Stress and glucocorticoid exposure as a critical environmental risk factor for Alzheimer’s disease

Abstract

The hypothalamic-pituitary-adrenal (HPA) axis coordinates the physiological stress response by driving glucocorticoid (GC) secretion. While acute glucocorticoid signaling supports adaptive cognitive and metabolic responses, sustained activation of this system can produce hippocampal atrophy, impaired synaptic plasticity, and long-term cognitive dysfunction. Alzheimer’s disease (AD), characterized by extracellular amyloid beta (Aβ) plaques and neurofibrillary tangles (NFTs), arises from interactions among genetic variants, epigenetic regulation, and environmental exposures. Here, we review clinical, epidemiological, and translational evidence implicating chronic stress and glucocorticoid exposure as a potent environmental risk factor for AD. Human observations linking stress exposure to cognitive impairment are complemented by clinical hypercortisolemia paradigms, which highlight the sensitivity of hippocampal circuits to prolonged cortisol exposure. The detrimental impact of stress on cognition is mediated by GC-responsive genes implicated in AD or, conversely, by AD genes that also respond to GCs. Animal models further support causality, showing that chronic stress or glucocorticoid exposure exacerbates Aβ accumulation, tau pathology, synaptic dysfunction, and cognition in AD-relevant systems. Finally, stress and GC exposure regulate GC or AD genes through epigenetic mechanisms, which can encode cumulative stress exposure and bias trajectories toward neurodegeneration. The involvement of stress in AD elucidates molecular mechanisms underlying its complex pathology but also provides a promising avenue of investigation for AD therapy and intervention strategies.

1. Introduction – Stress as a significant risk factor for Alzheimer’s disease (AD)

The physiological stress response is an evolutionarily conserved system designed to preserve internal stability when faced with actual or perceived threats to homeostasis (Godoy et al., 2018). Central to this response is the hypothalamic-pituitary-adrenal (HPA) axis (Russell and Lightman, 2019), whose activation culminates in the secretion of glucocorticoids (GCs) (primarily cortisol in humans) from the adrenal cortex. The release of GCs enables the organism to make glucose available for executing the “fight or flight” response. Acute HPA axis activation and cortisol signaling via the glucocorticoid receptor (GR) support adaptive cognitive and metabolic responses. However, sustained activation of this system or chronic stress exposure leads to a host of organ system impairments. Organs and tissues that are repeatedly engaged during the “fight or flight” response become overburdened (i.e., cardiometabolic diseases) whereas those not necessary for immediate survival, such as reproductive and digestive systems, are suppressed. The brain, which provided adaptive cognitive response, muscle coordination, and memory consolidation during acute stress exposure, undergoes hippocampal atrophy, impaired synaptic plasticity, and long-term cognitive dysfunction following chronic stress exposure (Lupien et al., 2009).

Cortisol diffuses across the plasma membrane to engage intracellular mineralocorticoid receptor (MR) and GR that bind to cortisol at low and high concentrations, respectively, and cortisol-GR signaling is tightly regulated by chaperone proteins such as FK506-inding protein 5 (FKBP5) and heat shock proteins (HSP70 and HSP90). GC signaling can modulate cellular behavior either by the GR functioning as a transcription factor or engaging with cytoplasmic proteins independent of direct gene regulation. These actions affect GC sensitivity through negative feedback mechanisms and regulate multiple organ-specific pathways that are typically associated with GC action and disease (Su et al., 2025). One organ that is particularly sensitive to GC exposure is the brain, as chronic exposure to stress or GCs is associated with psychiatric and neurological disorders and neurodegenerative diseases (Menard et al., 2017; Vreeburg et al., 2009; Vyas et al., 2016). Emerging evidence establishes chronic stress as a significant modifiable risk factor for AD through both epidemiological associations and biological mechanisms (Fujii et al., 2014). In this review, we propose that chronic stress acts as a modifiable environmental risk factor that accelerates the progression of Alzheimer’s disease-related neuropathology. This theoretical framework serves to reconcile largely observational human studies with the mechanistic insights revealed by animal and cellular models, where human data substantiate correlations and risk-modifying effects, while experimental models elucidate the molecular relationship between chronic stress and AD-like pathological changes.

1.1. Overview of Alzheimer’s disease

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the leading cause of dementia in older adults. It impairs memory, learning, and cognitive functions, affecting an estimated 7.4 million Americans today (Miao et al., 2026). Dementia is a clinical syndrome of cognitive decline that impairs functional independence and can arise from multiple pathologies, including cerebrovascular insufficiency, α-synuclein deposition (Lewy body disease), and frontotemporal neurodegeneration; AD, defined neuropathologically by extracellular amyloid-β (Aβ) plaques and intracellular neurofibrillary tangles (NFTs), is the most common underlying pathology, accounting for approximately 60–80% of all dementia diagnoses (Emmady et al., 2025). AD arises from an intricate interplay of inherited genetic variants, epigenetic regulatory modifications, and external environmental influences. Two primary forms of AD exist: familial and sporadic. Familial AD (FAD), contributing to approximately 1% of all cases, is caused by autosomal-dominant mutations in genes encoding proteins involved in amyloid protein processing, such as amyloid precursor protein (APP) itself, presenilin 1 (PSEN1), and presenilin 2 (PSEN2) (Bekris et al., 2010). These mutations lead to aberrant proteolytic processing of APP, favoring the production of neurotoxic Aβ peptides. In contrast, sporadic AD or late-onset AD (LOAD) is a more common form of the disease that arises from the combined influence of multiple common genetic variants (such as ABCA7, TREM2, etc.) (Shi et al., 2012; Bellenguez et al., 2022) together with non-genetic contributors such as aging, cardiovascular health, and environmental exposures. Among these inherited risk factors, the APOE-ε4 (APOE4) allele confers the strongest common genetic risk and influences Aβ aggregation, cholesterol metabolism, and neuroinflammation (Serrano-Pozo et al., 2021).

1.2. Aβ-plaque and NFTs in the pathogenesis of AD

The neuropathological signatures of AD include the deposition of Aβplaques and intracellular aggregates of hyperphosphorylated tau that impair neuronal signaling integrity (Mila-Aloma et al., 2020). Aβ is generated by the sequential proteolysis of the transmembrane protein APP by the β-secretase (BACE1) and γ-secretase complex (of which PSEN1 and PSEN2 are catalytic components), producing mostly hydrophobic Aβ42 peptides that aggregate into oligomers and plaques, disrupting synaptic signaling and triggering neuroinflammation (Zhang and Song, 2013). In a parallel, non-amyloidogenic pathway, α-secretase cleaves APP within the Aβ region to yield soluble APPα and the C83 fragment, preventing Aβ generation.

Tau, a microtubule-associated protein encoded by the MAPT gene, normally binds microtubules to maintain axonal architecture and supports the transport of cargoes required for synaptic connectivity (Wang and Mandelkow, 2016). In AD, aberrant tau hyperphosphorylation at serine and threonine residues (Ser202, Thr231, and Ser396) is mediated by several kinases, including glycogen synthase kinase-3β (GSK-3β), cyclin-dependent kinase 5 (CDK5), and mitogen-activated protein kinases (MAPKs), whose activities are significantly modulated by the presence of Aβ peptides but also by chronic stress and GC exposure (Sotiropoulos et al., 2011; Alvarez et al., 1999). Hyperphosphorylated tau detaches from microtubules, misfolds, and self-assembles into paired helical filaments that give rise to NFTs (Sotiropoulos et al., 2011; Takashima, 2008). This pathological transformation impairs axonal transport, compromises synaptic plasticity, and ultimately precipitates neuronal death, with tau pathology distribution correlating more closely with cognitive impairment severity than amyloid burden (Bejanin et al., 2017).

