Introduction
Alzheimer's disease (AD) is the most prevalent type of dementia globally, accounting for approximately 60%–80% of all dementia cases [1]. According to statistics from the World Health Organization, the number of AD patients worldwide has exceeded 50 million, and this figure is projected to rise to 152 million by 2050. The disease not only causes progressive cognitive decline and loss of self‐care ability in patients but also imposes a heavy medical and care burden on families and society, making it a major global public health issue.
The pathological progression of AD is highly insidious: from the onset of core pathological changes to the manifestation of obvious cognitive impairment, it typically undergoes a preclinical stage lasting 10–20 years [2]. During this stage, patients show no typical clinical symptoms, but characteristic changes such as Aβ deposition and abnormal phosphorylation of tau protein have already begun in the brain. However, massive irreversible neuronal damage has not yet occurred, making this the optimal window for intervention.
Scientific interventions can delay or even block the pathological process [3].
Currently, traditional diagnostic and screening methods for AD still have significant limitations: neuropsychological scales can only assess the degree of cognitive impairment but cannot reflect early pathological changes, and the results are susceptible to interference from multiple factors; cerebrospinal fluid (CSF) testing is an invasive procedure with low patient acceptance and high detection costs, making it difficult for large‐scale clinical application [4]; although PET‐CT can intuitively display the pathological distribution in the brain with high diagnostic accuracy, its routine clinical application is greatly restricted by issues such as expensive equipment, high testing costs, and radiation exposure [5]. As the “gold standards” for AD detection, CSF testing and PET‐CT are both based on the ATN biomarker classification framework proposed by the National Institute on Aging and Alzheimer's Association (NIA‐AA) in 2018. Centered on amyloid β (Aβ), tau protein, and neurodegeneration (N), this framework provides a core theoretical basis for the screening and classification of AD biomarkers and points out a clear direction for subsequent research on blood biomarkers.
With the continuous innovation of detection technologies, blood testing has gradually become an ideal method to address the dilemmas of traditional diagnosis. A variety of highly sensitive and specific AD blood biomarkers have been successively discovered and clinically validated. The revised Diagnostic and Staging Criteria for Alzheimer's Disease issued by NIA‐AA in 2024 officially incorporated core blood biomarkers such as plasma p‐tau217 into the AD diagnosis and staging system, updating the original ATN biomarker classification framework to the ATNIVS framework.
It clearly defines these biomarkers as core indicators for the independent diagnosis of AD, and detection methods are gradually evolving toward automation, high throughput, and precision. Research findings presented at the 2025 Alzheimer's Association International Conference (AAIC) and the first clinical practice guidelines for blood biomarkers further clarified that the sensitivity and specificity of AD blood testing indicators have both exceeded 90% (the “dual 90%” standard), enabling them to replace invasive CSF testing and expensive PET‐CT. The release of these guidelines has further expanded the application scenarios of blood testing in clinical AD diagnosis and provided a new path for early AD diagnosis [6].
Based on this background, this paper provides a comprehensive overview of the existing types of AD blood biomarkers and their detection technologies.
In addition, the diagnosis of AD also requires systematic exclusion of treatable or potentially reversible causes of cognitive impairment through routine laboratory evaluation. Routine laboratory evaluations therefore include complete blood count to screen for anemia or systemic infection; metabolic panel to assess liver and kidney function, electrolyte imbalance, and glucose abnormalities; thyroid function tests (TSH and FT4) to exclude thyroid dysfunction; vitamin levels (vitamin B12, folate, and homocysteine) to identify deficiency‐related cognitive decline; infection screening such as syphilis serology and HIV testing; and inflammatory or autoimmune markers when acute cognitive or psychiatric symptoms are present. These laboratory assessments help exclude alternative causes of cognitive decline and improve the diagnostic accuracy of AD.
Types and Pathological Significance of Blood Biomarkers in Alzheimer's Disease
Based on the core pathological changes and related pathological responses of AD, blood biomarkers can be divided into three categories: core pathological biomarkers, neuronal damage biomarkers, and genetic risk biomarkers. The dynamic changes of representative blood‐based biomarkers across different stages of Alzheimer's disease pathogenesis are illustrated in Figure 1.
Highlighting their relationships with amyloid deposition, tau pathology, and neurodegeneration. These biomarkers complement each other and provide multidimensional insights into early screening, accurate diagnosis, and prognosis of AD. Among them, core pathological biomarkers are the most well‐developed and have the greatest clinical application potential.
A structured classification of major blood‐based biomarkers, together with their pathological significance and clinical application value, is summarized in Table 1.
In addition, miRNAs, exosomes, and some inflammation‐related factors have also been studied for AD blood testing, but these biomarkers have significant gaps in specificity and accessibility compared with the aforementioned markers and are generally used for auxiliary diagnosis or scientific exploration.
Core Pathological Biomarkers
Core pathological biomarkers are directly associated with the two core pathological features of AD: β‐amyloid (Aβ) deposition and excessive phosphorylation of tau protein, serving as direct molecular manifestations of the AD pathological process. They mainly include Aβ‐related biomarkers and tau‐related biomarkers.
Aβ‐related biomarkers serve as early indicators of the AD pathological process. Aβ is produced through the hydrolysis of amyloid precursor protein (APP) by proteases, primarily existing in two subtypes: Aβ40 and Aβ42. Among them, Aβ42 has stronger aggregability, is prone to deposition in brain tissue, and induces abnormal phosphorylation of tau protein and neuronal damage [7].
The core of blood Aβ‐related biomarker detection includes Aβ42 concentration, Aβ40 concentration, and the Aβ42/Aβ40 ratio. When pathological Aβ deposition occurs in the brain, the blood Aβ42 concentration and Aβ42/Aβ40 ratio decrease [8]. The blood Aβ42/Aβ40 ratio has been found to decrease significantly in the preclinical stage of AD, preceding cognitive impairment, and can be used for early risk screening of AD.
