APOE3 Christchurch Heterozygosity and Autosomal Dominant Alzheimer’s Disease

Abstract

BACKGROUND

Variants in APOE and PSEN1 (encoding apolipoprotein E and presenilin 1, respectively) alter the risk of Alzheimer’s disease. We previously reported a delay of cognitive impairment in a person with autosomal dominant Alzheimer’s disease caused by the PSEN1^E280A variant who also had two copies of the apolipoprotein E3 Christchurch variant (APOE3^Ch). Heterozygosity for the APOE3^Ch variant may influence the age at which the onset of cognitive impairment occurs. We assessed this hypothesis in a population in which the PSEN1^E280A variant is prevalent.

METHODS

We analyzed data from 27 participants with one copy of the APOE3^Ch variant among 1077 carriers of the PSEN1^E280A variant in a kindred from Antioquia, Colombia, to estimate the age at the onset of cognitive impairment and dementia in this group as compared with persons without the APOE3^Ch variant. Two participants underwent brain imaging, and autopsy was performed in four participants.

RESULTS

Among carriers of PSEN1^E280A who were heterozygous for the APOE3^Ch variant, the median age at the onset of cognitive impairment was 52 years (95% confidence interval [CI], 51 to 58), in contrast to a matched group of PSEN1^E280A carriers without the APOE3^Ch variant, among whom the median age at the onset was 47 years (95% CI, 47 to 49). In two participants with the APOE3^Ch and PSEN1^E280A variants who underwent brain imaging, ¹⁸F-fluorodeoxyglucose positron-emission tomographic (PET) imaging showed relatively preserved metabolic activity in areas typically involved in Alzheimer’s disease. In one of these participants, who underwent ¹⁸F-flortaucipir PET imaging, tau findings were limited as compared with persons with PSEN1^E280A in whom cognitive impairment occurred at the typical age in this kindred. Four studies of autopsy material obtained from persons with the APOE3^Ch and PSEN1^E280A variants showed fewer vascular amyloid pathologic features than were seen in material obtained from persons who had the PSEN1^E280A variant but not the APOE3^Ch variant.

CONCLUSIONS

Clinical data supported a delayed onset of cognitive impairment in persons who were heterozygous for the APOE3^Ch variant in a kindred with a high prevalence of autosomal dominant Alzheimer’s disease. (Funded by Open Philanthropy and Good Ventures and others.)

In Antioquia, Colombia, there is a large family of approximately 6000 blood relatives, including more than 1000 carriers of the E280A variant of the PSEN1 gene (encoding the protein presenilin 1). Autosomal dominant Alzheimer’s disease is destined, with near 100% certainty, to develop in these carriers of PSEN1^E280A. Most of the carriers in this kindred have mild cognitive impairment in their mid-forties and dementia in their late forties.¹

Common variants of the apolipoprotein E gene (APOE) influence the risk of Alzheimer’s disease: APOE4 is linked to high risk, APOE3 is considered to confer a neutral risk, and APOE2 is associated with relative protection.^2–5 In addition to APOE, several other genes, when variant, cause susceptibility to Alzheimer’s disease; one of these is PSEN1.⁶ We previously reported the case of a person with the PSEN1^E280A variant who also had two copies of the rare Christchurch variant (R136S) in APOE3 (APOE3^Ch); in this person, mild cognitive impairment developed during her seventies — almost three decades after the expected age at onset.⁷ In vivo and postmortem analyses showed that the APOE3^Ch variant was linked to reduced tau accumulation and reduced neuroinflammation, which led to less neurodegeneration and less cerebral amyloid angiopathy within the context of higher cortical amyloid burden than is observed in persons with the PSEN1^E280A variant and without the APOE3^Ch variant.^7,8 We aimed to ascertain whether heterozygosity for the APOE3^Ch variant (APOE3^Ch/e3, APOE3^Ch/e2, or APOE3^Ch/e4) would delay the age at onset of mild cognitive impairment or dementia in 27 persons with this genotype among members of the Colombian kindred with the PSEN1^E280A variant, which is associated with autosomal dominant Alzheimer’s disease.

