|Home | About | Journals | Submit | Contact Us | Français|
Alzheimer's disease (AD) causes devastating cognitive impairment and an intense research effort is currently devoted to developing improved treatments for it. A minority of cases occur at a particularly young age and are caused by autosomal dominantly inherited genetic mutations. Although rare, familial AD provides unique opportunities to gain insights into the cascade of pathological events and how they relate to clinical manifestations. The phenotype of familial AD is highly variable and, although it shares many clinical features with sporadic AD, it also possesses important differences. Exploring the genetic and pathological basis of this phenotypic heterogeneity can illuminate aspects of the underlying disease mechanism, and is likely to inform our understanding and treatment of AD in the future.
Observations that cases of early-onset Alzheimer's disease (AD) may cluster within families were made as early as the 1930s, and almost two decades have now passed since the first reports confirming that the disease could be caused by autosomal dominantly inherited genetic mutations . Initially identified in the amyloid precursor protein (APP) gene on chromosome 21 , mutations were also subsequently found in the genes encoding presenilin (PSEN)1 on chromosome 14  and PSEN2 on chromosome 1 . Over 200 changes have now been described in these genes, although some are thought to represent nonpathogenic polymorphisms and in other cases, the pathogenic nature of the change is as yet unclear. More recently, duplications of the APP gene have been identified as an additional cause of familial AD (FAD) . Mutations in PSEN1 account for the majority of cases of FAD and it is in PSEN1 that the greatest numbers of mutations have been found, with 177 recorded to date compared with 32 and 14 in APP and PSEN2, respectively (Figures 1, ,2,2, ,3)3) . Collectively, these mutations are thought to account for up to 0.5% of AD in general . The degree to which FAD may be a paradigm for the more common, late-onset, sporadic form of the disease is uncertain but the insights gleaned from the study of FAD have catalyzed much of our current knowledge of the pathobiological basis of the disease. The development of cellular models, and transgenic mice harboring mutations in APP and the presenilins has allowed the disease mechanism to be explored in detail and to be probed with potential therapeutic agents. As these therapies come closer to clinical application, the debate regarding how generalizable the changes of FAD are to those of sporadic AD grows in significance. Unraveling the pathway through which specific genetic mutations translate into the manifestations of clinical AD and how the resulting phenotypes may vary, are clearly important areas of investigation. Although much has been published describing novel FAD gene mutations and exploring their functional effects at a molecular level, there have been far fewer detailed reports of the resulting clinical phenotypes [6,7]. In this article, we examine how closely the phenotype of FAD resembles and differs from that of sporadic AD. We emphasize the importance of appreciating the degree of phenotypic heterogeneity within FAD. Finally, we discuss several mechanisms by which the clinical picture, and underlying pathology, may be influenced by genotype.
The pathological hallmarks of AD in both familial and sporadic cases are extracellular plaques of amyloid-β (Aβ) and intraneuronal inclusions of hyperphosphorylated tau. It is now well recognised that mutations in all three genes known to cause FAD ultimately result in the enhanced production and/or deposition of Aβ . The mechanisms through which they achieve this are complex and variable. APP is a transmembrane protein whose function is not yet fully determined, although it appears to play a role in neural plasticity and the regulation of synapse formation. APP may be proteolytically cleaved by α-, β- and γ-secretases, and during its normal metabolism, it can undergo cleavage via alternative pathways. Cleavage by α-secretase is considered to be part of the nonamyloidogenic pathway in APP processing. It precludes Aβ formation as it cleaves within the segment of APP that would otherwise give rise to Aβ. Upon cleavage by α-secretase, APP releases its extracellular domain, a fragment called APPsα, which is known to have neurotrophic effects. Alternatively, APP may undergo sequential cleavage by β- and γ-secretase. Extracellular cleavage by β-secretase (also known as β-site of APP cleaving enzyme) generates a soluble extracellular fragment and is followed by the cleavage of APP within its transmembrane domain by γ-secretase. The unusual property of γ-secretase is that it