Introduction
The Hispanic minority group in the US is projected to reach 111 million by 2060 (U.S. Census Bureau, 2017). Hispanic elders have about a 50% greater risk for developing Alzheimer’s disease (AD) as compared to white non-Hispanics (WNHs), highlighting a need for identifying factors that contribute to and protect against the development of AD for this group. Studies examining neurodegeneration have shown less atrophy among Hispanics with AD than in WNHs (Burke et al., Reference Burke, Hu, Fava, Li, Rodriguez, Schuldiner, Burgess and Laird2019). Variables documented to be related to neurodegeneration in early and prodromal stages include age, sex, educational level, and genetic biomarkers (Kim et al., Reference Kim, Yeo, Park, Choi, Lee, Park, Ock, Eo, Kim and Cha2014; Vila-Castelar et al., Reference Vila-Castelar, Guzmán-Vélez, Pardilla-Delgado, Buckley, Bocanegra, Baena, Fox-Fuller, Tirado, Muñoz, Giraldo, Acosta-Baena, Rios-Romenets, Langbaum, Tariot, Lopera, Reiman, Quiroz and Kirkland Caldwell2020; Qiu et al., Reference Qiu, Bäckman, Winblad, Agüero-Torres and Fratiglioni2001), but much knowledge remains to be discovered for understanding specific factors that contribute to neurodegenerative disparities across ethnic groups.
AD has well-known genetic factors, including autosomal dominant inheritance, with early-onset disease in a small minority (<1%) of cases (De Oliveira et al., Reference De Oliveira, Chen, Smith and Bertolucci2017; Bird, Reference Bird, Adam, Ardinger, Pagon, Wallance, Bean, Stephens and Amemiya1993). Early onset familiar AD is known to be related to mutations in Amyloid Precursor Protein, Presenilin-1, and Presenilin-2 (Kim et al., Reference Kim, Yeo, Park, Choi, Lee, Park, Ock, Eo, Kim and Cha2014; Rogaeva et al., Reference Rogaeva, Tandon and Georger-Hyslop2001). Late-onset AD (LOAD) is known to have higher heritability (74% among twins) (De Oliveira et al., Reference De Oliveira, Bertolucci, Chen and Smith2014; Gatz et al., Reference Gatz, Pederson, Berg, Johansson, Johansson, Mortimer, Posner, Viitanen, Winblad and Ahlbom1997), leading to an increased interest about other genetic and non-genetic factors which may pose a risk for LOAD.
Indeed, the e4 allele of the apolipoprotein E (APOE) gene is recognized as a major risk factor for LOAD (Bilbul & Schipper, Reference Bilbul and Schipper2011; De Oliveira et al., Reference De Oliveira, Chen, Smith and Bertolucci2017; De Oliveira et al., Reference De Oliveira, Bertolucci, Chen and Smith2014; Michaelson, Reference Michaelson2014; Schipper, Reference Schipper2011). Importantly, any association between the e4 allele, brain morphology, metabolism, and amyloid load may have implications for developing preventive/treatment alternatives.
In addition to the genetic variability already highlighted, the influence of ApoE4 across age, sex, race, and nationality can also vary (Riedel et al., Reference Riedel, Thompson and Brinton2016; Liu et al., Reference Liu, Liu, Kanekiyo and Bu2013; Farrer et al., Reference Farrer, Cupples, Hainess, Hyman, KuKull, Mayeux, Myers, Pericak-Vance, Risch and Dujin1997). For instance, women carrying the ApoE4 may be at higher risk for AD in comparison to men (Riedel et al., Reference Riedel, Thompson and Brinton2016; De Oliveira et al., Reference De Oliveira, Bertolucci, Chen and Smith2014; Beydoun et al., Reference Beydoun, Boueiz, Abougergi, Kitner-Triolo, Beydoun, Resnick, O'Brien and Zonderman2012; Breitner et al., Reference Breitner, Wyse, Anthony, Welsh-Bohmer, Steffens, Norton, Tschanz, Plassman, Meyer, Skoog and Khachaturian1999). Women with two alleles increase their chances 10-fold, with the risk being fourfold in women with only one allele. On the other hand, men do not generally exhibit an increased risk with one allele, but the risk increases fourfold with two alleles (Farrer et al., Reference Farrer, Cupples, Hainess, Hyman, KuKull, Mayeux, Myers, Pericak-Vance, Risch and Dujin1997; Kim et al., Reference Kim, Castellano, Jiang, Basak, Parsadanian, Pham, Mason, Paul and Holtzman2009).
