Onset of Alzheimer disease in apolipoprotein ɛ4 carriers is earlier in butyrylcholinesterase K variant carriers

Study eligibility criteria

Eligible study participants were between the ages of 50 and 74; had probable mild AD (amnestic or non-amnestic), defined by a Mini-Mental State Examination score (MMSE) (35) of 20-27, inclusive (representative of mild AD), and either a Clinical Dementia Rating (36) Overall Global Score of 1, or a Global Score of 0.5 with a Memory Score of 1; a CSF pattern of low Aβ_1-42 (≤ 1200 pg/ml), elevated total-tau (> 200 pg/ml) and p-tau (> 18 pg/ml), and a total-tau to Aβ_1-42 ratio > 0.28; and a diagnosis of probable AD based on National Institute of Aging-Alzheimer Association (NIA-AA) criteria (37).

Assessments

CSF from participants was analyzed for markers of amyloid accumulation (inversely indexed by Aβ₄₂), tau pathophysiology (p-tau₁₈₁), neuroaxonal degeneration (NfL), synaptic injury (Ng), glial activation (YKL-40). The CSF analytes and assays are detailed in Table 1. Of 46 patients entering the study, 45 were characterized for common SNPs of APOE (ie, APOE4, APOE3, and APOE2), and BCHE rs1803274 (ie, BCHE-K).

The study required 3-dimensional (3D) T1-weighted structural magnetic resonance imaging (MRI) scans of the head, volumetric analyses calculated using VivoQuant^TM, which is comprised of a preprocessing module and a multi-atlas segmentation module, followed by visual inspection and manual editing if needed (38). The mean baseline ventricular volume and hippocampal volume were expressed as a percentage of the total intracranial volume (%TIV).

Statistical analysis

Patient baseline characteristics were summarized according to genotype and sex (Tables 2-4). Quantitative assessments were summarized using descriptive statistics, including number of patients, mean, and standard deviation. Qualitative assessments were summarized using frequency counts and percentages. The exact test was used to examine the Hardy-Weinberg equilibrium (HWE) in the distribution of APOE and BCHE alleles in the study population. All exact tests were performed using the R package “Hardy Weinberg” (40). The HWE P value, which measures the strength of evidence against the null hypothesis that the distribution does not follow Hardy-Weinberg equilibrium, was provided. A large P value is consistent with the distribution following HWE.

Both analysis of variance (ANOVA) and analysis of covariance (ANCOVA) were used to test whether the mean age-at-diagnosis of AD (primary analysis) or the mean age-at-baseline differed across two or more genotype groups, ie, by BCHE-K carrier status in APOE4 carriers, and in APOE4 homozygotes and heterozygotes. When the ANCOVA model was applied, the model included BCHE-K carrier status and sex as factors, and baseline MMSE total score as covariate. Prior to performing ANOVA and ANCOVA, the normality assumption of residuals was tested using the Kolmogorov-Smirnov test. If significant departures from normality were observed, the Wilcoxon Rank Sum test was applied. Both ANOVA and ANCOVA were applied to test baseline CSF Aβ₄₂ across two or more genotype groups. When ANCOVA was applied, age-at-baseline was included as an additional covariate in the model. If the normality assumption was not satisfied, both ANOVA and ANCOVA model were fitted to the log-transformed data. Box plots were used to visualize data by group.

Relationships between CSF Aβ₄₂, CSF p-tau₁₈₁, other CSF biomarkers, and brain volumes (in terms of total hippocampal and total brain ventricular volumes as %TIV) were explored in the overall population and in each genotype group in a simple linear correlation analysis with a Pearson correlation coefficient. The squared Pearson correlation coefficient (R²) and P value were provided in the correlation analysis and interpreted by descriptors to indicate the strength of the relationship. Correlation coefficients with a magnitude of 0.81 < R² < 1 indicate variables were very highly/strongly correlated; 0.49 < R² < 0.81 indicates highly/strongly correlated; 0.25 < R² < 0.49 indicates moderately correlated; 0.09 < R² < 0.25 indicates a low/weak correlation; and R² < 0.09 indicates negligible correlation. Scatterplots with a simple linear regression line were produced to depict the relationships between two quantitative variables.