However, due to the poor stability and easy degradation of Aβ protein, the diagnostic accuracy of detecting Aβ42 or Aβ40 concentration alone is limited, whereas the Aβ42/Aβ40 ratio, especially when combined with plasma p‐tau217‐related measures, can improve the accuracy of amyloid PET classification in preclinical AD [9, 10].
Tau‐related biomarkers are key indicators of neuronal dysfunction in AD. Tau protein normally functions to maintain the stability of neuronal microtubules. In the pathological state of AD, tau protein undergoes excessive phosphorylation, aggregates to form neurofibrillary tangles, and leads to impaired neuronal function [11].
Blood tau‐related biomarkers mainly include total tau protein (t‐tau) and phosphorylated tau protein (p‐tau). Among them, p‐tau is a specific indicator of abnormal phosphorylation of tau protein, and the most widely studied subtypes are p‐tau181, p‐tau217, and p‐tau231 [12].
Among phosphorylated tau protein detections, p‐tau217 is one of the most accurate AD blood biomarkers currently available. Its concentration is highly correlated with the degree of cerebral Aβ accumulation, tau deposition, and cognitive decline, and can effectively distinguish AD patients from healthy individuals, as well as AD from other types of dementia [13]. However, due to the barrier of the blood–brain barrier, the level of plasma p‐tau217 is only at the fg/mL level, requiring extremely high detection sensitivity.
Clinical studies have shown that plasma p‐tau217 and p‐tau217‐related ratios are significantly associated with CSF‐ and PET‐defined AD pathology, supporting their potential value as minimally invasive biomarkers for identifying central AD pathological changes [14, 15]. In contrast, p‐tau181 has relatively lower diagnostic specificity than p‐tau217, whereas p‐tau231 may be more closely associated with early pathological changes and cognitive decline. A systematic review and meta‐analysis showed that plasma and CSF p‐tau181 and p‐tau231 levels are associated with cognitive function in AD, supporting their potential value in disease assessment and progression monitoring [16].
In clinical practice, combined detection with different p‐tau subtypes may help improve biomarker interpretation [17].
Neuronal Damage Biomarkers
Neuronal damage biomarkers reflect secondary pathological changes such as neuronal injury, axonal degeneration, and glial activation during the progression of Alzheimer's disease (AD). They can complement core pathological biomarkers and provide information on disease severity, progression, and prognosis. The representative blood‐based neuronal damage biomarkers include neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP).
NfL is a major structural protein of neuronal axons. When axonal injury or neuronal degeneration occurs, NfL is released into the cerebrospinal fluid and subsequently into the blood. Its concentration is positively associated with the degree of neuroaxonal damage.
In AD, plasma or serum NfL levels generally increase with disease progression and correlate with cognitive decline, brain atrophy, and clinical severity. Therefore, NfL has potential value in disease staging, prognosis evaluation, and longitudinal monitoring.
However, NfL is not an AD‐specific biomarker. Rather than reflecting amyloid or tau pathology directly, it indicates general neuroaxonal injury. Elevated NfL levels can also be observed in multiple neurological conditions [18], including traumatic brain injury, Parkinson's disease, stroke, multiple sclerosis, frontotemporal dementia, amyotrophic lateral sclerosis, and other neurodegenerative or inflammatory diseases.
In acute neurological injuries such as traumatic brain injury and stroke, blood NfL may increase dynamically after axonal damage and remain elevated for weeks or months. Therefore, although NfL is sensitive to neuronal injury, its diagnostic specificity for AD is limited.
This non‐specificity has important clinical implications. First, NfL should not be used alone to diagnose AD or distinguish AD from other neurological disorders. Second, when elevated NfL is detected in patients with cognitive impairment, clinicians should carefully consider alternative or coexisting causes of neuroaxonal injury, such as recent head trauma, cerebrovascular disease, Parkinsonian disorders, inflammatory demyelinating diseases, peripheral neuropathy, renal dysfunction, and advanced age.
Third, the clinical interpretation of NfL should be combined with core AD pathological biomarkers, especially plasma p‐tau217 and Aβ42/Aβ40, as well as cognitive assessment, neuroimaging, medical history, and neurological examination. In this context, NfL is more suitable as a marker of disease intensity, neuronal injury burden, and progression rate than as an independent diagnostic marker for AD [19].
GFAP is an intermediate filament protein mainly expressed in astrocytes and is commonly used as a marker of astrocytic activation and glial injury. In AD, Aβ deposition and tau aggregation can activate astrocytes and microglia, leading to chronic neuroinflammation and increased GFAP release into blood. Plasma GFAP levels may rise in the preclinical or early stages of AD and are associated with amyloid pathology and subsequent cognitive decline, suggesting potential value for early risk assessment and disease monitoring.
Nevertheless, GFAP also has limitations in disease specificity. Although it is more closely related to astrocytic activation than to neuronal axonal damage, elevated blood GFAP can be observed in several neurological conditions, including traumatic brain injury, stroke, multiple sclerosis [20], neuromyelitis optica spectrum disorders, and other inflammatory or degenerative central nervous system diseases. Therefore, GFAP should be interpreted as a marker of astroglial activation and neuroinflammatory response rather than a disease‐specific marker for AD.
In clinical practice, its diagnostic value is enhanced when combined with AD core pathological biomarkers, cognitive assessment, and imaging evidence.
Genetic Risk Biomarkers
Genetic factors play a central role in the pathogenesis of Alzheimer's disease (AD), particularly in familial early‐onset forms. Pathogenic mutations in APP, PSEN1, and PSEN2 can directly cause autosomal dominant early‐onset AD, although these cases account for only a small proportion of all AD patients. In contrast, sporadic late‐onset AD is influenced by multiple genetic risk variants with different effect sizes.
Among them, the ε4 allele of the apolipoprotein E gene (APOE ε4) remains the most established genetic risk factor for late‐onset AD. APOE ε4 is associated with increased Aβ aggregation, impaired Aβ clearance, altered lipid metabolism, neuroinflammation, and earlier disease onset. Therefore, APOE genotyping has important value in genetic risk assessment, risk stratification, and clinical trial enrollment, although it should not be used alone as a diagnostic marker for AD.