may cleave APP at a variety of different sites, generating a peptide from 39–43 amino acids in length: the Aβ protein. PSEN1 forms part of the γ-secretase complex and the majority of pathogenic PSEN1 mutations are located in areas thought to lie in the vicinity of PSEN1's transmembrane domains. Under normal conditions, the majority of Aβ takes the form of the 40-amino acid peptide Aβ40. There is some additional production of the longer moiety, Aβ42, which accounts for 5–15% of the total pool of Aβ. Smaller quantities of other Aβs, both longer and shorter, may also be observed. Aβ peptides, particularly Aβ42, can cause toxic effects on neuronal and synaptic function as intracellular oligomers, and seed to form extracellular precipitates in the form of amyloid plaques. Aβ42 is the most amyloidogenic form of the peptide; its hydrophobic nature makes it more susceptible to conformational change leading to amyloid fibrillogenesis. Most FAD-causing mutations appear to either increase Aβ40 and Aβ42, or alter the ratio between them. However, the discovery of the ‘Arctic’ APP mutation (E693G) in a Swedish family, which is located away from the secretase sites, led to the suggestion of an alternative pathogenic mechanism. Aβ40 and Aβ42 are actually found in decreased levels with this mutation; however, they have an increased propensity to form proto-fibrils, which may accelerate their deposition in intracellular or extracellular locations, or both .
The most striking difference between familial and sporadic AD is the younger age at onset in individuals harboring FAD mutations. That the disease should manifest at an earlier age is not surprising, given the increased Aβ deposition that these mutations cause. The youngest ages at onset are seen in families with PSEN1 mutations, which typically fall within the range 35 to 55 years. APP mutations tend to give rise to symptom onset between 40 and 65 years and PSEN2 mutations between 40 and 70 years. It has been suggested that, as well as being associated with a younger age at the onset of disease, PSEN1 mutations give rise to shorter disease durations (4.8–6.8 years) than PSEN2 and APP mutations (range: 4.4–10.8 and 9.0–16 years, respectively) . However, the survival data of different series' demonstrate considerable variation. Although the disease course of FAD is often considered to be more aggressive, some reports find comparable disease durations for the familial and sporadic forms of AD . Drawing direct comparisons between disease duration in familial and sporadic AD is problematic as survival is likely to be affected by age at onset itself. Younger, healthier patients may survive longer at more severe stages of disease than older patients with sporadic AD, who are generally frailer and often have additional comorbidities. Another major difficulty in interpreting survival data is that estimates for age at symptom onset are highly subjective and unreliable. Symptoms tend to develop insidiously and, in some cases, may pass unrecognized for quite some time. Conversely, as awareness of the heritable nature of the disease heightens with subsequent generations, symptoms are likely to be noticed at an earlier stage. This probably contributes to the range in age at onset that is witnessed both between and within families with the same mutation. However, other genetic and epigenetic factors are also likely to be important in modulating the age at onset.
Pathological studies have demonstrated that the proportion of brain tissue occupied by Aβ42-containing plaques is significantly higher in PSEN1-associated AD than sporadic AD. Although there does not appear to be a difference in the amount of tissue containing Aβ40 between these two disease groups, both PSEN1-associated AD and sporadic AD show the same relationship between Aβ deposition and APOE genotype; individuals who carry an ε4 allele demonstrate a greater proportion of Aβ40 deposition in both familial and sporadic disease. It has been suggested that, in the presence of the APOε4 isoform, the threshold for Aβ40 fibrillization is decreased, fostering the deposition of Aβ40 onto Aβ42 -containing plaques . It has also been demonstrated that possession of the APOε4 allele in FAD, secondary to APP mutations, is associated with a decreased age at onset, as is the case in sporadic AD. In PSEN1 mutations, this modifying effect on age at onset by the ε4 allele was not previously thought to occur . However, when a single large Colombian kindred with the E280A PSEN1 mutation was examined, possession of the APOε4 allele did appear to be associated with an earlier age at onset . The true nature of the relationship between APOE status and FAD mutations will probably only be elucidated when much larger cohorts are studied.
