OUP user menu

Novel presenilin 1 mutation with profound neurofibrillary pathology in an indigenous Southern African family with early‐onset Alzheimer’s disease

Jeannine M. Heckmann, Wee‐Chuang Low, Cora de Villiers, Stuart Rutherfoord, Alvera Vorster, Harpal Rao, Christopher M. Morris, Raj S. Ramesar, Raj N. Kalaria
DOI: http://dx.doi.org/10.1093/brain/awh009 133-142 First published online: 21 October 2003


Genetically determined Alzheimer’s disease (AD) is virtually unknown in Africa. We report clinicopathological findings and a presenilin 1 (PS1) mutation associated with early‐onset AD in a large Xhosa family from Southern Africa. Twelve individuals spanning four generations were affected, four of whom underwent clinical and psychometric evaluation. Their phenotype was characterized by memory impairment beginning in the early part of the fifth decade, with progressive dementing illness lasting 6–7 years that did not appear to be modified by the presence of an apolipoprotein E (APOE)‐ϵ4 allele. Initial linkage‐based analysis using known DNA markers suggested allele cosegregation with a locus on chromosome 14. Direct sequencing of the PS1 gene disclosed a novel I143M (ATT to ATG at nucleotide 677) mutation that lies in a cluster in the second transmembrane domain of the protein. Examination of the proband’s brain at autopsy revealed severe AD pathology characterized by neuronal loss, abundant β amyloid (Aβ) neuritic plaques (Aβ42) and neurofibrillary degeneration extending into the brainstem. The phenotype of the I143M mutation was clearly associated with a high degree of neurofibrillary change compared with early‐onset sporadic AD cases. Although sporadic cases of AD do exist in African populations, our study confirms the existence of early‐onset familial AD among indigenous Southern Africans.

  • Africa; Alzheimer’s disease; apolipoprotein E; neurofibrillary tangles; presenilin
  • Aβ = β amyloid peptide; AD = Alzheimer’s disease; APOE = apolipoprotein E; APP = β amyloid precursor protein; DECO = détérioration cognitive observée; MMSE = Mini‐Mental State Examination; PS1 = presenilin 1 gene


Early‐onset familial Alzheimer’s disease (AD) accounts for less than 5% of all cases of AD in Western populations. It is transmitted in an autosomal dominant manner and thus far has been linked to pathogenic mutations in three genes. The first locus identified was that on chromosome 21 encoding the β amyloid precursor protein (APP) (Goate et al., 1991), but mutations in the APP gene only account for ≤1% of early‐onset familial AD (Hardy, 1997; Tanzi and Bertram, 2001). The chromosome 14 locus linked to early‐onset familial AD encodes presenilin 1 (PS1) (Alzheimer’s Disease Collaborative Group, 1995; Sherrington et al., 1995). To date, close to 80 different mutations have been identified in the PS1 gene (www.alzforum.org/res/com/mut/pre/table1.asp). Chromosome 1 bears a third locus that harbours the presenilin 2 (PS2) gene, in which at least six mutations have been described (Levy‐Lahad et al., 1995a; Rogaev et al., 1995). Although allelic variation encoding apolipoprotein E (APOE) influences age at onset in sporadic AD (Farrer et al., 1997), it does not appear to affect age at onset in PS1‐linked familial AD (Van Broeckhoven et al., 1994; Levy‐Lahad et al., 1995b).

Significant heterogeneity has been documented with PS1‐linked familial AD, particularly in age at disease onset (Haltia et al., 1994; Lampe et al., 1994; Kennedy et al., 1995; Crook et al., 1998; Queralt et al., 2002). Comparing kindreds with known mutations may allow the opportunity to correlate the genotypes and phenotypes of mutations in similar clusters or even identical mutations. Such comparisons may assist in determining the influence of epigenetic factors or environmental influences, even in different populations.

This report describes an indigenous South African family of Xhosa origin with early‐onset autosomal dominant PS1‐linked familial AD. The prevalence of AD amongst indigenous Southern African or black populations is contentious and, to our knowledge, familial AD has not been reported in Africans. Previous studies estimate the prevalence of AD to be low (∼1%) in some African populations (Hendrie et al., 1995; Farrag et al., 1998), although pathologically verifiable sporadic cases probably exist in sub‐Saharan countries (Ogeng’o et al., 1996).

