Tau biology, biomarkers, and therapeutics

Abstract

As Alzheimer's disease (AD) research advances, tau has emerged as both a critical biomarker and a promising therapeutic target, central to understanding disease mechanisms, tracking progression, and guiding treatment development. The Fall 2024 Alzheimer's Association Research Roundtable convened experts from academia, industry, National Institutes of Health (NIH), and the United States Food and Drug Administration (FDA) to explore current progress and future directions in tau‐centered diagnostics and therapeutics. Discussions addressed the integration of tau biomarkers into the 2024 Revised Diagnostic Criteria for AD, updates to amyloid and tau positron emission tomography (PET) imaging, and modeling of biomarker trajectories. Presenters highlighted tau‐targeting therapeutics including antisense oligonucleotides, monoclonal antibodies, and small molecules, alongside innovations in drug delivery. The interplay between anti‐amyloid and anti‐tau therapies and strategic design of combination trials were key themes. Regulatory insights facilitated discussions on drug approval pathways. The meeting highlighted the rapid evolution of tau research and emphasized opportunities to improve diagnostics, trial design, and treatment strategies in AD.

Highlights

• The Alzheimer's Association Research Roundtable (AARR) convened leaders from industry, academia, and government, to explore current progress and future directions in tau‐centered diagnostics and therapeutics.

• Tau has emerged as both a critical biomarker and a promising therapeutic target, central to understanding disease mechanisms, tracking progression, and guiding treatment development.

• Discussions addressed the integration of tau biomarkers into the 2024 revised diagnostic criteria for AD, updates to amyloid and tau PET imaging, modeling of biomarker trajectories, tau‐targeting therapeutics, and interplay between anti‐amyloid and anti‐tau therapies and strategic design of combination trials.

1. INTRODUCTION

The microtubule‐associated protein tau has emerged as a key biomarker and therapeutic target in Alzheimer's disease (AD) and other tauopathies as it plays a central role in the disease's pathophysiology, progression, and clinical manifestations. In the adult human brain, tau exists in multiple isoforms generated through alternative splicing of the MAPT gene, which vary by the presence of either three (3R) or four (4R) microtubule‐binding repeats and one or two N‐terminal exons. ¹ These isoforms differ in their structure, function, and propensity to aggregate. Tau is subject to extensive post‐translational modification, including phosphorylation, truncation, ubiquitination, and acetylation, which generate distinct proteoforms with unique biological properties. ² Tau protein is predominantly axonal under physiological conditions, but redistributes to somatodendritic compartments in settings of pathology. ³ Pathogenic forms of tau can form β‐sheet–rich aggregates with varying structures depending on disease context, as shown by cryo‐EM studies. ⁴ , ⁵ Some tau species are neurotoxic and capable of propagating pathology via a prion‐like mechanism known as “seeding,” in which pathological tau templates the misfolding of native tau in recipient cells. ⁶ , ⁷ Tau seeds vary in potency and structure, which may underlie disease heterogeneity among tauopathies. Notably, tau fluid and imaging biomarkers have been incorporated into the 2024 Revised Criteria for Diagnosis and Staging of AD: Alzheimer's Association Workgroup, ⁸ reinforcing the critical role of tau in refining the biological understanding of AD, improving diagnostic precision, and guiding disease staging.

Although recent advances in anti‐amyloid‐beta (Aβ) therapies have marked a significant milestone in AD treatment, there remains an urgent need to translate the growing body of knowledge on tau biology into clinical applications with the aim of enhancing patient outcomes in dementia care. Bridging the gap between fundamental tau biology and its clinical applications is essential for the development of biomarkers that improve early and differential diagnosis, prognostic assessment, and treatment response monitoring, as well as the development of effective tau‐targeted therapeutic strategies.

The Fall 2024 Alzheimer's Association Research Roundtable (AARR) provided a unique forum for leading experts from the AARR membership, academia, the National Institutes of Health (NIH), and the United States Food and Drug Administration (FDA) to discuss the latest advances in tau‐related diagnostics and therapeutics. The meeting facilitated critical discussions on recent insights into tau function, the evolution of tau biomarkers in both imaging and biofluid platforms, and the progress of tau‐targeted interventions. By integrating these developments, the meeting aimed to advance the understanding of tau's role in AD pathophysiology and foster continued innovation in biomarker discovery and tau‐based therapeutic development for AD and other neurodegenerative diseases. This manuscript synthesizes key discussions from the meeting, offering an overview of state‐of‐the‐art research, emerging biomarkers and novel treatments, as well as ongoing challenges in the field.

2. Tau Biomarkers

2.1. Tau biomarkers and the 2024 revised criteria for diagnosis and staging of AD

Blood‐ and cerebrospinal fluid (CSF) ‐based tau biomarkers form part of the Core biomarkers outlined in the 2024 Revised Criteria for Diagnosis and Staging of AD. ⁸ Among tau blood‐based biomarkers, plasma phosphorylated tau (p‐tau), particularly p‐tau217, shows significant promise in detecting AD pathology. A recent meta‐analysis comprising 113 studies with 29,625 unique individuals compared the accuracy of multiple plasma p‐tau assays for detecting biologically confirmed AD. The analysis found p‐tau217 to be the most accurate plasma p‐tau biomarker, reporting a pooled sensitivity of 88.1% (95% confidence interval; [CI]: 86.7‐89.5%), a specificity of 88.7% (95% CI: 87.4‐89.9%), and a diagnostic odds ratio of 50.7 (95% CI: 40.6‐63.4). ⁹ Studies have demonstrated that %p‐tau217 (ratio of p‐tau217 to non‐p‐tau) is comparable to or surpasses FDA‐approved CSF tests in predicting amyloid pathology in symptomatic individuals with mild cognitive impairment (MCI) or mild dementia. ¹⁰ Furthermore, plasma p‐tau217 has been shown to outperform standard clinical assessments (clinical examination, cognitive testing, and a computed tomographic scan) in identifying patients with AD pathology, demonstrating high diagnostic accuracy in both primary and secondary care settings. These findings highlight its potential for early and accurate detection of AD pathology across diverse healthcare settings. ¹¹

In a study from the Mayo Clinic Study of Aging, plasma p‐tau217 concentrations were estimated to increase after amyloid positron emission tomography (PET) and before tau PET positivity. ¹² Accordingly, plasma p‐tau217 assays demonstrate stronger associations with amyloid than with tau PET. ¹³ Some clinically available p‐tau217 immunoassays have been optimized for the prediction of amyloid pathology using a 2‐cutpoint model that categorizes results as negative, intermediate, or positive. ¹⁴ Using this approach, 15%–20% of patients in the intermediate category would need additional testing to confirm amyloid pathology while maintaining high and low probabilities of an abnormal amyloid PET result for positive and negative outcomes, respectively. ¹⁵ Plasma p‐tau217 also has the potential to guide referrals for tau PET, as low p‐tau217 concentrations predict low amyloid and, therefore, a low likelihood of sufficient tangle pathology for tau PET to provide clinically useful spatial information. ¹⁶ This approach is particularly valuable in clinical trials, where precise staging is essential for evaluating the efficacy of disease‐modifying therapies.

Blood‐based tau biomarkers also provide utility in predicting amyloid plaque accumulation over time. Longitudinal studies involving cognitively unimpaired individuals with subthreshold amyloid PET levels showed that elevated p‐tau217 and reduced amyloid‐beta (Aβ) 42/40 ratios at baseline predict future amyloid plaque progression, as measured by serial PET imaging. When combined, these biomarkers enhance predictive accuracy, offering a promising approach to identifying individuals at risk for amyloid accumulation. ¹⁷

Anti‐Aβ therapy trials report reductions in plasma p‐tau181 and p‐tau217 with lecanemab and donanemab treatment, respectively. ¹⁸ , ¹⁹ After discontinuation of lecanemab, increases in p‐tau181 concentrations were observed more quickly than changes in amyloid PET. However, a deeper understanding of how changes in these biomarkers at the individual level reflect treatment response in relation to amyloid pathology, cognitive changes, and other relevant clinical endpoints is necessary before p‐tau biomarkers can be adopted for treatment monitoring.