The majority of individuals with FAD have similar clinical presentations to those with sporadic AD, comprising an early impairment of episodic memory, which gradually progresses to involve multiple cognitive domains. Longitudinal studies of individuals at risk of FAD demonstrate that the earliest neuropsychological manifestations are a fall in verbal memory and performance IQ scores, occurring even 2–3 years presymptomatically . The neuropsychological profile of FAD at a group level has also been found to show relatively well-preserved naming, spelling and visuoperceptual skills compared with sporadic AD in the early stages [11,16]. There can therefore be a particularly long prodromal phase in FAD, where subtle memory problems are the only clinical feature for several years . Presymptomatic cognitive deficits are accompanied by neuroimaging changes, with accelerating rates of whole-brain and hippocampal atrophy in the 5 years prior to symptom onset [17,18].
The early stages of sporadic AD have been harder to study. Historically, individuals with memory concerns have tended to only seek medical attention when their symptoms are well established. However, we are currently witnessing a change in the tide of public opinion towards an appreciation of the importance of early diagnosis. There is increased recognition that most patients with sporadic AD are likely to have experienced a period of relatively isolated, usually amnestic, deficits termed mild cognitive impairment (MCI), prior to their diagnosis of AD . The term MCI is usually used to refer to individuals with cognitive decline, which is more than anticipated from normal aging but insufficient to fulfill criteria for dementia. The term will thus be applied to a heterogeneous group, not all of whom will develop AD, and there is currently a great research effort aiming to improve the characterization of these individuals and refine the definition of the MCI label. Nevertheless, there are now many lines of evidence supporting the notion that sporadic AD is preceded by a long prodromal stage. The canonical FAD phenotype, manifesting with early progressive impairment of episodic memory, can perhaps then be viewed as equivalent to the typical amnestic AD phenotype, preceded by MCI, which is commonly seen in sporadic disease.
Many atypical presentations of FAD have, however, been reported, and there can be significant heterogeneity both between and within families regarding clinical phenotype. Heterogeneity, however, also exists within sporadic AD and as recognition of these more unusual variants increases, they may be found to occur more frequently than was previously appreciated.
Several PSEN1 mutations have been reported to present with language impairment, including G209V, H163Y and R278I, which was detected in one patient given an initial clinical diagnosis of progressive nonfluent aphasia . Language deficits have been described in association with many other mutations later in the disease course, as is also often the case in sporadic AD. There is a growing body of research aiming to better characterize individuals with neurodegenerative disease who have predominant impairments of language function at presentation. Primary progressive aphasia is traditionally thought to signify a disease in the spectrum of frontotemporal lobar degenerations. However, a ‘logopenic’ variant of primary progressive aphasia has recently been proposed, in which AD appears to be the more common underlying pathology .
In sporadic AD, symptoms of agitation, depression, delusions and hallucinations occur commonly, particularly in the later stages of disease. Many PSEN1 mutations are associated with behavioral and psychiatric symptoms during the course of the disease and, on rare occasions, these can be the presenting features. Prominent early behavioral symptoms have also been reported in association with several PSEN2 mutations, including M239V, T122R and Y231C . An underlying pathological diagnosis of AD has been obtained for some PSEN1 mutations where a behavioral presentation has been described, including V89L , L226F  and M139V . It has been suggested that a distribution of pathology involving structures in frontal–subcortical circuits, such as the thalamus and striatum, may contribute to the behavioral phenotype in some of these patients . Pathological confirmation for other PSEN1 mutations, such as L113P, remains to be established . In some cases, the presence of behavioral and language features has been prominent enough to prompt a clinical diagnosis of frontotemporal dementia (FTD), as was the case for the PSEN1 insR352 genotype [27,28]. It was initially hypothesized that this PSEN1 genotype may be capable of inducing a distinct neurodegenerative illness, more akin to FTD than AD. Intriguingly, not only did the members of the kindred have an FTD phenotype but when an affected individual eventually came to autopsy, ubiquitin pathology typical of FTD was found. The subsequent discovery of the progranulin gene as a cause of familial FTD led to a re-evaluation of this case, which was indeed found to harbor a progranulin mutation . With the benefit of hindsight, the change in PSEN1 was then reconsidered to be a nonpathogenic polymorphism. This case highlights the importance of exercising caution when ascribing pathogenicity to rare PSEN1 variants, especially when the phenotype is unusual.