For nearly two decades, it was believed that women who carried copies of the Apoe4 gene were at greater risk for developing AD, with the current literature indicating that two-thirds of those diagnosed with AD are women (Snyder et al., Reference Snyder, Asthana, Bain, Brinton, Craft, Dubal, Espeland, Gatz, Mielke, Raber, Rapp, Yaffe and Carrillo2016). However, some experts maintain that sex differences in dementia research are largely explained by life expectancy (Hebert et al., Reference Hebert, Scherr, McCann, Beckett and Evans2001). Female e4 carriers have been noted to more rapidly deteriorate from a cognitive standpoint (Lehmann et al., Reference Lehmann, Refsum, Nurk, Warden, Tell, Vollset, Engedal, Nygaard and Smith2006) and to be at a greater risk of converting from mild cognitive impairment (MCI) to AD (Altmann et al., Reference Altmann, Tian, Henderson and Greicius2014). Altmann et al. (Reference Altmann, Tian, Henderson and Greicius2014) further found an ApoE4 sex interaction with biomarker levels stronger in females with MCI. The e4 allele increased total tau levels significantly more in women than in men, even after controlling for Abeta levels. Despite similar beta-amyloid levels among men and women MCI participants, ApoE4 increased the ratio of total tau to beta-amyloid significantly more in women (Altmann et al., Reference Altmann, Tian, Henderson and Greicius2014). These findings could be explained by amyloid changes occurring earlier in women as an effect from ApoE4 on tau biomarker, the e4 allele resulting in more tau-related pathology, or triggering a more accelerated tau pathology in women (Altmann et al., Reference Altmann, Tian, Henderson and Greicius2014). However, a recent metanalysis of the literature identified that males and females with ApoE3/E4 between the ages of 55-85 have equal risk of developing MCI or AD, but authors also suggested women may be at greater risk of developing MCI at younger ages (Neu et al., Reference Neu, Pa, Kukull, Beekly, Kuzma, Gangadharan, Want, Romero and Arneric2017).
The mixed literature on sex differences in dementia research may be explained by a combination of environmental, medical, and lifestyle variables. It is believed that estrogen is associated with the pathophysiology of AD in women. Specifically, the estrogen hypothesis claims that a normal amount of estrogen serves as a protective factor against AD. When there is a deficit, or dysfunction of estrogen, it is believed to exacerbate or precipitate AD (Rahman et al., Reference Rahman, Jackson, Hristov, Isaacson, Saif, Shetty, Etingin, Henchcliffe, Brinton and Mosconi2019). While estrogen is a hormone present in both sexes, it is primarily considered to be a female sex hormone, and estrogen deficit in women has been associated with neuroinflammation, synaptic decline, cognitive impairment, and risk of age-related disorders (Zárate et al., Reference Zárate, Stevnsner and Gredilla2017). Estrogen has also been shown to have neuroprotective properties in spinogenesis, metabolic regulation, DNA deterioration; among other neuroregulatory processes (Rahman et al., Reference Rahman, Jackson, Hristov, Isaacson, Saif, Shetty, Etingin, Henchcliffe, Brinton and Mosconi2019).
Despite the projected expansion of the Hispanic population, most research on AD biomarkers and sex differences continues to be conducted on WNH samples, and the literature focusing on ethnic and racial groups remains scarce. Results from Farrer et al. (Reference Farrer, Cupples, Hainess, Hyman, KuKull, Mayeux, Myers, Pericak-Vance, Risch and Dujin1997)’s meta-analytic study indicated that the attenuated effect of ApoE4 in Hispanics and African Americans (AA) merited further investigation. Duara et al. (Reference Duara, Barker, Lopez-Alberola and Loewenstein1996) studied a sample of Ashkenazi Jews, Hispanics, WNHs, and African Americans (AA) and found no significant variability among ethnic groups in terms of ApoE4 allele frequency in comparison to the reported population mean. Tang et al. (Reference Tang, Stern, Marder, Bell, Gurland, Lantigua, Andrews, Feng, Tycko and Mayeux1998) found that AA and Hispanics with E4 (+) status were as likely as WNHs to develop probable or possible AD, but those with E4 (−) status were two-to-four times more likely than WNHs to develop AD, potentially explained by nongenetic contributory factors (i.e., cardiovascular disease) to AD’s pathophysiology.
