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Early-onset Alzheimer disease and ADAD

Available figures measure different things and should not be confused. The Rouen study in France, strictly defining onset before 61 years, found an early-onset AD prevalence of 41.2 per 100,000 people at risk (40–59 years) and ADAD prevalence, defined by early-onset cases across three generations, of 5.3 per 100,000; 61% of early-onset patients had a positive family history and 13% had affected relatives across three generations [1, 20]. This is a regional population, not a national estimate. Orphanet assigns early-onset ADAD a prevalence class of 1–9 per 100,000 in Europe, recorded as a class without a validated underlying study [12]. As a proportion of all AD cases, early-onset familial disease represents less than 2% according to GeneReview and less than 3% according to the retired version; the ACMG/NSGC guideline places early-onset AD at 1–5% of all cases, ADAD below 5%, and cites an Alzheimer's Association estimate of approximately 250,000 people with early-onset AD in the United States [1, 11, 17]. Within early-onset familial AD, PSEN1 accounts for 20–70%, APP 10–15% and PSEN2 approximately 5%, leaving 20–40% without an identified cause; in autosomal dominant families, variant detection rates range from 40% to 80% depending on the definition used [11, 17].

9.2. APOE and ancestry

Approximately 20–25% of the population has at least one ε4 allele, and homozygotes represent approximately 2% of the population and 10–15% of AD patients [21, 63, 64]. In the meta-analysis by Farrer and colleagues (5930 patients and 8607 controls), odds ratios versus ε3/ε3 in people of European ancestry were approximately 3.2 for ε3/ε4, 14.9 for ε4/ε4 and 0.6 for ε2/ε3; the association was weaker in African Americans and Hispanics and stronger in Japanese people (approximately 5.6 and 33.1), was present from 40 to 90 years, attenuated after 70 and varied by sex [2, 65]. Risks therefore cannot be transferred between ancestries. Data underpinning the proposal of ε4 homozygosity as a genetic form come predominantly from people of European ancestry, with little information in African populations [21].

9.3. Frontotemporal dementia

Orphanet records an FTD point prevalence of 3.0 per 100,000 in Europe (class 1–9 per 100,000, unvalidated, based on a review) and an annual incidence of 3.05 per 100,000 in Italy; the behavioural variant has a validated prevalence of 4.0 per 100,000 in the Netherlands [14, 15, 22]. A population-based study in two UK counties (1.69 million inhabitants) measured a combined FTD, progressive supranuclear palsy and corticobasal syndrome prevalence of 10.8 per 100,000, with incidence of 1.61 per 100,000 person-years, peak prevalence at 65–69 years (42.6 per 100,000), estimated lifetime risk of 1 in 742 and a relevant genetic variant in 15% of screened patients [23]. FTD accounts for 5–10% of all dementias and 10–20% of dementias beginning before 65 years [4]. Approximately one-third of FTD is genetic, predominantly involving GRN, MAPT and C9orf72, and approximately 30% has a family history [24, 77]. GRN accounts for approximately 5% of all FTD and 20% of FTD with a family history, and 3.2% of isolated cases in a French series [4]. The C9orf72 expansion was the most frequent genetic cause in the founding series (11.7% of familial FTD and 23.5% of familial ALS) and is found in 30–50% of familial ALS and 4–10% of ALS without a family history [3, 53]. Gene-specific prevalence estimates (C9orf72-FTD between 0.04 and 134 per 100,000; MAPT-FTD approximately 0.8 per 100,000; GRN-FTD approximately 0.15 per 100,000) are calculations derived from proportions rather than population measurements [2, 3, 5]. Multisystem proteinopathy has an estimate of approximately 1 in 300,000 in the UK, probably underdiagnosed, which differs from Orphanet's figure of below 1 per 1,000,000 [2, 7]. CHMP2B accounts for less than 5% of all FTD [2]. The studies cited by the French PNDS (National Diagnosis and Care Protocol) on the cognitive and behavioural presentations of FTLD (bvFTD, nfvPPA and svPPA) indicate a prevalence of 15–22 per 100,000 and an age-adjusted incidence of 1.6–2.9 per 100,000 per year, probably underestimated, higher before 65 years and peaking between 55 and 70 years, although some series find a peak after 70 [34]. In the French national memory centre database (Banque Nationale Alzheimer), FTLD accounted for 3.6% of diagnoses of major neurocognitive disorder between 2018 and 2025, with under-representation of the motor forms, and bvFTD for 52.9% of FTLD; FTLD is the third degenerative cause of major neurocognitive disorder and the second before 65 years [34]. In communities with high acceptance of brain donation, 5–9% of older people have FTLD pathology at death, and about 40% of bvFTD cases are thought not to be correctly diagnosed [34]. These figures refer to bvFTD of all causes and not to the genetic forms alone.

9.4. Founder populations and geographical variation

PSEN1 variants have also been described in Japanese, African American and black African families [1]. A German early-onset AD cohort studied by exome sequencing showed a substantial frequency of PSEN2 variants, whose pathogenicity requires segregation and current criteria [2, 60].

9.5. Portugal and Brazil

The sources consulted include no prevalence studies of these genetic forms in Portugal or Brazil, or founder variants specific to these countries, and do not describe anti-amyloid antibody access in Portugal. In Brazil, ANVISA registered donanemab in April 2025 for MCI and mild dementia associated with AD, contraindicated during anticoagulant treatment or with CAA on MRI [93]. Absence of data in the sources consulted does not establish the absence of published data.

9.6. Special populations

In Down syndrome, virtually everyone has sufficient AD neuropathological changes by their mid-40s, with onset age and mortality consistent with complete genetic penetrance, and the biomarker sequence is predictable; the 95% prediction interval for onset age is 32 years, comparable to that of PSEN1 [21, 49]. Donanemab AUR recommend excluding these patients from anti-amyloid antibodies, and UK recommendations state there is no evidence for lecanemab, owing to more frequent CAA [31, 64]. Women of reproductive age carrying ADAD or FTD-ALS spectrum variants need reproductive counselling before pregnancy, and there are no safety data for anti-amyloid antibodies in pregnancy [1, 28]. Minority representation in registration trials was low: in the active CLARITY AD arm, approximately 12.5% of participants were Hispanic, 17.1% Asian and 2.3% Black, limiting generalisation of efficacy and ARIA risks [63]. Most younger patients with AD have at least one ε4 allele, and female carriers have a higher risk, particularly at younger ages [63].

9.7. European networks, registries and patient associations

The European Reference Network for Rare Neurological Diseases (ERN-RND) includes a frontotemporal dementia group, uses the CPMS (Clinical Patient Management System) for virtual case discussion and maintains a knowledge platform for each disease group [35]. The sources consulted identify no European reference network specifically for ADAD. Registries and cohorts preparing trials exist for carriers: DIAN and its expanded registry, GENFI and ALLFTD [124–126]. The Alzforum mutation database is recommended by the donanemab AUR to assess the association of ADAD variants with CAA [31]. Patient associations include Alzheimer Europe, the Alzheimer's Association, Alzheimer Portugal and Associação Brasileira de Alzheimer (ABRAz) for AD, and AFTD (Association for Frontotemporal Degeneration), FTD Talk and Rare Dementia Support for FTD [4, 17, 127–130]. ALS patient associations exist [3].

