Tests for both the diagnosis and for the determination of the development and effectiveness of drugs are of critical importance for treating AD as well as important aids in determining the effectiveness of drugs. However, there is no single definitive test for the diagnosis of AD, and a combination of diagnostic tools, medical history review, cognitive and functional assessments, as well as brain imaging, cerebrospinal fluid analysis, and blood tests are collectively employed (Schachter and Davis, 2000; Tsoi et al., 2015). The main cognitive and functional tests employed include the Mini-Mental State Examination (MMSE); the Montreal Cognitive Assessment (MoCA); and the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog); Addenbrooke’s Cognitive Examination (ACE) A summary of the advantages and limitations of these tests as well key references is provided in Supplementary Table S1. The ADAS-Cog test is designed to specifically measure the cognitive performance of subjects who already carry a diagnosis of AD and is a widely used primary outcome measure in both clinical research and drug trials to track the progression of moderate to severe AD (Rosen et al., 1984). However, ADAS-Cog is not a suitable assessment tool for broader screening or for detecting mild cognitive impairment (MCI) and the subtle cognitive changes associated with the early stages of cognitive decline; in contrast MMSE, MoCA and ACE are broad screening tools that assess a wider range of cognitive functions used in a wide range of dementias (Podhorna et al., 2016).

Along with cognitive tests, several biomarkers have value in the diagnosis of AD–see Figure 2 and Supplementary Table S2. Biomarkers can be classified as linked to Aβ deposition, pathologic tau and neurodegeneration (Jack et al. (2018). Assessing CSF is a relatively invasive and carries risk, therefore blood biomarkers that could track AD progression would be preferable; however, there are limitations related to quantification and reproducibility as only limited amounts of brain proteins diffuse into the bloodstream (Blennow and Zetterberg, 2015). For instance, although the longitudinal data has indicated an association between elevated plasma tau levels and subsequent cognitive decline, the slight increase in plasma tau may not be diagnostically significant (Mattsson et al., 2016; Blennow and Zetterberg, 2018). However, in a recent study of 786 patients a commercially available kit for p-tau217 provided comparable diagnostic evidence to that obtained from CSF biomarkers (Ashton et al., 2024). Furthermore, an analysis of blood samples in the UK Biobank identified four other plasma proteins linked to AD (Guo et al., 2024). Replication of these findings would prove beneficial to using blood tests to aid in the diagnosis of AD.

In conclusion, current data indicates that CSF biomarkers have a greater diagnostic value than plasma biomarkers; however, improvements in the kits available for assessment of plasma biomarkers may change this conclusion and thereby enhance prospects for earlier diagnosis, assessment of therapeutic treatment and, perhaps, prevention of AD.

A deficiency of acetylcholine (ACh) in the brain as the pathophysiological basis of AD was first proposed in 1976 and based on the observation that choline acetyltransferase, the enzyme responsible for the synthesis of ACh, was greatly reduced in the amygdala, cortex and hippocampus in postmortem brains from patients with AD compared to brains from non-AD subjects (Davies and Maloney, 1976). Cholinergic neurons, particularly in the basal forebrain, play critical roles in memory, attention, and learning (Whitehouse et al., 1982) and degeneration of ACh-producing neurons in AD patients affects neuronal communication, resulting in memory deficits (Hasselmo, 2006; Chen et al., 2022). As summarized in Figure 3 and based on co-immunoprecipitation data acetylcholinesterase (AChE), the enzyme responsible for the degradation of ACh, binds to and interacts with PS-1 in the same intracellular compartment in CNS neurons (Silveyra et al., 2008).

