Sex-dependent role of inflammatory biomarkers in Alzheimer's disease-related cognitive performance

Abstract

Background

Inflammatory factors are widely recognized as contributors to Alzheimer's disease (AD), along with pathological changes in amyloid-β (Aβ) and tau proteins.

Objective

We aimed to determine whether sex influences the inflammation-related cognitive performance.

Methods

This cross-sectional study included 317 patients (260 cognitively impaired and 57 cognitively normal) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) between September 7, 2005, and August 7, 2024. Interaction analysis was used to examine sex-specific effects of interleukins, while logistic regression examined associations between interleukins and clinically relevant worsening on non-memory Mini-Mental State Examination (MMSE) items (BCNMMMS). Linear regression assessed associations with hyperphosphorylated tau proteins (P-tau181). Finally, mediation analysis was conducted to investigate the role of sex hormone-binding globulin (SHBG) in interleukins, BCNMMMS, and P-tau181.

Results

In BCNMMMS, sex interactions were observed for IL-1β (p = 0.048), IL-21 (p = 0.020), and complement component 3 (C3) (p = 0.041). After adjustment for confounding factors, IL-18 was associated with BCNMMMS in males (OR = 0.520 [0.274, 0.983]), while C3 was associated with BCNMMMS in females (OR = 0.452 [0.215, 0.950]). IL-21 and C3 levels were significantly associated with P-tau181 in females (IL-21: p = 0.002, C3: p = 0.047) but not in males. Besides, SHBG mediated the effect of IL-21 on cognitive performance in females (proportion of mediation = 19.72%; p = 0.032), suggesting a role for SHBG in sex-dependent inflammatory pathways related to cognitive performance.

Conclusions

Peripheral inflammatory biomarkers, particularly IL-21, showed sex-specific links to plasma P-tau181 and non-memory impairment in ADNI, supporting sex-stratified analyses.

Keywords

Alzheimer's disease Alzheimer's Disease Neuroimaging Initiative cognitive performance inflammatory factors sex

Introduction

As the most common cause of dementia, Alzheimer's disease (AD) has become an increasingly serious health problem and has a significant influence on individuals and society.¹ Experts believe that AD dementia, develops because of multiple factors rather than a single cause.² Investigating the diverse factors contributing to AD and their impact on cognitive function is therefore of great scientific and clinical importance. A deeper understanding of these relationships is crucial for improving prevention, facilitating early diagnosis, and informing effective therapeutic interventions to mitigate the disease's widespread impact.

Although inflammation may not typically represent an initiating factor in neurodegenerative disease, there is emerging evidence in animal models that sustained inflammatory responses involving microglia and astrocytes contribute to disease progression.³ At present, a chronic inflammatory status appears to be responsible for the activation of glial cells, triggering a more persistent inflammatory status leading to neuronal damage and neurodegeneration.⁴ In addition to amyloid-β (Aβ) pathology, various studies have implicated neuroinflammation as an important factor in AD pathophysiology.⁵ It is now understood that during AD, the abnormal accumulation of soluble amyloid oligomers triggers excessive release of proinflammatory factors, such as IL-1, IL-6, cytokines, and other acute-phase reactants, overwhelming regulatory components, including IL-4, IL-10, receptor antagonists, inflammatory factor inhibitors and others, ultimately leading to neuronal and synaptic injury, loss and cognitive decline.⁶ Understanding how different inflammatory factors affect cognitive performance is essential for the development of therapeutic strategies and interventions for AD. In recent years, studies targeting inflammatory factors have identified many potential biomarkers to help identify early signs of cognitive decline and to guide interventions. In the future, the inhibition of inflammatory factors could be considered a potential therapeutic strategy to reduce the risk of neurodegenerative diseases.⁷

The higher prevalence of dementia among women than men has been reported in most epidemiology studies.⁸ Sex and gender differences should be one of priorities in the development of AD therapeutics from preclinical to clinical studies.⁹ Beyond statistics that expose this greater global burden of AD in females, empirical data about sex differences in AD pathogenesis are warranted to guide clinical and scientific approaches to AD for the benefit of both women and men.¹⁰ Further elucidation of the role of sex in disease-associated processes is required to design effective treatments.¹¹ Previous studies have shown that cognitive performance differs between men and women, and that such differences may also extend to sex-specific patterns of cognitive change. In clinically normal older adults, women generally outperform men in verbal memory, while men exhibit superior visuospatial abilities across the lifespan, and women often demonstrate greater resilience to age-related cognitive decline.^12,13 However, this pattern shifts across the AD spectrum. Among individuals with MCI due to AD, women have been shown to experience significantly greater cognitive worsening than men.¹⁴ Furthermore, evidence suggests that the convergence of female sex, APOE ε4 status, and high amyloid burden results in a more precipitous rate of cognitive decline compared to their male counterparts.¹⁵

Inflammatory factors may affect cognitive performance differently across the sexes. Many of these inflammatory medical conditions occur at different rates among men and women.¹⁶ Accumulating evidence demonstrates that many inflammatory conditions exhibit sex-specific prevalence rates. For instance, women are significantly more susceptible to both peripheral inflammatory diseases, such as rheumatoid arthritis, and central neuroinflammatory disorders, like multiple sclerosis.^17,18 These disparities are largely attributed to the heightened immune reactivity observed in females, characterized by distinct innate and adaptive immunological responses to both foreign and self-antigens.¹⁹ Importantly, this heightened female immunoreactivity may lower the threshold for inflammation-induced neurotoxicity, potentially accelerating the transition from systemic immune activation to cognitive impairment. Systemic inflammation has been identified as a critical early driver of cognitive decline in the decades preceding older adulthood.²⁰ Furthermore, the strength and nature of neuroinflammatory responses may differ significantly between sexes. Recent postmortem evidence across the AD spectrum indicates that women are more susceptible than men to AD/tau-related neuroinflammation, highlighting distinct sex-dependent patterns in neuroinflammatory signatures.²¹

