Abstract
INTRODUCTION
Establishing interchangeability between blood biomarkers and amyloid‐positron emission tomography (PET) could help identify patients who would benefit from novel amyloid‐targeting therapies for Alzheimer's disease.
METHODS
Adult participants from the Global Alzheimer's Platform Foundation's Bio‐Hermes study with clinical diagnosis of mild cognitive impairment/mild dementia and available amyloid‐PET evaluations and Aβ42/40‐based PrecivityAD® (by C2N Diagnostics) and/or phosphorylated tau at threonine 217 (p‐Tau217 (research use Meso Scale Discovery, Lilly) were included. Interchangeability between plasma‐based tests (per pre‐established thresholds) and amyloid‐PET stratifications was evaluated.
RESULTS
PrecivityAD (N = 537) and p‐Tau217 (N = 531) plasma‐based tests had high agreement with florbetapir‐PET (overall percent agreement: 80.7% and 90.1%, respectively) and accurately selected patients identified as florbetapir‐PET‐positive (positive predictive value: 86.0% and 88.0%, respectively). Further, they were non‐inferior to quantitative florbetapir‐PET (at 37 Centiloids) for identifying positive florbetapir‐PET visual reads.
DISCUSSION
Results support the hypothesis that blood biomarkers may be interchangeable with amyloid‐PET criteria for selecting patients who may benefit from treatment with novel amyloid‐targeting therapies.
Highlights
Interchangeability of plasma‐based tests and amyloid‐PET stratifications was demonstrated.
Non‐inferiority of plasma‐based biomarkers to amyloid‐PET for identifying patients with AD pathology was observed.
High agreement between plasma‐based test stratification and florbetapir‐PET expert visual read was observed.
Keywords: AD pathology, Alzheimer's disease, amyloid beta, blood biomarkers, concordance analysis, plasma biomarkers, positron emission tomography, tau
1. BACKGROUND
Alzheimer's disease (AD) is characterized by the presence of amyloid beta (Aβ) plaques and neurofibrillary tau tangles. 1 , 2 , 3 Specifically, abnormal aggregation of neocortical Aβ, a pathological hallmark of AD, is thought to occur prior to the accumulation of neurofibrillary tau tangles and neurodegeneration. 1 These neuropathological changes in AD are thought to begin decades before the onset of clinical symptoms, 4 and recent trials suggest that early intervention may provide the most clinical benefit to patients. 5 , 6 Therefore, the early and accurate identification of AD pathology is critical. AD pathology may be detected with positron emission tomography (PET) imaging 2 ; however, the manufacturing and supply logistics for radiopharmaceuticals are complex, and the short half‐life of radiolabeled drugs restricts access to within a short distance of the production site. 7 Additionally, the high costs associated with PET imaging are often not reimbursed. 8
Although several cerebrospinal fluid (CSF) biomarkers obtained through lumbar puncture have been validated and clinically certified for assessment of Aβ and tau pathology, 9 its negative patient perception, inconvenience of lumbar punctures, and unavailability of trained personnel to perform the procedure may also limit the use of CSF biomarkers in routine clinical care. 10 Blood biomarkers are simpler, more cost‐effective, and easier‐to‐implement alternatives to CSF and neuroimaging biomarkers. 2 , 3 , 8 Therefore, blood biomarkers represent a more accessible and affordable modality for assessing AD pathology that could lead to more equitable access to an AD diagnosis and, in turn, disease‐modifying therapy.
The aim of our study was to assess the agreement and interchangeability between plasma‐based tests and amyloid‐PET, as well as the non‐inferiority of the plasma‐based biomarkers to amyloid‐PET Centiloid (CL) values for the identification of patients with AD pathology.
2. METHODS
2.1. Patients and design
Patients included a subset of participants from the Global Alzheimer's Platform Foundation's Bio‐Hermes study (NCT04733989). 11 The study was launched in April 2021 and ended in November 2022. It was conducted across 17 research sites in the United States, all of which have extensive experience in recruiting participants from the community‐based population for clinical trials investigating potential new drug treatments for AD. The primary objective of the Bio‐Hermes study was to have a head‐to‐head comparison of results from various blood and digital biomarkers against brain amyloid‐PET scans and traditional cognitive tests. Details on the methodology and design of the Bio‐Hermes study have been published previously in Mohs et al. 11
Patients were included in our study if they were 60 to 85 years of age (inclusive), had a clinical diagnosis of mild cognitive impairment (MCI) or mild dementia, available amyloid‐PET evaluations, and plasma Aβ42/40‐based PrecivityAD® (C2N Diagnostics) results and/or plasma phosphorylated tau at threonine 217 (p‐Tau217) evaluations (research use Meso Scale Discovery [MSD], Lilly).