Subjects and methods

The proband and family

The proband was first referred for diagnosis in 1997 at the age of 50. Out of five other affected family members of similar age, three were subsequently also evaluated with full clinical examination. In addition, six unaffected family members were personally interviewed. One older unaffected individual was interviewed by telephone and his clinical status was confirmed in conversation with his family practitioner. The pedigree of this large Xhosa family (Fig. 1) was established after several interviews. The age at onset was taken as the first time that different family members collectively considered the individual to be definitely affected, as judged by increased forgetfulness and difficulties in daily living activities.

Fig. 1 A Southern African family with early‐onset AD. The family tree structure has been modified to preserve confidentiality. Filled symbols indicate affected individuals either living or who have since died (stroked symbol). Subjects I‐2 and II‐1 were said to have been affected in the sixth decade of life. The arrow identifies the proband (IV‐19), who presented at age 50. Four affected individuals IV‐10, IV‐13, IV‐16 and IV‐19 underwent full clinical and psychometric evaluation by two investigators (JH, CdV). The mean (± SD) age of onset in these four was 48.5 ± 5.9 years. The age of onset in two other affected members (IV‐4 and IV‐11) was determined to be 47 and 48 years. These two members had exhibited progressive symptoms for 7 years. Linkage‐based analysis using DNA from 13 affected and unaffected members in generations III and IV suggested cosegregation of alleles close to the PS1 locus on chromosome 14. A missense mutation, Ile143Met, was disclosed in PS1 in the four affected but not in the unaffected individuals whose their individual ages at the time of interview or blood collection were 60 (III‐5), 76 (III‐6), 62 (IV‐1), 56 (IV‐2), 56 (IV‐14), 63 (IV‐18), 57 (IV‐20), 53 (IV‐21) and 54 (IV‐23) years. In contrast to the early onset in affected members, the mean age (± SD) of the seven unaffected siblings and cousins determined at interview in the fourth generation was 57.3 ± 3.5 years.

Blood samples were collected from a total of 13 individuals, including four affected and nine unaffected members, for genetic analysis (Fig. 1). Informed consent was obtained from all the subjects, following the guidelines for ethical research approved by the Research Ethics Committee of the Faculty of Health Sciences, University of Cape Town.

Neuropsychological screening could only be completed in two subjects, IV‐13 and IV‐16. The other two affected subjects (IV‐10 and IV‐19) could not be fully tested due to the severity of their impairment (Fig. 1). A Xhosa translation (unpublished) of the détérioration cognitive observée (DECO), an informant questionnaire (Lenger et al., 1996), was administered to relatives to determine whether there was any decline in the subject’s everyday functioning over the past year. The DECO is a 19‐item questionnaire with a three‐point Likert Scale consisting of ‘better’ (2), ‘about the same’ (1) and ‘much worse’ (0). In addition, the Mini‐Mental State Examination (MMSE) was performed on one subject.


Axial CT brain scans were performed on all four affected individuals at the time of clinical evaluation. The scan angle was tilted to optimize the view of the medial temporal lobes. Starting with a lateral tomogram, sections were taken passing through the front of the hard palate parallel with the orbitomeatal line. Subject IV‐19 had a single MRI brain scan that only included axial and sagittal T1‐weighted sequences. Single‐photon emission computed tomography imaging was performed in subject IV‐13 at the time of clinical evaluation.

Autopsy and histopathology

The proband died at age 50 years, 5 months after she presented in the clinic. The autopsy was performed at the Department of Anatomical Pathology, Tygerberg Hospital. Death was attributed to bronchopneumonia. The lungs showed some anthracosis, occasional focal scars and variable bronchopneumonia. The heart, spleen and the rest of the abdominal organs were unremarkable. The brain was removed in its entirety and immersion‐fixed in buffered formalin for detailed evaluation. Paraffin‐embedded tissue sections were cut and stained by conventional histopathological and immunocytochemical methods from the four cortical lobes, hippocampal formation, basal ganglia, cerebellum and brainstem. We also evaluated in parallel brain tissue from three sporadic AD cases of Caucasian origin and with relatively early‐onset disease. Their ages at death were 58, 64 and 67 years. The specimens were obtained from the Newcastle Brain Tissue Resource centre at the MRC building, Newcastle General Hospital.