1.3. Human studies linking stress and AD

With the basic molecular pathology of AD summarized, we now consider how stress has been associated with cognitive impairment, dementia, and AD. Human epidemiological evidence, while necessarily observational, provides the most direct support for a link between chronic stress and AD. A recent population-based cohort study of more than a million individuals reported an association in which chronic stress was linked with a 2.45-fold increase in AD risk, with the additive effects of depression yielding a 4-fold increased odds ratio (Wallensten et al., 2023). Supporting this relationship, a longitudinal study of 777 women followed for 50 years found that midlife stress was associated with nearly three-fold higher risk of developing dementia at a relatively younger age than women without stress-related exhaustion and showed more cognitive impairment 24 years later even without dementia (Guo et al., 2024). Interestingly, another population study and meta-analysis reported an increase of up to 6-fold in the incidence of dementia in spousal caregivers of dementia, which implicated chronic psychosocial and caregiving-related stress in increased dementia risk in what the authors termed “ironic tragedy” (Norton et al., 2010; Vitaliano et al., 2003; Vitaliano, 2010). Moreover, iatrogenic data of psychiatric outcomes in more than 700,000 chronic GC-exposed patients who received GC therapy compared to non-exposed patients for the same medical conditions reported more than 5-fold increase in acute neuropsychiatric symptoms including delirium, confusion, and disorientation (Fardet et al., 2012), supporting the development of neuropsychiatric vulnerabilities that can be associated with dementia. A recent biomarker-based study of 305 participants with normal cognitive function from the Framingham Heart Study showed that elevated cortisol levels in midlife were associated with increased amyloid protein accumulation 15 years later, especially in postmenopausal women. Amyloid proteins were concentrated primarily in the posterior cingulate gyrus, precuneus, and frontal lobe regions, providing evidence for the long-term effects of high-level cortisol. In men, however, midlife cortisol was not associated with amyloid protein accumulation or with tau pathology, suggesting that cortisol effects are sex- and pathology-specific and may be modulated by hormonal status (Salardini et al., 2025).

1.4. Cushing’s syndrome – A model for chronic cortisol excess and cognitive impairment

Clinical studies help to uncover the relationship between chronic stress exposure and AD by providing a more granular insight on the physiological consequences of stress on the brain. Cushing’s syndrome (CS) provides a clinical paradigm for studying how persistent cortisol excess impairs cerebral processes and may facilitate the evolution to AD pathology. The disease is characterized by hypercortisolemia, usually caused by a pituitary adenoma (Cushing’s disease), adrenal tumors, or ectopic adrenocorticotropic hormone (ACTH) secretion (Lacroix et al., 2015), and is associated with a constellation of neuropsychiatric symptoms, including depression, anxiety, and cognitive impairment (Kelly, 1996). A recent systematic review reported that patients with CS have persistent impairments in memory, attention, executive function, depression/anxiety symptoms, and structural brain alterations (Katragadda et al., 2025). Moreover, although surgical remission of hypercortisolism can lead to partial recovery of brain structure and function, residual cognitive deficits often persist, suggesting that vulnerable neuronal populations may be irreversibly damaged (Forget et al., 2002).

One of the major targets of GC signaling and hypercortisolemia is the hippocampus. As the hippocampus is central to memory encoding and consolidation and expresses abundant GR and MR, it exhibits responsiveness to circulating GCs. Under normal physiological conditions, GCs help coordinate adaptive responses to stress and support memory formation, in part by regulating synaptic plasticity mechanisms such as long-term potentiation (LTP) (de Kloet et al., 2005). However, sustained elevations in GCs arising from prolonged activation of the HPA axis can lead to impaired induction and maintenance of LTP in hippocampal networks (Kim and Diamond, 2002). This is due to alterations in the expression and function of glutamate receptors (including NMDA and AMPA receptors) and disruption of calcium signaling, as well as interference with critical intracellular pathways for synaptic strengthening (Popoli et al., 2011). Furthermore, in adults, the dentate gyrus of the hippocampus is one of the few brain regions where neuronal progenitors keep generating new neurons throughout life (Toda et al., 2019). Chronic exposure to elevated GC levels significantly diminishes the proliferation rate of neuronal progenitor populations and undermines the viability of nascent neurons and their incorporation into pre-existing circuits. Reduced neurogenesis in adults is directly associated with cognitive deficits, as newly generated neurons contribute to pattern separation, learning, and memory encoding (Klyubin et al., 2022). Chronic hypercortisolism also negatively impacts hippocampal resilience by interfering with brain-derived neurotrophic factor (BDNF) signaling, a pathway fundamental to maintaining neuronal viability, dendritic growth, and synaptic plasticity. Chronic elevation of GCs results in a significant downregulation of BDNF in the hippocampus, which is mediated by GC receptor activation that directly inhibits BDNF gene expression, disrupts BDNF mRNA stability, and interferes with BDNF protein trafficking and secretion (Chen et al., 2017). In addition to inhibiting BDNF, GCs modulate multiple molecular cascades that can further impair neuronal plasticity and resilience (Russo et al., 2012; Vyas et al., 2016; de Kloet and Meijer, 2024). GCs have a deleterious impact on Aβ and tau formation, and this topic will be discussed in greater detail below.

1.5. Psychiatric disorders (depression, anxiety) that increase risk for dementia and AD

Emerging epidemiological studies show a correlation between major psychiatric disorders and subsequent neurodegenerative disease risk. It is important to mention that since many psychiatric disorders also involve HPA axis dysregulation, untangling their roles as either comorbid or preceding conditions to AD remains a challenge. Despite these common underlying conditions, it is important to consider many studies that have examined these disorders per se in the context of AD risk.

A recent longitudinal study of older adults found that depression was associated with an increased risk of AD, all-cause dementia, and vascular dementia in which depression may function as both a long-term risk factor and a prodromal symptom of AD. AD risk was elevated when depression occurred within 2 years before AD diagnosis and also when depression had been present for 6–8 years (Xu et al., 2025). Generalized anxiety disorder manifests elevated risk for mild cognitive impairment (MCI) and an approximately 30% increased likelihood of progressive dementia across prospective cohorts (Santabarbara et al., 2019). Clinical-based surveys have shown that major depressive disorder (MDD) has shown a significant longitudinal association with subsequent AD, and a longer interval between depression diagnosis and AD onset was associated with higher risk, suggesting that a history of depression may increase the likelihood of developing AD later in life.

Post-traumatic stress disorder (PTSD) also constitutes a potential risk determinant for neurodegenerative disease, with a 61% elevated dementia incidence among individuals with documented PTSD relative to unexposed populations, implicating PTSD in heightened vulnerability to both AD and vascular dementia (Gunak et al., 2020). Neuroimaging studies provide mechanistic evidence by demonstrating spatially convergent structural abnormalities in PTSD patients with concurrent cognitive impairment, involving bilateral frontal, temporal, and parietal regions (de Araujo Jr et al., 2023). These neuroanatomical patterns demonstrate remarkable overlap with canonical AD neuropathological distributions, suggesting that PTSD-associated neurodegeneration may preferentially target identical brain networks vulnerable to AD pathology. Hippocampal structures, which are the most frequently implicated regions in PTSD-related dementia, represent precisely those anatomical substrates exhibiting earliest tau accumulation and atrophic changes in AD progression (de Araujo Jr et al., 2023).

Bipolar disorder (BD), or manic-depressive disorder, is also characterized by impairments in attention, memory, and executive function, and demonstrates polygenic overlap with AD pathology (Huang et al., 2023). Genome-wide association analyses identified two jointly associated loci implicating MARK2, a gene encoding microtubule affinity regulating kinase 2 involved in neuronal migration during corticogenesis and tau phosphorylation, with exhibiting concordant directional effects across both BD and AD phenotypes, while the VAC14 gene, which regulates endosomal homeostasis and signaling via lipid phosphatidylinositol 3,5-bisphosphate kinase (PI3K), showed opposite effects (Drange et al., 2019) with loss-of-function mutations producing neurodegeneration in experimental models (Zhang et al., 2007). The identification of these pleiotropic loci points to possible shared processes governing both neurodevelopmental migration anomalies and neurodegenerative tau dysregulation.

