INTRODUCTION
The evaluation and treatment of individuals with cognitive impairment in clinical practice is changing rapidly. The identification of amyloid, the core pathology of Alzheimer’s disease (AD), has become easier after approval of 18F-labelled tracers for positron emission tomography (PET) [1]. Biomarkers in cerebrospinal fluid (CSF), such as total tau (t-tau), phosphorylated tau (p-tau), and amyloid-β42 (Aβ42), are elevated in patients with pathologically confirmed AD [2]. Blood-based biomarkers (BBMs), which were previously considered less accurate compared to CSF biomarkers [2], have recently shown marked development [3-5]. Moreover, anti-amyloid antibody treatments, such as lecanemab and donanemab, have been approved in many countries since 2023, making the biological diagnosis of AD essential in the clinical practice rather than optional for research purposes [6]. While clinicians previously predicted the cause of cognitive impairment based on clinical symptoms and manifestations reported by patients or informants, it is now necessary to identify the underlying pathology to decide on the optimal treatment. Meanwhile, anti-amyloid antibodies as disease modifying treatment can induce significant adverse events [7,8]. Therefore, clinicians need to be more vigilant to patients’ comorbidities and concomitant medications to minimize the unfavored results.
In this review, we summarized the major changes and clinical implications in the diagnostic criteria of AD, particularly in light of the recently revised criteria released in 2024 [9]. We also reviewed the fluid biomarkers (CSF biomarker and BBM) for AD that are currently or will soon be available in real-world clinical practice. Furthermore, we explored various scenarios that may arise when applying anti-amyloid treatment.
MAJOR CHANGES IN DIAGNOSTIC CRITERIA FOR AD
Changes in diagnostic criteria for AD-emphasis on confirming pathology
The diagnostic criteria for AD have shifted from reliance on clinical symptom observation in the 20th century to a biologically defined framework grounded in neuroimaging and fluid biomarkers in the 21st century. Since Alzheimer [10] first identified amyloid plaques and neurofibrillary tangles in the postmortem brain tissue of a patient in 1906, the medical community has long regarded “AD” and “AD dementia” as essentially synonymous concepts [11-17]. This was largely because there had been no means of directly confirming pathology in living patients, and diagnosis was possible only after typical clinical symptoms appeared at the dementia stage. A turning point came with the advancement of PET and CSF analyses, which enabled visualization and quantification of amyloid and tau pathology in the living brain. These technologies revealed that AD-related pathological changes begin at least 10-20 years before the onset of clinical symptoms [18]. This discovery spurred efforts to distinguish AD from dementia due to AD [19-22], and, with the subsequent demonstration of clinical efficacy for anti-amyloid antibody therapies, pathological confirmation has emerged as a central task not only in research but also in clinical practice. Consequently, the diagnostic paradigm of AD has been shifting its emphasis from clinical symptoms to pathological confirmation (Table 1).
Amyloid, tau, and neurodegeneration classification
The National Institute on Aging-Alzheimer’s Association (NIA-AA) criteria, released in 2011, explicitly recognized that the pathological process of AD begins years before dementia symptoms appear [12,15,23]. Importantly, biomarkers were incorporated into the diagnostic algorithm to facilitate early and more accurate detection. These biomarkers included amyloid PET imaging, CSF Aβ42, CSF t-tau and p-tau, glucose metabolism on fluorodeoxyglucose PET, and characteristic patterns of cortical atrophy on structural MRI [15,17]. Nevertheless, significant barriers to the use of biomarkers in clinical practice include a lack of standardization, variability in interpretation, and limited availability, leading the NIA-AA criteria in 2011 to not recommend biomarkers for routine diagnostic purposes.
Building on these advances, the 2018 NIA-AA research framework redefined AD as a biological construct, distinct from clinical syndromes, by introducing the AT(N) classification system [24]. Within this model, individuals are categorized according to biomarker status for amyloid (A), tau (T), and neurodegeneration (N), with each component rated as positive or negative (A+/A-, T+/T-, N+/N-). A central tenet of this framework is that individuals who are both amyloid- and tau-positive (A+T+) are classified as having AD, irrespective of clinical presentation. In contrast, those with amyloid positivity but no tau pathology (A+T-) are considered to fall within the “Alzheimer’s continuum,” reflecting a biologically defined disease process that may or may not yet manifest clinically [25]. Prior to the 2018 framework, when an individual was suspected to have dementia owing to AD based on past history and clinical symptoms, 2011 NIA-AA criteria regarded him or her having “probable AD dementia” or “possible AD dementia.” [15] However, under this framework, the diagnosis of AD is reserved only for cases where both amyloid and tau biomarkers are confirmed [24]. Cases exhibiting typical AD clinical symptoms but with unknown biomarker status are recommended to be labeled “Alzheimer’s clinical syndrome.” This framework is characterized by its emphasis on distinguishing between syndromes based on clinical symptoms and diseases based on biomarker status. However, this research framework was still intended to be used only in research, not in routine clinical practice.