Sex hormone-binding globulin (SHBG) is a circulating glycoprotein with established clinical relevance across multiple physiological contexts.²² Emerging evidence suggests that SHBG may also reflect inflammatory status. Serum SHBG concentrations have been shown to be inversely associated with high-sensitivity C-reactive protein (hs-CRP) in men, premenopausal women, and postmenopausal women independently.²³ And SHBG could be a prodromal biomarker to predict disease progression in AD.²⁴

Although existing studies have suggested a potential association between SHBG, inflammatory responses, and cognitive function, there is currently a lack of systematic investigation into whether SHBG mediates sex differences in inflammation-related cognitive decline. Most previous studies have been confined to single-sex cohorts or have only explored the direct relationship between SHBG and cognitive function, without examining its potential mediating effect.^25,26 Therefore, this study aimed to fill this gap by testing the hypothesis that SHBG may mediate the relationship between inflammation levels and cognitive decline, and that this mediating effect may differ by sex. Such exploration may help elucidate the biological mechanisms underlying inflammation-driven cognitive impairment and provide a scientific basis for sex-specific intervention strategies.

Methods

Participants

We recruited 317 cases of participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Cognitively normal participants and patients diagnosed with mild cognitive impairment (MCI) or AD were included in our cohort. Individuals with incomplete data for the Mini-Mental State Examination (MMSE), plasma P-tau181, or inflammatory markers, as well as those aged below 60 years, were excluded. More details about participant selection are available at ADNI (https://adni.loni.usc.edu/). Detailed methodologies for clinical and cognitive assessments have been described previously.²⁷ CSF biomarker measurements and protocols are detailed in Shaw et al. (2009).²⁸ Although both plasma P-tau181 and P-tau217 were quantified in a subset of participants, P-tau181 was selected for the present analyses. Preliminary analyses did not identify stable or consistent sex-dependent associations for P-tau217 comparable to those observed for P-tau181. Therefore, P-tau217 was not included in subsequent analyses, and all reported results are thus based on plasma P-tau181. Specifically, we used Alzheimer's Disease Sequencing Project (ADSP)-harmonized cognitive composite scores, available for ADNI participants, for exploratory domain-specific analyses, including harmonized composite visuospatial (VSP), memory (MEM), language (LAN), and executive function (EXE) domains. These composite scores were standardized summary measures derived from multiple neuropsychological tests within each cognitive domain and were developed using ADSP harmonization procedures to improve robustness and cross-study comparability. All ADSP-harmonized cognitive composite scores were standardized with higher values indicating better cognitive performance. All scores were obtained directly from the ADNI phenotype files.

Definition

Clinically relevant worsening on non-memory Mini-Mental State Examination (MMSE) items (BCNMMMS) was defined as a binary variable indicating whether participants were judged by clinicians to have non-memory cognitive impairment at the time of a cross-sectional assessment. This determination was based on performance in the non-memory domains of the MMSE. The BCNMMMS classification was directly obtained from ADNI and reflects clinician-verified ratings; no re-derivation was performed by the authors. While BCNMMMS is a clinician-rated rather than a quantitative measure—and is therefore susceptible to potential rater variability—any misclassification is expected to be nondifferential with respect to inflammatory biomarker levels. Prior studies underscored the clinical importance of non-memory cognitive domains, which may capture distinct spatiotemporal pathological processes, including differential neocortical versus medial temporal tau involvement, and exhibit prognostic trajectories that differ from those of memory decline.²⁹ The inflammatory factors mentioned in this article include Interleukin-1β (IL-1β), IL-6, IL-10, IL-18, IL-21, and C3.

Statistical analysis

All statistical analyses were conducted on SPSS (Statistical Product and Service Solutions, version 27.0). Descriptive statistics were used to summarize baseline characteristics, with comparisons made according to BCNMMMS. Following normality testing, continuous variables were compared using the Student's t-test or the Mann-Whitney U test, and categorical variables were compared using the chi-square test. Participants were stratified by sex (male/female). An interaction analysis examined the relationship between inflammatory factors and sex. Subsequently, logistic regression was used to evaluate associations between inflammatory factors and BCNMMMS in each sex group, reported as odds ratios (ORs) with 95% confidence intervals (CIs). Linear regression was used to assess the association between inflammatory factors and P-tau181 levels, reported as beta coefficients (β) with 95% CIs. Covariates were selected a priori based on established associations with cognitive outcomes and AD pathology and included age, years of education, smoking status, alcohol use, hypertension, cancer, stroke, and clinical diagnosis (CN, MCI, or AD, entered as indicator variables). A mediation analysis was conducted to explore the relationship between SHBG, cognitive performance, and inflammatory factors using the mediation package in R (version 4.4.1). We fitted (i) a mediator model (linear regression for SHBG) and (ii) an outcome model (logistic regression for BCNMMMS; linear regression for P-tau181), adjusting for the prespecified covariates. The average causal mediation effect (ACME), average direct effect (ADE), and total effect were estimated using non-parametric bootstrapping with 1000 simulations. Statistical significance was defined as a two-tailed p-value <0.05. All p-values across subgroups were pooled and adjusted using the Benjamini-Hochberg false discovery rate (FDR) procedure to control the overall false discovery rate. Given the cross-sectional design, mediation findings should be interpreted as exploratory and hypothesis-generating rather than causal.