In recent years, genome‐wide association studies (GWAS) have greatly expanded the understanding of the genetic architecture of AD beyond APOE. Large‐scale GWAS and meta‐analyses have identified multiple susceptibility loci associated with late‐onset AD, including BIN1, CLU, PICALM, CR1, ABCA7, MS4A, CD33, EPHA1, CD2AP, SORL1, TREM2, INPP5D, FERMT2, PTK2B, HLA‐DRB1, ADAM10, ACE, ADAMTS1, IQCK, and WWOX [21]. These loci are involved in multiple AD‐related biological processes, such as amyloid precursor protein processing, tau pathology, lipid metabolism, endocytosis, immune regulation, complement activation, microglial function [22], and synaptic homeostasis.
Therefore, GWAS findings not only provide additional genetic risk markers but also reveal the complex biological networks underlying AD pathogenesis.
The 11p11.2 locus, also referred to as the CELF1/SPI1 region, is one representative susceptibility locus that has attracted increasing attention. Functional genomic studies suggest that this locus may involve the coexistence of multiple functional variants and multiple candidate genes, rather than a single causal variant or gene [23]. Potential genes in this region, such as CELF1, SPI1, MTCH2, ACP2, NDUFS3, PSMC3, C1QTNF4, and MADD, may participate in immune regulation, mitochondrial function, protein degradation, and microglial‐related pathways.
These findings suggest that AD genetic risk loci may influence disease susceptibility through complex regulatory mechanisms and cell‐type‐specific effects.
From a clinical perspective, genetic risk biomarkers are mainly used for risk prediction and population stratification rather than direct diagnosis. Their diagnostic value is limited when used alone because carrying a risk allele does not necessarily indicate current AD pathology, and the absence of a risk allele does not exclude AD. However, when combined with blood‐based pathological biomarkers such as plasma p‐tau217 and Aβ42/Aβ40, as well as cognitive assessment, neuroimaging, and clinical information, genetic markers may improve individualized risk prediction, early identification of high‐risk populations, and selection of participants for prevention or disease‐modifying treatment trials.
Future studies should further validate polygenic risk scores and GWAS‐derived loci in ethnically diverse populations to improve the generalizability and equity of genetic risk assessment in AD.
Other Biomarkers
In addition to core pathological biomarkers, neuronal damage biomarkers, and genetic risk biomarkers, AD involves multisystem pathological changes such as inflammation, immunity, and gene regulation, leading to the emergence of various auxiliary biomarkers. However, most of them are limited by specificity or accessibility and are only used for scientific research or combined auxiliary diagnosis. Among inflammation‐related biomarkers, pro‐inflammatory factors such as CRP and IL‐6 are elevated in AD brain inflammatory responses, while anti‐inflammatory factors such as IL‐10 are decreased, reflecting inflammatory imbalance.
However, these indicators are susceptible to interference from infections, cardiovascular diseases, etc., and have insufficient specificity [24]; exosomes can carry brain‐derived proteins such as Aβ and Tau, which can reflect brain pathology. Selective encapsulation of tau filaments has been found in brain‐derived exosomes of AD patients [25], but the separation and purification technology is complex, the detection cost is high, and accessibility is poor; miRNAs (such as miR‐132 and miR‐124) are involved in AD pathology by regulating gene expression, and some subtypes are associated with Aβ deposition, but there are problems such as low tissue specificity, unstable expression, and lack of detection standardization; complement system biomarkers (C3, C4, etc.) are involved in Aβ clearance and neuronal damage. The expression of complement genes (C1q, C3, C4, etc.) in the brains of AD patients is significantly upregulated, and the classical pathway is continuously activated [26], but it is easily affected by the systemic immune status and has limited specificity; others such as GSN (gelsolin) and DBP (vitamin D‐binding protein) can regulate Aβ aggregation, directly interact with Aβ, and reduce its toxic damage [27], but the clinical validation sample size of both is small, the detection methods are not popularized, and there is a lack of unified reference standards.
In recent years, several emerging AD‐related blood biomarkers have been investigated to complement established core biomarkers such as plasma p‐tau217, Aβ42/Aβ40, NfL, and GFAP. These emerging candidates are mainly associated with tau‐related pathological changes, synaptic dysfunction, neuroinflammation, proteomic alterations, and multimodal disease stratification. Although most of these markers remain at the exploratory or research‐validation stage, they may provide additional information on disease staging, prognosis, and individualized risk prediction when combined with core AD biomarkers, cognitive assessment, neuroimaging, and genetic risk information.
A summary of representative emerging AD‐related blood biomarkers and their potential clinical value is shown in Figure 2.
Therefore, these emerging biomarkers should not be interpreted as independent diagnostic markers at the current stage. Their greatest potential may lie in combination with established core pathological biomarkers and clinical information, thereby improving disease staging, prognosis evaluation, and individualized prediction in AD.
Detection Technologies for Blood Biomarkers in Alzheimer's Disease
The core requirements for AD blood biomarker detection are “high sensitivity, high specificity, convenience, and low cost,” and the performance of detection technologies directly determines their clinical application value. Currently, AD blood biomarker detection technologies at home and abroad show a diversified development trend, covering multiple technical paths such as immunoassays, mass spectrometry, and proteomics screening. Among them, the fully automated chemiluminescence detection technology platform based on immunoassays is more in line with clinical detection needs due to its adaptability to clinical standardized processes, while high‐throughput analysis methods play an important role in scientific research and broad‐spectrum screening.
The most widely used technologies in the latest clinical studies on AD blood biomarkers are chemiluminescence detection technologies based on immunoassays (mainly represented by electrochemiluminescence immunoassay, photo‐induced chemiluminescence immunoassay, and chemiluminescent enzyme immunoassay), high‐sensitivity single molecule array technology (Single Molecule Array, SIMOA), and SomoScan technology, mass spectrometry, and Olink technology suitable for high‐throughput detection. A comparison of the major detection technologies for AD blood biomarkers, including their detection limits, advantages, and limitations, is presented in Table 2.