The glutamate hypothesis is based on evidence that there is a reduction in the binding of l-[3H] glutamate in the postmortem brains of subjects with AD (Greenamyre et al., 1987; Maragos et al., 1987) thus linking a defect in glutaminergic neurotransmission, the principle excitatory pathway in the brain, to AD (Maragos et al., 1987; Hladky and Barrand, 2022). As summarized in Figure 4 glutaminergic transmission and synaptic NMDARs are important for synaptic plasticity, LTP and neuronal survival, and also play an essential role in memory (see Abraham et al., 2019), whereas the activation of extra-synaptic NMDARs is associated with cell death and AD. This association form the basis for the use of the NMDAR antagonist, memantine, for the treatment of AD (Hardingham et al., 2002; Hardingham, 2006; Hardingham and Bading, 2010).

The most widely promoted hypotheses to explain the pathogenesis of AD is that neuronal damage is caused by aberrations in the processing of APP and the accumulation of Aβ due to a failure of the brain to clear Aβ (see Figure 5). Aβ was first isolated from the postmortem brains of subjects with AD and also Down’s Syndrome by Glenner and Wong in 1984, and in 1991, with the discovery of a mutation in the APP gene, (now associated with FAD) the amyloid hypothesis was proposed by Hardy and Allsop (Hardy and Allsop, 1991). Importantly, the presence of the ApoE4 allele significantly increases the risk of late-onset AD and increases the neurotoxicity of Aβ proteins (Kaplitt et al., 1996; Farrer et al., 1997).

APP is a transmembrane protein that is present in many types of cells, including neurons, and is known to play a major role in neuron synaptogenesis (Wang et al., 2009). In healthy humans without evidence of cognitive decline, Aβ is produced by the cleavage of APP by two major types of secretases: α- and β-secretase (Figure 5). When APP is cleaved by α-secretase, two fragments are generated: (i) soluble amyloid precursor protein alpha (sAPPα) and (ii) C83 (a membrane-bound C-terminal fragment) (Chow et al., 2010; Hare J, 2010). sAPPα has a role in neuroprotection and maintaining synaptic function and plasticity, which are essential for neuronal communication, learning, and memory formation (Mattson, 1997). In addition, a rat model of AD has shown that sAPPα promotes neurogenesis, which is also important for learning and memory formation (Huber et al., 1997). Most importantly, sAPPα inhibits aggregation of Aβ peptides by shifting the APP processing pathway into an anti-amyloidogenic pathway, thus preventing formation of toxic amyloid plaques generated by β-secretase. Cleavage of APP by β-secretase generates two different fragments: (i) soluble amyloid precursor protein beta (sAPPβ) and (ii) C99 (also called β-CTF), a membrane-bound fragment that is further processed by γ-secretase (Qiang et al., 2017) to yield several different beta-amyloid peptides, of which Aβ42 is the most common, and the basis of the amyloidogenic pathway (de Paula et al., 2009). γ-secretase is a protein complex made of many subunits including PSEN1, PSEN2, nicastrin, anterior pharynx-defective 1 (APH-1), and presenilin enhancer 2 (PEN-2) (Zhang et al., 2014). As mentioned previously, although a mutation in the PSEN1 gene is more commonly associated with earlier onset of FAD, a mutation in PSEN2 can also contribute. In fact, both PSEN1 and PSEN2 play important roles in forming γ-secretase’s catalytic subunit that if mutated, leads to alternations in the APP cleavage process and results in increased production of longer and more aggregation-prone forms of Aβ (Kabir et al., 2020).

Co-morbidities, such as insulin resistance in patients with T2D, enhance the buildup of Aβ plaques and advanced glycation end-products (AGEs) can modify Aβ peptides and accelerate aggregation of soluble Aβ peptides (Vitek et al., 1994). There is a strong link between insulin deficiency, insulin resistance, diabetes, and AD, with some research referring to AD as ‘type 3 diabetes’ (Steen et al., 2005; de La Monte and Ward, 2008). In addition, patients who are subjected to elevated oxidative stress, or inflammation, such as during frequent infection (also see section 3.6), are prone to developing abnormal Aβ plaques (Sharma and Kim, 2023), and Aβ42 itself has pro-oxidant activity (Ganguly et al., 2017; Butterfield and Boyd-Kimball, 2018). As previously discussed, studies have shown that Aβ42 levels decrease in CSF but increase in plasma (Teunissen et al., 2018), implying the measuring of plasma Aβ42 to monitor AD progression; however, the data has questioned its diagnostic utility.

