Introduction
The strongest risk factor for Alzheimer’s disease (AD) is increasing age [
38]. The impact of sex differences on AD risk between women and men, however, has been variably interpreted [
3,
20,
28,
31,
37]. Evidence from multi-site studies suggests that women make up nearly two-thirds of patients with AD dementia in North America [
28]. The prevalence of AD dementia among women and men has also been increasing steadily in European ethnic groups, particularly for women after age 90 [
31,
37].
Postmortem studies have revealed that women tend to have greater AD pathology [
5,
13], but men more often present with atypical clinical syndromes (e.g., frontotemporal dementia, corticobasal syndrome) [
14,
24]. We developed a mathematical algorithm based upon density and distribution of neurofibrillary tangle counts to classify three neuropathologic subtypes of AD: hippocampal sparing AD, typical AD, and limbic predominant AD [
24]. Further investigation revealed that 63% of the hippocampal sparing AD were men, while 69% of limbic predominant AD were women. Moreover, the hippocampal sparing AD subtype had an average age at onset of cognitive symptoms at 63 years and shorter average disease duration of 8 years. In contrast, limbic predominant AD had an older average age at onset of 76 years old and longer disease duration of 10 years. We found that nearly 50% of hippocampal sparing AD had an atypical (non-amnestic) clinical syndrome, whereas more than 90% of limbic predominant AD had an AD dementia clinical presentation.
We hypothesized that atypical clinical presentations may lead to more frequent clinical misdiagnosis of AD in men, and thereby result in an apparent reduced frequency of AD in men. Thus, we proposed to investigate differences in the frequency of AD utilizing an autopsy-confirmed series from the Florida Autopsied Multi-Ethnic (FLAME) cohort. The objectives of this study were: (1) to examine differences between women and men, with respect to demographics, clinical course, genetics, and neuropathologic measures; and (2) to examine sex differences in neuroanatomic distributions of amyloid-β plaques and neurofibrillary tangles stratified by age.
Discussion
In the FLAME cohort, the overall frequency of autopsy-confirmed AD did not differ between women and men; however, men less often had an amnestic, multidomain clinical syndrome typical of AD dementia. As expected based upon societal gender-bias, women had lower educational levels. Women also had lower brain weights, older age at onset, longer disease duration, and older age at death. Severity of amyloid-β plaque pathology and neurofibrillary tangle pathology was greater in women compared to men. Interestingly, the frequency of autopsy-confirmed AD was highest among men in their 60 s, while women had an exponential increase from their 60 s into their late 90 s.
Our study investigating a large autopsy-confirmed AD cohort did not find an overrepresentation of women, but did find that men were less likely than women to present clinically with an AD dementia syndrome. A number of previous clinical studies, however, have found women are at a higher risk of AD than men [
3,
20]. Others have shown that higher risk is more specific to older women [
10,
15,
28,
31], supporting our observation of overrepresentation of women in their 80 s onward in an autopsy-confirmed AD cohort. Still others found no difference in incidence risk of AD in women, or once they controlled for age the risk was mitigated [
19,
28,
29]. Collectively, these studies support our finding of overall similar frequency of AD in women and men.
Our study is one of the first to highlight over representation of autopsy-confirmed AD in men in the 7th decade (i.e., 60–69). Men were more likely to present clinically with a non-amnestic or atypical clinical syndrome (e.g., corticobasal syndrome, frontotemporal dementia) and to have a younger age at onset. Diagnostic accuracy of autopsy-confirmed AD was lower in men than women in this study, especially in the 6th and 7th decades, when men more commonly came to autopsy. Men were also found to have greater neocortical neurofibrillary tangle pathology, relative to hippocampal tangles compared to women. In contrast, women were overrepresented in later decades and had greater hippocampal neurofibrillary tangle pathology relative to neocortex. The above mentioned findings, which are also mirrored in tau PET biomarker studies [
26], may be in concordance with the observation in epidemiological studies that mild cognitive impairment is more frequently diagnosed among men than in women [
27]. One interpretation may be that women who have more severe memory impairment may abruptly transition from normal to a more severely impaired cognitive state, whereas men with atypical clinical syndromes may have a slower transition from a mild cognitive impairment state to dementia.
We and others have shown that the rate of cognitive decline does not differ between women and men, but that women are more likely to have overt dementia [
5,
6]. Our data further demonstrate that neurofibrillary tangle counts in the hippocampus, across all ages, are more severe among women than men and that limbic predominant AD subtype is more common among women. Taken together with findings from neuroimaging studies demonstrating hippocampal atrophy occurs at a faster rate in women [
2], our study supports the observation that women are more likely than men to present with a typical amnestic dementia. We hypothesize that women in later decades have a high frequency of correct diagnosis as AD dementia because AD neuropathology is more frequent in women and they have greater hippocampal involvement relative to the neocortex, leading to an amnestic syndrome.
In this study we extend our current knowledge by examining quantitative neurofibrillary tangle counts measures in association cortices, primary cortices, and hippocampal subsectors. Our findings suggest that there is an interaction between age and sex, resulting in selective neuroanatomic susceptibility to neurofibrillary pathology in different decades of life. Importantly, the neocortex may be especially vulnerable to neurofibrillary pathology in younger onset individuals, especially men, and to have a more rapid progression. Conversely, older aged individuals were found to have significantly reduced cortical pathology [
8]. Women with autopsy-confirmed AD have been consistently found to be older at death [
4,
5,
13,
32], but our study additionally suggests that men may be younger at onset of cognitive symptoms and have a shorter disease duration.
Our primary goal was to investigate the hypothesis that epidemiological studies showing greater prevalence of AD dementia in women may be biased by the sex- and age-influenced distribution of Alzheimer type pathology, which results in a disproportionate number of men with atypical clinical presentations. Our results support this hypothesis and point to differences in selective vulnerability to neurofibrillary tangle pathology as a plausible biological factor underlying differences in frequency of clinical diagnoses. Other biological factors that may have a bearing on our findings include the impact of sex hormones on neuronal vulnerability [
30], as well as developmental differences in brain reserve [
23,
33,
35]. One of the most important considerations when investigating sex differences in an aging cohort is differential survival of men and women, relating to their cardiovascular risk profile [
10]. Thus, survival differences, resulting from differing cardiac risk profiles, could contribute to the observed lower frequency of AD in men in their later decades, as well as a greater frequency of men diagnosed with mild cognitive impairment.
This study has several limitations inherent to autopsy-based studies derived from specialty clinics or from self-referral. Although data were provided on clinical progression, the FLAME cohort is a cross-sectional and retrospective study. We did not observe overall sex differences in women and men, but there may be inherent bias among individuals who elected to participate in autopsy program [
21]. Interestingly, we found that the overall FLAME cohort was comprised of 51% men and 49% women, suggesting brain donations from specialty clinics or from self-referral may not be sex-biased. Another potential limitation is that patients or the caregivers of patients with atypical clinical syndromes may be more likely to seek autopsy confirmation. Thus, our clinical findings may not be generalizable to population or epidemiologic-based cohorts. It should also be noted that our reported findings on cognitive decline were derived from a much smaller cohort with three or more available MMSE scores used to calculate rate of change. Thioflavin-S is a fluorescent dye that binds to β-pleated sheets and thus readily binds to both amyloid-β plaques and neurofibrillary tangles. Although immunohistochemical methods were used to validate thioflavin-S findings, it should be noted that differences in sensitivity may occur when assessing presence of pathology. An unbiased stereology approach was not employed to quantify neurofibrillary tangle and senile plaque data. To offset this limitation, neuropathologic lesions were quantified as counts per microscopic field across a large series of autopsy-confirmed AD cases. We did not observe sex-based frequency differences within black Americans or Hispanic Americans, but further examination of these underserved cohorts should be performed to identify whether our findings remain consistent in other diverse cohorts of autopsy-confirmed AD. Last, information on cause of death is rarely available, thus limiting further investigation into survival bias.
From a historical perspective, sex differences in research studies have not been adequately investigated and in some circumstances women have even been excluded from participating in many studies. This bias against inclusion of women ultimately prompted a federal mandate in the United States requiring inclusion of women in research studies [
11]. Sex is often used as a covariate with balanced cohorts being investigated, but by statistically adjusting for sex, important biological clues underlying disease mechanisms may be neglected. Our data derived from autopsy-confirmed AD did not reveal a sex bias in terms of recruitment and consent for autopsy and they demonstrate that women and men have a similar frequency of AD. Sex-based differences in the age at onset, selective vulnerability of neocortical and limbic brain regions and the resulting clinical presentations may contribute to previous observations in clinical studies of AD being more common in women. These present observations should not take away from progress being made toward women’s health initiatives [
12], but instead should be used to emphasize the importance of considering neuropathologic and biological differences in young-onset AD (i.e., < 65 years) compared to AD in the oldest-old (i.e., > 85 years). The bearing of sex and age, as well as differential rates of progression of memory impairment, may have important implications for therapeutic trials, which may benefit from stratification of women and men, and/or by age at onset when assessing efficacy.
Acknowledgements
Mrs. Liesinger, Dr. Duara, Dr. Aziz, Mrs. Hinkle, Ms. DiLello, Ms. Johnson, Dr. Ross, Dr. Ertekin-Taner, and Dr. Dickson have nothing to disclose. Dr. Graff-Radford reports grants from Biogen, grants from Lilly, grants from Novartis, grants from Axovant, grants from Janssen during the conduct of the study. Dr. Murray reports grants from Florida Department of Health, Ed and Ethel Moore Alzheimer’s Disease Research Program (6AZ01, 8AZ06), grants from National Institute on Aging (R01 AG054449), grants from Gerstner Family Career Development Award during the conduct of the study; personal fees from Avid Radiopharmaceuticals, Inc., outside the submitted work. We thank the patients and their families for their generous brain donations to help further our knowledge of Alzheimer’s disease.