Results

Differences in baseline characteristics between CN and BCNMMMS

After stratification by BCNMMMS status, diagnosis (p < 0.001) and plasma P-tau181 levels (p < 0.001) significantly differed between the two groups. Other risk factors, such as sex, age, education level, smoking, drinking, hypertension, stroke, and cancer, remained comparable (p > 0.05). None of the six inflammatory factors showed a significant difference (p > 0.05) (Table 1).

Table 1.

Comparative analysis of groups with and without clinically relevant worsening on non-memory MMSE items. Group A, individuals without clinically relevant worsening on non-memory MMSE items; and Group B, individuals with clinically relevant worsening on non-memory MMSE items.

Variables	Total (n = 317)		p
Variables	Group A (n = 279)	Group B (n = 38)	p
Sex, (male), n (%)	172 (61.65)	24 (63.16)	0.857
Age, (years)	75.00 (71.00, 80.00)	74.00 (70.00, 78.25)	0.431
Education, (years)	16.00 (13.00, 18.00)	16.00 (12.75, 18.00)	0.687
Smoking, n (%)	121 (43.37)	17 (44.74)	0.873
Drinking, n (%)	13 (4.66)	1 (2.63)	0.568
Hypertension, n (%)	134 (48.03)	15 (39.47)	0.322
Cancer, n (%)	74 (26.52)	9 (23.68)	0.709
Stroke, n (%)	5 (1.8)	2 (5.3)	0.172
SHBG	1.76 (1.63, 1.88)	1.78 (1.66, 1.90)	0.430
Testosterone	0.33(−0.23, 0.49)	0.33 (−0.23, 0.49)	0.796
P-tau, (×pg/mL)	29.00 (19.00, 41.00)	39.00 (32.50, 54.25)	<0.001
Diagnosis, n (%)			<0.001
CN	57 (20.43)	0 (0.00)
MCI	160 (57.35)	16 (42.11)
AD	62 (22.22)	22 (57.89)
Inflammatory factors
IL-1β	2.15 (2.05, 2.25)	2.12 (2.01, 2.21)	0.410
IL-6	3.95 (2.87, 5.49)	3.92 (3.22, 5.02)	0.625
IL-10	5.23 (4.18, 7.23)	5.10 (4.36, 6.96)	0.953
IL-18	2.39 (2.30, 2.51)	2.36 (2.27, 2.47)	0.213
IL-21	7.02 (1.99, 17.11)	9.38 (0.87, 21.06)	0.805
C3	0.26 (0.20, 0.30)	0.26 (0.22, 0.31)	0.351

CN: Cognitively normal; MMSE: Mini-Mental State Examination; BCNMMMS: Clinically relevant worsening on non-memory MMSE items; SHBG: sex hormone-binding globulin; HC: Health control; MCI: Mild cognitive impairment; AD: Alzheimer's disease; IL: Interleukin; C3: Complement component 3.

The data for SHBG, testosterone and inflammatory factors were log-transformed (base 10) to reduce skewness and improve normality.

Analysis of sex-dependent effect of inflammatory factors

To investigate the interaction between sex, inflammatory factors, BCNMMMS, and P-tau181 levels, an interaction effect analysis was conducted. In BCNMMMS, we found that IL-1β (p = 0.048), IL-21 (p = 0.020), and complement component 3 (C3) (p = 0.041) exhibited significant interaction effects with sex, whereas other inflammatory factors did not show any significant interaction with sex (p > 0.05). None of the inflammatory factors showed a significant correlation with P-tau181 levels. In models without sex stratification, none of the inflammatory factors showed a statistically significant main effect on P-tau181 levels. However, significant interactions with sex were observed for IL-1β, IL-21, and C3, which motivated subsequent sex-stratified analyses (p > 0.05) (Table 2).

Table 2.

Interaction analysis of sex and interleukins.

Variables	p for interaction
Variables	BCNMMMS	P-tau
IL-1β	0.048	0.219
IL-6	0.730	0.621
IL-10	0.634	0.700
IL-18	0.183	0.637
IL-21	0.020	0.819
C3	0.041	0.759

MMSE: Mini-Mental State Examination; BCNMMMS: Clinically relevant worsening on non-memory MMSE items; IL: Interleukins; C3: Component 3.

Upon stratifying the cohort by sex, significant between-group differences were observed for several baseline characteristics, including age (p = 0.019), education (p < 0.001), smoking (p = 0.001), cancer (p = 0.005), SHBG (p < 0.001), and testosterone levels (p < 0.001). Other risk factors, such as drinking, hypertension, stroke, clinically relevant worsening on MMSE (BCMMSE), BCNMMMS, diagnosis, and P-tau181 level, suggested no significant relationship (p > 0.05) (Supplemental Table 1). Analysis of inflammatory factors revealed that IL-1β, IL-10, IL-21, and C3 remained comparable between males and females (all p > 0.05). In contrast, IL-6 and IL-18 levels were significantly higher in males than in females (both p < 0.001; Supplemental Figure 1).