Multiple regression analysis was also utilized to assess the functional relationships between the biomarker of interest and amyloid and tau pathophysiology. A multiple regression model was applied with CSF Aβ₄₂ and CSF p-tau₁₈₁ as the response variables, and APOE4, BCHE-K, age-at-baseline, sex, baseline MMSE total score, and the biomarker of interest (ie, CSF p-tau₁₈₁ or Aβ₄₂, NfL, Ng, YKL-40; total hippocampal volume, total brain ventricular volume) as independent variables. This determined the strength of association of CSF Aβ₄₂ or CSF p-tau with parameters of interest, in conjunction with the other independent variables included in the model.

Age-at-diagnosis of AD in APOE4 carriers was reduced in carriers of BCHE-K alleles

There were no significant APOE4 carrier or allele frequency-associated differences on the participants’ age-at-diagnosis of AD or on their age-at-baseline of study (Table 2). APOE4 homozygotes (n = 10) had a mean age-at-diagnosis of AD of 64.2 years versus heterozygotes (n = 23) of 65.2 years, versus noncarriers (n = 12) of 63.3 years. In contrast, BCHE-K homozygotes (n = 3) had a lower mean age-at-diagnosis of AD of 59.4 years versus heterozygotes (n = 13) of 62.2 years, versus noncarriers (n = 29) of 66.0 years (P = .048, ANOVA) (Table S1).

In the primary analysis of this investigation, amongst APOE4 carriers, BCHE-K carriers (n=11) versus noncarriers (n=22) were associated with differences in mean age-at-diagnosis of AD of 6.4 years (60.6 versus 67.0; P < .001, ANOVA; P = .001, ANCOVA) (Figure 1A; Table 3), and mean age-at-baseline of study of 6.2 years (62.1 versus 68.3; P = .001, ANOVA; P = .002, ANCOVA) (Figure 1C). The difference in mean age-at-diagnosis of AD in BCHE-K carriers versus noncarriers was similar in APOE4 heterozygotes at 6.7 years (60.5 versus 67.3 years) and homozygotes at 5.7 years (60.8 versus 66.4 years) (P = .013, ANOVA; P = .019, ANCOVA) (Figure 1B). Significant differences of a similar magnitude were also seen in age-at-baseline of study (Figure 1D).

• Download figure
• Open in new tab

Figure 1. Age-at-diagnosis of AD and age-at-baseline by BCHE-K carrier status in APOE4 carriers, homozygotes, and heterozygotes

A. Age-at-diagnosis of AD in APOE4 carriers by BCHE-K carrier status

B. Age-at-diagnosis of AD in APOE4 homozygotes and heterozygotes by BCHE-K carrier status

C. Age-at-baseline of study in APOE4 carriers by BCHE-K carrier status

D. Age-at-baseline of study in APOE4 homozygotes and heterozygotes by BCHE-K carrier status

* P < .001, ANOVA; P = .001, ANCOVA with BCHE-K carrier status and sex as factors and baseline MMSE total score as covariate for mean age-at-diagnosis of AD.

** P = .013, ANOVA; P = .019, ANCOVA with APOE4 homozygotes and heterozygotes by BCHE-K carrier status and sex as factors and baseline MMSE total score as covariate for mean age-at-diagnosis of AD.

† P = .001, ANOVA; P = .002, ANCOVA with BCHE-K carrier status and sex as factors and baseline MMSE total score as covariate for mean age-at-study-baseline.

†† P = .015, ANOVA; P = .025, ANCOVA with APOE4 homozygotes and heterozygotes by BCHE-K carrier status and sex as factors and baseline MMSE total score as covariate for mean age-at-study-baseline.

In APOE4 noncarriers (N = 12), the mean age-at-diagnosis and age-at-baseline of the study were similar between BCHE-K carriers (n = 5) at 63.9 and 64.8 years, respectively, and noncarriers (n = 7) at 62.8 and 64.4 years, respectively (Table 3).

