Neurobiology of Aging

Conflicting findings exist regarding the differences in amyloid burden and cognitive performance based on sex and apolipoprotein E ( APOE ) genotype. This study aimed to investigate the brain amyloid-β (A β ) burden and cognitive performances by sex and APOE genotype in a cohort of A β -positron emission tomography (PET)–positive participants. Brain A β burden was assessed using 18F-florbetapir PET standard uptake value ratios. Cognitive performance was evaluated using standardized neuropsychological tests. In the cognitively normal participants, females had a higher A β burden than males in APOE ε 4 noncarriers, whereas APOE ε 4 carriers had a higher A β burden than noncarriers in males. In the cognitively impaired participants, APOE ε 4 carriers were more likely to have a higher A β burden than noncarriers in the brain regions of the lateral parietal gyrus, frontal gyrus, and precuneus. In addition, females were more likely to have poorer language and visuospatial performance compared to males, while the APOE genotype did not significantly impact cognitive performance. These findings further elucidate the impact of sex and APOE genotype on brain A β burden and sex-related cognitive performance should be considered in the Alzheimer’s Continuum. © 2023 The Author(s


Introduction
Alzheimer's disease (AD) is the most prominent cause of dementia in which cerebral amyloid-β (Aβ) deposition precedes progressive neurodegeneration and cognitive dysfunction (Long and Holtzman, 2019). A vast majority of epidemiologic studies indicate that the prevalence and incidence of AD dementia are higher in females compared to males (Jia et al., 2020;Niu et al., 2017), though autopsy studies show similar amyloid plaque by sex but slightly more neurofibrillary tangles (NFTs) in females (Barnes et al., 2005;Filon et al., 2016). Previous studies have also demonstrated that the sex-specific prevalence of Aβ positive does not differ by age (Jack et al., 2017), and brain Aβ deposition plateaus at the end stage of the disease (Jack et al., 2013). Even so, as a primary driving force of AD that may lead to subsequent NFTs and neurodegeneration (Bakota and Brandt, 2016), whether the brain Aβ burden differs between males and females in different disease stages of AD has not been well investigated. The apolipoprotein E (APOE) ε4 genotype is the most common genetic risk factor for AD, with no significant difference in the distribution between males and females (Munoz et al., 2019). In the pathology of AD, APOE ε4 may increase Aβ aggregation and decrease Aβ clearance, which leads to the formation of Aβ plaques in the brain (Kim et al., 2009). Currently, there are conflicting findings regarding the sex and APOE genotype differences of amyloid burden in different cognitive stages. In a population-based study with cognitively normal (CN) individuals, sex was not associated with brain Aβ burden, while APOE ε4 was associated with a greater Aβ burden in individuals aged > 70 years (Jack et al., 2015). In contrast, 1 study showed that male carriers of APOE ε4 had lower cerebrospinal fluid Aβ compared to females across the diagnosis of CN, mild cognitive impairment (MCI), and probable AD (Duarte-Guterman et al., 2021). However, utilizing the Alzheimer's Disease Neuroimaging Initiative database, 1 study found that females have higher 18F-florbetapir (AV45) positron emission tomography (PET) standard uptake value ratios (SUVRs) than males in CN and AD dementia subjects but not in MCI subjects, and APOE ε4 is related to greater Aβ burden in all the CN and MCI subjects but only in males with AD (Sundermann et al., 2018). The discrepancies among these reports of sex-related and APOE genotype-related brain Aβ burden may be due to the differences in study designs and sample sizes. Additionally, not all the participants in these studies were Aβ positive.
As a core feature of AD, the Aβ deposition in the brain is estimated to take decades to go from the levels observed in healthy controls to the levels observed in AD dementia (Villemagne et al., 2013). Though the brain Aβ burden is less well correlated with cognitive impairment than neurodegenerative biomarkers (Jack et al., 2010), a higher baseline Aβ burden indicates faster cognitive decline (Farrell et al., 2017;Ottoy et al., 2019). However, previous studies that have investigated the relationships between sex/APOE ε4 and cognitive performance did not take the brain Aβ burden fully into account, and inconsistent results remained (Duarte-Guterman et al., 2021;Jack et al., 2015;Sundermann et al., 2018). In addition, most of these studies have focused on memory performance but not nonmemory performance.
The aims of this study were to evaluate the sex and APOE ε4 effects on the brain Aβ burden in an Aβ-PET-positive cohort with different cognitive states and further evaluate the sex and APOE ε4 effects on cognitive performance assessed by standardized neuropsychological tests.