METHODS

STUDY DESIGN

In this retrospective study, we investigated data from a cohort of 1077 descendants in a family with the PSEN1^E280A variant,⁹ which was assessed by our group from 1995 to 2022 (Fig. S1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). The study was approved by the institutional review boards of the University of Antioquia, Colombia, and the Mass General Brigham integrated health care system. All the participants (or their representatives) provided written informed consent before the initiation of study procedures. (The document was read to participants who were unable to read, and they were asked to sign, if possible, or to provide a fingerprint.) Enrolled participants were 18 years of age or older and had at least one parent with the PSEN1^E280A variant, and therefore, mild cognitive impairment followed by autosomal dominant Alzheimer’s disease was destined to occur. During data collection, investigators were unaware of the genetic status of the participants.

Participants underwent regular clinical and neuropsychological assessments. The intervals between testing have been analyzed previously and were homogeneous across APOE3^Ch carriers and noncarriers.^1,10 In one report involving this kindred,¹ testing in a group of 309 persons with the PSEN1^E280 variant took place at a mean interval of 2 years (range, 1 to 11), and in another report,¹⁰ at a median interval of 2 years (interquartile range, 1 to 3).

Neuropsychological assessments were conducted in Spanish with the use of a battery that has been adapted and validated for the detection of Alzheimer’s disease–related cognitive impairment in this kindred, which included the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), the Trail Making Test, and the Rey–Osterrieth complex figure, among other assessments.^10–12 Using these data, neurologists and neuropsychologists, who were unaware of the genetic variants in the specific participants, retrospectively classified participants as having normal cognitive status, mild cognitive impairment, or dementia, according to the National Institute on Aging and the Alzheimer’s Association Workgroup criteria for mild cognitive impairment due to Alzheimer’s disease¹³ and the related Workgroup criteria for dementia due to Alzheimer’s disease.¹⁴

We also used the Functional Assessment Staging (FAST) system, which is a scale to assess the level of functional impairment in patients with dementia (scores range from 1 [no impairment] to 7 [total dependence]), and the Mini-Mental State Examination (MMSE; scores range from 0 to 30, with lower scores indicating greater cognitive impairment). The criteria for mild cognitive impairment were as follows: for persons with less than 9 years of education, a FAST score of at least 3 and an MMSE score of no more than 24; and for persons with 9 or more years of education, an MMSE score of no more than 26. The criteria for mild dementia were as follows: for persons with less than 9 years of education, a FAST score of at least 4 and an MMSE score of no more than 22; and for persons with 9 or more years of education, an MMSE score of no more than 24.

PARTICIPANTS

In an extended family of 1077 PSEN1^E280A carriers in whom autosomal dominant Alzheimer’s disease was destined to develop, we identified 121 carriers of the APOE3^Ch variant (Fig. S2), among whom 1 had the previously reported PSEN1^E280A variant and was homozygous for APOE3^Ch and 27 were carriers of PSEN1^E280A and were heterozygous for the APOE3^Ch variant. Of these 27 persons who were heterozygous for PSEN1^E280A and APOE3^Ch, 1 died at 57 years of age without cognitive symptoms (last examination at 2 years before death), 13 had mild cognitive impairment or dementia, and 13 did not meet the criteria for mild cognitive impairment or dementia (and are not identified in the family tree in order to protect privacy) (see the Supplementary Appendix).

These 27 persons who were heterozygous for the APOE3^Ch and PSEN1^E280A variants (i.e., the group of interest) had extensive clinical data and underwent neuropsychological testing. Two participants also had neuroimaging data, and 4 also underwent postmortem brain examinations. Genotyping was performed for the PSEN1^E280A and APOE variants, as previously described.⁷

NEUROIMAGING METHODS

One participant underwent amyloid (¹¹C-Pittsburgh compound B) and tau (¹⁸F-flortaucipir) positron-emission tomographic (PET) imaging at Massachusetts General Hospital in Boston. Two participants (Participants 1 and 2) underwent structural magnetic resonance imaging and ¹⁸F-fluorodeoxyglucose (FDG) PET imaging at the Hospital Pablo Tobón Uribe in Medellín, Colombia.