Fully Automated Chemiluminescence Detection Technologies Based on Immunoassays
Chemiluminescence detection technologies based on immunoassays take the specific binding of antigens and antibodies as the core principle, and realize the quantitative detection of low‐abundance biomarkers through signal amplification technology. They have the advantages of simple operation, high degree of automation, and adaptability to clinical batch detection, and are one of the mainstream technical paths for AD blood testing. They mainly include electrochemiluminescence immunoassay (ECLIA), photo‐induced chemiluminescence immunoassay (LiCA), and chemiluminescent enzyme immunoassay (CLEIA).
Electrochemiluminescence immunoassay (ECLIA) uses tris(2,2′‐bipyridine)ruthenium(II) (Ru(bpy)32+)‐tripropylamine (TPA) as the core luminescent system, combined with magnetic microsphere capture technology to achieve quantitative detection of biomarkers [28]. Its core principle is to trigger an electrochemical reaction by applying a positive voltage to the electrode, causing TPA and Ru(bpy)32+ to undergo an oxidation reaction at the anode of the electrode, and form an excited state Ru(bpy)32+* through electron transfer. When the latter returns to the ground state, it releases a 620 nm red light signal, and the intensity of the light signal is positively correlated with the concentration of the target substance; the capture system uses 2–4 μm silanized magnetic microspheres, which improve the antigen capture efficiency and detection specificity by virtue of their large specific surface area and rapid magnetic separation characteristics.
According to the difference in the molecular weight of the target substance, the detection mode adopts a two‐step sandwich method (suitable for macromolecular antigens) and a competitive method (suitable for small molecular substances). This technology has excellent performance, with a detection limit of pg/mL level and a wide dynamic range, which can detect the low‐abundance AD core biomarker p‐tau217 in blood [29]. However, the stability of its core Ru(bpy)32+‐TPA labeling system depends on refrigerated storage, temperature and light can easily lead to a decrease in its activity, and the core hardware such as electrodes requires high maintenance frequency and cost.
Photo‐induced chemiluminescence immunoassay (LiCA), also known as reactive oxygen species pathway homogeneous chemiluminescence technology, innovatively integrates the triple advantages of laser excitation, homogeneity, and immune reaction. With a dual‐microsphere system of nanoscale photosensitive microspheres and luminescent microspheres as the core, relying on specific luminescent reactions mediated by reactive oxygen species, combined with a homogeneous washing‐free design, it constructs a high‐precision quantitative detection system for biomarkers. Compared with other traditional chemiluminescence technologies, this technology has outstanding advantages: no electrode triggering, no enzyme‐catalyzed substrate decomposition, and no washing and separation steps, which effectively improves detection efficiency and consistency.
Nanoscale photosensitive microspheres can efficiently absorb 680 nm laser energy and generate reactive oxygen species, while paired luminescent microspheres can specifically receive reactive oxygen species energy. Under multilevel energy transfer, the rare earth element europium is finally excited to emit 610–630 nm orange‐red light signals. The dual microspheres have small diameter and high specific surface area, and their surfaces can be immobilized with a higher load of antibodies after silanization modification.
During the detection process, the distance between the dual microspheres is controlled through the specific binding of antigens and antibodies to ensure efficient transfer of reactive oxygen species and trigger luminescent reactions. The intensity of the light signal is well positively correlated with the concentration of the target substance, enabling accurate quantification of the analyte [30, 31, 32]. The detection limit of this platform technology can reach fg/mL level, which can efficiently detect low‐abundance biomarkers (such as blood NfL, p‐tau217, etc.), with intra‐batch and inter‐batch coefficients of variation (CV) ≤ 5% and a wide dynamic range (also applicable to the detection of blood GFAP, Aβ protein, etc.).
Chemiluminescent enzyme immunoassay (CLEIA) technology mainly relies on the alkaline phosphatase (ALP)‐AMPPD luminescent system and magnetic bead capture. This technology selects alkaline phosphatase (ALP) as the specific labeling enzyme and uses AMPPD [3‐(2′‐spiroadamantane)‐4‐methoxy‐4‐(3″‐phosphoryloxy)phenyl‐1,2‐dioxetane] as the exclusive luminescent substrate, forming a highly matched luminescent system. Its luminescent mechanism is that ALP catalyzes AMPPD to remove the phosphate group, generating an unstable intermediate AMPD.
During the spontaneous decomposition of AMPD, energy is released, producing blue‐green fluorescence with a maximum wavelength of 477 nm for signal capture and quantitative analysis [33]. The capture system uses magnetic microspheres with a diameter of 1–5 μm as the solid‐phase carrier, and their surfaces are specially treated for immobilizing capture antibodies. The detection process mainly adopts a two‐step sandwich method: first, the capture antibody on the magnetic beads binds to the target antigen in the sample, and after magnetic field separation of impurities, ALP‐labeled detection antibody is added to form a complex.
After washing again, AMPPD substrate is added to initiate the luminescent reaction, and finally, quantitative analysis is completed through the photoelectric detection system. This technology has a detection limit of pg/mL to fg/mL level (can effectively detect indicators such as p‐tau217/Aβ42 and Aβ42/Aβ40) [34, 35]. However, this technology also has certain drawbacks: the core reaction enzyme (ALP) is susceptible to degradation by environmental factors, leading to a decrease in catalytic activity and affecting detection accuracy; at the same time, sample collection requires the use of special blood collection tubes to avoid additives in the blood collection tubes interfering with enzyme activity and antigen–antibody binding reactions, which increases the detection cost to a certain extent.
High‐Sensitivity Single Molecule Array Technology Based on Immunoassays
Single molecule array (Simoa) technology is based on single‐molecule immunoassay technology. With dual‐antibody sandwich immune reaction as the core, it integrates magnetic bead capture, femtoliter‐scale microwell array separation, and single‐molecule signal detection technology. The core logic is to capture and seal individual immune complexes in microwells, and generate signals through β‐galactosidase‐catalyzed substrates to realize the counting and quantitative analysis of individual biomarker molecules [36].