Visuospatial and visuoperceptual deficits commonly occur in the later stages of both sporadic AD and FAD. In sporadic AD, a posterior cortical atrophy variant may occur, where marked visual processing deficits are the presenting features and memory is relatively well preserved until later in the disease . To date, this phenotype has not been observed in FAD.
Neurological symptoms and signs tend to be more prominent in FAD than sporadic AD, and have been recognized as a feature of the disease ever since it was first reported. Myoclonus and seizures are common. Spastic paraparesis, extrapyramidal signs and cerebellar ataxia have also been described, although less frequently. All of these features are particularly prominent in kindreds with a very early-onset of symptoms – below the age of 40 . However, it is important to note that significant phenotypic heterogeneity is observed, both between families sharing a common mutation and within the affected individuals of a single family. Whilst all of these unusual features do appear to occur more frequently in FAD than sporadic AD, the fact that younger patients often survive to more severe stages of disease may, in part, contribute to the trend. Furthermore, the preference for reporting atypical cases of FAD over those with sporadic AD may also play a role. In some reports, particularly those including longitudinal assessments, it is clear that neurological signs and symptoms developed at an early stage of the disease, and there are cases where these features predate cognitive impairment (Table 1). However, other reports of ‘early’ atypical features should be read with caution as sometimes the disease may be well established by the time a patient first presents to a clinician.
Myloclonus can be a particularly prominent and early sign in FAD, in contrast to sporadic AD, where it tends to become apparent later in the course of the disease. The same pattern appears to occur with seizures, which can occur more frequently and at an earlier stage in the disease in FAD than in sporadic AD . Myoclonus and seizures are observed more often in very early-onset cases of FAD. In fact, in one family with the PSEN1 L166P mutation and a young age at onset (between 29 and 35 years), tonic-clonic seizures predated the onset of cognitive impairment . Fine finger myoclonus has also been noted to occur prior to the onset of objective memory impairment, although it may not necessarily be apparent to the patient or their family at this mild early stage . Interestingly, there is a cluster of mutations associated with early myoclonus and seizures in the region encoding transmembrane domain III of the PSEN1 gene (S169L, S169P and S170F). Early myoclonus and seizures are also observed in individuals with APP mutations and particularly with APP duplications (see later).
Spastic paraparesis has been reported to occur in association with over 20 different PSEN1 mutations . Symptoms of spastic paraparesis may occur simultaneously with cognitive deficits or may predate them by several years. Progressive spastic paraparesis occurring for over a decade prior to the onset of cognitive impairment has been reported in association with the PSEN1 R278K mutation . There is a particular association of spastic paraparesis with PSEN1 exon 9 deletions, although not all individuals with these deletions demonstrate the phenotype . In those who do, there are frequently characteristic histological findings of large, rounded plaques in the frontal cortex known as ‘cotton wool plaques’ . Composed of Aβ40 and Aβ42, cotton wool plaques usually lack a compact amyloid core and demonstrate very little neuritic change or glial cell component. The spastic paraparesis phenotype of FAD is often not associated with any specific imaging findings. However, significant white-matter abnormalities on cranial MRI, presumed secondary to amyloid angiopathy, have been observed in a family exhibiting spastic paraparesis and cotton wool plaques secondary to the E280G PSEN1 mutation .
Several hypotheses have been proposed to explain why certain PSEN1 mutations may give rise to the spastic paraparesis phenotype. The observation that most of the mutations associated with spastic paraparesis cause exceptionally high levels of Aβ42 suggests that a dosage effect may be operating . Intraneuronal Aβ may also play a role; this is a more frequent finding in AD patients with cotton wool plaques. As intraneuronal Aβ is known to induce only a weak immune response, it has been postulated that cotton wool plaques may result from a combination of increased Aβ production and decreased clearance . However, the presence of cotton wool plaques per se is not enough to explain the phenotype as these have also been found in patients lacking signs of spastic paraparesis . Furthermore, the presence of spastic paraparesis is highly variable, even between the affected individuals of a single family sharing identical mutations and similar environmental exposures. Therefore, there may be additional, as yet unidentified, factors operating in these patients that drive the pathology towards structures, like the corticospinal tracts and motor cortex, which are usually spared in both familial and sporadic AD.