Schizophrenia (SCZ) also shares modest but significant polygenic and phenotypic overlap with AD. A 2025 cross-trait GWAS analysis reported 16 shared risk loci in immune and synaptic genes between SCZ and AD, indicating some convergent molecular architecture underlying cognitive decline and neurodegeneration (Liu et al., 2025a). A meta-analysis study demonstrated that individuals with SCZ were at significant risk of developing dementia (Cai and Huang, 2018). While SCZ confers heightened vulnerability to later-life neurodegeneration, its genetic architecture may interact with AD pathology through neurodevelopmental and synaptic regulatory networks, implying that early-onset SCZ reflects a sensitized neurobiological state predisposing to subsequent AD.

1.6. Non-psychiatric diseases associated with HPA axis dysregulation and AD

There are multiple non-psychiatric medical conditions that independently increase AD risk through shared HPA axis dysregulation, including type 2 diabetes, cardiovascular disease, rheumatoid arthritis, and chronic kidney disease. Studies on type 2 diabetes, obesity, and metabolic syndrome consistently demonstrate HPA axis hyperactivity and mechanistic links to AD pathology. T2DM (type 2 diabetes mellitus) patients show significantly elevated basal cortisol, impaired negative feedback after dexamethasone suppression, and exaggerated responses to corticotropin releasing hormone (CRH) stimulation, and this elevation is independent of age, BMI, or cardiovascular factors (Bruehl et al., 2007). Obesity’s neurological impact operates through multiple cortisol-dependent pathways. For example, midlife obesity increases dementia risk by visceral adiposity’s promotion of low-grade chronic inflammation that dysregulates the HPA axis (Al-Kuraishy et al., 2023). This implicates a positive feedback from adipose-tissue-inflammation-increased cortisol secretion, resulting in reciprocal adipokine dysregulation. In both T2DM and obesity, chronic HPA activation impairs insulin receptor function, reduces glucose uptake, and creates brain insulin resistance. Insulin-degrading enzyme (IDE), which degrades both insulin and Aβ42, decreases under insulin-resistant conditions, leading to increased brain amyloidosis (Ahmad et al., 2019), and GSK-3β pathway activation by insulin resistance promotes tau phosphorylation (Goncalves et al., 2019). Conversely, higher Aβ decreases insulin receptor expression, worsening insulin resistance and further elevating cortisol.

Hypertension, coronary artery disease, heart failure, and stroke all demonstrate HPA axis dysregulation that accelerates cognitive decline through overlapping mechanisms. In coronary artery disease (CAD), high anticipatory-stress-cortisol predicted CAD severity as determined by angiography, independent of age and sex. Also, the heightened GC sensitivity predicted significantly higher carotid intima-media thickness, suggesting that tissue hypersensitivity to GCs promotes the progression of atherosclerosis (Alevizaki et al., 2007). Heart failure (HF) also induces cognitive impairment. The REGARDS study found that after controlling for stroke, other cardiovascular diseases, and baseline cognitive scores, patients hospitalized for HF had a significantly faster rate of overall cognitive decline than those who had not been hospitalized for HF (Sterling et al., 2024). A quarter of outpatients with chronic heart failure have severe cognitive impairment, primarily in executive and processing speed, that is predicted by heart failure severity and genetic (APOE-ε4) status (Vogels et al., 2007). Stroke triggers acute and immediate HPA axis activation, and cortisol level will remain elevated for days to weeks. Elevated GCs after acute ischemic stroke cause “remote hippocampal neurodegeneration” (damage distant from the infarct site), contributing to post-stroke cognitive decline and dementia risk through cortisol-induced hippocampal atrophy, neuroinflammation, and impaired neurogenesis (Gulyaeva et al., 2021). It is important to note that there exist HPA axis-independent, disease-specific mechanisms, such as impaired insulin signaling in diabetes and obesity, that also play integral roles in AD (Yoon et al., 2023). We merely emphasize the role of the HPA axis in psychiatric and non-psychiatric disorders, as it will be difficult to delineate the independent contribution of these factors by themselves.

2. Stress response genes and AD

So far, we have considered epidemiological, population, and clinical studies that have linked chronic stress or GC exposure to AD pathology. At the molecular level, harmful effects of stress are achieved via genomic or non-genomic mechanisms. Genomic mechanism involves the binding of GR at specific regions in the genome and initiating the transcriptional response of GC-regulated genes. GR can either activate or suppress target gene transcription, and this ability depends on the binding site sequence, which can affect receptor conformation and interaction with other proteins (Weikum et al., 2017). Non-genomic mechanism refers to protein-protein interactions between GC-bound receptor and other proteins in the cytoplasm. The GR has a bidirectional regulatory relationship with signaling molecules of the immune system, and this is sometimes achieved by protein-protein interactions (Silverman and Sternberg, 2012). Here, we consider several exemplary genes that are directly regulated by GC action through genomic mechanisms and play pivotal roles in HPA axis dysregulation, brain function, and AD.

2.1. FK506-binding protein 5 (FKBP5)

There is substantial evidence identifying FKBP5 as a critical molecular link between chronic stress and HPA axis dysregulation and is the primary regulator of intracellular GC signaling besides GR (Wochnik et al., 2005; Scammell et al., 2001). As mentioned before, the FKBP5 gene modulates GR sensitivity and negatively regulates the HPA axis by anchoring GR within an HSP90-based chaperone complex (Grad and Picard, 2007). During periods of prolonged stress or GC exposure, the FKBP5 gene becomes upregulated due to the binding of the GR to multiple GC response elements (GREs) scattered throughout its genomic region, and the increase in FKBP5 protein level suppresses additional GC signaling, leading to GC resistance and extracellular hypercortisolism (Zannas, 2023; Zannas et al., 2016). Interestingly, a single nucleotide polymorphism (SNP), rs1360780, in FKBP5, which is associated with higher GC-induced gene expression, has been linked to cognition (Klengel et al., 2013; Fujii et al., 2014). Furthermore, epigenetic mechanisms, including stress- and age-associated DNA hypomethylation at FKBP5 regulatory regions further potentiate its expression and create a long-lasting pro-inflammatory and stress-sensitized state (Zannas, 2023). In the context of AD, elevated FKBP5 expression has been linked to impaired synaptic plasticity and accumulation of pathogenic tau species (Blair et al., 2013), potentially accelerating neurodegeneration. Experimental models have shown that trauma-like stress combined with Aβ exposure increases FKBP5 expression in the hippocampus and amygdala, exacerbating neuronal apoptosis and cognitive decline (Faborode et al., 2022).

2.2. Glucocorticoid receptor (GR or NR3C1)

A 2022 study reported altered cortisol-binding GR phosphorylation patterns in AD prefrontal cortex. Analyzing 79 post-mortem samples, researchers discovered reduced phosphorylated-GR (p-GR) at BDNF signaling-associated amino acid residue S267, which positively correlated with Mini Mental State Examination (MMSE) scores prior to death. Conversely, p-GR at cortisol pathway-associated residue S226 negatively correlated with MMSE. This demonstrated that GR dysfunction in AD is not due to the loss of function or loss of expression, but altered phosphorylation status, with downregulated BDNF signaling and upregulated cortisol pathway activation (Dromard et al., 2022). GR also functions as a critical negative regulator of the HPA axis by suppressing CRH and ACTH secretion at hypothalamic and pituitary levels. In AD, this autoregulatory mechanism becomes progressively impaired, leading to elevated circulating GC and impairment of GR-mediated negative feedback, confirmed in a clinical study (Elgh et al., 2006), which further correlated with hippocampal atrophy and specific cognitive impairments.

