Alzheimer disease blood biomarkers: considerations for population-level use

Abstract

In the past 5 years, we have witnessed the first approved Alzheimer disease (AD) disease-modifying therapy and the development of blood-based biomarkers (BBMs) to aid the diagnosis of AD. For many reasons, including accessibility, invasiveness and cost, BBMor neuroimaging. However, many questions remain regarding how best to utilize BBMs at the population level. In this Review, we outline the factors that warrant consideration for the widespread implementation and interpretation of AD BBMs. To set the scene, we review the current use of biomarkers, including BBMs, in AD. We go on to describe the characteristics of typical patients with cognitive impairment in primary care, who often differ from the patient populations used in AD BBM research studies. We also consider factors that might affect the interpretation of BBM tests, such as comorbidities, sex, and race or ethnicity. We conclude by discussing broader issues such as ethics, patient and provider preference, incidental findings, and dealing with indeterminate results and imperfect accuracy in implementing BBMs at the population level.

Introduction

Owing to rising life expectancies, we are seeing a global increase in the number of individuals aged 60 years and older, who currently comprise approximately 12% of the population¹. Because age is the strongest risk factor for developing Alzheimer disease (AD) and related dementias (ADRD), the number of people diagnosed with AD/ADRD is also expected to increase worldwide, from an estimated 57.4 million people in 2019 to more than 152 million in 2050² Notably, ageing is not homogeneous across the world’s population. High-income countries (HICs) that already have long life expectancies will not experience as dramatic an increase in AD/ADRD as low-income and middle-income countries (LMICs), which are facing rapid increases in life expectancies. The epidemic increase of AD/ADRD with a global ageing population has led to an intensified effort to promote early detection and diagnosis, to identify appropriate pharmacological and nonpharmacological treatments and to prevent or delay disease onset.

AD is a progressive neurodegenerative condition characterized clinically by amnestic memory impairment and impairment in other cognitive domains, ultimately leading to loss of activities of daily functioning. Amyloid plaques, neurofibrillary tangles and neurodegeneration are the main pathological hallmarks of the disease. Given the heterogeneity of clinical symptoms in AD and the overlap with other types of dementia, the use of biomarkers of AD pathology combined with clinical information contributes to increased diagnostic accuracy The first AD biomarkers included cerebrospinal fluid (CSF) amyloid-β (Aβ) and phosphorylated tau (p-tau) levels, Aβ PET and MRI. Although these biomarkers have excellent diagnostic accuracy compared with neuropathology observed at autopsy, their use at the population level for diagnosing AD has many limitations. Blood-based biomarkers (BBMs) have the potential to transform the AD diagnostic pathway, but many questions remain regarding their implementation and utilization in this context.

This Review provides a broad perspective on the factors that warrant consideration when implementing and interpreting AD BBMs at the population level. First, we provide an overview of the current use of biomarkers in AD, including BBMs. In addition, we describe the characteristics of typical patients with cognitive impairment in primary care and how they differ from the populations used in most AD BBM research studies. We consider factors that might affect the interpretation of BBM tests, including comorbidities, sex, and race or ethnicity. Finally, we discuss broader issues such as ethical aspects, patient and provider preference and other considerations for the use of AD BBMs at the population level. We focus on blood Aβ₄₀, Aβ₄₂, Aβ_42:40 ratio, p-tau181, p-tau217 and neurofilament light (NfL), because these markers are currently available for clinical use, at least in some parts of the world. However, it is important to highlight that these biomarkers should be used not in isolation to diagnose AD but in the context of objective cognitive impairment, a clinical examination and potential confirmatory testing with CSF or PET when warranted. Also, most studies of AD BBMs to date have been conducted in certain geographical areas of HICs, which has influenced the thinking behind current implementation strategies. We will need to explore AD BBMs in LMICs and other understudied populations to fully understand how they can be interpreted and implemented at the worldwide population level and to obtain patient and provider perspectives across different cultures.

Incorporating biomarkers into AD diagnosis

Historically, AD has been defined as a clinicopathological entity. Clinical symptoms such as amnestic memory impairment and a decline from prior cognitive functioning could indicate possible or probable AD, but a definite diagnosis could only be made at autopsy with evidence of Aβ plaques and neurofibrillary tau tangles³. A clinical syndrome does not indicate a specific aetiology; for example, amnestic memory impairment could be caused by AD, other progressive neurodegenerative diseases, brain injury, an acute vascular event or potentially reversible conditions such as major depression, vitamin B deficiency or medication interactions. Therefore, on the basis of clinical symptoms alone, determining whether someone with memory impairment has AD pathology is difficult. Assessing the efficacy of anti-Aβ or anti-tau disease-modifying therapies (DMTs) was also challenging, because some patients enrolled in clinical trials were ultimately found not to have AD pathology at autopsy⁴.