Extrapyramidal signs have been reported in association with several different PSEN1 mutations, particularly those causing very early onset of the disease and other neurological signs, such as spastic paraparesis . A novel PSEN2 mutation (A85V) has also recently been reported where Parkinsonism was a feature . In the majority of cases, Parkinsonism does not become apparent until several years into the clinical course. However, an extrapyramidal syndrome characterized by rigidity, bradykinesia and an abnormal posture and gait was present from the onset of cognitive symptoms in the first reported kindred harboring the PSEN1 G217D mutation . The autopsy findings in this family included, in addition to neurofibrillary tangles, amyloid angiopathy and neuritic plaques, neuronal rarefaction in the substantia nigra and numerous cotton wool plaques. These were distributed throughout the cerebral cortex but were also concentrated in subcortical structures, particularly the striatum. As the authors suggest, the topography of pathology in these patients could account for their clinical presentations. Furthermore, these cases illustrate how the separate, yet often overlapping, phenotypes of spastic paraparesis and Parkinsonism in FAD may share a common pathological substrate of cotton wool plaques located in specific anatomical areas.
Another histological feature that can underlie a Parkinsonian presentation of FAD is Lewy body pathology (LBP). Both LBP and cotton wool plaques were detected in a patient exhibiting a dopa-responsive Parkinsonian syndrome, who developed cognitive symptoms 7 years after the onset of his movement disorder. The patient was found to possess a novel three base-pair deletion (DelT440) in PSEN1 . Widespread neocortical LBP has been demonstrated in a kindred with very early-onset FAD secondary to an S170F PSEN1 mutation, in which the age at symptom onset was as young as 26 years . In this family, myoclonus, seizures and extrapyramidal signs were features of the disease. Parkinsonism, seizures, a very early onset of disease and LBP were also reported in association with the M233V mutation in PSEN1 . However, not all FAD patients with extrapyramidal signs are found to have LBP and equally, many patients with LBP do not manifest Parkinsonism clinically. In fact, concomitant LBP is a frequent finding in neuropathological examinations of patients with both familial and sporadic AD, occurring in up to 60% of patients with each disease according to some series [47,48]. Studies including analysis of the frequency of LBP in different FAD gene mutations have suggested that PSEN1 mutations are significantly more likely to have associated LBP than PSEN2 mutations . Furthermore, there appear to be mutation-specific effects on the development of LBP in FAD. Certain PSEN1 mutations, such as A260V, are associated with diffuse LBP whereas in others, like G209V, the LBP seems to occur in the amygdala only. In other PSEN1 mutations, such as A246E, there tends to be very little LBP at all . LBP has also been observed in association with APP mutations and, with both the presenilins and APP, no correlations between mutation position and the development of this pathological picture have been identified . It has been suggested that the presence of diffuse LBP in FAD is associated with longer durations of disease than when the LBP is restricted to limbic areas . However, a detailed understanding of the significance of LBP in FAD, and how LBP relates to duration and manifestation of disease remains to be elucidated, particularly as many studies were conducted before synuclein antibodies became available.
Variable cerebellar Aβ deposition has been observed in both familial and sporadic AD [50,51]. In several PSEN1 mutations, particularly E280A , I143T , M139V  and G209V , severe involvement of the cerebellum by Aβ plaque deposition has been noted. This has also been found in FAD secondary to the APP717 mutations  and in the brains of elderly patients with Down's syndrome [57,58]. Even in cases of FAD with similar amounts of Aβ deposition to sporadic AD in the cerebellum, cerebellar Purkinje cell loss and reactive astrocytosis are significantly more severe in FAD than sporadic AD . Pathological involvement of the cerebellum may occur without cerebellar signs featuring in the clinical presentation . However, progressive cerebellar ataxia has been reported as preceding cognitive impairment in the S170F  and P117A  mutations in PSEN1. Furthermore, cerebellar signs have been observed during the disease course in a range of other mutations in PSEN1, including L282V . This mutation is notable for its association with an unusually high amount of amyloid angiopathy, which was found to involve the cerebellum in the individual manifesting cerebellar signs.