Study participants
A total of 291 AV45 PET-positive individuals were retrospectively enrolled from an outpatient memory clinic and the general community from January 2019 to March 2022. All participants were Mandarin-speaking Han Chinese individuals aged 45-80 years, with the ability to finish a battery of standardized neuropsychological tests. Routine laboratory tests and cranial magnetic resonance imaging scanning were carried out. Individuals with evident abnormalities in folic acid, vitamin B12, and thyroid function or with a history of cerebral infarction, subdural hematomas, hydrocephalus, intracranial tumors, intracranial infections, head trauma, and psychiatric disorders were excluded. Participants were diagnosed with AD dementia by experienced neurologists if they met the criteria for probable AD dementia proposed by the National Institute on Aging and Alzheimer's Association (McKhann et al., 2011). The diagnosis of MCI was made if the participants met the actuarial neuropsychological criteria put forward by Jak and Bondi (Bondi et al., 2014). Participants who had no apparent cognitive complaint or self-reported cognitive decline experiences but performed normally on the standardized neuropsychological tests were classified as CN. Due to the small number of participants with MCI (n = 68) and AD (n = 92) dementia after being grouped by sex and APOE genotype, and the mean AV45 PET SUVRs of MCI and AD dementia being similar, the MCI and AD dementia subjects were combined and assigned to a subgroup of cognitively impaired (CI) subjects. The APOE genotype was confirmed by polymerase chain reaction assay and followed by direct sequencing. The APOE genotype was categorized into APOE ε4 carriers and noncarriers, given the low prevalence of APOE ε4 homozygotes when stratified by sex and cognitive status. All the participants underwent AV45 PET scans within 3 months after cognitive assessments. Written informed consent was obtained from all the participants or their caregivers. The ethics committee of Shanghai Sixth People's Hospital, affiliated with Shanghai Jiao Tong University School of Medicine, approved this study.

Cognitive assessment
The Chinese version of the Mini-Mental State Examination (Katzman et al., 1988) and the Montreal Cognitive Assessment-Basic  were used as the overall indices of general cognitive performance. The status of global function was assessed by Activities of Daily Living (Chen et al., 1995) and Functional Assessment Questionnaire (Pfeffer et al., 1982). Furthermore, a battery of standardized neuropsychological tests was carried out to assess the abilities of participants in different cognitive domains, including the Auditory Verbal Learning Test (Zhao et al., 2015) and Brief Visuospatial Memory Test-Revised (Benedict et al., 1996) for memory, Boston Naming Test (Guo et al., 2006) and Animal Verbal Fluency Test (Zhao et al., 2013) for language, Symbol Digit Modalities Test (Wechsler, 2008) and Digit Span Test (Wechsler, 2008) for attention, Clock Drawing Test (Huang et al., 2023) and Judgement of Line Orientation (Qualls et al., 2000) for visuospatial ability, and Shape Trail Test Part A and Shape Trail Test Part B for executive function. For each standardized neuropsychological test mentioned above, we generated the z-scores using 269 CN controls with an Aβ-PET negative as the reference. Following this, for each subject, the average z-scores for the tests of Auditory Verbal Learning Test and Brief Visuospatial Memory Test-Revised, Boston Naming Test and Animal Verbal Fluency Test, Symbol Digit Modalities Test and Digit Span Test, Clock Drawing Test and Judgement of Line Orientation, and Shape Trail Test Part A and Shape Trail Test Part B were calculated to yield the composite memory, language, attention, visuospatial, and executive scores, respectively.

Amyloid PET imaging
The cerebral amyloid-β deposition of all participants was measured with a PET/computed tomography system (Biograph mCT Flow PET/CT; Siemens, Erlangen, Germany). 18F-florbetapir was the first 18F-labeled PET tracer developed for measuring the accumulation of amyloid protein in the brain. In short, PET scans were performed 50 minutes after the intravenous injection of 7.4 MBq/kg (0.2 mCi/ kg) 18F-florbetapir and lasted for 20 minutes. PET images were reconstructed using filtered back projection algorithm with corrections for decay, normalization, dead time, photon attenuation, scatter, and random coincidences. Partial volume correction was not applied to the PET data. According to the guidelines of visual rating (Lundeen et al., 2018), 3 nuclear medicine physicians with specialized training carried out the interpretation of the PET images. The positive results of PET images were determined if more than 2 physicians made the same judgment. SUVR has been widely used as a proxy for the quantitative analysis of amyloid PET. All the brain amyloid burdens were expressed as the SUVR scores using cerebellar crus as a reference. The SUVR for the global cortical region was calculated by weighted averaging of the cortical regions of interest (ROIs), including frontal gyrus, lateral parietal gyrus, lateral temporal gyrus, medial temporal gyrus, posterior cingulate gyrus, occipital gyrus, and precuneus (Akamatsu et al., 2016).