Revised criteria for diagnosis and staging of AD by AA in 2024
Background for evolving revised criteria
With anti-amyloid antibody treatment gaining approval in multiple countries, starting with the United States and including South Korea, the need to standardize the diagnosis and staging of AD has grown significantly to ensure smooth communication in actual clinical practice. The accuracy of BBMs that detect AD-related pathologies, such as amyloid and tau, has improved remarkably. This also means that biologically diagnosing AD has become easier compared to existing methods like PET and CSF biomarkers. As existing brain imaging, CSF, and BBMs within the AT(N) category have become somewhat interchangeable, the need for updates has been raised.
Major changes in 2024 revised criteria
The 2018 criteria emphasized that they were a “research framework” intended solely for research purposes and not designed for clinical use [24]. However, the 2024 diagnostic criteria are intended for use in actual clinical practice [9]. The 2024 revised criteria defined A and T biomarkers as the “core biomarkers” for AD. These are subdivided further into Core 1 biomarkers, which change early in the disease course, and Core 2 biomarkers, which change relatively later. Notably, while 2018 criteria required both A and T biomarkers for diagnosing AD [24], revised 2024 criteria can diagnose AD using Core 1 biomarkers (representing A) alone [9]. In revised 2024 criteria, BBMs were recognized as AD biomarkers, enabling improved cost-effectiveness and accessibility in the biological diagnosis of AD. Furthermore, as in the 2018 research framework, the 2024 revised criteria clearly distinguished between biological stages (A-D) and clinical stages (1-6) to enable integrated use (see sections “Biological staging of AD,” “Clinical staging of AD,” and “Integration of biological and clinical staging”).
Biomarker categorization-A, T, N, inflammatory/immune mechanism, vascular brain injury, and alpha-synucleinopathy
The previous AT(N) biomarkers in 2018 have been expanded to include the categories of inflammatory/immune mechanisms (I), vascular brain injury (V), and alpha-synucleinopathy (S) (Table 2). T biomarkers have been further subdivided into T1 or T2 since the specific types increase at different stages of AD. T1 biomarkers are those that change at a time point similar to that of early-stage amyloid PET in AD (p-tau217, 181, and 231). T2 biomarkers are tau that change at a time point similar to tau PET, i.e., at a relatively later stage (e.g., microtubule binding region [MTBR]-tau243).
Core biomarkers represent key indicators of neuropathological changes in AD, encompassing the A and T categories (Table 2). They are subdivided into Core 1 and Core 2 biomarkers based on the timing of abnormality and intended use [9]. Core 1 biomarkers are those that change at a time point similar to early-stage amyloid PET in AD (A, T1, hybrid combination). Core 2 biomarkers are those that change relatively later compared to Core 1 (tau PET and biofluid T2). The detection of abnormalities in Core 1 biomarkers allows for a diagnosis of AD, irrespective of clinical symptoms. However, not all Core 1 biomarkers are yet sufficiently accurate for routine clinical use.
Clinical implication of biomarker use
The results of AD biomarker tests should always be interpreted within the appropriate clinical context, as current biomarkers still have important limitations (e.g., lack of validated biofluid biomarkers, lower sensitivity and accuracy compared to neuropathological confirmation). Although biomarker testing could theoretically allow the diagnosis of AD in asymptomatic individuals, this revised criteria clearly emphasize that such testing should not be used for clinical purposes in asymptomatic patients at the present time [9].
Coexisting pathologies beyond AD are common, so clinical judgment is always required to answer the question: “Can AD fully explain this patient’s symptoms?” For instance, if a patient presents with Parkinsonian symptoms and hallucinations and tests positive for Core 1 biomarkers, the clinician must determine whether the cognitive decline arises from AD or from neuronal synuclein disease [26]. In such situations, Core 2 biomarker testing can provide valuable clarification. A positive Core 2 biomarker result indicates a substantial contribution of Alzheimer’s pathology to the clinical picture, whereas a negative result suggests a limited role. Clinical judgment is equally essential when Core 1 biomarker findings do not align with the clinical presentation. For example, if the clinical features strongly suggest AD but the Core 1 biomarker is negative, further testing is required to resolve the inconsistency.
Biological staging of AD
According to the 2024 revised criteria, patients diagnosed with AD using Core 1 biomarkers can have their biological staging (A-D) determined based on amyloid and tau PET results (Table 3) [9]. Biological staging based on fluid biomarkers is currently defined as “conceptual” and is only available for research purposes; it cannot be used in clinical practice. Revised 2024 criteria workgroup judged that biofluid biomarkers such as p-tau205, MTBR-tau243, and non-phosphorylated tau (np-tau) fragments as early, intermediate, and advanced-stage fluid markers required further validation for routine clinical use. In South Korea, neither tau PET nor fluid biomarkers are currently available for clinical use (only CSF fluid Core 1 biomarkers are available, see section “Regulatory approval status and future direction of CSF biomarkers”), making biological stage classification limited.
Clinical staging of AD
Clinical staging of AD is defined as stages 0 through 6 (Table 4) [9]. This is largely consistent with what was first presented in the 2018 research framework [24]. While similar to the Global Deterioration Scale [27], this clinical staging differs in that it was developed after the concept of disease-specific AD biomarkers was established. Compared to existing clinical stage terminology, subjective cognitive decline [28] corresponds to stage 2, mild cognitive impairment (MCI) [12] to stage 3, and mild, moderate, and severe dementia to stages 4, 5, and 6, respectively. Currently, lecanemab is approved for use in patients with MCI or mild dementia (with confirmed amyloid pathology) [29], corresponding to clinical stages 3 and 4.