Two distinct histopathological profiles have been identified in FAD caused by PSEN1 mutations, and the type of pathology appears to be driven by the position of the mutation within the gene . Mutations occurring before codon 200 are more frequently associated with a type 1 pattern of pathology, whilst mutations beyond codon 200 tend to have type 2 pathology, which includes more severe amyloid angiopathy. Type 1 pathology more closely resembles the pathology seen in sporadic AD, with many diffuse and cored plaques, and few white-matter plaques. Amyloid angiopathy is only mild to moderate, and is usually confined to leptomeningeal vessels. In type 2 pathology, the number and distribution of diffuse and cored plaques is similar but these tend to be much larger and concentrated around amyloidotic arteries. Amyloid angiopathy is nearly always severe, involving both leptomeningeal and intraparenchymal arteries. Aβ40 deposition tends to be increased in those with type 2 pathology, and this effect is independent of the APOE genotype. The distinction between the two types of pathology is not absolute; some mutations occurring before codon 200 have been associated with widespread amyloid angiopathy, such as E184D  and DelI83/M84 . Equally, the N405S mutation, occurring after codon 200, appears to produce limited amyloid angiopathy . However, there do appear to be clinical implications of the different histopathological signatures. Individuals with type 1 pathology have a younger age at onset (on average 5 years) and shorter disease duration (~2.5 years) than those with type 2 pathology.
In order to explain the dichotomous manner by which the mutation site influences the development of amyloid angiopathy, it has been suggested that PSEN1 mutations may actually act through two separate pathways . First, they increase the production and deposition of Aβ42 and second, they affect Notch signaling, inducing breakdown of the vascular epithelium. The subsequent leakage of Aβ into brain tissue may seed onto existing deposits of Aβ, producing the large plaques seen around vessels in type 2 pathology and promoting the development of amyloid angiopathy. Interestingly, the residues of PSEN1 that are thought to be functionally most important, namely aspartyl residues at the active site and PSEN1 cleavage site, are located in the region of the gene where mutations are associated with severe amyloid angiopathy. Also important is the observation that these type 2 cases have less aggressive phenotypes, with later ages at onset and longer disease durations.
The majority of APP mutations reported to date lie in the vicinity of the β- and γ-secretase cleavage sites. APP residue 717, located just distal to the C-terminus of the Aβ domain by the γ-secretase site, was recognized early on to represent a mutation hotspot. The ‘London’ APP V717I mutation was the first described, comprising valine substituted by isoleucine. However, valine substituted by glycine, phenylananine or leucine at codon 717 also results in a similar phenotype. Families with mutations at this site were found to share common features of early impairment of episodic memory, dyscalculia, lack of insight and prominent myoclonus and seizures . A double point mutation at APP 670/671 (G to T and A to C), resulting in a lysine methionine to asparagine leucine double substitution, was later identified in two Swedish families . This double mutation frames the other end of the Aβ domain, lying at the N-terminus by the β-secretase site.
Some mutations have also been found within the Aβ coding sequence, and these appear to have specific pathological and clinical features, particularly in relation to amyloid angiopathy (Table 2). The ‘Dutch mutation’ in APP, a missense mutation resulting in a glutamate to glutamine substitution at position 693 (APP E693Q) gives rise to a highly distinct phenotype . These individuals have severe amyloid angiopathy leading to recurrent cerebral hemorrhage. The main clinical features are focal neurological symptoms and signs relating to these cerebrovascular events, although most patients also develop dementia. Pathologically, diffuse parenchymal deposits of Aβ are observed but the hallmark neuritic plaques and neurofibrillary tangles of AD are absent. A neighboring mutation at residue 692 (an A to G substitution) has been reported in a separate family and labeled the Flemish mutation . The affected individuals suffered either hemorrhages or a progressive dementia, or both. A biopsy of a family member with hemorrhages revealed amyloid angiopathy with sparse neuritic plaques but no neurofibrillary tangles. Presymptomatic carriers of the Flemish mutation have also been investigated and found to have significantly increased periventricular and subcortical white-matter lesions on cranial MRI .