In France, the rare disease reference centre “Démences Rares ou Précoces”, part of the BRAIN-TEAM network (rare diseases of the central nervous system), brings together three reference centres (Paris, coordinating; Lille; Marseille) and 15 competence centres, including Pointe-à-Pitre, and may be contacted directly or as a tertiary referral by memory centres [34]. The France-DFT association, founded in 2010, supports patients with the behavioural variant or primary progressive aphasia and their caregivers, with a telephone helpline, discussion groups led by a psychologist, groups for patients' children, staff training, research support and delegates in Belgium and Switzerland; it belongs to the international World FTD United network (associations from 15 countries), which has promoted an awareness week every September since 2016, and funds the Neuro-CEB brain bank [34]. France Alzheimer, the Association Française des Aidants, therapeutic education programmes and the BRAIN-TEAM AIDAN programme complete the resources for caregivers [34].

10. Illustrative Clinical Case

10.1. Resistance to ADAD in a PSEN1 p.(Glu280Ala) carrier homozygous for APOE3 Christchurch

The case, published by Arboleda-Velasquez and colleagues in 2019, comes from the largest known ADAD family, in Antioquia, Colombia, where approximately 1200 PSEN1 p.(Glu280Ala) carriers develop MCI and dementia at median ages of 44 years (95% CI 43–45) and 49 years (95% CI 49–50) [59]. The proband, a confirmed carrier of this variant that increases Aβ42 production, was described by informants as cognitively intact until her eighth decade, approximately three decades beyond the expected onset age; exact age was withheld to protect anonymity [59]. At that time she met MCI criteria, with memory impairment limited to recent events and normal neurological examination, and maintained relative cognitive stability over 24 months of assessment [59]. Because her earlier history partly relied on informants, the authors could not distinguish delayed MCI onset from prolonged MCI [59].

Exome sequencing confirmed the PSEN1 variant and identified homozygosity for the rare APOE3 Christchurch variant, R136S in mature-protein numbering, corresponding to NM_000041.4:c.460C>A, p.(Arg154Ser) in precursor numbering; Sanger sequencing confirmed the result, and whole-genome analysis identified this homozygosity as the most likely modifier [9, 59]. The patient had dyslipidaemia treated with atorvastatin 40 mg/day, and type III hyperlipoproteinaemia was confirmed, common in variants affecting the APOE low-density lipoprotein (LDL) receptor-binding region; atorvastatin was increased to 80 mg/day and ezetimibe 10 mg/day added [59]. This management addressed dyslipidaemia and does not constitute ADAD treatment.

Biomarkers showed dissociation. Pittsburgh compound B PET revealed unusually high amyloid burden (distribution volume ratio 1.96, versus 1.49–1.60 in carriers with MCI in their 40s), but flortaucipir tau PET was restricted to medial temporal and occipital regions, glucose metabolism was preserved, including the precuneus, hippocampal volume was within the range of carriers with MCI in their 40s, and plasma NfL was relatively low for age (20.24 pg/mL) [59]. Figure 1 of the article shows amyloid, tau and FDG PET images and hippocampal volume (https://pmc.ncbi.nlm.nih.gov/articles/PMC6898984/figure/F1/); the pedigree is in Extended Data Figure 1 (https://pmc.ncbi.nlm.nih.gov/articles/PMC6898984/figure/F3/) and Sanger confirmation of R136S in Extended Data Figure 2 (https://pmc.ncbi.nlm.nih.gov/articles/PMC6898984/figure/F4/) [59].

In functional studies, ApoE3 Christchurch had a lower capacity to promote Aβ42 aggregation in vitro, similar to ApoE2, and the lowest heparin and heparan sulphate proteoglycan binding among the isoforms tested; ApoE2 and ApoE3 Christchurch reduced LDL receptor binding by 98% and 60%, respectively (Figure 2: https://pmc.ncbi.nlm.nih.gov/articles/PMC6898984/figure/F2/) [59]. An antibody targeting ApoE amino acids 130–143 reduced normal ApoE3 binding to heparin, which the authors propose as a hypothetical therapeutic target [59]. No other Christchurch homozygote carrying the PSEN1 variant was found; among 117 family members analysed, 6% were Christchurch heterozygotes, including four PSEN1 variant carriers who progressed to MCI at a mean age of 45 years, suggesting protection requires homozygosity [59].

10.2. Lessons for practice

The case illustrates three points. First, penetrance of an ADAD variant considered complete may be modified by variants in other genes, reinforcing the rule against predicting a carrier's onset age from genotype alone [11, 59]. Second, amyloid burden and clinical disease may dissociate: in this patient, very high amyloid burden coexisted with limited tau and neurodegeneration [59]. Carrier trials removing amyloid without clinical benefit point in the same direction (section 7.9) [102, 104]. Third, this is a single observation: other factors may have contributed, and association does not prove causality [59]. None of this report's management is presented as routine, and APOE testing remains unrecommended as a predictive test in asymptomatic people [11].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis of autosomal dominant Alzheimer disease

AD has a typical clinical duration of 8–10 years, ranging from 1 to 25 years, and death usually results from starvation, malnutrition or pneumonia [17]. In PSEN1-related forms, progression over 6–7 years is frequent, whereas PSEN2-related forms have a mean duration of 11 years [17]. Onset before 45 years may indicate faster progression, and severe cortical atrophy on MRI or marked metabolic deficits on PET suggest advanced disease [1]. In ε4 homozygotes in the NACC (National Alzheimer's Coordinating Center) neuropathological cohort, mean age at death was 77.2 years, 7–10 years earlier than in ε3 homozygotes [21]. These data describe disease without disease-modifying therapy; anti-amyloid antibodies modestly alter progression in predominantly sporadic AD, without survival data, and have no demonstrated ADAD efficacy (sections 7.5 and 7.6) [31, 90].

