Lewy Body Dementia: Building a Blood Biomarker Strategy Without Serial CSF
A sponsor opening a phase 2 trial in Lewy body dementia faces a problem that has nothing to do with assay sensitivity. Clinical criteria will admit patients whose dominant pathology is Alzheimer’s, or limbic TDP-43, or all three at once. A cerebrospinal fluid seeding assay confirms that misfolded α-synuclein is present, then stops. And in a cohort defined by fluctuating cognition, dense longitudinal CSF sampling is impractical to schedule and harder to consent. The measurement problem here is compositional, not analytical.
What blood biomarkers work for Lewy body dementia trials?
No single analyte is sufficient. An LBD blood panel needs three functional components: an α-synuclein measure whose signal is confirmed to originate in brain tissue, a phosphorylated tau measure to identify Alzheimer’s co-pathology at screening, and a neurodegeneration marker for longitudinal tracking. Brain specificity comes from how the plasma is prepared, not from which analyte is selected.
Key Takeaways
- Clinical diagnosis is the weakest link. Pooled sensitivity of clinical DLB criteria against autopsy is roughly 60% [2] — meaning a criteria-only enrollment strategy admits a substantial fraction of participants without the pathology the drug targets.
- Co-pathology is the rule. Alzheimer’s and limbic TDP-43 changes coexist with Lewy pathology often enough that “pure LBD” is a minority phenotype in autopsy series [1][15].
- A seeding assay answers one question, once. CSF α-synuclein seed amplification is a presence/absence readout, has low sensitivity for focal or low-burden Lewy pathology [7], and shows accuracy that varies measurably between laboratories [6].
- Plasma phosphorylated tau already changed one LBD program. Stratifying a phase 2a DLB cohort by pre-treatment plasma p-tau181 raised effect sizes to 0.7 or higher in the co-pathology-free stratum, and the follow-on phase 2b excluded elevated p-tau181 at screening [10][13].
- Total plasma α-synuclein is not a CNS measurement. Circulating α-synuclein is dominated by red blood cell content [9], so a mixed-origin plasma value cannot support a target engagement or pharmacodynamic claim.
The LBD enrollment problem is misclassification, not detection
Lewy body dementia is the umbrella term for dementia with Lewy bodies (DLB) and Parkinson’s disease dementia (PDD), separated by the one-year rule: in DLB, cognitive impairment precedes or arrives within a year of parkinsonism; in PDD, it follows by more than a year [1]. That distinction is temporal, not biological, and it is the first place an LBD trial inherits ambiguity from its own diagnostic framework.
Clinical criteria miss roughly 40% of cases
A meta-analysis of 22 clinicopathological studies covering 1,585 patients put pooled sensitivity of clinical DLB criteria at 60.2% and specificity at 93.8% [2]. The 2017 consensus criteria improved the framework by promoting dopamine transporter imaging to an indicative biomarker and formalizing REM sleep behaviour disorder as a core feature [3], but no revision to a symptom-based framework can resolve what is fundamentally a molecular question.
For a sponsor, high specificity and mediocre sensitivity has a specific consequence. Screening on criteria alone does not fill the cohort with the wrong disease — it fills it with a distribution of underlying pathologies, only some of which the investigational agent addresses. Patient stratification failure of this kind is not visible in baseline demographics. It surfaces at the primary endpoint, as diluted effect size.
Co-pathology is the rule, not the exception
Lewy pathology rarely travels alone. Alzheimer’s neuropathologic change coexists with Lewy body disease frequently enough that the mixed phenotype outnumbers the pure one in most autopsy series, and mixed cases decline faster than either pure group [1]. Limbic-predominant age-related TDP-43 encephalopathy (LATE) adds a third layer: in an analysis of four autopsy-confirmed cohorts including 313 Lewy body disease cases, the presence and extent of LATE changes associated with cognitive impairment independently of both Lewy body subtype and Alzheimer’s neuropathologic change [15].
Antemortem evidence points the same direction. CSF α-synuclein seed amplification is positive in about 30% of patients with a clinical Alzheimer’s diagnosis, against 9% of controls, and positivity tracks with faster cognitive decline [5]. The traffic runs both ways: Alzheimer’s cohorts contain synucleinopathy, and Lewy body cohorts contain Alzheimer’s pathology and TDP-43 proteinopathy.