Integration of biological and clinical staging
Patients diagnosed with AD using Core 1 biomarkers can be classified by combining their clinical stage (1-6) and biological stage (A-D). The clinical stage is listed first, followed by the biological stage (e.g., stage 2C). As shown in Table 5 [9], the two stages do not always correspond: advanced clinical symptoms (e.g., stage 6A) may coexist with an early biological stage. This reflects the principle of the revised 2024 criteria distinguishing biological pathology from clinical syndrome [9]. More severe clinical symptoms than biological stage suggest other pathologies additional to AD, whereas milder symptoms relative to biological stage may indicate greater cognitive reserve [30].
International Working Group recommendation on revised AA criteria
According to the AA criteria, AD is diagnosed based solely on biomarker abnormalities [9]. In contrast, the International Working Group (IWG) conceptualized AD as a “clinical-biological construct” rather than a merely biological phenomenon [31]. The IWG articulated apprehensions regarding the AA’s criteria that label cases as “AD” solely based on biomarker positivity regardless of symptoms, recommending the diagnosis only when biomarker positivity is confirmed alongside cognitive decline. While the AA also advised caution regarding biomarker testing in asymptomatic individuals, the IWG places even greater emphasis on this point. The diagnosis of AD in asymptomatic biomarker-positive individuals could be potentially problematic, as it is unclear when symptoms might emerge. Individuals who are cognitively normal yet exhibit biomarker positivity may live their entire lives without developing dementia. The diagnosis of such individuals offers no benefit when considering the social and psychological consequences.
Application of the revised criteria in clinical practice: recommendations from the Korean Association for Geriatric Psychiatry
The revised 2024 diagnostic criteria from the AA placed greater emphasis on the biological definition of AD than previous versions. This shift reflects the growing accessibility of biomarkers and the development of therapies targeting them. However, the IWG opposes this biomarker-only framework proposed by AA, recommending that the presence of clinical symptoms (cognitive impairment) remain essential for the diagnosis of AD. Moreover, in real-world clinical practice in South Korea, the use of Core 1 or Core 2 biomarkers remains limited. PET and CSF tests are costly or invasive, and among the Core 2 biomarkers, tau PET and amyloid/tau BBMs are currently available only for research purposes. Therefore, at present, the Korean Association for Geriatric Psychiatry (KAGP) currently recommends that patients with cognitive symptoms be evaluated for clinical severity of cognitive decline using Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR) criteria [29,32], with biomarker testing considered only when AD is suspected as the underlying cause.
FLUID BIOMARKERS FOR AD
CSF biomarker
Comparison with PET biomarker
As well as a PET scan, the CSF test can accurately reflect the pathology of the brain, including Aβ and tau. As the CSF test examines the body fluid surrounding the brain directly, it is known to reveal abnormal increases at an earlier stage of AD than PET scans [33]. The pros and cons of the CSF test compared to PET scans are presented in Table 6.
Major CSF biomarkers for diagnosis and staging of AD
Several CSF biomarkers are used for identifying AD-related pathology. CSF Aβ42/40 ratio is known to reflect the early cerebral amyloid deposit. Aβ42, the isoform of Aβ comprising 42 amino acids, is prone to aggregate more readily than Aβ40 and to form Aβ oligomers, fibrils, and neuritic plaques [34]. Aβ40 is the most abundant isoform and is more soluble than Aβ42 [35]. In AD, pathologic Aβ42 remains aggregated in the brain parenchyma and not secreted to CSF, probably through glymphatic system, resulting in the reduced level of Aβ42 in CSF [36]. Thus, a decreased CSF Aβ42/40 ratio demonstrates a high degree of concordance (over 90%) with amyloid PET positivity [34,36,37].
CSF t-tau is the first biomarker reported for tau pathology in AD [38]. Although the level of CSF t-tau is higher in AD patients than in healthy controls, it is also increased in other neurodegenerative diseases (e.g., frontotemporal dementia, dementia with Lewy bodies, and vascular dementia) [39] and in acute brain injuries that are not clearly related to tauopathy (e.g. Creutzfeldt-Jakob disease [40], traumatic brain injury [41], and ischemic stroke [42]). Due to this lack of clarity surrounding ttau, which can increase in both AD-related neurodegeneration and non-specific brain damage, the revised criteria from the AA did not include t-tau in either T or N categories [9].
CSF p-tau is thought to indicate the pathologic form of tau [39], as hyperphosphorylation results in the aggregation of tau and the formation of neurofibrillary tangles [43]. Following the initial CSF p-tau study was presented in 1995 by Blennow et al. [44], CSF p-tau181, tau phosphorylated at threonine 181, has been studied extensively for biomarker of AD [45,46]. However, recent studies reported that p-tau217 is more accurate than p-tau181 for AD diagnosis such as amyloid and tau PET positivity [47,48]. Other observational studies identified that CSF p-tau231 levels increased in the very early stage of AD at a similar time to, or faster than, p-tau217 [49,50]. Although they are tau biomarkers, p-tau181, p-tau217, and p-tau231 are observed to rise during the early stages of AD, when Aβ begins to deposit and amyloid PET scans become positive. Meanwhile, CSF p-tau205 is known to increase in later stage of AD, together with t-tau, when neuronal dysfunction occurs [51,52].