Another APP mutation within the Aβ sequence, known as the Iowa mutation (APP D694N) has also been found to be associated with severe amyloid angiopathy on pathological investigation. Widespread neurofibrillary tangles and amyloid plaques with particularly abundant Aβ40 deposition are seen with this mutation. However, clinically, these affected individuals lacked symptoms of recurrent cerebrovascular events and presented with progressive cognitive impairment . A similar picture was found with the recently identified ‘Arctic’ mutation APP E693G, which has a purely cognitive phenotype, typical of AD, despite the presence of marked amyloid angiopathy . The pathological findings in the Arctic mutation also include neurofibrillary tangles and abundant amyloid plaques with characteristic ring-like structures. Why the presence of amyloid angiopathy should have clinical manifestations with some of these mutations but not with others, is intriguing. This is particularly true for the Arctic and Dutch mutations, which occur at exactly the same position (amino acid 693) but give rise to totally different phenotypes. It has been proposed that the explanation may lie in the polarity of the substituted amino acid. The Dutch mutation involves a replacement of a glutamic acid with glutamine, which is polar and hydrophilic. Therefore, Dutch Aβ may have a greater affinity for, and propensity to disrupt, the vessel wall than Arctic Aβ, which is nonpolar and hydrophobic .
The well-established link between Down's syndrome and AD provided early evidence that copy number variation may play a role in AD pathogenesis and it is now known that duplications of the APP gene can give rise to FAD . The phenotype associated with APP duplication can be diverse, even within a single family . However, characteristic features include a high frequency of seizures, up to 57% in one series, and prominent amyloid angiopathy . The size of the duplication does not appear to influence the clinical presentation. The cognitive phenotype resembles that of typical AD, with early progressive impairment of episodic memory. In some families, the clinical presentation is solely cognitive, with no distinguishing neurological features . Knowledge of the prevalence of APP duplications and spectrum of associated clinical features is likely to expand as increasing numbers of centers search for copy-number variation in their cohorts of patients.
Interestingly, a recent postmortem study of an individual with an APP duplication revealed diffuse neocortical LBP with additional amyloid angiopathy and AD pathology . The clinical phenotype of this patient, comprising cognitive decline associated with visual hallucinations and extrapyramidal signs, was also compatible with a diagnosis of dementia with Lewy bodies. As the authors suggest, this case adds weight to the hypothesis that accumulation of Aβ can induce LBP. However, further work must be done in larger cohorts of patients to elucidate more about the clinical manifestations of coexistent LBP in FAD, and to investigate the relationship between β-amyloid and α-synuclein in neurodegenerative disease in general.
In FAD, genetic factors clearly modify the clinical phenotype in striking ways. The manifestations of sporadic AD are also likely to be influenced by genetic factors, although generally in a far subtler manner. The involvement of the APOE gene in sporadic AD has long been known, with possession of the ε4 allele conferring both an increased risk of developing the disease and possibly driving the pathology towards limbic areas . Genome-wide association studies have recently revealed three new ‘risk genes’ for sporadic AD: CLU, PICALM and CR1 [75,76]. An important direction for future work will be to investigate whether these variants have phenotypic associations, particularly in sporadic AD but also in FAD.
Genes known to cause FAD may also play a role in sporadic disease. There are certain PSEN1 polymorphisms, for example E318G, whose pathogenic nature remains uncertain. It has been suggested that some PSEN1 polymorphisms may modify the phenotype of sporadic AD by altering the rate of cognitive decline .