Why serial CSF cannot carry an LBD program
CSF has been the reference matrix for CNS biomarkers because it sits next to the tissue of interest. But CSF biomarker limitations in clinical trials are structural rather than incidental, and in LBD specifically three of them constrain what the matrix can deliver across the life of a program.
A seeding assay is a switch, not a dial
Seed amplification assays detect misfolded α-synuclein by nucleating aggregation in vitro, and they perform well as diagnostic classifiers. In the largest analysis to date, across the Parkinson’s Progression Markers Initiative cohort, the assay separated Parkinson’s disease from controls with high sensitivity and specificity and identified prodromal individuals before diagnosis [4]. That is a real advance for enrollment confirmation.
It is also a categorical output. A binary positive tells a sponsor that seeding-competent α-synuclein is present. It does not quantify burden, and it does not move interpretably over sixteen weeks of dosing — which is what a pharmacodynamic endpoint requires. The assay’s blind spot is also structural: in autosomal dominant Alzheimer’s disease, where Lewy pathology tends to be amygdala- or olfactory-predominant, CSF seed amplification detected seeding activity in only about 10% of symptomatic cases, and sensitivity for focal, low-burden Lewy pathology was low [7]. Early or regionally restricted pathology is exactly the population a disease-modifying LBD trial wants to enrich for.
Accuracy varies between laboratories
A four-laboratory comparison of CSF α-synuclein seed amplification in DLB found variation in diagnostic accuracy across sites in distinguishing patients from healthy controls, and the authors framed protocol standardization as the barrier to clinical implementation [6]. For a multi-site trial, between-laboratory variance in a screening assay translates directly into inconsistent cohort composition across regions — a source of noise that no statistical adjustment recovers.
Sampling density is capped by the population
The practical constraint is the least technical and the most binding. Repeat lumbar puncture is tolerated by many patients, but in a cohort characterized by fluctuating attention, visual hallucinations and caregiver-dependent scheduling, dense serial CSF collection is difficult to consent and difficult to complete. Programs end up with sparse CSF timepoints and therefore no ability to resolve a within-patient trajectory. Venipuncture carries none of these constraints, which is why the search for a lumbar puncture alternative in neurology research is a trial-operations question before it is a scientific one.
Figure 1. What each sampling modality can and cannot support in an LBD program
| Design requirement | CSF seed amplification | Total plasma analyte | Brain-enriched plasma compartment |
|---|---|---|---|
| Confirm synucleinopathy at screening | Yes — categorical [4] | No | Yes — quantitative [8][9] |
| Detect focal or low-burden Lewy pathology | Limited [7] | No | Under evaluation |
| Identify Alzheimer’s co-pathology | Requires separate CSF panel | Yes — p-tau181/231 [11][12] | Yes |
| Repeat monthly or more often | No | Yes | Yes |
| Support a target engagement claim | No — not quantitative | No — mixed tissue origin [9] | Yes — provenance defined |
| Consistent across multi-site labs | Variable [6] | Assay-dependent | Requires pre-specified precision data |
Table 1. Illustrative synthesis of design requirements against sampling modality. Cell entries are supported by the cited literature; the composite framing is editorial.
What plasma already does — and where it stops
The blood-based CNS biomarker field has moved faster in dementia than most sponsors have updated their protocols. Two things are worth separating: what plasma measures well today, and what it cannot measure without a provenance step.
Phosphorylated tau is a working co-pathology filter
Plasma phosphorylated tau species do a specific job in an LBD cohort: they flag the Alzheimer’s co-pathology that dilutes a synuclein-targeted or cholinergic-targeted treatment effect. In the European DLB consortium cohort, plasma Aβ42/40, p-tau181 and p-tau231 associated with amyloid status, while plasma GFAP and NfL tracked cognitive symptoms — supporting p-tau as a non-invasive read on amyloid co-pathology and the glial and axonal markers as monitoring biomarkers [11]. In probable DLB, plasma p-tau181 and p-tau231 concentrations associated with subsequent cognitive decline [12].
