Frequency of Mutations in the Presenilin and Amyloid Precursor Protein Genes in Early-Onset Alzheimer Disease in Spain FREE

ALZHEIMER DISEASE (AD) is the most common cause of dementia and may be present in familial and sporadic forms. The age of onset in AD may vary widely and this is the basis for the classification into early- and late-onset, with 60 or 65 years being the usual cutoff point.^{1 - 2} Although the underlying causes of AD remain unknown, molecular analysis in recent years has suggested that the causes may be heterogeneous. Among genetic factors, the apolipoprotein E ϵ4 (APOE ϵ4) allele is a risk factor for familial and sporadic early- and late-onset AD.^{3 - 5} In addition, in autosomal dominant early-onset AD, mutations in 3 causative genes have been identified, the amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2) genes. However, the clinical presentation of AD is more complex, since some patients may present with an early-onset but in sporadic or familial nonautosomal dominant manner, or they may have familial AD but be within the range of late-onset AD. The relative contribution of APP and PSEN mutations to autosomal dominant early-onset AD is the subject of considerable controversy. While in some studies^{2 ,6} virtually all cases could be explained by mutations in these 3 genes, others^{1 ,7} found a low mutation frequency suggesting the involvement of other genetic factors. We report the experience of our center in mutational analysis of patients who were referred because of a high suspicion of a genetic disorder. Although because of its design and referral bias this study cannot be considered an epidemiological population-based study, it does reflect the experience observed in clinical practice. The aim of this study was to clarify the respective contribution of the APP and PSEN mutations to autosomal dominant AD and to determine its contribution to sporadic and familial nonautosomal dominant early-onset AD and familial late-onset AD. We will also discuss its relation with the APOE ϵ4 allele status.

From 1996 through 2001 we ascertained 94 white patients who had AD (92 with probable AD and 2 with definite AD) from 82 independent families. Most patients were recruited in the Neurology Service at the Hospital Clinic, Barcelona, Spain, although some cases were referred from other centers (see "Acknowledgment" section). All patients fulfilled the National Institute for Neurological and Communication Disorders and Stroke–Alzheimer's Disease and Related Disorders Association criteria for probable or definite AD.⁸ We included patients with familial or sporadic early-onset AD. We also included patients with familial late-onset AD but limited enrollment to the first 30 families. Patients with vascular dementia, a history of alcoholism, or incomplete information about age of onset or family history were excluded from this study. To exclude frontotemporal dementia, patients with early behavioral disturbances or an early language disorder were not included. Patients with prominent atrophy of the frontal lobes in the brain computed tomographic or magnetic resonance imaging scans were also excluded from this study. Age of onset was determined as the age of the appearance of the first cognitive symptoms. Early onset and late onset were defined by an age of onset younger or older than 65 years, respectively. Based on family aggregation, patients were subdivided into familial or autosomal dominant groups according to the pedigree. Familial was defined by at least 1 first-degree relative with a clinical picture suggestive of AD. Autosomal dominant transmission was defined by at least 3 members with dementia in 2 generations and at least 1 with detailed clinical information supporting the diagnosis of AD.¹ Assymptomatic members with a family history of AD were not included. All subjects gave written informed consent. A group of 110 healthy controls from our center was used to compare APOE ϵ4 allele frequencies.⁹

Genomic DNA was isolated from the peripheral blood lymphocytes according to standard procedures. The coding regions of the PSEN1 (exons 3-12), PSEN2 (exons 3-12), and APP (exons 16 and 17) genes were amplified using described specific primers.^{1 ,10} The polymerase chain reaction products were analyzed through single-strand conformation polymorphism and subsequent sequencing of the samples with abnormal mobility using the dye terminator cycle sequencing kit (Perkins-Elmer Inc, Wellesley, Mass). Sequencing was performed on an automatic sequencer (ABI PRISM model 377; Applied Biosystems, Foster City, Calif). Apolipoprotein genotyping was performed through polymerase chain reaction amplification and HhaI restriction enzyme digestion as described.¹¹ Evidence for cosegregation with the disease was required in all cases to consider a novel pathogenic mutation.

The frequency of mutations was calculated as the number of mutations divided by the number of families in each group. When analyzing the age of onset and the APOE ϵ4 allele frequency only the proband from each family was included. As the age of onset followed a normal distribution, we used a 1-way analysis of variance with Bonferroni correction and a t test to compare the age of onset among AD groups based on familial aggregation and mutation status, respectively. We carried out a χ² analysis to compare APOE ϵ4 allele frequency. We used the receiver operating characteristic curve to determine the age of onset that identifies the higher number of mutations testing the minimal number of patients without mutations. We included in the receiver operating characteristic curve all sampled individuals, whether or not they belonged to the same family. Although we acknowledge that the age of onset is not independent among family members, from a clinical perspective we considered it relevant to include all individuals in the analysis.