Statistical analyses
Demographic and clinical characteristics of the groups stratified by sex, APOE genotype, and cognitive status were compared by oneway analysis of variance or χ 2 analyses based on the data types. In different cognitive states, general linear models were used to determine the effects of sex and APOE genotype on AV45 PET SUVRs (global cortical region and each ROI), with age and education years as covariates. In all the participants, sex and APOE genotype were also used as predictor variables, and cognitive performance (z-scores Table 1 Demographic and clinical characteristics of the study participants Index CN (Aβ+) for memory, language, attention, visuospatial, and executive abilities, respectively) as dependent variables, with age, education years, and AV45 PET SUVRs as covariates. All models initially included 2way interaction between sex and APOE genotype. The main effects of sex and APOE genotype were estimated if no significance was found for 2-way interactions. A significant interaction between sex and APOE genotype would then be followed up by Bonferroni multiple comparison tests regarding the sex and APOE genotype stratified groups. Analyses were performed using IBM SPSS Statistics 23.0. GraphPad Prism (version 8.0) was used to create figures. The level of significance was defined as p = 0.05 (2 sided).

Demographic and clinical characteristics
Demographic and clinical characteristics between the groups stratified by sex, APOE genotype, and cognitive status are presented in Table 1. There was no significant difference in the prevalence of hypertension, hyperlipidemia, and diabetes between the groups. Significant differences in age and education years were found between the groups, in which CN male APOE ε4+ subjects were relatively older than CI female APOE ε4− subjects, and CI female subjects had relatively lower education years than other groups. For the cognitive assessments, poorer cognitive outcomes in Mini-Mental State Examination, Montreal Cognitive Assessment-Basic, memory, language, attention, visuospatial and executive abilities, Functional Assessment Questionnaire, and Activities of Daily Living were observed for the majority groups of CI compared to CN.

Associations between sex, APOE genotype, and brain Aβ burden
The effects of sex and APOE ε4 genotype on the brain Aβ burden were investigated in different cognitive states. In the participants with CN, the results of general linear models are listed in Table 2 to determine the effects of sex and APOE ε4 genotype on AV45 PET SUVRs in the global cortical region and each cortical ROI. Significant 2-way interaction between sex and APOE genotype for AV45 PET SUVR was found in the global cortical region and each cortical ROI (p = 0.009, 0.017, 0.008, 0.007, 0.009, 0.026, 0.008, and 0.016, respectively). Comparisons of AV45 PET SUVRs between the sex and APOE genotype stratified groups were displayed in Figure 1. In the global cortical region and cortical ROIs, AV45 PET SUVRs were all significantly higher in female APOE ε4 noncarriers than male APOE ε4 noncarriers (p = 0.0001, 0.0002, 0.0002, 0.0002, 0.0063, 0.0005, 0.0001, and 0.0059, respectively), but no sex difference was detected in APOE ε4 carriers. Except for the brain region of the medial temporal gyrus, the differences in AV45 PET SUVRs between APOEε4 carriers and noncarriers were found in males (p = 0.0014, 0.0031, 0.0019, 0.0002, 0.0010, 0.0466, and 0.0005, respectively) but not females. Except for the brain regions of the medial temporal gyrus and occipital gyrus, female APOE ε4 carriers also had higher AV45 PET SUVR than male APOE ε4 noncarriers (p = 0.0081, 0.0096, 0.0190, 0.0211, 0.0023, and 0.0175, respectively). In addition, age showed significant positive associations with AV45 PET SUVR in the global cortical region and cortical ROIs except for the medial temporal gyrus. Education level also showed significant positive associations with AV45 PET SUVR in the global cortical region and each cortical ROI.
In the participants with CI, no significant 2-way interaction between sex and APOE ε4 genotype for AV45 PET SUVR was found either in the global cortical region or in cortical ROIs. Furthermore, no significant main effect of sex was detected on the AV45 PET SUVR in any brain region. However, though the APOE ε4 genotype had no significant effect on AV45 PET SUVR in the global cortical region and cortical ROIs of lateral temporal gyrus, medial temporal gyrus, posterior cingulate gyrus, and occipital gyrus, APOE ε4 carriers did have higher AV45 PET SUVR levels than noncarriers in the cortical ROIs of lateral parietal gyrus, frontal gyrus, and precuneus (p = 0.046, 0.042, and 0.033, respectively). In addition, though no significant correlation between age and AV45 PET SUVR was observed in the global cortical region and each cortical ROI, education was positively correlated with AV45 PET SUVRs in these areas.