NEUROPATHOLOGICAL METHODS

Neuropathological material, which was obtained with the use of previously described techniques,⁸ was available from four persons who were heterozygous for the PSEN1^E280A and APOE3^Ch variants (Participants 3 through 6), from five carriers of PSEN1^E280A and APOE3 (Participants 7 through 11), and from one person who was a PSEN1^E280A carrier and was homozygous for APOE3^Ch (Participant 12). Immunohistochemical staining was performed uniformly for pathological markers of Alzheimer’s disease in cortical areas, followed by quantitative morphologic evaluation of vascular pathologic features in all the participants. ImageJ software, version 1.52p (National Institutes of Health), was used for image analysis for all findings. A description of these methods is provided in the Supplementary Appendix.

STATISTICAL ANALYSIS

We retrospectively estimated the cumulative incidence function, which represents the probability of mild cognitive impairment or dementia over time, considering the time from the date of birth to the onset of mild cognitive impairment or dementia in all the participants available from the cohort described above. We calculated the restricted mean and median survival times (i.e., the time to mild cognitive impairment or dementia) with the standard error and 95% confidence intervals, respectively, with categorization according to APOE3^Ch genotype. For participants with censored data, the times to these end points were calculated to the date of the last medical evaluation. We considered death to be a competing event for the analyses because death could occur before the onset of mild cognitive impairment or dementia.

To control for potential confounding factors of sex, APOE genotype, and years of formal education, we established a matched sample between carriers and noncarriers of the APOE3^Ch variant. For this matching, we used the nearest-neighbor method with the propensity score as a measure of similarity,¹⁵ resulting in a ratio of approximately 12 noncarriers to 1 APOE3^Ch carrier, which was determined by the MatchIt package in R software, after the maximum ratio was initially set at 20:1. To examine the balance of the groups before and after matching, we used a standard mean (or proportion) difference (Fig. S3). A subhazard ratio¹⁶ was calculated in an exploratory analysis to estimate the association between APOE3^Ch status and the time to mild cognitive impairment or dementia (Table S1 and Fig. S4). Statistical analyses were performed with the use of R software, version 4.3.1.¹⁷

RESULTS

CLINICAL AND NEUROIMAGING FINDINGS IN TWO LIVING PERSONS HETEROZYGOUS FOR APOE3^Ch

The first participant was a man with the genotype PSEN1^E280A–APOE3^Ch/e3 and 11 years of formal education. At 47 years of age, he had no subjective or objective cognitive problems or changes in daily functioning, and neurologic examinations and testing were normal (FAST score, 1; MMSE score, 28). At 49 years of age, his cognitive test performance was within normal limits for his age and educational level, and he reported having mild subjective memory concerns (FAST score, 2; MMSE score, 27). At 51 years of age, he received a diagnosis of mild cognitive impairment (FAST score, 3; MMSE score, 25) and had a decline in semantic fluency (decline in the CERAD Animal Fluency score from the 79th to 53rd percentile [range in normal populations, 25th to 75th]), executive functioning (decline in the Trail Making Test score from the 45th to 42nd percentile [range in normal populations, 25th to 75th]) and memory recall (decline in the CERAD Word List Recall score from the 23rd to 1st percentile [range in normal populations, 25th to 75th]). At 54 years of age, he received a diagnosis of mild dementia (FAST score, 4; MMSE score, 21).