Normally, Aβ is rapidly cleared by microglia, which act as the phagocytic cells in the CNS (Liu et al., 2021). However, in elderly people, the ability of microglia to phagocytose and degrade Aβ is decreased, resulting in decreased clearance and accumulation of Aβ inside neurons, eventually triggering inflammation, reduced neuronal communication, tau tangles, and ultimately death and degeneration (Lee and Landreth, 2010; Mucke and Selkoe, 2012; Bloom, 2014).

The argument that the buildup of amyloid plaques is the cause of AD has been vigorously disputed (Kepp, 2017), and this skepticism is supported by disappointments in the results of clinical trials with anti-Aβ MABs (Reiss et al., 2021). Skepticism supports the need to explore outside of the amyloid hypothesis for alternative targets (Karran and Hardy, 2014; Tse and Herrup, 2017; Herrup, 2022; Kaur et al., 2024). Amyloid plaques may start 20–30 years prior to evidence of cognitive function thereby raising issues over early detection, when to initiate treatment, and the potential of significant side-effects arising from long-term chronic treatment with drugs (Jansen et al., 2015). In addition, a significant percentage of patients with dementia are amyloid negative (Beach et al., 2012; Serrano-Pozo et al., 2014); in contrast, there are reports that elderly people with normal cognitive function have elevated levels of Aβ plaques (Aizenstein et al., 2008; Kepp, 2017)

β-secretase (BACE-1) inhibitors have been developed to decrease the level of Aβ proteins. However, Phase 2/3 clinical trials of subjects with mild-moderate AD with atabecestat and verubecestat were stopped due to low clinical efficacy with cognitive decline greater than in the placebo group (Egan et al., 2019; Henley et al., 2019). Similarly, lanabecestat also failed to slow cognitive decline and raised concerns over psychiatric adverse events (Wessels et al., 2020). Conceivably the failure of BACE1 inhibitors is linked to β-secretase targeting not only APP but other proteins, thus contributing to toxic side effects (Bazzari and Bazzari, 2022). It is argued that targeting γ-secretase with specific modulators (GSMs) will provide better specificity (Hur, 2022).

As summarized in Figure 6 the tau hypothesis is based on observations that AD is associated not only with an accumulation of Aβ but also aggregates of misfolded tau protein (see Lee et al., 2001; Frost et al., 2009). In AD, aggregates of Aβ are implicated in causing hyperphosphorylation of tau proteins (Gong and Iqbal, 2008), which interact with the nuclear pore complex producing structural disruption and functional loss culminating in neurotoxicity (Eftekharzadeh et al., 2019). Tau, which in the human brain exists in six isoforms, is a microtubule-associated protein that stabilizes the microtubules that serve as the highway for transporting cellular components.

The vascular hypothesis, as originally proposed in 1993 by De La Torre and Mussalvand, states that any disruption of blood supply that compromises cerebral perfusion will result in microglial activation and the build-up of neurofibrillary tangles and elevate the risk of a decline in cognitive function. This hypothesis is supported by epidemiological data as well as ultrastructural data from postmortem brains of subjects with AD that show extensive pathological changes in cerebral capillaries.

Cerebral hypoperfusion not only results in hypoxia and reduced nutrient delivery to the brain but also hinders the adequate clearance of metabolic waste products, including Aβ proteins. The primary route for Aβ clearance from the brain is via the glymphatic system - a combination of astrocyte (a type of glial cell) and lymphatic system that serves as a perivascular transit network linking the CSF and interstitial solutes that serves as the brain’s clearance system (Iliff et al., 2012). As illustrated in Figure 7, CSF produced in the choroid plexus passes through the subarachnoid space and crosses the periarterial space to the interstitial fluid space via aquaporin-4 (AQP4) channels present on astrocytes. Dysfunction of the glymphatic system has been associated with neurodegenerative disease (de Leon et al., 2017; Lv et al., 2021). Ablation of the meningeal glymphatic system in mice results in the accumulation of Aβ proteins (Da Mesquita et al., 2018a; 2018b).

