← Writing

Biomarkers of Alzheimer's Disease

Alzheimer'sbiomarkersneuroscience

Alzheimer's Disease (AD) is a neurodegenerative disease that affects over 55 million people worldwide. It leads to dementia, cognitive decline, memory loss, and impaired judgment, and there is no effective cure for it. Unfortunately, since the symptoms of AD are often conflated with those of normal aging, clinical AD diagnosis is often delayed, and only patients in later stages are diagnosed, when any mitigating treatment is too late [1]. This has motivated almost a century of research on biomarkers for Alzheimer's, which serve two main purposes. First, as a framework for early detection and prediction: the appearance of biomarkers like amyloid and tau are early indicators of the onset of Alzheimer's [2, 3], and establishing these biomarkers perhaps allows us to narrow down which patients should get confirmatory testing, or to predict which patients may experience cognitive decline so support services can be allocated accordingly. Second, if causal relationships are established between a given biomarker and AD, this could inspire cures — inhibitors or enzymes that disrupt biomarker formation and therefore slow down or stop AD pathology. This essay reviews the main classes of AD biomarkers and explores how promising they are for therapeutics.

Classes of Biomarkers

Aβ

Aβ is a molecule formed from the cleavage of Amyloid Precursor Protein (APP). There are two ways it is normally cleaved: amyloidogenic and non-amyloidogenic. The non-amyloidogenic cleavage of APP involves α-secretase, producing sAPPα, which is soluble and released into the extracellular space. This is followed by γ-secretase cleavage that forms a P3 protein, whose significance is barely known, and AICD, the intracellular component. The amyloidogenic pathway instead involves sequential cleavage by β- and γ-secretase. β-secretase produces sAPPβ, a slightly shorter variant of the non-amyloidogenic product, followed by γ-secretase cleavage that produces Aβ40 or Aβ42, the latter being more prone to aggregation. The driving factor behind this propensity is believed to be the solubility of the Aβ fragments — in the monomeric stage, Aβ42 is more prone to aggregation than Aβ40, since it is less soluble.

As the amyloid protein aggregates, it forms soluble oligomers and, eventually, fibrils and plaques. Recent literature suggests oligomers ranging from dimers to dodecamers show a stronger causal implication in AD than the amyloid plaques classically believed to be the cause. There are two types: type 1 AβOs are of high molecular weight, and type 2 AβOs are of lower molecular weight [4]. Type 2 oligomers, which are less toxic, are part of the amyloidogenic aggregation pathway contributing toward fibril and plaque formation, while type 1 oligomers stay in the same misfolded form and are thus more pathogenic. Aggregation into oligomers is a complex, multistep process. The first step, nucleation, is the collection of multiple monomers into a dimer, trimer, or tetramer species. More recent evidence, however, suggests a significant contribution from a second step, "secondary nucleation," in which oligomers are primarily formed from monomers dissociated off existing fibrils — supported by the fact that the hydrophobic surface of fibrils is an optimal catalytic site for Aβ oligomer formation and elongation. There are also differences among intermediate oligomeric species: the dimer undergoes larger conformational changes in its C-terminus than trimers and tetramers, involving more β-sheet structure shielding hydrophobic residues, and while hydrogen bonds in the β-sheets are present throughout, larger oligomers are additionally stabilized by D23–K28 salt bridges between chains, whereas the dimer shows a D23–N27 interaction instead.

As oligomers grow, they begin to form protofibrils — soluble, fibril-like structures with a higher number of β-sheets [5] — which exist in dynamic equilibrium with smaller oligomers. Oligomers are mostly spherical at first, then elongate by merging those spherical subunits into a "bead-like" appearance, forming protofibrils, which twist around each other to form insoluble fibrils. It remains unclear whether this aggregation pathway is direct or involves other proteins and molecular machinery. Nevertheless, fibrils consist of parallel and antiparallel β-sheets, which are highly conserved; the addition of just two residues going from Aβ40 to Aβ42 increases toxicity significantly, with the latter forming fibrils with maximally buried hydrophobic side chains [6].