In APOE4 carriers with BCHE-K, younger mean age-at-baseline was accompanied by slightly higher amyloid and tau accumulations (Table 3)

Across genotype groups defined by APOE4 and BCHE-K carrier status the proportion of female participants fell as the burden of APOE4 and BCHE-K alleles diminished (Table 3). Carriers of both alleles had the earliest age-at-diagnosis, generally greater memory deficits, and the most amyloid accumulation. APOE4 carriers with BCHE-K had slightly higher amyloid and tau accumulations than APOE4 carriers without BCHE-K, despite a mean age-at-baseline that was 6.2 years earlier. In APOE4 carriers, neurodegeneration and glial activation were similar between carriers and noncarriers of BCHE-K (Table 3). On the other hand, noncarriers of both APOE4 and BCHE-K had the least amyloid pathology and multidomain cognitive deficits, with relatively less memory impairment. Individuals in these contrasting groups, defined by the presence or absence of both APOE4 and BCHE-K alleles, had similar levels of tau and neurodegenerative pathology (Table 3).

Higher APOE4 allele frequency associates with amyloid pathology and the limbic-amnestic phenotype (Table 2)

Female participants exhibited a higher APOE4 allele frequency than male participants. In APOE4 carriers, there was a more temporo-limbic (hippocampal atrophy > ventricular expansion) and amnestic (memory > visuospatial impairment) phenotype relative to APOE4 noncarriers (Table 2). APOE4 carriers relative to noncarriers had higher levels of amyloid pathology (inversely indexed by the CSF Aβ₄₂ = 664 versus 799 pg/mL, respectively; P = .028 ANOVA, P = .028 ANCOVA) (Table 2). APOE4 homozygotes had the highest levels of amyloid pathology (CSF Aβ₄₂ = 581 pg/mL) and the lowest levels of tau pathophysiology (CSF p-tau₁₈₁ = 35.2 pg/mL), neuroaxonal damage (NfL), and synaptic injury (Ng). Glial activation (YKL-40) showed apparent APOE4 allele frequency-dependent decreases to the lowest levels in homozygotes. Amyloid pathology was further increased in APOE4 homozygotes with BCHE-K (CSF Aβ₄₂ = 554 pg/mL), with further reductions of tau (CSF p-tau₁₈₁ = 34.0 pg/mL), neurodegenerative pathology, and glial activation (Table 2).

More amyloid pathology associated with less tau pathophysiology especially in carriers of both APOE4 and BCHE-K (Figure 2)

Multiple regression analyses indicated that associations with amyloid pathology (inversely indexed by CSF Aβ₄₂) included APOE4 carrier status (P < .029), larger total brain ventricle volume (P < .021), less synaptic injury (Ng, P < .001), less tau (p-tau₁₈₁, P = .005), and showed a trend for an association with less glial activation (P = .097). In simple linear correlation analyses in the overall population, amyloid pathology showed weak to moderate inverse correlations with tau pathophysiology, synaptic injury, and glial activation (Figures 2, 3B, and 3C). In carriers of both APOE4 and BCHE-K, inverse correlations of amyloid with both synaptic injury and tau pathophysiology strengthened to moderate and weak inverse correlation with glial activation persisted but was not statistically significant (Figures 2, 3B, and 3C). In APOE4 carriers without BCHE-K, inverse correlations of amyloid with tau pathophysiology were weak, and with synaptic injury and glial activation were moderate (Figures 2 and 3C).

• Download figure
• Open in new tab

Figure 2. Correlations of CSF Aβ₄₂ versus CSF p-tau₁₈₁ in the overall population and in APOE4 and BCHE-K subgroups

Simple linear correlation analyses: R square and P value were obtained by fitting a simple linear regression model.

• Download figure
• Open in new tab

Figure 3. Correlations of neurodegenerative and glial activation markers with CSF Aβ₄₂ and CSF p-tau₁₈₁ in the overall study population and in genotype groups

A. CSF neurofilament light chain (NfL) (higher levels index more neuroaxonal injury)

B. CSF neurogranin (Ng) (higher levels index more synaptic injury)

C. CSF YKL-40 (higher levels index more glial activation)

Simple linear correlation analyses: R square and P value were obtained by fitting a simple linear regression model. Only those analytes associated with either CSF Aβ₄₂ or CSF p-tau₁₈₁ in multiple regression analyses are shown.