In CSF, the 2024 Revised Criteria uses hybrid ratios of p‐tau181/Aβ42 and total tau (t‐tau)/Aβ42 to identify amyloid pathology in symptomatic individuals. These FDA‐approved ratios are utilized in clinical practice and demonstrate excellent concordance with amyloid PET. ²⁰ However, in a small percentage of participants (<5%), discrepancies between CSF levels of p‐tau181/Aβ42 and amyloid PET, as well as between p‐tau181/Aβ42 and Aβ42/Aβ40, ²¹ are observed. Such inconsistencies may introduce uncertainty regarding treatment decisions for anti‐Aβ therapies. Potential causes for these discrepancies may include early disease stage, co‐pathologies, or preanalytical aspects of sample collection. ²² , ²³ , ²⁴ Identifying the underlying causes of these discrepancies and adherence to the manufacturers’ recommended pre‐analytical protocol for collection and sample handling that follows the protocol standardized by the Alzheimer's Association is crucial to enhancing confidence in reliably detecting AD pathology.

Although tau PET imaging is also an integral component of the 2024 Revised Criteria, tau PET outcome data for individuals with AD cannot currently be meaningfully compared or combined due to differences in tracer properties, instrumentation, and analytical methods. A Working Group of experts across industry and academia led by the Critical Path Institute's (C‐Path) Critical Path for Alzheimer's Disease Consortium (CPAD) was convened at the 2022 Alzheimer's Association International Conference (AAIC) to discuss standardization of quantitative tau PET imaging. The Working Group developed the Joint Propagation Model (JPM), which simultaneously models the relationships of data from anchor point subjects and data from subjects in head‐to‐head studies, providing conversion equations for multiple tracers to a common scale in units termed “CenTauRs”. ²⁵

Using a set of five cortical regions of interest (ROIs), a strong linear relationship was observed between CenTauR values across cohorts. Using the JPM approach, CenTauR estimates were similar to but more accurate than those derived using a linear regression approach akin to that used for the Centiloid scale for amyloid PET. Across cognitively impaired individuals scanned with various tau PET tracers and matched for age, Mini‐Mental State Examination (MMSE) scores, and amyloid burden (measured in Centiloids), the prevalence of tau PET positivity was found to be remarkably consistent across tracers (flortaucipir, 78%; MK‐6240, 76%; RO948, 75%). This consistency was determined using a JPM‐based CenTauR threshold of 18 in the meta‐temporal region. The cutoff was derived through ROC analysis, identifying the CenTauR value that most effectively distinguished between FDA‐approved positive and negative visual reads for flortaucipir.

In summary, the integration of plasma and CSF tau biomarkers, alongside imaging modalities, provides a comprehensive framework for diagnosing and staging AD. Plasma biomarkers, particularly p‐tau217, demonstrate high diagnostic accuracy and predictive capabilities, paving the way for broader acceptance in clinical and research settings. However, in order to fully access the diagnostic accuracy of plasma p‐tau217 assays, additional studies using predefined thresholds and that incorporate ethnic diversity as well as comorbidities known to affect p‐tau217 levels are still needed. ²⁶ , ²⁷ As biomarker research progresses, implementation of standardized methodologies is key to harmonizing tau PET results across distinct AD research and interventional clinical trials. Development of certified reference materials and methods for the standardization of p‐tau217 measurements is also underway, and would be crucial for accurate and reliable measurements of plasma p‐tau217 across different assays and laboratories. ²⁸ Furthermore, the development of blood‐based staging methods is expected to transform the diagnosis and management of AD, enhancing precision and scalability across diverse populations.

2.2. Updated appropriate use criteria for amyloid and tau PET

The first Appropriate Use Criteria (AUC) for amyloid PET imaging was published in 2013, led by the Amyloid Imaging Taskforce of the Alzheimer's Association and the Society of Nuclear Medicine and Molecular Imaging (SNMMI). ²⁹ , ³⁰ In light of significant advances over the past decade, the Alzheimer's Association and SNMMI convened a new multidisciplinary workgroup to update these criteria for amyloid PET and, for the first time, develop corresponding AUC for tau PET in clinical practice. ³⁰

An independent, comprehensive literature review conducted by Oregon Health Sciences served as the evidence base for the workgroup's deliberations. Using a Delphi procedure, the group rated 17 real‐world clinical scenarios on a nine‐point scale, where 7–9 indicates an appropriate use, 4–6 indicates uncertainty, and 1–3 indicates rare appropriateness. Amyloid and tau PET tests were considered separately and given their own rating for each scenario. Overall, the strongest evidence for their use includes: assessment and prognosis for individuals with mild cognitive impairment, assessment of people with dementia when the cause is not clearly known, determining eligibility for treatment with new disease‐modifying therapies, and monitoring response to these treatments. ³⁰

Overall, these updated AUC were designed to help clinicians determine when amyloid or tau PET could be beneficial in guiding diagnosis of individuals who have, or are at risk of, cognitive decline, and management of individuals with MCI or dementia. Although they are primarily aimed at dementia specialists who devote a substantial portion of their clinical practice to patients with cognitive issues, these revised AUCs also serve as a helpful reference for a broader group of professionals interested in using amyloid and tau PET in clinical settings.

Researchers are also investigating how different tau PET thresholds (from trials such as Clarity AD and TRAILBLAZER) influence patient enrichment in AD studies. Applying the Clarity AD thresholds for tau PET, most CELIA participants fall into the low‐to‐intermediate tau range. When using the TRAILBLAZER thresholds, which rely on a different tau tracer (flortaucipir, Tauvid) and a harmonization approach, about 30% of CELIA participants register as no/very low tau, around 50% as low/intermediate, and roughly 19% as high. Comparing these two threshold systems suggests that the classification of participants varies depending on which set of cutoffs and tracers is used, highlighting the importance of selecting appropriate tau enrichment criteria in clinical trials.

2.3. Temporal modeling of AD biomarker trajectories to predict outcomes of tau therapies

Recent developments in temporal modeling methods and applications of these approaches to observational longitudinal biomarker studies indicate that several AD biomarkers, including amyloid PET, tau PET, and plasma p‐tau217 accumulation trajectories are highly predictable once above critical levels (i.e., positivity thresholds) regardless of common dementia risk factors such as age, sex, and apolipoprotein E (APOE) ‐e4 status. ¹² , ³¹ , ³² , ³³ , ³⁴ The predictable nature of AD biomarker accumulation can be leveraged to characterize biomarker accumulation over time, which can then be used to estimate the age at which key biomarker transitions occur in a given individual (e.g., A‐ to A+, T‐ to T+).

Anchoring biomarker, clinical, and cognitive observations to biomarker‐based timelines provides an intuitive approach to study, characterize, and communicate biomarker progression and biomarker‐informed dementia risk, and to identify factors that contribute to accelerated disease progression. ³² , ³⁵ , ³⁶ Studies using these methods recapitulate the expected AD biomarker and clinical sequence on average ³⁷ but also reveal heterogeneity in the ages at which individuals become A+ and T+, the time from A+ to T+, and the time from A+ to dementia, with the latter ranging from nearly coincident to >20 years. Relative to amyloid, there is less heterogeneity from meta‐temporal T+ to dementia onset (Clinical Dementia Rating [CDR] Global = 1) and fewer factors that appear to impact this timing. ³⁸ For both amyloid and tau biomarkers, older age of A+ and older age of T+ are associated with shorter time to dementia.

Important for clinical trial planning and clinical implementation, estimated A+ and T+ ages provide a metric to interpret biomarker‐based dementia risk in terms of estimated years to symptom onset such as the probability of remaining dementia‐free 5, 10, or more years after biomarker positivity. Portraying biomarker risk in terms of time is likely more readily understood by patients and caregivers than centiloid values or biomarker concentrations/ratios. When determining optimal populations and timing for amyloid‐targeting and emerging tau‐targeting treatments, it will be important to consider the age of A+, the age of T+, and other health factors in the context of competing risks of death to ascertain the likelihood that individuals will experience dementia within their lifetime.