Another challenge for understanding the effect and risk associated with the e4 allele is that, broadly speaking, the Hispanic/Latino genetic diversity is understudied. Most of the existing research on ApoE4 focused on Hispanics is limited to individuals from Mexico, Dominican Republic, and of Puerto-Rican origin, with small sample sizes that limit generalizability of findings to diverse Latino groups, or fail to detect any AD risk associated with ApoE4 likely for being underpowered (Campos et al., Reference Campos, Edland and Peavy2014; O’Bryant et al., Reference O'Bryant, Johnson, Reisch, Edwards, Hall, Barber, Devous, Royall and Singh2013; Sevush et al., Reference Sevush, Peruyera, Crawford and Mullan2000; Tang et al., Reference Tang, Cross, Andrews, Jacobs, Small, Bell, Merchant, Lantigua, Costa, Stern and Mayeux2001; Tang et al., Reference Tang, Stern, Marder, Bell, Gurland, Lantigua, Andrews, Feng, Tycko and Mayeux1998). With increased recognition of existing variability related to presence of ApoE4 attributed to genetic diversity, González et al. (Reference González, Tarraf, Jian, Vásquez, Kaplan, Thyagarajan, Daviglus, Lamar, Gallo, Zeng and Fornage2018) explored the presence of e4 allele in a large and diverse Latino sample with well-defined ancestry background. They reported the frequency of ApoE4 was highest among Caribbean Latinos, with a lower frequency in Mainlander Latinos from Central and South America, and Mexicans (González et al., Reference González, Tarraf, Jian, Vásquez, Kaplan, Thyagarajan, Daviglus, Lamar, Gallo, Zeng and Fornage2018). The distribution of ApoE genotype in this study was consistent with continental ancestry patterns identified by Conomos et al. (Reference Conomos, Laurie, Stilp, Gogarten, McHugh, Nelson, Sofer, Fernández-Rhodes, Justice, Graff, Young, Seyerle, Avery, Taylor, Rotter, Talavera, Daviglus, Wassertheil-Smoller, Schneiderman and Laurie2016), with Mainlanders having lower e2 and e4 frequencies seen in Ameridian ancestry compared to Cuban Latinos with more European ancestry, whereas frequency of alleles 2 and 4 was highest among Dominicans, consistent with known higher frequencies of these alleles in those with African ancestry.
Moreover, there is ample literature exploring the relationship between ApoE4 and brain variables that contribute in significant ways to brain disease. Among these, cortical thickness and ApoE4 positivity have been extensively investigated, with studies indicating global reductions in cortical thickness in the areas of cornu ammonis, dentate gyrus, subiculum, entorhinal cortex (ERC), perirhinal cortex, parahippocampal gyrus (PHG), and fusiform gyrus for ApoE4 carriers with normal cognition (6.8% lower volume than in non-carriers) (Burgren et al., Reference Burgren, Zeineh, Ekstrom, Braskie, Thompson, Small and Bookheimer2008). The ERC and the subiculum area show greater thinness in carriers than non-carriers (14.8% and 12.6% respectively). Studies have also shown cortical thinness of the hippocampus in ApoE4 carriers relative to non-carriers among cognitively normal subjects (O'Dwyer et al., Reference O'Dwyer, Lamberton, Matura, Tanner, Scheibe, Miller, Rujescu, Prvulovic, Hampel and Stamatakis2012). Nonetheless, the limited cultural epidemiological studies exploring the relationship of ApoE4 and brain morphology with social and demographic variables among ethnic groups inspired the current investigation.
This study seeks to examine the effect of these factors on regional brain volume focused on medio-temporal structures including the hippocampi, entorhinal cortex, and parahippocampal gyri given their susceptibility to AD comparing Hispanics and WNHs. The current study has the following hypotheses: (1) ApoE4 will strongly predict medial temporal volume among WNHs with aMCI (as confirmed with neuropsychological focused diagnostic criteria) and (2) sociodemographic factors (i.e., age, education, and sex) will be stronger predictors of medial temporal volume among Hispanics.
Materials and Methods
Participants and recruitment
The current sample was recruited from subjects enrolled in the 1Florida Alzheimer’s Disease Research Center (1Florida ADRC), Clinical Core in Miami Beach, FL between 2015 and 2018 and were diagnosed with amnestic MCI (aMCI). One hundred and twenty-three total participants were stratified into two ethnic groups, Hispanic (n = 75) and White non-Hispanic (WNH, N = 48). The aMCI diagnosis included a mixture of early and late onset. The data used for this study were collected at the baseline visit of a longitudinal study of aging and Alzheimer’s disease (the 1Florida ADRC). Written informed consent was gathered from all subjects and/or their study partners. The Institutional Review Board at Mount Sinai Medical Center, Miami Beach approved this study. All human data included in this manuscript were obtained in compliance with the Helsinki Declaration.
Clinical and neuropsychological assessment
All participants underwent a standardized protocol that included a detailed clinical and neurological evaluation, a comprehensive neuropsychological assessment, and MRI scan to accurately identify normal cognition vs impairment. Laboratory exams included ApoE4 analysis as described below. Clinical and functional history was derived from a structured interview with the participant and study partner/caregiver focused on symptom identification (e.g., cognitive and emotional) and functional baseline using the Clinical Dementia Rating Scale (CDR).
