Background
According to the World Health Organization (WHO), Alzheimer’s disease (AD) is the most common form of neurodegenerative dementia affecting nearly 50 million people worldwide with higher incidence rates as age increases [
1]. Among various described genetic risk factors [
2], the ε4 allele of the
APOE gene (
APOE ε4) confers the strongest genetic risk to develop AD and dementia with Lewy bodies (DLB) [
3]. Apolipoprotein E (apoE) is a 34-kDa glycoprotein, produced in the brain mainly by astrocytes [
4] whereas in the periphery, more than 90% of the circulating apoE is derived from hepatocytes [
5]. In humans, the
APOE gene is polymorphic and exists in three common variants: ε2, ε3, and ε4 [
6], where the ε3 allele is far more common than the ε2 and ε4 variants [
7]. Harboring the ε4 allele is linked to a higher risk of AD and a younger age of disease onset [
8]. Although the mechanisms that promote the increased risk of AD risk are not fully understood, various mechanisms suggesting loss of apoE neuroprotective function or a gain of neurotoxic function through amyloid-β (Αβ) dependent and/or independent cascades were suggested [
9]. Despite the established connection between the ε4 variant and AD, cerebrospinal fluid (CSF) levels of total apoE and apoE isoforms appear not to be altered in AD patients versus controls nor differ between subjects with different
APOE genotypes [
10,
11]. Instead, low plasma apoE levels were shown to be directly associated with an increased risk of AD and all types of dementias, whereas higher levels appeared protective [
12]. We and others have further documented that the presence of ε4 was associated with lower levels of total plasma apoE, which was mainly due to a specific reduction of the apoE4 isoform [
10,
13,
14]. Interestingly, although plasma apoE is unable to cross the blood–brain barrier [
5], we have also found that a higher ratio of plasma apoE4 to apoE3 levels was linked to negative brain imaging findings including gray matter atrophy and lower cerebral glucose metabolism [
15], worse cognition, and more pathological CSF AD biomarker levels [
16]. In a mouse model with humanized livers, we recently demonstrated that carrying ε4 specifically in the liver was associated with various pathological changes in the brain, and plasma apoE4 levels were significantly and negatively correlated with various synaptic marker levels in the hippocampus [
17].
Despite the well-documented connection between
APOE ε4 and neuropathological processes implicated in AD [
18], it is important to highlight that also non-ε4 carriers develop AD which illustrates that other factors (i.e., environment, race/ethnicity, sex) may independently or synergistically promote or modify the risk of AD. There is a higher prevalence of the ε4 allele in Black/African American (B/AA) than non-Hispanic white (NHW) adults [
19], which was previously discussed to partially explain the higher incidence of AD in older B/AA individuals [
20‐
23]. However, subsequent studies found that B/AA ε4 carriers are at a relatively lower risk of developing AD than NHW ε4 carriers [
24‐
26], possibly due to a protective variant (rs10423769) located on chromosome 19 [
27]. More recently, it was proposed that ε4 ancestry (European compared to African local genetic ancestry) influenced
APOE ε4 expression in the brain and that this expression difference may underlie previously documented variation in ε4-induced AD risk between populations of different races/ethnicities [
28]. Even though the ε4-promoted risk of AD appears to be lower in B/AA adults, there is a higher general risk of AD and cognitive impairment in this population compared to NHW individuals [
29‐
31].
Overall, since low plasma apoE levels were associated with a higher risk of AD [
12] and specifically the
APOE ε4 genotype appears to contribute to this higher risk by its association with lower plasma apoE levels [
10,
32], we hypothesized that the difference in AD risk between populations might be attributed to corresponding differences in their plasma apoE levels, hence individuals with higher AD risk exhibiting lower plasma apoE levels and vice versa. In the current study, we aimed to assess the plasma apoE levels (total and isoform levels) in a cohort of older B/AA and NHW participants with normal cognition, mild cognitive impairment (MCI), or mild AD dementia with detailed neuropsychological, CSF, and magnetic resonance imaging (MRI) analysis [
33]. Our main objective was to test whether, similar to NHW ε4 carriers, B/AA ε4-carriers also exhibit lower plasma apoE levels than non-carriers [
10,
13,
14,
16]. Furthermore, we aimed to assess potential associations between plasma apoE levels, global cognition, AD CSF biomarkers (amyloid-β40 (Aβ
40), amyloid-β42 (Aβ
42), total tau (t-tau), and phosphorylated tau at threonine (Thr) 181 (p-tau181)) as well as CSF levels of neurofilament light (NfL) chain, soluble TREM2 (sTREM2), and plasma lipids. Last, we investigated whether the distribution of plasma apoE monomers and dimers in non-ε4-carriers differed between the studied diagnostic and ethnic groups.