The most prominent advantage of this platform is its ultrahigh sensitivity, with a detection limit of fg/mL level, which can detect trace biomarkers such as NfL and p‐tau217 in peripheral blood that cannot be captured by traditional methods, and has certain application value in basic research on neurodegenerative diseases. However, its core shortcoming is the extremely complex operation, which requires multiple precise steps such as sample incubation, magnetic separation and washing, precise loading of magnetic beads into microwells, oil film sealing, enzymatic reaction, and imaging counting. The large inter‐batch difference makes it difficult for clinical application.
In summary, although the Simoa platform is irreplaceable in basic research on trace biomarkers, it is still difficult to achieve routine clinical transformation due to shortcomings such as complex operation, poor sample adaptability, and high maintenance costs.
High‐Throughput Detection Technologies
High‐throughput detection technologies realize biomarker detection based on the physical and chemical properties of analytes or proteomics analysis, and have unique advantages in scenarios such as multi‐biomarker screening and new biomarker discovery.
Mass spectrometry (MS) is mainly classified according to ionization methods and is core divided into matrix‐assisted laser desorption/ionization mass spectrometry (MALDI‐MS) and electrospray ionization mass spectrometry (ESI‐MS). It is often used in combination with mass analyzers such as quadrupole, time‐of‐flight (TOF), and orbitrap (such as MALDI‐TOF‐MS, LC‐ESI‐Q‐TOF‐MS) to optimize detection efficiency. This technology requires first digesting protein samples into peptides with trypsin, then processing them through corresponding ionization methods, followed by separating peptide ions according to differences in ion mass‐to‐charge ratio through a mass analyzer.
The detector captures ion signals and converts them into mass spectra and finally completes the qualitative identification and absolute quantification of peptides and corresponding proteins by searching protein databases. This technology can accurately distinguish tau protein isoforms with different phosphorylation sites (such as p‐tau181, p‐tau217, p‐tau231) and Aβ fragments [37], and can quantify multiple modified proteins simultaneously, which can discover potential new biomarkers in proteomics research. The specificity is significantly better than that of immunoassay methods, but the detection sensitivity is greatly affected by the quality of sample pretreatment, the detection efficiency of low‐abundance biomarkers depends on enrichment steps, the sample processing is cumbersome, and the detection cycle exceeds 4 h.
The core technology of Olink technology is the proximity extension assay (PEA). It amplifies signals through PCR after the combination of antibody‐oligonucleotide probes, realizing high‐throughput simultaneous detection of proteins. The sample demand is only 1–5 μL, which is suitable for micro‐sample analysis.
Although this platform still relies on antibodies, it can simultaneously analyze hundreds of proteins in a single detection through innovative nucleic acid signal amplification technology, including AD core biomarkers (p‐tau217, Aβ42/Aβ40) and inflammation‐related factors. It can construct a multi‐biomarker combined diagnostic model and has certain advantages in AD risk stratification [38]. However, the detection cost is high, and result interpretation needs to be combined with bioinformatics analysis.
SomaScan detection technology uses chemically modified single‐stranded DNA aptamers (SOMAmer) to specifically bind to target proteins, combined with nucleic acid amplification and detection technology, to achieve high‐sensitivity and high‐specificity quantitative analysis of thousands of proteins (up to 11,000) in biological samples. Targeted capture is achieved through the specific binding of aptamers to target proteins, and quantitative analysis is completed through microarray chips and signal amplification technologies (such as fluorescence quantitative or electrochemical detection) without the participation of antibodies, avoiding antibody cross‐reaction and batch difference interference in principle. Its core feature is the ability to accurately distinguish AD‐related protein isoforms with similar structures (such as p‐tau181/p‐tau217/p‐tau231 subtypes, Aβ40/Aβ42 fragments and modified derivatives) [39].
It can simultaneously quantify hundreds of proteins in a single detection, and the aptamers have strong stability and can maintain binding activity in complex plasma matrices. However, the detection sensitivity is at the ng/mL to pg/mL level, which is significantly lower than other platforms, making it unsuitable for high‐precision detection of low‐abundance proteins.
Comprehensive analysis of the above technologies shows that AD blood biomarker detection technologies generally present a dual‐track development trend of “clinical application” and “scientific research expansion”: fully automated immunoassay platforms (electrochemiluminescence immunoassay, photo‐induced chemiluminescence technology, chemiluminescent enzyme immunoassay, etc.) dominate clinical applications with their advantages of high automation and high consistency, while Simoa, Olink, mass spectrometry, and SomoScan technologies have become key tools for AD pathological mechanism research and new biomarker discovery.
Clinical Application and Performance Evaluation
Due to the high adaptability of fully automated immunoassay platforms to clinical standardized processes in the field of AD blood biomarker detection, some studies have conducted clinical validation of relevant electrochemiluminescence immunoassays, photo‐induced chemiluminescence technology, and chemiluminescent enzyme immunoassays.
In Chinese population‐based cohorts, detection method performance varies significantly. For example, using light‐initiated chemiluminescent assay (LiCA) to detect plasma Aβ42/Aβ40 achieved an AUC of 0.81 (95% CI: 0.77–0.84) for identifying amyloid PET‐positive individuals, whereas Simoa yielded an AUC of 0.72 (95% CI: 0.68–0.76; p < 0.01), suggesting that LiCA may provide better discriminative performance in Chinese populations.
Population Heterogeneity and Cutoff Calibration
Population heterogeneity is an important factor affecting the clinical translation of blood‐based biomarkers for Alzheimer's disease. Although plasma biomarkers such as p‐tau217, Aβ42/Aβ40, NfL, and GFAP have shown promising diagnostic performance, their concentrations and diagnostic accuracy may vary across populations with different racial, ethnic, genetic, clinical, and socioeconomic backgrounds. A recent systematic review summarized that ethnic and racial factors may influence blood biomarker levels and their diagnostic interpretation in AD, suggesting that clinical thresholds established in one population may not be directly applicable to another population without further validation [40].
Importantly, racial and ethnic differences should not be interpreted as purely biological determinants. They may reflect complex interactions among genetic ancestry, APOE genotype distribution, comorbidities, renal function, vascular risk factors, body mass index, educational background, healthcare access, and other social determinants of health. Therefore, before blood‐based biomarkers are widely implemented in routine clinical practice, cutoff values should be validated in large, multicenter, and ethnically diverse cohorts, preferably using amyloid PET, tau PET, or CSF biomarkers as reference standards.