Cognitive status was evaluated with a comprehensive battery that included the Mini-Mental Status Exam (MMSE), (i) episodic verbal memory function with the Logical Memory delayed recall from the Wechsler Memory Scale 3rd edition (WMS-III; Wechsler, Reference Wechsler1997) and the serial Hopkins Verbal Learning Test delayed trial (HVLT-II; Brandt, Reference Brandt1991), (ii) visuo-spatial and motor abilities with Block Design (Wechsler Adult Intelligence Scale, fourth edition, Reference Wechsler2008), (iii) language function with the Category fluency (C-Flu; Benton, Reference Benton1968) and the letter verbal fluency test (FAS; Controlled Oral Word Association Test; Benton et al., Reference Benton, Sivan, Hamsher, Varney and Spreen1994), and (iv) attention and executive functions with the Trail Making Test-Part B (TMT-B; Reitan, Reference Reitan1958) and the Stroop Color Word Association test (Stroop, Reference Stroop1935).
Ethnic classification was based on subject’s self-report. All Hispanic participants were either born in a Spanish-speaking country and/or spoke Spanish as their primary language. Hispanic subjects were evaluated at baseline and annually with a comprehensive neuropsychological battery of tests in their primary language. All these tests have been used for diagnostic determination in several studies with English and Spanish-speaking subjects (Loewenstein et al., Reference Loewenstein, Curiel, DeKosky, Bauer, Rosselli, Guinjoan, Adjouadi, Peñate, Barker, Goenaga, Golde, Greig-Custo, Hanson, Li, Lizarraga, Marsiske and Duara2018; Loewenstein et al., Reference Loewenstein, Curiel, Wright, Sun, Alperin, Crocco, Czaja, Raffo, Penate, Melo, Capp, Gamez and Duara2017). Tests had appropriate age, education, and cultural/language normative data (Arango-Lasprilla, Rivera, Aguayo, et al., Reference Arango-Lasprilla, Rivera, Aguayo, Rodríguez, Garza, Saracho, Rodríguez-Agudelo, Aliaga, Weiler, Luna, Longoni, Ocampo-Barba, Galarza-del-Angel, Panyavin, Guerra, Esenarro, García de la Cadena, Martínez, Perrin and Arango-Lasprilla2015; Arango-Lasprilla, Rivera, Garza, et al., Reference Arango-Lasprilla, Rivera, Garza, Saracho, Rodríguez, Rodríguez-Agudelo, Aguayo, Schebela, Luna, Longoni, Martínez, Doyle, Ocampo-Barba, Galarza-del-Angel, Aliaga, Bringas, Esenarro, García-Egan, Perrin and Arango-Lasprilla2015; Benson et al., Reference Benson, Felipe, Xiaodong and Sano2014; Ostrosky-Solís et al., Reference Ostrosky-Solís, Ardila and Rosselli1999). Proficient and bilingual Spanish/English psychometricians performed the testing.
Diagnostic classifications
aMCI participants were included in the analyses. The sample of 1Florida ADRC participants who were cognitively normal or diagnosed with dementia, and who completed both genetic testing and MRI was too small to be included in the analysis, as any result generated from these groups would have likely been underpowered for making any meaningful conclusions. Thus, CN and participants with dementia were excluded altogether from the study. Criteria for aMCI was based on Petersen’s guidelines (Reference Petersen2004), which are (a) memory complaint (s), preferably confirmed by an informant; (b) a CDR global scale score of .5; (c) MMSE score of 24+; (d) no impairment in social and/or occupational function; (e) no evidence of DSM 5 criteria for major neurocognitive disorder; (f) confirmation of memory impairment at 1.5 SD or greater, below expected levels, based on age and education adjusted normative data for each cultural/language group on the Hopkins Verbal Learning Test (HVLTR: Brandt, Reference Brandt1991) delayed recall or on the delayed paragraph recall of the Wechsler Memory Scale – 3rd edition (WMSIII: Wechsler, Reference Wechsler1997). Abnormal scores were considered 1.5 standard deviations below normal limits consistent with age, education, and language related norms, with a CDR global score equal to .5. Clinical diagnoses were determined by a neurologist, neuropsychologist, and clinician completing the psychiatric and functional measures.
The final exclusion criteria were: (1) CN, AD, or dementia secondary to other etiologies that met DSM-V criteria for major cognitive impairment; (2) sensory-perceptual loss or motor impairments, which interfered with an individual’s ability to complete a standardized battery of neuropsychological tests; (3) current or past history of major psychiatric disturbance with hospitalization, active psychosis, bipolar disorder, current major depressive episode, current or past history of alcohol or substance abuse within the six months preceding the first appointment; and (4) use of medications with anticholinergic properties (i.e., antipsychotics and sedatives).