MTBR-tau243 is a recently identified CSF biomarker that shows an abnormal increase in a late stage of AD and tauopathy [53-55]. According to Horie et al. [54], CSF p-tau181, p-tau217, and p-tau231 levels were associated with amyloid PET rather than tau PET, while CSF MTBR-tau243 levels were associated with tau PET rather than amyloid PET. Similarly, CSF MTBR-tau243 was linearly associated with tau PET, while CSF p-tau217 exhibited a steep increase during the early Braak stage of tau PET, eventually forming a relative plateau [54].
CSF np-tau, different from t-tau that includes both np-tau and p-tau, serves as a biomarker that represents tau tangle formation and neurodegeneration in a later stage of AD [56].
Sequence of CSF biomarker abnormality by AD progression
Based on the incorporation of observational studies, CSF biomarkers seem to become abnormal in the following sequence as AD progresses: CSF Aβ42/40 ratio, p-tau217 (or p-tau181 or p-tau231), p-tau205, MTBR-tau243, and np-tau [49,50,56]. The former biomarkers represent amyloid pathology, and the latter represent tau pathology.
Regulatory approval status and future direction of CSF biomarkers
Three hybrid combination biomarkers, including CSF Aβ42/40 ratio, t-tau/Aβ42 ratio, and p-tau181/Aβ42 ratio, have received regulatory approval from the United States Food and Drug Administration (FDA) and European Medicines Agency [50]. These are also Core 1 biomarkers in the revised criteria by AA (Table 2). Of those, the Korean Ministry of Food and Drug Safety approved Roche’s Elecsys Aβ42 and p-tau181 in 2023, and t-tau in 2024.57,58 As anti-amyloid antibody therapies are introduced, confirming AD pathology is now clinically essential. CSF testing is therefore expected to become more common among patients who are burdened by PET scans and uncomfortable with radiation exposure. The regulatory approval status of CSF and PET biomarkers in Korea, the United States, Europe, and Japan is presented in Table 7.
BBM
Comparison with PET and CSF biomarker
PET and CSF biomarkers demonstrate high accuracy in detecting brain amyloid or tau pathology. However, PET imaging is costly and requires specialized equipment (PET scanner) for each clinic. CSF biomarker testing requires a lumbar puncture, which is invasive. BBMs, on the other hand, do not need specialized equipment and are less invasive, making them more accessible than PET or CSF biomarkers. Although attempts to detect AD pathology in plasma or serum via conventional technique (such as ELISA or xMAP) showed mixed results [2], recent studies presented high accuracy comparable to PET or CSF [3,59,60]. This development is owing to the introduction of new assay techniques, such as immunoprecipitation mass spectrometry (IP/MS), single molecule array (Simoa), electrochemiluminescence immunoassay, and chemiluminescent enzyme immunoassay [37,61,62].
Major BBMs for diagnosis and staging of AD
Plasma Aβ42/40 ratio can identify the early cerebral amyloid deposition, as can the CSF Aβ42/40 ratio. It is known to convert abnormal earlier than amyloid PET in the course of AD progression [63]. However, when comparing amyloid PET (+) and (-) cases, CSF Aβ42/40 ratio decreased by approximately 50%, whereas plasma Aβ42/40 decreased by only about 10%-15% [64]. This smaller fold change in plasma Aβ42/40 ratio leads to the relatively lower diagnostic accuracy for confirming the presence or absence of cerebral amyloid pathology.
Plasma p-tau217 is a tau isoform that is elevated in the presence of early amyloid deposition and can even reflect the presence of later tau pathology [9,60,65,66]. To date, it is the most accurate BBM for detecting amyloid PET positivity [66,67]. Unlike plasma Aβ42/40 ratio, for which there is only a 10%-15% difference between amyloid PET (+) and (-) individuals, p-tau217 shows a difference of 300%-700%, making it a more suitable diagnostic tool for AD [64,68]. Like amyloid PET, plasma p-tau217 is a strong predictor of progression to MCI in cognitively unimpaired individuals [69]. Recent studies indicate that the ratio of plasma p-tau217 to np-tau217 (%p-tau217) is more accurate than p-tau217 alone, and that the accuracy of %p-tau217 is not affected by age, unlike p-tau217 [70]. Plasma %p-tau217 can be equal or even superior to FDA-approved CSF biomarkers for detection of amyloid PET positivity [60]. Other studies suggest that plasma p-tau217/Aβ42 ratio may be more accurate than p-tau217 alone in detecting amyloid pathology [71-73]. In light of this high accuracy, plasma p-tau217 is considered as a Core 1 biomarker in the revised criteria of AD from AA [9]. In addition, plasma p-tau217 has potential to reflect later tau pathology and cognitive impairment [74,75]. This indicates that plasma p-tau217 could be used as a staging biomarker of AD.