Associations between sex, APOE genotype, and cognitive performance
In all the participants, cognitive performance in memory, language, attention, visuospatial, and executive abilities was normalized (z-scored) with standardized neuropsychological tests. Regarding the continuous characteristics of these cognitive performances, we investigated the effects of sex and APOE ε4 genotype in the participants with different cognitive states at the same time. As presented in Table 4, no significant 2-way interaction between sex and APOE genotype was found in any of the models. The main effects of sex showed that females had poorer language and visuospatial performance (beta = −0.363, −0.439; p = 0.005, 0.001) than males. No significant main effect of the APOE ε4 genotype was observed on any cognitive domain. In addition, age was associated with decreased memory and attention performance, and education was associated with increased cognitive performance in all cognitive domains. AV45 PET SUVR had a significantly negative correlation with cognitive performance except for visuospatial performance.

Discussion
In our Aβ positive cohort with different cognitive states, we demonstrate that sex and the APOE ε4 genotype display different effects on brain Aβ burden in different cognitive states. In the CN participants, a significant 2-way interaction between sex and APOE genotype is observed for AV45 PET SUVR. We find that females have significantly higher AV45 PET SUVR than males in APOE ε4 noncarriers, and APOE ε4 carriers have significantly higher AV45 PET SUVR than noncarriers in males except for the brain region of the medial temporal gyrus. In the CI participants, no significant 2-way interaction between sex and APOE genotype is observed. However, though no main effect of sex is observed on AV45 PET SUVR, APOE ε4 genotype has main effects on AV45 PET SUVR in the cortical ROIs of lateral parietal gyrus, frontal gyrus, and precuneus, in which APOE ε4 carriers are more likely to have higher AV45 PET SUVRs than noncarriers.
APOE ε4 genotype is the most common genetic risk factor for AD. In our study, sex and APOE genotype differences of global AV45 PET SUVR were not significant in the CI participants. This is consistent with the finding that the APOE ε4 allele plays a greater role in the initial seeding of Aβ deposition compared to subsequent plaque growth (Liu et al., 2017). However, in the CN participants, females had a significantly higher AV45 PET SUVR compared to males in APOE ε4 noncarriers but not in APOE ε4 carriers. Furthermore, male APOEε4 carriers but not female APOE ε4 carriers had a significantly higher AV45 PET SUVR compared to APOE ε4 noncarriers. These results would appear to be contradictory to the previous report that APOE ε4 may generate a greater effect of AD-related changes in women versus men (Neu et al., 2017). A reasonable explanation is that AD-related disease severity may be more associated with the abnormal amyloid-generated tau pathology, which was found to be more pronounced in female carriers of APOE ε4, even with similar amyloid deposition compared to males (Buckley et al., 2019;Duarte-Guterman et al., 2021;Hohman et al., 2018). Either way, this higher level of Aβ deposition in the early stage of AD may at least partly    contribute to the relatively more NFTs in the late stage of AD and lead to the higher prevalence and incidence of AD dementia in females (Filon et al., 2016;Jia et al., 2020). Thus, it might be essential to suggest distinct early detection and preventive strategies regarding sex and APOE ε4 genotype in response to the discrepancies of brain amyloid changes, including a possible sex and APOE ε4 genotype-specific cutoff score. To be noted, in the CN Aβ+ subjects with no APOE ε4 allele, the significantly higher level of Aβ burden in females compared to males may be due to other sex differences of biological underpinnings and relevant risk factors, such as sex-specific hormonal changes, socioeconomic status, and comorbidities of cardiovascular or cerebrovascular diseases (Ferretti et al., 2018;Nebel et al., 2018), which need to be fully elucidated in future studies. Besides the global cortical region, we also investigated the sex and APOE genotype differences of Aβ burden in 7 brain ROIs. As a result, similar findings were observed in the majority of brain regions. To be noted, in the CN participants, the APOE ε4 associated higher AV45 PET SUVR was not found in males within the medial temporal gyrus. This may be due to the APOE ε4-associated greater medial temporal atrophy in asymptomatic individuals (Taylor et al., 2014). Even so, females still had a higher AV45 PET SUVR than males in APOE ε4 noncarriers. Among the CI participants, though no sex difference of AV45 PET SUVR was found in all the examined brain regions either, APOE ε4 carriers were more likely to have higher AV45 PET SUVR than noncarriers within the lateral parietal gyrus, frontal gyrus, and precuneus. Given the relatively higher Aβ burden in CI participants, this may be due