APOE genotyping
Samples were genotyped for the APOE e2, e3, and e4 alleles using predesigned TaqMan SNP Genotyping Assays for SNPs rs7412 and rs429358 (Thermo Fisher Scientific, Massachusetts, USA) on the QuantStudio 7 Flex Real-Time PCR system (Applied Biosystems, California, USA) following the manufacturer’s protocol. ApoE genotypes were available for 242 subjects.
MRI
MRI scans were performed using a Siemens Skyra 3T MRI scanner at Mount Sinai Medical Center, Miami Beach. MRI scans were evaluated by visual inspection as well as with T2-weighted FLAIR (5 mm thick sequential axial slices), and a 3D T1-weighted volumetric magnetization-prepared rapid gradient-echo (MP-RAGE) sequence (which provides high tissue contrast and high spatial resolution with whole brain coverage) to quantitate brain atrophy.
Normalized values for hippocampi (HPC), entorhinal cortex (ERC), and parahippocampal gyri (PHG) volumes were obtained by dividing each regional volume by total intracranial volume to adjust for variation in head size. Brain regions were analyzed as combined, rather than as split hemisphere, consistent with a similar approach in previous studies (Burke et al., Reference Burke, Hu, Fava, Li, Rodriguez, Schuldiner, Burgess and Laird2019). This approach is also more appropriate for the current study’s aims.
Statistical methods
Multiple linear regression analyses were conducted with age, education, sex, and ApoE4 status as predictor variables and left and right combined MRI volumes of the hippocampus, parahippocampus, and entorhinal cortex as the dependent variables. The ApoE4 status variable was dummy coded as 1 when there was the presence of an e4 allele and 0 when there was no presence of an e4 allele. Analyses were conducted among two aMCI groups: Hispanic and WNH. Due to multiple comparisons, Bonferroni correction was applied to minimize Type 1 error and the p value was set at <.017 for each sample in which associations with three MRI measures were assessed. All statistical analyses were performed using IBM SPSS (Version 26; Armonk, NY: IBM Corp.).
Results
The demographic distribution of Hispanic and WNH aMCI groups is summarized in Table 1. There were significantly less years of education among the Hispanic aMCI group (M = 14.27, SD = 3.70) when compared to the WNH group (M = 16.15, SD = 2.94) at the .01 α level. The Hispanic group was significantly younger (M = 70.39, SD = 7.24) than the WNH group (M = 73.31, SD = 7.07). There were also significantly higher hippocampal volumes among the Hispanic group [F(1, 121) = 14.49, p < .001]. The groups did not differ in parahippocampal or ERC volume, sex, or ApoE4 status.
Table 1. Demographic variables among ethnic aMCI groups
Note. ApoE4: Apoliprotein e4 allele; WNH: White non-Hispanic; aMCI: Amnestic Mild Cognitive Impairment.
The results of multiple linear regressions are summarized in Table 2. Bonferroni-corrected results indicated that when controlling for ApoE4 status, education, and age, sex was a significant predictor of hippocampal volume among the Hispanic group (β = .000464, R 2 = .196, p < .01), as well as among the WNH group (β = .000455, R 2 = .195, p < .05). Figures 1 and 2 illustrated the regression equation. Education (β = .000028, R 2 = .168, p < .01) and sex (β = .000261, R 2 = .168, p < .01) were significant predictors of parahippocampal volume only among the Hispanic aMCI group when controlling for the effects of ApoE4 status and age (Figure 3). No significant predictors were identified for either group in the entorhinal cortex.
Figure 1. Linear regression model between predictor variable and hippocampal volume among the Hispanic aMCI group.
Figure 2. Linear regression between predictor variables and hippocampal volume among the WNH aMCI group.
Figure 3. Linear regression between predictor variables and parahippocampal volume among the Hispanic aMCI group.
Table 2. Results of linear regression between ApoE4 status and MTA among ethnic aMCI groups
Note. ApoE4: Apoliprotein e4 allele; WNH: White non-Hispanic; aMCI: Amnestic Mild Cognitive Impairment.
Table 3 illustrates results of one-way ANCOVAs, controlling for age and education, comparing hippocampal and parahippocampal volume between male and females within each ethnic group. Among the WNH aMCI group, females were found to have significantly larger hippocampal volumes at an alpha level of p < .05. Among Hispanic aMCI subjects, females had significant larger hippocampal (p < .001) and parahippocampal (p < .05) volume when compared to males. No sex differences in parahippocampal volume were noted among the WNH aMCI group.