To confirm the patterns of AD pathology, we used polyclonal and monoclonal antibodies to the following β amyloid (Aβ) peptides: residues 8–17 (Aβ; 6F3D; Dako, Denmark), 17–24 (4G8; Signet, MA, USA), 36–40 (Aβ40; Ter 40; courtesy of H. Mori, Osaka, Japan) and 38–42 (Aβ42; Ter 42; courtesy H. Mori). The distribution of APP was studied by immunostaining with the 22C11 antibody (Boehringer Mannheim, Germany). Neurofibrillary pathology, Lewy bodies and APOE‐containing lesions were detected with antibodies to tau protein (AT8; Endogen, UK), α‐synuclein (Novacastra, UK), ubiquitin (Dako) and APOE (courtesy of H. Mori). Specificities of Aβ end‐terminal antibodies and other AD pathology‐associated antibodies have been well established previously. Upon immunoblotting, the Aβ antibodies recognized the expected band(s) of proteins in solubilized preparations of neocortical homogenates from sporadic AD subjects (Kalaria et al., 1996).

Linkage and mutation analysis

Genomic DNA was extracted from whole blood using standard methods. Linkage‐based analysis was used to identify the suspected early‐onset gene contributing to the disease. DNA was amplified by PCR to determine cosegregation of particular alleles with the following markers (Sherrington et al., 1995) on chromosomes 1 and 14: D1S459, D1S490, D1S479, D14S77, D14S53 and D14S43. Based upon evidence of shared alleles, the PS1 gene was amplified using intronic primers as described previously (Hutton et al., 1996). In the first attempt, exons 3–10 of PS1 were sequenced directly and analysed using an ABI 377 automated DNA sequencer (Perkin Elmer Applied Biosystems, UK). The colour‐coded electropherograms were checked for nucleotide ambiguity by manual inspection as well as by display on Sequence Analysis version 3.1 and comparison with published sequences.

To confirm the suspected mutation, restriction digestion analysis of exon 6 of PS1 (ordering based on current database; Yasuda et al., 1999) was performed in a series of samples, including those of affected and unaffected family members. An allele‐specific nested PCR was run on the original exon 6 products with the following primer sequences: forward 5′ GCCATCATGATCAGTGCCAT 3′; reverse 5′ CCAACCATGGGAAGAACAG 3′. The PCR conditions included 58°C for annealing and 30 s extension time for 35 cycles using the Hotstar PCR kit (Qiagen, UK). The expected PCR product of size 124 bp was subjected to restriction digest with NcoI endonuclease (New England Nuclear) at 37°C for 12 h. The products were then separated on a 20% Tris–boric acid–EDTA gel and detected by silver staining (Condie et al., 1993).

APOE genotyping of DNA extracted from blood or formalin‐fixed brain tissue was performed essentially as described previously (Premkumar et al., 1996).


Clinical findings

The Xhosa form part of the Bantu (indigenous people of south and central Africa), who lived predominantly in the homelands of Ciskei and Transkei in Eastern Cape until 1994. The family was traced back five generations to the east coast of Southern Africa and started relocating further south in the last two generations. There was no history of consanguinity or Caucasian admixture in the preceding generations of this Xhosa family. We recorded 12 affected individuals spanning four generations (Fig. 1). The first known affected members (I‐2 and II‐1) were said to have died between 60 and 70 years of age, although this was an approximation as their dates of birth were unknown. The ages of affected individuals in generation III were best estimates as many of them were not literate. Of the four affected individuals examined in the fourth generation, two (IV‐10 and IV‐19) were smokers and had used alcohol excessively at times. None had used illicit drugs or other medications. Subject IV‐10 had also suffered a head injury 1 year before symptom onset, with brief loss of consciousness followed by confusion, which persisted for a few hours. There was no other significant medical history in any of the subjects. Due to sociopolitical factors at the time, the family members received less than 4 years, in some cases, and a maximum of 7 years (average 5.2 years) of formal school education. The matriarch of the family was interviewed 4 years after all the family members were first seen and confirmed that unaffected members of similar age (i.e. in the fourth generation) to the affected members were still normal with respect to remembering and their daily activities. The mean (± SD) age of unaffected members (n = 7) in the fourth generation at the time of the interviews was 57.3 ± 3.5 years (Fig. 1).