Tau pathophysiology associated with synaptic injury and glial activation, especially in noncarriers of both APOE4 and BCHE-K, and with neuroaxonal degeneration, especially in APOE4 carriers without BCHE-K (Figure 3)

Multiple regression analyses indicated associations with tau pathophysiology (p-tau₁₈₁) with higher CSF levels of NfL (P = .002), Ng (P < .001), and YKL-40 (P = .01). Thus, more tau pathophysiology associated with neuroaxonal and synaptic degeneration, and glial activation. In simple linear regressions in the overall population, tau pathophysiology was moderately correlated with neuroaxonal damage, highly correlated with synaptic injury, and weakly correlated with glial activation (Figures 3A-C). Simple linear regressions showed correlations of tau pathophysiology with synaptic injury that was moderate in APOE4 carriers without BCHE-K, strong in carriers of both APOE4 and BCHE-K, and very strong in APOE4 non-carrier subgroups (Figure 3B). Simple linear regression showed tau pathophysiology had a moderate correlation with the neuroaxonal degeneration in the overall group, and this strengthened to highly correlated in APOE4 carriers without BCHE-K (Figure 3A). In simple linear regressions, correlation of glial activation with tau pathophysiology was weak in the overall population, weak in APOE4 carriers without BCHE-K, and very strong in noncarriers of both APOE4 and BCHE-K (Figure 3C). YKL-40 is proposed as a marker of tau pathology (41).

Amyloid pathology may accumulate at an earlier age in APOE4 carriers with BCHE-K alleles

APOE4/E4 individuals start accumulating amyloid at an earlier age, followed by APOE3/E4, APOE2/E4, APOE3/E3, and APOE2/E3 (13). The age of symptom onset may strongly correlate with the age at which an individual reaches a tipping point in fibrillar Aβ accumulation (44). In the current study, amyloid accumulation reached higher levels in genotype groups with lower levels of glial activation and an inverse correlation was observed between amyloid pathology and glial activation (Table 2; Figure 3C). Levels of amyloid pathology across genotype subgroups are compatible with APOE4 allele frequency-dependent accumulation beginning at an earlier age with the earliest start in APOE4 carriers with BCHE-K alleles, particularly in APOE4 homozygotes with BCHE-K (Table 2).

An age- and genotype-dependent phenotypic extreme of limbic-amnestic and amyloid-predominant mild AD is associated with low levels of glial activation (Figures 3 and 4; Table 2)

Optimally, glial activation may balance needs to attenuate amyloid accumulation and limit the spread of tau pathology (45, 46). This is an important determinant of the pathology mix and clinical features of early AD, including the age-of-onset of AD, and it links amyloid and tau pathology, cholinergic signaling, and glial activation (33).

• Download figure
• Open in new tab

Figure 4. Amyloid pathology facilitating phenotype of mild AD exemplified by APOE4 homozygotes with BCHE-K aged < 75 years

In preclinical and prodromal AD, lower BuChE activity, especially in APOE4 homozygote BCHE-K carriers below 75 years of age, results in higher extracellular ACh and increased signaling through nAChR on glial cells (1). “Functionally underactive” glia with decreased phagocytic function and homeostatic responsiveness (2), impair Aβ clearance and induce earlier and greater amyloid accumulation (3). In prodromal and mild AD, limbic cholinergic denervation due to ApoE4-, Aβ- and tau-mediated damage to basal forebrain cholinergic neurons, removes the cholinergic “brake” on glia to increase glial activation, tau pathology and neurodegeneration in the MTL (4), but the spread of this pathology outside of the MTL is initially limited until cholinergic denervation has progressed to include other neocortical areas (5). Thus, high levels of amyloid accumulation develop at a younger age, and basal forebrain cholinergic denervation of MTL structures results in hippocampal atrophy and a rapidly progressing limbic-amnestic presentation in mild AD with good response to AChE-Is (6). The progression of corticolimbic cholinergic denervation to neocortical areas beyond the MTL results in the spread of glial activation, tau pathology, and neurodegeneration.