Table 3. Results of ANCOVAs for hippocampal and parrahippocampal volume between male and female by ethnic group
Note. WNH: White non-Hispanic; aMCI: Amnestic Mild Cognitive Impairment.
Discussion
Our study sought out to examine the effect of ApoE4 allele and sociodemographic variables (i.e., age, education, and sex) on regional brain volume loss specific to medio-temporal structures comparing Hispanic and WNH groups with aMCI. The authors chose stratified analysis as our method to evaluate and control for confounding variables. This method is simpler than analyzing interaction effects, easier to understand, and therefore optimal for clarity and dissemination of our results to a wider audience. Our findings suggest that sex is a better predictor of hippocampal volume, with females showing larger volumes than males. Hispanic females showed higher volumes in both the hippocampus and parrahippocampus when compared to Hispanic males, and these differences were less evident between sex groups in the WNH group. Hypothesis two, which anticipated sociodemographic factors as stronger predictors of medio-temporal volume loss, was supported by our findings, with both education and sex significantly predicting parahippocampal volume among the Hispanic group only. These results highlight the varying predictive value from genetic and demographic variables on regional volume among different racial/ethnic groups. These findings also add to the mixed literature on sex differences in dementia research.
Our results are in line with a previous study that reported an overall protective factor among females with greater hippocampal volume (Burke et al., Reference Burke, Hu, Fava, Li, Rodriguez, Schuldiner, Burgess and Laird2019). Burke’s study, which examined biological sex differences in the development of MCI and AD, noted that the odds for disease progression in both diagnostic groups were equal for men and women. However, the progression was slower in women. Burke and colleagues also found that less hippocampal atrophy significantly decreased the odds of AD and MCI development for women. Kim et al. (Reference Kim, Yeo, Park, Choi, Lee, Park, Ock, Eo, Kim and Cha2014) identified e4 allele and depressive symptoms at baseline as risk factors for AD development in women. However, in their sample, e4 carriers displayed greater white matter hyperintensity (WMH) volume. Our study did not examine vascular factors such as WMH, so we are unable to determine any influence from cardiovascular disease that could exert an effect on some of our differences. This may be an important step in future research, especially when considering that an association between ApoE4 and carotid wall thickening has been established in e4 women carriers as a function of oral estrogen use, which may support a mechanism of interaction between APOE and estrogen among e4 carriers (Yaffe et al., Reference Yaffe, Haan, Byers, Tangen and Kuller2000).
Despite our small sample size, our study highlights that sex and gender differences may be important mediating variables in dementia research. Among women, there is inherent variability related to hormones, chromosomal, and gonadal differences. Estrogen receptors esterone (E1), estradiol (E2), and estriol (E3) are responsible for regulating molecular and genomic responses required for survival at the level of cells, genes, and organs. Dysregulation of estrogen signaling and transcriptional pathways have been shown to have severe neurological consequences. These alterations are displayed through changes in estrogen concentration, or through modifications of estrogen receptor activities (Rahman et al., Reference Rahman, Jackson, Hristov, Isaacson, Saif, Shetty, Etingin, Henchcliffe, Brinton and Mosconi2019). The neuroanatomical changes may be seen in the hippocampus, hypothalamus, locus coeruleus, posterior cingulate cortex, prefrontal cortex, raphe nucleus, and thalamus. Decreased responsiveness of estrogen receptor 1 (E1) may mediate cognitive decline and dementia risk (Rahman et al., Reference Rahman, Jackson, Hristov, Isaacson, Saif, Shetty, Etingin, Henchcliffe, Brinton and Mosconi2019). E1 splice variants are also localized throughout the brain as we age, highly concentrated in the hippocampus. These splice variants cause E1 to become nonfunctional, which occurs more often in women (Ishunina et al., Reference Ishunina, Fischer and Swaab2007). E1 polymorphism is also associated with having a higher risk for AD in women, especially when they are ApoE4 carriers (Ryan et al., Reference Ryan, Carrière, Carcaillon, Dartigues, Auriacombe, Rouaud, Berr, Ritchie, Scarabin and Ancelin2014; Zissimopoulos et al., Reference Zissimopoulos, Barthold, Brinton and Joyce2017).
One hypothesis that deserves consideration relates to accelerated disease progression/clinical transition happening later for women. Age-specific risk is lower in women for various cardiovascular conditions while prognosis worsens (Gao et al., Reference Gao, Chen, Sun and Deng2019), which could also be true for brain pathology. Estrogen and progesterone affect the reproductive system, central and peripheral nervous systems, growth and development (Edwards, Reference Edwards2005). Although production of these reduces as women age, many women remain in hormonal treatment for years, and a thorough understanding of mechanisms responsible for hormone-related neuroprotection are still largely undetermined.