11.2. Prognosis of genetic frontotemporal dementia

In the international cohort of 3403 carriers, mean disease duration was 6.4 years (SD 4.9) for C9orf72, 7.1 years (SD 3.9) for GRN and 9.3 years (SD 6.4) for MAPT, with mean ages at death of 65.3, 68.8 and 58.5 years, respectively [24]. For C9orf72, duration ranges from 0 to 36 years; the ALS form lasts 2.9 ± 2.8 years, the FTD form 7.5 to 14 years depending on the cohort, and survival after ALS onset is approximately 1.8 years, making motor neurone disease the principal poor prognostic marker [3]. For GRN, duration is 3 to 12 years and mean age at death 65 ± 8 years in one series [4]. For MAPT, some patients live with the disease for more than 30 years [5]. For CHMP2B, duration ranges from 3 to more than 20 years, institutionalisation occurs at a mean age of 65 years, death at 68 years, and carriers lose an average of 13 years of life expectancy [6, 72]. In multisystem proteinopathy, death usually occurs in the sixth or seventh decade from respiratory insufficiency [2]. In TARDBP, ALS lasts 3–5 years and FTD 1–16 years [18]. In the UK population study, survival after diagnosis ranged from 2.9 years in progressive supranuclear palsy to 9.1 years in semantic primary progressive aphasia [23]. None of the sources describes alteration of natural history by treatment, except respiratory and nutritional support in motor neurone disease [3].

In bvFTD of all causes, the French PNDS (National Diagnosis and Care Protocol) considers a family history or an identified variant, associated ALS and parkinsonism with neuroleptic intolerance to be poor prognostic factors [34]. In the regional memory centre network of Nord-Pas-de-Calais (690 patients with FTLD syndromes referred between 2010 and 2016), median survival from symptom onset to death was 6.5 years in bvFTD (about 440 patients, including a minority with associated ALS), 6.1 years in the language variants, 4.0 years in the motor group, made up of patients with progressive supranuclear palsy and corticobasal syndrome, and 5.5 years in AD; after adjustment for age and sex, only the motor group had shorter survival than AD [123]. The French PNDS presents these values as the survival of probable bvFTD without and with motor signs, but the 4-year value refers to the motor forms (PSP and corticobasal syndrome) and not to bvFTD with parkinsonism [34, 123]. When ALS is present (about 15% of cases), life expectancy falls to about 3 years, according to a meta-analysis cited by the PNDS [34]. Possible bvFTD has a better prognosis, because the disease is less advanced or because there is no underlying FTLD [34]. Death results mainly from respiratory failure due to aspiration pneumonia or asphyxia from choking [34]. These are data for disease under symptomatic treatment, without disease-modifying therapy, in a predominantly sporadic population.

11.3. Surveillance

The historical early-onset AD GeneReview proposed monthly surveillance for secondary complications; this recommendation predates current care organisation and is not a guideline-based interval [1]. In the C9orf72 GeneReview, reassessment every 2–3 months applies to the ALS phase, and the FTD-only phase has no fixed intervals [3]. For the behavioural variant of any cause, including the genetic forms, the French PNDS (National Diagnosis and Care Protocol) fills this gap with a specialist consultation every 6 months and a general practice consultation at least every 3 months [34].

12. Summary and Final Considerations

Early-onset autosomal dominant AD and familial FTD share the adult genetics clinic for cognitive or behavioural decline before 65 years, but differ in genetic architecture and practical consequences. ADAD is rare (less than 2% of all AD), results from PSEN1, APP or PSEN2 variants altering amyloid precursor protein processing, has high but incomplete penetrance, and usually begins in the fourth or fifth decade [11, 17, 61]. APOE is a risk and modifier gene rather than a Mendelian cause under the 2024 criteria, although ε4 homozygosity has been proposed as a highly penetrant genetic form [21, 49]. In FTD, approximately one-third of cases are genetic; the C9orf72 GGGGCC expansion, GRN haploinsufficiency and MAPT variants account for most, and rare genes (CHMP2B, VCP, TBK1, TARDBP and others) link dementia to ALS, myopathy and Paget disease of bone [3–5, 7, 24].

Genetic diagnosis requires choosing the correct method: a panel with copy-number analysis to avoid missing APP duplication, and dedicated C9orf72 expansion testing, which short-read sequencing does not reliably detect [3, 17, 27]. PSEN2 variants require careful interpretation because many described as pathogenic in older literature have not been confirmed [2, 60]. Historical APP, APOE and MAPT nomenclature differs from HGVS on MANE transcripts and should be presented in both forms [9, 10].

No gene-directed treatment is approved for any of these entities. Anti-amyloid antibodies are authorised for AD at the MCI or mild dementia stage, with modest benefit, ARIA risk depending on ε4 allele count and rules differing between jurisdictions: ε4 homozygotes may be treated in the US and are excluded in the EU and UK, and APOE genotyping has become a pharmacogenetic test with family implications [28–30, 64]. In ADAD, efficacy data are lacking, APP variants with CAA favour exclusion, and carrier access should occur through trials [2, 31]. Prevention trials in ADAD carriers and the first phase 3 GRN-FTD trial were negative on clinical outcomes despite biomarker effects, and next-generation trials with other antibodies, anti-tau agents and gene therapies are ongoing [33, 102, 105, 108, 115]. In FTD, treatment is symptomatic and multidisciplinary, with SSRIs as the first pharmacological option for behaviour and avoidance of donepezil [3, 4]. The 2025 French PNDS (National Diagnosis and Care Protocol) for the behavioural variant sets reference doses for SSRIs, trazodone, agomelatine and aripiprazole, advises against antipsychotics outside psychosis and against anti-Alzheimer drugs, proposes genetic testing for all patients, preceded by plasma progranulin, and treats the diagnosis as a medico-social emergency, with specialist follow-up every 6 months [34].

Counselling rests on a 50% risk for first-degree relatives of a carrier, the predictive testing protocol inherited from Huntington disease, rejection of testing asymptomatic minors and reproductive discussion before pregnancy [3, 11, 17]. APOE-associated risks should be explained using the correct Mendelian rule: every child of an ε4 homozygote inherits an ε4 allele, contrary to the 50% stated in the UK recommendations [64]. Participation in cohorts such as DIAN, GENFI and ALLFTD and in prevention trials provides structured follow-up for asymptomatic carriers, for whom sources define no clinical surveillance intervals outside research [124–126].

