Early onset familial Alzheimer’s disease

Methods.

Patients with AD, an AAO of <61 years, and a family history of at least one affected first-degree relative were recruited from clinical and research referrals to the Dementia Research Group at the National and St. Mary’s hospitals, London, UK. The patients were categorized into those with a clinical antemortem diagnosis of AD based on their history, examination, and investigations (probable AD using National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association [NINCDS-ADRDA] criteria18) and those who had additionally undergone neuropathologic examination and showed the signature lesions of AD (definite AD18). The investigations included screening blood tests for the reversible causes of dementia, brain CT or MRI, EEG, neuropsychology, and CSF examination. The AAO was defined as the age at which cognitive impairment was sufficient to interfere with social functioning. Patients from families with previously identified mutations were excluded. Following approval from both the National and St. Mary’s hospitals ethics committees and informed consent from the patients (or where appropriate assent from the caregivers) we obtained whole blood specimens from 31 unrelated probands. Genomic DNA was prepared from peripheral blood leukocytes using a Nucleon II extraction kit (Nucleon Bioscience, Strathclyde, UK).

Mutation analysis was done on genomic DNA by direct sequencing of both strands of PCR-amplified coding exons of PSEN1 (complete open reading frame), exons 16 and 17 of APP, and exons 4, 5, and 7 of PSEN2. Primer sequences were designed according to Genbank entries and are available on request. Amplification was done using a standard protocol, and PCR products were purified using spin columns before sequencing of exons as described previously.19 Mutation screening was stopped if a known causative mutation was demonstrated in any proband. In all other patients all three genes were sequenced. Confirmation of missense mutations was performed by differential enzymatic restriction digests, or by allele-specific oligonucleotide hybridization (ASOH) when a suitable restriction site was absent (table 1).

When a novel sequence variant was found in the proband, additional pedigree members were genotyped by sequencing the appropriate exon to establish cosegregation between the mutation and the disease. In most families, however, this analysis was not possible, because there were no other living affected relatives. In addition, 100 normal white controls were genotyped to determine the prevalence of a novel finding in the normal population. For probands in whom we failed to identify a sequence variant we also determined the APOE genotype using a standard one-stage PCR technique.20

Results.

Seventeen patients with probable and 14 patients with definite AD (NINCDS-ADRDA criteria) were recruited.18 The mean AAO was 46.9 years (median 45 years; range 33 to 60 years). The majority of patients (23 of 31) belonged to families with three or more affected family members in at least two generations and thus had a family history fulfilling recognized criteria for autosomal dominant inheritance.11 The remaining eight patients had a single affected first-degree relative; for six patients this was a parent and for two patients a sibling. Demographic characteristics of the probands and details of the sequence variants detected are shown in table 1. Sixteen PSEN1 sequence variants were identified in 17 (55%) probands including eight recognized PSEN1 mutations (see AD Mutation Database [http://molgen-www.uia.ac.be/ADMutations/] for details): Y115C, M146I, L153V, L171P, G184D, A260V, R269H, and the Δ4 mutation (Thr 113/114 ins), a novel three–base pair deletion resulting in the effective deletion of codon 167 (Δ167) and seven novel missense point changes (Y154C, I229F, F237L, L235V, C263F, R377M, and G378V). The Δ167 deletion was characterized by the deletion of the third base pair of codon 167 and the first and second base pairs of codon 168. Three APP sequence variants were identified in 4 (13%) probands, including the recognized V717I (London) and V715A (German) mutations (see AD Mutation Database) and a novel missense point change: H677R.

All novel sequence variants identified were absent from 100 healthy unrelated white control patients. In three families DNA was available from affected relatives, allowing us to test cosegregation of the sequence variant found in the proband with the disease. In families 134 (PSEN1 G378V) and 267 (PSEN1 L235V) we were able to demonstrate the presence of the sequence variant in all affected patients for whom DNA was available. In family 134 the G378V mutation was present in three patients, and in family 267 the L235V mutation was present in two patients. However, in family 209, the APP H677R sequence variant was absent from the affected sibling; both siblings were APOE ε3 homozygotes. Thus in our series 21 of 31 (68%) probands were associated with a sequence variant in APP or PSEN1. If because of the inherent uncertainty of a probable AD diagnosis we confine ourselves to those with definite AD who also fulfill criteria for autosomal dominant inheritance, then 9 of 11 (82%) probands are associated with a sequence variant in APP or PSEN1 (six recognized mutations, two mutations that cosegregate with disease, and one novel sequence variant only demonstrated in a single proband). Similarly, 10 out of 13 (77%) probands with probable AD (NINCDS-ADRDA) who also fulfilled criteria for autosomal dominant inheritance were associated with a sequence variant in APP or PSEN1 (seven recognized mutations and three novel sequence variants only demonstrated in a single proband).