The extent to which the e4 allele is a susceptible gene for LOAD in Hispanics continues to be a subject of scientific inquiry. One hypothesis is that, among Hispanics, other variables rather than presence alone of e4 allele may mediate medial temporal atrophy, which is consistent with results from the current investigation. Langbaum et al. (Reference Langbaum, Chen, Caselli, Lee, Reschke, Bandy, Alexander, Burns, Kaszniak, Reeder, Corneveaux, Allen, Pruzin, Huentelman, Fleisher and Reiman2010), for example, explored hypometabolism in AD-affected regions in e4 carrier Latinos with healthy cognition, with authors confirming that E4 (+) was in fact associated with hypometabolism but not with atrophy among this ethnic group. Since atrophy or cortical thinning occurs later that presynaptic or post-synaptic dysfunction, functional neuroimaging measures may be more sensitive to the early effects of E4 (+).
At the same time, ApoE4 is thought to be implicated in many pathological processes, ranging from neurotoxicity, increased amyloid pathology, to synaptic and mitochondrial dysfunction (Liu et al., Reference Liu, Liu, Kanekiyo and Bu2013), with available literature proposing a differential effect from e4 allele in either amyloid plaque burden (Duara et al., Reference Duara, Loewenstein, Lizarraga, Adjouadi, Barker, Greig-Custo, Rosselli, Penate, Shea, Behar, Ollarves, Robayo, Hanson, Marsiske, Burke, Ertekin-Taner, Vaillancourt, De Santi, Golde and DeKosky2019; Huynh et al., Reference Huynh, Davis, Ulrich and Holtzman2017) or other pathological processes such as tau pathology, neuroinflammation, vascular lesions, among others (Richey et al., Reference Huynh, Davis, Ulrich and Holtzman1995). ApoE4 effect on these pathological processes is believed to contribute to the hypometabolism observed in neuroimaging studies (Fouquet et al., Reference Fouquet, Besson, Gonneaud, La Joie and Chetelat2014), but this effect has also been reported as a direct contribution from amyloid burden. This implies that more studies are needed to fully understand the independent and combined effects of AD biomarkers on the brain and cognition.
Interestingly, Hispanics are 1.5 times more likely to develop AD than WNHs (Alzheimer’s Association, 2020), and the e4 allele is known to pose a major risk factor for sporadic LOAD (Chartier-Harlin et al., Reference Chartier-Harlin, Parfitt, Legrain, Perez-Tur, Brousseau, Evans, Berr, Vidal, Roques and Gourlet1994). Considering this, results from the current investigation may indicate that, contrary to WNHs, females may have a decreased risk of neurodegeneration, and factors like sex and education differentially contribute to AD pathophysiology among individuals of Hispanic origin. It is also possible that that presence of the e4 allele may indirectly, rather than directly, exert an effect on other mechanisms (i.e., vascular lesions and glucose metabolism) implicated in brain dysfunction among Hispanics. However, studies comparing AD biomarkers among Hispanics is slowly emerging.
The current study additionally detected education as predictive of medial temporal atrophy among Hispanic women. This is consistent with previous research focused on sex differences and the contribution of education on cognitive decline, with the latter having been identified as a moderator for cognitive deterioration, with higher levels of education posing a protective/cognitive reserve role against developing dementia across most ethnic groups (Mortensen & Høgh, Reference Mortensen and Høgh2001). Similarly, Mungas et al. (Reference Mungas, Gavett, Fletcher, Farias, DeCarli and Reed2018) examined closely education, brain degeneration, and cognitive decline in diverse older adults comparing Hispanics, AA, and Caucasian groups and found that education moderated the relationship between cognitive decline and brain atrophy. That is, a higher level of education was associated with slower cognitive decline in subjects with less cortical atrophy, and faster cognitive deterioration among those with greater atrophy. Mungas et al. (Reference Mungas, Gavett, Fletcher, Farias, DeCarli and Reed2018) noted these findings were more robust among Hispanics, which may favor the cognitive reserve hypothesis for this ethnic group. Other researchers have reported similar findings (Nyberg et al., Reference Nyberg, Magnussen, Lundquist, Baaré, Bartrés-Faz, Bertram, Boraxbekk, Brandmaier, Drevon, Ebmeier, Ghisletta, Henson, Junqué, Kievit, Kleemeyer, Knights, Kühn, Lindenberger, Penninx and Fjell2021) but have suggested that education does not offer further protection once a degenerative process has begun. Interestingly, biological sex was also found as predictive of medial temporal volume in Hispanic females, with this group having greater parahippocampal volume than Hispanic males and WNHs. Previous studies have explored age-associated volume decline comparing men and women (Driscoll et al., Reference Driscoll, Davatzikos, An, Wu, Shen, Kraut and Resnick2009; Gur et al., Reference Gur, Gunning-Dixon, Turetsky, Bilker and Gur2002), but these studies have not included diverse racial/ethnic cohorts for meaningful comparisons and for achieving a thorough understanding of biological/sex differences in the context of diversity and ethnicity, making this an imperative gap area in AD biomarkers research.