3. Animal Studies and Tau Pathology

3.1. Humanized knock‐in mice for amyloidosis and tauopathy

The development of humanized amyloid precursor protein (APP) and tau knock‐in mice represents a significant advancement in modeling AD‐type amyloidosis and tauopathy, addressing limitations associated with early transgenic models that relied on artificial overexpression systems. APP knockin mice harbor various mutations, including the Swedish, Iberian, and Arctic mutations, that lead to elevated levels of Aβ40 and Aβ42, modeling the amyloidosis seen in human AD. ²⁷ Tau knockin mice were developed by replacing the entire murine microtubule‐associated protein tau (MAPT) gene with the human ortholog, creating an in vivo platform for studying human tauopathy. ²⁸ Attempts to generate an AD‐like model by crossing APP knock‐in mice with wild‐type MAPT mice resulted in limited success. Currently, no available mouse model fully replicates the human AD brain, which requires the presence of both wild‐type APP and MAPT.

Murine models harboring MAPT mutations that are associated with human frontotemporal dementia (FTD) provide a valuable resource to investigate mechanisms that are specifically relevant to tau‐induced neurotoxicity. FTD‐associated MAPT mutations primarily affect exon 10 splicing, altering the ratio of 3R to 4R tau isoforms. Pathological tau formation in FTD is often associated with a shift from the normal 1:1 ratio toward an excess of either 3R or 4R tau. Among the most studied mutations in MAPT are S305N and P301S, each of which have distinct effects on tau pathology. Research presented at the meeting showed that S305N MAPT mutant mice exhibit increased tau phosphorylation, synaptic loss, and cognitive deficits, despite a lack of tau seeding and aggregation. In contrast, P301S MAPT mutant mice show early tau seeding and progressive tangle pathology, but no overt phosphorylation at earlier stages.

New mouse models are being developed in the Model AD Consortium using gene replacement technology, where key genes implicated in AD are fully replaced with their human orthologs. These models include control lines with wild‐type human alleles and variant lines carrying high‐impact pathogenic mutations, allowing for precise investigation of genetic contributions to disease pathology. ²⁹

Overall, humanized knock‐in mouse models provide valuable insights into amyloidosis and tauopathy in AD. These models allow for the investigation of specific disease mechanisms, including tau hyperphosphorylation, aggregation, synaptic dysfunction, and cognitive decline, without the confounding effects of artificial overexpression. They also underscore the importance of aligning preclinical models with specific therapeutic targets and highlight the potential for knock‐in mice to facilitate biomarker discovery and drug development in AD research.

3.2. Role of soluble multimeric species of tau in AD

Tau exists in multiple forms within the brain, including abundant physiological forms of tau, insoluble aggregated fibrillar species that are visible with tau‐specific PET imaging, and soluble multimeric species (also called high‐molecular‐weight [HMW] tau) that remain invisible to current in vivo imaging techniques. Although neurofibrillary tangles (NFTs) are a defining neuropathology of AD, emerging evidence highlights the pathogenic significance of soluble, HMW tau species.

Studies using post mortem AD tissue, brain interstitial fluid from tau transgenic mice, and cellular models report that phosphorylated HMW tau is released from neurons into the brain interstitial fluid, where it demonstrates efficient uptake by neurons, transport along axons, and propagation to synaptically‐connected neurons. ³⁹ , ⁴⁰ , ⁴¹ Although sarkosyl‐insoluble tau is abundant (∼150 µg/g) in AD brains, HMW tau is present at only ∼1 µg/g. ³⁹ This rare species can be identified through super‐resolution microscopy or amplification assays. Despite its scarcity, HMW tau exhibits potent bioactivity, making it traceable through seeding assays.

Post‐translational modification studies showed that HMW tau differs molecularly from paired helical filaments (PHFs). Key modifications indicate that while HMW tau shares some post‐translational modifications with sarkosyl‐insoluble tau, it is not simply fragmented fibrils but a distinct species with a unique biochemical profile. ⁴² , ⁴³

Bioactive oligomeric forms of tau have been observed at pre‐ and postsynaptic sites in AD patients and experimental models, unlike filamentous tau. ⁴⁴ For example, the long studied conformational tau antibody Alz50 recognizes synaptic terminal zones in Alzheimer's tissue. ⁴⁵ A direct effect on synaptic function has also been observed. Microinjection of soluble HMW tau derived from human brain affected by Alzheimer's disease into neurons led to decreased bursting, ⁴⁶ consistent with earlier observations of diminished neural activity in tau transgenic mice independent of tangle formation. ⁴⁷ Furthermore, a cross‐sectional study of 40 post mortem brains with early‐stage AD demonstrated that synapses containing tau oligomers are selectively and excessively engulfed by microglia and astrocytes in individuals with dementia but not those without dementia and identical intermediate Braak III to IV stages. ⁴⁸

Longitudinal in vivo imaging in tau transgenic mice report that neuronal loss is more closely associated with tau expression than with tangle formation, with non–tangle‐bearing neurons showing a threefold higher risk of death compared to neurons containing tangles. Suppressing tau synthesis via doxycycline reduces neuronal loss by approximately half despite the persistence of tangles, as demonstrated in both rTg4510 and Thy22 mouse models of tauopathy, reinforcing the pathogenic role of soluble tau species over fibrillar aggregates. ⁴⁹

Taken together, soluble multimeric forms of tau accumulate at synapses, impair synaptic function, predispose neurons to death, and contribute to the propagation of misfolded tau between neurons. Although animal models and human data suggest that soluble misfolded tau aggregates may be more harmful than the tangles themselves, measuring and targeting these rare intracellular tau species remains a significant challenge for future therapeutic strategies.

4. Tau‐Targeted Therapies in AD

4.1. Emerging Tau‐Targeted Therapeutic Strategies

Targeting tau in AD is compelling because tau is both a neuropathological hallmark of the disease and a driver of neurodegeneration that correlates closely with, and even predicts, cognitive decline. ⁵⁰ , ⁵¹ To date, tau‐targeted therapeutic strategies have ranged from immunotherapies aimed at blocking inter‐neuronal spread to intracellular approaches targeting aggregated or modified tau species, as well as interventions designed to lower levels of tau RNA. ⁵² Largely inspired by progress in the amyloid field, innovations in clinical development have accelerated this work, including the recently published 2024 Revised Criteria for Diagnosis and Staging of AD ⁸ and tools such as fluid and imaging biomarkers, particularly blood‐based diagnostics that enable more precise hypothesis testing and streamlined participant recruitment in clinical trials at various stages of the disease.

Although extracellular tau species, thought to be a major contributing factor to the pathological spread of tau in AD, have been the target of several antibody‐based clinical trials, the as‐yet‐undefined nature of the specific tau species that are secreted from cells, taken up by synaptically connected neurons and drive templated aggregation in the recipient neuron presents a challenge to this approach. ⁵³ The initial failure of antibodies targeting the N‐terminus of tau protein, for example, may be due to cleavage of this domain from the core aggregation domain prior to cellular secretion. ⁵⁴ Subsequent approaches have thus shifted focus toward the “mid region” of tau near the microtubule‐binding region (MTBR). ⁵⁵ , ⁵⁶ Recent data from the mid‐region tau antibody bepranemab was reported to show reduced accumulation of tau, alongside reduced cognitive decline in ApoE4 non‐carrier participants or those with low tau pathology at baseline, suggesting a therapeutic window preceding substantial tau accumulation. ⁵⁷

Antisense oligonucleotides (ASOs) targeting tau RNA species present an alternative therapeutic strategy designed to reduce tau levels more generally. ASO‐based reduction of human tau in the PS19 mouse model of tauopathy was found to protect against neuronal death and reverse pre‐existing tau pathology; intrathecal delivery of tau‐targeting ASOs was also reported to reduce tau RNA and protein in non‐human primates. ⁵⁸ These and other preclinical studies enabled testing of the tau ASO BIIB080 administered intrathecally to enter AD clinical trials. Recent Phase 1b data show significant CSF tau protein reduction (∼60%). Data in a small subset of patients who underwent tau PET also suggest reduction of tau aggregates – if confirmed in the ongoing Phase 2 study, this would be a human first. ⁵⁹ , ⁶⁰ Although this study was not powered to measure changes in cognition, a comparison of a propensity‐matched external control arm suggested trends toward slowing of cognitive decline, which will need to be confirmed in Phase 2. Other intrathecally administered oligonucleotides undergoing Phase 1 clinical trials of AD patients are the anti‐MAPT ASO NIO752, and the anti‐MAPT siRNA LY3954068.