Eventually, fibrils gather to form plaques, which include not just Aβ deposits but other cellular components as well. The first common type is the primitive plaque, in which amyloid fibrils weave together with dystrophic neurites — axons without a myelin sheath, swollen in size — along with gatherings of organelles and cellular structures such as vesicles, tubules, and mitochondria [7]. Primitive plaques are believed to mature into "cored plaques," which contain an amyloid-dense core in addition to the coalescence of other material. Plaque formation is not unique to fibrillar aggregation, either: there are correlations between amyloid plaques and cerebral amyloid angiopathy, the deposition of plaques near blood vessels.

T-tau and P-tau

Tau is a microtubule-associated protein important in neurons for functions such as microtubule binding and stabilization. It is highly soluble and unfolded in solution under normal physiological conditions. The abnormal accumulation of tau aggregates in neurons, however, is a strong biomarker for Alzheimer's [8], and can occur for a range of reasons. First, through post-translational modifications (PTMs): in AD brains, regulation of tau modification breaks down, leading to hyperphosphorylation — AD-brain tau carries three to four times more phosphate than healthy adult tau, reducing its affinity for microtubules among other mechanistic faults. Tau also undergoes acetylation and caspase truncation. Second, through cellular stress factors like oxidative stress and neuroinflammation: reactive oxygen species (ROS) damage tau through carbonylation and activate stress kinases that also drive hyperphosphorylation, while neuroinflammation causes cytokine release that activates similar kinases. This is coupled with simultaneous malfunction of the clearance and chaperone pathways. The chaperone network, involving Hsp70 and Hsp90, binds tau to promote its degradation when misfolded and to stabilize pathological tau conformations. Tau is cleared by two main systems: the ubiquitin–proteasome system, which specifically targets monomeric, modified, still-soluble tau, and the autophagy–lysosome system, which engulfs aggregated oligomeric tau. Aging, excessive PTMs, and tau aggregation all inhibit these processes — blocking ubiquitination and inhibiting proteasomes — resulting in the accumulation of soluble oligomers and insoluble filaments. Both total tau (t-tau) and phosphorylated tau (p-tau) are therefore biomarkers in AD, since p-tau is a key PTM with exaggerated impact while other tau modifications, encapsulated in t-tau, also contribute [8].

Tau aggregates in a manner very similar to Aβ, where modifications to soluble monomers cause the formation of oligomers (increasingly recognized as the most toxic species) and, eventually, filaments and tangles. Specifically, tau forms neurofibrillary tangles that include paired helical fragments and straight filaments, where the repeat region in tau segments forms a cross-β filament core. Once filaments form, propagation is almost prion-like: misfolded tau can be released and taken up by connected cells, recruiting endogenous tau to misfold as well and propagating the pathology.

NVIS

Beyond Aβ and tau, a recent revision of the AT(N) framework now includes neurodegeneration, vascular, inflammation, and synaptic dysfunction (NVIS) markers, acknowledging that AD involves diverse pathological processes beyond the amyloid–tau duopoly [2, 9]. These are used less as standalone diagnostics, however, and more to complement the two core biomarkers.

CSF Biomarkers

There are, broadly, three main modalities through which AD biomarkers are accessed. The first is cerebrospinal fluid (CSF) — the classical biomarker set first discovered in patients. CSF from AD patients shows significantly more Aβ42, t-tau, and p-tau [10]. Accessing CSF is invasive, often requiring a lumbar puncture, but it yields highly accurate measurements of brain-derived proteins. Aβ42 begins decreasing in the CSF a decade before clinical symptoms relative to Aβ40, so the Aβ42/Aβ40 ratio is lower in CSF [11]. Increased t-tau in the CSF serves as a general neurodegeneration marker, since it is also elevated in stroke, traumatic brain injury, and other neurodegenerative conditions, while p-tau in the CSF is a much more specific biomarker for AD [10]. Patients positive for both Aβ42/Aβ40 and p-tau are overwhelmingly more likely to develop AD — roughly 80–90% convert to AD dementia within five years [10].