2.3. Heat shock protein 70 (HSP70) family

HSP70 family members are direct regulators of GR function, maintaining GR in a ligand-responsive conformation and controlling its nuclear translocation, and HSP70 expression is modulated by GC exposure (Wadekar et al., 2004). Although there are no genetic variants of the HSP70 family found with a major effect on AD, nominal associations in prefoldin families suggest that polygenic small-effect variation in protein homeostasis pathways might contribute to AD risk, consistent with models of AD as a complex, multifactorial disease involving protein misfolding and aggregation (Broer et al., 2011). A gene expression study revealed complex patterns of “hub genes” within the HSP70 family that are immune-related and differentially expressed in AD. In the immune-related HSP70 family genes, HSPA1A & HSPA2 were significantly upregulated in AD and HSPA8 was downregulated, in which HSPA1A showed a negative correlation with APP and APOE. HSPA2 showed a positive correlation with PSEN1, whereas HSPA8 was negatively correlated with β- and γ-secretase activity and with APP. This differential regulation of HSP70 family genes in AD possibly reflects compensatory stress responses (HSPA1A/HSPA2) or failure of chaperone systems (HSPA8), which may affect proteostasis in AD (Dong et al., 2022).

2.4. Brain-derived neurotrophic factor (BDNF)

BDNF is a neurotrophic factor that is essential for neuronal survival, differentiation, and synaptic plasticity (Kowianski et al., 2018). It is abundantly expressed in the hippocampus, cortex, and basal forebrain, all of which are involved in memory and learning processes (Egan et al., 2003). Animal and human studies have shown that chronic elevation of GCs reduces BDNF expression through a GR-mediated mechanism, thereby impairing neurogenesis and long-term potentiation (LTP), which are essential for cognitive resilience (Egan et al., 2003; Alboni et al., 2011; Suri and Vaidya, 2013). AD is also associated with reduced BDNF levels, demonstrated in a comprehensive meta-analysis of 15 studies involving more than 2000 participants, and reduced BDNF levels were associated with decreased MMSE scores and increased age (Ng et al., 2019). This meta-analysis also found that individuals with MCI did not exhibit significantly lower BDNF concentrations than controls, implying that the decline in BDNF may occur later in the disease continuum. Mechanistically, sustained stress and high cortisol levels inhibit the phosphorylation of CREB, a cAMP-responsive metaprotein and a key transcription factor that regulates BDNF transcription, reducing its activity and thereby reducing BDNF expression. BDNF is also a direct target of activated GR (Chen et al., 2017). Stress-induced BDNF reduction, by CREB or GR, can exacerbate tau hyperphosphorylation and Aβ toxicity, creating a feed-forward loop that accelerates neurodegeneration. BDNF signals through its receptor TrkB to activate the PI3K-Akt pathway, which negatively regulates GSK-3β. BDNF deficiency leads to the disinhibition of GSK-3β, and impairment of synaptic plasticity, making neurons more susceptible to Aβ oligomer-induced toxicity (DaRocha-Souto et al., 2012; Mai et al., 2002). Aβ itself has been shown to inhibit BDNF expression, further impairing neurotrophic support and synaptic plasticity, especially in hippocampal circuits essential for memory formation (Garzon and Fahnestock, 2007). Evidence from a community-based cohort study also suggests that higher peripheral BDNF levels are protective, as Weinstein et al. reported that for every standard deviation increase in serum BDNF levels, the risk of dementia was reduced by 33% (Weinstein et al., 2014).

2.5. Dopamine signaling, tyrosine hydroxylase, and MAOA

Dopaminergic dysfunction is also closely related to stress-induced cognitive decline and the pathogenesis of AD (Storga et al., 1996). Tyrosine hydroxylase (TH) is a rate-limiting enzyme that converts tyrosine to levodopa (L-DOPA), the precursor to dopamine. In dementia patients, TH levels are reduced in the hippocampus and cortical regions (Marui et al., 2003), impairing the synthesis of dopamine (DA), thereby impairing synaptic plasticity, attention, and memory consolidation (Jay, 2003). Importantly, stress and elevated GC levels exacerbate this dopaminergic dysfunction. A study by Núñez and colleagues showed that although stress-induced corticosterone (CORT) release can increase TH mRNA expression and phosphorylation in specific brain regions such as the nucleus tractus solitarius (NTS-A2) by activating the HPA axis (Nunez et al., 2009), this upregulation appears to be transient and region-specific. Over time, chronic stress and GC exposure impair TH phosphorylation at key regulatory sites (e.g., Ser31, Ser40), reducing its catalytic efficiency and reducing overall dopamine production in key regions (e.g., prefrontal cortex and hippocampus). In addition, the TH locus is also a genomic target of GR signaling, as a study linked stress-induced increase in TH methylation and decrease in mRNA as a mediating factor for the effect of chronic stress on behavior (Niwa et al., 2013).

Monoamine oxidase A (MAOA) is a mitochondrial enzyme involved in the oxidative deamination of monoamine neurotransmitters such as dopamine, serotonin, and norepinephrine and plays an important role in stress-induced AD-related neurodegeneration. Chronic stress and GC exposure lead to increased MAOA activity, resulting in excessive neurotransmitter catabolism, significantly reducing the synaptic availability of these neuromodulators that are critical for mood regulation, cognition, and neuroplasticity (Jones and Raghanti, 2021). From a mechanistic perspective, increased MAOA activity also leads to the formation of reactive oxygen species (ROS), especially hydrogen peroxide, through the oxidative deamination of monoamines (Bortolato et al., 2008). The accumulation of ROS directly leads to neuronal damage and apoptosis. This oxidative stress cascade impairs neuronal integrity and promotes the accumulation of Aβ and hyperphosphorylation of tau (Riederer et al., 2004). Furthermore, increased MAOA activity can disrupt mitochondrial function, further contributing to neuronal vulnerability and functional decline in AD patients. A 2021 study by Jones and Raghanti demonstrated that chronic stress and elevated GC levels further amplify MAOA expression through transcriptional regulation, which may be mediated by a GRE within the regulatory region of the MAOA gene (Jones and Raghanti, 2021). Increased MAOA activity promotes inflammation and neuroinflammatory responses, which exacerbates the progression of AD neuropathology (Jones and Raghanti, 2021). These genes play critical roles in neurotransmitter function, are controlled by GC signaling, and are dysregulated in AD.

2.6. Corticotropin-releasing hormone (CRH)

CRH released from the hypothalamus in response to stress acts as both hormone and neuromodulator to counteract stressors. Research on CRH and its association with AD began in the mid-1980s and has established a consistent pattern. Post-mortem AD brain studies beginning in 1985 document a 50–70% CRH reduction in frontal, temporal, and striatal regions, accompanied by compensatory CRH receptor 1 (CRHR1) upregulation in affected cortical areas (De Souza et al., 1986). And lower CRH concentrations in cerebrospinal fluid are associated with the severity of cognitive impairment (May et al., 1987). This historical study implicates CRH’s neuroprotective role, which has been confirmed in a more recent study showing CRH protected neurons from oxidative stress-induced cell death through CRHR1 activation and PKA signaling (Pedersen et al., 2002; Bayatti et al., 2005). CRH also promoted sAPPα and non-amyloidogenic APP processing, shifting away from pathological amyloidogenic processing (Lezoualc’h et al., 2000). CRHR1 genetic variants shed mechanistic insight on how stress induces cognitive dysfunction and cellular aging, and can potentially serve as a biomarker for early intervention, identifying vulnerable populations before clinical AD symptoms emerge. A 2014 study of 466 subjects across the adult age span examined SNPs rs110402 and rs242924 in CRHR1 and discovered that GG-homozygotes showed severe working memory dysfunction onset during middle age and identified a genetically based subtype of premature cognitive impairment attributed to altered hippocampal CRHR1 activation (Grimm et al., 2015). A more recent meta-analysis combining 562,668 individuals identified CRHR1 as showing a strong regulatory function (common pathway), genome-wide significance, and shared genetic architecture between AD and frailty through colocalization analysis (Enduru et al., 2024).