13. References

1. Bird TD. Early-Onset Familial Alzheimer Disease [RETIRED CHAPTER, FOR HISTORICAL REFERENCE ONLY]. 1999 Sep 24 [updated 2012 Oct 18]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026.
2. GeneGuidance (locally curated database; records PSEN1, PSEN2, APP, APOE, C9orf72, GRN, MAPT, TREM2, CHMP2B, VCP, TBK1).
3. Gossye H, Engelborghs S, Van Broeckhoven C, van der Zee J. C9orf72 Frontotemporal Dementia and/or Amyotrophic Lateral Sclerosis. 2015 Jan 8 [updated 2020 Dec 17]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK268647/
4. Hsiung GYR, Feldman HH. GRN Frontotemporal Dementia. 2007 Sep 7 [updated 2020 Feb 6]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1371/
5. Rohrer J, Ryan B, Ahmed R. MAPT-Related Frontotemporal Dementia. 2000 Nov 7 [updated 2022 Aug 18]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1505/
6. Roos P, Holm IE, Nielsen JE, Ejlerskov P, Toft A, Isaacs AM. CHMP2B-Related Frontotemporal Dementia-Amyotrophic Lateral Sclerosis. 2007 Aug 23 [updated 2026 Mar 25]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1199/
7. Kimonis V, Romano C. Multisystem Proteinopathy. 2007 May 25 [updated 2026 Sep 3]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1476/
8. Forrest SL, Kril JJ, Stevens CH, Kwok JB, Hallupp M, Kim WS, et al. Retiring the term FTDP-17 as MAPT mutations are genetic forms of sporadic frontotemporal tauopathies. Brain. 2018;141(2):521-534. doi:10.1093/brain/awx328. PMID: 29253099. PMC5888940.
9. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166. PMC5765404. Tool: https://rest.variantvalidator.org (GRCh38, MANE Select).
10. National Center for Biotechnology Information. ClinVar. Records VCV000014245, VCV000014253, VCV000014256 (MAPT) and VCV000031151 (C9orf72). Available at: https://www.ncbi.nlm.nih.gov/clinvar/ .
11. Goldman JS, Hahn SE, Catania JW, LaRusse-Eckert S, Butson MB, Rumbaugh M, et al. Genetic counseling and testing for Alzheimer disease: joint practice guidelines of the American College of Medical Genetics and the National Society of Genetic Counselors. Genet Med. 2011;13(6):597-605. doi:10.1097/gim.0b013e31821d69b8. PMID: 21577118. PMC3326653.
12. Orphanet. Early-onset autosomal dominant Alzheimer disease (ORPHA:1020). Available from: https://www.orpha.net/en/disease/detail/1020
13. Orphanet. Familial Alzheimer-like prion disease (ORPHA:280397). Available from: https://www.orpha.net/en/disease/detail/280397
14. Orphanet. Frontotemporal dementia (ORPHA:282). Available from: https://www.orpha.net/en/disease/detail/282
15. Orphanet. Behavioral variant of frontotemporal dementia (ORPHA:275864). Available from: https://www.orpha.net/en/disease/detail/275864
16. Orphanet. Frontotemporal dementia with motor neuron disease (ORPHA:275872). Available from: https://www.orpha.net/en/disease/detail/275872
17. Bird TD. Alzheimer Disease Overview. 1998 Oct 23 [updated 2018 Dec 20]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1161/
18. Manohar V, Crowley L, Sreedharan J. TARDBP-Related Amyotrophic Lateral Sclerosis-Frontotemporal Dementia. 2009 Apr 23 [updated 2023 Jan 5]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK5942/
19. Zerr I, Schmitz M. Genetic Prion Disease. 2003 Mar 27 [updated 2021 Jan 7]. In: Adam MP, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1229/
20. Campion D, Dumanchin C, Hannequin D, Dubois B, Belliard S, Puel M, et al. Early-onset autosomal dominant Alzheimer disease: prevalence, genetic heterogeneity, and mutation spectrum. Am J Hum Genet. 1999;65(3):664-670. doi:10.1086/302553. PMID: 10441572. PMC1377972.
21. Fortea J, Pegueroles J, Alcolea D, Belbin O, Dols-Icardo O, Vaqué-Alcázar L, et al. APOE4 homozygozity represents a distinct genetic form of Alzheimer's disease. Nat Med. 2024;30(5):1284-1291. doi:10.1038/s41591-024-02931-w. PMID: 38710950. PMC13310155.
22. Seelaar H, Rohrer JD, Pijnenburg YA, Fox NC, van Swieten JC. Clinical, genetic and pathological heterogeneity of frontotemporal dementia: a review. J Neurol Neurosurg Psychiatry. 2011;82(5):476-486. doi:10.1136/jnnp.2010.212225. PMID: 20971753.
23. Coyle-Gilchrist IT, Dick KM, Patterson K, Vázquez Rodríquez P, Wehmann E, Wilcox A, et al. Prevalence, characteristics, and survival of frontotemporal lobar degeneration syndromes. Neurology. 2016;86(18):1736-1743. doi:10.1212/wnl.0000000000002638. PMID: 27037234. PMC4854589.
24. Moore KM, Nicholas J, Grossman M, McMillan CT, Irwin DJ, Massimo L, et al. Age at symptom onset and death and disease duration in genetic frontotemporal dementia: an international retrospective cohort study. Lancet Neurol. 2020;19(2):145-156. doi:10.1016/s1474-4422(19)30394-1. PMID: 31810826. PMC7007771.
25. Tang M, Ryman DC, McDade E, Jasielec MS, Buckles VD, Cairns NJ, et al. Neurological manifestations of autosomal dominant familial Alzheimer's disease: a comparison of the published literature with the Dominantly Inherited Alzheimer Network observational study (DIAN-OBS). Lancet Neurol. 2016;15(13):1317-1325. doi:10.1016/s1474-4422(16)30229-0. PMID: 27777020. PMC5116769.
26. Grangeon L, Charbonnier C, Zarea A, Rousseau S, Rovelet-Lecrux A, Bendetowicz D, et al. Phenotype and imaging features associated with APP duplications. Alzheimers Res Ther. 2023;15(1):93. doi:10.1186/s13195-023-01172-2. PMID: 37170141. PMC10173644.
27. Roggenbuck J, Eubank BHF, Wright J, Harms MB, Kolb SJ. Evidence-based consensus guidelines for ALS genetic testing and counseling. Ann Clin Transl Neurol. 2023;10(11):2074-2091. doi:10.1002/acn3.51895. PMID: 37691292. PMC10646996.
28. Eisai Inc. LEQEMBI (lecanemab-irmb) injection, for intravenous use; LEQEMBI IQLIK (lecanemab-irmb) injection, for subcutaneous use. Prescribing information. BLA 761269 / BLA 761375. Effective version 21/07/2026. Available at: https://api.fda.gov/drug/label.json?search=openfda.generic_name:lecanemab (set_id 9d1ff786-e577-410a-a273-c4d7d0e4e975).