At one end of the age- and genotype- and disease stage-dependent spectrum of early AD is the limbic-amnestic and amyloid accumulation-predominant phenotype, exemplified APOE4 homozygotes with BCHE-K alleles aged less than 75 years (Table 2). In APOE4 carriers with BCHE-K, greater accumulation of Aβ pathology at younger ages is likely due to greater deficits in glial-mediated clearance mechanisms (7, 47) (Figures 3C and 4). These functionally underactive glia may produce proinflammatory cytokines and be classified as “inflammatory,” but their endolysosomal and phagocytic functions may be greatly reduced (48). Putatively lower BuChE activity in BCHE-K carriers with APOE4 alleles results in higher extracellular ACh, further reduces the phagocytic function and responsiveness of glial cells, and further impairs Aβ clearance. While excessive cholinergic signaling encourages high levels of amyloid accumulation, the spread and accumulation of tau pathology is kept in check (Figures 2 and 4). However, in early AD, APOE4 is associated with rapidly increasing tau pathology as the combination of ApoE4 and Aβ pathology induces sufficient degeneration of basal forebrain corticolimbic cholinergic neurons to release the cholinergic “brake” on glial activation in the medial temporal lobe (MTL) and eventually also in other cortical regions (Figure 4). Cholinergic denervation and tau pathology accumulating in the MTL results in a rapidly progressing limbic-amnestic presentation with response to AChE-I that is apparent in the mild stage of AD (32, 33). The focal MTL/limbic denervation is indexed by a more limbic-amnestic presentation (memory > visuospatial impairment, and hippocampal atrophy > ventricular expansion) (Tables 2 and 3).

Both Aβ and ApoE modulate the cholinergic system

The cholinergic system acts to control glial reactivity and function though α7-nAChRs via rapid focal synaptic signaling and slow diffuse extracellular signaling (6). ApoE forms soluble and highly stable complexes with cholinesterase enzymes and Aβ, that can oscillate between slow and ultrafast ACh hydrolysis, depending on Aβ availability (29). Reduced cholinesterase activity and lowering of glial activation markers is seen in APOE4 carriers, particularly in individuals with polymorphic variants of genes encoding cholinesterase enzymes with lower activity, such as BCHE-K (24, 29). In addition, lower cholesterol delivery by ApoE4 to the long and extensively arborized axons of the metabolically taxed cholinergic neurons requires them to expend energy on cholesterol synthesis and to consume acetyl coenzyme A (acetyl-CoA) (49). Acetyl-CoA is an essential substrate for the synthesis of both ACh and lipids essential for myelin formation and maintenance of cellular membranes (50). The ascending white matter projections of the basal forebrain cholinergic system may be particularly vulnerable to the combination of Aβ pathology and ApoE4 (51–53).

In a concentration- and aggregation-dependent manner, Aβ targets cholinergic synapses, signals through α7-nAChRs, and also influences the extracellular fluid equilibrium between breakdown and synthesis of ACh via effects on ACh-hydrolyzing capacity of cholinesterase (22) and choline acetyltransferase (ChAT) activity (54). In both aging and AD, the intraneuronal accumulation of oligomeric assemblies of Aβ₄₂ is a relatively selective trait of basal forebrain cholinergic neurons (55, 56), and endocytic internalization of Aβ-nAChR complexes may underlie intracellular accumulation of Aβ₄₂ and its neurotoxic consequences, such as tau phosphorylation (57). In AD, α7-nAChR expression on astrocytes is positively correlated with neuritic plaque burden (58). In an amyloid mouse model, loss of α7-nAChRs reduced Aβ₄₂ plaque load, but increased levels of soluble Aβ₄₂ oligomers and accelerated memory decline (59). Thus, Aβ and ApoE have physiological and disease roles in the tuning of cholinergic activity and in the vulnerability of the basal forebrain cholinergic system to degeneration, and these actions influence the functional status of cholinoceptive neuronal and nonexcitable cells (17).

Hypofunctional glial-mediated clearance of Aβ likely underlies the “amyloid accumulating” phenotype exemplified by carriers of APOE4 and BCHE-K aged < 75 years

A balanced activation of glia may be needed to stimulate Aβ clearance and avoid “hypofunctional” glia promoting amyloid accumulation versus “hyperactivated” glia that facilitate the dissemination of tau (45, 46, 60). In amyloid mouse models, inhibition of reactive astrogliosis increases Aβ₄₂ plaque burden (61), whereas shifting microglia to an interferon-responsive state boosts ApoE expression, phagocytosis, containment of plaques, and lessens damage to nearby neurons and synapses (62). However, further shifting of microglia to an overactivated state may increase synaptic engulfment and accelerate the dissemination of tau pathology (63). Mouse models indicate that ApoE controls glial activation, with ApoE4 locking microglia in a homeostatic state with failure to clear neurodegenerative pathology (7, 47). Glia in APOE4 mice demonstrate a significant decrease in phagocytic capacity (64). Microglial and astrocyte coverage of plaques is likely protective for surrounding neurons, and ApoE4 is associated with decreased coverage and more neuronal dystrophy (65–67).