The PrecivityAD blood test, by C2N Diagnostics, is a clinically available test that uses liquid chromatography mass spectrometry to measure the concentration of Aβ42 and Aβ40 protein isoforms and detect the presence of apolipoprotein E‐specific peptides in the blood. These values are combined with age in an algorithm to estimate an Amyloid Probability Score (APS) (with a score range between 0 and 100) that identifies the likelihood of brain amyloid pathology by amyloid‐PET. 12 , 13 , 14 Patients were stratified based on their APS (C2N) into three groups (absence of amyloid plaques, presence of amyloid plaques, or intermediate) using prespecified APS cutpoints. The APS cutpoints were determined against an amyloid‐PET reference standard stratified at 25 CL. Additionally, patients were stratified into three groups (positive, negative, and indeterminate) based on their plasma p‐Tau217 measured using a proprietary assay on the Meso Scale Diagnostics platform (Lilly Research Laboratories, Indianapolis, IN, USA) using prespecified cutpoints. The prespecified cutpoints were selected against an amyloid‐PET reference standard stratified at 24.1 CL.
18F Florbetapir‐PET images were acquired from a 10‐min scan per the label dosage and administration instructions. 15 , 16 Visual read was performed centrally following the manufacturer's indications, 15 , 16 and the plaque burden was quantified using MIMneuro software (MiM Software Inc., Cleveland, OH, USA). 17 The detailed process of florbetapir‐PET scan was published previously in Mohs et al. 11 Overall, amyloid‐PET scans were uploaded to an electronic imaging portal and centrally read by an expert trained in the manufacturer's florbetapir‐PET read process. The reader had access to each subject's standardized uptake value ratio but made the final determination according to the manufacturer's standards. Additionally, the scans were quantified using the Lilly florbetapir standard analysis pipeline 18 and stratified at ≥37 CL for the non‐inferiority comparison. The stratification of 37 CL was chosen to align with previously adopted clinical trial inclusion criteria. 5 , 19 Details on processing steps for PET data and Centiloid measurements are listed in Supplementary Appendix 1.
The study was carried out in accordance with the ethical principles of the Declaration of Helsinki and Good Clinical Practice guidelines. All patients provided informed consent.
2.2. Agreement between plasma‐based tests and amyloid‐PET visual read
The agreement between APS (C2N) and florbetapir‐PET visual read was assessed using three plasma‐based stratifications: (1) APS (C2N) (after excluding 14.0% intermediate samples) and florbetapir‐PET (visual read), (2) APS (C2N) (intermediate considered negative) and florbetapir‐PET (visual read), and (3) APS (C2N) (intermediate considered positive) and florbetapir‐PET (visual read).
The agreement between p‐Tau217 (MSD, Lilly) and florbetapir‐PET visual read was also assessed using three plasma‐based stratifications: (1) p‐Tau217 (MSD, Lilly) (after excluding 18.6% indeterminate samples) and florbetapir‐PET (visual read), (2) p‐Tau217 (MSD, Lilly) (indeterminate considered negative) and florbetapir‐PET (visual read), and (3) p‐Tau217 (MSD, Lilly) (indeterminate considered positive) and florbetapir‐PET (visual read).
Outcome measures included positive percent agreement (PPA), negative percent agreement (NPA), overall percent agreement (OPA), positive predictive value (PPV), and negative predictive value (NPV). Formulas for each of these outcome measures are detailed in Table S1. Two‐sided 95% confidence intervals (CIs) were calculated using the Wilson score method. As per the hypothesis, PPV > 80% and NPV > 80% were used to indicate high agreement between plasma‐based tests and PET.
For both APS (C2N) and p‐Tau217 (MSD, Lilly) datasets, the agreement between florbetapir‐PET visual read and florbetapir‐PET quantitation, stratified at 37 CL, was also evaluated.
2.3. Non‐inferiority of plasma‐based biomarkers to amyloid‐PET quantitation
The non‐inferiority of APS (C2N) and p‐Tau217 (MSD, Lilly) to florbetapir‐PET clinical trial criteria (≥37 CL) 5 , 19 for identifying AD pathology was assessed (Figure S1). To determine non‐inferiority, both plasma assays and the quantified PET image, stratified at 37 CL, were compared with the reference standard, that is, the presence of AD amyloid pathology as determined by the amyloid‐PET visual read. The agreements between the florbetapir‐PET quantitation, stratified at 37 CL, and florbetapir‐PET visual read (PPVc1r, comparator 1 vs. reference) and the fluid biomarkers and florbetapir‐PET visual read (PPVc2r, comparator 2 vs. reference) were calculated. 20 The non‐inferiority margin was prespecified as 15%. The difference in agreements delta, PPVc1r − PPVc2r, was computed, and sample‐level bootstrap methods were used to construct two‐sided 95% CIs of the delta. The hypothesis was that the upper 95% CI of PPVc1r − PPVc2r would be within the non‐inferiority margin of 15%. An assumed positive prevalence of 65% was used; the rationale for this assumed prevalence is detailed in Supplementary Appendix 2.