Over the past two decades, the development and standardization of CSF biomarkers (low Aβ_42:40 ratio and high p-tau levels) and PET biomarkers (elevated Aβ) has contributed to a more accurate clinical diagnosis of AD. Moreover, these biomarkers can confirm that participants with clinical symptoms enrolled in DMT trials targeting AD pathology actually have the pathology^5,6. However, these biomarkers have several limitations, including little to no availability in many geographical areas, both in HICs and LMICs; invasiveness, given that a lumbar puncture is needed for CSF collection and an intravenous infusion is required for PET); contraindications, such as spinal stenosis and other chronic conditions that increase with age; and prohibitive cost owing to lack of coverage by many insurance companies and/or high out-of-pocket copayments. In addition, CSF and PET biomarkers often need to be requested by dementia specialists, such as neurologists, psychiatrists or geriatricians, who are in short supply worldwide, resulting in long waiting times^7,8. The latter issue is especially troublesome given that current DMTs seem to be most effective early in the disease process, and the long wait to see dementia specialists could result in individuals no longer being eligible for the treatments⁹. This situation has prompted calls for greater involvement of primary care providers (PCPs) in the identification, diagnostic evaluation and care of patients with AD so that only those who need more comprehensive assessments need to be referred to dementia specialists^10–13.

Given the limitations of CSF and PET AD biomarkers for real-world clinical populations, the technological advances over the past 5 years, which have enabled the measurement of the much lower concentrations of Aβ and p-tau in the blood compared with CSF, have been a much-needed accomplishment. These BBMs are already available for clinical use in both specialty and primary care in some countries, and have been incorporated into clinical trials¹⁴. However, despite the considerable potential of BBMs to enable more accurate and earlier diagnosis of AD, many questions remain regarding their implementation and utilization at the population level.

Overview of AD blood-based biomarkers

Current clinically available AD BBMs, most of which are measured in plasma, include a low Aβ_42:40 ratio (owing to Aβ deposition in the brain) and elevated p-tau181, p-tau217 and NfL levels.

Amyloid-β_42:40 ratio

Differences in Aβ_42:40 ratios between individuals with and without elevated brain amyloid pathology, as determined by PET, are less pronounced in plasma (approximately a 15% reduction) than in CSF (approximately a 50% reduction)¹⁵. Although Aβ concentrations are naturally lower in blood than in CSF, studies have consistently demonstrated that the plasma Aβ_42:40 ratio has excellent diagnostic accuracy to detect elevated brain amyloid across the AD cognitive spectrum^15,16.

Phosphorylated tau181 and tau217

Plasma p-tau181 and P-tau217 are biomarkers that are specific for AD brain pathology and are not increased in other tauopathies such as frontotemporal dementia¹⁷. Levels of these proteins correlate with CSF and PET biomarkers of AD pathology across the clinical AD spectrum and have been found to predict progression from MCI to AD ^18–20. Although both p-tau181 and p-tau217 biomarker tests are clinically available, several studies now suggest that p-tau217 is a better indicator than p-tau181 of the presence of elevated brain amyloid ^21,22. In addition to p-tau181 and p-tau217 measurements, the ratio of p-tau217 to non-phosphorylated tau has also shown high accuracy for the prediction of brain amyloid, as measured using amyloid PET or CSF Aβ_42:40 and p-tau181:Aβ₄₂ ratios²³.

Neurofilament light chain

Neurofilament light chain (NfL) in plasma or serum is a biomarker of degeneration of large-calibre axons²⁴. NfL is considered to be a non-specific neurodegeneration biomarker because its levels are elevated in many neurodegenerative diseases²⁵. Changes in blood NfL correlate more strongly with cognitive symptoms and brain atrophy than do Aβ or p-tau biomarkers²⁶.

Platforms and assays

Ultrasensitive immunoassays and antibody-free mass spectrometry platforms are available to measure BBMs in the clinic. Mass spectrometry measures include plasma Aβ_42:40 and p-tau217: non-phosphorylated tau ratios. Immunoassays, some of which are fully automated, can be used to measure the Aβ_42:40 ratio and p-tau181, p-tau217 and NfL levels. In-depth descriptions and comparisons of the various assays are provided in other review articles^27,28. Importantly, the accuracy of AD BBMs, compared with amyloid PET, varies substantially across platforms and assays^22,29. However, access to a specific assay for clinical use might be limited by workflows and/or the laboratory with which the health system has a contract. An important consideration for the use of AD BBMs at the population levels is throughput and scalability. From this perspective, automated immunoassays have advantages over mass spectrometry methods.