Such population‐calibrated validation will be essential for improving the generalizability, diagnostic accuracy, and equity of blood‐based biomarker testing in Alzheimer's disease.
Electrochemiluminescence Immunoassay (ECLIA)
In studies related to AD blood biomarkers, ECLIA focuses on the accurate quantification of plasma core pathological protein p‐tau217, and relevant studies have verified its clinical value in multi‐population and multicenter scenarios.
Hibar et al. [41] verified its performance in five major clinical cohorts (N = 2148, including populations with cognitive impairment and normal cognition). The results showed that the AUC value of this method for detecting amyloid pathology was 0.878 (95% CI: 0.840–0.915) in the cognitively impaired population and 0.907 (95% CI: 0.885–0.929) in the cognitively normal population; when the cutoff value was set at < 0.189 pg/mL, the negative predictive values (NPV) were 92.51% (cognitively impaired) and 98.60% (cognitively normal), which could reduce the demand for amyloid PET scans by 8.55%–41.68%, providing an efficient tool for clinical pre‐screening. However, the study also found that its specificity was relatively low (29%–51%).
In another multicenter validation study, blood p‐tau217 based on ECLIA performed excellently in distinguishing AD from other neurodegenerative diseases, with an AUC value of 0.92, and was highly correlated with cerebrospinal fluid p‐tau217 levels (r = 0.83) [42]. Dyer et al. [43] applied ECLIA to detect blood p‐tau217 in a memory clinic cohort (N = 523) and proved that its sensitivity for identifying amyloid positivity was 96.3% and specificity was 82.1%, suggesting that this technology still maintains stable performance in real clinical scenarios and has the potential to be a minimally invasive alternative to CSF testing.
Photo‐Induced Chemiluminescence Technology (LiCA)
Based on the singlet oxygen energy transfer mechanism triggered by laser, LiCA achieves washing‐free homogeneous detection, with both high sensitivity and anti‐interference ability. In recent years, it has been systematically validated in domestic AD cohorts, and detection parameters have been optimized especially for the characteristics of biomarkers in the Chinese population.
Wang et al. [30] simultaneously detected six biomarkers including Aβ42/Aβ40, p‐tau181, and p‐tau217 using LiCA technology in a Shanghai multicenter cohort (N = 1254). The results showed that the AUC value of plasma p‐tau217 based on LiCA technology for predicting amyloid positivity was 0.92 (95% CI: 0.90–0.94), which was significantly better than other biomarkers; when a dual‐threshold strategy was adopted (≤ 0.37 pg/mL as negative, > 0.47 pg/mL as positive), only 16% of samples required further PET verification, greatly reducing diagnostic costs. The study also confirmed that the AUC value of LiCA for detecting Aβ42/Aβ40 (0.81) was significantly higher than that of the Simoa platform (0.72, p < 0.01), which was speculated to be related to the fact that LiCA detection did not require sample dilution, reducing the adsorption and degradation of Aβ protein.
In another study based on the C‐PAS cohort (N = 1275), the association between plasma p‐tau217 detected by LiCA technology and AD pathological progression was explored. The plasma p‐tau217 level gradually increased with the elevation of Braak staging (median 0.36 pg/mL in Braak I–II stages vs. 1.13 pg/mL in V–VI stages, p < 0.001), and could effectively distinguish A+T− and A+T+ subgroups (AUC = 0.876), further verifying the sensitivity and stability of plasma p‐tau217 detection based on LiCA technology [31].
Tu et al. [32] further expanded the application scenario of LiCA technology and found that in populations with abnormal sleep quality, the p‐tau217 level detected by LiCA was positively correlated with the rate of cognitive decline (β = 0.12, p = 0.004), providing a new dimension for screening high‐risk AD populations. In the latest study on AD blood biomarker detection using LiCA technology, it was found that plasma p‐tau217 levels were positively correlated with CL values measured by PET‐CT, with the strongest positive correlation in the range of 10 < CL ≤ 30 (p = 0.001), and could effectively distinguish populations with CL > 30 (AUC = 0.919 and 0.926 for CN/MCI populations, respectively) and CL > 10 (AUC = 0.959 for dementia populations) [44].
Chemiluminescent Enzyme Immunoassay (CLEIA)
CLEIA realizes the detection of AD biomarkers through enzyme‐catalyzed chemiluminescent signal amplification, and detection platforms based on this technology have been applied in clinical detection.
Dakterzada et al. [34] used CLEIA technology to detect plasma p‐tau217 in 276 patients with cognitive impairment. Its correlation with cerebrospinal fluid Aβ42/Aβ40 reached r = −0.707, and the AUC value for identifying amyloid positivity was 0.924, which was significantly better than p‐tau181 (AUC = 0.808); longitudinal follow‐up showed that blood p‐tau217 levels could predict the progression of MCI to AD (AUC = 0.876), and the dual‐threshold strategy (< 0.156 and ≥ 0.385 pg/mL) could reduce the need for lumbar puncture by 72%. Arranz et al. [35] validated the performance of CLEIA technology in a Spanish cohort (N = 389), and the AUC value of combined detection of p‐tau217, p‐tau181, and Aβ42/Aβ40 reached 0.94, providing higher accuracy for clinical diagnosis.
However, there are also reports that the core indicator p‐tau217/Aβ42 of CLEIA has poor stability, and certain false‐positive detection cases have occurred [45], requiring further optimization and verification of the technology.
In summary, ECLIA technology has completed international multicenter validation, showing strong negative exclusion ability, but its specificity is relatively insufficient, and the overall consistency with Aβ‐PET needs to be improved (especially in cognitively normal populations), and the overall diagnostic performance still needs further optimization; LiCA technology has been applied in multiple studies as a clinical research tool, and the detection results are highly consistent with Aβ‐PET, with excellent performance. Large‐scale external validation across multiple centers and ethnic groups is underway. After multicenter validation, CLEIA technology has good consistency between various AD‐related indicators and CSF detection results and shows potential for predicting AD progression.