3. RESULTS
3.1. Patients
Of the 1001 participants in the Bio‐Hermes study, a total of 547 patients had amyloid‐PET evaluations available. Of these, 537 patients had APS results, and 531 patients had plasma p‐Tau217 (MSD, Lilly) evaluations available (Figure S2). Patient demographics for the substantially overlapping populations with available APS (C2N) and p‐Tau217 (MSD, Lilly) were evaluated (Table 1). The mean age across groups was approximately 73.2 years, with over 85% of patients identifying as White. Notably, approximately 47% of patients were positive for AD amyloid pathology based on amyloid‐PET visual read. In comparison, 32% and 42% of participants were positive for AD pathology based on APS (C2N) and p‐Tau217 (MSD, Lilly) plasma‐based tests, respectively.
TABLE 1.
Patient demographics and baseline characteristics.
| APS (C2N) and amyloid‐PET N = 537 | Plasma p‐Tau217 (MSD, Lilly) and amyloid‐PET N = 531 | |
|---|---|---|
| Clinical diagnosis | ||
| MCI due to AD, n (%) | 289 (53.8) | 289 (54.4) |
| AD dementia, n (%) | 248 (46.2) | 242 (45.6) |
| Age, mean (SD), years | 73.2 (6.6) | 74.0 (6.6) |
| Females, n (%) | 290 (54.0) | 286 (53.9) |
| Race, n (%) | ||
| White | 458 (85.3) | 453 (85.3) |
| Asian | 10 (1.9) | 10 (1.9) |
| Black or African American | 62 (11.5) | 61 (11.5) |
| Other | 7 (1.3) | 7 (1.3) |
| Ethnicity, n (%) | ||
| Not Hispanic or Latino | 459 (85.5) | 453 (85.3) |
| Hispanic or Latino | 71 (13.2) | 71 (13.4) |
| Unknown | 7 (1.3) | 7 (1.3) |
| Prevalence PET positive (visual read), n (%) | 254 (47.3) | 247 (46.5) |
| Prevalence fluid positive, n (%) | 171 (31.8) | 224 (42.2) |
| MMSE score, mean (SD) | 25.4 (2.99) | 25.4 (3.00) |
| APOE ε4 status, n (%) | ||
| Non‐carrier | 317 (59.0) | 315 (59.3) |
| Carrier | 220 (41.0) | 216 (40.7) |
Abbreviations: AD, Alzheimer's disease; APOE4, apolipoprotein E4; APS, Amyloid Probability Score; MCI, mild cognitive impairment; MMSE, Mini‐Mental State Examination; MSD, Meso Scale Discovery; PET, positron emission tomography; SD, standard deviation.
RESEARCH IN CONTEXT
Systematic review: The authors evaluated the literature for studies comparing blood‐based biomarkers of AD pathology and amyloid‐PET in early symptomatic AD populations.
Interpretation: Plasma‐based tests were interchangeable with florbetapir‐PET stratifications for detecting patients with AD pathology by expert visual read. Using blood‐based biomarkers as an alternative to PET may increase access to both an AD diagnosis and novel disease‐modifying therapies.
Future directions: Even though this study incorporated a relatively high prevalence of typically underrepresented groups, the generalizability of blood‐based biomarkers being interchangeable with amyloid‐PET should be further evaluated. Considerations could include ethnic and racial diversity, socioeconomic status, populations with high levels of comorbidity, and impact in real‐world clinical care settings. Further, this work evaluated two plasma‐based assays for identifying AD pathology; however, other blood‐based biomarkers also warrant evaluation.
3.2. Agreement between APS (C2N) and florbetapir‐PET
Of 537 patients, 147 were positive and 226 were negative for both APS (C2N) (after removing 14.0% of patients classified as intermediate) and florbetapir‐PET visual read, yielding an OPA of 80.7% (95% CI: 76.9 to 84.1); further a PPA of 69.3% (95% CI: 62.8 to 75.2) and a NPA of 90.4 (95% CI: 86.1 to 93.5) were observed (Table 2, Table S2). The PPV was 86.0% (95% CI: 79.9 to 90.4), and the NPV was 77.7% (95% CI: 72.5 to 82.1) (Table 2). As expected, a slightly lower agreement between APS (C2N) and florbetapir‐PET visual read was observed when a single cutpoint was employed, that is, patients considered intermediate were recast as either having the presence of amyloid plaques or the absence of amyloid plaques, positive or negative to florbetapir‐PET visual read (Table 2, Table S2).