Together, these developments underscore the complexity of tau biology. Pathological forms of tau have been a challenging target in drug development. The susceptibility of tau to differential splicing, proteolytic cleavage, post‐translational modifications, conformational changes, and aggregation presents major challenges to drug development, as does a lack of longitudinal biomarkers and bypassing of the blood–brain barrier. ⁵³ Development of precision medicine tools, tailored clinical trial designs, and a deeper understanding of which tau species and delivery routes are clinically relevant are poised to accelerate the field. By matching therapies to the specific stage of tau pathology and measuring outcomes with increasingly sensitive biomarkers, researchers may finally overcome the hurdles that have long impeded effective tau‐targeted treatments for AD. Several novel treatment strategies targeting tau pathology were highlighted at the 2024 Fall AARR meeting, a summary of which is provided below.

4.1.1. Vectorized anti‐tau siRNA and VY‐TAU01 tau antibody

Voyager Therapeutics is developing two complementary approaches to target pathological tau: an anti‐tau antibody, VY7523, and a tau‐silencing gene therapy, VY1706.

VY7523, an intravenously administered monoclonal antibody, was designed to inhibit the spread of pathological tau by selectively binding its C‐terminal domain. The team screened over 700 antibodies before identifying VY7523, which showed over 200‐fold selectivity for pathologic tau over human wild‐type tau. In P301S transgenic mice, VY7523 reduced propagation of pathogenic tau by ∼70%; anti‐tau antibodies such as gosuranemab and zagotenemab failed to suppress tau propagation in the same model. When compared to bepranemab, VY7523 demonstrated similar inhibition of pathologic tau spread, suggesting potential biological efficacy. VY7523 is currently in a first‐in‐human clinical trial, starting with a single ascending dose (SAD) study in healthy volunteers, followed by a multiple ascending dose (MAD) trial in early AD patients, which will assess its impact on pathological tau spread.

The second approach, VY1706, is a gene‐silencing therapy using a novel adeno‐associated virus (AAV) vector designed to cross the blood–brain barrier intravenously, a challenge for traditional AAV9 and AAV5 vectors. Voyager developed bar‐coded capsid libraries, screening them in primates to identify capsids that efficiently penetrate the brain while minimizing off‐target effects, particularly in the liver. Using these capsids, they engineered a small interfering RNA (siRNA) therapy to silence tau expression, which, when administered intravenously to P301S mice, lowered tau mRNA by ∼80% in the cortex and brainstem and significantly reduced the total and pathological tau in these regions. A similar effect was observed in human tau transgenic mice, demonstrating robust tau suppression at both mRNA and protein levels.

By pursuing both antibody‐based clearance and gene therapy‐driven tau suppression, these therapeutics aim to determine whether halting or reversing tau pathology can yield meaningful clinical benefits in AD.

4.1.2. ASOs targeting tau

As mentioned previously, the ASO therapy BIIB080 targets tau RNA and is thus designed to lower overall tau protein levels. In a Phase 1 study (CS1, NCT03186989), multiple intrathecal doses administered to patients with mild AD produced robust and sustained decreases in CSF total and p‐tau. ⁶⁰ Tau pathology, measured by MK6240 tau PET, also declined. ⁵⁹ Tau PET analyses across various Braak composite regions and subcortical structures showed clear decreases in the standardized uptake value ratio (SUVR) compared to external matched controls from the gosuranemab BIIB092 trial, which served as a natural history comparison. An exploratory whole‐brain evaluation likewise showed broad reductions across temporal, parietal, occipital, and frontal regions in BIIB080 recipients, whereas matched controls from the BIIB092 trial exhibited SUVR increases over time.

To expand upon these findings, the Phase 2 CELIA trial (NCT05399888) is investigating BIIB080 in participants with MCI due to AD or mild AD dementia. This randomized, double‐blind, placebo‐controlled study includes multiple‐dose regimens and tracks clinical outcomes such as CDR Sum of Boxes (CDR‐SB), ADCS‐ADL‐MCI, and ADAS‐Cog‐13, as well as biomarker endpoints (tau PET, amyloid PET, plasma, and CSF markers). The 18‐month placebo‐controlled portion will be followed by a 24‐month open‐label extension, allowing for a comprehensive assessment of long‐term safety and efficacy.

Overall, the BIIB080 clinical program is positioned to test the impact of reducing both intracellular and extracellular tau in early AD. Its Phase 2 results will help better understand the temporal and spatial dynamics of biomarkers and clinical effects in a larger and broader population.

4.1.3. Anti‐MTBR‐tau antibody (E2814)

E2814, a human IgG1 antibody that binds to MTBR‐tau, is being developed as a therapeutic agent for AD to inhibit the spread of tau aggregates between cells and to slow the subsequent decline in cognitive function. An open‐label study (Study 103: NCT04971733) in patients with mild to moderate cognitive impairment due to dominantly inherited AD (DIAD) is currently ongoing to evaluate the safety, pharmacokinetics, and the target engagement of E2814 by determining the ratio of antibody‐bound and unbound forms of MTBR‐tau299 and 354 peptides (including the E2814 epitopes) in CSF. In addition, T1 and T2 biomarkers of the 2024 Revised Criteria for Diagnosis and Staging of AD, ⁸ including p‐tau217 and MTBR‐tau243 ⁶¹ as well as tau PET, will be assessed.

In Study 103 in DIAD patients, E2814 was safe and well tolerated over a broad dose range, confirming antibody concentration‐dependent changes in target engagement in the CSF. Furthermore, E2814 robustly reduced CSF MTBR‐tau243 (−50% and −75% at 3 and 9 months, respectively) as well as p‐tau217 (−50% at 24 months). Three DIAD participants for whom tau PET was acquired at 108 weeks demonstrated either a trending reduction in tau from baseline or no further tau accumulation. These results support further evaluation in the ongoing Phase 2/3 DIAN‐TU Tau NexGen Platform Study (NCT05269394) in DIAD and future trials in sporadic AD populations.

4.1.4. Anti‐p‐tau monoclonal antibody posdinemab (JNJ63733657)

Posdinemab (JNJ‐63733657) is a high‐affinity anti‐p‐tau IgG1 monoclonal antibody (mAb) generated via an immunization campaign directed toward human AD brain‐derived paired‐helical filaments of tau. ⁶² It is hypothesized to slow the spread of tau seeds in the brain by targeting extracellular tau phosphorylated at Ser212 and Ser217. It differentiates from first‐generation N‐terminal targeting tau mAbs in epitope (located in the proline‐rich domain) and specificity for p‐tau, and by its different isotype (IgG1 vs. IgG4). In an in vitro cell depletion assay and in in vivo studies in transgenic mice, posdinemab interfered with seeding better than first‐generation tau mAbs. ⁶³ In a Phase 1 single and multiple ascending dose study, posdinemab was well tolerated and exhibited linear pharmacokinetics that were similar in healthy volunteers and patients with early AD (MCI or mild dementia due to AD). The half‐life of posdinemab was 18–26 days, the CSF‐to‐serum ratio was 0.2%, and dose‐dependent reductions in free CSF p‐tau217 were observed in healthy volunteers and AD patients. ⁶² Posdinemab is currently being tested in a randomized, double‐blind, placebo‐controlled, common close Phase 2b trial, Auτonomy (NCT04619420) in patients with early symptomatic AD with intermediate tau burden, as identified by plasma p‐tau217 and confirmed with tau PET. Two doses are compared with placebo in a 1:1:1 randomization ratio. The primary endpoint is the change from baseline to week 104 on the integrated AD Rating Scale MCI version (iADRS‐MCI; composite of ADAS‐Cog13 plus ADCS‐ADL‐MCI). Secondary endpoints include change from baseline on tau PET, ADAS‐Cog13, ADCS‐ADL‐MCI, Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), and CDR‐SB. Enrollment was completed in 2023, with 523 patients randomized across 115 sites in 10 countries. Using plasma p‐tau217 as an initial screening test reduced required tau PET scans by approximately 50%. ⁶⁴