ApoE, a glycoprotein that is instrumental in cholesterol transport, participates in lipid metabolism, and aids in the removal of Aβ proteins (Martel et al., 1997; Mahley and Rall, 2000; Marais, 2019). Humans express three ApoE genotypes, with ApoE3 being the most prevalent genotype, and ApoE2 the most protective genotype for AD, with higher risk in ApoE2/4 and ApoE3/4 heterozygotes (McCorkindale et al., 2022). Risk is also higher in females who carry ApoE4 (Riedel et al., 2016). Approximately 25% of the population express the ApoE4 genotype and it is a major contributor to LOAD (Crean et al., 2011; Di Battista et al., 2016; Montagne et al., 2020), and lowers the clearance of Aβ (Ma et al., 2018; Zhou et al., 2023), increases the transcription of AP-1, promotes tau phosphorylation (Huang et al., 2017; Barthélemy et al., 2020), and neuroinflammation (Fitz et al., 2021; Iannucci et al., 2021). Targeting ApoE4 offers another approach to the treatment of AD and a gene therapy trial directed at ApoE4 is currently being pursued (see Table 2).

A number of studies link atherosclerosis to dementia and AD; however, it is unclear whether the enhanced risk applies to all subjects with atherosclerosis. For example, a meta-analysis reported a link to carotid artery intima-media thickness (CMIT) and the risk of AD (Xie et al., 2020). Dolan et al. (2010) also concluded that the risk of AD was enhanced in those with cerebral atherosclerosis, but no significant risk was noted in those with peripheral atherosclerosis including coronary artery disease. Similarly, Dearborn et al. (2017) reported that intracranial atherosclerosis was associated with MCI and dementia, but not specifically with AD; however, the ApoE4 genotype is associated with an elevated risk of atherosclerosis and AD (Davignon, 2005). Gender also affects the risk of atherosclerosis: women are also at higher risk of developing AD and higher for those who have the ApoE4 genotype (Mielke et al., 2014; Riedel et al., 2016). Nonetheless, the between lipid levels and AD and benefits of lipid lowering drugs has been questioned (see Table 2; Reitz et al., 2004). Matsuzaki et al. (2011) demonstrated an association between hyperlipidemia and neuritic plaques seen in AD but it did not show any relationship with the levels of neurofibrillary tangles. A 2023 report of a cohort study of over 15,500 subjects with dementia and an average age of approximately 80 reported that statin use had positive effects of cognition as based on MMSE scores and favouring those taking simvastatin (Petek et al., 2023). A 2024 population-based cohort study in Hong Kong that statin use decreased the risk of dementia in patients with heart failure by ∼20% (Ren et al., 2024), and the most recent American Diabetes Association (ADA) guidelines recommend that to avoid cognitive risk in the age group 40–75, LDL levels should not exceed 70 mg/dL (1.8 mmol/L) (see also Chou et al., 2022).

Hypertension, which affects approximately 1.3 billion people worldwide, is a significant risk factor associated with AD. Pathological changes in the vasculature result in endothelial dysfunction, a reduction in vasoprotective factors including nitric oxide (NO), elevated levels of ROS, the promotion of vasoconstriction and atherosclerotic plaque formation, and a heightened risk of the formation of thrombi (Brandes, 2014). High mid-life, but not late-life, blood pressure has been correlated with AD (Gabin et al., 2017) but lower blood pressure in those older than 75 is likely a secondary phenomenon not directly linked to AD (Skoog et al., 1998). Data suggests that angiotensin receptor blockers may be the most beneficial antihypertensive drugs for patients with AD and hypertension inferring a contributing role for aberrant angiotensin-signaling in the pathogenesis of AD (Adesuyan et al., 2022, see Table 2).