TABLE 2.
Agreement parameters between APS (C2N) and florbetapir‐PET.
| Reference standard |
Florbetapir‐PET (visual read) (N = 537) |
||
|---|---|---|---|
|
Plasma assay (prespecified stratification) |
APS (C2N) (remove intermediate: 14.0%) % (95% CI) | APS (C2N) (intermediate considered negative) % (95% CI) | APS (C2N) (intermediate considered positive) % (95% CI) |
| PPA | 69.3 (62.8 to 75.2) | 57.9 (51.7 to 63.8) | 74.4 (68.7 to 79.4) |
| NPA | 90.4 (86.1 to 93.5) | 91.5 (87.7 to 94.3) | 79.9 (74.8 to 84.2) |
| OPA | 80.7 (76.9 to 84.1) | 75.6 (71.8 to 79.1) | 77.3 (73.5 to 80.6) |
| PPV | 86.0 (79.9 to 90.4) | 86.0 (79.9 to 90.4) | 76.8 (71.1 to 81.7) |
| NPV | 77.7 (72.5 to 82.1) | 70.8 (65.9 to 75.2) | 77.7 (72.5 to 82.1) |
Abbreviations: APS, Amyloid Probability Score; CI, confidence interval; NPA, negative percent agreement; NPV, negative predictive value; OPA, overall percent agreement; PET, positron emission tomography; PPA, positive percent agreement; PPV, positive predictive value.
The PPV for APS (C2N) after removing intermediate and when intermediate was recast as absence of amyloid plaques was greater than the prespecified 80%. However, when intermediate was considered as the presence of amyloid plaques, the PPV (76.8%) did not meet this prespecified requirement. In all comparisons, the NPV did not meet the prespecified threshold of 80%. In the APS (C2N) dataset, high concordance was observed between the two florbetapir‐PET stratifications: visual read and quantitation stratified at 37 CL (Table S3).
3.3. Agreement between p‐Tau217 (MSD, Lilly) and florbetapir‐PET
Of 531 patients, 197 were positive and 192 were negative for both visual assessment of florbetapir PET and p‐Tau217 (MSD, Lilly) (18.6% of indeterminate participants removed) with an OPA of 90.1% (95% CI: 86.9 to 92.5) (Table 3, Table S4). A PPA of 92.5% (95% CI 88.1 to 95.4) and a NPA of 87.7% (95% CI: 82.6 to 91.4) were observed. Further, a PPV of 88.0% (95% CI: 83.0 to 91.6) and NPV of 92.3% (95% CI: 87.8 to 95.2) was observed. Similar to APS, a slightly lower OPA was observed when a single cutpoint was used, and indeterminates were recast as negative or positive for p‐Tau217 (MSD, Lilly) (Table 3, Table S4).
TABLE 3.
Agreement parameters between p‐Tau217 and florbetapir‐PET.
| Reference standard |
Florbetapir‐PET (visual read) N = 531 |
||
|---|---|---|---|
|
Plasma assay (prespecified stratification) |
p‐Tau217 (MSD, Lilly) (remove indeterminate: 18.6%) % (95% CI) | p‐Tau217 (MSD, Lilly) (indeterminate considered negative) % (95% CI) | p‐Tau217 (MSD, Lilly) (indeterminate considered positive) % (95% CI) |
| PPA | 92.5 (88.1 to 95.4) | 79.8 (74.3 to 84.3) | 93.5 (89.7 to 96.0) |
| NPA | 87.7 (82.6 to 91.4) | 90.5 (86.5 to 93.4) | 67.6 (61.9 to 72.8) |
| OPA | 90.1 (86.9 to 92.5) | 85.5 (82.2 to 88.3) | 79.7 (76.0 to 82.9) |
| PPV | 88.0 (83.0 to 91.6) | 88.0 (83.0 to 91.6) | 71.5 (66.3 to 76.2) |
| NPV | 92.3 (87.8 to 95.2) | 83.7 (79.2 to 87.5) | 92.3 (87.8 to 95.2) |
Abbreviations: CI, confidence interval; MSD, Meso Scale Discovery; NPA, negative percent agreement; NPV, negative predictive value; OPA, overall percent agreement; PET, positron emission tomography; PPA, positive percent agreement; PPV, positive predictive value; p‐Tau217, phosphorylated tau at threonine 217.