3. AD Genes that are relevant for stress

Increasing evidence shows that chronic stress and GC dysregulation can directly modulate key AD-related genes. While human mechanistic studies remain limited by ethical and practical constraints, evidence from human cell culture, clinical biomarker studies, and translational models reveal that stress hormones can regulate genes that have already been implicated in AD pathogenesis, such as those involved in the formation of Aβ plaques and NFTs. Together, these studies further establish that chronic stress exposure may contribute to AD molecular pathology.

3.1. Effect of stress on Aβ and NFTs

GCs can mediate increases in BACE1, PSEN1, or γ-secretase activity, which enhance Aβ production. For instance, a study has shown that cortisol administration in the human neuroblastoma SK-N-MC cell line can increase BACE1 expression via membrane-associated glucocorticoid receptors (mGR) coupled with Gαs-induced (G protein αs subunit) activation of the cAMP-PKA-CREB signaling cascade, leading to CREB binding to the cAMP CRE site in the BACE1 promoter (Choi et al., 2017). Another study suggested that GCs exert more direct transcriptional regulatory effects, subsequently leading to elevated protein levels and enhanced amyloidogenic processing (Zhang et al., 2020). Non-transcriptional mechanisms reinforce this pathway. In human SH-SY5Y neuronal cells, GCs enhance Aβ production at the mitochondria-associated membrane (MAM) of ER by enhanced γ-secretase activity in MAM fractions, which then promotes PSEN1 localization to MAM and driving mitochondrial accumulation of Aβ before substantial extracellular Aβ deposition (Choi et al., 2023). Collectively, stress or GCs can wield genomic and non-genomic effects on secretase protein complexes and Aβ production. Human studies, while often lacking mechanistic insight, have shown significant relationships between cortisol vs. Aβ-and NFT-related biomarkers. Cortisol levels were positively correlated with increased amyloid deposition in post-menopausal women (Salardini et al., 2025). Higher awakening cortisol levels were also associated with lower CSF Aβ₄₂ levels indicative of greater Aβ₄₂-induced plaque burden, while bedtime cortisol levels were positively associated with higher CSF tau (Holleman et al., 2022). However, another clinical study reported that cortisol levels are negatively associated with CSF total tau (T-tau), phosphorylated tau, and T-tau/Aβ₄₂ ratio in patients with mild AD dementia, suggesting a possible neuroprotective effect of cortisol (Laske et al., 2009). Limited clinical findings warrant additional studies using larger cohort sizes to provide a clearer clinical insight on the mechanistic link between stress/cortisol exposure and AD biomarkers.

3.2. Apolipoprotein (APOE)

As previously mentioned, APOE-ε4 allele is the strongest risk factor for AD, with those carrying one copy having a 3–4 likelihood of developing AD and those with two copies having an 8–12 times the likelihood (Farrer et al., 1997; Yamazaki et al., 2019). APOE shows both direct transcriptional regulation by GCs and reciprocal effects of APOE genotype on cortisol levels. A study identified a functional GRE at the human APOE promoter and demonstrated that GC effects at the APOE locus appear to be cell-type specific (Trusca et al., 2017). Dexamethasone can induce ~2–6-fold increase in APOE mRNA and protein level in peritoneal macrophages via transcriptional activation of its promoter (Trusca et al., 2017), while treatment with the GC prednisone reduced APOE mRNA levels in SLE (systemic lupus erythematosus) patients (Song et al., 2013). It remains unclear if and how GCs regulate APOE expression in neuronal and glial cells. Interestingly, clinical evidence demonstrates that APOE genotype modifies HPA axis activity, by showing that CSF cortisol follows a gene-dose pattern, where higher CSF cortisol concentrations in AD patients were associated with increased frequency of the APOE-ε4 allele and decreased frequency of the APOE-ε2 allele (Peskind et al., 2001). In human frontal cortex tissue, increased cortisol levels were associated with APOE-ε4 genotype, negatively correlated with cognitive test scores, and positively correlated with frontal cortex Aβ levels (Gil-Bea et al., 2010; Duarte-Zambrano et al., 2023). Together, these studies suggest reciprocal associations between GC signaling and APOE-associated AD, where GCs can regulate APOE expression in some cellular contexts, while APOE genotype may influence HPA axis activity in AD. However, the directionality, cell-type specificity, and relevance of this interaction in humans remain incompletely defined.

4. GC resistance and immune dysregulation under chronic stress

Cortisol and related GCs are fundamental in suppressing immune and inflammatory responses during stress, ensuring homeostatic balance by inhibiting pro-inflammatory cytokines (e.g., IL-1, TNF-α) and promoting anti-inflammatory mediators like IL-10 (Silverman and Sternberg, 2012). However, persistent stress can lead immune cells to become progressively resistant to GCs, which is known as GC resistance (GC-R). Over time, immune cells downregulate or desensitize GRs, impairing their response to cortisol’s regulatory effects. In individuals under chronic stress, this results in a “paradoxical” situation of high circulating GC levels and elevated pro-inflammatory cytokines, as GR-mediated negative feedback on immune activation becomes ineffective (Yang et al., 2012).

This GC-R phenotype sustains low-grade inflammation linked to depression, PTSD, and anxiety, because elevated cortisol cannot suppress inflammatory signaling (Tong et al., 2023). Moreover, the chronic inflammatory state associated with GC-R is becoming increasingly recognized as a risk factor for neurodegenerative diseases, by promoting neuroinflammation and synaptic dysfunction. The immune system’s “resistance” to stress hormones creates a vicious cycle: stress induces immunosuppression, with chronic stress exposure leading to GC-R, which leads to uncontrolled inflammation, which then increases the risk of psychiatric disorders and neurodegeneration (Perrin et al., 2019). At the same time, in the AD brain, where microglia are already exposed to Aβ, tau, lipid dyshomeostasis, and neuronal injury signals, this state could lower the threshold for sustained inflammatory activation.

4.1. Immune dysregulation and inflammation in AD

Stress and AD are clinically associated with immune dysregulation. Meta-analytic evidence from 17,245 participants demonstrates moderate-to-large cortisol elevations across blood, saliva, and CSF compartments in AD patients (Zheng et al., 2020). Prospective data also show that a high urinary free cortisol (UFC) to creatinine (Cr) ratio predicts a 1.31–1.38-fold increased AD risk years before clinical onset (Ennis et al., 2017). Sustained HPA axis activation elevates cortisol in AD patients, showing 83% higher cortisol level that primes microglia toward hyperreactive states through GR-mediated NF-κB nuclear translocation, which induces a pro-inflammatory phenotype (Piirainen et al., 2017). The release of cytokines IL-1β, IL-6, and TNF-α activates parallel kinase cascades in p38 MAPK and GSK-3β that directly hyperphosphorylates tau at multiple pathologically relevant sites. An inflammasome, a cytosolic complex containing the protein NLRP3 (NLR family pyrin domain-containing protein 3) associated with microglial activation, serves as an amplification point, where Aβ oligomers trigger caspase-1-mediated IL-1β maturation, while complement cascade activation drives CR3-mediated synaptic elimination by downregulated homeostatic markers (CX3CR1, P2RY12) and upregulated phagocytic receptors (Chen and Yu, 2023). This integrated stress-inflammation-neurodegeneration process creates multiple positive feedback loops wherein stress-induced cytokines enhance Aβ production while impairing clearance, hyperphosphorylated tau activates additional microglial TLRs, and activated inflammasomes nucleate new Aβ aggregates, supporting a model in which chronic stress may amplify neuroimmune processes that interact with Aβ, tau, and synaptic injury.