4.1.5. Phase 2 study of ceperognastat, an orally available O‐linked N‐acetyl glucosaminidase inhibitor for the treatment of early symptomatic AD

O‐Linked N‐acetyl glucosaminidase (OGA) is a glycoside hydrolase that catalyzes the removal of O‐linked β‐N‐acetyl glucosamine (O‐GlcNAc) from intracellular proteins. OGA inhibition increases O‐GlcNAc on tau protein and slows the accumulation of hyperphosphorylated, insoluble tau, including neurofibrillary tangles. ⁴⁷ , ⁴⁸ , ⁴⁹

PROSPECT‐ALZ (NCT05063539) is a multicenter, double‐blind, randomized, placebo‐controlled Phase 2 study of ceperognastat, an orally available OGA inhibitor. The study enrolled participants with early symptomatic AD. Participants were randomized 1:1:1 to receive once daily 0.75 or 3.0 mg ceperognastat or placebo. The primary endpoint evaluated clinical progression in participants with low‐medium tau PET levels at baseline, measured by the integrated AD Rating Scale. Secondary endpoints included cognitive and functional scales and AD biomarkers.

Baseline characteristics were balanced across cohorts. Primary and secondary endpoints failed to show significant dose‐dependent benefit of ceperognastat versus placebo. The 3.0‐mg arm demonstrated greater cognitive decline compared with placebo. In contrast, volumetric magnetic resonance imaging (MRI) showed less decrease in whole brain and hippocampal volume and less increase in ventricular volume in both treatment arms compared with placebo at 76 weeks. Depending on the region, tau PET demonstrated less increase from baseline at 76 weeks in the primary cohort with 3.0 mg. Plasma p‐tau217 levels showed less increase from baseline at multiple time points in both treatment arms compared with placebo. The glial fibrillary acidic protein showed an initial decrease from baseline and remained below placebo in both treatment arms. Common adverse events with ceperognastat included headache, cardiac‐related events, and weight decrease. More serious and severe treatment‐emergent adverse events were associated with ceperognastat. There were more study discontinuations in the 3.0‐mg arm compared with placebo.

Despite evidence of potential disease‐modifying biomarker effects, the study did not demonstrate a clinical benefit, making it unclear whether this drug class has an effective therapeutic dosing window. PROSPECT‐ALZ is ongoing, with a 6‐month safety follow‐up extension study underway.

4.2. Relationship of anti‐amyloid therapies and tau pathology

A panel of tau pathology biomarkers is advancing in the clinic and allows for the evaluation of anti‐Aβ therapies on tau pathology. Biomarkers such as p‐tau217, p‐tau205, and MTBR‐tau243 associate differently with amyloid PET versus tau PET, with p‐tau217 having a stronger relationship with amyloid PET and MTBR‐tau243 having a stronger relationship with tau PET. ⁶¹ , ⁶⁵ These relationships are at least in part driven by the time at which these biomarkers become abnormal in the disease continuum. ²² , ⁶⁶

The effects of lecanemab on tau biomarkers could give us insight into how lowering amyloid pathology, such as toxic oligomers, protofibrils, and plaques, can impact the tau pathology cascade. In the Clarity AD Phase 3 study, ⁶⁷ lecanemab reduced amyloid‐associated p‐tau181 in CSF by approximately 20% after 18 months of treatment. ⁶⁷ This observation translated to plasma, with reductions observed in both p‐tau181 and p‐tau217. ⁶⁸ Reductions in plasma p‐tau concentrations differ from the increase in concentrations observed in placebo patients over time, and the difference from placebo reached significance as early as 6 months after treatment. ⁶⁷ , ⁶⁸ Lecanemab treatment slowed the accumulation of CSF MTBR‐tau243 and tau PET. ⁶⁹ The pharmacodynamic response on these tangle‐associated biomarkers occurred later than the amyloid‐associated biomarkers, with separation from the placebo trajectory being observed after 12 months of treatment. ⁶⁹

The level of tau pathology at baseline could influence the treatment response, as it has been shown that tau PET is prognostic of clinical progression. In the Clarity AD tau PET substudy, 342 patients were assigned to a no/low (SUVR <1.06, n = 141, 41%), intermediate (SUVR = 1.06–2.91, n = 191, 56%), or high tau group (SUVR >2.91, n = 10, 2.9%) based upon their baseline MK‐6240 SUVR image. Whole cortical gray matter SUVR cutoffs were derived from the analysis of amyloid PET‐positive individuals with a mean MMSE (SD) of 23.4 (3.5) from the Lantheus/Cerveau database. When tau was present in the no/low tau subgroup, it was predominantly restricted to the entorhinal cortex (Braak I). The signal was also detected in the hippocampus (Braak II) and early limbic regions (Braak III), but there was minimal signal in the later Braak regions (IV‐VI). Lecanemab was found to slow tau PET accumulation regionally as predicted by the tau pathology location at baseline, with the no/low tau subgroup having a nominal effect in the medial temporal lobe. Lecanemab has an impact on tau PET more broadly in the intermediate and high tau subgroups. ⁷⁰

As tau PET was only available in a subset of the larger ClarityAD population, amyloid PET was used to estimate tau levels at baseline in order to determine if the no/low tau subgroup had a differential response on clinical outcomes in the larger Clarity AD population. The no/low tau cutoff translated to an amyloid PET cutoff <60 Centiloids. In the overall population, lecanemab (n = 714) slowed CDR‐SB by 27% vs. placebo (n = 757) at 18 months. In the no/low tau/amyloid <60 Centiloids subgroup, lecanemab (n = 146) had 51% less decline than placebo (n = 161) at 18 months. Results suggest that in no/low tau individuals when tau pathology is still contained within the early Braak regions, including the entorhinal cortex, anti‐amyloid therapies may have a greater effect on slowing cognitive decline. ⁷¹ Once spreading begins, additional therapy, such as an anti‐tau, may be needed to stop tau pathology and clinical progression.

Taken together, tau biomarkers can be used to demonstrate the contribution of each therapeutic when delivered in combination. The growing repertoire of available fluid and imaging biomarkers could enable more rapid advancements in tau therapeutics as monotherapies or combination therapies.

4.3. Advanced delivery methods for AD therapies

The blood–brain barrier poses a significant challenge to the delivery of therapeutics to the brain. Systemically administered large molecules such as antibodies have very limited brain access (<0.1%). ⁷² , ⁷³ Large polar molecules such as ASOs have virtually no brain penetrance and require delivery through invasive intrathecal administration, which places a major burden on patients. ⁷⁴ , ⁷⁵ Among various brain delivery technologies, transferrin receptor‐mediated transcytosis has been the most widely investigated, noninvasive mechanism. ⁷³ , ⁷⁶ Roche's proprietary Brainshuttle technology platform harnesses this mechanism to deliver therapeutic cargos like monoclonal antibodies to the central nervous system. ⁷⁷ The transport of the therapeutic payload occurs at the level of the brain capillaries; it is active and rapid and ensures not only a substantially increased exposure and target engagement but also a more widespread and homogeneous biodistribution of the therapeutic agent in the brain. ⁷⁷

In the case of the Brainshuttle anti‐amyloid antibody trontinemab, this translates into rapid and robust amyloid plaque clearance at dose levels at which standard monoclonal antibodies usually have little effect on amyloid plaque pathology. ⁷⁸ It is hypothesized that trontinemab's unique and more direct access to the brain via the capillaries (with relative bypassing of vascular amyloid deposits), as well as the ability to use lower systemic doses, may be associated with additional safety benefits, that is, a reduced incidence of amyloid‐related imaging abnormalities (ARIA).