To assess potential sex-dependent effects of inflammatory factors on BCNMMMS status, logistic regression models were employed. Given the large heterogeneity observed, z-scores were used to analyze IL-1β, IL-6, IL-18, IL-21, and C3 levels. The multivariate logistic regression analyses identified IL-18 levels in males (p = 0.044) and C3 levels in females (p = 0.036) as potential protective factors for BCNMMMS. In univariate analysis, high IL-21 levels in females were associated with increased risk, although this association was not observed in the multivariate regression model (Figure 1). Regarding memory-specific decline (BCMMSREC), no significant associations were observed for all inflammatory factors (p > 0.05) (Supplemental Table 2).

Figure 1.

Logistic regression models for inflammatory factors and BCNMMMS. *Adjusted for age, education, smoking, drinking, hypertension, cancer, stroke, diagnosis. Due to the large heterogeneity that was observed, IL-1β, IL-6, IL-18, IL-21, and C3 were included in the analysis using standardized scores (Z-scores). MMSE: Mini-Mental State Examination; BCNMMMS: clinically relevant worsening on non-memory MMSE items; IL: interleukins; C3: complement component 3.

Given that plasma P-tau181 serves as a biomarker of cortical tau deposition and cognitive function, linear regression was used to examine sex-specific associations with inflammation. We found that IL-21 levels in the female population showed a significant correlation with P-tau181, both before and after adjusting for confounders (unadjusted: p = 0.004; adjusted: p = 0.002), while higher C3 levels in the female population were identified as a potential protective factor against elevated P-tau181 levels after adjusting for confounders (p = 0.047). This suggests that inflammatory factors and complement components exhibit sex-specific associations with P-tau181 levels (Table 3). The association between IL-21 and P-tau181 showed inverse linear trends in male versus female participants. Although IL-21 was not significantly negatively correlated with P-tau181 in males (p = 0.337), it emerged as a statistically significant risk factor for elevated P-tau181 levels in females (p = 0.002). A similar pattern was observed in C3, which demonstrated a non-significant opposing trend that became apparent only after covariate adjustment. These findings suggest that inflammatory factors and complement exert contrasting effects on P-tau181 synthesis across sex groups (Supplemental Figure 5).

Table 3.

Linear regression models for the P-tau181 and inflammatory factors.

Variables	Unadjusted analysis						Adjusted analysis
	Male			Female			Male			Female
	B	95%CI	p	B	95%CI	p	B	95%CI	p	B	95%CI	p
IL-1β	1.194	−1.441, 3.830	0.372	0.766	−1.861, 3.392	0.565	0.851	−1.630, 3.331	0.500	0.967	−1.621, 3.554	0.461
IL-6	−0.369	−2.971, 2.233	0.780	1.055	−1.642, 3.751	0.440	−0.621	−3.117, 1.875	0.624	0.750	−1.820, 3.321	0.564
IL-10	−0.034	−0.132, 0.065	0.502	−0.510	−1.512, 0.492	0.316	−0.035	−0.129, 0.060	0.469	−0.504	−1.486, 0.479	0.312
IL-18	−1.360	−4.116, 1.395	0.331	−0.147	−2.741, 2.446	0.911	−0.637	−3.319, 2.046	0.640	−0.692	−3.250, 1.866	0.593
IL-21	−2.165	−4.881, 0.551	0.117	3.625	1.160, 6.090	0.004	−1.280	−3.904, 1.343	0.337	3.827	1.393, 6.261	0.002
C3	0.894	−1.672, 3.460	0.493	−1.662	−4.380, 1.055	0.228	0.159	−2.299, 2.618	0.898	−2.677	−5.317, −0.037	0.047

IL: Interleukin; C3: Component 3.

*Adjusted for Age, Education, Smoking, Drinking, Hypertension, Cancer, Stroke, Diagnosis.

Mediation effects in different sexes

Given the observed sex-specific effects in SHBG and testosterone levels, we further investigated their potential mediating roles in the sex-dependent effects of these inflammatory factors. It was suggested that SHBG mediated this effect in females (proportion of mediation = 19.72%; p = 0.032), whereas this effect was not significant in males (proportion of mediation = 0.37%; p = 0.898) (Figure 2). We also analyzed the associations of IL-18 and IL-21 with cognitive performance and P-tau181 levels and explored the potential mediating role of testosterone in these relationships. In males, IL-18 showed a marginal association with non-memory cognitive performance (p = 0.084). While the results suggested a potential mediating role for testosterone, this association did not reach statistical significance (proportion of mediation = 0.02%, p = 0.968). This suggested that testosterone may not be a key mediator in the relationship between IL-18 and cognitive performance. In females, a significant positive correlation was found between IL-21 and P-tau181 levels (p < 0.001). However, testosterone did not appear to significantly mediate this relationship (proportion of mediation = 2.67%, p = 0.488). This result suggested that IL-21 may influence P-tau181 levels independently of testosterone (Supplemental Figure 4).

Figure 2.

Mediating effects of SHBG for outcomes. (A) The mediating effects of IL-18 for SHBG in the male group. (B) The mediating effects of IL-18 for SHBG in the female group. (C)The mediating effects of IL-21 for SHBG in the male group. (D) The mediating effects of IL-21 for SHBG in the female group. MMSE: Mini-Mental State Examination; BCNMMMS: clinically relevant worsening on non-memory MMSE items; SHBG: sex hormone-binding globulin; IL: interleukin; C3: complement component 3.