The PPV for p‐Tau217 (MSD, Lilly) after removing indeterminate and when indeterminate was considered negative was the same (88.0% [95% CI: 83.0 to 91.6]) and met the prespecified threshold of 80%. On the other hand, the PPV when indeterminate was considered positive was 71.5% (95% CI: 66.3 to 76.2) and did not meet the prespecified threshold. NPV for p‐Tau217 (MSD, Lilly) after removing indeterminate and when indeterminate was considered positive were the same (92.3% [95% CI: 87.8 to 95.2]) and was 83.7% (95% CI: 79.2 to 87.5) when indeterminate was considered negative; all categories met the prespecified threshold of 80% for NPV. In the p‐Tau217 (MSD, Lilly) dataset, high concordance was observed between the two florbetapir‐PET stratifications: visual read and quantitation stratified at 37 CL (Table S5).
3.4. Non‐inferiority of plasma‐based biomarkers to florbetapir‐PET quantitation (stratified at 37 CL)
Both APS (C2N) and p‐Tau217 (MSD, Lilly) were non‐inferior to florbetapir‐PET quantitation, stratified at 37 CL, for identifying patients with AD pathology as defined by florbetapir‐PET visual read (Table 4). A PPVc1r (comparison of quantitated [37 CL stratification] to visual read of florbetapir‐PET) of 95.9% and a PPVc2r (comparison of blood biomarker to florbetapir‐PET visual read) of 91.9% were obtained, a difference of 4.0% (95% CI: 1.0 to 6.8) between APS (C2N) and florbetapir‐PET stratified at 37 CL for identifying AD pathology (Table 4).
TABLE 4.
Non‐inferiority of plasma‐based biomarkers to florbetapir‐PET.
| Reference standard | Comparator 1 | Comparator 2 | N | PPVc1r | PPVc2r | PPVc1r‐PPVc2r (95% CI) |
|---|---|---|---|---|---|---|
| Amyloid‐PET visual read |
Amyloid‐PET quantitation (37 Centiloids) |
APS (C2N) | 537 | 95.9 | 91.9 | 4.0 (1.0 to 6.8) |
|
p‐Tau217 (MSD, Lilly) |
531 | 97.4 | 93.2 | 4.2 (1.4 to 6.7) |
Note: Assumes amyloid positive prevalence of 65%; intermediate/indeterminate plasma samples removed.
Abbreviations: APS, Amyloid Probability Score; CI, confidence interval; MSD, Meso Scale Discovery; PET, positron emission tomography; PPV, positive predictive value; PPVc1r : PPV between amyloid‐PET quantitation (37 centiloids) and amyloid‐PET visual read; PPVc2r : PPV between fluid biomarker and amyloid‐PET visual read; p‐Tau217, phosphorylated tau at threonine 217.
A PPVc1r of 97.4% and a PPVc2r of 93.2% were obtained, giving a difference of 4.2% (95% CI: 1.4 to 6.7) between p‐Tau217 (MSD, Lilly) and florbetapir‐PET stratified at 37 CL for identifying AD pathology (Table 4).
As the upper CIs of the differences were less than the prespecified non‐inferiority margin of 15%, both APS (C2N) and p‐Tau217 (MSD, Lilly) were deemed non‐inferior to florbetapir‐PET stratified at 37 CL for identifying AD pathology (defined as positive visual read).
4. DISCUSSION
Results from our study demonstrate that both plasma‐based tests (APS, PrecivityAD test [C2N] and p‐Tau217 [MSD, Lilly]) were in high agreement and non‐inferior to amyloid‐PET clinical trial criteria for selecting patients with amyloid pathology.
In the current study, both a clinically available (PrecivityAD [C2N]) and a research use (p‐Tau217 assay [MSD, Lilly]) plasma‐based tests were non‐inferior to amyloid‐PET quantitation, stratified at 37 CL for identifying patients with AD pathology (amyloid‐PET visual read positive). Given the alignment between amyloid‐PET visual read and quantitation, stratified at 37 CL, was very high (>95% PPV), meeting the non‐inferiority margin provides high confidence in the ability of plasma‐based tests to identify patients with AD amyloid pathology. Additionally, a clinically available plasma‐based test (PrecivityAD [C2N]) was able to rule in patients with MCI or AD (indeterminate/intermediate removed or considered negative) who would also be selected by amyloid‐PET (PPV: 86% at the observed prevalence rate). However, it was not able to consistently rule out (per predefined NPV >80%) patients who were negative by amyloid‐PET visual read. Similarly, a research use p‐Tau217 assay (MSD, Lilly) was able to identify (rule in) patients with MCI or AD (indeterminate/intermediate removed or considered negative) who were positively identified by amyloid‐PET (PPV: 88% at the observed prevalence rate). In addition, p‐Tau217 (MSD, Lilly) was able to consistently identify (rule out) the same patients who were not positively identified by amyloid‐PET visual read.