Of the 94 patients with AD (39 men, 55 women), 60 patients presented with early-onset AD and 34 with late-onset AD (Table 1). In the early-onset AD group, 44 patients from 36 families had a positive family history for AD (11 families with autosomal dominant AD and 25 families with familial nonautosomal dominant AD) and 16 were sporadic cases of AD. The mean age of the different groups is given in Table 1. The age at onset of the autosomal dominant cases was earlier than the familial nonautosomal dominant cases (P<.01). We found 8 mutations in the PSEN1 or PSEN2 genes among the autosomal dominant group (54.6 %, Table 1). In contrast, we found only 1 mutation (4%) in the familial nonautosomal dominant group¹² and 1 mutation (6.2%) in the sporadic group. In both families 1 of the proband's parents had died young of unrelated medical problems and were considered the possible mutation carriers. The age of onset in patients with PSEN mutations (mean [SD] age, 43.9 [7.2] years) was earlier than in patients without PSEN mutations (52.9 [5.7] years, P<.001) among those with early-onset AD. The presence of mutation was associated with an autosomal dominant pattern (P<.001). In all cases with PSEN mutations, segregation analysis was performed and cosegregation was confirmed. No mutations were found in the group with late-onset AD. We did not find mutations in the APP gene. We identified 5 different mutations in the PSEN1 gene^{12 - 15} and 2 (Table 2) in the PSEN2 gene.¹⁶ The M139T mutation in the PSEN1 gene was identified in 2 independent families.⁶ The T430M mutation in the PSEN2 gene was detected in a 45-year-old individual and was also detected in the affected proband's parent. The H163R mutation had been previously described in several families with early-onset AD.^{7 ,13} We found several nonpathogenic polymorphisms in the PSEN1 and PSEN2 (Table 3) genes. Most families presented a typical clinical picture, although some mutations were associated with a particular phenotype. The family carrying the V89L mutation in the PSEN1 gene¹⁵ presented early behavioral disturbances and the families carrying the H163R and S169P¹³ mutations showed early myoclonus and generalized tonic-clonic seizures.

The APOE ϵ4 genotype and allele frequencies in patients with AD is indicated in Table 4. Among autosomal dominant early-onset AD, the APOE ϵ4 allele frequency was significantly increased in patients without PSEN mutations compared with healthy control subjects. In this group 3 of the 5 individuals without PSEN mutations carried at least 1 APOE ϵ4 allele (2 with APOE ϵ4ϵ4 and 1 with APOE ϵ4ϵ3). None of the patients with PSEN mutations carried an APOE ϵ4 allele. The APOE ϵ4 allele frequency was increased in familial early-onset cases without PSEN mutations compared with healthy controls (P<.004) and was similar to the familial late-onset group. In sporadic early-onset cases without PSEN mutations the APOE ϵ4 allele was also overrepresented compared with healthy controls but without statistical significance (P = .06, Fisher exact test). The receiver operating characteristic curve (Figure 1), performed to determine the best cutoff age for genetic analysis, showed that if an age of 52 years or younger is used to select patients who will be tested, then 85% of PSEN mutations would have been detected with the inclusion of a low number of patients without mutations (1 − specificity = 25%). If we had chosen a cutoff age of 58 years or younger, we would have detected 92% of PSEN mutations although with an increase in the number of patients without mutations (44%) tested. Finally, if we had chosen an age of 64 years or younger, we would have detected 100% of PSEN mutations but testing 58% of the patients without mutations.

In this study we assessed the relative contribution of the PSEN and APP gene mutations in a group of patients with sporadic and familial nonautosomal AD, focusing on the age of onset and family history. We found that the presence of mutation was clearly related to the family aggregation pattern. The frequency of mutations in the PSEN genes was high among patients with autosomal dominant AD in contrast with the low frequency found in the sporadic and familial nonautosomal dominant AD groups. Our frequency of 54.6% among autosomal dominant AD is in the upper zone of the reported frequencies, which ranges from 16.5% to 71.0%.^{6 - 7} However, the specific familial pattern and segregation analysis is not always available in some studies.^{2 ,7} This suggests that in most studies, the real frequency of mutations could have been underestimated by the inclusion of familial nonautosomal dominant cases. In addition, the absence of segregation analysis in some studies^{2 ,7} could have included rare nonpathogenic polymorphisms as pathogenic mutations.¹⁷ Our results confirm the notion that most PSEN mutations are associated with an autosomal dominant pattern and early onset, and point out that mutations are uncommon in patients with different presentation patterns. Nevertheless, in our familial nonautosomal dominant and sporadic AD cases carrying PSEN mutations, the parents' premature deaths could have masked the autosomal dominant presentation. The existence of phenocopies, incomplete penetrance, censoring effect, and false paternity may jeopardize the original autosomal dominant pattern.¹⁸ In addition, some sporadic cases carrying a PSEN mutation^{1 ,19 - 20} and cases with a probable de novo PSEN mutation have also been described.²¹ In any case, the careful study of the family pedigree is essential in determining the indication for genetic testing. Despite some patients with late-onset AD who had PSEN2 mutations having been described,²² we did not find mutations in this group. In this study, we have found common and uncommon nonpathogenic polymorphisms in the PSEN genes. From a genetic perspective, the presence of these nonpathogenic polymorphisms should always be considered in the genetic testing of AD. In particular, uncommon polymorphisms may have important implications for genetic counseling in AD.¹⁷