Further mediation analysis of the relationship between C3 and P-tau181 showed that, although C3 yielded an indirect effect on P-tau181 levels in females through SHBG (p = 0.032), this effect was not statistically significant (proportion of mediation = 26.92%, p = 0.066). This marginal significance suggested that SHBG may modulate the effect of C3. However, the effect was relatively weak (Supplemental Figure 2), which may be due to the small sample size. Besides, testosterone levels in females mediated the relationship between C3 and P-tau181 to some extent, although the mediating effect was not strong (Proportion of mediation = 1.84%, p = 0.065). This indicated that testosterone may play a role in mediating the effects of C3 on P-tau181 levels (Supplemental Figure 3).

Exploratory multidomain analyses

Finally, to further characterize potential sex-specific patterns across cognitive domains, exploratory consistency analyses examined associations between inflammatory biomarkers and memory, executive function, language, visuospatial abilities, and the plasma Aβ_40/42 ratio. In sex-stratified analyses, differential association patterns between inflammatory markers and cognitive or biomarker outcomes were observed across sexes (Supplemental Table 4). In male participants (sex = 1), higher IL-6 levels were positively associated with higher language and visuospatial composite scores, and IL-18 was positively associated with the plasma Aβ_40/42 ratio. In contrast, among female participants (sex = 2), IL-21 showed an inverse association with language composite scores, whereas C3 was positively associated with the plasma Aβ_40/42 ratio. Corresponding estimates in the opposite sex were not statistically significant.

Discussion

In the present study, the distribution of diagnostic categories (CN, MCI, AD) showed no statistically significant sex difference (p = 0.056), despite a non-significant trend toward a higher MCI proportion in males and a higher proportion of healthy controls and AD in females. Conversely, inflammatory markers exhibited marked sex-related differences. Male patients showed significantly higher levels of IL-6 and IL-18 compared with female patients (IL-6: p < 0.001; IL-18: p < 0.001).

In the exploration of interaction effects, we found statistically significant sex-dependent interactions between BCNMMMS and IL-1β, IL-21, and C3. However, no such interaction was observed for plasma P-tau181 levels. Stratified logistic regression analysis showed that higher C3 levels in females were associated with a lower risk of BCNMMMS worsening, while higher IL-18 levels in males also acted as a protective factor. No significant association between IL-1β and BCNMMMS was detected in either sex group. In linear regression analyses of plasma P-tau181, IL-21 and C3 levels in females were significantly associated with P-tau181, whereas this trend appeared to be reversed in males. These findings collectively indicate that inflammatory factors and complement components exhibit sex-specific associations, with differences not only in significance but also in direction and magnitude, suggesting distinct patterns of inflammatory influence between female and male participants. After adjusting for multiple comparisons using the Benjamini-Hochberg FDR procedure, only the association between IL-21 and P-tau181 remained statistically significant in the female subgroup (FDR-adjusted p = 0.024). Following adjustment, the other initially significant associations were no longer statistically significant. Consequently, these results may require careful interpretation. Nevertheless, the observed trends for other inflammatory factors, though not statistically significant after adjustment, may still hold clinical relevance and warrant further investigation in larger cohorts.

Baseline characteristics stratified by sex revealed significant differences in SHBG and testosterone levels. Further mediation analysis indicated that the relationship between female IL-21 and plasma P-tau181 levels was mediated by SHBG. This suggests that inflammatory biomarkers such as IL-18, IL-21, and C3 may show sex-specific patterns of association with non-memory cognitive impairment and tau pathology. Rather than indicating direct treatment targets, our findings highlight the need to consider sex when designing and interpreting biomarker studies in AD and when generating hypotheses for future interventional research. This may lead to the development of clinically differentiated cognitive performance drugs tailored to sex-specific treatments.

The exploratory consistency analyses indicated sex-specific patterns in the associations between inflammatory or complement markers and cognitive domains, as well as the plasma Aβ_40/42 ratio. In males, IL-6 and IL-18 were positively associated with non-memory cognitive composite scores and the plasma Aβ_40/42 ratio. Conversely, in females, IL-21 showed an inverse association with language performance, while C3 exhibited a positive association with the plasma Aβ_40/42 ratio. Due to the exploratory nature of these analyses, which were adjusted only for pre-specified covariates and are inherently limited by multiple testing and modest subgroup sizes, these findings warrant cautious interpretation.

Several studies have suggested that the inflammatory factors IL-18, IL-21, and C3 are potential therapeutic targets. The NOD-like receptor thermal protein domain-associated protein 3 (NLRP3) inflammasome is a key component of the innate immune system and plays an essential role in inflammation-related diseases such as atherosclerosis, AD, and inflammatory bowel disease (IBD) by promoting the release of IL-1β and IL-18.³⁰ At the same time, IL-18 increases Aβ production and the expression of Cdk5 and GSK-3β, which are involved in P-tau181. IL-18 also induces mitochondrial stress and oxidative stress (OS) in neurons, either directly or indirectly, including via Aβ.³¹ This suggests that IL-18 may be considered a risk factor in previous reports; however, in our study, IL-18 and C3 appeared to act as protective factors for non-memory cognitive performance in males.