Exclusionary versus confirmatory test

Whether BBMs should be used solely as triage tools to rule out a diagnosis of AD or as confirmatory diagnostic tests is the subject of much discussion. Some BBMs are reported to be as accurate as FDA-approved CSF tests for classifying amyloid PET status²³. However, these conclusions were based on studies that used batch testing to assay AD BBMs and did not prospectively measure the biomarkers in real time to establish a clinical diagnosis. Also, the study cohorts are not necessarily generalizable, as they tended to be healthier, younger and less racially and ethnically diverse than real-world AD populations. Consequently, in this Review, we primarily consider AD BBMs as a triaging tool.

Effects of chronic conditions on AD blood-based biomarkers

In the USA, the average age at diagnosis of dementia, of which AD is the most common cause, has been estimated as 83 years³⁰, although the exact age varies according to geographical location and differential risk factors and access to care. Despite this high average age, most research studies of BBMs to date have focused on younger and healthier cohorts. Approximately 60% of older adults aged 70 years and older with AD/ADRD have three or more chronic conditions^31,32. Among older adults with AD/ADRD, 52–82% also have hypertension^33,34, 16–39% have diabetes^33–35, 11–18% have chronic kidney disease (CKD)³⁶, 9–17% have chronic obstructive pulmonary disease^33,34,37 and 9–28% have congestive heart failure³⁴ (Figure 1). The prevalence of these conditions is even higher among racialized groups³⁸ and individuals with low socioeconomic status³⁹. These chronic conditions, as well as frailty, are also risk factors for AD/ADRD and can affect the expression of AD pathology with regard to cognitive function, disease stage and neuropathological burden^40–42. In addition, polypharmacy is common in patients with multiple chronic conditions and can further affect cognitive function, making it difficult to accurately diagnose AD/ADRD and predict disease progression. As a result of this complexity, an estimated 50–70% of patients with AD/ADRD are not recognized or incorrectly diagnosed in primary care⁴³.

Figure 1 |.

Chronic conditions in typical patients presenting with cognitive impairment. The figure shows the percentages of patients presenting to primary care with cognitive impairment who also have various chronic conditions^33–37.

The use of BBMs in primary care could help provide a better understanding of the aetiology of cognitive impairment. However, evidence suggests that the presence of certain chronic conditions can affect AD BBM levels, but not CSF or PET biomarkers — an important consideration when interpreting BBM levels for diagnosis of AD. Chronic conditions could influence BBMs through three different mechanisms. First, they could be risk factors for AD pathology and, thus, contribute to AD pathology or exacerbate the clinical manifestations. Second, they could physiologically affect the levels of the BBMs, for example, by altering renal function or blood volume. Last, the conditions themselves could cause peripheral changes in BBMs; for example, as discussed below, some cardiovascular conditions and medications affect plasma amyloid. It is important to make a clear distinction between these three mechanisms, as only the last two can influence the interpretation of AD BBM levels, resulting in false-positive or false-negative results. In the sections that follow, we describe three chronic conditions that must be considered when interpreting AD BBM results.

Chronic kidney disease

Multiple studies have shown that a diagnosis of CKD, low estimated glomerular filtration rate (eGFR) and high blood creatinine levels are all associated with elevated blood levels of Aβ₄₀,Aβ₄₂, p-tau181, p-tau217 and NfL^19,44–48. Even among patients with a total eGFR >90 ml/min, indicating the absence of stage ≥3 CKD, robust correlations are observed between total eGFR and BBM levels⁴⁹. The increased BBM levels have been attributed to reduced renal clearance.

Epidemiological and clinical studies indicate that the associations between BBMs and cognition or neuroimaging outcomes are the same regardless of whether CKD or eGFR is included as a covariate in regression models⁴⁶, leading to speculation that CKD does not affect the interpretation of the BBMs. However, although CKD seems to have no effect at the group level, failure to consider CKD at the individual level could result in a false-positive diagnosis (if measuring p-tau181 or p-tau217) or a false-negative diagnosis (if measuring Aβ₄₀ and Aβ₄₂) of AD pathology⁵⁰.

We do not yet know precisely how to use BBMs for the diagnosis of AD in patients with CKD or how to consider renal function in the interpretation of BBM levels. Different BBM cut-off points depending on the severity of CKD might be warranted. Work is ongoing to determine these cut-off points, especially for individuals who have additional chronic conditions, which are common among patients with CKD. One potential way to mitigate the effects of CKD on BBMs might be to use ratios. For example, CKD has less effect on plasma Aβ42:40 or p-tau217:non-phosphorylated tau rations than on individual biomarker (Aβ40, Aβ42 or P-tau) levels^44,46,51,52. However, a study of patients with CKD reported that levels of Aβ₄₀ and Aβ₄₂, and the Aβ_42:40 ratio, changed significantly when measured before and after dialysis⁵³. Therefore, even if ratios are used, time after dialysis might need to be considered.