BCHE-K carriers, who have lower levels of glial activation markers and higher levels of proinflammatory cytokines (24, 68), may exhibit deficient glial responses to neurodegeneration. Rapid and appropriately tuned changes in the catalytic activity of BuChE, and necessary adaptive changes in cholinoceptive nonexcitable cells may be more difficult to achieve with BuChE-K, particularly in APOE4 carriers. Carriers of APOE4 and BCHE-K may have more senescent microglia that are associated with blocked endolysosomal processing, impaired phagocytosis, and with acceleration of Aβ₄₂ pathology (69). YKL-40 is a context-dependent modulator of glial phagocytic activity in both mice and humans (70). In the current study, APOE4 noncarriers had higher mean levels of CSF YKL-40, at 318 ng/mL, relative to APOE4 carriers at 247 ng/mL (Table 2). Mean levels were further reduced in APOE4 homozygotes at 202 ng/mL and in APOE4 homozygotes with BCHE-K alleles at 190 ng/mL (Table 2). This likely explains the inverse correlation between amyloid pathology and YKL-40 and synaptic injury seen in simple linear regression analyses overall and more specifically in APOE4 carriers (Figure 3B and 3C). In addition, a weak correlation between YKL-40 and synaptic injury overall was evident in BCHE-K non-carriers (Figure S1). The correlation persisted as weak in APOE4 carriers without BCHE-K and strengthened to high in non-carriers of both APOE4 and BCHE-K.

Increases in functional glial activation with age might explain why APOE4-associated risk for AD reduces over the age range of 70-80 years, and why progression to dementia in prodromal AD carriers of both APOE4 and BCHE-K is at least 2-fold higher in those aged below 75 years relative to older patients (33). In a longitudinal study of prodromal AD, progression to AD over 3-4 years in APOE4 and BCHE-K carriers was 39% in participants aged less than 75 years and decreased to 18% in those aged 75 years or more (33). This was contrary to findings in the overall study, where progression to AD was greater in older participants (29%), relative to those aged less than 75 years (13%). Younger APOE4 carriers have accelerated progression of hippocampal atrophy in prodromal and early stage AD, but in individuals more advanced in age or in progression of disease, any influence of APOE4 on hippocampal atrophy is lost (71). Moreover, global cerebral atrophy in AD patients with a mean age of 70 years was reduced in an APOE4 allele frequency-dependent manner (72), whereas in older patients, with a mean age of 80 years, atrophy was no different by genotype (71, 73).

The “amyloid accumulating and initially tau spread limiting” phenotype supports the amyloid cascade hypothesis

Carriers of APOE4 and BCHE-K with prodromal AD, in addition to progression rates to AD that inversely correlated with age, had the greatest hippocampal atrophy and declines in short- and long-term retrieval from verbal memory and overall cognitive impairment (27, 33). Likewise, in the current study of mild AD patients, APOE4 carriers relative to noncarriers had greater memory deficits, hippocampal atrophy, and amyloid accumulation, and these findings in APOE4 carriers occurred over 6 years earlier in BCHE-K carriers compared to noncarriers (Figure 1C, Table 3). The amyloid cascade hypothesis of AD implies that reaching the threshold for parenchymal amyloid positivity at an earlier age should drive secondary effector tau pathology, with an earlier age-at-onset of AD (74). Support for this hypothesis comes from slower progression of tau tangle accumulation and cognitive decline following antibody-induced removal of amyloid plaque below key threshold levels in patients with early symptomatic AD (75, 76). However, in the current study, correlations of amyloid pathology with tau pathophysiology in APOE4 carriers were negative, particularly in carriers of both APOE4 and BCHE-K (Figure 2). Across groups defined by APOE4 allele frequency, APOE4 homozygotes had the highest amyloid pathology and lowest tau and neurodegenerative pathology (Table 2). Amyloid pathology was further increased in APOE4 homozygotes with BCHE-K, with further reduction of tau and neurodegenerative pathology (Table 2).