29. European Medicines Agency. Leqembi (lecanemab): EPAR — product information (Summary of Product Characteristics). Revision 5; latest EC decision 11/06/2026. Amsterdam: EMA. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/leqembi .
30. European Medicines Agency. Kisunla (donanemab): EPAR — product information (Summary of Product Characteristics). Revision 3; latest EC decision 26/02/2026. Amsterdam: EMA. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/kisunla .
31. Rabinovici GD, Selkoe DJ, Schindler SE, Aisen P, Apostolova LG, Atri A, et al. Donanemab: Appropriate use recommendations. J Prev Alzheimers Dis. 2025;12(5):100150. doi:10.1016/j.tjpad.2025.100150. PMID: 40155270. PMC12180672.
32. Alector Inc. A Phase 3 Study to Evaluate Efficacy and Safety of AL001 in Frontotemporal Dementia (INFRONT-3). ClinicalTrials.gov NCT04374136. Updated 21/09/2026. Available at: https://clinicaltrials.gov/study/NCT04374136 .
33. Sevigny J, Uspenskaya O, Heckman LD, Wong LC, Hatch DA, Tewari A, et al. Progranulin AAV gene therapy for frontotemporal dementia: translational studies and phase 1/2 trial interim results. Nat Med. 2024;30(5):1406-1415. doi:10.1038/s41591-024-02973-0. PMID: 38745011. PMC11108785.
34. Haute Autorité de Santé; Centre de référence Démences Rares ou Précoces. Protocole National de Diagnostic et de Soins (PNDS): Dégénérescence frontotemporale – variante comportementale. Saint-Denis La Plaine: HAS; 2025 Dec 5 (October 2025 text). Available at: https://www.has-sante.fr/jcms/p_3764530
35. European Reference Network for Rare Neurological Diseases (ERN-RND). Disease groups — Frontotemporal dementia. Available at: https://www.ern-rnd.eu/ .
36. Firth HV, Hurst JA. Dementia—early onset and familial forms. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 386–9.
37. Baker M, Mackenzie IR, Pickering-Brown SM, Gass J, Rademakers R, Lindholm C, et al. Mutations in progranulin cause tau-negative frontotemporal dementia linked to chromosome 17. Nature. 2006;442(7105):916-919. doi:10.1038/nature05016. PMID: 16862116.
38. Cruts M, Gijselinck I, van der Zee J, Engelborghs S, Wils H, Pirici D, et al. Null mutations in progranulin cause ubiquitin-positive frontotemporal dementia linked to chromosome 17q21. Nature. 2006;442(7105):920-924. doi:10.1038/nature05017. PMID: 16862115.
39. Goate A, Chartier-Harlin MC, Mullan M, Brown J, Crawford F, Fidani L, et al. Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer's disease. Nature. 1991;349(6311):704-706. doi:10.1038/349704a0. PMID: 1671712.
40. Crawford F, Hardy J, Mullan M, Goate A, Hughes D, Fidani L, et al. Sequencing of exons 16 and 17 of the beta-amyloid precursor protein gene in 14 families with early onset Alzheimer's disease fails to reveal mutations in the beta-amyloid sequence. Neurosci Lett. 1991;133(1):1-2. doi:10.1016/0304-3940(91)90042-r. PMID: 1791986.
41. Sherrington R, Rogaev EI, Liang Y, Rogaeva EA, Levesque G, Ikeda M, et al. Cloning of a gene bearing missense mutations in early-onset familial Alzheimer's disease. Nature. 1995;375(6534):754-760. doi:10.1038/375754a0. PMID: 7596406.
42. Tsuda T, Chi H, Liang Y, Rogaeva EA, Sherrington R, Levesque G, et al. Failure to detect missense mutations in the S182 gene in a series of late-onset Alzheimer's disease cases. Neurosci Lett. 1995;201(2):188-190. doi:10.1016/0304-3940(95)12170-6. PMID: 8848249.
43. Levy-Lahad E, Wasco W, Poorkaj P, Romano DM, Oshima J, Pettingell WH, et al. Candidate gene for the chromosome 1 familial Alzheimer's disease locus. Science. 1995;269(5226):973-977. doi:10.1126/science.7638622. PMID: 7638622.
44. Rogaev EI, Sherrington R, Rogaeva EA, Levesque G, Ikeda M, Liang Y, et al. Familial Alzheimer's disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer's disease type 3 gene. Nature. 1995;376(6543):775-778. doi:10.1038/376775a0. PMID: 7651536.
45. Strittmatter WJ, Saunders AM, Schmechel D, Pericak-Vance M, Enghild J, Salvesen GS, et al. Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proc Natl Acad Sci U S A. 1993;90(5):1977-1981. doi:10.1073/pnas.90.5.1977. PMID: 8446617. PMC46003.
46. Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science. 1993;261(5123):921-923. doi:10.1126/science.8346443. PMID: 8346443.
47. Rovelet-Lecrux A, Hannequin D, Raux G, Le Meur N, Laquerrière A, Vital A, et al. APP locus duplication causes autosomal dominant early-onset Alzheimer disease with cerebral amyloid angiopathy. Nat Genet. 2006;38(1):24-26. doi:10.1038/ng1718. PMID: 16369530.
48. Jonsson T, Atwal JK, Steinberg S, Snaedal J, Jonsson PV, Bjornsson S, et al. A mutation in APP protects against Alzheimer's disease and age-related cognitive decline. Nature. 2012;488(7409):96-99. doi:10.1038/nature11283. PMID: 22801501.
49. Jack CR, Andrews JS, Beach TG, Buracchio T, Dunn B, Graf A, et al. Revised criteria for diagnosis and staging of Alzheimer's disease: Alzheimer's Association Workgroup. Alzheimers Dement. 2024;20(8):5143-5169. doi:10.1002/alz.13859. PMID: 38934362. PMC11350039.
50. Hutton M, Lendon CL, Rizzu P, Baker M, Froelich S, Houlden H, et al. Association of missense and 5'-splice-site mutations in tau with the inherited dementia FTDP-17. Nature. 1998;393(6686):702-705. doi:10.1038/31508. PMID: 9641683.
51. Hardy J, Duff K, Hardy KG, Perez-Tur J, Hutton M. Genetic dissection of Alzheimer's disease and related dementias: amyloid and its relationship to tau. Nat Neurosci. 1998;1(5):355-358. doi:10.1038/1565. PMID: 10196523.
52. Watts GD, Wymer J, Kovach MJ, Mehta SG, Mumm S, Darvish D, et al. Inclusion body myopathy associated with Paget disease of bone and frontotemporal dementia is caused by mutant valosin-containing protein. Nat Genet. 2004;36(4):377-381. doi:10.1038/ng1332. PMID: 15034582.
53. DeJesus-Hernandez M, Mackenzie IR, Boeve BF, Boxer AL, Baker M, Rutherford NJ, et al. Expanded GGGGCC hexanucleotide repeat in noncoding region of C9ORF72 causes chromosome 9p-linked FTD and ALS. Neuron. 2011;72(2):245-256. doi:10.1016/j.neuron.2011.09.011. PMID: 21944778. PMC3202986.