to the nature of the brain regions with predominant Aβ deposition (La Joie et al., 2012). The general linear models in our study showed that age was positively correlated with AV45 PET SUVR in the majority of brain regions except for the medial temporal gyrus in CN participants, which may be due to its relatively lower carrying capacity for aggregated amyloid protein, leading to a different spatiotemporal distribution from other brain regions (Whittington et al., 2018). In addition, age-associated increased AV45 PET SUVR was not found in CI participants. This is consistent with the finding that the rate of Aβ deposition slowed toward a plateau in the late stages of AD (Villemagne et al., 2013). It is well known that education level is a protective factor for the onset of AD. However, we found that education level was positively correlated with AV45 PET SUVR in both CN and CI participants. A reasonable assumption is that the protective effect of higher education would conceal the severity of the disease, leading to an advanced stage of Aβ deposition at the first medical contact. Given the possible relationships between brain Aβ burden and cognitive function, the AV45 PET SUVR in the global cortical region, together with age and education years, was used as a covariate to evaluate the effects of sex and APOE genotype on different cognitive performances. As a result, the global Aβ burden was negatively correlated with cognitive performance except for visuospatial ability. Furthermore, the association between Aβ burden and memory performance was the most robust. This is consistent with the previous study and indicated that impaired memory performance might be an early marker of brain Aβ deposition (Jansen et al., 2018;Pike et al., 2007). In our study, no significant 2-way interaction between sex and APOE genotype was found with regard to cognitive performance. In addition, the main effect of the APOE ε4 genotype was not observed on cognitive performance. This can be explained by the fact that APOE ε4 is more associated with the pathology of amyloidosis than tau aggregation and synapse loss, which are tightly linked to cognitive impairment (Kim et al., 2009). Previous studies found better memory performance (Duarte-Guterman et al., 2021;Jack et al., 2015) but poorer verbal and visuospatial abilities (Laws et al., 2016) in females compared to males. In our Aβ-positive participants, this poorer language and visuospatial performance in females compared to males was confirmed. However, the femaleassociated better memory performance was not observed. This may be attributed to the brain Aβ burden we adjusted, which showed a most robust negative association with memory performance. In the general linear models, age was negatively associated with memory and attention but not with other aspects of cognitive performance. This is consistent with the observation that besides memory impairment, attentional deficits also occur in the early stage of AD (Malhotra, 2019). As a protective factor for cognitive decline, we found that a higher education level was significantly associated with better performances in all cognitive domains.
Limitations should be noted in this study. First, it should be noted that the generalizability of our findings related to sex and APOE genotype needs to be validated in different populations, as ethnicity might influence the magnitude of the APOE ε4 associated risk of developing AD (Farrer et al., 1997). Second, the risk factors for brain amyloid deposition in female APOE ε4 noncarriers were not fully elucidated in this study and should be further investigated. Third, a small number of participants with an average young age were recruited in our study. Though the multiple comparisons of brain Aβ burden in participants aged ≥60 years was consistent with the result for all the participants, considering the heavier brain Aβ burden was found in early-onset Alzheimer's disease patients versus late-onset Alzheimer's disease patients in previous studies, further investigations should be carried out in an age-specific population.
In conclusion, the interactive effects of sex and APOE genotype on brain Aβ deposition were found in the CN population, with females having significantly higher Aβ burden than males among APOE ε4 noncarriers, and APOE ε4 carriers having significantly higher Aβ burden than APOE ε4 noncarriers among males. This suggests more attention should be paid to the sex and APOE genotype differences in amyloid deposition in the early stage of AD. In addition, though no significant effect of APOE genotype was observed on cognitive performance, females had poorer verbal and visuospatial performance than males in our Aβ burden-adjusted analyses. Given the negative correlations between Aβ burden and the majority of cognitive performance, an earlier detection of brain Aβ pathology in females should be done for optimal timing of early antiamyloid therapeutics.

Disclosure Statement
There were no actual or potential conflicts of interest in regard to this study.