The Brainshuttle technology platform represents a highly versatile platform that may, in principle, be expanded to a wide range of targets and payloads beyond monoclonal antibodies, including ASOs. Brainshuttle ASOs represent a novel and highly promising modality that may enable druggability of a wide range of central nervous system targets, including tau, which could, for example, be targeted with intravenously delivered Brainshuttle ASO conjugates to decrease expression of the tau protein in the brain. ⁷⁹

5. Tau and Combination Therapies

Although a large body of data suggests that amyloid initiates tau pathology in AD, an overwhelming body of literature suggests that tau pathology drives neurodegeneration and is more directly linked to cognitive decline. ⁸⁰ Focusing on a single pathology thus leaves significant disease‐driving mechanisms unaddressed. A more effective and comprehensive approach to AD treatment will likely require targeting of amyloid and tau pathways, similar to the multi‐pronged combination approach successful in other therapeutic areas such as human immunodeficiency virus (HIV) and oncology.

Several lines of evidence support a dual‐treatment approach. First, recent clinical studies with anti‐Aβ therapies demonstrated that clearing amyloid plaque can reduce disease progression but does not stop or reverse the AD course, suggesting that amyloid clearance alone may not be sufficient to address the neurotoxicity that occurs in AD. ⁶⁷ In addition, since a leading model of AD progression suggests that amyloid promotes the spread and pathological effects of tau, simply targeting tau without addressing amyloid may allow the amyloid‐driven cascade to persist. ⁸¹ Finally, genetic evidence shows that mutations leading to amyloid accumulation result in early AD onset. ⁸² Once tau pathology begins to propagate, however, it can drive disease progression independently of amyloid.

Comparing combination therapies directly against an approved amyloid‐targeted therapy can provide actionable insights while reducing the burden on patients. Importantly, several combination treatment trials are currently ongoing, including two studies of E2814, an anti‐tau therapy, dosed with lecanemab, an anti‐amyloid treatment, in individuals with dominantly inherited AD (DIAN‐TU NextGen study; NCT05269394) and early sporadic AD (Eisai Study 202; NCT06602258). Moving forward, a combination therapy approach that includes both anti‐amyloid and anti‐tau agents, possibly with other interventions addressing neuroinflammation or neuroprotection, may slow disease progression more effectively. However, challenges remain in balancing feasibility, accessibility, and efficacy.

It should be noted that while combination therapy offers promise, monotherapy trials remain essential to fully evaluate tau‐targeting drugs, particularly given the stronger correlation between tau pathology and AD symptoms. Placebo‐controlled trials remain ethical and necessary to establish the efficacy and safety of tau therapies before they are incorporated into combination strategies. Large‐scale combination trials could limit the independent evaluation of tau‐targeting drugs, potentially restricting their use for patients ineligible for amyloid‐targeting therapies. The logistics of combination trials also introduce complexities. These studies require larger sample sizes, increased costs, and longer durations, making recruitment and retention challenging. ⁸³ Ultimately, while combination therapy holds promise for AD, a strategic approach is necessary to ensure both amyloid and tau‐targeting therapies are adequately studied, accessible, and optimized for diverse patient populations.

5.1. Strategic design of combination therapy trials in AD

Designing clinical trials for combination therapies requires careful design considerations to ensure internal validity, feasibility, and meaningful conclusions. Clinical trial designs for combination therapies generally fall into two categories: fixed designs, such as parallel‐arm and factorial trials, which evaluate a single drug under a structured protocol, and platform trials, which allow multiple therapies to be tested simultaneously under a single master protocol with the flexibility to add, modify, or remove treatments. Trials can be further categorized into simultaneous, sequential, or factorial designs, each introducing unique complexities in execution and analysis.

Different platform trial designs have distinct advantages and challenges. A simultaneous platform trial in AD could involve a control arm receiving an approved anti‐amyloid drug and an intervention arm receiving both anti‐amyloid and anti‐tau therapies. This design closely aligns with current clinical practice and facilitates recruitment since all participants receive an active intervention. However, it does not allow isolation of the independent effects of anti‐tau therapy without a monotherapy arm. A sequential platform trial could start with all participants receiving an anti‐amyloid drug before randomization to either a placebo or anti‐tau therapy. The choice of randomization timing—at baseline or after initial treatment—affects data analysis and interpretation (e.g., levels of amyloid clearance may impact levels of tau). This approach also helps mitigate the safety concerns of introducing multiple interventions but, like the simultaneous design, does not allow for evaluating the specific effects of anti‐tau therapy.

On the other hand, a factorial platform trial introduces multiple arms, incorporating placebo, anti‐amyloid monotherapy, anti‐tau monotherapy, and combination therapy. In theory, this design provides the most optimal approach by allowing direct comparisons between interventions, allowing assessment of both individual drug effects and potential synergy. However, it requires larger sample sizes, longer durations, and greater resources, making it more complex and expensive. Recruitment and retention can also be challenging, as participants and physicians may be reluctant to accept randomization to a particular arm. Therefore, ethical and practical feasibility must be considered when applying all intervention levels.

The DIAN‐TU Tau NextGen trial serves as an example of a sequential platform trial evaluating E2814. Originally designed as a monotherapy study in autosomal dominant AD mutation carriers, the trial evolved as emerging data on amyloid‐lowering therapies highlighted the need for combination approaches. The trial design now evaluates E2814 both as a monotherapy and in combination with lecanemab, an anti‐amyloid therapy. Participants are stratified into symptomatic and asymptomatic groups based on their CDR score: Symptomatic participants receive lecanemab (an approved anti‐amyloid drug) for 6 months, followed by randomization to E2814 or placebo. The primary endpoint is tau spread, measured via tau PET, with additional cognitive and biomarker assessments. Asymptomatic participants receive E2814 or a placebo for 12 months, after which lecanemab is introduced. The primary endpoints include soluble biomarkers, such as CSF p‐tau217/total tau ratio.

Several challenges emerged during the design of the Tau NextGen trial. The full effect of amyloid immunotherapy on tau biomarkers was initially unclear, and recent tau monotherapy trials yielded mixed results, complicating the interpretation of tau therapy efficacy. Although amyloid immunotherapies reduce tau biomarkers, it remains uncertain whether complete amyloid clearance is necessary before introducing tau therapy. Furthermore, distinguishing between the contributions of amyloid and tau therapies to biomarker changes presents additional complexities, as some biomarkers are affected by both therapies.

Statistical considerations play a crucial role when designing combination therapy trials. The primary objective of the study must guide the design, ensuring that the study structure allows for a clear mechanistic understanding of how each therapy contributes to treatment effects. Biomarker selection also plays a critical role, yet questions remain about how well tau biomarkers can identify appropriate participants and whether heterogeneity within the selected population could affect outcomes. The inclusion of comparative arms is another crucial decision, particularly in trials evaluating combination therapies. Although placebo and monotherapy arms provide valuable insights, their ethical justification must be assessed carefully to balance scientific rigor with patient welfare.

Randomization and masking introduce further complexities, particularly in platform and factorial trials, as complex consent forms, patient and physician preferences, and challenges in maintaining blinding can impact recruitment and feasibility. Placebo allocation must also be carefully managed, particularly in trials testing multiple anti‐tau drugs, to ensure statistical balance and maintain study integrity.

Sample size considerations are equally critical. Traditional clinical trials assume a homogeneous population with consistent treatment effects across all subgroups, but emerging evidence suggests that treatment responses may vary. If effects differ significantly across subpopulations, adjustments may be necessary to prevent underpowering the study. Similarly, in combination trials, determining whether the treatment effect is additive or synergistic is essential, as even a small but meaningful combination effect could necessitate a larger sample size.