Discussion
The risk of developing AD is approximately two-fold higher in B/AA subjects compared to NHWs [
20]. Nevertheless, the AD risk-promoting effect of the
APOE ε4 genotype, the strongest genetic risk factor for sporadic AD, appears lower in B/AA compared to white individuals [
44]. With strong support for a direct association between low plasma apoE levels and an increased risk of AD and other types of dementia [
12], we aimed to assess whether the lower ε4-induced risk of AD in B/AAs may be attributed to a different plasma apoE profile than that previously reported in whites [
10,
14,
16].
Contrary to studies that have illustrated a higher frequency of ε4 allele in B/AAs [
19,
23], no such observation was made in the current cohort, possibly due to the low sample size. In line with previous reports, we found no difference in plasma apoE levels between the diagnostic groups in either racial/ethnic group [
10,
13], confirming that plasma apoE levels per se are not suitable as an AD diagnostic biomarker. In line with our hypothesis of B/AA
APOE ε4-carriers exhibiting a different plasma apoE profile compared to NHWs, potentially underlying their lower
APOEε4-promoted risk of AD, we indeed found significantly higher levels of plasma apoE in B/AA
APOE ε4/ε4-carriers compared to NHWs with the same genotype, despite the very small sample size. There was also no significant difference in plasma apoE levels between
APOE ε4 non-carriers, heterozygous, and homozygous B/AA subjects whereas there was a strong trend towards lower plasma apoE levels in the NHW
APOE ε4 homozygous subjects compared to non-carriers, in line with earlier reports [
10,
14,
32,
45]. That the latter difference did not reach statistical significance we argue is due to the small sample size.
In both racial/ethnic groups, the apoE4 isoform was less prominent and contributed the least to the total apoE levels in
APOE ε4 heterozygotes, which is in line with previous studies [
10,
13,
15,
32]. Previous studies have reported a higher turnover rate of the apoE4 isoform in plasma, compared to that of apoE2 and apoE3 [
46]; however, potential differences in hepatic
APOE allele expression in heterozygous individuals have yet to be investigated. In the human brain, it was shown that the
APOE ε4 allele was significantly higher expressed than the non-
APOE ε4 allele in heterozygous subjects; however, another study suggested significantly higher expression levels of
APOE ε4 due to polymorphisms in the
APOE gene promotor [
47,
48]. A recent study by Reddy and colleagues reported plasma cell-free mRNA levels of 50 AD-relevant genes, including
APOE, in B/AA AD patients and cognitively unimpaired controls; however, it remains to be determined how plasma transcripts of
APOE relate to protein levels of the same. Previous results from Griswold and colleagues proposed the notion of race/ethnicity-driven differences in
APOE expression showing that the central nervous system expression of
APOE, specifically in the frontal cortex, differed between
APOE ε4/ε4 AD patients with African local genomic ancestry versus European local ancestry [
28], with significantly higher levels of apoE in the latter group. Based on our findings, we speculate that ancestry-related effects may also influence plasma apoE levels, specifically in
APOE ε4/ε4 subjects despite the low sample numbers, although our observed effect is in plasma and in the opposite direction to that reported by Griswold et al. A measurement of genomic ancestry on the subjects included in the current cohort was not performed but such an analysis may indeed clarify a potential relationship with plasma apoE levels. Furthermore, we did not find any significant difference in plasma apoE levels between B/AA males and females contrasting previous findings of higher plasma apoE levels in female versus male Caucasian subjects [
15]. In NHW females, we did however observe a trend of higher plasma apoE levels compared to males (
p = 0.062).