PET imaging in this participant at 51 years of age showed slightly higher levels of brain cortical amyloid plaque (Aβ) burden than was seen in PSEN1^E280A carriers in whom Alzheimer’s disease had developed at the expected age in this kindred (distribution volume ratio [DVR], 1.68 vs. a mean [±SD] of 1.50±0.12 among PSEN1^E280A carriers with cognitive impairment) but a more limited tau burden in brain regions related to Alzheimer’s disease, including the entorhinal cortex (standardized uptake value ratio [SUVR], 1.33 vs. a mean of 1.60±0.29) (Fig. 1 and Fig. S5). When this participant was 53 years of age, ¹⁸F-FDG PET imaging showed a cerebral metabolic rate for glucose in the precuneus within the range of younger PSEN1 variant carriers in whom mild cognitive impairment had developed at a typical age in this kindred (SUVR, 1.15 vs. a mean of 1.23±0.17) (Fig. 1).

Figure 1. Limited Tau Pathological Findings and Relatively Preserved Regional Cerebral Metabolic Activity in Persons Heterozygous for the PSEN1^E280A and APOE3^Ch Variants.

Shown are baseline amyloid-β positron-emission tomographic (PET) images (Panel A; ¹¹C-Pittsburgh compound B [PiB]) and tau PET images (Panel B; ¹⁸F-flortaucipir). Sample coronal views (upper images) and sagittal views (lower images) are shown for each. For visual comparison, samples that were obtained from two participants are shown: from a typical person with the E280A variant of presenilin 1 (PSEN1^E280A) who had mild cognitive impairment (Panels A and B, left images), and from a person with the PSEN1^E280A variant who was also heterozygous for the APOE3^Ch variant (right images). The color-coded scale bar indicates the PiB distribution volume ratio (DVR) or flortaucipir standardized uptake value ratio (SUVR) values; blue represents the lowest binding and red represents the highest binding, with values ranging from 1.0 to 2.0. The person who was heterozygous for the APOE3^Ch variant had greater amyloid-β plaque burden and a relatively limited tau burden as compared with persons with the PSEN1^E280A variant in whom the onset of mild cognitive impairment occurred at the typical age in this kindred, as exemplified by the person whose images are shown here. Panel C shows ¹⁸F-fludeoxyglucose (FDG) PET images indicating the precuneus cerebral metabolic rate for glucose. In all three panels, the sample was obtained from a person with the PSEN1^E280A variant. The leftmost image in Panel C shows the sample of a person in whom the onset of mild cognitive impairment occurred at a typical age in this kindred, the middle image a sample from a person heterozygous for the APOE3^Ch variant, and the rightmost image a sample from a second person heterozygous for the APOE3^Ch variant. The color-coded scale bar indicates FDG SUVR values; blue represents lowest values and red represents highest values, with values ranging from 0.5 to 1.8. SUVR values represent the regional cerebral metabolic rate for glucose.

The second participant was a man with genotype PSEN1^E280A–APOE3^Ch/e2 who was illiterate and had received no formal education. His evaluation at 38 years of age showed slow processing speed and low verbal memory recall, although the scores were within the normal limits for his educational level (FAST score, 2; MMSE score, 27). He reported subjective cognitive concerns at 42 years of age, and his performance on the MMSE was reduced as compared with previous scores (FAST score, 2; MMSE score, 22) in the context of depression (Geriatric Depression Scale-15 score, 6; on a scale from 0 to 15, with a score above 5 indicating depression). His overall objective performance, however, was within the normal limits for his age and educational level. His scores were similar to those on previous verbal and nonverbal memory tests, and he did not have functional decline, so he did not meet the criteria for mild cognitive impairment at that time.