In the third generation, the mean (n = 4) age at disease onset, based on estimates from family members, was 49 ± 3.37 years, with a mean disease duration of 6.33 ± 2.31 years, and death was at 56.67 ± 3.21 years. In the fourth generation the mean (n = 6) age at onset was 50 ± 4.98 years and the mean age at death for three members was 55.33 ± 5.51 years, after a disease duration of 7.33 ± 2.10 years. The remaining three are still alive 3, 8 and 12 years after onset of dementia (Fig. 1).

Table 1 provides a summary of the various neurological and psychiatric features in the four affected individuals. The disease onset was remarkably similar in all affected individuals. The first reported symptom in each case was memory impairment; the patient would first either forget to hide their money or forget where they had hidden it. Progressive memory impairment was characterized by difficulty in remembering the names of their own children and eventually being unable to recognize them. Loss of memory of such characteristics is considered unacceptable in African cultures (Kalaria et al., 1997). Later, spontaneous speech became limited and the subjects would wander and get lost in the neighbourhood. Eventually they were unable to perform any activities of daily living, became mute and were bedridden. Myoclonus was not present in any individual at the time of examination (disease duration 8 ± 2.45 years) and only the proband suffered one seizure in the last 5 months prior to death (disease duration ∼5.5 years). According to the informants, seizures did not appear to be a feature of the phenotype.

View this table:
Table 1

Neurological, cognitive and behavioural features in demented family members

Subject identity (see Fig. 1)IV‐10IV‐13IV‐16IV‐19
Approximate age at onset (years)52415645
Years of symptoms at time of examination7575
Age at death (years)6150
PS1 mutationI143MI143MI143MI143M
Head injuryMinor+
Naming children+
Preserved insight++
Presence of seizures+
Orientation person/time/place+/–/–+/–/++/+/––/–/–
Activities of daily living (feed self/bath/toilet)+/–/–+/+/++/+/+–/–/–
MMSE (maximum = 30)9NDNDUT
DECO (maximum = 38)ND133UT
Visuospatial skillsPoorNPoorUT
Recognition memoryND+UT
Naming N?NN
LanguageSparse; intelligibleNNMinimal output
Impaired visual learningND++UT
Impaired verbal learningND++UT
Psychiatric manifestations++
Primitive reflexes++
Extensor plantar responses

+ = presence; – = absence; N = normal. N? = simple objects only; ND = not determined. UT = untestable due to severity of dementia. Head injury: Minor = no loss of consciousness.

The proband (IV‐19, Fig. 1) was right‐handed. She presented at 50 years of age 5 years after onset, and was found to have profound dementia. She was unable to communicate or recognize her family members, and unable to perform simple tasks of daily living. She was uncooperative, irritable and displayed repetitive behaviour, using either objects in the close vicinity or repeatedly cleaning when given an object. At times she was aggressive and at other times exhibited inappropriate laughter. Spontaneous speech was occasional, with loss of coherence. She could not follow simple instructions or name everyday objects. She was easily distracted and did not consistently respond to her name. Her blood chemistry values were within normal limits. The patient was discharged to a nursing home and died 5 months later.

Patient IV‐13 was examined 5 years after the onset of her memory disturbance. She was pleasant and cooperative but disorientated for time. She could remember her children’s names but not their ages. She showed grossly impaired visual and verbal learning and poor recognition memory. She repeatedly perseverated on a word‐generation test. Naming, language and visuospatial skills were still intact. The DECO questionnaire completed by her sister revealed a score of 13/38, confirming severe dementia. Four years later she started wandering from home and is currently awaiting institutional placement.

Patient IV‐16 was first examined 7 years after the onset of symptoms. She often repeated herself during the examination. Like her cousin (IV‐13), she had difficulty remembering the number and names of her children. Memory and visuospatial skills were very poor but naming and language skills were still intact. The DECO questionnaire was 3/38. She was institutionalized 3 years later.