In addition to their relevance to dementia, plasma apoE levels have also been implicated in ischemic heart disease [
49]. Studies of white individuals have revealed a positive link between elevated levels of plasma apoE and cardiovascular-associated mortality [
50] as well as the risk of ischemic heart disease in males [
51]. Heart diseases are more frequent in B/AAs compared to NHWs [
52], which could be attributed to environmental (i.e., lifestyle, socioeconomic factors) as well as genetic reasons [
53,
54] and share similar risk factors with sporadic AD [
55], including altered plasma lipids. In our study, B/AAs exhibited higher occurrence of diabetes and hypertension as well as higher state and national deprivation index scores compared to NHWs suggesting that B/AAs were living in a less affluent block area. To our knowledge, no studies have assessed any potential associations between plasma apoE levels and cardiovascular-associated mortality in B/AAs. We did not observe any association between plasma apoE levels and statin treatment, any of the known clinical comorbidities (CAD, CHF, hypertension, hyperlipidemia, cancer, diabetes, Afib, chronic renal failure, COPD, stroke/TIA), or scores on the area deprivation index, which is a better reflection of social economic status than household income as older people may have more assets than incomes.
In our study, B/AA MCI patients exhibited lower plasma triglyceride levels compared to the NHW patients. Higher plasma HDL and lower triglycerides in B/AAs versus white subjects were previously reported and implicated a differential risk of cardiovascular disease in the two groups [
56]. That plasma lipid levels are of relevance to neurodegenerative disease processes and AD has been demonstrated in several studies [
57]. For instance, higher plasma triglyceride levels at mid-life were linked to AD brain pathology two decades later [
58]. A recent lipidomics study of AD patients and healthy subjects further highlighted an altered plasma lipidome alongside AD pathology [
59]. In the current study, we observed higher levels of total cholesterol in AD patients compared to MCI patients and controls. Increased total and LDL cholesterol levels were also observed in the studied
APOE ε4 carriers. High triglyceride levels or low HDL levels were previously associated with cognitive impairment [
60‐
62] and a recent meta-analysis indeed demonstrated a positive link between AD risk and elevated levels of total cholesterol and LDL particles [
63]. In our study, we noted that total plasma apoE levels were significantly and positively correlated to plasma LDL levels in the B/AA study subjects whereas plasma apoE levels were significantly associated with HDL in the NHWs. Specifically, plasma apoE4 isoform levels in
APOE ε3/ε4 subjects were significantly associated with plasma LDL in B/AAs only. Whereas plasma LDL levels are causally associated with vascular disease and atherosclerosis, it appears the protective function of HDL against the same and can be lost [
64]. If the lipid composition may vary in an ancestry-dependent manner needs to be established. Importantly, we repeatedly noticed positive correlations between different plasma lipids and CSF tau levels, regardless of clinical diagnosis and
APOE genotype, indicating an unfavorable association between plasma lipids and tau pathology. Furthermore, gender is well known to influence plasma lipid levels [
27], and we also documented, within each racial/ethnic groups, higher levels of cholesterol in females compared to males. Only in B/AAs we further found higher levels of triglycerides in males compared to females. In the B/AA controls, MMSE scores were negatively correlated with triglycerides and positively associated with HDL in support of a link between cognition and the plasma lipid profile.