5. A summary of similarities between stress response and AD

The literature suggests a significant mechanistic overlap between the pathophysiology of AD and neurobiological alterations induced by chronic stress, implicating shared susceptibility pathways that increase AD risk and promote its progression. As shown in the table below, both AD and chronic stress exposure exhibit HPA axis dysfunction, characterized by persistent hypercortisolism and impaired GC feedback mechanisms, i.e., GC-R. Chronic stress and AD also overlap in several pathophysiological domains, including hippocampus-dependent cognitive deficits, cardiometabolic dysfunction, and neuroinflammatory cascades involving microglial activation and cytokine signaling. Both conditions demonstrate inadequate neurotrophic support, such as inhibition of BDNF and synaptic dysfunction, manifested as reduced neurogenesis and impaired LTP, respectively. Parallel dysregulation of neurotransmitter systems, such as upregulation of MAOA in both cases, further underscores the co-occurring role of monoaminergic dysfunction, which can contribute to oxidative stress. Furthermore, the transcriptional regulation of AD-related genes (APP, BACE1, and PSEN1) and stress-response genes in both cases, along with the combined promotion of tau protein hyperphosphorylation and Aβ accumulation, suggests that chronic stress may contribute to AD vulnerability or progression by engaging overlapping molecular and cellular pathways. The Table 1 below lists many of the similarities observed between chronic stress exposure and AD.

Table 1.

Characteristic similarities between AD and chronic stress exposure.

Disease Characteristics	AD	Chronic Stress
Aging	Strongest risk factor	Cumulative across age
Sex difference	Higher prevalence in females	Higher response in females
Cognitive deficits	Progressive impairment	Yes (hippocampal atrophy, memory deficits)
Psychiatric disorders	Comorbid	Primary risk factor
Cardiovascular dysfunction	Yes (e.g., APOE-ε4)	Yes (hypertension, obesity, diabetes)
HPA axis dysfunction	Yes (e.g., CRH ↓, cortisol ↑)	Yes (sustained activation of cortisol ↑)
Immune activation/neuroinflammation	Yes (microglia, cytokines)	Yes (GC resistance, inflammation, and microglial priming)
Neurotrophic factor deficiency	BDNF ↓, impaired synaptic plasticity	GC-mediated BDNF suppression
Synaptic dysfunction	Synaptic loss, impaired LTP	Impaired LTP, reduced neurogenesis
AD-related gene/pathway regulation	APP/BACE1/PSEN1 (amyloidogenic processing); MAPT (tau); APOE (lipid metabolism, Aβ clearance, and neuroimmune responses)	Stress-mediated transcription of APP-processing genes, and APOE- and MAPT-related pathways
Tau and Aβ pathology	NFTs, Aβ accumulation	NFTs, Aβ accumulation
Neurotransmitter dysregulation	MAOA ↑, altered monoaminergic, cholinergic, and noradrenergic signaling	MAOA ↑, TH dysfunction, altered monoamine availability
Oxidative/metabolic stress	ROS production, lipid peroxidation, mitochondrial dysfunction	ROS production, mitochondrial stress, insulin resistance, inflammatory oxidative injury

6. Animal models of stress and cognitive impairment

Studies of humans, particularly those with Cushing’s syndrome, have highlighted the neurotoxic effects of cortisol on learning and memory. Except for biomarker or imaging studies, studies of AD molecular pathology in humans are cross-sectional in nature due to their dependence on postmortem brain tissues. To properly understand the interaction between chronic stress and AD, researchers have increasingly relied on animal models that recapitulate hallmark pathologies of AD, such as Aβ accumulation, tau hyperphosphorylation, synaptic dysfunction, and cognitive impairment. Benefits of using animal models of stress or AD are that they provide readily accessible brain tissues and can establish controlled investigations and experimental interventions into causality and gene-environment interactions. However, they are not without shortfalls, such as the limited ability to make inferences about human AD from animal models in terms of AD-related behavior, molecular pathology, and genetics. These models vary widely in species, genetic modifications, and relevance to human AD, allowing researchers to investigate different facets of the disease from spontaneous onset to engineered pathology. This section summarizes animal models used to examine how experimentally imposed stress or GC exposure modifies AD-like phenotypes.

When it comes to AD-related animal models of Aβ, non-human primates such as the rhesus macaque (Macaca mulatta) and cynomolgus monkey (Macaca fascicularis) represent some of the most translationally relevant systems. Aged rhesus macaques naturally exhibit Aβ plaque formation, cognitive deficits, and age-related decline, and are similar to human AD populations, with APOE-ε4 expression present in about 20% of senile rhesus macaques (Bu, 2009; Uno and Walker, 1993; Voytko, 1999). Likewise, M. fascicularis shows plaque deposition in the superior and inferior gyri of the temporal cortex and amygdala by age 20. In comparison to other non-human primates, cynomolgus monkeys have a longer lifespan, typically exceeding 35 years of age (Nakamura et al., 1998). These models offer key advantages in mirroring the gradual, spontaneous progression of pathology seen in humans.

Canine models of cognitive dysfunction have also emerged as valuable tools for studying naturally occurring AD-like symptoms. Senior-aged dogs develop Aβ plaques composed of the same amino acid sequence found in human Aβ, and the severity of deposition correlates with cognitive deficits in both species (Braidy et al., 2015; Chambers et al., 2012). Specific Aβ subspecies, such as the highly neurotoxic pyroglutamyl Aβ (pE3Aβ), have been identified in the hippocampus of affected dogs (Schutt et al., 2016; Schmidt et al., 2015), further supporting their translational relevance.

Among rodent models, a range of transgenic mice have been developed to express human genes associated with familial AD. The PDAPP mouse, carrying the Indiana APP mutation at the γ-cleavage site of AβPP, develops amyloid plaques at 6–9 months of age (Games et al., 1995). Tg2576 and APP23 mice carry a Swedish double mutation at the β-cleavage site and show memory deficits and plaque formation at ages varying from 3 to 11 months (Hsiao et al., 1996; Sturchler-Pierrat et al., 1997). The TgCRND8 model includes both AβPP Swedish and Indiana mutations, leading to an earlier onset of AD pathology when compared to single-mutation mice (Chishti et al., 2001). These first-generation single- and double-mutation amyloid models effectively reproduce amyloid pathology but generally lack robust tau tangles and exhibit only limited overt neurodegeneration, constraining their capacity to fully recapitulate human AD. Later multi-mutation models such as 5xFAD partially overcome this limitation, as discussed below, producing synaptic and neuronal loss in the subiculum, cortex, and hippocampus by 4–6 months of age.

There are also several animal models that have investigated taupathies. For instance, transgenic mice that express the human tau gene with P301L or P301S mutations produce tau proteins that form NFTs and show brain atrophy and cognitive deficits (Ramsden et al., 2005; Dumont et al., 2011). The THY-Tau22 mouse model develops taupathies and cognitive impairment without the motor dysfunction observed in other models (Schindowski et al., 2006). Knock-in models that replace the mouse Mapt gene with mutated forms of human MAPT, such as the FTDP-17 triple mutant knock-in model, show similar taupathies as transgenic models (Umeda et al., 2013). Other studies have examined prion-like tau propagation in mice by injecting and seeding their brains with pathological tau and documenting its spread and consequent effect on cognition (Clavaguera et al., 2009; Houze et al., 2025). An interesting study has used synthetic tau-like proteins to induce taupathy without the need for mutations or tau overexpression (Guo and Lee, 2013). Similar studies have been performed in several classic animal models such as Drosophila, zebrafish, C.elegans, rats, and non-human primate species (Li and Le, 2013; Bilen and Bonini, 2005; Paquet et al., 2010; Langer Horvat et al., 2023; Beckman et al., 2021). Despite the intriguing observation that tau proteins can be seeded, propagate in the brain like prions, and cause neuronal dysfunction, many studies failed to perfectly recapitulate human taupathy. Logically, investigators have also constructed animal models combining both Aβ and tau pathologies. The commonly used 3xTg-AD mouse model carries two Aβ- and one tau-related mutations, and these animals develop both amyloid plaques and NFTs (Oddo et al., 2003). The APP/MAPT knock-in mice that carry these human genes show increased propagation of tau, implicating an interaction between these two genes (Huang et al., 2022).