Integration of blood biomarkers for the diagnosis of AD presents a promising avenue for advancing diagnostic tools and management, beyond the limited use of CSF biomarkers and PET due to accessibility issues and their invasive nature. Increasing evidence points toward the utility of blood biomarkers for detecting AD pathology. For example, findings from an analysis of two independent cross‐sectional cohort studies in 686 patients suggest that the APS result is clinically valid for identifying the presence of brain amyloidosis when compared with amyloid‐PET as reference standard. 12 These results are in line with the observations from our study (APS [C2N] had PPV of 86% and NPV of 77.7%). However, in another Bio‐Hermes study, which compared the performance of different tests to identify amyloid‐PET positivity, p‐Tau217 was the strongest predictor of brain amyloid positivity. 11 Similarly, in our study, a slightly higher concordance was observed between p‐Tau217 (MSD, Lilly) and amyloid‐PET in comparison to the APS (C2N) (OPA [95% CI]: 90.1% [86.9 to 92.5] vs 80.7 [76.9 to 84.1], respectively). However, the difference in indeterminate/intermediate prevalence for PrecivityAD (C2N) and p‐Tau217 assay (14.0% vs 18.6%) should be noted when considering differences in OPA. Additional diagnostic testing for AD pathology may be required for patients with an indeterminate/intermediate test result. 21 Overall, these results suggest that p‐Tau217 may be more accurate than Aβ‐based blood tests for identifying AD pathology among individuals with cognitive impairment due to AD. Results from our study build on the existing evidence that points to a high concordance between blood biomarkers and amyloid‐PET. 22 , 23 , 24 , 25 , 26 Of note, findings from our study do not point to the efficacy of plasma biomarkers for identifying treatment‐related amyloid clearance.
It should be noted that Aβ plasma biomarkers and amyloid‐PET imaging capture different pools of amyloid and represent distinct biological measurements. The proteins detected in plasma, such as those measured by PrecivityAD, reflect more soluble, easily diffusible amyloid, whereas PET imaging visualizes aggregated, fibrillar amyloid deposits in the brain. 27 Therefore, differences between plasma and imaging results are expected; however, the high PPVs reported here support the idea that either of the tests would be appropriate to identify the presence of AD pathology to support diagnosis.
One of the greatest strengths of this study is the novel non‐inferiority analysis demonstrating that plasma assays would select a patient population similar to that of typical amyloid‐PET trial criteria. In addition, the use of prespecified criteria instead of “within‐study‐derived cutpoints” suggests the applicability of these cut‐off points in future studies. Although in our study only one approved amyloid‐PET tracer was included, data were acquired from multiple centers and scanners and used CL thresholds, which are transferable across tracers. However, this study also has certain limitations. The use of only two blood biomarkers in our study may limit the generalizability of results to other biomarkers. Another limitation of our study is the patient population; participants of the Bio‐Hermes study may not be representative of the general population due to the selection procedure employed by the Global Alzheimer's Platform Foundation and the selection criteria for this study. Further, the underlying assumptions, including the 65% prevalence of AD pathology, as well as the modest sample size of our study, could limit the extrapolation of these results. Lastly, the lack of confirmatory tau PET or CSF data prevents definitive assessment of tau pathology in the study cohort. Although research suggests that amyloid positivity is associated with tau positivity in individuals with MCI or dementia, 28 it remains possible that the performance of the assays assessed in this study could vary in amyloid‐positive individuals who are tau‐negative, thereby limiting the generalizability of the findings. Large prospective studies with data collected longitudinally over a long period of time and development/validation of biomarkers based on prespecified cut points are required for broader implementation of blood biomarkers in clinical practice.
Overall, the findings support the hypothesis that plasma‐based tests may be interchangeable with amyloid‐PET stratifications for identifying patients with AD amyloid pathology who may be suitable for and could benefit from treatment with novel amyloid‐targeting therapies. As clinicians gain experience and become comfortable using emerging high‐performance blood biomarkers as part of the routine clinical evaluation of individuals with cognitive impairment, a significant opportunity exists for elevating the quality of care, equitable diagnostic and treatment access, and improved clinical outcomes.