The inverse correlations of amyloid pathology with tau pathophysiology and synaptic injury may be a consequence of tau and neurodegenerative pathology localized to the MTL (which includes the entorhinal cortex, amygdala, and hippocampus). This MTL tau and neurodegenerative pathology may be responsible for transitioning APOE4 and BCHE-K carriers into AD at an earlier age with less global tau and neurodegenerative pathology (Table 3; Figure 4) (77, 78). Such neuroanatomical distinctions are not discernible on CSF indices that summarize pathology changes across the brain. Spatial resolution requires tau tangle-ligand positron-emission tomography (tau-PET) neuroimaging. In the presence of global amyloid pathology in APOE4 carriers, tau-PET indicates that tau pathology is more severe with a focal MTL distribution (78, 79). Furthermore, younger age is associated with PET-tau signal in the MTL in early AD APOE4 carriers but not in APOE4 noncarriers (78). Across the aging and AD spectrum, APOE4 carriers present increased microglial activation relative to noncarriers in early Braak stage regions within the MTL and this microglial activation mediates Aβ-independent effects of APOE4 on tau accumulation that are further associated with neurodegeneration and clinical impairment (80).

Degeneration of basal forebrain cholinergic neurons, that project to the MTL and other cortical structures, precedes and predicts longitudinal entorhinal/MTL degeneration (81, 82) and preclinical APOE4 carriers exhibit the greatest loss of basal forebrain volume (83). The ascending neuronal projections of the basal forebrain cholinergic system may be particularly vulnerable to the combination of ApoE4-mediated glial hypofunction, high levels of Aβ, and tau pathologies (51–53). The impact of focal basal forebrain pathology is magnified as it causes widespread presynaptic cholinergic corticolimbic denervation. Both amyloid and tau pathologies may be required for substantial impairment of cholinergic synaptic plasticity and memory, and continuous destruction of the projecting branches of the cholinergic nuclei in the basal forebrain (84, 85).

Implications for future clinical research and development of therapeutics

Findings in the current study, if confirmed, could have implications for the conceptualization of Alzheimer pathological cascades, therapeutic targets, and the application of existing and future treatments. The genetic architecture of prognosis in AD, that is fundamental for patients and the design of clinical trials, is less well established than the genetics of susceptibility. BCHE-K may join APOE4 allele frequency, age, and sex as a foundational component of predictive modelling of early AD phenotypes (86, 87). Moreover, the cholinergic hypothesis appears seamlessly interlinked with the amyloid cascade hypothesis. APOE and BCHE genotypes appear to exert critical influence on the functional activation of glia as indexed by the microglial and astrocyte activation marker, YKL-40, in the CSF. Levels of extracellular cholinergic signaling to cholinoceptive cells, including glia, depend on the enzymatic activity of BuChE, that relies on BuChE levels and polymorphic variation, ApoE levels and polymorphic variation, and soluble Aβ levels. The health of cholinergic neurotransmission and extracellular signaling systems may be crucial to healthy brain aging (88).

The quantitative removal of amyloid pathology with anti-amyloid antibodies is dependent on Fc receptor-mediated phagocytosis and clearance of Aβ (75, 76). Response to this targeted immune activating therapeutic approach may vary depending on the individual’s predominant microglial state. While some beneficial effects might be due to antibody-mediated stimulation of glia with improved performance of homeostatic functions (89), stimulating the clearance of Aβ prior to substantial levels of corticolimbic cholinergic denervation and spreading MTL tau pathology may produce the best outcomes in APOE4 carriers below the age of 75 years, but this may require intervention in asymptomatic individuals. In addition, longer treatment durations may be necessary in substantial amyloid accumulators, such as APOE4 homozygotes with BCHE-K alleles, to get below the amyloid threshold to prevent or slow further corticolimbic cholinergic denervation and tau pathology.

While retuning of innate immune responses may be required to harness protective and beneficial effects and to attenuate negative effects, the required changes will differ across a genotype, age, and disease stage continuum. Tuning in the wrong direction will simply make matters worse. Considerations may be further complicated by the nuances and complexity of glial cell phenotypes across different brain regions, between adjacent glia, and in different disease contexts (90). The challenge in the development of potential amyloid pathology reducing therapeutics may not lie in simply upregulating the activation state of glia.