Plasma p-tau181 is a biomarker that has been studied earlier than other biomarkers, as well as CSF p-tau181. It has demonstrated high accuracy in detecting amyloid and tau pathology [76]. However, compared with plasma p-tau217, plasma p-tau181’ s accuracy to detect amyloid PET positivity is lower [67]. Plasma p-tau231 is known to be elevated in the early stage of amyloid deposition [65,77]. After increasing in the early stage, it remains at a plateau in the later stage of AD [78].
Plasma p-tau205 and endogenous MTBR-tau243 (eMTBR-tau243) reflect tau pathology in the later stage of AD [65,79]. Especially, a recent study demonstrated that plasma eMTBRtau243 showed strong association with tau PET uptake rather than amyloid PET uptake, outperforming other plasma tau biomarkers including p-tau217 [80]. Plasma p-tau205 and eMTBR-tau243 are Core 2 biomarkers which can be used to represent AD-related tauopathy [9].
Neurofilament light chain (NfL) is a subunit of axonal neurofilament which is released into CSF or blood during neurodegeneration or axonal injury. As a “N (neurodegeneration)” biomarker in the revised criteria for AD [9], increased plasma NfL can predict future cognitive decline or hippocampal atrophy [81]. Plasma NfL increases not only in AD but also in other CNS-related diseases such as frontotemporal dementia, vascular dementia, HIV-associated dementia, amyotrophic lateral sclerosis, Creutzfeldt-Jakob disease, and parkinsonian disorders [82,83].
Glial fibrillary acidic protein (GFAP) is a component of the cytoskeleton in astrocytes. Plasma GFAP is an indicator of the degree of neuroinflammation since the number and size of astrocytes increase due to inflammatory or immune responses. GFAP is an “I (inflammation)” marker in the revised criteria for AD [9]. It does not increase specifically in Alzheimer’s pathology and, like NfL, also increases in other degenerative brain diseases such as frontotemporal dementia or Lewy body dementia [82].
Two cut-off approach using BBMs
When using a single cut-off value to determine amyloid positivity via BBM, some patients will have BBM levels close to the threshold. Since measurements may vary slightly between repeat tests for the same patient, if a measurement is near the cut-off threshold, the result could change if repeat testing is performed. To minimize the risk of false positive or false negative results, Global CEO Initiative on Alzheimer’s Disease BBM workgroup recommended using two cut-off approach to classify BBM tests into three categories: positive, intermediate, and negative (Figure 1) [3,4,84]. Patients with a “positive” BBM result are interpreted as having amyloid pathology, while those with a “negative” result are interpreted as having no amyloid pathology. Patients with an “intermediate” BBM result may consider confirmatory amyloid PET or CSF testing, or repeat BBM testing after 1 year [4]. When using two cutoff approach, the BBM workgroup recommended that the BBM should be accurate enough to minimize the number of individuals with an “intermediate” result to no more than 15%-20% of the entire test population.
Application of BBMs in real-world clinical settings
BBMs can be used as a triage test to determine the presence or absence of amyloid pathology and to screen candidates for anti-amyloid antibody treatment [3,5]. As a triage test, the BBM test can produce two results: a negative result rules out AD pathology with high probability, while a positive result must be confirmed by amyloid PET or CSF tests [5]. Patients with cognitive symptoms can undergo a clinical evaluation, including a physical examination, history taking, and neuropsychological testing. If AD is suspected as a result of the initial assessment, BBM testing alongside a brain MRI or CT scan can be further assessed. BBM test results require CSF or PET tests to confirm amyloid pathology. If amyloid pathology is confirmed, anti-amyloid antibody treatment may be considered [3]. The BBM Workgroup organized by the Global CEO Initiative on AD recommended that, in secondary care, BBM tests need to be conducted under the following conditions: objective cognitive impairment, suspected AD pathology after comprehensive assessment, and expected improvement in management of cognitive impairment [4]. In primary care, since younger patients in primary care usually have a low chance of amyloid pathology, this BBM Workgroup recommended using BBM as a triage test in primary care only for those who are 55 years old or older [4]. The Alzheimer’s Association Clinical Practice Guideline similarly recommended the use of BBM only when knowing whether AD pathology is useful for shared decision making with the patient in secondary care or specialized memory clinic where memory disorder specialists are involved [5].
BBMs can be used as a confirmatory test in which a negative result excludes amyloid pathology, and a positive result confirms AD pathology without the need for further confirmatory tests. After physical examination, history taking, and neuropsychological testing in patients with cognitive symptoms, a confirmatory BBM test is available if AD is suspected. When amyloid pathology is confirmed by BBM, anti-amyloid antibody treatment can be considered [3]. The recommendation from the BBM Workgroup organized by the Global CEO Initiative on a confirmatory BBM test in a secondary care is same as a triage test [4]. In primary care, this BBM Workgroup recommended use of BBM for confirming amyloid pathology for those aged 65 or older [4]. The Alzheimer’s Association Clinical Practice Guideline did not distinguish between confirmatory and triage tests in terms of clinical use recommendations. Regardless of its purpose (triage or confirmation), BBM may be inappropriate if knowing whether AD pathology is present would be of little utility [5]. This utility of the test could be either patient preference, diagnostic or prognostic value, or the treatment decision-making. The Alzheimer’s Association Clinical Practice Guideline only focused on secondary care settings, due to need for gathering further evidence in primary care [5].