The fourth subject (IV‐10) was also examined 7 years after first symptoms. Two years previously he had started wandering from home and experienced auditory hallucinations. In the preceding month he had been intermittently incontinent. He could not wash or dress but could still feed himself. The patient was restless and irritable but was orientated for self. His speech was sparse but intelligible. He could name simple objects presented to him in the clinic room. The MMSE score was 9/30. Routine laboratory investigations were normal apart from a reactive fluorescent treponemal antibody‐absorbed test in the blood and CSF but negative venereal disease research laboratory (VDRL) test for neurosyphilis. The CSF was acellular but protein was slightly elevated (0.7 g/dl; normal = 0.45 g/dl). These findings were not interpreted to be due to neurosyphilis; however, a course of therapy was given. The patient deteriorated and died 2 years later.


In all affected subjects, head CT imaging showed generalized cerebral atrophy, most marked in the medial temporal lobes and tempoparietal regions, with ex‐vacuodilatation of the lateral ventricles (not shown). The severity of cerebral atrophy paralleled the degree of cognitive dysfunction. Atrophy with ventricular dilatation was most marked in the proband and least in patient IV‐13. However, the posterior fossa appeared normal in all the subjects. A T1‐weighted MRI of the proband was of poor quality due to movement artefact but essentially showed the same features as the CT image, with marked cerebral atrophy most prominent in the medial temporal lobe.

Single‐photon emission computed tomography in patient IV‐13 showed markedly reduced cerebral perfusion in the left medial temporal region, right parietal region and both parieto‐occipital regions (not shown).

Neuropathological findings

Gross examination of the brain of the proband (IV‐19) confirmed atrophy of the temporal lobes and less marked atrophy in the frontal lobes. The total brain weight at autopsy was 1334 g. The vessels of the circle of Willis and the vertebrobasilar system only showed focal mild atheroma. Consistent with the CT, the lateral and third ventricles were moderately enlarged. The substantia nigra was only mildly depigmented.

Histopathological examination of the brain revealed areas of focal neuronal loss with marked astrogliosis. Silver staining revealed large numbers of neuritic plaques and neurofibrillary tangles, particularly in the inferiomedial temporal lobes (Fig. 2). Neuritic plaques were abundant in all cortical lobes (Fig. 2A, B), including the striate cortex at the occipital pole. The mean density of amyloid plaques was 41 per mm2 in the cortical lobes. This compared to 48, 15 and 47 plaques per mm2 in three sporadic AD cases, who died at 58, 64 and 67 years. Hippocampal pyramidal neurons showed widespread granulovacuolar degeneration with marked neuronal loss, spongiosis and astrogliosis. Numerous compact and diffuse plaques in the neocortical lobes and hippocampal formation were further revealed by antibodies to Aβ42 peptide, although very few were evident with antibodies to the more soluble Aβ40 peptide (Fig. 2C, D). Sections from the cerebellum also showed diffuse Aβ42 deposits, predominantly in the molecular layer (Fig. 2E). Focal amyloid angiopathy was evident in several meningeal vessels, as revealed by antibodies to Aβ peptides (Fig. 2E). Antibodies to amyloid associated proteins, such as ubiquitin and APOE, also characteristically stained the deposits, suggesting no apparent qualitative differences between the pattern of immunostaining in this case and in sporadic AD cases. The periventricular and deep white matter showed diffuse demyelination that was consistent with sporadic AD cases. However, patchy accumulation of APP immunoreactivity was evident in the deep white matter, indicating diffuse axonal pathology (not shown).

Fig. 2 Neuropathological findings in the brain of the proband. Neuritic plaques in the hippocampal formation are demonstrated with (A) Holmes stain and (B) ubiquitin immunocytochemistry. Immunocytochemical localization of (C) Aβ42 and (D) Aβ40 deposits in adjacent sections of the medial temporal cortex. (E) Profound Aβ42‐positive diffuse deposits and cerebral amyloid angiopathy was also evident in the cerebellum. Antibodies to tau (AT8) revealed neurofibrillary pathology in (F) the lateral geniculate nucleus, (G) inferior colliculus and (H) medullary nuclei. The bar represents 100 µm (A, B) or 50 µm (CH).