PET Imaging

PET imaging visualizes pathology directly in living brain tissue and is therefore more anatomically specific than fluid biomarkers. Tracers bind to Aβ plaques or to paired helical filaments in tau neurofibrillary tangles, detecting or ruling out amyloid or tau pathology [12, 13]. Given the strong correlation these biomarkers have with Alzheimer's, this is useful clinically: Dubois et al. found that amyloid PET predicts AD with 92% sensitivity and 100% specificity [14].

BBMs

Blood-based biomarkers (BBMs) are among the most transformative recent advances in AD diagnosis [15, 16]. Previously, questions about whether brain biomarkers could pass the blood–brain barrier made BBMs seem unlikely to be effective, but recent studies have shown promise. Since blood draws are much less invasive and more scalable than the other biomarker modalities, BBMs are a growing focus in AD screening [15]. The most important class so far is plasma p-tau species, specifically p-tau181, which distinguishes AD from non-AD patients with an AUC greater than 0.85 [17]. The FDA has also recently approved a test for p-tau217 [18, 19]. There are additionally blood measurements for Glial Fibrillary Acidic Protein (GFAP), which reflects inflammation, and Neurofilament Light Chain (NfL), which reflects neuron damage [15].

How Accurate and How Early?

Many studies have examined the accuracy of biomarker tests for AD. A meta-analysis by Therriault et al. found plasma p-tau to be the best indicator, with p-tau217 achieving 91.1% AUROC and p-tau181 achieving 81.5% AUROC [17]. Jiao et al. found p-tau217 extremely powerful at diagnosing Alzheimer's from healthy controls (AUC 0.95), but less powerful at differentiating Alzheimer's from other neurodegenerative diseases [20]. Older studies show that core CSF biomarkers also differentiate AD from control patients strongly [10]. The overall limitation of biomarker screening, however, is that while it is generally good at ruling out amyloid pathology — false negatives are rare — it is less good at confirming true positives [21], because many Alzheimer's biomarkers are not unique to the condition. Amyloid plaques, for instance, are found in healthy brains as well [22], and neuroinflammation and synaptic dysfunction are present in other neurodegenerative diseases too. Different studies reach different conclusions about which biomarker is best for diagnosis, so its accuracy is still not fully settled.

In terms of timing, AD is split into multiple stages. Preclinical AD can be detected through biomarkers indicating amyloid or tau pathology, but this stage can last decades, with only some patients progressing to prodromal AD [3, 11], where biomarkers become more severe and patients experience mild cognitive impairment before eventually progressing to dementia. Current evidence therefore suggests biomarkers are stronger at detecting pathology than at predicting an individual's exact timeline [14], since other factors — genome and lifestyle — also affect the speed of onset in ways an instantaneous biomarker snapshot cannot capture. Nevertheless, biomarkers remain a good indicator of likelihood [23].

Limitations and Benefits

There are three layers of limitation for biomarker-based diagnosis. First, the supporting studies are often flawed or biased in various ways. Cutoffs for any biomarker are generally somewhat arbitrary — how much amyloid constitutes AD pathology? — and many studies report this as a source of bias, since a fixed, predetermined cutoff may not suit every patient demographic [17]. Additionally, people who volunteer for AD biomarker studies are often already experiencing symptoms, inflating the apparent prevalence of biomarkers regardless of whether AD is the cause, and study populations rarely reflect true population demographics [21].

Second, standardization of biomarker testing is incomplete. Different labs use different analytical approaches — for instance, blood measurements of p-tau217 are performed with different antibodies depending on the manufacturer [8] — and CSF and PET testing each have their own issues with partial sensitivity (binding to off-target compounds) and cost [1, 8].