The mean age of probands with mutations was 44.7 years, whereas those without mutations were significantly older at 55.4 years (p = 0.001). The mean age of probands with PSEN1 mutations was not significantly younger than those with APP mutations (43.7 years vs 50.3 years; NS).

In 10 probands we did not identify a sequence variant in PSEN1 or the examined exons of APP and PSEN2; three probands were homozygous and five patients were heterozygous for APOE ε4 (table 2). The frequency of the APOE ε4 allele was 0.80 in these probands without mutations.

Discussion.

AD is characterized by progressive deposition of Aβ₄₀₍₄₂₎ (the 40–42 residue amyloid β protein) in brain parenchyma and cerebral blood vessels. Several mutations in APP have been identified (for review, see AD Mutation Database and reference 21). These mutations are located close to the major APP processing sites, either adjacent to the Aβ domain (the β and γ secretase cleavage sites) or within the Aβ domain itself (the α secretase cleavage site) as shown in the figure. The pathophysiologic mechanisms of these mutations vary. The only known mutation near to the β secretase cleavage site is the 670/671 (Swedish) double mutation, which consists of the substitution KM → NL. This mutation results in increased total Aβ, by increasing both Aβ₄₀ and, to a lesser extent, Aβ₄₂.16 Those near the γ secretase cleavage site generally increase production of the more amyloidogenic Aβ₄₂,16 although the V715M (French)17 mutation results in a reduction of Aβ₄₀ without affecting Aβ₄₂ production, suggesting that it is the increase in the ratio of Aβ₄₂ to Aβ₄₀ that is important rather than the absolute amount of Aβ₄₂. A more recent study of six mutations near the γ secretase cleavage site, including the V715A (German) mutation, confirmed this finding and further demonstrated an inverse correlation between these ratios and AAO.22

• Download figure
• Open in new tab
• Download powerpoint

Figure. The amyloid precursor protein molecule with localization of the Aβ and p3 proteins, showing pathogenic mutations and the novel sequence variant identified in this study.

Pathogenic intra-Aβ mutations (see the figure) are generally associated with amyloid accumulation in cerebral blood vessels in addition to amyloid plaque formation. Carriers of the E693Q (Dutch) mutation have intracerebral hemorrhages,23 whereas patients with the A692G (Flemish) mutation frequently have intracerebral hemorrhage but may survive to develop a progressive AD-like dementia.24 The E693G (Arctic) mutation is associated with a more typical early onset AD phenotype without evidence of intracerebral hemorrhage. The three known mutations at codon 693 (see the figure) result in reduced Aβ₄₂ levels in conditioned media, which contrasts with increased levels of both Aβ₄₀ and Aβ₄₂ in media for the A692G (Flemish) mutation.25 For the E693G (Arctic) mutation this reduction in Aβ was observed even in presymptomatic mutation carriers 20 to 30 years before the expected onset of the disease. This together with the finding that Arctic Aβ formed protofibrils at a much higher rate and in larger quantities than wild type Aβ is the basis of a proposed alternative pathogenic mechanism for this kindred where protofibrils result in the accelerated buildup of insoluble Aβ.25 We are not certain of the pathogenicity of our novel intra-Aβ sequence variant H677R, which was present in one of two siblings. The diagnosis in both siblings is secure because they both underwent neuropathologic examination that confirmed the signature lesions of AD. Neither sibling had an APOE ε4 allele, a mutation in PSEN1, or the sequenced exons of PSEN2 to account for their disease. Nothing is known about the previous generation in this family so it remains a possibility that the sequence variant is an autosomal dominant pathogenic mutation accounting for the disease in one of the siblings with the other sibling being a phenocopy; i.e., having sporadic AD. Alternatively, the sequence variant is a rare nonpathogenic polymorphism, absent from 100 normal white controls, and both siblings have sporadic AD. We await further structural modeling and Aβ metabolism studies for this novel sequence variant.