Obesity

Obesity has been associated with reductions in plasma but not CSF levels of Aβ₄₀, Aβ₄₂, p-tau181, p-tau217 and NfL^{19,44,46,49,54}. A potential explanation for these findings is that individuals with obesity have increased blood volumes, resulting in lower concentrations of AD BBMs⁵⁴. As an example of this phenomenon, one study examined plasma NfL levels in patients with obesity before (mean BMI 41.2 kg/m²) and six months after (mean BMI 29.6 kg/m²) gastric bypass surgery in comparison with control individuals of healthy weight (mean BMI 23.4 kg/m²). Before surgery, the patients with obesity had significantly lower levels of plasma NfL than the controls. However, approximately 6 months after surgery, with considerable weight loss, no differences were observed between the groups⁴⁹. Therefore, plasma NfL levels significantly increased in patients with obesity who underwent gastric bypass surgery and experienced considerable weight loss. This finding seems to contradict studies showing that weight loss is beneficial for brain structure and function and would, therefore, be expected to decrease blood NfL levels⁵⁵.

Many patients with cognitive impairment lose weight before and during the early symptomatic and later phases of dementia, and low BMI is associated with AD and vascular pathology in later life^56,57. Therefore, interpreting AD BBM levels in the context of obesity and weight loss is difficult. Different cut-off points based on the BMI category are unlikely to be needed, but knowledge of recent considerable weight loss might aid the interpretation of AD BBMs. Similar to the CKD situation, the use of ratios could mitigate the confounding effects of obesity on BBM levels^44,46. New anti-obesity medications that result in rapid weight loss are being developed, such as glucagon-like peptide 1 agonists, and we will need to understand how to interpret BBM tests in patients who use these drugs if they develop cognitive impairment.

Cardiovascular disease

Aβ is intricately related to both AD and cardiovascular disease. In patients with AD dementia, Aβ can accumulate in the heart and induce AD-related cardiac amyloidosis^58,59. In addition, accumulation of plasma Aβ₄₀ in vascular walls and heart tissue is associated with cardiac dysfunction, coronary heart disease, heart failure and cardiovascular disease mortality^60,61. Moreover, high levels of plasma Aβ₄₂ are associated with an increased risk of heart failure⁶². One study showed that sacubitril/valsartan, a neprilysin inhibitor and receptor angiotensin blocker that is used to treat heart failure, increased levels of Aβ₄₀ and Aβ₄₂ but significantly decreased the Aβ_42:40 ratio in the blood⁶³. By contrast, this drug did not affect the CSF Aβ_42:40 ratio ⁶⁴.Therefore, use of the plasma Aβ_42:40 ratio to detect AD pathology in patients with cognitive impairment who use sacubitril/valsartan could result in a false-positive diagnosis of AD because the medication lowers the ratio.

The prevalence of cognitive impairment is high among patients with heart failure and coronary heart disease^65,66, and the ability to accurately determine whether the cause is vascular, AD or both is important. AD BBMs might contribute to a better understanding of the underlying aetiology of the cognitive impairment, but heart failure, coronary heart disease and other cardiovascular diseases complicate the utilization and interpretation of these biomarkers. Plasma p-tau could be a better biomarker than Aβ to detect AD pathology in this population: despite the effect of sacubitril/valsartan on the plasma Aβ_42:40 ratio, the drug did not affect levels of plasma p-tau181, p-tau217 or NfL⁶³. However, one study reported that levels of p-tau181 and p-tau217 were elevated in individuals with a history of myocardial infarction, so the possibility remains that certain vascular conditions can affect p-tau levels²¹. Comprehensive longitudinal studies of AD BBMs among cohorts with cardiovascular disease are needed to understand how best to use and interpret the BBMs in this context.

Multiple chronic conditions

Although several studies have reported that specific chronic conditions or medications can affect the levels or interpretation of AD BBMs, the effects of combinations of multiple chronic conditions on BBM levels have not been examined. As we have already stated, around 60% of older adults with cognitive impairment have three or more chronic conditions^31,32. Some of the conditions described above increase whereas others decrease AD BBM levels, and the net effects of coexistence of these conditions on BBM levels remain unknown.

An algorithm that incorporates multiple chronic conditions and medications to allow physicians to better interpret the BBM levels in the context of the presence of these conditions is in development. However, several nuances need to be considered. Some of the conditions will be risk factors or contributing factors for AD pathology, thereby affecting both blood and brain biomarkers. Other conditions or medications have physiological effects that result in altered BBM levels but do not affect CSF or PET AD biomarkers. Ideally, both blood and brain biomarkers should be measured. However, many patients with chronic conditions have contraindications to a lumbar puncture for CSF collection or to an amyloid PET scan. In addition, patients either with and without multiple chronic conditions might not want to undergo the more invasive tests, making it difficult to understand whether the effects of chronic conditions and medications on BBMs originate in the brain or the periphery. Patients who do undergo a lumbar puncture and/or an amyloid PET scan will not be fully representative of the broader population.