Although oncology trials are often referenced when discussing combination strategies in AD, AD presents unique challenges that limit direct comparisons. In cancer, surgical resection followed by adjuvant chemotherapy has a clear rationale, but an equivalent approach in AD is less defined. Determining whether dramatic amyloid reduction can serve as an equivalent step remains an open question.

Looking ahead, key questions remain. For example, should combination therapies be introduced in asymptomatic individuals who harbor AD pathology? Is complete amyloid removal necessary? Is our understanding of tau biomarkers sufficient to predict the outcomes for tau‐amyloid combination therapies? Advances in tau biomarkers have improved the understanding of disease progression, but their role in guiding combination therapy remains uncertain. Thoughtful trial design, strategic cohort selection, and rigorous statistical planning will be critical in determining the future of amyloid/tau combination treatments in AD.

5.2. Selection of participants for anti‐amyloid and anti‐tau combination trials

Results from recently completed anti‐amyloid monotherapy trials suggest that intervention at an earlier stage of pathology may offer more opportunity to slow the pathophysiological process of AD and confer greater clinical benefit. Although conventional wisdom previously held that tau therapy might be able to be initiated later than anti‐amyloid interventions, recent observational studies highlight the importance of early tau accumulation in medial temporal lobe regions in setting the stage for “ca‐tau‐strophe”, the rapid rise in neocortical tau and cognitive decline. It is challenging to identify participants with sufficient levels of amyloid and tau who are likely to progress during relatively short trials but prior to the stage of tauopathy that may be too late to halt progression.

Although the exact characteristics of “goldilocks” participants will depend on the specific mechanism of action of the therapeutic intervention to be tested, several guiding principles may be useful. First, paradoxically, older age is associated with slower rates of tau accumulation ⁸⁴ in the setting of elevated amyloid and evidence of cognitive impairment, so selection of participants below age 75 may be appropriate. Second, rapid neocortical tau accumulation is predicted by both baseline amyloid and baseline tau levels, particularly medial temporal lobe tau levels. Finally, selecting participants on the basis of biomarker or imaging levels of amyloid and tau, rather than the clinical stage, may be more useful in predicting the likelihood of progression and response to therapeutic intervention. The National Institute on Aging (NIA) ‐funded Alzheimer's disease Clinical Trials Consortium Alzheimer Tau Platform will test combinations of amyloid and tau therapeutics across the continuum of preclinical and prodromal AD in cohorts defined by baseline amyloid and tau biomarkers.

6. Biomarkers and Regulatory Perspectives on Drug Approval Pathways for AD

6.1. Perspective of the US FDA

At the Fall 2024 Research Roundtable meeting, representatives from the US FDA shared insights on the critical role of biomarkers in AD drug development and regulatory approval. The BEST (Biomarkers, EndpointS, and other Tools https://www.ncbi.nlm.nih.gov/books/NBK326791/) criteria categorize biomarkers by risk/susceptibility, safety, monitoring, pharmacodynamic response, diagnosis, prognosis, and treatment prediction. Biomarkers in drug development can be used for proof of concept, enrollment criteria, efficacy assessment, and safety monitoring. Scientific evidence supporting biomarkers must include biological rationale, analytical validation, clinical validation, and quality data.

Surrogate endpoints are used in clinical trials as a substitute for a direct measure of how a patient feels, functions, or survives and are categorized into validated, reasonably likely, and candidate surrogates. A validated surrogate endpoint is an endpoint supported by a clear mechanistic rationale and clinical data providing strong evidence that an effect on the surrogate endpoint predicts a clinical benefit. A surrogate endpoint can thus be used to support traditional approval without the need for additional efficacy information. A reasonably likely surrogate endpoint is an endpoint supported by clear mechanistic and/or epidemiologic rationale but insufficient clinical data to show that it is a validated surrogate endpoint. Such endpoints can be used for accelerated approval for drugs or expedited access for medical devices. A candidate surrogate endpoint is an endpoint still under evaluation for its ability to predict clinical benefit.

The traditional (Full) approval pathway of the FDA requires substantial evidence of effectiveness demonstrated on a clinically meaningful endpoint (e.g., how a patient feels, functions, or survives) or validated surrogate. On the other hand, accelerated approval (21 CFR 314.500‐subpart H, accelerated approval regulations) may be considered for serious or life‐threatening diseases with an unmet need. Accelerated approval requires substantial evidence of effectiveness demonstrated on an endpoint that is not itself a direct measure of the clinical benefit of interest but is instead reasonably likely to predict that clinical benefit (e.g., surrogate or intermediate clinical endpoint) based on “epidemiologic, therapeutic, pathophysiologic, or other evidence.” The FDA may require further adequate and well‐controlled clinical trials to verify and describe clinical benefits for products approved under accelerated approval in order to obtain traditional approval.

Accelerated approval can expedite drug development, particularly in situations where it is challenging to conduct a trial of reasonable size or duration to detect a clinical benefit. It is typically used in rare or slowly progressive diseases where an extended period of time is required to measure the intended clinical benefit of a drug. Accelerated approval necessitates adequate data to assess safety and typically requires clinical trials to verify clinical benefit. If a clinical benefit can be measured in a trial of reasonable size and duration, the traditional approval pathway may be the most efficient path for development and provides the greatest access.

Regulatory representatives from the FDA present at the meeting noted that combination therapy reviews at the FDA follow similar principles, with specific regulatory requirements applied only to fixed‐dose combinations, where two products are developed, packaged, and marketed together. Furthermore, they underscored the essential role of biomarkers in AD development, encouraging high‐quality assays even as exploratory endpoints due to their potential to become more informative as scientific knowledge advances.

6.2. Insights from global regulatory agencies

Discussions with regulatory representatives from the FDA, European Medicines Agency (EMA), Health Canada, and Japan's Pharmaceuticals and Medical Devices Agency (PMDA) addressed key lessons from recent amyloid targeting drug approvals and their implications for anti‐tau therapy approvals. The EMA emphasized the necessity of demonstrating a positive benefit‐risk balance, highlighting the need for well‐defined clinical endpoints, comprehensive long‐term safety and efficacy data, and clear guidelines for dosing and discontinuation in both short‐ and long‐term settings. Health Canada underscored the importance of defining patient populations, robust statistical analysis plans, and demonstrating clinically meaningful changes while noting that post hoc analyses are often insufficient for approval. PMDA Japan indicated that the review of anti‐tau therapies would mirror that of anti‐amyloid therapies, focusing on disease progression suppression and biomarker evaluation. The agency also highlighted the importance of looking into the correlation between biomarker reduction and slowing of disease progression.

Regarding approval pathways for combination therapies, PMDA emphasized the need for clinical trials demonstrating superiority over monotherapy and managing potential adverse events. Health Canada highlighted that while polypharmacy is not regulated by them, data may be requested to ensure the safety of combined therapies, particularly regarding drug‐drug interactions. Additionally, new combination formulations require full regulatory review, and if combination use is studied exclusively in clinical trials, approval may be limited to add‐on therapy rather than standalone use. The EMA highlighted the complexity of approving combination therapy for AD, noting that current guidelines in EMA lack specific instructions. They stressed that trial design, especially Phase 3, would influence approval scope.

The EMA further emphasized the need for biomarker validation specific to the mechanism of action, noting that data from amyloid‐targeting therapies may not apply to tau or combination therapies. Health Canada highlighted that biomarkers can serve as promising evidence under its Notice of Compliance with Condition (NOCC) pathway. The NOCC policy offers an accelerated review pathway with a 200‐day time frame for promising new drugs targeting serious, life‐threatening, or severely debilitating conditions lacking marketed treatments in Canada. The NOCC requires confirmatory trials to validate the benefits initially demonstrated by promising evidence, which may include surrogate or clinical endpoints reasonably likely to predict clinical benefit.