There is also a racial link between expression of the ApoE4 genotype and Aβ protein levels in the brain as well as the CSF biomarkers T-tau, which are reported to be lower in African Americans who are diagnosed with AD (Morris et al., 2019). Furthermore, subjects expressing ApoE4 who also have other risk factors such as hypertension, atherosclerosis, and hypercholesterolemia, experience reduced protective benefits from anti-hypertensive drug therapy against AD (Hajjar et al., 2002; Stampfer, 2006; de Oliveira et al., 2018). Risk is further heightened by body mass index (BMI) and leptin signaling (Blautzik et al., 2018). Collectively, these findings highlight the importance of targeting modifiable risk factors as has been emphasised by several reports (see Femminella et al., 2018; Wang et al., 2018; Livingston et al., 2020; Zhang et al., 2021).

As a consequence of COVID-19 and in part because therapies for other targets such as acetylcholine and glutamate excitotoxity have given ambivalent results, there has been renewed interest in the role of viral infections, and not just SARS-CoV-2, in the development of AD, and also the potential preventive benefit of vaccinations (De Vlieger et al., 2022). (SCOPUS data are presented Supplementary Figures S2, 3 for the role of viruses and vaccinations.)

As early as 1952 links were made between the role of viral infections and AD (Sjogren et al., 1952). In 1982, it was proposed that recurring infections with human herpesvirus (HSV)-1 might be involved in the development and the progression of AD (Ball, 1982). Several other viruses including SARS-CoV-2, HIV, and spirochetal Gram-negative bacterial infections such as syphilis have been linked with the development of AD, however, a key question is: “How does a viral infection initiate or worsen AD?”. Two possible answers are summarized in Figure 8. According to the direct infection hypothesis (Figure 8A), the virus directly enters the CNS and causes neuronal death or activates an antiviral response (Seaks and Wilcock, 2020), causing neuroinflammation and AD pathology (De Vlieger et al., 2022). More frequent re-infection produces the greater cumulative damage seen in AD patients (Itzhaki et al., 2016). The indirect infection theory suggests that the virus does not necessarily need to enter the CNS but rather a peripheral viral infection induces a systemic inflammation that can cause AD pathology (De Vlieger et al., 2022). An additional mechanism (Figure 8B) involves extracellular vesicles (EVs). EVs are small sized vesicles that transport components between cells, including proteins, lipids, nucleic acids, and misfolded proteins such as Aβ and tau proteins (Hill A.F., 2019), thereby supporting the hypothesis of a prion-like transmissible process as a cause for AD proposed in 1984 (Prusiner, 1984).

Once the virus enters the CNS, an inflammatory response occurs resulting in neuroinflammation, Aβ aggregation, and formation of neurofibrillary tangles. The process becomes amplified if the microglia fail to clear the virus, enhancing the of release inflammatory cytokines and oxidative compounds (De Vlieger et al., 2022). Data from mice show that proinflammatory cytokines induce the expression of interferon-induced transmembrane protein 3 (Ifitm3) in neurons; Ifitm3 is a transmembrane molecule known to increase the activity of γ-secretase and thereby raise levels of Aβ proteins (Hur et al., 2020). The proinflammatory molecules released as a result of viral infection shift the APP processing towards the amyloidogenic pathway. Viral infection also increases the activity of Tau kinases and increases the formation of neurofibrillary tau tangles (Kitazawa et al., 2011). Viral infection is also associated with increased quinolinic acid (QA) (De Vlieger et al., 2022), produced by microglia and activated macrophages (Braidy et al., 2009). In the nematode Caenorhabditis elegans, QA acts as a neurotoxin and has been shown to contribute to neurodegeneration via the activation of NMDA receptors, increasing cytosolic calcium and ROS levels and ultimately depleting ATP (da Silveira et al., 2018). Collectively, these data prompt further investigations of the link between viral infections, neuroinflammation and the development of AD.