Sex differences and AD blood-based biomarkers

Most studies that compared the Aβ_42:40 ratio or p-tau181, p-tau217 or NfL levels in serum or plasma between men and women across the AD clinical spectrum did not find any sex differences^{26 67–70}. However, some studies reported higher plasma p-tau181 or p-tau217 levels in men than in women^21,71,72. Two of these studies^21,71 reported a higher frequency of CKD in men than in women; when the individuals with CKD were excluded from the analysis, the sex difference was partially attenuated but remained statistically significant. Despite the general lack of cross-sectional sex differences in the levels of AD BBMs, however, some studies suggest that for a given biomarker level, women are more adversely affected. For example, elevated p-tau181 or p-tau217 levels have been associated with greater cognitive decline and changes in brain neuroimaging measures in women than in men^69,73,74.

Currently, no strong justification exists to develop sex-specific AD BBM cut-off points. Future studies will need to examine whether the prognostic value of the biomarkers differs by sex. Many studies of diagnostic or prognostic AD BBMs have not explored the possibility of sex differences, and the reporting could be biased towards positive findings. To determine whether sex differences in AD BBMs exist, uniform reporting across studies, including negative findings, will be crucial. In addition, it will be important to consider whether study populations show sex differences in the frequency of chronic conditions and whether these differences influence AD BBM levels.

Racial and ethnic differences in AD blood biomarkers

In the USA, AD/ADRD disproportionately impacts Black and Hispanic individuals, with Black Americans twice as likely to be diagnosed — frequently at a later disease stage — than non-Hispanic white Americans, and Hispanic Americans 1.5 times as likely to be diagnosed^75,76. Some studies of racial and ethnic differences in AD biomarkers have reported lower CSF p-tau levels in Black versus non-Hispanic white individuals for the same clinical severity of AD, and a lower prevalence of elevated brain amyloid on PET in Hispanic and Asian individuals than in non-Hispanic white individuals^77–80. Examination of screening data for the Anti-Amyloid in Asymptomatic AD study found that Black and Asian participants were more likely to be ineligible for the trial than were non-Hispanic white participants, owing to an absence of elevated brain amyloid⁸¹. These results have led to suggestions that racial and ethnic differences in these biomarkers of AD pathology reflect differences in the underlying aetiology of cognitive impairment⁷⁹. An alternative, non-mutually exclusive explanation is that hesitancy to undergo a lumbar puncture or amyloid PET is common among racialized groups, and those who elect to undergo such tests are not necessarily representative of the entire population.

Compared with CSF and PET AD biomarkers, studies of racial and ethnic differences in AD BBMs are less consistent, with some^77,82,83 but not all^68,71 reporting differences in BBM levels by race and ethnicity. Race is a social construct, and many questions remain about how to utilize race in biomedical research and to biologically interpret the findings⁸⁴. The inconsistent results could have multiple explanations. First, the frequencies of certain chronic conditions or medications that affect the levels of AD BBMs vary by race and ethnicity (and between study cohorts) and could contribute to observed differences in BBM levels. Second, other social determinants of health could contribute to racial and ethnic differences in AD BBMs. To date, examination of the effects of racialization, including environmental and sociocultural exposures, on AD BBMs has been limited^84,85. However, one 2024 study reported that exposure to greater levels of racial discrimination was associated with greater increases in plasma p-tau 181 and NfL levels among Black Americans⁸⁶. Last, penetrance of the apolipoprotein E (APOE) ε4 allele, a strong genetic risk factor for late-onset AD, differs by ancestry. Among people with this allele, the risk of AD dementia is lower in individuals with African or American Indian ancestry than in those with European ancestry^87,88. How differences in the penetrance of the APOE ε4 allele affect AD BBM levels and contribute to any observed racial or ethnic differences is unclear. The accuracy of BBM-based diagnostic algorithms for AD that include the APOE genotype has not been determined in racialized groups and, therefore, the use of these algorithms in such populations is not advisable at the present time.

Given the inconsistency between studies in racial and ethnic differences in AD BBMs and the lack of explanation for any observed differences, the use of race-specific or ethnicity-specific cut-off points to diagnose AD is not recommended at this time. It is important to note, however, that the existing cut-off were primarily derived from studies of participants who were white, younger than the typical patient with dementia, and healthier and wealthier than the general population. Therefore, the generalizability and accuracy of these cut-off points in more heterogenous and racially and ethnically diverse populations remain to be determined.