Neurofibrillary pathology was evident in the form of dystrophic neurites, neuropil threads and extracellular tangles in all cortical lobes, including the primary visual cortex. Such features were also invariably present in the basal ganglia, particularly in the putamen and both segments of the pallidum. The density of tangles was >50 per mm2 in the cortical lobes and hippocampus. This compared to 38, 29 and 22 cortical tangles per mm2 in the three sporadic AD cases. The severity in the proband readily accorded with Braak stage VI. Several neurons also contained intracellular tangles. These were strikingly apparent in the lateral geniculate (Fig. 2F). The brainstem contained widespread neurofibrillary pathology, including extracellular and intracellular globose tangles in the central tegmentum, colliculi and medullary nuclei, revealed by antibodies to tau (Fig. 2G, H). These brainstem regions also contained several neuritic plaques stained by Aβ42 antiserum. The nigral nuclei showed patchy neuronal loss, spongiosis, astrocytosis and occasional tangles in pigmented neurons. Lewy bodies were generally absent in the neocortex, substantia nigra and locus coeruleus, but a few α‐synuclein‐positive neurons and punctate extracellular deposits were evident in the molecular layer of the hippocampus proper. We could not detect any Pick bodies or Pick cells. All these features were consistent with the diagnosis of severe AD according to the criteria of the Consortium to Establish a Registry for Alzheimer’s Disease (Mirra et al., 1991).

Genetic analysis

Initial linkage‐based analysis had suggested the involvement of PS1 but not the presenilin 2 gene in the occurrence of disease. Thus there was no evidence of cosegregation with markers on chromosome 1. However, this was not true with co‐inheritance of the chromosome 14 markers, which pointed to an association. We also ruled out the possibility of the AD in this family being linked to known APP gene mutations by direct sequencing of exons 16 and 17 of APP. We found no apparent sequence variation in any of the four affected subjects compared with all the unaffected siblings or cousins in the fourth generation.

Initial direct sequencing of the PS1 gene in the proband (IV‐19) and an affected sibling (IV‐16) revealed a novel ATT to ATG mutation at nucleotide 677 in exon 6 (Fig. 3A, B). This was predicted to lead to an isoleucine‐to‐methionine missense substitution at codon 143 (I143M) in the second transmembrane region of PS1. This mutation or other exon changes were not apparent in the seven unaffected siblings or cousins and one other disease‐free member of the family from the third generation (Fig. 3).

Fig. 3 Demonstration of a novel mutation by direct sequencing in PS1 of affected members. The electropherograms indicate (A) a T to G change (asterisk) at nucleotide 677 in exon 6 of the proband, and (B) no mutation in the unaffected individual. The same change was also evident in subjects IV‐10, IV‐13 and IV‐16. This translated to an Ile to Met mutation at codon 143. (C) Restriction digestion of allele‐specific nested PCR product for confirmation of the mutation. PCR product digestion with NcoI from affected members (IV‐10, IV‐16) (lanes 1 and 3) resulted in a 98‐bp fragment plus a smaller 17‐bp fragment. The 17‐bp fragment migrated at the front (not shown). These fragments were not evident in DNA from seven unaffected family members of similar age and 50 unrelated Xhosa control subjects when similarly incubated with NcoI, but retained the original 115‐bp amplicon (lane 2). (Lane M) Marker V from Roche Molecular Biochemicals (UK).

To confirm this novel finding, we also evaluated restriction digests of allele‐specific nested PCR‐amplified products of exon 6. The mutation created a NcoI restriction enzyme site in the mutant allele. The NcoI digest resulted in 98‐ and 17‐bp fragments from the original 115‐bp amplicon in the affected individuals with this mutation but not in the unaffected members (Fig. 3C). In further analysis we confirmed that the mutation was not present in 100 chromosomes of the normal Xhosa control sample.

In an attempt to evaluate if APOE genotype status modulated the onset or symptoms of disease in the affected members, we also screened 11 individuals whose DNA was available subsequent to the exhaustive linkage and mutation analysis. The APOE genotype of the proband (IV‐19) was determined to be ϵ2/ϵ4 (Table 2). Interestingly, both affected and unaffected siblings carried one APOE ϵ4 allele. The ϵ4 allele frequency was calculated to be 50% in both groups. As expected, the ϵ2 and ϵ3 alleles were almost equally distributed in the two groups. There was no apparent relationship between the presence of the APOE ϵ4 allele and age of onset or age at death in the affected individuals (Table 2).