IL-17, IL-21, and IL-22 levels are elevated in patients with AD.³² In AD patients, Aβ-induced IL-21 production in CD4+ T cells is significantly increased, and the extent of brain atrophy is positively correlated with an increase in IL-21-secreting CD4+ T cells.³³ This suggests that IL-21 may be the pathological result of Aβ amyloid deposition in AD. It is now understood that patients with AD mount a strong immune response to Aβ, possibly to clear excess Aβ. Enhanced production of inflammatory factors (such as IL-21) leads to a cascade of responses. However, a strong immune response that fails to effectively clear Aβ can exacerbate inflammation, worsen AD neuropathology, and cognitive dysfunction.³⁴ This phenomenon suggests that IL-21 may mediate the worsening of cognitive performance resulting from AD.

In the present study, we found that C3 is associated with cognitive performance in AD, consistent with the literature. For instance, a study by Litvinchuk et al.³⁵ found that the expression of C3 and the C3a receptor (C3aR1) was positively correlated with cognitive performance and Braak staging of the human AD brain, indicating that C3 affects cognitive ability and P-tau181 levels.

Analysis by western blotting showed that Aβ₄₂ levels were increased in AD brain as compared to healthy people.³⁶ Inflammatory factors may have region-specific effects on non-memory cognitive performance rather than memory-related cognitive performance. Spatial correlation analyses have demonstrated that neuroinflammation shows stronger co-localization with tau pathology than with Aβ deposition, and inflammatory burden in AD-related regions has been associated with cognitive impairment. These findings underscore the potential role of inflammation in modulating tau-related neurodegeneration and support its relevance as a therapeutic target in AD.³⁷

Sex differences in occupancy of the SHBG steroid-binding sites can be explained easily by the much higher testosterone and dihydrotestosterone levels in men than in women, coupled with the fact that SHBG levels are approximately 50% lower in men than in women.³⁸ SHBG may influence inflammatory responses by mediating the bioavailability of sex hormones, thereby indirectly affecting cognitive functions. Studies have shown that estrogens have several neuroprotective effects related to the prevention of attention deficits, particularly by promoting neuronal vitality and reducing Aβ accumulation.³⁹ Circulating SHBG is a major determinant of the metabolic clearance of sex steroid hormones, and it modulates their access to target tissues.⁴⁰ Therefore, changes in SHBG levels may directly influence the biological effects of sex hormones. This study also explored the mediating role of SHBG and testosterone.

In addition to these essential features, several other pathological changes such as inﬂammation, sustained activation of microglia and other immune cells are commonly observed in association with AD.⁴¹ There are many mechanisms of AD formation, and inflammation is one of them. It involves complex processes, including both complement components and inflammatory cells and the substances they produce, such as cytokines and chemokines.⁴² Research has shown that downregulation of SHBG expression is closely associated with elevated levels of circulating pro-inflammatory factors. High levels of inflammation are often accompanied by changes in SHBG levels, possibly through mechanisms that promote or inhibit SHBG synthesis and metabolism. For example, studies have found a negative correlation between hsCRP levels and testosterone and SHBG levels, whereas IL-6 was positively correlated with SHBG levels.⁴³ However, in this study, no association between IL-6 and SHBG was found, likely due to insufficient statistical power stemming from the limited patient sample.

Sex steroid hormones generally have anti-inflammatory effects.⁴⁴ The differences in sex hormone and SHBG levels between males and females may lead to differences in inflammatory factors, which could, in turn, impact cognitive performance and AD pathology. Our analysis revealed sex-specific differences in IL-21 and C3 levels, and identified SHBG as a mediator of the relationship involving IL-21.

To our knowledge, this study is among the first to investigate sex-specific associations between peripheral inflammatory biomarkers, a clinically defined worsening on non-memory MMSE items, and plasma P-tau181 within the ADNI cohort, while also exploring the potential mediating role of SHBG. By considering cognitive performance, tau pathology, and sex hormone-related traits, our work extends previous studies that examined these components in isolation. However, the study also has several limitations that should be acknowledged. The study's statistical power is constrained by the sample size, which allows only for the identification of correlations rather than the establishment of causality. Furthermore, the class imbalance present in the BCNMMMS outcome may have biased the statistical models. Although domain-specific MMSE analyses are reported in the literature, our cohort size prevented such stratified examinations, leading to the use of the composite BCNMMMS measure from ADNI. The panel of inflammatory markers analyzed was also limited; factors such as tumor necrosis factor-α (TNF-α), IL-3, IL-7, and IL-9 did not yield significant results in our models. A significant methodological constraint was the low rate of CSF (n = 27) collection due to the invasiveness of lumbar puncture. Consequently, CSF-based analyses were underpowered, and the generalizability of findings involving CSF biomarkers may be limited.

Future research could include a broader range of inflammation-related factors and investigate other potential contributors to this sex-dependent effect, such as testosterone and luteinizing hormone, to better understand these specific relationships. Longitudinal studies could also be conducted to explore whether these sex-dependent effects of inflammatory factors occur at which stage of neurodegeneration. In AD, the interaction between immune responses and sex hormones may become a key factor. Therefore, understanding the roles of inflammatory factors such as IL-18, IL-21, and C3 in sex differences can help us provide more effective treatments.

Conclusion

Our findings indicate that inflammatory biomarkers such as IL-18, IL-21, and C3 may exhibit sex-specific associations with non-memory cognitive impairment and plasma P-tau181 in ADNI. These results do not imply therapeutic targets but rather highlight the importance of considering sex when interpreting biomarker profiles related to inflammation and tau pathology. Given the cross-sectional design and the loss of statistical significance after FDR correction for several associations, these findings provide a basis for new hypotheses; however, they require confirmation in prospective longitudinal cohorts and via studies designed to elucidate the underlying mechanisms.