However, there are reports that the core indicator p‐tau217/Aβ42 ratio has a high false‐positive rate, and the need for special blood collection tubes may increase the clinical use cost, and relevant technical details need to be improved.
Clinical Significance of Blood Biomarker Detection in Alzheimer's Disease
AD blood biomarker detection breaks the limitations of traditional AD diagnostic models and has important clinical significance in early AD screening, accurate diagnosis, disease assessment, epidemiological surveys, and new drug development, providing strong support for early diagnosis and treatment of AD.
Realizing Early AD Screening and Risk Stratification
Blood biomarkers can capture early signals of pathological changes in the preclinical stage of AD. Combined with high‐sensitivity detection technologies, they can accurately identify high‐risk AD populations, perform risk stratification, and implement personalized interventions [46]. For preclinical AD populations with positive biomarkers and normal cognition, lifestyle or drug interventions can delay the pathological process; for negative populations, regular follow‐up can reduce unnecessary medical interventions and psychological burdens, realizing precise prevention and control [47].
Assisting Accurate AD Diagnosis and Differential Diagnosis
Early AD symptoms are similar to those of other types of dementia and benign age‐related cognitive decline, making traditional diagnosis prone to misdiagnosis and missed diagnosis. p‐tau217 and Aβ42/Aβ40 ratio can effectively distinguish AD from other diseases. Combined with high‐specificity detection technologies, they can significantly improve the accuracy of differential diagnosis and provide support for formulating precise treatment plans for patients [13, 48].
Evaluating Disease Severity and Progression Rate
The concentration of AD blood biomarkers is highly correlated with disease severity and pathological progression. The concentrations of p‐tau217 and NfL increase with disease progression, and the concentration of GFAP reflects the degree of neuroinflammation. Through long‐term follow‐up detection, the changes in biomarker concentrations can be monitored in real time, the effect of intervention treatment can be evaluated, the treatment plan can be adjusted in a timely manner, and the treatment strategy can be optimized [49].
Promoting Large‐Scale Epidemiological Surveys of AD
Blood testing has the advantages of convenience, low cost, and high throughput, and can be used for large‐scale population‐based AD epidemiological surveys to accurately count the epidemiological characteristics of AD and provide data support for the formulation of prevention and control strategies [50]. At the same time, accurate identification of high‐risk populations through large‐scale screening and centralized medical resources for targeted interventions can optimize the allocation of medical resources and reduce the social burden [51].
Assisting in AD New Drug Development
A core difficulty in AD new drug development is the lack of accurate screening and efficacy evaluation indicators, leading to long clinical trial cycles and low success rates [52]. Blood biomarkers can be used as screening indicators for new drug clinical trials to accurately select participants who meet the trial conditions, improving the efficiency and success rate of clinical trials [53]; at the same time, they can be used as efficacy evaluation indicators to real‐time monitor the intervention effect of drugs on the AD pathological process, shorten the clinical trial cycle, and reduce R&D costs [54]. In addition, blood biomarkers can also be used to distinguish AD patients with different pathological subtypes, providing a basis for the personalized development of new drugs, promoting the clinical transformation of precision treatment drugs, and providing more new paths for AD treatment.
Prospects
AD blood biomarkers have initially formed a detection scheme dominated by core pathological biomarkers (Aβ42/Aβ40, p‐tau217, etc.), supplemented by neuronal damage biomarkers (NfL, GFAP, etc.) and genetic risk biomarkers (such as APOE ε4); detection technologies have also developed in a dual‐track manner based on the above schemes: generally divided into immunoassay platforms suitable for clinical detection (ECLIA, LiCA, CLEIA, etc.) and high‐throughput platforms suitable for scientific research and new biomarker discovery (mass spectrometry, Olink, SomoScan, etc.).
Although substantial progress has been made in the development of blood‐based biomarkers for Alzheimer's disease (AD), several challenges remain unresolved. Integrating blood biomarker testing with cognitive assessments in clinical practice may provide a promising approach to further enhance the efficiency of AD diagnosis and management.
Challenges and Future Directions of Blood‐Based Biomarkers for Alzheimer's Disease
Although blood‐based biomarkers have shown great potential for the early screening, diagnosis, disease monitoring, and clinical management of Alzheimer's disease (AD), several challenges remain before their routine clinical implementation. These challenges can be summarized at three levels: technical challenges, biomarker‐related challenges, and clinical application challenges.
First, at the technical level, the standardization and reproducibility of blood biomarker assays still need further improvement. Different detection platforms, such as ECLIA, LiCA, CLEIA, Simoa, mass spectrometry, Olink, and SomaScan, differ in analytical sensitivity, detection limits, antibody specificity, sample processing procedures, and calibration systems. These methodological differences may lead to inconsistent results across laboratories and clinical centers.
In addition, pre‐analytical factors, including blood collection tubes, anticoagulants, centrifugation protocols, storage temperature, freeze–thaw cycles, and sample transportation conditions, can also affect biomarker stability and measurement accuracy. Therefore, future research should focus on establishing unified pre‐analytical procedures, standardized assay calibration systems, external quality control programs, and internationally comparable reference ranges.
Second, at the biomarker level, the biological specificity and optimal combination of blood biomarkers remain important issues. Core pathological biomarkers, especially plasma p‐tau217 and Aβ42/Aβ40, are closely associated with amyloid deposition and tau pathology and have relatively high diagnostic value. However, some biomarkers, such as NfL, GFAP, inflammatory factors, complement proteins, exosomes, and miRNAs, may reflect broader pathological processes including neurodegeneration, neuroinflammation, immune activation, and systemic metabolic changes.
Their concentrations may be influenced by age, vascular risk factors, kidney function, systemic inflammation, and other neurological diseases. Therefore, single‐biomarker interpretation may not be sufficient for accurate diagnosis. Future studies should further clarify the temporal dynamics, disease‐stage specificity, and pathological relevance of different biomarkers, and develop optimized multi‐marker panels that combine core pathological biomarkers with neuronal injury, inflammatory, genetic, and peripheral metabolic indicators.
Third, at the clinical application level, the translation of blood biomarkers into routine practice requires further validation in diverse real‐world populations. Most current studies have been conducted in selected research cohorts or specialized memory clinics, whereas clinical practice involves patients with different ages, ethnic backgrounds, educational levels, comorbidities, disease stages, and cognitive profiles. Therefore, diagnostic cutoff values and clinical decision thresholds need to be validated in large‐scale, multicenter, and ethnically diverse cohorts.
In addition, blood biomarker testing should not be viewed as a complete replacement for clinical evaluation. Instead, it should be integrated with cognitive assessment, neurological examination, neuroimaging, CSF biomarkers, genetic risk assessment, and longitudinal follow‐up when necessary. In primary care settings, blood biomarkers may serve as an efficient tool for risk stratification and referral decision‐making, while in specialized clinics they may assist in differential diagnosis, disease staging, treatment selection, and therapeutic monitoring.
Overall, the future development of AD blood biomarkers should move toward standardized detection, biologically interpretable biomarker panels, population‐calibrated thresholds, and integrated diagnostic workflows. With further technological optimization and multicenter clinical validation, blood‐based biomarkers are expected to become an important component of precision diagnosis and management of Alzheimer's disease.
Integrating Blood Biomarkers With Cognitive Assessment to Enhance Diagnostic and Clinical Management Efficiency
Currently, the area under the curve (AUC) of plasma p‐tau217 for identifying Aβ positivity can reach 0.96. However, its accuracy in detecting Aβ pathology among cognitively normal older adults or individuals with subjective cognitive decline (SCD) remains relatively limited. To address this bottleneck in early screening, integrating blood‐based testing with cognitive assessment has been proposed as a comprehensive diagnostic strategy.
Three main approaches have been suggested.
First, the workflow optimization strategy. Unlike the traditional pathway in which cognitive assessment precedes laboratory testing, Brum et al. [55] proposed a stepwise strategy in which plasma p‐tau217 measurement is performed as the initial step. Individuals are stratified according to their probability of Aβ positivity.
Subsequently, high‐risk individuals receive therapeutic interventions, those at intermediate risk undergo cerebrospinal fluid (CSF) analysis or positron emission tomography (PET) imaging, and low‐risk individuals receive detailed neuropsychological assessment and follow‐up.
Second, the gatekeeping or threshold‐based strategy. Tideman et al. [56] reported that BioCog, a self‐administered digital cognitive test including a 10‐word list learning task, cognitive processing speed assessment, and time orientation evaluation, when combined with blood biomarkers, can efficiently identify AD‐related cognitive impairment in primary care settings. Establishing threshold values for blood testing allows biomarker testing to be performed within a defined intermediate range, thereby substantially improving the positivity rate and diagnostic accuracy.
This strategy provides an efficient and reliable solution for early diagnosis and referral for precision treatment of AD.
Third, the additive strategy. Several studies have shown that combining blood biomarkers with specific cognitive paradigms can improve early detection accuracy for AD. For example, Chen et al. [57] used MemTrax, a 2‐min working memory assessment; Lu et al. [58] employed the semantic intrusion score of the double‐list learning test; Jauregi‐Zinkunegi et al. [59] applied the serial position effect in the logical memory test (including the primacy effect in the first third and the recency effect in the last third); and Li et al. [60] incorporated visuospatial memory assessments.
These approaches significantly improved the accuracy of blood biomarkers for early AD screening.
Overall, integrating blood biomarkers with cognitive assessment paradigms represents a promising strategy to improve early screening efficiency and optimize diagnostic workflows for Alzheimer's disease in both primary care and specialized clinical settings.
Artificial Intelligence‐Assisted Biomarker Discovery and Multimodal Diagnostic Modeling
With the rapid expansion of proteomics, genomics, metabolomics, lipidomics, exosome profiling, and digital phenotyping, the challenge of AD biomarker research is gradually shifting from the discovery of a single novel marker to the selection, integration, and clinical interpretation of high‐dimensional biomarker information. Artificial intelligence and machine learning may help address this challenge by integrating established blood biomarkers, emerging molecular candidates, cognitive assessment, eye movement, speech and language features, gait parameters, wearable device data, neuroimaging, genetic risk factors, and clinical comorbidities. Such multimodal models may improve early risk prediction, identify individuals who require further PET or CSF confirmation, and support individualized disease monitoring.
However, AI‐assisted diagnostic models still require high‐quality clinical samples, standardized data collection, external validation, and prospective clinical testing. In particular, collaboration between biomarker researchers, clinicians, data scientists, and industry partners will be essential for developing robust, interpretable, and clinically feasible models. Therefore, future AD biomarker research should not only focus on discovering new molecular markers but also on building clinically meaningful multimodal prediction systems based on accessible biological, cognitive, behavioral, and digital indicators.
In addition to established core pathological biomarkers, emerging synaptic biomarkers may also be incorporated into AI‐assisted multimodal models. Synaptic proteins such as SNAP‐25, GAP‐43, VAMP‐2, synaptotagmin‐1, neurogranin, and neuronal pentraxins may provide information more directly related to synaptic dysfunction and cognitive decline. When combined with plasma p‐tau217, Aβ42/Aβ40, NfL, GFAP, cognitive assessment, neuroimaging, genetic risk factors, and digital behavioral indicators, these synaptic markers may help improve disease staging, progression prediction, and treatment‐response monitoring.
Therefore, the future value of synaptic biomarkers may lie not only in their independent diagnostic performance, but also in their contribution to integrated, interpretable, and clinically applicable risk prediction models.
Author Contributions
Shuangshuang Li: writing – original draft. Lin Huang: methodology, writing – review and editing. Qihao Guo: conceptualization, project administration, funding acquisition.
All authors have read and approved the final manuscript.
Funding
This work was supported by the National Natural Science Foundation of China (grant numbers 82171198 and 82501892) and the SJTU Trans‐med Awards Research Program (grant number 20250202).
Conflicts of Interest
The authors declare no conflicts of interest.