Although these models have been invaluable in studying amyloid-driven or tau-related pathologies, their application to the intersection of chronic stress and AD remains a small and emerging area. A subset of these models has nonetheless been specifically utilized to examine how chronic stress and GC exposure influence AD-related outcomes, offering valuable early insights into gene–environment interactions; however, considerably more work is needed to systematically characterize how stress modifies AD pathogenesis across genetic backgrounds, sexes, and developmental stages and the molecular pathways through which these factors affect learning and memory.

Several transgenic mouse models have been used specifically to explore how chronic stress and GC exposure interact with AD pathology. The 3xTg-AD mouse, which includes human mutations AβPPSwe, PS1M146V (PSEN1 mutation), and tauP301L, develops both amyloid plaques and NFTs. This model is widely used to investigate the interplay between amyloid and tau pathology (Oddo et al., 2003). Importantly, studies using the 3xTg-AD mice exposed to GCs through DEX administration show increases Aβ and tau accumulation, accelerating synaptic dysfunction and cognitive deficits (Green et al., 2006). The 5xFAD model, which is one of the most aggressive and well-characterized transgenic models of AD, incorporates five familial AD mutations: the Swedish (K670N/M671L), Florida (I716V), and London (V717I) mutations in the APP gene, along with M146L and L286V mutations in the PSEN1 gene. These mice produce increased level of Aβ42, the most aggregation-prone form of the amyloid peptide, and exhibit intraneuronal Aβ accumulation as early as 1.5 months of age, with dense plaque formation, gliosis, and robust neurodegeneration by 4–6 months. Behaviorally, 5xFAD mice demonstrate progressive deficits in memory and learning, along with synaptic and neuronal loss - particularly in the subiculum, cortex, and hippocampus. Due to the early onset and severity of pathology, this model is widely used to evaluate experimental therapies and study early disease mechanisms. Recently, this model demonstrated that chronic stress accelerates cognitive decline, possibly via enhanced Aβ deposition, increased neuroinflammation, and downregulation of the BDNF-TrkB signaling pathway (Oakley et al., 2006; Shlomi-Loubaton et al., 2025). Similarly, APOE transgenic models, particularly those expressing the human APOE-ε4 allele, have demonstrated increased vulnerability to stress-induced AD pathology. These models show enhanced APP expression, altered Aβ clearance, and greater tau cleavage, which together contribute to synaptic dysfunction and cognitive decline (Wang et al., 2022; Holtzman et al., 1999). In addition, chronic stress in APOE-ε4 carriers has been linked to increased neuroinflammatory markers and behavioral disturbances, reinforcing the idea that APOE-ε4 not only confers genetic risk but also amplifies the brain’s sensitivity to environmental stress (Parhizkar and Holtzman, 2022).

Bigenic mouse models, such as the PS/APP line—which combines mutant human APP and PSEN1 genes—develop amyloid pathology at an earlier age and in a more aggressive manner than single transgenic models (Holcomb et al., 1998). These models have been used to assess the long-term effects of early life stress, revealing that stress exposure during critical developmental periods increases Aβ accumulation and worsens cognitive performance. Notably, these effects appear reversible: treatment with GR antagonists has been shown to prevent stress-induced pathology, suggesting that stress hormone signaling may play a direct role in AD progression (Lesuis et al., 2018).

In rat models, the McGill-R-Thy1-APP and TgF344-AD lines are notable for their relevance to stress-based AD research. The McGill model, which expresses APP751 with the Swedish and Indiana mutations, shows early intracellular Aβ accumulation and has been used to demonstrate that GC exposure disrupts hippocampal synaptic plasticity (Leon et al., 2010; Qi et al., 2021). The TgF344-AD rat expresses both mutant human APP and PSEN1, developing progressive amyloid and tau pathology alongside behavioral phenotypes resembling anxiety and fear dysregulation, particularly when exposed to chronic stress (Cohen et al., 2013; Pentkowski et al., 2022). These models are especially valuable due to rats’ larger brains and greater neuroanatomical similarity to humans, which enable more nuanced behavioral testing and pharmacological manipulation than mouse models (Jacob and Kwitek, 2002).

Finally, zebrafish show promise for future studies of the intersection of chronic stress and AD. Zebrafish share approximately 70% of their protein-coding genes with humans, and 84% of human disease-related genes have zebrafish orthologs, including APP, PSEN1, PSEN2, and MAPT (Howe et al., 2013; Groth et al., 2002). They exhibit several AD-like behavioral phenotypes such as avoidance, memory deficits, and impaired habituation to startle, making them useful for drug testing and cognitive assessment (Blaser and Gerlai, 2006; Best et al., 2008). Aβ injection in zebrafish has been shown to impair cognition and increase tau phosphorylation, effects that can be pharmacologically reversed (Nery et al., 2014). Although zebrafish have been used to study AD and stress independently, they have not yet been widely used to examine both in tandem. However, emerging research suggests they may be a valuable model for exploring gut-brain interactions, most specifically in how tryptophan metabolism is linked to AD and dementia (Demin et al., 2021; Kalueff et al., 2014). This connection is mechanistically relevant to the stress–AD intersection because chronic stress and elevated GCs induce the tryptophan-catabolizing enzymes indoleamine 2,3-dioxygenase (IDO) via pro-inflammatory cytokines and tryptophan 2,3-dioxygenase (TDO) (Moffett and Namboodiri, 2003), shunting tryptophan away from serotonin synthesis and toward the kynurenine pathway (Gibney et al., 2014). The resulting neuroactive metabolites, including the NMDA agonist quinolinic acid and the free-radical generator 3-hydroxykynurenine, contribute to neuroinflammation, excitotoxicity, and oxidative damage implicated in AD pathogenesis, while reduced synthesis of the NMDA antagonist kynurenic acid may further tip the balance toward neurodegeneration (Mithaiwala et al., 2021).

7. Epigenetic mechanisms mediating the effect of stress on AD

Chronic stress and GC exposure can produce durable changes in gene regulation that persist beyond the initial hormonal stimulus. Although acute GR activation produces transient transcriptional responses, repeated or prolonged exposure may cause epigenetic changes or DNA sequence-independent mechanisms that regulate gene activity, including DNA methylation (DNAm), histone modifications/chromatin accessibility, and non-coding RNA expression, thereby creating a longer-lasting molecular “memory” of stress exposure (Cox et al., 2021; Martinez-Feduchi et al., 2024). In the context of chronic stress, these mechanisms are important because they can encode persistent or cumulative effects of GC exposure on gene expression and provide a plausible link between chronic HPA axis activation, persistent changes in stress-responsive genes, and AD-relevant processes including neuroinflammation, synaptic dysfunction, Aβ accumulation, and tau pathology (Thomassin et al., 2001).

DNAm has been the most extensively studied stress-responsive epigenetic mark because it is relatively stable and measurable in archived tissues. GC-associated DNAm changes can persist across developmental periods, accumulate with repeated exposure, and alter future transcriptional responsiveness by changing GR access to regulatory elements (Niwa et al., 2013; Lee et al., 2010; Lee et al., 2011). Thus, DNAm can help explain how early-life or chronic stress produces long-term changes in genes involved in HPA axis regulation, neuronal function, and AD vulnerability.