CONFLICT OF INTEREST STATEMENT
J.W., L.H., T.L., H.E.M., and M.J.P. are employees of Eli Lilly and Company and hold company stock. S.C.B. is an employee of Eli Lilly and Company and holds company stock. She also holds a patent on a method for the detection of a neurological disease. J.B.B. is an employee of C2N Diagnostics and holds company stock. R.W. is an employee of IXICO and holds company stock. R.M. is an employee of the Global Alzheimer's Platform Foundation and is the Vice President for Clinical Development at AgeneBio, Inc. He received a grant from the National Institute on Aging and received consulting fees from AgeneBio, Inc. and Amyriad Therapeutics, Inc. He is a study section member for the SBIR Review Panel at the National Institute on Aging and a DSMB member at Noah Pharmaceuticals. He is a member of the board of directors of Cogstate, Ltd., a member of the board of governors of Alzheimer's Drug Discovery Foundation, and holds stock in Eli Lilly and Co. Author disclosures are available in the Supporting Information.
CONSENT STATEMENT
All patients provided informed consent. The study was carried out in accordance with the ethical principles of the Declaration of Helsinki and Good Clinical Practice guidelines.
Supporting information
Supporting information
Supporting information
ACKNOWLEDGMENTS
The authors would like to thank all participants involved in the research studies and clinical trials, as well as their care partners and families. The authors thank the Global Alzheimer's Platform Foundation, Bio‐Hermes study team, and the research partners, investigators, and support staff involved in their research studies and clinical trials. Medical writing and editorial support were provided by Preetinder Kaur of Certara and were funded by Eli Lilly and Company, Indianapolis, IN, USA. Data used in the preparation of this article were obtained from BioHermes.
Wang J, Braunstein JB, Wolz R, et al. Interchangeability of blood‐based biomarkers and PET to identify Alzheimer's disease pathology. Alzheimer's Dement. 2026;18:e70196. 10.1002/dad2.70196
REFERENCES
- 1. Hansson O. Biomarkers for neurodegenerative diseases. Nat Med. 2021;27(6):954‐963. doi: 10.1038/s41591-021-01382-x [DOI] [PubMed] [Google Scholar]
- 2. Dubois B, Villain N, Frisoni GB, et al. Clinical diagnosis of Alzheimer's disease: recommendations of the international working group. Lancet Neurol. 2021;20(6):484‐496. doi: 10.1016/S1474-4422(21)00066-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Jack CR Jr, Bennett DA, Blennow K, et al. NIA‐AA research framework: toward a biological definition of Alzheimer's disease. Alzheimers Dement. 2018;14(4):535‐562. doi: 10.1016/j.jalz.2018.02.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Trejo‐Lopez JA, Yachnis AT, Prokop S. Neuropathology of Alzheimer's disease. Neurotherapeutics. 2022;19(1):173‐185. doi: 10.1007/s13311-021-01146-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Sims JR, Zimmer JA, Evans CD, et al. Donanemab in early symptomatic Alzheimer disease: the TRAILBLAZER‐ALZ 2 randomized clinical trial. JAMA. 2023;330(6):512‐527. doi: 10.1001/jama.2023.13239 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Van Dyck CH, Swanson CJ, Aisen P, et al. Lecanemab in early Alzheimer's disease. N Engl J Med. 2023;388:9‐21. doi: 10.1056/NEJMoa2212948 [DOI] [PubMed] [Google Scholar]
- 7. Jie C, Treyer V, Schibli R, Mu L. Tauvid: the first FDA‐approved PET tracer for imaging tau pathology in Alzheimer's disease. Pharmaceuticals (Basel). 2021;14(2):110. doi: 10.3390/ph14020110 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Hampel H, Au R, Mattke S, et al. Designing the next‐generation clinical care pathway for Alzheimer's disease. Nat Aging. 2022;2(8):692‐703. doi: 10.1038/s43587-022-00269-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers for the diagnosis of Alzheimer's disease: a systematic review and meta‐analysis. Lancet Neurol. 2016;15(7):673‐684. doi: 10.1016/S1474-4422(16)00070-3 [DOI] [PubMed] [Google Scholar]
- 10. Neto‐Rosa P, Zimmer E. Chapter 11: Spinal Fluid. World Alzheimer Report 2021 by Alzheimer's Disease International. Accessed May 24, 2024. https://www.alzint.org/u/World‐Alzheimer‐Report‐2021‐Chapter‐11.pdf