Supplemental Material

sj-docx-1-alz-10.1177_13872877261440932 - Supplemental material for Sex-dependent role of inflammatory biomarkers in Alzheimer's disease-related cognitive performance

Supplemental material, sj-docx-1-alz-10.1177_13872877261440932 for Sex-dependent role of inflammatory biomarkers in Alzheimer's disease-related cognitive performance by Dehao Yang, Bo Zhang, Shiyue Wang, Jiaxuan Chen, Jinrong Zhu, Ruting Wei, Hejia Cai, Bingxin Teng, Xiaoya Xie, Bohao Li, Xicheng Yu, Suwen Huang, Guangyong Chen and Yiyun Weng in Journal of Alzheimer's Disease

Footnotes

Acknowledgements

We gratefully acknowledge the ADNI for providing the valuable data used in this study. Their support and resources were essential to our research.

ORCID iDs

Dehao Yang

Guangyong Chen

Yiyun Weng

Ethical considerations

This study used publicly available, anonymized data and did not require ethics approval.

Consent to participate

All participants in the ADNI study provided written informed consent at the time of enrollment.

Consent for publication

Not applicable.

Author contribution(s)

Dehao Yang: Formal analysis; Funding acquisition; Supervision; Writing – original draft; Writing – review & editing.

Bo Zhang: Data curation; Formal analysis; Writing – original draft; Writing – review & editing.

Shiyue Wang: Data curation; Formal analysis; Writing – original draft; Writing – review & editing.

Jiaxuan Chen: Data curation; Formal analysis.

Jinrong Zhu: Data curation.

Ruting Wei: Data curation.

Hejia Cai: Data curation.

Bingxin Teng: Data curation.

Xiaoya Xie: Data curation.

Bohao Li: Data curation.

Xicheng Yu: Data curation.

Suwen Huang: Data curation.

Guangyong Chen: Conceptualization.

Yiyun Weng: Conceptualization.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the National Natural Science Foundation of China (No. 82302081), the Zhejiang Provincial Natural Science Foundation of China (No. LQ24H090003) and the National Innovation and Entrepreneurship Training Program for College Students (No.202510343022).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

Data may be obtained from a third party and are not publicly available. The original data are held by the ADNI () and access is available to any bona fide researcher proposing a health-related project that is in the public's interest. Researchers must be registered with the ADNI's Access Team.

Supplemental material

Supplemental material for this article is available online.

References

Lane

Hardy

Schott

. Alzheimer's disease. Eur J Neurol 2018; 25: 59–70.

2024 Alzheimer's disease facts and figures. Alzheimers Dement 2024; 20: 3708–3821.

Glass

Saijo

Winner

, et al. Mechanisms underlying inflammation in neurodegeneration. Cell 2010; 140: 918–934.

Calsolaro

Edison

. Neuroinflammation in Alzheimer's disease: current evidence and future directions. Alzheimers Dement 2016; 12: 719–732.

Zimmer

Leuzy

Benedet

, et al. Tracking neuroinflammation in Alzheimer's disease: the role of positron emission tomography imaging. J Neuroinflammation 2014; 11: 120.

Weisman

Hakimian

. Interleukins, inflammation, and mechanisms of Alzheimer's disease. Vitam Horm 2006; 74: 505–530.

Costantini

Carrarini

Calisi

, et al. Search in the periphery for potential inflammatory biomarkers of dementia with Lewy bodies and Alzheimer's disease. J Alzheimers Dis 2024; 99: 1147–1158.

Cao

Tan

, et al. The prevalence of dementia: a systematic review and meta-analysis. J Alzheimers Dis 2020; 73: 1157–1166.

Nebel

Aggarwal

Barnes

, et al. Understanding the impact of sex and gender in Alzheimer's disease: a call to action. Alzheimers Dement 2018; 14: 1171–1183.

10.

Scheyer

Rahman

Hristov

, et al. Female sex and Alzheimer's risk: the menopause connection. J Prev Alzheimers Dis 2018; 5: 225–230.

11.

Lopez-Lee

Torres

ERS

Carling

, et al. Mechanisms of sex differences in Alzheimer's disease. Neuron 2024; 112: 1208–1221.

12.

McCarrey

Kitner-Triolo

, et al. Sex differences in cognitive trajectories in clinically normal older adults. Psychol Aging 2016; 31: 166–175.

13.

Murre

Janssen

Rouw

, et al. The rise and fall of immediate and delayed memory for verbal and visuospatial information from late childhood to late adulthood. Acta Psychol (Amst) 2013; 142: 96–107.

14.

Sohn

Shpanskaya

Lucas

, et al. Sex differences in cognitive decline in subjects with high likelihood of mild cognitive impairment due to Alzheimer's disease. Sci Rep 2018; 8: 7490.

15.

Buckley

Mormino

Amariglio

, et al. Sex, amyloid, and APOE ε4 and risk of cognitive decline in preclinical Alzheimer's disease: findings from three well-characterized cohorts. Alzheimers Dement 2018; 14: 1193–1203.

16.

Dolsen

Crosswell

Prather

. Links between stress, sleep, and inflammation: are there sex differences? Curr Psychiatry Rep 2019; 21: 8.