BBMs can also be used to monitor treatment response of anti-amyloid antibody treatment. Lecanemab elevated plasma Aβ42/40 ratio and reduced p-tau181 and GFAP in phase 3 clinical trials [7]. Donanemab reduced plasma p-tau217 [8]. After successfully reducing amyloid with antibody therapy, periodic BBM scans can be used to determine if amyloid pathology is re-accumulating, and then discuss whether to restart antibody therapy.
Required accuracy of BBM for detecting amyloid pathology
According to the revised diagnostic criteria by the AA in 2024, an “accurate” BBM eligible for inclusion in Core 1 biomarkers must have an amyloid PET positive detection accuracy of 90% or higher in intended-use population, which is equivalent to CSF biomarkers [9]. Accuracy is defined as (true positive+true negative)/(true positive+true negative+false positive+false negative) at a given threshold or cut-off.
In July 2024, the BBM Workgroup by the Global CEO Initiative on AD recommended that the minimum performance of a triage BBM test should be a sensitivity of 90% and a specificity of 85% in primary care, and 75%-85% in secondary care. As a confirmatory test, the BBM test should have sensitivity and specificity of at least 90% [4].
In the Alzheimer’s Association Clinical Practice Guideline revealed in July 2025, the performance of BBM detecting amyloid pathology should be at least 90% sensitivity and a 75% specificity to be used as a triage test [5]. This is intended for limited use in specialized memory clinics, where the prevalence of amyloid-positive patients is expected to be higher than in primary care settings. To perform as a confirmatory test, the BBM test requires a minimum sensitivity and specificity of both 90% [5]. The BBM tests that currently meet the criteria are listed in Table 8.
Regulatory approval status of BBMs
In May 2025, FDA cleared the Lumipulse G p-tau217/Aβ42 Plasma Ratio with two cut-off approach as the first in vitro diagnostic device to aid the diagnosis of AD [85]. Additionally, various BBMs are under FDA’s Breakthrough Device Program, which can accelerate the approval process (Table 9) [86]. Currently, BBM is being used in clinical trials for monitoring the response of anti-amyloid antibody treatments such as lecanemab, donanemab, and trontinemab [7,8,87]. Beyond those listed in Table 9, BBMs meeting criteria suggested by the Alzheimer’s Association Clinical Practice Guideline or the Global CEO Initiative could be used in future clinical trials. The revised diagnostic criteria in 2024 stated that amyloid PET detection with an accuracy of 90% or higher can be used for AD diagnosis and biological stage assignment [9].
SPECIAL CONSIDERATIONS
Our previous guideline publication addressed the fundamental principles of patient selection for anti-amyloid antibody therapy, including inclusion and exclusion criteria and major comorbidities relevant to amyloid-related imaging abnormalities (ARIA) monitoring [29]. Those recommendations remain essential for routine practice. In the present review, we extend this framework by exploring issues that were only briefly mentioned or not sufficiently elaborated previously, such as long-term maintenance strategies, genotype-specific dosing, and the nuanced challenges faced by individuals with Down syndrome or serious mental illness. Drawing upon the Practical Guide of the KAGP to Anti-Amyloid Monoclonal Antibody Therapy for Alzheimer’s Disease (2025) and newly available clinical evidence, we aim to provide a more detailed discussion of these special considerations and to highlight additional dilemmas likely to emerge in real-world settings.
Long-term maintenance therapy after 18 months
In phase 2 study 201 and its open-label extension, 18 months of lecanemab treatment led to robust reductions in amyloid PET burden. When treatment was discontinued, however, amyloid re-accumulated, with approximately 21% of the reduction lost by 24 months [88]. These findings indicate that while lecanemab is highly effective at reducing amyloid, continuous dosing is likely required to sustain low amyloid levels.
Pharmacokinetic and pharmacodynamic modeling based on the extension data further suggested that monthly intravenous dosing could maintain these reductions after the induction phase. On January 2025, the U.S. FDA approved a maintenance regimen of intravenous infusion once every 4 weeks for patients completing the 18-month biweekly course [89]. Thus, both every 2 weeks and every 4 weeks of intravenous dosing are now considered viable options for maintenance therapy, with local regulatory review expected in South Korea. Moreover, the U.S. FDA approved weekly subcutaneous dosing for maintenance treatment in August 2025 [90]. Regarding that a weekly subcutaneous dosing as an initial treatment is in the course of approval review after being granted fast track status [91], more convenient methods will be available in the near future.
For comparison, donanemab followed a different maintenance strategy. In the TRAILBLAZER-ALZ 2 trial, donanemab was administered every 4 weeks, and patients underwent amyloid PET every 6 months. When plaques were reduced to minimal levels, treatment was switched to placebo, reflecting that its therapeutic target is amyloid plaque itself [8]. The FDA similarly advises that donanemab may be discontinued in practice once amyloid PET confirms clearance, with biomarkers such as blood-based measures used to monitor for re-accumulation. Together, these findings illustrate two distinct approaches: lecanemab considering continuous maintenance to prevent rebound, and donanemab adopting a biomarker-driven discontinuation paradigm.