- 11. Mohs RC, Beauregard D, Dwyer J, et al. The Bio‐Hermes study: biomarker database developed to investigate blood‐based and digital biomarkers in community‐based, diverse populations clinically screened for Alzheimer's disease. Alzheimrs Dement. 2024;20(4):2752‐2765. doi: 10.1002/alz.13722 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Hu Y, Kirmess KM, Meyer MR, et al. Assessment of a plasma amyloid probability score to estimate amyloid positron emission tomography findings among adults with cognitive impairment. JAMA Netw Open. 2022;5(4):e228392. doi: 10.1001/jamanetworkopen.2022.8392 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kirmess KM, Meyer MR, Holubasch MS, et al. The PrecivityAD™ test: accurate and reliable LC‐MS/MS assays for quantifying plasma amyloid beta 40 and 42 and apolipoprotein E proteotype for the assessment of brain amyloidosis. Clin Chim Acta. 2021;519:267‐275. doi: 10.1016/j.cca.2021.05.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Fogelman I, West T, Barunstein JB, et al. Independent study demonstrates amyloid probability score accurately indicates amyloid pathology. Ann Clin Transl Neurol. 2023;10(5):765‐778. doi: 10.1002/acn3.51763 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. U.S. Food & Drug Administration . Amyvid® D (Florbetapir F 18 Injection) Full Prescribing Information. Accessed May 24, 2024. https://pi.lilly.com/us/amyvid‐uspi.pdf
- 16. European Medicines Agency . Summary of Product Characteristics. Amyvid, INN‐florbetapir [18F]. Accessed May 24, 2024. https://www.ema.europa.eu/en/documents/product‐information/amyvid‐epar‐product‐information_en.pdf
- 17. Harn NR, Hunt SL, Hill J, et al. Augmenting amyloid PET interpretations with quantitative information improves consistency of early amyloid detection. Clin Nucl Med. 2017;42(8):577‐581. doi: 10.1097/RLU.0000000000001693 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Navitsky M, Joshi AD, Kennedy I, et al. Standardization of amyloid quantitation with florbetapir standardized uptake value ratios to the centiloid scale. Alzheimer's Dement. 2018;14(12):1565‐1571. doi: 10.1016/j.jalz.2018.06.1353 [DOI] [PubMed] [Google Scholar]
- 19. Mintun MA, Lo AC, Duggan Evans C, et al. Donanemab in early Alzheimer's disease. N Engl J Med. 2021;384(18):1691‐1704. doi: 10.1056/NEJMoa2100708 [DOI] [PubMed] [Google Scholar]
- 20. Meijun Li. Statistical methods for clinical validation of follow‐on companion diagnostic devices via an external concordance study. Statistics in biopharmaceutical research. 2016;8(3):355‐363. doi: 10.1080/19466315.2016.1202859 [DOI] [Google Scholar]
- 21. Brum WS, Cullen NC, Janelidze S, et al. A two‐step workflow based on plasma p‐tau217 to screen for amyloid β positivity with further confirmatory testing only in uncertain cases. Nat Aging. 2023;3(9):1079‐1090. doi: 10.1038/s43587-023-00471-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Palmqvist S, Janelidze S, Quiroz YT, et al. Discriminative accuracy of plasma phospho‐tau217 for Alzheimer disease vs other neurodegenerative disorders. JAMA. 2020;324(8):772‐781. doi: 10.1001/jama.2020.12134 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Janelidze S, Mattsson N, Palmqvist S, et al. Plasma P‐tau181 in Alzheimer's disease: relationship to other biomarkers, differential diagnosis, neuropathology and longitudinal progression to Alzheimer's dementia. Nat Med. 2020;26(3):379‐386. doi: 10.1038/s41591-020-0755-1 [DOI] [PubMed] [Google Scholar]
- 24. Mattsson‐Carlgren N, Janelidze S, Palmqvist S, et al. Longitudinal plasma p‐tau217 is increased in early stages of Alzheimer's disease. Brain. 2020;143(11):3234‐3241. doi: 10.1093/brain/awaa286 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Palmqvist S, Tideman P, Mattsson‐Carlgren N, et al. Blood biomarkers to detect Alzheimer disease in primary care and secondary care. JAMA. 2024;332(15):1245‐1257. doi: 10.1001/jama.2024.13855 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Schindler SE, Petersen KK, Saef B, et al. Head‐to‐head comparison of leading blood tests for Alzheimer's disease pathology. Alzheimer's Dement. 2024;20(11):8074‐8096. doi: 10.1002/alz.14315 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Roberts BR, Lind M, Wagen AZ, et al. Biochemically‐defined pools of amyloid‐β in sporadic Alzheimer's disease: correlation with amyloid PET. Brain. 2017;140(5):1486‐1498. doi: 10.1093/brain/awx057 [DOI] [PubMed] [Google Scholar]
- 28. de Leeuw DM, Trieu C, Vromen EM, et al. Cerebrospinal fluid amyloid and tau biomarker changes across the Alzheimer disease clinical spectrum. JAMA Netw Open. 2025;8(7):e2519919. doi: 10.1001/jamanetworkopen.2025.19919 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting information
Supporting information