There is recent evidence demonstrating that FAD-causing gene mutations may sometimes be even more directly implicated in cases of apparently sporadic AD: through somatic and germ-line mosaicism . An individual who underwent negative genetic testing on a peripheral blood sample was subsequently found to have the PSEN1 P436Q mutation in DNA extracted from the cerebral cortex. Although this patient had no family history of dementia herself, one of her daughters subsequently developed symptoms and was found to have inherited the mutant haplotype from her mother. Interestingly, the two affected individuals had a difference in age at onset of approximately 15 years and strikingly different clinical phenotypes, although both featured prominent neurological signs. As the authors suggest, these findings have several important implications. Differing levels of mosaicism may account for some of the phenotypic variation seen in FAD. Furthermore, there may be many patients with AD who have pathogenic somatic mutations expressed in brain tissue that are not found on conventional sequencing of DNA extracted from blood lymphocytes.
In the past two decades, significant advances have been made in our understanding of FAD at a molecular, genetic, pathological and clinical level. The number of known mutations has risen exponentially and, with them, an expanding spectrum of clinical phenotypes has been identified. Observations of how certain symptoms and signs tend to cluster with particular mutations has led to an appreciation of how the location of a mutation within a gene can dictate the type of pathology that develops, particularly in relation to amyloid angiopathy. Familial and sporadic AD have been shown to share many similarities. They commonly present with identical symptoms of impaired episodic memory but can both, more rarely, have initial deficits in other cognitive domains. The factors dictating the topography of pathology in both familial and sporadic disease remain to be elucidated. However, possible mechanisms have been proposed. As has always been the case, the insights gained from studying FAD continue to inform our understanding of AD in general. With the prospect of disease-modifying therapy focusing interest on detecting the earliest stages of disease, longitudinal study of the presymptomatic phase in individuals destined to develop FAD will be more important than ever.
There is a growing appreciation of how knowledge gleaned from the study of FAD has the potential to increase our understanding of AD as a whole. In recognition of this, the first international collaborative study of individuals carrying FAD mutations has recently been launched: the dominantly inherited Alzheimer network (DIAN). This multicenter study will collect longitudinal data to explore the clinical, imaging, neuropsychological, blood and cerebrospinal fluid changes that occur in the long period leading up to symptom onset in FAD. By establishing the largest cohort of FAD patients studied to date, it will allow some of the unresolved questions surrounding the relationship between genotype and phenotype in FAD to be addressed. Aided by initiatives like the DIAN, research into FAD is likely to progress at an accelerating pace over the coming decade. The advent of disease-modifying therapy provides great momentum for driving this progress, as it creates an unprecedented need to understand the pathogenesis of AD in more detail. Issues such as the relationship between AD pathology and amyloid angiopathy, for example, have been brought to the forefront by recent immunotherapy trials. It has been found that, in patients treated with Aβ42 immunotherapy, removal of amyloid plaques may be accompanied by accumulation of Aβ42 in perivascular drainage pathways, with an associated transient increase in amyloid angiopathy . It appears that this plaque-derived Aβ42 can be cleared, given sufficient time . However, factors including APOE genotype, are likely to influence how effectively this is achieved. Understanding more about the mechanisms underlying the variable occurrence and clinical manifestation of amyloid angiopathy in FAD may therefore have far-reaching implications. It is clear that in both familial and sporadic AD, a range of atypical clinical phenotypes can occur, and by understanding more about how genetic factors modify the pathological and clinical picture, we may find that the phenotype contains clues suggesting the most appropriate treatment for an individual.
Financial & competing interests disclosure: This work was undertaken at University College London and University College London Hospitals, which received a proportion of funding from the Department of Health's NIHR Biomedical Research Centres funding scheme. The Dementia Research Centre is an Alzheimer's Research Trust Coordination Centre. N Ryan is supported by a Clinical Research Training Fellowship from the Medical Research Council (UK). M Rossor is a site PI for the NIA-funded DIAN study, which receives additional support from DeNDRON, the NIHR clinical research network for dementia and neurodegenerative diseases. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Papers of special note have been highlighted as: