Sex Differences in Biofluid Biomarkers for Alzheimer’s Disease Open Access

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the most common cause of dementia. Sex is increasingly recognised as an important factor in AD and a key variable for precision medicine approaches [1]. The prevalence of AD is higher in women, and the lifetime risk of AD for women is approximately 20%, twice that of men [2]. This is partially explained by the relatively greater life expectancy in women, as higher age is a major risk factor for AD [2]. Studies on age-adjusted incidence have found conflicting results, and sex differences in AD risk appear to depend on age, birth cohort, and across different geographical regions [3]. Several studies in Europe and Asia suggest higher incidence of AD in women, while many studies in North and South America have not found sex differences in incidence [4, 5]. Nonetheless, biological sex can contribute to differences in the pathophysiological process, presentation of symptoms, and treatment responses [6]. In parallel with the increasing focus on sex differences and sex-specific risk factors for AD, recent years have seen considerable development in biofluid biomarkers for AD pathology, with the emergence of highly precise blood-based biomarkers and their incorporation in the diagnosis of AD [7].

The pathophysiological hallmarks of AD are extracellular amyloid-β (Aβ) plaques, intracellular neurofibrillary tangles composed of phosphorylated tau (p-tau) proteins, and neurodegeneration [8] see (Fig. 1). This is accompanied by changes in several molecular pathways, including inflammation and mitophagy [9, 10]. AD pathology can be identified by analysis of specific proteins in cerebrospinal fluid (CSF) and blood plasma (Fig. 1) years before onset of clinical symptoms [7]. Thus, recently revised criteria for diagnosis and staging of AD by the Alzheimer’s Association Workgroup (AAW-criteria) define AD by the presence of biomarkers for Aβ proteinopathy and p-tau [7] and propose that biomarkers of AD tau proteinopathy, neurodegeneration, and inflammation may inform disease staging, prognosis, or serve as indicators of treatment responses [7].

Biological sex may affect both the concentration and interpretation of biofluid biomarkers [6] (see Fig. 2). Certain markers may vary in concentration between males and females, which would necessitate sex-specific cut-offs. Sex might also influence interpretation of biomarkers due to sex-specific associations with clinical presentation, pathology, comorbidities, or progression. While sex differences in biofluid biomarkers have been reviewed previously [6], the last 5 years have been prolific for the AD biomarker field: novel and highly accurate blood-based biomarkers for amyloid and tau pathology have been developed and are already tested in primary care settings [11], and new assays specific to neurofibrillary tangles are emerging in both CSF and plasma [12‒14]. As such, this review expands previous work by considering the impact of biological sex on blood-based biomarkers and recently developed assays. In line with the AAW-criteria, focus will be on sex differences in fluid biomarkers for Aβ proteinopathy; phosphorylated and secreted AD tau; AD tau proteinopathy; neuronal injury; and inflammation (see Fig. 1).

This review will focus on the effects of biological sex, defined by the presence and expression of XX (female) and XY (male) sex chromosomes. However, it should be noted that few studies explicitly state how sex is defined in their research, and some still use the terms sex and gender – the social roles associated with being a man or a woman – interchangeably.

Biological sex can influence the concentration and interpretation of biofluid biomarkers directly, by affecting normal CSF and plasma protein concentrations, or indirectly, through interactions with AD pathology (see Fig. 2). For example, sex differences in brain anatomy may affect biofluid biomarkers. Males have about 10% larger brain volume than females, but whether there are further sex differences across brain regions is debated [15]. However, due to their relatively larger brains, males have greater CSF volumes and larger ventricles. There are also apparent sex differences in the relative percentages of the brain composed of white and grey matter, with the former being higher in males and the latter being higher in females [16]. Sex differences in brain structure and function may contribute to differences in biofluid results for neurodegeneration biomarkers such as CSF neurofilament light chain (NfL), a marker of axonal degeneration. Moreover, these differences may contribute to sex differences in risk for cerebrovascular disease that may interact with or contribute to AD pathology [17].

There are several genetic risk factors for AD, with the apolipoprotein E (APOE) ε4 allele being the most common. Notably, APOE ε4 appears to interact with sex and increase AD risk more in heterozygous female carriers compared to male carriers, with no sex differences in homozygous carriers [18]. Moreover, female APOE ε4 carriers have higher levels of the AD biomarker CSF p-tau compared to male carriers and female noncarriers [19, 20]. The effect of APOE ε4 on sex differences in AD biomarkers appears to be stage dependent, with stronger sex differences in CSF p-tau at early stages for ε4 carriers and at later stages for noncarriers [21]. The APOE ε4 allele influences the relationship between biofluid biomarkers of Aβ and tau pathology, particularly in female carriers with stronger associations between CSF and blood Aβ biomarkers and p-tau biomarkers in carriers [22, 23].

Biological sex appears to interact with and affect AD pathology, in turn affecting the concentration and interpretation of associated biomarkers. Indeed, higher levels of tau deposition are apparent in female brains [24], even when accounting for Aβ plaque loads [25]. Moreover, Aβ and tau pathology are more closely related to brain atrophy and cognitive decline in females [26, 27]. As such, matched levels of Aβ and tau biomarkers might have different implications for males and females in terms of disease progression and clinical presentation, and sex-specific guidelines may be useful for clinical trials and treatment with anti-amyloid therapies [28].

The field of biofluid biomarkers for AD is highly active and in constant development. The discovery of accurate blood-based biomarkers of AD and their incorporation into diagnostic frameworks and clinical practice are poised to revolutionise AD research [7, 11]. Unfortunately, while studies of CSF and blood-based biomarkers typically include both male and female participants, sex differences are rarely examined directly, despite the potential impact of sex on fluid biomarker levels [5]. Rather, it is a common practice to adjust for sex in multivariate analysis, thus eliminating the impact of any sex difference on the results. Moreover, interactions between different AD biomarkers and biological or demographic variables such as sex, race, or socioeconomic status are rarely considered, despite their potential impact on disease stage, prognosis, or treatment responses. The sex differences in biomarkers considered herein are summarised in Table 1, highlighting both sex differences in absolute concentrations (e.g., consistently higher levels in men than in women) or sex-specific interactions with relevant variables such as cognitive decline (e.g., consistently worse memory performance for women with high biomarker levels compared to men).

Low levels of Aβ₄₂ in the CSF are indicative of the presence of amyloid plaques in the brain. CSF Aβ₄₂, or the CSF Aβ₄₂/₄₀ ratio, is used clinically to diagnose the presence of Aβ proteinopathy [30]. In CSF, there is about a 50%-fold change in the Aβ_42/40 ratio in the presence of Aβ pathology, whereas in plasma the fold change is limited to 10–15% [7, 31‒33]. This limited fold change makes the Aβ_42/40 ratio in plasma a less robust indicator of Aβ proteinopathy; other plasma assays discussed below are preferred for the diagnosis of AD pathology [7].

Several studies have found no sex differences in the concentrations of CSF Aβ₄₂ or the Aβ_42/40 ratio in sporadic AD [23, 34‒37] or autosomal dominant AD (ADAD) [38]. However, CSF Aβ₄₂ appears to be differentially associated with hippocampal atrophy, memory performance, and CSF p-tau concentrations in males and females [23, 27]. One study found that while baseline levels of CSF Aβ_42/40 did not differ between the sexes, age and sex interacted on the association between CSF Aβ_42/40 and disease progression; lower CSF Aβ_42/40 was associated with increased risk of disease progression in males, particularly at younger ages, but not in females, irrespective of age [39]. As such, CSF Aβ₄₂ could have different prognostic performances, and in this context of use, sex-specific cut points or even alternative prognostic biomarkers might be required.

In plasma, significant sex differences have not been found for SIMOA or LC-MS plasma measurements of Aβ₄₂ [40, 41]. One study found no differences in the Aβ42/40 ratio [41], whereas another found lower Aβ_42/40 ratios in females compared to males [42]. Notably, the sex difference in plasma Aβ_42/40 levels did not extend to patients with mild cognitive impairment (MCI) or dementia, suggesting that discrepancies between studies might be due to the disease stages studied [42]. Like findings in CSF, sex appears to influence the association between plasma Aβ₄₂ and prognosis. In a longitudinal community-based study, higher plasma Aβ₄₂ levels were associated with reduced future risk of AD in females, but not in males; conversely, lower plasma Aβ₄₂ levels were associated with yearly memory decline exclusively in females [43].

High levels of phosphorylated (mid-region) tau 181 in the CSF are used clinically as a biomarker of AD pathology [30]. Recently, different phosphorylated mid-region tau fragments (p-tau181, p-tau217, and p-tau231) have been identified as promising plasma biomarkers of AD pathology; especially, p-tau217 is emerging as an accurate and precise biomarker [44, 45]. Notably, these p-tau fragments become abnormal around the same time as amyloid PET and before the emergence of neurofibrillary tau tangles on tau PET [46, 47]. Secretion of tau fragments phosphorylated at the residues 181, 217, and 231 to plasma could thus represent a reaction to Aβ pathology, linking Aβ proteinopathy to the onset of tau proteinopathy in AD [48]. In the AAW-criteria, these p-tau biomarkers are noted T1 – biofluid analytes of soluble tau fragments that may reflect a reaction to amyloid plaques or soluble Aβ species in plaque penumbra.

Cross-sectional CSF studies of p-tau concentrations generally do not find sex differences [23, 34, 35, 37, 38]. However, females with low CSF Aβ may be more susceptible to increased p-tau concentrations [23, 27]. In line with this, female APOE ε4 carriers with CSF Aβ pathology appear to have higher levels of CSF p-tau181 compared to male carriers [20]. Nonetheless, another study found stronger associations between CSF p-tau181 and progression to MCI in males compared to females, but that this sex difference diminished with age [39].

Analysis of sex difference in plasma p-tau biomarkers has resulted in conflicting findings. Several studies have found no sex differences in plasma p-tau181 in cognitively unimpaired (CU) or MCI participants [40, 49‒51]; in extension, one multiethnic community study and one study of different neurodegenerative diseases also found no sex differences [41, 52]. However, two studies have found lower plasma p-tau181 levels in females compared to males, irrespective of cognitive status or diagnosis [53, 54]. Moreover, while one study found no interaction between sex and amyloid load on plasma p-tau181 levels [40], another found that despite similar plasma p-tau181 levels in males and females, females with higher p-tau181 at baseline had greater Aβ deposition, more entorhinal cortex tau deposition, higher CSF p-tau181, and faster cognitive decline in comparison with males [51]. In dementia-free participants with Aβ pathology, females with higher baseline plasma p-tau181 levels had higher rates of incident dementia relative to males [51].

For plasma p-tau217, several studies have found no sex differences in CU individuals [50, 55], a multiethnic community study [41], patients with different neurodegenerative diseases [52], or PSEN1 carriers [56]. However, one large population-based study found that plasma p-tau217 levels were significantly increased in males compared to females in CU participants, while there were no sex differences in the MCI and dementia groups [49]. In individuals with ADAD, sex and p-tau217 interacted on global cognitive and memory performance among CU mutation carriers, with female carriers performing better than males [56]; this effect was not present among CU noncarriers, suggesting a sex-specific cognitive resilience in female carriers to early accumulation of plasma p-tau217. One small study on 85 CU and 78 MCI participants suggests that a sex-specific association between cognitive performance and p-tau217 may also be present in sporadic AD. This study found that baseline plasma p-tau217 above the threshold indicating early levels of amyloid aggregation predicted verbal memory and medial temporal lobe atrophy in CU females, while CU males exhibited similar rates of decline and atrophy independent of plasma p-tau217 levels [55]. In line with findings from ADAD, these sex-specific interactions were specific to early stages and not present at the MCI stage [55]. Few studies have examined associations between biological sex and p-tau231, but one study examining several p-tau assays found no sex differences in p-tau231 in their cohort of CU and MCI participants over 70 years [50].

While the phosphorylated mid-region fragments at residues 181, 217, and 231 become abnormal around the same time as Aβ plaques begin to form, and correlate both with Aβ proteinopathy and tau proteinopathy, novel assays of different tau fragment analytes such as p-tau205, N-terminal-associated tau (NTA-tau) fragments, or microtubule-binding region tau (MTBR-tau243) become abnormal later and correspond better to tau PET [13‒15]. These, together with tau PET, are categorised as T2 markers – markers of AD tau proteinopathy, in the AA framework [7]. These novel biomarkers hold great promise for the study of AD tau proteinopathy across disease stages and in response to treatment interventions, but unfortunately, no study has yet examined sex differences in NTA-tau, p-tau205, or MTBR-tau243, nor if there are sex-specific interactions between these biomarkers and other markers of AD pathology, AD clinical symptoms, prognosis, or other relevant factors. Given accumulating evidence suggesting sex differences in tau proteinopathy in AD evident on tau PET [26, 28], this is a fruitful avenue for further research.

In the 2018 NIA-AA research framework for AD, CSF levels of total tau were considered biomarkers of neurodegeneration [11]. However, in the most recent AA framework, total tau is not included as a fluid biomarker of neurodegeneration [57]. Increased levels of plasma or CSF total tau are complex to interpret: on the one hand, CSF and plasma total tau increase early in AD and correlate closely with fluid p-tau, suggesting that total tau could be a biomarker of tau proteinopathy. On the other hand, total tau is not specific to AD and increases in response to, for example, head trauma, cerebral infarction, and proteinopathies such as Creutzfeldt-Jacob disease, suggesting that it is rather a marker of neurodegeneration. As such, sex differences in CSF or total tau could be indicative of sex differences in tau proteinopathy, neurodegeneration, or potentially other comorbidities that influence total tau levels. Highlighting this challenge is a host of contrasting findings on sex differences in total tau biomarkers, with some studies finding lower CSF total tau in males [27, 37] and others in females [38]. Similarly, while some studies have reported higher plasma total tau levels in females [42, 58, 59], other studies have found no sex differences [60, 61].

Neurodegeneration is a key step in the progression of AD pathology but is also central in other forms of dementia such as Lewy body disorder or frontotemporal dementia. As such, abnormalities in biomarkers of neuronal injury – the N category – are not specific to AD and occur in several other conditions such as non-AD neurodegenerative diseases, after traumatic injury and after ischemic injury among others [7]. NfL is a marker of axonal injury and a reliable marker of neurodegeneration that may be measured in plasma and CSF. Plasma and CSF levels of NfL are strongly correlated with each other and similarly associate with markers of brain atrophy such as cortical thinning or white matter atrophy [62].

Several studies have reported higher levels of CSF NfL in males with AD and in other neurodegenerative diseases [63‒66]. This difference could be due to greater vascular pathology in males or a larger proportion of brain white matter, but the exact cause is not known.

Notably, despite the strong correlation between plasma and CSF NfL levels, the sex differences in CSF NfL levels do not appear to translate to findings in plasma, with several studies reporting no baseline sex differences in CU, MCI, or dementia patients in plasma NfL levels [41, 42, 55, 58, 67]. The cause of discordance between CSF and plasma NfL levels is not known. This underscores that beyond examining correspondence between plasma, CSF, and neuroimaging biomarkers, examining sex differences in these modalities is important to understand their utility in picking up sex-specific disease trajectories and the need for sex-specific guidelines.

Few studies have examined sex-specific associations between plasma NfL and AD-related outcomes such as brain atrophy or clinical symptoms. One study found that higher NfL was associated with more medial temporal lobe atrophy over time exclusively in females but did not find evidence of sex-specific effects on cognitive trajectories [55]. This highlights potentially different prognostic utilities of the plasma NfL as a marker of neurodegeneration in males and females. In ADAD, female PSEN1 carriers had lower NfL levels at younger ages but show greater increases with age compared to male carriers [56].

Like markers of neuronal injury, biomarkers of neuroinflammation represent markers of nonspecific processes involved in AD pathology. A growing body of evidence shows that immune and inflammatory mechanisms are important in AD, with both reactive astrocytes and active microglia being linked to amyloid and tau pathology [9]. Biomarkers of inflammation might reflect either astrocyte or microglia activity. Biomarkers of a vast range of inflammatory pathways have been intensively studied in CSF, serum, and plasma, with apparent sex differences in, for example, systemic inflammation [68], the kynurenine pathway [69], matrix metalloproteinases [70], and microglial activity [71]. However, there is still no consensus on reliable biomarkers of AD-associated inflammation. One marker that has been incorporated in the AAW-criteria under the I – inflammation category – is glial fibrillary acid protein (GFAP), a marker of astrocytic reactivity.

GFAP can be measured in plasma or CSF, but plasma levels appear to be a more robust indicator of Aβ pathology [72]. While altered GFAP levels are not specific to AD, they are associated with early amyloid PET, higher risk of incident dementia, and faster rates of cognitive decline [47, 73, 74]. Some studies report no significant associations between sex and plasma GFAP levels [66, 73], while others have found elevated plasma GFAP in CU females compared to males [55, 72] and female-specific associations with future medial temporal lobe atrophy [55].

While some AD biomarkers display consistent differences in concentration between males and females, others do not appear to differ by sex (Table 1). Remarkably, even biomarkers that show no baseline difference display sex-specific associations with brain atrophy, cognitive performance, and/or disease progression (Table 1). For biomarkers with consistent sex differences in concentration, such as CSF NfL levels, sex-specific cut-offs may be appropriate. For the biomarkers showing sex-specific associations with relevant clinical variables such as disease progression, more research is required to develop sex-specific guidelines for different contexts of use, such as clinical trial enrolment or disease staging. This has key implications for future research.

Firstly, biomarker data should be reported separately for males and females, and group differences should be assessed if the sample size is adequate. This is especially important for emerging biomarkers such as NTA-tau or MTBR-tau243, where little is known about sex differences in concentration. Centrally, beyond examining group differences in biomarkers, sex-specific associations with variables of interest should be explored in well-powered studies. Furthermore, research should consider the upstream factors influencing these sex differences. From a precision medicine perspective, research should reach beyond enrolling equal numbers of males and females towards considering sex-specific biology that may interact with disease pathology [1]. Factors such as menopausal status; hormonal contraception or hormone treatment; reproductive events; genetics and sex-genotype interactions; and sex differences in the pattern of comorbidities [6, 20, 75] may be relevant for disease aetiology, prognosis, and treatment responses. For example, in cognitively normal populations, perimenopausal and postmenopausal women exhibit increased amyloid deposition and reduced brain volume in AD-related brain regions compared to premenopausal females and age-matched males; this difference is more pronounced in APOE ε4 carriers [76]. This indicates a progressive increase in AD risk as females undergo menopausal changes, implying that endocrine ageing may exacerbate the effects of chronologic ageing in the female brain [76]. Females who experience early menopause (naturally or iatrogenic) represent the highest risk cluster for late-life cognitive decline and dementia within the menopause population [1, 77]. As a result, studies examining sex differences at the preclinical stage or in ADAD may find divergent results if their proportion of pre-, peri-, and postmenopausal females differ notably.

Beyond examining the effects of biological sex, an intersectional approach may be required to fully determine the impact of sex and gender on AD. The concepts of sex and gender interact to influence health and disease, and their effects may also differ depending on social and demographic variables such as geography, culture, race and ethnicity, age, or socioeconomic status [78‒80]. Disentangling these concepts in research is a significant challenge but may enrich our understanding of AD and other diseases. Indeed, attempts to create gender metrics based on data such as education levels and self-reported depression have been linked to cardiovascular health, with higher “femininity scores” being associated with an angina diagnosis predating a heart attack in men in the UK biobank [81]. It should be noted that the current review is not a systematic review and, therefore, may exclude relevant articles. As such, future systematic reviews and meta-analysis on sex differences in biofluid biomarkers should be conducted. While sample size and power are central challenges when studying the effects of biological sex differences and interacting identities such as gender and race, more nuanced reporting of these categories could lead to future well-powered meta-analyses.

Biological sex influences the interpretation of biofluid biomarkers, even for biomarkers where baseline sex differences in absolute levels are not consistently reported. Potential sex-specific guidelines may be required depending on context of use; this is becoming increasingly relevant with the growing emphasis on biomarkers in the diagnosis and staging of AD. Intensive work is ongoing to validate and harmonise blood-based biomarkers across different assays as diagnostic tools in specialist memory clinics and primary care. The effects of sex should be considered in this work. The biomarker field should move beyond controlling for the effects of sex to viewing it as a variable of interest, and sex differences and sex-specific interactions should be explored. Ultimately, future research should aim to identify the upstream mechanisms driving interactions between sex biology and disease pathology.

The author has no conflicts of interest to declare.

This work was funded by the Norwegian Health Association (Grant No. 25633). The funder had no role in the manuscript conception, planning, writing, or decision to publish.

M.A. conceived and wrote the manuscript.