Finally, as noted, AD pathology — the mere presence of a biomarker — indicates likelihood rather than certainty [14], since too many other factors influence individual disease onset.

Nevertheless, biomarkers have real therapeutic potential for detecting Alzheimer's. Advances in BBMs are particularly promising for confirmatory testing, since they are far less invasive and expensive than gold-standard PET and CSF analyses [18, 19]. Achieving earlier diagnosis remains challenging; biomarker-based prediction currently seems to have similar efficacy to gene-based early diagnosis for early-onset AD, for example, with the best-case payoff being disease-modifying treatment and lifestyle change in time to matter.

Biomarkers have also been a focus for prevention, not just detection. Unfortunately, most drugs attempting to disrupt the amyloidogenic pathway — by inhibiting BACE1 (β-secretase), for example — have failed to actually prevent cognitive decline. Recent progress in anti-amyloid immunotherapies, however, such as monoclonal antibodies that bind Aβ oligomers and promote their clearance, has produced FDA-approved drugs (aducanumab, lecanemab, donanemab) [1].

Conclusion

Alzheimer's Disease is defined by pathologies like Aβ and tau that offer promising directions for early detection and treatment. This comes with significant limitations, however, in distinguishing AD from other neurological conditions with similar biomarkers that differ mainly by intensity. Ultimately, biomarkers indicate likelihood rather than certainty, because pathology overlaps with normal aging and other neurodegenerative conditions [22], and because genetic and lifestyle factors modulate progression as well [14]. Alzheimer's is far too complicated for any single molecular indicator to define it. Still, biomarkers remain useful as probabilistic tools that can shift AD care toward earlier intervention and more effective treatment [21, 23].