Population use of AD blood-based biomarkers

Understanding how to interpret AD BBM levels in the context of multiple chronic conditions is an important first step in the implementation of these biomarkers for the diagnosis of AD in older adults at the population level. However, several additional questions will need to be addressed, including how to deal with incidental findings, indeterminate results and imperfect accuracy, and how to account for the possible effects of stigma and discrimination. Additional considerations include the value of biomarkers in individuals who have subjective cognitive complaints or are cognitively healthy but have a family history of dementia, the potential impact of a positive biomarker test on the eligibility to drive or obtain long-term care insurance, and the implications of including the biomarker test in the medical record. Some of these issues are discussed below. Notably, legal and ethical issues will vary depending on the geographical location and domestic laws.

Most studies of AD BBMs have focused on specialty clinic populations of patients or research participants who are non-Hispanic white, of high socioeconomic status, healthier than the general population with fewer co-morbid conditions and, on average, younger than the typical patient presenting to primary care with cognitive concerns. In addition, most of the clinical experience with AD BBMs to date has been in specialty clinics or in research settings as part of a comprehensive work-up of symptoms. Although the current recommendation is that AD BBMs should not be used in isolation from cognitive assessments, and confirmation of AD pathology with CSF or PET is suggested⁸⁹, the limitations of the latter biomarkers and the availability and advantages of the BBMs are likely to lead to their widespread adoption in primary care. Indeed, most individuals with cognitive impairment worldwide will never be referred to dementia specialists for various reasons, including access, availability, cost and patient preference to not see a specialist even if referred^65,90–92. Thus, widespread education regarding timely methods to assess cognitive impairment, when to use AD BBMs and how to interpret the results in the context of clinical symptoms is imperative for health-care providers. Given the limited time that PCPs have to spend with patients, new patient care models will need to be developed and appropriate reimbursement systems will be essential.

Real-world accuracy and implementation

As discussed above, few studies to date have examined AD BBMs for the diagnosis and prognosis of AD in primary care, particularly among racialized groups, in diverse populations of older adults with multiple chronic conditions or among individuals with multiple risk factors associated with social determinants of health. Therefore, the current cut-off points for the clinically available tests have not been confirmed using clinically representative or worldwide populations, and the accuracy of AD BBMs in the real world is unknown. Moreover, observational studies of diverse populations generally analyze all samples in batches, which produces less variability than prospectively testing individual blood samples for patient care. Cut-off points for clinical use will need to be refined and updated over time⁹³.

The use of one versus two cut-off points for AD BBMs is the subject of an ongoing discussion. One complication of using a single cut-off point is that some individuals have values near the cut-off point and might show discordant results on repeat testing. Furthermore, the choice of a cut-off point to optimize a ‘rule-in’ approach (resulting in high sensitivity) at the population level will greatly increase the number of individuals that will need follow-up tests and visits compared with a ‘rule-out’ test (resulting in high specificity). Some tests use two cut-off points to indicate positive, intermediate and negative results, with intermediate results prompting additional testing and follow-up. The use of two cut-off points for an AD BBM could enhance diagnostic accuracy by identifying individuals with and without brain pathology and might work better than a single cut-off point to diagnose AD⁹⁴. However, from a population perspective, the use of two cut-oof points could result in a larger number of patients with indeterminate result, who will require additional testing and referrals, thereby impeding timely care in already strained health-care systems.

The conditions for blood collection are also more standardized in research studies than in real-world clinical settings. Although the measurement of plasma p-tau species and NfL seem to be relatively robust to variations in pre-analytical handling factors, the Aβ_42:40 ratio is more sensitive^93,95. Location and ability to standardize blood collection might affect the choice of an AD BBM in a particular clinical setting⁹⁶.

Access to biomarkers and confirmatory diagnostics

Although AD BBMs are less invasive and costly and more accessible than amyloid PET or CSF biomarkers, access issues remain that could exacerbate health-care disparities. Individuals who are racialized, have low socioeconomic status or live in rural areas have limited access to health-care providers who, at present, have sufficient knowledge or confidence to interpret AD BBMs, or have the infrastructure and resources to provide the comprehensive diagnostic clinical work-up that is needed in conjunction with these biomarkers. Furthermore, in insurance-based health-care systems, although immunoassays are reimbursed to some extent, out-of-pocket costs might be prohibitive for some patients.

Currently, the main application of AD BBMs is as part as a triaging process, such that a positive test would require further confirmatory testing such as amyloid PET or CSF analysis. Practitioners have been shown to overestimate the probability of disease both before and after testing and regardless of whether the results are positive or negative⁹⁷. This overestimation is likely to increase the number of people who are referred for additional confirmatory tests or specialty care, leading to a lengthening of the waiting time to see a specialist. The costs of the additional tests might also be insurmountable for some individuals, thereby widening health disparities. This scenario also raises the question of whether a patient should undergo an AD BBM test if they are not interested in the result, cannot afford the test or have contraindications to further confirmatory testing.