7. Summary/Conclusion

The research and discussions highlighted during the Fall 2024 AARR underscore the increasing importance of tau as both a biomarker and a therapeutic target in AD. Advances in blood‐based and imaging tau biomarkers have facilitated earlier and more precise diagnosis and have provided critical insights into disease staging and progression. Meanwhile, new humanized knock‐in mouse models offer more physiologically relevant platforms for dissecting the mechanisms of amyloidosis and tauopathy. In parallel, emerging evidence on oligomeric tau highlights rare yet highly pathogenic species that could have more detrimental effects on neurons than insoluble fibrillar tau aggregates.

Developing effective tau‐directed therapies remains challenging due to the complexities around the tau protein and the difficulty of bypassing the blood–brain barrier. An individual affected by tauopathy most certainly harbors a constellation of tau species that fall within the spectrum between microtubule‐bound monomer and insoluble fibrillar aggregate. Although evidence suggests that multimeric soluble tau protein is particularly harmful, the relative contributions of specific species of post‐translational modified monomer and insoluble fibril remain unclear, with some studies suggesting that fibrillar tau may be benign or even protective. This uncertainty poses a challenge for therapeutic development, as small molecules and antibodies may target a specific tau conformation but miss other species that contribute to neurotoxicity. ASO‐based approaches resolve this limitation by reducing overall levels of tau protein, but may risk negative consequences due to loss of normal tau function. Addressing this gap is essential for the design of interventions with the highest likelihood of clinical benefit. Precision medicine tools, more tailored clinical designs, and a deeper understanding of which tau species and delivery routes are most relevant are nevertheless poised to accelerate the field. By matching therapies to the specific stage of tau pathology and measuring outcomes with increasingly sensitive biomarkers, researchers may finally overcome the hurdles that have long impeded effective tau‐targeted treatments for AD. Several novel tau therapeutic strategies are advancing the field, such as monoclonal antibodies, nucleic acid‐based therapies, and gene therapy approaches that target distinct forms or phases of tau pathology. In parallel, innovations in drug delivery platforms, such as transferrin receptor‐mediated transcytosis, offer the prospect of overcoming the challenges of crossing the blood–brain barrier to enhance therapeutic efficacy.

Discussions on combination therapies, particularly those pairing anti‐amyloid and anti‐tau agents, reflect a new era of multimodal intervention in AD. Robust biomarker‐based staging has become central to optimizing trial design, identifying the most appropriate patient populations, and evaluating treatment efficacy. Nonetheless, multiple challenges persist, including the complexity of determining each therapy's independent contribution, the logistical and statistical hurdles of large‐scale combination trials (e.g., larger sample sizes, higher costs, and extended timelines), and the considerations around integrating biomarkers as surrogate endpoints into regulatory pathways. Addressing these gaps will require continued investment in standardized assays, large‐scale collaborative studies, and novel trial frameworks that can efficiently test multiple therapeutic mechanisms.

Taken together, interdisciplinary collaboration, rigorous validation of tau biomarkers, and strategic trial designs that consider both monotherapy and combination approaches are essential. As the field moves forward, the synergy of innovative biomarker technologies, targeted pharmacological interventions, and supportive regulatory environments has the potential to accelerate meaningful progress. Lessons from early setbacks in amyloid‐targeting therapies have demonstrated the importance of refining trial designs, selecting the right patients, and aligning interventions with the disease stage, an approach that ultimately led to meaningful breakthroughs. We are now in a similar phase with tau therapies, where early challenges have sharpened our understanding of the right mechanisms of action, patient populations, and timing of intervention. Encouraging early signs of success suggests that the field is on the point of translating these insights into transformative therapies. By applying these lessons, researchers and clinicians are better poised than ever to modify the trajectory of AD and improve the lives of individuals affected by this devastating disorder.

CONFLICT OF INTEREST STATEMENT

Bess Frost has received grants or contracts from the NIA (R01AG057896, R01AG078964), MD Anderson Belfer Neurodegeneration Consortium, and Transposon Therapeutics, consulting fees from MD Anderson Belfer Neurodegeneration Consortium, support for attending the 2024 Fall AARR, has served as Guest Editor, Alzheimer's & Dementia, Board Member, Alzheimer's Association RI Chapter, CurePSP Scientific Advisory Board member, American Federation for Aging Research Scientific Advisory Council member, Belfer Neurodegeneration Consortium Member and Funded Investigator, Tau Consortium Member and Funded Investigator.

Hartmuth Kolb is a current full‐time employee of Enigma Biomedical Group and stockholder of Enigma and former employer Johson and Johnson, has patents planned used or pending for US20190271710A1 and WO2009102498A1.

Alicia Algeciras‐Schimnich has received payment or honoraria for presentations from Roche Diagnostics and Eli Lilly, participated on a Data Safety Monitoring Board for Fujirebio Diagnostics and Roche Diagnostics, is a full‐time employee of Mayo Clinic in Rochester, MN.

Tobey J. Betthauser has received NIH/NIA grants (R01AG080766 paid to institution), honorariums from NIH and Intermountain Healthcare, travel support from University College London, Alzheimer's Association, and NIH.

Sarah DeVos is a full time employee of Curie.Bio.

Fiona Elwood is a full time employee of Johnson & Johnson.

Adam S. Fleisher is a stockholder and full time employee of Eli Lilly and Co.

David Henley is a full‐time employee and stockholder of Johnson & Johnson.

Kanta Horie has received grants from Washington University, royalties from C2N Diagnostics, patents planned issued or pending for Methods to Detect MTBR‐tau Isoforms and use Thereof (PCT/US2020/046224), is a stockholder and full‐time employee of Eisai Co, Ltd.

Bradley Hyman has received grants from the NIH, Abbvie, Cure Alzheimer Fund, Rainwater Foundation, JPB Foundation, consulting fees or scientific advisory board compensation from Dewpoint, Abbvie, Alexion, Arvinas, AstaZenica, Biogen, BMS, Cell Signaling, Lilly, Merck, Novartis, Pfizer, Sanofi, Takeda, TD Cowen, Vigil, Violet, Voyager, Wavebreak, support for ADPD organizing committee and attending AARR, particiapted on Data Safety Monitoring Board for Biogen, and is a stockholder for Novartis and Dewpoint.

William Charles Kreisl is a full time employee at Eisai, Inc.

Luka Kulic is full‐time employee and owns stocks of F. Hoffmann‐La Roche Ltd.

Antoine Leuzy serves as a consultant to CPAD and Enigma Biomedical Group.

Jose‐Alberto Palma is a full time employee of Eli Lilly and Co.

Sophie Parmentier‐Batteur is a full‐time employee at Novartis.

Maria‐Magdalena Patru is a full‐time employee of Roche Diagnostics.

Gil D. Rabinovici has served on scientific advisory boards and/or as a consultant for Eli Lilly, Genenetech/Roche, GE Healthcare, Alector and Merck, payments from Efficient LLC, JAMA Neurology (Associate Editor), Miller Medical Communications, paid for participation on a Data Safety Monitoring Board or Advisory Board for Johnson & Johnson;

Larisa Reyderman is a full‐time employee of Eisai Inc.

Reisa A. Sperling is employed by Brigham and Women's Hospital, received consulting fees from Abbvie, AC Immune, Acumen, Alector, Alnylam, Biohaven, Bristol‐Myers Squibb, Cytox, Genentech, Ionis, Janssen, Merck, NervGen, Neuraly, Neurocentria, Oligomerix, Prothena, Roche, Shionogi, and Vaxxinity.

Serge Van Der Geyte is a stockholder and full‐time employee of Johnson and Johnson.

Kristin R. Wildsmith is a full‐time employee of Eisai, Inc.

Rebecca M. Edelmayer, Simin Mahinrad, Maria C. Carrilloand Christopher J. Weberare full‐time employees of the Alzheimer's Association. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

Consent (i.e., all human subjects provided informed consent) was not applicable.

Supporting information

Supporting Information

Frost B, Kolb H, Algeciras‐Schimnich A, et al. Tau biology, biomarkers, and therapeutics. Alzheimer's Dement. 2025;11:e70165. 10.1002/trc2.70165