APOE ε4 homozygotes and dosing strategies
APOE ε4 homozygotes are known to have a markedly elevated risk of ARIA compared with noncarriers, a fact consistently demonstrated across anti-amyloid antibody trials [92]. Accordingly, although various dosing strategies have been considered to mitigate this risk, no definitive guideline has yet been established.
The TRAILBLAZER-ALZ 6 randomized trial tested a modified titration schedule of donanemab against the standard regimen. Patients in the titration group experienced significantly fewer ARIA-E events (15.6% vs. 24.2%), and radiographic severity was lower, while amyloid reduction and overall safety outcomes remained comparable between the groups [93]. Although the published report did not provide detailed subgroup analyses for APOE ε4 homozygotes, earlier observations suggest that this population may derive particular benefit from titration. Nevertheless, in the absence of definitive trial data, it remains uncertain whether modified dosing strategies reliably mitigate the excess risk in ε4 homozygotes.
For lecanemab, recent appropriate use recommendations and updated safety analyses have underscored the increased risk of ARIA in APOE ε4 homozygotes, but no titration-based dosing strategies have been proposed to date [92,94]. Ongoing prevention trials such as AHEAD 3-45 are exploring alternative dosing regimens, including lower or less frequent administration in cognitively unimpaired individuals, but these studies are not specifically designed to establish genotype-based titration strategies [95]. Due to the emergence of ARIA in the early treatment phase, the U.S. FDA recently (in August 2025) recommended an additional MRI monitoring before 3rd IV infusion, whereas the pre-existing recommendation was for MRI scans before 5th, 7th, and 14th infusions [96]. Future studies will find the risk factors of ARIA and the impact of dose adjustment in this high-risk group.
Down syndrome: at risk of being overlooked in treatment planning
The vast majority of individuals with Down syndrome develop amyloid pathology by mid-adulthood, and AD typically manifests at an earlier age compared with sporadic cases [97]. Comorbidities such as vascular risk factors, epilepsy, and autoimmune conditions are also more prevalent, complicating both diagnosis and therapeutic decision-making [98,99].
Despite this elevated risk, people with Down syndrome have been almost universally excluded from clinical trials of disease-modifying therapies for AD. As a result, there is little direct evidence on the safety or efficacy of monoclonal antibodies in this population. Concerns about heightened vulnerability to adverse effects, including ARIA or infusion-related reaction, must therefore be balanced against the ethical imperative to provide access to potentially disease-modifying treatment.
Systemic barriers and diagnostic overshadowing can further limit access to care. Intellectual disability and psychiatric comorbidities may delay recognition of dementia symptoms, leading to missed opportunities for timely diagnosis and intervention. These challenges underscore the need for tailored screening and treatment guidelines and dedicated research efforts. In the meantime, clinicians should adopt a cautious, individualized approach, emphasizing shared decision-making with families and caregivers.
Serious mental illness: challenges in treatment access and decision-making
Patients with serious mental illness, such as schizophrenia and bipolar disorder, pose unique challenges when considering anti-amyloid antibody treatment. While many individuals remain clinically stable for extended periods, episodes of symptom exacerbation can impair judgment and decision-making capacity, complicating both the recognition of cognitive decline and adherence to treatment protocols. In routine clinical settings, these issues may not be readily apparent, as psychiatry lacks objective laboratory markers for disease activity. Close collaboration with psychiatrists is therefore essential to ensure accurate assessment of mental status, evaluate treatment readiness, and anticipate fluctuations in capacity over time.
People with severe psychiatric disorders have also been disproportionately excluded from AD clinical trials, leading to a paucity of evidence on the safety and efficacy of monoclonal antibodies in this population. This exclusion may inadvertently reinforce disparities, particularly given the elevated burden of vascular and metabolic comorbidities that already increase dementia risk in these patients [100]. From an ethical perspective, clinicians face a dual challenge. On one hand, there are legitimate concerns regarding patients’ ability to provide ongoing informed consent, maintain adherence, and tolerate adverse events such as ARIA or infusion-related reactions. On the other hand, excluding patients solely because of a psychiatric diagnosis risks deepening existing inequities in care, particularly when individuals retain clear decision-making capacity. In such cases, opportunities for disease-modifying therapy should not be denied automatically but instead considered through individualized assessment and shared decision-making, with active involvement of caregivers and psychiatrists.
Other uncertainties: clinical dilemmas likely to arise in clinical practice
Despite accumulating trial data, several important clinical dilemmas remain insufficiently addressed by current guidelines. These scenarios were rarely represented in clinical studies, where exclusion criteria limited the inclusion of medically complex patients, but they are almost certain to emerge in real-world practice.
One such situation involves the new diagnosis of malignancy during anti-amyloid antibody therapy. In the absence of trial evidence, oncologic treatment is generally prioritized, with antibody therapy suspended while systemic disease is addressed. While this pragmatic approach avoids competing toxicities, there is no clear evidence on whether and when therapy can be safely restarted. In addition, many cancer therapies can alter hematopoietic function or increase bleeding risk, which may further compound the potential for ARIAH. The growing use of immuno-oncologic agents also raises the possibility of unforeseen pharmacologic or immunologic interactions, an area where virtually no data currently exist.