Diagnostic implications

As for other medical tests and procedures, guidelines and appropriate use criteria must be developed to indicate when and how AD BBMs should be incorporated into patient care and clinical decision-making. For a patient experiencing cognitive symptoms, a negative BBM test might provide reassurance that they do not have AD. However, a negative test cannot rule out another cause of dementia. Many pathologies contribute to dementia but only some have available biomarkers; therefore, further work-up will be needed to understand the aetiology.

A positive BBM test might indicate the presence of AD pathology but could also be an incidental finding or might be only one of many factors that is contributing to the person’s cognitive impairment. This is a particularly important consideration for individuals over the age of 80 years, who could have multiple brain pathologies, including cerebrovascular disease and/or other neurodegenerative pathologies^98,99. Moreover, even at younger ages, AD pathology and Lewy body pathology often co-occur. Thus, when contemplating AD treatments, a positive finding should not be used in isolation without also considering other causes and interventions, as well as the patient’s care goals. For some individuals, better control of vascular disease and risk factors, changes in diet or exercise, or interventions to enhance sleep and reduce stress might be more immediate options. In addition, the use of an AD BBM for an AD diagnosis in the context of life-limiting conditions or when a patient is ineligible for a DMT might not contribute to a change in care or care goals. As the time horizon for life expectancy changes, the calculus around taking action following a positive AD biomarker test also changes. Guidelines for such situations will need to be formulated, taking into account both ethics and patient preferences.

Documentation and communication of test results

In many health-care delivery settings, the results of an AD BBM test will be entered into the patient’s medical record and might be viewable by the patient in a health-care portal before they can discuss the results with their physician. Patients might not understand the distinction between AD pathology and a clinical diagnosis of AD, or that the results of the blood test alone are not conclusive¹⁰⁰. For example, even if a blood test has an accuracy greater than 90%, some individuals will be misdiagnosed. No biomarker is 100% sensitive and specific, and co-pathologies for which biomarkers are unavailable, such as Lewy bodies, often contribute to cognitive impairment. Therefore, reliance on biomarkers alone without clinical context will result in some degree of misdiagnosis.

A lack of patient or provider understanding of what the AD BBM is being used for and what the results might mean could cause distress. Therefore, before the blood test is conducted, additional PCP time will be needed to explain what the test is measuring, the possible next steps if the blood test is positive and the potential diagnostic uncertainty. The European Academy of Neurology and the European Alzheimer’s Disease Consortium released a position statement recommending that pre-biomarker and post-biomarker counselling for individuals with mild cognitive impairment should be delivered by a dementia specialist¹⁰¹. However, this approach is not amenable to large-scale implementation. PCPs have limited time with patients, so additional reimbursement models for the extra time to provide the background information and answer questions will be necessary if AD BBMs are to be used widely at the population level¹³. In addition, new care models incorporating the training and use of non-PCP health care providers, as well as for longitudinal dementia care following a diagnosis of AD/ADRD, are warranted.

During the PCP–patient discussion before disclosure of the AD BBM test result, the potential negative implications of a positive test for future lifestyle choices must be outlined so that an informed decision can be made by the patient and/or their family¹⁰². These implications will differ by country and region. For example, in the USA, a positive AD BBM test in the medical record could affect elibility for long-term care insurance or disability insurance, because insurers are permitted to use health information in the underwriting process¹⁰². The cost of life insurance might also be affected. In addition, a BBM indicating AD pathology could affect patients’ rights, including the ability to hold a driver’s licence or own a gun¹⁰³. Patients with a diagnosis of dementia or AD still experience stigma, the degree of which varies around the world and by culture^104–107. Some have suggested that the discovery and approval of DMTs for AD would reduce stigma, but a study that tested this hypothesis found no evidence for such a reduction¹⁰⁸.

Multiple studies have assessed the impact of disclosure of amyloid PET or APOE status to individuals with or without cognitive impairment in clinical trials or specialized centres (for example, AD research centres)^109,110. However, the participants in these types of studies are not representative of the older adults with cognitive impairment and multiple chronic conditions who present to primary care. As we have already highlighted, research study participants in general have higher education levels and socioeconomic status, are motivated to participate in clinical research, and are predominantly non-Hispanic white. At the population level, it remains unclear how many individuals across the AD clinical spectrum want to know whether they have AD pathology, based either on a blood test or another biomarker test. Qualitative data is needed to understand whether preferences vary by age, sex, race or ethnicity, culture, insurance, education or other factors related to social determinants of health.