At 52 years of age, with no intervening testing after the previous battery, this participant had a decline in verbal memory, and his family reported having increased concerns about his cognitive condition despite preserved functional independence. At that time, he met the criteria for mild cognitive impairment (FAST score, 3; MMSE score, 25). (When the numerical results on the MMSE and FAST diverged, the clinical study team placed more emphasis on the FAST score because it reflects functional capabilities; the study team retrospectively classified this person as having mild cognitive impairment at this time point.) When this participant was 57 years of age, functional decline led to a diagnosis of mild dementia (FAST score, 4; MMSE score, not available). At 62 years of age, his memory skills declined further, although his naming and semantic fluency skills remained similar to the results on previous tests. He required more support to complete instrumental activities of daily living and had progression to moderate dementia (FAST score, 5; MMSE score, 18). When this participant was 64 years of age, FDG PET imaging showed a relatively preserved cerebral metabolic rate for glucose in the precuneus (SUVR, 1.35) (Fig. 1) as compared with PSEN1^E280A carriers in whom mild cognitive impairment had developed at the expected age. Imaging for Aβ and tau was not conducted owing to a lack of local availability and travel restrictions to the United States.

AGE AT ONSET IN PERSONS WITH PSEN1^E280A HETEROZYGOUS FOR APOE3^CH

To evaluate the time to mild cognitive impairment or dementia, with accounting for known influencing factors, we matched APOE3^Ch carriers with noncarriers on the basis of sex, educational level, and APOE genotype, as described above. This matching process yielded a subgroup of 353 participants for this part of the study (Table S2). Figures 2A and 2B show the time to mild cognitive impairment or dementia among the 1077 PSEN1^E280A carriers, including 27 carriers of the APOE3^Ch variant. Figures 2C and 2D show the data for the matched samples that had been obtained from 326 persons who did not have the APOE3^Ch variant and from 27 carriers of the APOE3^Ch variant.

Figure 2. Cumulative Incidence of Mild Cognitive Impairment and Dementia among Persons with the PSEN1^E280A Variant.

Shown are the cumulative incidence functions of mild cognitive impairment and dementia among persons with the PSEN1^E280A variant, 27 of whom had the APOE3^Ch variant and 1050 of whom did not. The analyses of mild cognitive impairment shown in Panel A and of dementia shown in Panel B included all the participants. The analyses of mild cognitive impairment shown in Panel C and of dementia shown in Panel D included matched samples from 27 persons with the APOE3^Ch variant and 326 persons without this variant. Participants were matched for sex, APOE genotype, and years of formal education. Death without a diagnosis of mild cognitive impairment or dementia was a competing risk. Shading indicates the 95% confidence interval.

Survival analyses of the matched samples showed that the median age at the onset of mild cognitive impairment was 52 years (95% CI, 51 to 58) among APOE3^Ch carriers, as compared with approximately 47 years (95% CI, 47 to 49) in the matched sample of noncarriers. The median age at the onset of dementia among APOE3^Ch carriers was 54 years (95% CI, 49 to 57), which indicated an apparent delay as compared with the noncarriers, among whom the median age at onset was 50 years (95% CI, 48 to 51) (Table S3). Some imprecision in the ages at the onset of mild cognitive impairment and dementia in both groups may have been introduced by gaps in testing that could have been as long as 5 years.

PATHOLOGICAL FINDINGS IN PERSONS WITH APOE3^CH

The four participants who were heterozygous for APOE3^Ch and had available autopsy material had a greater amyloid-β plaque burden and a relatively limited tau burden as compared with PSEN1^E280A variant carriers in whom mild cognitive impairment had occurred at the kindred’s typical age as determined by qualitative visual inspection, but this observation was not quantified. The regional distribution patterns of amyloid and tau in the participants were typical of those in persons with Alzheimer’s disease. In our previous article, we reported that the participant, who was homozygous for the APOE3^Ch variant, had a distribution of tau deposition in the cortex that largely spared the frontal lobe and was prominent in the occipital lobe.^7,8 In contrast, the brains of the APOE3^Ch heterozygous carriers in the current study did not show this pattern of tau deposition. Findings were within the expected range of tau deposition variation and without visible association between the extent of tau pathological findings and the age at onset of dementia (Table S4 and Fig. S6).