17.

Vogel

Muhammad

Song

, et al. Sex differences in pain and quantitative sensory testing in patients with rheumatoid arthritis. Arthritis Care Res 2023; 75: 2472–2480.

18.

Ryan

Mills

KHG

. Sex differences regulate immune responses in experimental autoimmune encephalomyelitis and multiple sclerosis. Eur J Immunol 2022; 52: 24–33.

19.

Klein

Flanagan

. Sex differences in immune responses. Nat Rev Immunol 2016; 16: 626–638.

20.

Walker

Gottesman

, et al. Systemic inflammation during midlife and cognitive change over 20 years: the ARIC study. Neurology 2019; 92: e1256–e1267.

21.

Acosta-Martínez

Carter

Nessim

, et al. Sex- and region-dependent neuroinflammation in Alzheimer's disease. Alzheimers Dement 2025; 21: e14603.

22.

Thaler

Seifert-Klauss

Luppa

. The biomarker sex hormone-binding globulin - from established applications to emerging trends in clinical medicine. Best Pract Res Clin Endocrinol Metab 2015; 29: 749–760.

23.

Liao

, et al.

Low serum sex hormone-binding globulin: marker of inflammation?

Clin Chim Acta 2012; 413: 803–807.

24.

Shen

, et al. Plasma sex hormone-binding globulin predicts neurodegeneration and clinical progression in prodromal Alzheimer's disease. Aging 2020; 12: 14528–14541.

25.

Jin

Hua

, et al. Correlations between SHBG, sex hormones, inflammation, and neurocognitive decline in Alzheimer's disease: a retrospective study. Curr Alzheimer Res 2024; 21: 491–502.

26.

Yeap

Flicker

. Testosterone, cognitive decline and dementia in ageing men. Rev Endocr Metab Disord 2022; 23: 1243–1257.

27.

Crane

Carle

Gibbons

, et al. Development and assessment of a composite score for memory in the Alzheimer's disease neuroimaging initiative (ADNI). Brain Imaging Behav 2012; 6: 502–516.

28.

Shaw

Vanderstichele

Knapik-Czajka

, et al. Cerebrospinal fluid biomarker signature in Alzheimer's disease neuroimaging initiative subjects. Ann Neurol 2009; 65: 403–413.

29.

Wuestefeld

Xie

McGrew

, et al. Tau, atrophy, and domain-specific cognitive impairment in typical Alzheimer's disease. Alzheimers Dement 2025; 21: e70511.

30.

Chen

Escames

, et al. The NLRP3 inflammasome: contributions to inflammation-related diseases. Cell Mol Biol Lett 2023; 28: 51.

31.

Ojala

Sutinen

. The role of interleukin-18, oxidative stress and metabolic syndrome in Alzheimer's disease. J Clin Med 2017; 6: 55.

32.

Saresella

Calabrese

Marventano

, et al. Increased activity of Th-17 and Th-9 lymphocytes and a skewing of the post-thymic differentiation pathway are seen in Alzheimer's disease. Brain Behav Immun 2011; 25: 539–547.

33.

Baglio

Saresella

Preti

, et al. Neuroinflammation and brain functional disconnection in Alzheimer's disease. Front Aging Neurosci 2013; 5: 81.

34.

Baulch

Acharya

Agrawal

, et al. Immune and inflammatory determinants underlying Alzheimer's disease pathology. J Neuroimmune Pharmacol 2020; 15: 852–862.

35.

Litvinchuk

Wan

Swartzlander

, et al. Complement C3aR inactivation attenuates tau pathology and reverses an immune network deregulated in tauopathy models and Alzheimer's disease. Neuron 2018; 100: 1337–1353.e1335.

36.

Raha

Vaishnav

Friedland

, et al. The systemic iron-regulatory proteins hepcidin and ferroportin are reduced in the brain in Alzheimer's disease. Acta Neuropathol Commun 2013; 1: 55.

37.

Appleton

Finn

Zanotti-Fregonara

, et al. Brain inflammation co-localizes highly with tau in mild cognitive impairment due to early-onset Alzheimer's disease. Brain 2025; 148: 119–132.

38.

Hammond

. Diverse roles for sex hormone-binding globulin in reproduction. Biol Reprod 2011; 85: 431–441.

39.

Mrdjen

Fox

Bukhari

, et al. The basis of cellular and regional vulnerability in Alzheimer's disease. Acta Neuropathol 2019; 138: 729–749.

40.

Pugeat

Nader

Hogeveen

, et al. Sex hormone-binding globulin gene expression in the liver: drugs and the metabolic syndrome. Mol Cell Endocrinol 2010; 316: 53–59.

41.

Ozben

. Neuro-inflammation and anti-inflammatory treatment options for Alzheimer's disease. Clin Biochem 2019; 72: 87–89.

42.

Twarowski

Herbet

. Inflammatory processes in Alzheimer's disease-pathomechanism, diagnosis and treatment: a review. Int J Mol Sci 2023; 24: 6518.

43.

Osmancevic

Daka

Michos

, et al. The association between inflammation, testosterone and SHBG in men: a cross-sectional multi-ethnic study of atherosclerosis. Clin Endocrinol (Oxf) 2023; 99: 190–197.

44.

Uchoa

Moser

Pike

. Interactions between inflammation, sex steroids, and Alzheimer's disease risk factors. Front Neuroendocrinol 2016; 43: 60–82.