Herpes Simplex Virus type 1 (HSV-1), has been extensively studied for its role in AD. Upon initial infection, HSV-1 stays latent in the brain and in the trigeminal ganglia (Jamieson et al., 1991). During periods of acute stress or immunocompromised states, it can reactivate, resulting in viral shedding and initiating inflammation in nervous tissue leading to the release of ROS from microglial cells. HSV-1 infection itself may shift APP processing into the amyloidogenic pathway by increasing the expression of β-secretase and nicastrin, a component of γ-secretase, thus increasing the production of amyloid proteins (Figure 5) (Wozniak et al., 2007). HSV-1 may play a role in the abnormal phosphorylation of tau proteins that then form neurofibrillary tangles and cause AD (Alvarez et al., 2012). Supportive data shows recurrent HSV-1 infections in mice result in oxidative stress in the brain (Protto, et al., 2020). HSV-1 also interferes with proteasomal degradation of damaged DNA by downregulating Ku80 protein involved in the repair of double-strand DNA breaks, eventually resulting in neuronal death (De Chiara et al., 2010).

COVID-19 has also been linked to AD. Even though it is still too early to determine the long-term harmful effects of SARS-CoV-2, several studies report signs of AD after infection, including cognitive and motor difficulties, and memory loss issues (Helms et al., 2020; Bliddal et al., 2021). Patients who experience “Long COVID” describe sleep disruption, fatigue, anxiety, depression and a “brain-fog” (inability to focus, loss of memory, and difficulty conducting normal activities) that can persist for months (Venkataramani and Winkler, 2022). SARS-CoV-2, like HSV-1, induces neuroinflammation by either direct invasion into the CNS or through peripheral infection causing systemic inflammation (De Vlieger et al., 2022). Yang et al. (2021) reported that the gene expression pattern of the microglia subpopulation induced by COVID-19 overlaps with that of microglia seen in AD patients. In addition, increased levels of the AD biomarkers Aβ, T-tau, P-tau, and neurogranin were significantly increased in the EV of COVID-infected patients (De Vlieger et al., 2022). The fact that Aβ40 and 42 levels and P-tau are increased in the EV suggests that SARS-CoV-2 contributes to the accumulation of Aβ in infected patients as well as phosphorylation of tau proteins, thus increasing the risk of developing AD (Sun et al., 2021). Brain imaging data from a UK brain-bank of pre and post COVID infection of patients showed reductions in global brain size, thickness of grey matter, and cognitive function, plus damage to olfactory tissue, indicating that long COVID accelerates the onset of AD and can exacerbate the key symptoms (Douaud et al., 2022).

Vaccinations have also been shown to reduce the risk of AD (Wu et al., 2022). Both influenza and pneumonia vaccinations reduce the risk of AD (Lehrer, 2022), with a 46-month follow up study showing that influenza vaccination decreased the risk of AD by 3.4% (Bukhbinder et al., 2022). Similarly, a recent meta-analysis indicates vaccinations against Tdap (tetanus, diphtheria, and pertussis), influenza, hepatitis A and B, and typhoid) reduce the risk for dementia (Wu et al., 2022). Other herpetic vaccinations, for instance for shingles (varicella-zoster) have been associated with a 15% reduced risk of AD and other neurodegenerative diseases (Lehrer and Rheinstein, 2022). A prospective study (albeit proof of concept with n = 49) with the Bacillus Calmette–Guérin (BCG) vaccine for tuberculosis has also provided positive data that will stimulate additional interest in the role of vaccines as a protection against AD (Dow et al., 2022). Whether benefits are seen with COVID-19 vaccinations remains to be analysed.

Acyclovir, a first-line treatment for HSV-1 infection, has been shown to reduce tau phosphorylation and Aβ deposition in HSV-1 infected Vero cells (Wozniak et al., 2011). Hui et al. (2020) found the combination of acyclovir and dexamethasone protective against cognitive impairment in mice injected with Aβ. Finally, vaccines against common viral infections associated with increasing the risk of AD have been developed and tested (Lehrer and Rheinstein, 2022).