Of the seven novel PSEN1 sequence variants we were able to demonstrate cosegregation for two—G378V and L235V—and these are likely to be pathogenic mutations. A further sequence variant, C263F, occurred at a residue where mutations have previously been reported.26 In keeping with previous reports the remaining five missense point changes occur in transmembrane domains.21

Mutations of APP and PSEN1 show almost complete penetrance by the age of 60 years with the exception of four PSEN1 missense mutations: A79V,11 I143F,27 H163Y,28 and H163R.29 The level of penetrance and mechanism of pathogenicity remain to be determined for the novel sequence variants described here. We did not identify any mutations in exons 4, 5, and 7 of PSEN2, and none has yet been reported in the UK.

The proportion of PSEN1 and APP sequence variants described in the current study is much higher than the 18% identified in a previous report11 in which the same criteria were used to define autosomal dominant inheritance as in this study. Our findings are in accord with a French FAD study in which 24 of 34 (71%) families fulfilling stricter three-generational criteria to define autosomal dominant inheritance were found to have mutations in APP or PSEN1.12

Ten probands did not have a mutation in APP, PSEN1, or PSEN2. Despite there being two very early onset probands (families 278 and 373) in this group it is of interest to note that the mean age of these probands was 10 years older than those with mutations (p = 0.001) and that 3 (30%) of these probands were homozygous and 5 (50%) were heterozygous for APOE ε4. Thus the observed frequency of the APOE ε4 allele of 0.80 was higher than the 0.15 observed in the general white population.9,12⇓ In view of their older AAO it is plausible that some of these patients may in fact have early onset AD due to the presence of one or two ε4 alleles. The proband from family 278, with a three-generational family history, did not have an ε4 allele, and the absence of an APP, PSEN1, or PSEN2 mutation in this neuropathologically confirmed family could be explained by a mutation located outside the regions that we analyzed, or by the involvement of other causative and risk genes for FAD. Although the proband from family 373 was homozygous for ε4, it seems unlikely that this is APOE driven AD and again a mutation outside the regions analyzed or the involvement of another genetic factor seems more likely.

Early onset autosomal dominant AD is a rare disease with an estimated prevalence of 5.3 persons per 100,000 at risk.12 Our data are therefore from a highly selected population of patients who have been carefully characterized and fulfill either antemortem NINCDS-ADRDA criteria for a diagnosis of probable AD or have neuropathologically verified AD. For the subset of patients with neuropathologically confirmed definite AD, who in addition fulfilled recognized criteria for autosomal dominant inheritance, 9 of 11 (82%) had an associated sequence variant in either APP or PSEN1. Six of these variants were previously reported pathogenic mutations, two were novel changes for which we were able to demonstrate cosegregation, and one was a novel change demonstrated in the proband only. For the probands who had probable AD and who also fulfilled additional criteria for autosomal dominant inheritance, 10 of 13 (77%) had associated sequence variants in APP or PSEN1. Seven of these sequence variants were recognized pathogenic mutations and three were novel sequence variants where we were not able to demonstrate cosegregation with disease. It is these novel sequence variants demonstrated in only a single proband that cause the greatest difficulties in clinical practice, particularly for PSEN1 mutations, as many are private; i.e., they are only found in a particular individual or family.11,30⇓ It is therefore difficult to interpret the causative effects of the novel sequence variants described and this will require extending the families where possible, looking at the pathophysiology of the mutations, especially their effect on APP processing, and waiting for further reports.

Because a molecular genetic diagnosis of an inherited disorder affects not only the patient, but also the entire family, genetic counseling must be an essential component of the diagnosis of inherited disorders.31 In the case of predictive testing this will invariably involve trained clinical geneticists using a (modified) Huntington disease protocol.31 For diagnostic testing the literature11,12⇓ and our data would suggest that this should be confined to those with a clearly positive family history of autosomal dominant inheritance and an early AAO in all family members. Mutational screening of affected individuals without such an autosomal dominant family history is rarely indicated unless there is a very characteristic phenotype and inadequate family history.31 Routine counseling before diagnostic testing would include making the patient and the family aware that a mutation may not be found and that a novel finding may be difficult to interpret.

In a clinical setting the emphasis must remain on making an accurate clinical diagnosis and excluding the many reversible conditions that can masquerade as AD. However, it should not be forgotten that families who do not have a proven pathogenic mutation or who do not wish to undergo diagnostic genetic screening should not be denied the benefits of genetic counseling based on the autosomal dominant risk model. As in Huntington disease, in the absence of a disease-modifying treatment, the take up of predictive testing is likely to be low.32,33⇓