54. Renton AE, Majounie E, Waite A, Simón-Sánchez J, Rollinson S, Gibbs JR, et al. A hexanucleotide repeat expansion in C9ORF72 is the cause of chromosome 9p21-linked ALS-FTD. Neuron. 2011;72(2):257-268. doi:10.1016/j.neuron.2011.09.010. PMID: 21944779. PMC3200438.
55. Freischmidt A, Wieland T, Richter B, Ruf W, Schaeffer V, Müller K, et al. Haploinsufficiency of TBK1 causes familial ALS and fronto-temporal dementia. Nat Neurosci. 2015;18(5):631-636. doi:10.1038/nn.4000. PMID: 25803835.
56. Cirulli ET, Lasseigne BN, Petrovski S, Sapp PC, Dion PA, Leblond CS, et al. Exome sequencing in amyotrophic lateral sclerosis identifies risk genes and pathways. Science. 2015;347(6229):1436-1441. doi:10.1126/science.aaa3650. PMID: 25700176. PMC4437632.
57. Schultz SA, Liu L, Schultz AP, Fitzpatrick CD, Levin R, Bellier JP, et al. γ-Secretase activity, clinical features, and biomarkers of autosomal dominant Alzheimer's disease: cross-sectional and longitudinal analysis of the Dominantly Inherited Alzheimer Network observational study (DIAN-OBS). Lancet Neurol. 2024;23(9):913-924. doi:10.1016/s1474-4422(24)00236-9. PMID: 39074479. PMC11822357.
58. Firth HV, Hurst JA. Down's syndrome (trisomy 21). In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 666.
59. Arboleda-Velasquez JF, Lopera F, O'Hare M, Delgado-Tirado S, Marino C, Chmielewska N, et al. Resistance to autosomal dominant Alzheimer's disease in an APOE3 Christchurch homozygote: a case report. Nat Med. 2019;25(11):1680-1683. doi:10.1038/s41591-019-0611-3. PMID: 31686034. PMC6898984.
60. Blauwendraat C, Wilke C, Jansen IE, Schulte C, Simón-Sánchez J, Metzger FG, et al. Pilot whole-exome sequencing of a German early-onset Alzheimer's disease cohort reveals a substantial frequency of PSEN2 variants. Neurobiol Aging. 2016;37:208.e11-208.e17. doi:10.1016/j.neurobiolaging.2015.09.016. PMID: 26522186.
61. Thordardottir S, Rodriguez-Vieitez E, Almkvist O, Ferreira D, Saint-Aubert L, Kinhult-Ståhlbom A, et al. Reduced penetrance of the PSEN1 H163Y autosomal dominant Alzheimer mutation: a 22-year follow-up study. Alzheimers Res Ther. 2018;10(1):45. doi:10.1186/s13195-018-0374-y. PMID: 29747683. PMC5944151.
62. Ryman DC, Acosta-Baena N, Aisen PS, Bird T, Danek A, Fox NC, et al. Symptom onset in autosomal dominant Alzheimer disease: a systematic review and meta-analysis. Neurology. 2014;83(3):253-260. doi:10.1212/wnl.0000000000000596. PMID: 24928124. PMC4117367.
63. Cummings J, Apostolova L, Rabinovici GD, Atri A, Aisen P, Greenberg S, et al. Lecanemab: Appropriate Use Recommendations. J Prev Alzheimers Dis. 2023;10(3):362-377. doi:10.14283/jpad.2023.30. PMID: 37357276. PMC10313141.
64. Mummery CJ, Rasmussen J, Blackburn D, Coulthard E, Davies RR, Dorsey R, et al. Lecanemab appropriate use recommendations for clinical practice in the UK. J Neurol Neurosurg Psychiatry. 2026;97(5):372-384. doi:10.1136/jnnp-2025-336597. PMID: 41390238. PMC13151530.
65. Farrer LA, Cupples LA, Haines JL, Hyman B, Kukull WA, Mayeux R, et al. Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. JAMA. 1997;278(16):1349-1356. doi:10.1001/jama.1997.03550160069041. PMID: 9343467.
66. Guerreiro R, Wojtas A, Bras J, Carrasquillo M, Rogaeva E, Majounie E, et al. TREM2 variants in Alzheimer's disease. N Engl J Med. 2013;368(2):117-127. doi:10.1056/nejmoa1211851. PMID: 23150934. PMC3631573.
67. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 40, section 40.5 Neuronal Ceroid Lipofuscinoses.
68. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 44, Disorders of Cellular Trafficking (44.1.3; 44.3.1 Dementia and Others; Table 44.1).
69. Guerreiro RJ, Lohmann E, Brás JM, Gibbs JR, Rohrer JD, Gurunlian N, et al. Using exome sequencing to reveal mutations in TREM2 presenting as a frontotemporal dementia-like syndrome without bone involvement. JAMA Neurol. 2013;70(1):78-84. doi:10.1001/jamaneurol.2013.579. PMID: 23318515. PMC4001789.
70. Bateman RJ, Xiong C, Benzinger TL, Fagan AM, Goate A, Fox NC, et al. Clinical and biomarker changes in dominantly inherited Alzheimer's disease. N Engl J Med. 2012;367(9):795-804. doi:10.1056/nejmoa1202753. PMID: 22784036. PMC3474597.
71. Gorno-Tempini ML, Hillis AE, Weintraub S, Kertesz A, Mendez M, Cappa SF, et al. Classification of primary progressive aphasia and its variants. Neurology. 2011;76(11):1006-1014. doi:10.1212/wnl.0b013e31821103e6. PMID: 21325651. PMC3059138.
72. Roos P, Johannsen P, Lindquist SG, Brown JM, Waldemar G, Duno M, et al. Six generations of CHMP2B-mediated Frontotemporal Dementia: Clinical features, predictive testing, progression, and survival. Acta Neurol Scand. 2022;145(5):529-540. doi:10.1111/ane.13578. PMID: 34997757.
73. Al-Obeidi E, Al-Tahan S, Surampalli A, Goyal N, Wang AK, Hermann A, et al. Genotype-phenotype study in patients with valosin-containing protein mutations associated with multisystem proteinopathy. Clin Genet. 2018;93(1):119-125. doi:10.1111/cge.13095. PMID: 28692196. PMC5739971.
74. Chambers C, Lichten L, Crook A, Uhlmann WR, Dratch L. Incorporating Genetic Testing Into the Care of Patients With Amyotrophic Lateral Sclerosis/Frontotemporal Degeneration Spectrum Disorders. Neurol Clin Pract. 2023;13(5):e200201. doi:10.1212/cpj.0000000000200201. PMID: 37736067. PMC10511270.
75. Ducharme S, Dols A, Laforce R, Devenney E, Kumfor F, van den Stock J, et al. Recommendations to distinguish behavioural variant frontotemporal dementia from psychiatric disorders. Brain. 2020;143(6):1632-1650. doi:10.1093/brain/awaa018. PMID: 32129844. PMC7849953.
76. Frisoni GB, Festari C, Massa F, Cotta Ramusino M, Orini S, Aarsland D, et al. European intersocietal recommendations for the biomarker-based diagnosis of neurocognitive disorders. Lancet Neurol. 2024;23(3):302-312. doi:10.1016/s1474-4422(23)00447-7. PMID: 38365381.
77. Greaves CV, Rohrer JD. An update on genetic frontotemporal dementia. J Neurol. 2019;266(8):2075-2086. doi:10.1007/s00415-019-09363-4. PMID: 31119452. PMC6647117.
78. Rascovsky K, Hodges JR, Knopman D, Mendez MF, Kramer JH, Neuhaus J, et al. Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain. 2011;134(Pt 9):2456-2477. doi:10.1093/brain/awr179. PMID: 21810890. PMC3170532.
79. Finch N, Baker M, Crook R, Swanson K, Kuntz K, Surtees R, et al. Plasma progranulin levels predict progranulin mutation status in frontotemporal dementia patients and asymptomatic family members. Brain. 2009;132(Pt 3):583-591. doi:10.1093/brain/awn352. PMID: 19158106. PMC2664450.
80. Goldman JS, Hahn SE, Catania JW, LaRusse-Eckert S, Butson MB, Rumbaugh M, et al. ADDENDUM: Genetic counseling and testing for Alzheimer disease: joint practice guidelines of the American College of Medical Genetics and the National Society of Genetic Counselors. Genet Med. 2019;21(10):2404. doi:10.1038/s41436-019-0559-1. PMID: 31217590.
81. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 2, Inborn Errors of Metabolism in Adults: A Diagnostic Approach to Neurological and Psychiatric Presentations.
82. Firth HV, Hurst JA. Parkinson's disease. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 516–7.
83. Eisai Inc. ARICEPT (donepezil hydrochloride) tablets and orally disintegrating tablets. Prescribing information. NDA 020690/022568. Effective version 30/12/2021. Available at: https://api.fda.gov/drug/label.json?search=openfda.generic_name:donepezil .
84. Novartis. EXELON PATCH (rivastigmine transdermal system). Prescribing information. NDA 022083. Effective version 08/05/2024. Available at: https://api.fda.gov/drug/label.json?search=openfda.generic_name:rivastigmine .
85. European Medicines Agency. Exelon (rivastigmine): EPAR — product information. Authorisation 11/05/1998. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/exelon .
86. Galantamine hydrobromide extended-release capsules (ANDA 090178). Prescribing information. Effective version 12/09/2025. Available at: https://api.fda.gov/drug/label.json?search=openfda.generic_name:galantamine .
87. Memantine hydrochloride extended-release capsules (ANDA 214651). Prescribing information. Effective version 01/09/2026. Available at: https://api.fda.gov/drug/label.json?search=openfda.generic_name:memantine .
88. European Medicines Agency. Ebixa (memantine hydrochloride): EPAR — product information. Authorisation 15/05/2002. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/ebixa .
89. Eli Lilly and Company. KISUNLA (donanemab-azbt) injection, for intravenous use. Prescribing information. BLA 761248. Effective version 30/12/2025. Available at: https://api.fda.gov/drug/label.json?search=openfda.generic_name:donanemab (set_id 190352d4-ef62-4679-b4fa-e846e2766afa).
90. Wilkinson E. NICE rejects Alzheimer's disease drugs for third time. Pharm J. 19 Jun 2025. Available at: https://pharmaceutical-journal.com/article/news/nice-rejects-alzheimers-disease-drugs-for-third-time .
91. U.S. Food and Drug Administration. Drugs@FDA: BLA 761248 (Kisunla) — submission history. Available at: https://api.fda.gov/drug/drugsfda.json?search=application_number:BLA761248 .
92. U.S. Food and Drug Administration. Drugs@FDA: BLA 761269 (Leqembi) and BLA 761375 (Leqembi Iqlik) — submission history. Available at: https://api.fda.gov/drug/drugsfda.json?search=application_number:BLA761269 .
93. Agência Brasil. Anvisa aprova medicamento para retardar avanço do Alzheimer. 23 Apr 2025. Available at: https://agenciabrasil.ebc.com.br/saude/noticia/2025-04/anvisa-aprova-medicamento-para-retardar-avanco-do-alzheimer .
94. Alzheimer Europe. NICE to reconsider lecanemab and donanemab following successful appeals [news item]. 20 Mar 2026. Available at: https://www.alzheimer-europe.org/news/nice-reconsider-lecanemab-and-donanemab-following-successful-appeals .
95. European Medicines Agency. Cyproterone-containing medicinal products: Article 31 referral — restrictions in use of cyproterone due to meningioma risk (PRAC recommendation 2020 Feb 13; CMDh position 2020 Mar). Amsterdam: EMA; 2020. Available at: https://www.ema.europa.eu/en/medicines/human/referrals/cyproterone-containing-medicinal-products
96. European Medicines Agency. Hydroxyzine-containing medicinal products: Article 31 referral — Annex III, amendments to relevant sections of the summary of product characteristics and package leaflets. London: EMA; 2015. Available at: https://www.ema.europa.eu/en/documents/referral/hydroxyzine-article-31-referral-annex-iii_en.pdf
97. Viatris Specialty. ZOLOFT (sertraline hydrochloride). Prescribing information. NDA019839. Revised 2023 Aug [cited 2026 Oct 26]. Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=fda754f6-d0f3-4dce-a17a-927d64f912f7
98. Allergan Inc. Celexa (citalopram) tablets. Prescribing information. NDA020822. Effective 2023-10-09 [cited 2026 Oct 26]. DailyMed set ID 4259d9b1-de34-43a4-85a8-41dd214e9177. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=4259d9b1-de34-43a4-85a8-41dd214e9177
99. Allergan. LEXAPRO (escitalopram) tablets. Prescribing information. NDA021323. Revised 2023 Oct [cited 2026 Oct 26]. Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=13bb8267-1cab-43e5-acae-55a4d957630a
100. Otsuka America Pharmaceutical, Inc. ABILIFY (aripiprazole) — prescribing information [Internet]. NDA021436. SPL version effective 2025-01-29 (set id c040bd1d-45b7-49f2-93ea-aed7220b30ac). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=c040bd1d-45b7-49f2-93ea-aed7220b30ac
101. European Medicines Agency. Valdoxan (agomelatine): EPAR — product information (SmPC, sections 4.2–4.4). First authorisation 19/02/2009; last renewal 12/12/2018. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/valdoxan . Accessed 6 Nov 2026.
102. Salloway S, Farlow M, McDade E, Clifford DB, Wang G, Llibre-Guerra JJ, et al. A trial of gantenerumab or solanezumab in dominantly inherited Alzheimer's disease. Nat Med. 2021;27(7):1187-1196. doi:10.1038/s41591-021-01369-8. PMID: 34155411. PMC8988051.
103. Washington University School of Medicine. DIAN-TU-001 gantenerumab/solanezumab analysis study. ClinicalTrials.gov NCT04623242. Available at: https://clinicaltrials.gov/study/NCT04623242 .
104. Bateman RJ, Li Y, McDade EM, Llibre-Guerra JJ, Clifford DB, Atri A, et al. Safety and efficacy of long-term gantenerumab treatment in dominantly inherited Alzheimer's disease: an open-label extension of the phase 2/3 multicentre, randomised, double-blind, placebo-controlled platform DIAN-TU trial. Lancet Neurol. 2025;24(4):316-330. doi:10.1016/s1474-4422(25)00024-9. PMID: 40120616. PMC12042767.
105. Tariot PN, Lopera FS, Ríos-Romenets S, Sink KM, Giraldo-Chica M, Acosta-Baena N, et al. Safety and efficacy of crenezumab in cognitively unimpaired carriers of the PSEN1Glu280Ala mutation at risk for autosomal-dominant Alzheimer's disease in Colombia (API ADAD Colombia Trial): a phase 2, randomised, double-blind, placebo-controlled trial. Lancet Neurol. 2026;25(2):147-159. doi:10.1016/s1474-4422(25)00426-0. PMID: 41579901. PMC12994566.
106. Genentech, Inc. Crenezumab versus placebo in preclinical PSEN1 E280A carriers (API ADAD Colombia). ClinicalTrials.gov NCT01998841. Available at: https://clinicaltrials.gov/study/NCT01998841 .
107. Alector, Inc. Alector announces topline results from latozinemab phase 3 trial in individuals with frontotemporal dementia due to a GRN mutation and provides business update [press release]. 21 Oct 2025. Available at: https://investors.alector.com/news-releases/news-release-details/alector-announces-topline-results-latozinemab-phase-3-trial/ .
108. Mayorga A, Boeve B, Le Ber I, Borroni B, Mummery C, Rohrer J, et al. Efficacy and safety results of INFRONT-3: a phase 3 randomized, double-blind, placebo controlled study evaluating latozinemab (AL001) in FTD-GRN (LS2.002) [abstract]. Neurology. 2026;106(11 Suppl 1). doi:10.1212/WNL.0000000000217902. Available at: https://index.mirasmart.com/AAN2026/PDFfiles/AAN2026-005852.html .
109. Prevail Therapeutics. Phase 1/2 study of LY3884963 (PR006) in FTD-GRN. ClinicalTrials.gov NCT04408625. Available at: https://clinicaltrials.gov/study/NCT04408625 .
110. van den Berg LH, Rothstein JD, Shaw PJ, Babu S, Benatar M, Bucelli RC, et al. Safety, tolerability, and pharmacokinetics of antisense oligonucleotide BIIB078 in adults with C9orf72-associated amyotrophic lateral sclerosis: a phase 1, randomised, double blinded, placebo-controlled, multiple ascending dose study. Lancet Neurol. 2024;23(9):901-912. doi:10.1016/s1474-4422(24)00216-3. PMID: 39059407.
111. Mummery CJ, Börjesson-Hanson A, Blackburn DJ, Vijverberg EGB, De Deyn PP, Ducharme S, et al. Tau-targeting antisense oligonucleotide MAPTRx in mild Alzheimer's disease: a phase 1b, randomized, placebo-controlled trial. Nat Med. 2023;29(6):1437-1447. doi:10.1038/s41591-023-02326-3. PMID: 37095250. PMC10287562.
112. Shulman M, Wu S, Ziogas N, Edwards A, Collins J, Lin L, et al. Exploratory analyses of clinical outcomes from the BIIB080 phase 1b study in mild Alzheimer's disease. Nat Aging. 2026;6(2):445-453. doi:10.1038/s43587-025-01031-9. PMID: 41673497. PMC12920120.
113. Washington University School of Medicine. DIAN-TU Amyloid Removal Trial (ART), lecanemab. ClinicalTrials.gov NCT06384573. Available at: https://clinicaltrials.gov/study/NCT06384573 .
114. Washington University School of Medicine. DIAN-TU Tau NexGen (E2814 + lecanemab). ClinicalTrials.gov NCT05269394. Available at: https://clinicaltrials.gov/study/NCT05269394 .
115. Washington University School of Medicine. DIAN-TU primary prevention platform (remternetug). ClinicalTrials.gov NCT05552157 (and NCT06647498). Available at: https://clinicaltrials.gov/study/NCT05552157 .
116. Washington University School of Medicine. DIAN-TU-002 remternetug arm. ClinicalTrials.gov NCT06647498. Available at: https://clinicaltrials.gov/study/NCT06647498 .
117. Banner Health. Donanemab and RG6289 in PSEN1 E280A mutation carriers. ClinicalTrials.gov NCT06996730. Start 08/07/2026. Available at: https://clinicaltrials.gov/study/NCT06996730 .
118. AviadoBio Ltd. AVB-101 in FTD-GRN (ASPIRE-FTD). ClinicalTrials.gov NCT06064890. Available at: https://clinicaltrials.gov/study/NCT06064890 .
119. Passage Bio, Inc. PBFT02 in FTD with GRN or C9orf72 variants (upliFT-D). ClinicalTrials.gov NCT04747431. Available at: https://clinicaltrials.gov/study/NCT04747431 .
120. Vesper Biotechnologies ApS. VES001 SAD/MAD in healthy volunteers and asymptomatic GRN carriers. ClinicalTrials.gov NCT06226064. Available at: https://clinicaltrials.gov/study/NCT06226064 .
121. Kjolby M. Open-label multiple dosing study of VES001 in asymptomatic GRN carriers. ClinicalTrials.gov NCT06705192. Available at: https://clinicaltrials.gov/study/NCT06705192 .
122. Passage Bio, Inc. Corporate presentation, March 2026 (Exhibit 99.2, Form 8-K). Available at: https://www.sec.gov/Archives/edgar/data/1787297/000110465926022539/pasg-20260303xex99d2.htm .
123. Leroy M, Bertoux M, Skrobala E, Mode E, Adnet-Bonte C, Le Ber I, et al. Characteristics and progression of patients with frontotemporal dementia in a regional memory clinic network. Alzheimers Res Ther. 2021;13(1):19. doi:10.1186/s13195-020-00753-9. PMID: 33419472.
124. Dominantly Inherited Alzheimer Network (DIAN). DIAN — institutional website and DIAN Expanded Registry. Available at: https://dian.wustl.edu/ and https://dianexpandedregistry.org/ . Accessed 10 Oct 2026.
125. Genetic Frontotemporal Dementia Initiative (GENFI). Institutional website. Available at: https://www.genfi.org/ . Accessed 10 Oct 2026.
126. ARTFL LEFFTDS Longitudinal Frontotemporal Lobar Degeneration (ALLFTD). Institutional website. Available at: https://www.allftd.org/ . Accessed 10 Oct 2026.
127. Alzheimer Europe. Institutional website. Available at: https://www.alzheimer-europe.org/ . Accessed 10 Oct 2026.
128. Alzheimer Portugal. Institutional website. Available at: https://alzheimerportugal.org/ . Accessed 10 Oct 2026.
129. Associação Brasileira de Alzheimer (ABRAz). Institutional website. Available at: https://abraz.org.br/ . Accessed 10 Oct 2026.
130. The Association for Frontotemporal Degeneration (AFTD). Home page. Available at: https://www.theaftd.org/ . Accessed 10 Oct 2026.