GC-sensitive epigenetic gene regulation can be grouped into two broad categories. The first group includes genes that regulate HPA axis function or intracellular GC sensitivity, such as FKBP5, NR3C1/GR, CRH, and POMC (Westberry et al., 2006). The second includes genes involved in brain function and AD-relevant pathways, including neurotransmission, synaptic plasticity, neurodevelopment, amyloid processing, and neuroinflammation. Examples include TH, MAOA, GAD1, BDNF, and BACE1 (Makino et al., 2002; Xu et al., 2020; Zhang et al., 2010; Tsankova et al., 2006; Cordner and Tamashiro, 2016). The epigenetic regulation of these genes provides a mechanistic bridge between chronic stress exposure, altered GC signaling, and AD-relevant molecular pathology.

Epigenetic clocks offer another framework for linking stress, aging, and AD, because they estimate biological age from DNAm patterns (Conole et al., 2025). Since aging is the strongest risk factor for AD, accelerated epigenetic aging may represent a biomarker of reduced biological resilience rather than a direct causal mechanism (Bonham et al., 2025). Recent work has begun to distinguish correlative clocks from approaches that prioritize CpGs with stronger causal evidence, but the relevance of these measures to brain-specific AD mechanisms remains an active area of investigation (Ying et al., 2024).

Altogether, epigenetic changes, whether caused by stress or the aging process, can play causative roles in the development of AD symptoms. However, it is still unclear whether stress and aging cumulatively erode normal brain function and resilience in what is often referred to as cognitive reserve (Pettigrew and Soldan, 2019), which then predisposes the brain to AD, or they directly affect the expression of genes that govern AD-related pathways. The former concept ascribes a more preexisting, etiological role to stress, whereas the latter describes stress and AD as parallel processes where stress can exacerbate AD symptoms through convergent pathways. While our work has focused more on the latter concept, the literature does not preclude the former. Given the significant overlap between stress and cognition, it is likely that both processes are at play, with epigenetic mechanisms facilitating the impact of stress and time on cognitive function (Liu et al., 2025b).

7.1. Proposed model linking stress/GC exposure to AD-related pathology

When taken together, the evidence reviewed in this article supports a mechanistic model in which chronic stress activates the HPA axis and increases GC exposure levels; over time, this leads to dysregulation in GC signaling and HPA axis function. These changes may be consolidated or amplified through epigenetic mechanisms as described above, thereby altering the expression of genes associated with stress response, neuronal function, neuroinflammation, and AD. Concurrently, this stress-mediated epigenetic alteration may promote neuroimmune activation, microglial inflammatory signaling, and cytokine-driven pathways that ultimately converge to drive the accumulation of Aβ, tau hyperphosphorylation, synaptic dysfunction, and neurodegeneration. This theoretical framework, which links chronic stress, HPA axis dysfunction, epigenetic regulation, and neuroinflammation to AD-related pathology, is summarized in Fig. 1. However, it is important to note that the figure depicts one possible mechanism of how an environmental factor such as stress or GC exposure can contribute to AD symptoms.

Fig. 1.

Proposed Model Linking Stress/GC Dysregulation to AD-related Pathology.

8. Translational and clinical implications

A major translational question is whether reducing chronic stress or normalizing GC signaling can significantly alter AD risk in humans. At present, no randomized clinical trial has demonstrated that stress reduction prevents AD onset or slows AD progression. Nevertheless, stress-reduction interventions, such as cognitive-behavioral approaches, exercise, sleep improvement, and mindfulness-based interventions, may improve intermediate outcomes such as cortisol regulation (Rogerson et al., 2024), cardiovascular risk (El-Malahi et al., 2024), and inflammatory burden (Rosenkranz et al., 2016). These outcomes are clinically meaningful, but they should not yet be equated with AD prevention.

GC signaling pathways also represent potential therapeutic targets, but translation remains challenging. GR antagonists, FKBP5 modulators, and interventions targeting HPA axis feedback could theoretically reduce GC resistance, neuroinflammation, or stress-amplified amyloid and tau pathology. However, GCs also have essential homeostatic and anti-inflammatory functions, so the broad suppression of GC signaling may disrupt neuroimmune balance. Future therapeutic strategies will need to define disease stage, exposure timing, sex-specific effects, and cell-type specificity before GC-related pathways can be safely targeted in AD.

Cortisol and related HPA axis measures may also have biomarker value, but they are not currently validated as stand-alone predictors of AD. Cortisol varies by circadian rhythm, acute stress, sleep, medications, metabolic disease, psychiatric disorders, and sex hormone status. Therefore, future biomarker studies should integrate repeated cortisol measurements with AD biomarkers such as amyloid PET, tau PET, CSF or plasma phosphorylated tau, inflammatory markers, and cognitive assessments. Such designs would help determine whether HPA axis dysregulation is a marker of vulnerability, a consequence of prodromal disease, or a modifiable contributor to progression.

9. Limitations

Considering the evidence reviewed above, many human studies cited in this review can demonstrate robust associations between chronic stress exposure, psychiatric comorbidities, HPA axis dysregulation, and subsequent cognitive decline, but they do not establish causation. Residual confounding (e.g., shared genetic, socioeconomic, vascular, or lifestyle risk factors), reverse causation (prodromal AD altering stress perception and HPA axis function), and selection effects in registry- and volunteer-based cohorts all remain plausible alternative explanations. To our knowledge, no randomized controlled trial has demonstrated that reducing chronic stress or lowering cortisol exposure in humans prevents AD, and such trials are ethically and practically difficult to conduct.

Moreover, the mechanistic evidence linking GCs to AD-relevant molecular pathology is derived predominantly from transgenic rodent models and immortalized cell lines. These systems are essential for dissecting molecular pathways under controlled conditions, but they have translational limitations: (1) most rodent AD models rely on familial AD mutations and reproduce only a subset of human AD pathology, largely failing to capture the age-dependent, sporadic course of the majority of clinical disease; (2) species differences in HPA axis physiology, such as cortisol versus corticosterone, complicate direct translation; (3) experimental GC doses and exposure durations often exceed physiologically relevant human exposures; and (4) stress paradigms used across studies are highly heterogeneous and not standardized and might result in different patterns of HPA axis activation.

Taken together, the body of evidence reviewed here is best interpreted as supporting a potential association between chronic stress and AD. Future work integrating prospective human biomarker studies, interventional trials targeting HPA axis dysregulation, and more translationally rigorous animal models will be needed to move from association and mechanistic plausibility toward a rigorously established causal contribution of stress to AD.

10. Conclusion

Alzheimer’s disease is a debilitating neurodegenerative disorder with a wide range of risk factors, such as aging, genetics, cardiovascular disease, lifestyle, and environmental factors. One of the potent environmental factors is stress. Chronic exposure to stress or GCs contributes to a host of disease conditions. One of the more vulnerable organs is the brain, as GCs affect a multitude of processes involved in neurotransmission, mood, and cognition. In this review, we summarized evidence that in clinical studies, chronic GC or stress exposure is linked to cognitive impairment and can potentially contribute to AD symptoms. Animal stress paradigms further support this relationship by providing behavioral and molecular evidence of the impact of stress on cognition. Mechanisms by which stress may contribute to cognitive impairment are through shared molecular and cellular pathways between AD and stress exposure mediated by epigenetics. The ability of GCs and stress to cause persistent, dose-dependent, and cumulative epigenetic changes, which in turn can influence gene activity can explain how their chronic effects can erode gene function over time. These changes occur at genes that play critical roles in GC signaling, neuronal function, and AD. As the impact of wear and tear by stress is persistent and cumulative, individuals become more susceptible to AD as they age. Our goal was not to supplant currently existing models of AD etiology and pathology, but merely highlight the importance of chronic GC or stress exposure in AD.

Data availability

No data was used for the research described in the article.