Aside from a differential risk of AD, differences in the levels of CSF AD biomarkers have been observed between races/ethnicities. Black/African-Americans, including the subjects investigated here, have in several studies exhibited lower CSF levels of t-tau and p-tau compared to white patients, but also lower levels of CSF sTREM2 [
33,
65,
66]. We previously reported that low plasma apoE levels were unfavorably linked to cognition and to CSF AD biomarker levels in a sample of ethnic Norwegian subjects from a cohort of longitudinally followed MCI and AD patients [
16]. In a recent follow-up study of the same cohort, we in addition discovered a significant correlation between low plasma apoE levels, Aβ brain pathology, and transition from amnestic MCI to AD dementia [
32]. In the current study, we found correlations of opposite directions between plasma apoE and CSF t-tau levels in B/AA versus NHW controls, with the latter exhibiting a negative correlation in line with the notion that lower plasma apoE levels may not be beneficial in these subjects [
12,
16]. A significant race/ethnicity-by-
APOEε4 interaction was previously reported for CSF tau levels [
65]; however, whether plasma apoE contributes directly or indirectly to AD risk in ε4 negative B/AA by influencing the levels of CSF tau levels, indicative of brain tau pathology, needs further research. We did not attempt to classify the study subjects based on amyloid/tau/neurodegeneration (A/T/N) since a universal cutoff for mainly t-tau and p-tau may be misleading with B/AA patients reported to exhibit lower CSF t- and p-tau [
33,
65,
67,
68]. Furthermore, amyloid-β plaque pathology was proposed to be negatively linked to African ancestry [
69]; hence, the development of individualized biomarker cutoff in which race/ethnicity is taken into account, is needed.
The functional relevance of plasma apoE levels in relation to both AD risk and levels of CSF biomarkers is complicated as apoE in subjects with the ε2 and/or ε3 occurs in different monomer, dimer, and multimer formations. In vitro studies have shown that apoE-apoA-II heterodimers have a higher affinity for the Aβ
42 peptide compared to monomers [
70] preventing it from neuronal endocytosis [
71]. Furthermore, it was previously shown that in
APOE ε3/ε3 individuals, apoE dimers account for 55% of total plasma apoE [
39]. In our previous study of
APOE ε3/ε3 subjects included in a longitudinal followed Norwegian cohort, we replicated this finding and further observed that the plasma apoE distribution in monomers and dimers was altered in AD patients [
16]. Here, we instead found that approximately 45% of the plasma apoE in
APOE ε3/ε3 B/AAs and NHWs occurred as dimers and further noted higher amounts of monomers over dimers in subjects with
APOE ε2/ε3,
APOE ε2/ε4, and
APOE ε3/ε4. We also observed that the ratio of apoE monomers/dimers was unrelated to clinical diagnosis and CSF AD biomarkers. Interestingly, previously, it was shown that CSF and brain apoE dimers did not differ between AD patients and cognitively healthy controls [
41,
43,
72]. In our earlier work, we however observed that plasma apoE3 homodimers were linked to worse cognition and higher tau/Aβ
42 ratios in
APOE ε3/ε3 Norwegian individuals [
16], suggesting a positive association between plasma apoE3 homodimers and AD pathology. In the current study, we did not observe any significant correlations between CSF AD biomarkers or cognition with either plasma apoE monomers or dimers; hence, more efforts are needed to elucidate whether specifically less apoE3 dimers in
APOE ε3 homozygous AD patients replicate in other cohorts [
16]. Plasma apoE dimer formation is linked to lipoparticle formation [
16] and the observed strong correlations between apoE monomers, dimers, and plasma lipids would support that cohort differences in diet may in part explain the difference in apoE dimers observed in the current study compared to previous studies [
16,
39]. Plasma apoE dimers have a lower affinity for the LDL receptor [
39,
73], making them less capable of delivering lipids; thus, we speculate that a lipid-enriched western diet might require a higher presence of apoE monomers which may be more efficient in assisting lipid metabolism. Furthermore, we recently showed that plasma apoE3 isoform levels were negatively associated with plasma glucose levels in
APOE ε3/ ε4 subjects with body mass index (BMI) above 25 [
74]. The mechanistic underpinnings of the relationship between plasma apoE3 dimers and glucose levels remain to be investigated, specifically since higher dietary fat intake is associated with impaired glucose tolerance [
75].
Black/African Americans are significantly underrepresented in AD research and the limitations of our study in addition to the overall low sample size, include the low number of APOE ε2 and APOE ε4-carries in both racial/ethnic groups, the rather low number of APOE ε4 homozygotes and the complete absence of homozygous APOE ε2/ε2 individuals. Due to the low sample size, our results are more prone to type I and II statistical errors. Our study is also limited by the inability to assess whether plasma apoE levels were associated with lifestyle-associated factors like BMI, tobacco, and alcohol use.