View this table:
Table 2

APOE genotypes in affected and unaffected members of family*

Affecteds Unaffected members
(n = 4)(n = 7)
Mean (± SD) age at onset (year)48.5 ± 5.9
Mean (± SD) age at interview or death (years) 55.3 ± 7.157.3 ± 3.5
APOE genotypesϵ3/ϵ4 (IV‐10); ϵ2/ϵ4 (IV‐13); ϵ3/ϵ4 (IV‐16); ϵ2/ϵ4 (IV‐19)ϵ2/ϵ4 (n = 4); ϵ3/ϵ4 (n = 3)
APOE‐ϵ4 allele frequency50%50%

*Of the 13 DNAs collected, 11 were genotyped for APOE as described in Subjects and methods. All affected subjects had the I143M PS1 mutation; no unaffected subjects had mutations in PS1.


The PS1 locus is linked to the most aggressive form of familial AD, with disease presentation as young as 30 years of age (Fox et al., 1997; Houlden et al., 2001), and by 60 years almost 90% are affected (Rogaeva et al., 2001). Although there does not appear to be a clear correlation between specific mutations and age at onset, cases with PS1 mutations in codons preceding 200 appear to be on average 5 years younger than those above codon 200 (Mann et al., 2001). The I143M mutation described in this report is a novel mutation that is part of the cluster at the second transmembrane domain (Fraser et al., 2000). Two other families with mutations at codon 143 have been described: a Belgian family bearing the I143T mutation (Cruts et al., 1995) and an English family with relatively later onset bearing the I143F mutation (Rossor et al., 1996) (Table 3). Unlike in the English family, our study suggests complete penetrance of this mutation. A recent screening of the PS1 gene in 414 subjects revealed the I143T mutation in two sisters and a further, unrelated individual. Interestingly, both sisters exhibited a very early onset of disease (<35 years) but both had also inherited another PS1 mutation (I439V) in trans from their mother (Rogaeva et al., 2001)

View this table:
Table 3

Phenotypic features in familial AD with mutations at codon 143 of PS1

BelgianEnglishSouth African Xhosa
Mean age at onset (years)3555 50
Mean duration (years)4*Not known8
Mean age at death (years)39*6556
Penetrance+68 years of age (unaffected)+
Reference(Cruts et al., 1995)(Rossor et al., 1996)This study

*Details of only one individual (see also Mann et al., 2001).

Myoclonus may appear midway through the illness (Fox et al., 1997) or at a very early age (Mayeux et al., 1985) in familial AD. Unlike many other families with PS1 mutations and the I143T family (Table 2), who had both myoclonus and epilepsy (Mann et al., 2001), our family did not have myoclonus as a feature. Seizures also frequently occur in young‐onset familial AD (Frommelt et al., 1991; Lampe et al., 1994; Haltia et al., 1994; Fox et al., 1997; Janssen et al., 2001). The proband’s aggressive illness appeared to be associated with only one seizure, supporting the view that other influences may play a role in the phenotype. Spastic paraparesis associated with a number of different mutations (Crook et al., 1998; Houlden et al., 2000; Moretti et al., 2000; Rogaeva et al., 2001; O’Riordan et al., 2002) and ‘cotton wool’ plaques in the neocortex were also absent in this family.

Our South African family is from a socioeconomically disadvantaged community and the patients appear to have presented late in their disease. Due to limited schooling as well as cross‐cultural screening difficulties, the DECO informant questionnaire was used instead of Western psychometric tests. However, this is comparable to the modified informant questionnaire on cognitive decline in the elderly (IQCODE) used previously in other African settings (Sayi et al., 1997). The MMSE has more stringent literacy requirements and was not reliable in all subjects. Some of the unaffected members often displayed low scores. The DECO, proposed as an alternative approach for dementia screening, did not show bias with respect to education (Ritchie and Fuhrer, 1992) or social class (O’Connor et al., 1989).

Although assessed at different times in their disease, the pattern of cognitive decline appeared similar in all affected subjects. Memory loss was the earliest signs of the illness. There was preservation of naming at a stage when other cognitive skills were profoundly affected (Fox et al., 1997). However, the rate of decline varied; two affected individuals had preserved naming and language skills 5 and 7 years after onset whereas the other two, who were severely demented, exhibited minimal speech with similar disease duration. The question arises whether non‐genetic factors (Graves and Kukull, 1994) that reportedly influence sporadic AD, including a low level of education, a history of smoking, alcohol use and head trauma, could have resulted in the intrafamilial phenotypic variability evident in this family.