Acute vascular events raise even more urgent challenges. In myocardial infarction or ischemic stroke, reperfusion therapies such as percutaneous coronary intervention, anticoagulation, thrombolysis, or endovascular procedures must be initiated without delay, given the critical importance of time to treatment. One published case has documented fatal intracerebral hemorrhage in a patient on lecanemab who received thrombolytic agent for stroke, illustrating the risk [101]. Yet emergent reperfusion therapy cannot be withheld in golden-time emergencies, leaving clinicians in unavoidable dilemmas. Similar concerns extend to atrial fibrillation newly diagnosed during lecanemab therapy, where the need for anticoagulation conflicts with current recommendations to avoid therapeutic anticoagulation in patients receiving anti-amyloid antibodies. A recent case report described management of such a situation through left atrial appendage closure as an alternative to long-term anticoagulation [102]. These examples illustrate how real-world practice will repeatedly confront high-stakes decisions in the absence of trial data. Ongoing accumulation of clinical experience and registry-based evidence will be essential to inform future guidance.
DISCUSSION
In the emergence of disease-modifying treatments, this review focused on the major changes in the field of AD: diagnostic criteria, biomarkers for AD pathology, and potential scenarios for anti-amyloid monoclonal antibody treatments. Interpretation of biomarker results in combination with clinical presentation became much more important regarding the various treatment options. Clinicians who care for patients with cognitive symptoms need to know the recent paradigm shift of AD diagnosis and treatment.
According to the revised 2024 criteria by AA, AD is not a clinical syndrome anymore, but a biologically defined disease [9]. According to these criteria, asymptomatic individuals with biomarker positive are diagnosed to have AD. However, the psychological and ethical burden and the cost-benefit ratio of the diagnosis imposed on them remain uncertain. Therefore, there are concerns, notably raised by IWG, that referring to the preclinical stage as a “diagnosis” is problematic [31,103]. Actually, the effect of anti-amyloid antibody treatment, such as lecanemab and donanemab, on asymptomatic individuals is being investigated in clinical trials [6,95,104]. Routine clinical use of biomarker for AD diagnosis is not recommended until the results of these trials have been fully analyzed and the performance of the biomarkers is guaranteed.
CSF biomarkers are the gold standard surrogate for the diagnosis of AD. In real-world clinical practice in South Korea, CSF biomarkers remain the only way to evaluate the stage of AD via tau pathology, as tau PET is not yet available for routine clinical use (Table 7). Nevertheless, lumbar puncture is a relatively invasive, time-consuming, and requires a skillful physician supported by a well-trained assistant. Alternatively, BBMs are accumulating growing evidence to serve as a standalone marker for the diagnosis of AD. With the recent approval of the Lumipulse G p-tau217/Aβ42 plasma ratio by U.S. FDA (the first BBM to receive such authorization), multiple BBMs are expected to be implemented in routine clinical practice. However, BBM levels and performance can be affected by comorbidities such as cardiac disease and kidney function [81,105]. Moreover, the prevalence of amyloid pathology varies across races and ethnicities, which may influence predictive value [106]. Once these obstacles are addressed, BBMs can be adopted not only in dementia care setting but also in general health check-up examination in the future.
The introduction of anti-amyloid monoclonal antibodies, such as lecanemab and donanemab, into routine clinical practice highlights the increasing complexity of treatment decision-making in AD. Clinicians must now consider concomitant use with conventional symptomatic drugs, potential drug-drug interactions, and patient-specific contexts. While there is currently no clinical study evidence to support a contraindication, agents such as ginkgo biloba, selective serotonin reuptake inhibitors, and antipsychotics may raise theoretical concerns and require careful and sufficient monitoring. Beyond pharmacological considerations, ethical and clinical dilemmas remain regarding the use of lecanemab in asymptomatic individuals. Although the treatment paradigm may change based on the results of lecanemab and donanemab clinical trials in asymptomatic cognitively unimpaired individuals [6,95,104], the use of anti-amyloid monoclonal antibodies in this population is currently not recommended. Special populations, including adults with Down syndrome, seizure disorders, or autoimmune diseases, may represent unique candidates, yet require further evidence to guide safe implementation. These nuanced scenarios emphasize that while lecanemab offers a landmark treatment option, its integration into diverse real-world clinical settings demands careful balancing of efficacy, safety, and patient-centered care. As maintenance regimens by monthly dosing and subcutaneous injection rather than intravenous infusion are recently introduced [90], treatment options will be more available in future clinical settings.
In conclusion, the diagnosis of AD is now shifting from clinical symptom-oriented to biomarker-oriented, despite caution surrounding the use of biomarker alone. The emergence of anti-amyloid monoclonal antibodies as a disease-modifying treatment will facilitate the utilization of biomarker testing, especially BBMs, in the routine clinical settings. Guideline for the special but common cases will be updated based on evidence-based observational studies, which clinicians should follow closely.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