Pathological findings of cerebral amyloid angiopathy were less prominent in the frontal cortex in APOE3^Ch carriers (Fig. S7A) than in noncarriers. In addition, APOE3^Ch carriers had a numerically lower percentage of partial involvement of vessel walls (present only in a portion the annular circumferences of walls) than noncarriers, with pathological findings of cerebral amyloid angiopathy in frontal and occipital cortexes (Figs. S7B and S7C). The participant in the earlier study, who was homozygous for the APOE3^Ch variant, was less affected in all these measurements of cerebral amyloid angiopathy than the current cohort of participants who were heterozygous for the variant. However, direct comparisons between homozygous and heterozygous participants cannot be made because the homozygous participant in the earlier case report was much older than the participants in the current cohort. Furthermore, vessels that were analyzed in participants who were heterozygous or homozygous for APOE3^Ch showed greater similarity to controls in terms of length, branching pattern, and spacing between them than those obtained from PSEN1^E280A carriers without the APOE3^Ch variant (Fig. S8).

DISCUSSION

We report clinical, cognitive testing, neuroimaging, and neuropathological data from heterozygous APOE3^Ch variant carriers among participants with familial Alzheimer’s disease due to PSEN1^E280A from an extensively studied group of persons in Colombia. Within the limits of confidence that could be extracted from the data obtained from 27 participants who were heterozygous for the APOE3^Ch variant, as compared with 1050 participants who did not have this variant, we found that they had an onset of mild cognitive impairment and dementia, analyzed retrospectively, that was approximately 5 years later for mild cognitive impairment and 4 years later for dementia; we also found that they had different patterns on PET imaging. The precision of these estimates was probably not affected by irregular intervals between testing for the reasons noted above, including homogeneity of intervals across the entire group of persons with the PSEN1^E280A variant. The apparent delay in the onset of clinical features that were attributable to autosomal dominant Alzheimer’s disease in participants who were heterozygous for the APOE3^Ch variant was less than that observed in a previously reported case of APOE3^Ch homozygosity.

Our original report about one person in this population who was homozygous for APOE3^Ch7 also discussed four persons with the PSEN1^E280A variant who were heterozygous for APOE3^Ch, and we did not describe a delay in the age at onset of cognitive impairment. This finding led us⁷ and others¹⁸ to initially conclude that APOE3^Ch heterozygosity was not protective. The analyses that we present here encompassed a larger cohort with APOE3^Ch and PSEN1^E280A variants, with more comprehensive longitudinal clinical characterization, and indicated that APOE3^Ch heterozygosity is apparently linked to delay in the expected cognitive impairment.

The PET imaging findings in two participants heterozygous for the APOE3^Ch variant, which showed limited tau pathological findings and relatively preserved glucose metabolism, suggest that the delayed clinical onset that is associated with the APOE3^Ch variant may involve mechanisms that limit tau pathologic conditions and neurodegeneration, even in the presence of a high burden of Aβ amyloid plaque. These findings are consistent with our observations in the previously reported case of a person who was homozygous for the APOE3^Ch variant, but these are speculations and were not systematically studied owing to the small sample.

Limitations of this study arise from the relatively small number of persons who have both the APOE3^Ch and PSEN1^E280A variants, as well as the homogeneity of the population belonging to a genetic isolate.¹⁹ These limitations increase uncertainty around differences in the point estimates for the ages at the onset of mild cognitive impairment and dementia. Further studies involving larger and more ethnically diverse samples of persons with Alzheimer’s disease may shed light on any apparent protective effect of the APOE3^Ch variant. Furthermore, the biologic insight that the APOE3^Ch variant is protective in this group of persons from the Antioquia, Colombia, cohort may not translate to sporadic Alzheimer’s disease or to other groups.

The clinical, cognitive, neuroimaging, and neuropathological data that we present here provide evidence that APOE3^Ch heterozygosity delayed the onset of cognitive impairment in a form of autosomal dominant Alzheimer’s disease and may have a protective effect against Alzheimer’s disease and neurodegeneration in this population.