References

  1. Scheltens, P., Blennow, K., Breteler, M. M. B., de Strooper, B., Frisoni, G. B., Salloway, S., & Van der Flier, W. M. (2016). Alzheimer's disease. The Lancet, 388(10043), 505–517.
  2. Jack, C. R. Jr., Bennett, D. A., Blennow, K., Carrillo, M. C., Dunn, B., Haeberlein, S. B., Holtzman, D. M., Jagust, W., Jessen, F., Karlawish, J., et al. (2018). NIA-AA research framework: toward a biological definition of Alzheimer's disease. Alzheimer's & Dementia, 14(4), 535–562.
  3. Jack, C. R. Jr., Knopman, D. S., Jagust, W. J., Shaw, L. M., Aisen, P. S., Weiner, M. W., Petersen, R. C., & Trojanowski, J. Q. (2010). Hypothetical model of dynamic biomarkers of the Alzheimer's pathological cascade. Lancet Neurology, 9(1), 119–128.
  4. Lacor, P. N., Buniel, M. C., Chang, L., Fernandez, S. J., Gong, Y., Viola, K. L., Lambert, M. P., Velasco, P. T., Bigio, E. H., Finch, C. E., Krafft, G. A., & Klein, W. L. (2004). Synaptic targeting by Alzheimer's-related amyloid β oligomers. Journal of Neuroscience, 24(45), 10191–10200.
  5. Siddiqi, M. K., Majid, N., Malik, S., Alam, P., & Khan, R. H. (2019). Amyloid oligomers, protofibrils and fibrils. In Macromolecular Protein Complexes II (Subcellular Biochemistry, Vol. 93, pp. 471–503). Springer.
  6. Horn, A. H. C., & Sticht, H. (2010). Amyloid-beta42 oligomer structures from fibrils: a systematic molecular dynamics study. Journal of Physical Chemistry B, 114(6), 2219–2226.
  7. Kidd, M. (1964). Alzheimer's disease: an electron microscopical study. Brain, 87, 307–320.
  8. Leuzy, A., Mattsson-Carlgren, N., Palmqvist, S., Janelidze, S., Dage, J. L., & Hansson, O. (2022). Tau biomarkers in Alzheimer's disease: towards implementation in clinical practice and trials. Lancet Neurology, 21(8), 726–740.
  9. Jack, C. R. Jr., Bennett, D. A., Blennow, K., Carrillo, M. C., Feldman, H. H., Frisoni, G. B., Hampel, H., Jagust, W. J., Johnson, K. A., Knopman, D. S., et al. (2016). A/T/N: an unbiased descriptive classification scheme for Alzheimer disease biomarkers. Neurology, 87(5), 539–547.
  10. Olsson, B., Lautner, R., Andreasson, U., Ohrfelt, A., Portelius, E., Bjerke, M., Hölttä, M., Rosén, C., Olsson, C., Strobel, G., et al. (2016). CSF and blood biomarkers for the diagnosis of Alzheimer's disease: a systematic review and meta-analysis. Lancet Neurology, 15(7), 673–684.
  11. NEJM Long-Horizon Trajectories Study. (2024). Biomarker changes during 20 years preceding Alzheimer's disease. New England Journal of Medicine.
  12. Chételat, G., Arbizu, J., Barthel, H., Garibotto, V., Law, I., Morbelli, S., van de Giessen, E., Agosta, F., Barkhof, F., Brooks, D. J., et al. (2020). Amyloid-PET and 18F-FDG-PET in the diagnostic investigation of Alzheimer's disease and other dementias. Lancet Neurology, 19(11), 951–962.
  13. JNM Amyloid PET Review. (2022). The role of amyloid PET in imaging neurodegenerative disorders: a review. Journal of Nuclear Medicine.
  14. Dubois, B., Villain, N., Schneider, L., Fox, N., Campbell, N., Galasko, D., et al. (2024). Alzheimer disease as a clinical-biological construct — an international working group recommendation. JAMA Neurology.
  15. Hansson, O. (2021). Blood-based biomarkers for Alzheimer's disease: towards clinical implementation. Lancet Neurology, 20(9), 739–752.
  16. Nature Aging BBM Overview. (2023). Blood biomarkers for Alzheimer's disease in clinical practice and trials. Nature Aging.
  17. Therriault, J., et al. (2025). Blood phosphorylated tau for the diagnosis of Alzheimer's disease: a systematic review and meta-analysis. Lancet Neurology.
  18. Palmqvist, S., et al. (2024). Blood biomarkers to detect Alzheimer disease in primary care and secondary care. JAMA.
  19. Nature Medicine Implementation Study. (2025). Plasma phospho-tau217 for Alzheimer's disease diagnosis in primary and secondary care using a fully automated platform. Nature Medicine.
  20. Jiao, B., Ouyang, Z., Liu, Y., Zhang, C., Xu, T., Yang, Q., Zhang, S., Zhu, Y., Wan, M., Xiao, X., Liu, X., Zhou, Y., Liao, X., Zhang, W., Luo, S., Tang, B., & Shen, L. (2025). Evaluating the diagnostic performance of six plasma biomarkers for Alzheimer's disease and other neurodegenerative dementias in a large Chinese cohort. Alzheimer's Research & Therapy, 17(1), 71.
  21. Nature Aging Pretest Probability Study. (2024). Diagnosis of Alzheimer's disease using plasma biomarkers adjusted to clinical probability. Nature Aging.
  22. Jansen, W. J., Janssen, O., Tijms, B. M., Vos, S. J. B., Ossenkoppele, R., Visser, P. J., et al. (2022). Prevalence estimates of amyloid abnormality across the Alzheimer disease clinical spectrum. JAMA Neurology, 79(3), 228–243.
  23. JAMA Neurology Preclinical Prediction Study. (2023). Prediction of longitudinal cognitive decline in preclinical Alzheimer disease using plasma biomarkers. JAMA Neurology.