Our findings are subject to certain limitations. Our sample size for the WNH group was relatively small. This may be the reason why hypothesis 1, which anticipated that ApoE4 would strongly predict medial temporal volume among WNHs, was not supported by our findings. This is inconsistent with previous research that has in fact demonstrated a higher frequency of e4 allele associated with poorer cognitive performance and/or increased risk of conversion from normal cognition to brain disease among European-American cohorts (Beydoun et al., Reference Beydoun, Boueiz, Abougergi, Kitner-Triolo, Beydoun, Resnick, O'Brien and Zonderman2012; Yu et al., Reference Yu, Lutz, Wilson, Burns, Roses, Saunders, Yang, Gaiteri, De Jager, Barnes, Bennett and Toft2017; Borenstein et al., Reference Borenstein, Mortimer, Wu, Jureidini-Webb, Fallin, Small, Mullan and Crawford2006; Farrer et al., Reference Farrer, Cupples, Hainess, Hyman, KuKull, Mayeux, Myers, Pericak-Vance, Risch and Dujin1997; Tang et al., Reference Tang, Stern, Marder, Bell, Gurland, Lantigua, Andrews, Feng, Tycko and Mayeux1998). Another possible limitation to consider is that the between group comparisons revealed the Hispanic group were significantly younger and had less years of education than the WNH group. Both these factors may influence brain structure volume. It should therefore be considered that these sociodemographic variables may have biased or exerted an effect in our findings from the linear regression analyses. However, including age and education in the linear regression model allows for control when examining the predictive effects of other variables. Additionally, our aMCI population sample was clinic-based and had higher level of education when compared to other community-based Hispanic samples. Given the protective effects from educational attainment inferred from the literature, the education reality of our sample likely limits generalizability of findings and should be taken in consideration when interpreting results.
Moreover, the cross-sectional design used in this study only allowed us to establish morphological differences across ethnic groups after controlling for several variables, and although we would need a longitudinal design to look at disease progression and risk factors more robustly, future research will need to expand our understanding of demographic, cultural, and environmental factors, so we gain a more comprehensive contextual knowledge implicated in AD pathophysiology. The current study does not make any claims about therapeutic implications. It simply highlights the need for continued work with ethnically diverse populations considering sex differences in anticipation that this knowledge may help establish different treatment approaches for men and women. Future research should incorporate other neuropsychological variables and examine differences in delayed memory performance between Hispanic and WNH groups, and associations of delayed memory scores to medial temporal volume, as well as sex differences among these associations. Examining these relationships would shed light on our findings and provide further understanding of how these factors interact with each other.
Although not without limitations, this study can be considered a contribution to the AD literature given our focus on cross-ethnic differences relying on a diverse ethnic sample. One important reason for continued focus on culture and ethnic variables in AD research lies in the vast literature concerning the “Hispanic paradox,” which postulates a decreased mortality rate for Hispanic subgroups despite having a lower socioeconomic and health status than WNHs (Palloni & Arias, Reference Palloni and Arias2004). Despite increased longevity in this ethnic group, longitudinal evaluations of diverse patterns of progressive cognitive decline in the context of ApoE4 (+) among diverse clinical/ethnic groups are still lagging. Given prevalence of cardiovascular disease and white-matter burden among Hispanics (Mungas et al., Reference Mungas, Fletcher and de Carli2015; Williams et al., Reference Williams, An, Beason-Held, Huo, Ferrucci, Landman and Resnick2019; Brickman et al., Reference Brickman, Honig, Scarmeas, Tatarina, Sanders, Albert, Brandt, Blacker and Stern2008) future studies should aim at elucidating the relationship between cardiovascular risk factors and white matter degeneration among Hispanic e4 carriers. Finally, future studies should examine the independent and collective contribution of AD biomarkers such as amyloid accumulation vs atrophy as a function of e4 allele to brain pathology among ethnic groups.
Funding statement
This research was supported by the National Institute of Aging Grant numbers 5P50AG0477266021 Florida Alzheimer’s Disease Research Center (Todd Golde, PI) and 1P30AG066506-01 Florida Alzheimer’s Disease Research Center (Todd Golde, PI).
Competing interests
None.