The polymorphism in the APOE gene increases susceptibility to the development of AD later in life, with an ϵ4 allele dose‐dependent effect (Farrer et al., 1997). Although APOE‐ϵ4 may influence the age at onset in familial AD due to APP mutations (Levy‐Lahad et al., 1995b), the same effect has not been shown in AD due to PS1 mutations (Van Broeckhoven et al., 1994). We did not find any evidence to suggest that the ϵ4 allele modified the onset or aggressive course of the disease, although a high frequency of the allele was apparent in both the affected and unaffected individuals. This is consistent with the increased frequency (25%) of the APOE‐ϵ4 allele among Bantu Africans (Kalaria et al., 1996) and no or weak association of the allele with dementia in Africans, African–Americans and Hispanics (Kalaria et al., 1997; Sayi et al., 1997; Tang et al., 1998).

The neuropathological findings in the brain of the proband were generally consistent with observations in other individuals bearing PS1 mutations (Mann et al., 2001). There was a marked preponderance of Aβ42‐type deposits (De Jonghe et al., 1999; Singleton et al., 2000), but also widespread and intense neurofibrillary degeneration. Although unclear in previous studies, this feature also extended to the brainstem structures, with evidence of marked neuronal dystrophy in both the superior and inferior colliculi. The degree of pathology, including the lack of severe cerebral amyloid angiopathy (Dermaut et al., 2001), in the proband was consistent with the type 1 histological pattern, consisting of many diffuse senile plaques in the frontal lobes (Mann et al., 2001), although the age at onset did not coincide.

The prevalence of AD amongst indigenous South African subjects is not known. Local neurologists and old‐age psychiatrists rarely encounter it (de Villiers and Louw, 1996). There appears to be only a single previous preliminary report on pathologically confirmed AD, in two black South African women (Cole, 1977). Community‐based studies in Ibadan, Nigeria (Hendrie et al., 1995) and Upper Assiut, Egypt (Farrag et al., 1998) also suggested that sporadic AD was uncommon, with a significantly lower prevalence amongst Nigerian Africans (1%) compared with African–Americans (6%) over the age of 65 years (Hendrie et al., 1995). The investigators of these studies reason that these groups have similar origins but they acknowledge an estimated 25% white European admixture in the existing genetic pool of African–Americans (Chakraborty et al., 1991). This may also apply to Caribbean Hispanic families (Athan et al., 2001). A follow‐up report confirmed the prevalence results, with the age‐standardized incidence rate of AD amongst the West African population at 1.15% compared with 2.52% amongst African–Americans (Hendrie et al., 2001). There was a weak association of the APOE‐ϵ4 allele and a much lower prevalence of vascular disease, which could contribute to a lower incidence of AD in Africa (Hendrie et al., 2001). There may be several other reasons why AD is not seen in indigenous South Africans; a decade ago the elderly (>65 years) constituted 8.7% of the white South African population whereas only 3% of the black South African population were elderly (Prinsloo, 1991). Given that rural communities still have difficulties in accessing tertiary health care and that memory impairment may be accepted as part of normal ageing, patients do not readily interface with clinicians (de Villiers and Louw, 1996). The proband in this family was not brought to medical attention until she had reached an advanced state of dementia, when the caregivers found it increasingly difficult to provide care.

In summary, we report a large African family with PS1‐linked early‐onset AD. Most of the clinical and pathological features are in keeping with those previously described in Caucasians. Although as‐yet unknown environmental influences cannot be discounted, the variability in the course of the disease evident in this Southern African family suggests other influences play a role.


We are indebted to Dr Marc Combrink for referring the proband for clinical evaluation. We thank Cindy‐Lee Cupido for the initial linkage analysis and Janet Slade for technical assistance and are grateful to Julie Hind and Linda Cawley for secretarial assistance. Our research programmes are supported by grants from the Medical Research Council (UK), the Alzheimer’s Research Trust (UK), the Alzheimer’s Association, USA (IIRG Award), the National Institutes of Health (NINDS), USA and EU Framework 5 project.


View Abstract