Screening applications

Much discussion has been centered around the possibility of using AD BBMs for screening purposes, including risk assessment, similar to current recommendations for cholesterol and blood glucose screening, or to identify individuals for early treatment initiation if DMTs prove to be effective at preclinical stages. Despite the potential future benefits, however, the idea of using AD BBMs for screening raises several concerns. AD pathology begins decades before the appearance of clinical symptoms and increases with age: about 30% of individuals who are cognitively unimpaired or report subject cognitive complaints aged 70 years and older have brain Aβ pathology, and this figure increases to 40% by the age of 80 years¹¹¹. Although positivity on both amyloid and tau PET has been strongly associated with future cognitive impairment among cognitively healthy individuals¹¹², a subset of individuals remain cognitively unimpaired until death despite showing moderate-to-severe AD pathology at autopsy^113,114. These individuals might have developed symptoms if they had lived long enough, but uncertainty about if and when an individual will develop AD dementia could cause psychological distress^115,116.

Some studies have suggested that plasma AD biomarkers can predict the risk of cognitive impairment or dementia 5–6 years into the future; however, these studies did not use the current clinically available tests and cut-off points, so the generalizability of the results at the population level is unclear^117,118. Moreover, numerous chronic conditions can influence the risk of dementia, but studies have not assessed the additional prognostic value of AD BBMs for patients who are frail and/or have multiple chronic conditions. Not all individuals will want to know their risk, and this preference might be informed by several factors, including culture, race, ethnicity, chronic comorbidities or life expectancy. Also, as mentioned above, evidence of AD pathology from BBMs in individuals with cognitive impairment might have legal repercussions and exacerbate stigma^102,119.

Another important consideration is how the availability of direct-to-consumer (DTC) tests might influence AD diagnostic pathways and care. Given the potential stigma and discrimination associated with AD BBM test results in medical records, some individuals might prefer a DTC test. However, these tests have a number of drawbacks as diagnostic aids for AD. A poorly performing test, the difficulty of interpreting the results in the context (or absence) of clinical symptoms and/or inadequate guidance on false-positive results could cause substantial distress for individuals both with and without cognitive symptoms. Furthermore, many health care providers might lack the expertise to use or interpret them..

Conclusions

The technological advancements that have led to the development of BBMs of AD pathology have provided an unprecedented opportunity to improve the timeliness and accuracy of AD diagnosis at the population level, including in low-resource settings. However, much research is still needed to understand how best to implement BBMs in the primary care setting. Many older adults with cognitive impairment have multiple chronic conditions that need to be considered when interpreting BBM levels to mitigate the potential for false-positive or false-negative results. Several issues relating to the use of AD BBMs remain to be resolved, including the potential for incidental findings and increased stigma and discrimination, dealing with indeterminate results and imperfect accuracy, the implications of a positive biomarker test for driving or insurance, and the consequences of including the biomarker test results in the medical record. New policies and guidelines surrounding the use of AD BBMs will be needed to ensure appropriate protections for patients.

Box 1 |. Considerations for population-level use of AD BBMs.

Factors that might affect AD BBM interpretation

The following factors might alter AD BBM levels, leading to false-positive or false-negative results:

• Obesity.

• Race or ethnicity.

• Sex.

• Presence of multiple chronic conditions.

• Chronic kidney disease.

• Cardiovascular conditions and medications.

Broader issues

The following additional factors need to be considered when implementing AD BBMs in the global population:

• Patient and health-care provider preferences.

• Potential for incidental findings.

• Degree of diagnostic benefit for a given patient.

• Possible repercussions from inclusion of test results in a patient’s medical record.

AD, Alzheimer disease; BBMs, blood-based biomarkers.

Key points.

• Numerous studies have demonstrated the clinical utility and accuracy of plasma measures of the amyloid-β_42:40 ratio and phosphorylated tau (p-tau) 181 and p-tau217 levels for the detection of Alzheimer disease (AD) pathology among clinically well-characterized patients.

• Most AD blood-based biomarker (BBM) studies focused on specialty clinic populations are not generalizable to typical patients with dementia, and an urgent need exists to test the BBMs at the population-level.

• Chronic kidney disease, obesity and cardiovascular conditions or medication can elevate or lower AD BBM levels and need to be taken into consideration to avoid false-positive or false-negative diagnoses; understanding how to interpret AD BBM levels in the context of multiple chronic conditions is crucial for diagnosing AD among older adults in the population.

• Other factors that might influence AD BBM levels include sex and race or ethnicity, although findings on these associations have been inconsistent to date.

• A positive BBM test might indicate the presence of AD pathology but could be an incidental finding; the test must be considered in the context of all other symptoms and with the potential for co-pathologies.

• Policies must be developed to protect patients who have AD BBM results added to their medical records so that they do not lose access to insurance or to disability or other rights.