Although the family history is an essential factor to indicate genetic testing, this information may sometimes be unavailable. For this reason, we studied whether the age of onset could have been used as the unique variable to predict the presence of mutation. We found that all PSEN mutations would have been determined with a cutoff age of 64 years. However, in our opinion, the optimal cutoff age (considering only the age of onset) is 58 years as most mutations would have been detected by testing an acceptable number of individuals without mutations.

Most families with PSEN mutations presented a typical phenotype not different from the common form of AD. However, 2 families presented early myoclonus and generalized tonic-clonic seizures, a phenotype already associated with PSEN mutations.^{23 - 24} In addition, the family with the V89L mutation showed early and prominent behavioral disturbances.¹⁵ These data support the presence of clinical heterogeneity in autosomal dominant early-onset AD in addition to genetic and allelic heterogeneity.

In this study we found that in the group of autosomal dominant AD, patients without PSEN mutations presented an increased APOE ϵ4 allele frequency compared with healthy controls. Although autosomal dominant AD is mainly due to PSEN mutations, this result may reflect that in a small number of families the disease may be related to the APOE ϵ4 allele. In fact, an increased APOE ϵ4 allele frequency has been reported in families with autosomal dominant inheritance.³ Most autosomal dominant families carried either a mutation or at least 1 APOE ϵ4 allele, but in 2 families neither factor was identified. This suggests that either intronic mutations in these genes or mutations in other genes can be expected to explain the disease in these families. On the other hand, in familial and sporadic early-onset AD, the frequency of PSEN mutations was low and the APOE ϵ4 allele was overrepresented compared with healthy controls. These findings are consistent with other studies, which have also documented that APOE ϵ4 allele is overrepresented in early-onset familial and sporadic AD.^{3 ,25 - 26} In addition, the study of Houlden et al²⁷ showed that the increased APOE ϵ4 allele frequency was only observed in non-PSEN families. Our data confirm the notion that different causes coexist in early-onset AD and they can be related to the familial pattern. Once a mutation in the PSEN genes has been excluded, the APOE ϵ4 allele emerges as a risk factor in these non-PSEN groups reflecting the beginning of the spectrum of its effect.

Our study suggests that the screening of PSEN and APP mutations in patients with early-onset AD is likely to be successful only if there is a clear autosomal dominant pattern, but not in other familial or sporadic cases. In sporadic and familial cases, the indication for genetic testing should rely on the careful analysis of the family pedigree. When the family history is not available, 58 years could be used as a cutoff age as most PSEN mutations are detected below this age.

Accepted for publication June 20, 2002.

Author contributions: Study concept and design (Drs Lleó, Blesa, Molinuevo); acquisition of data (Drs Lleó, Blesa, Queralt, Peña-Casanova); analysis and interpretation of data (Drs Lleó, Rojo, Oliva); drafting of the manuscript (Dr Lleó); critical revision of the manuscript for important intellectual content (Drs Blesa, Queralt, Ezquerra, Molinuevo, Peña-Casanova, Rojo, Oliva); statistical expertise (Dr Lleó); obtained funding (Drs Lleó, Blesa); administrative, technical, and material support (Drs Lleó, Queralt, Ezquerra, Peña-Casanova, Rojo); study supervision (Drs Blesa, Molinuevo, Oliva).

This work was supported by the "II Beca Proyecto de Investigación" from the Fundación Sociedad Española de Neurología (Dr Lleó), and by grant 1999 SGR-00226 from the Generalitat de Catalunya, Barcelona (Dr Oliva), Barcelona.

We are grateful to Joan Martí-Aliaga, PhD, for his statistical assistance. We also thank the following physicians who contributed with additional cases: Nolasc Acarín, MD; Miguel Aguilar, MD; Ruth Djaldetti, MD; Carmen Antúnez, MD; Begoña Berlanga, MD; Cristóbal Carnero, MD; Esther Cubo, MD; and Jordi Gendre, MD.

Corresponding author and reprints: Rafael Oliva, MD, Genetics Service, Hospital Clínic, Villarroel 170, 08036 Barcelona, Spain (e-mail: roliva@clinic.ub.es).