One DLB program has already acted on this. A 16-week randomized phase 2a study of a p38α kinase inhibitor (AscenD-LB, NCT04001517) was re-analyzed post hoc after stratifying participants by pre-treatment plasma p-tau181 against a 2.2 pg/mL cut-point for Alzheimer’s pathology. Treatment effects were substantially larger in participants without elevated p-tau181, with effect sizes of 0.7 or higher on clinical and functional measures [10]. The follow-on phase 2b (RewinD-LB, NCT05869669) carried that forward as a design change, excluding patients with elevated screening p-tau181 and adopting a clinical rather than cognitive primary endpoint; trial simulations put statistical power above 95% in the co-pathology-free population at 160 participants [13].
That sequence is the clearest available demonstration that in LBD, a blood biomarker’s first job is cohort composition. The drug did not change between phase 2a and 2b. The population definition did.
Figure 2. Effect size by pre-treatment plasma p-tau181 stratum
| Stratum | Interpretation | Reported treatment effect size vs. placebo |
|---|---|---|
| Overall phase 2a population | Unstratified, mixed co-pathology | Positive but smaller than the stratified subgroup |
| Plasma p-tau181 not elevated | No evidence of Alzheimer’s co-pathology | ≥0.7 on clinical and functional measures |
| Plasma p-tau181 elevated | Alzheimer’s co-pathology present | Lower effect size |
Table 2. Data from post hoc stratified analysis of the AscenD-LB phase 2a trial [10], with design consequences reported in the phase 2b rationale [13]. Suggested rendering: horizontal bar chart, three bars, x-axis “Cohen’s d vs. placebo” from 0 to 1.0, with the stratified bar visually emphasized.
Where total plasma measurement fails
Extend the same logic to α-synuclein (alpha-synuclein) and it breaks. Circulating α-synuclein concentration is strongly influenced by red blood cells, which account for the overwhelming majority of the protein in blood [9]. A total plasma α-synuclein assay is therefore mostly a haematological measurement with a small CNS contribution buried inside it. Pre-analytical haemolysis moves the number. Brain pathology, in a mixed-origin sample, barely does.
This is the structural difference between the tau and synuclein cases. Plasma p-tau181 works as an Alzheimer’s co-pathology filter because the peripheral background is low enough for the CNS signal to dominate. Plasma α-synuclein does not work the same way, and no improvement in analytical sensitivity fixes it — a more sensitive assay on a mixed-origin sample measures the confound more precisely. Plasma p-tau specificity and plasma α-synuclein specificity are different problems with different solutions.
Provenance: making a plasma measurement brain-specific
The workaround is to change what is in the tube before measuring it. Neuron-derived extracellular vesicles (NDEVs) are membrane-bound particles released by neurons that cross the blood-brain barrier and circulate in plasma, carrying intracellular cargo from their cell of origin. Enriching plasma for this compartment before immunoassay converts a mixed-tissue measurement into a tissue-of-origin measurement.
The supporting literature in synucleinopathy is now substantial enough to plan against. In a cross-sectional study of 664 serum samples across three cohorts — including DLB, multiple system atrophy, frontotemporal dementia and progressive supranuclear palsy — neuron-derived exosomal α-synuclein was roughly twofold higher in prodromal and clinical Parkinson’s disease than in comparison groups, separated Parkinson’s from controls with an AUC of 0.86 across populations, and remained stably elevated in longitudinal samples [9]. A second group, working in two independent cohorts, found that α-synuclein measured in blood extracellular vesicles enriched by cell-type-specific surface capture distinguished Parkinson’s disease from multiple system atrophy — and that the ratio between putative glial and putative neuronal vesicle populations was the most discriminating readout [8].
That second result matters for LBD design specifically. It demonstrates that the informative variable is not the analyte concentration but the compartment it was drawn from. Two patients can carry the same total plasma α-synuclein and a different disease.
What to validate before committing a program to it
Enrichment is a sample preparation step, and sample preparation steps are where multi-site trials lose reproducibility. Before an NDEV-based readout goes into a protocol, ask for four things in writing:
- Precision, stated as within-batch and between-batch CV on pooled plasma across a realistic number of independent runs — not a single-day replicate experiment.
- Characterization consistent with MISEV2023, the ISEV consensus document on minimal information for extracellular vesicle studies, which sets expectations for how vesicle preparations are separated, characterized and reported [14]. A provenance claim without characterization data is an assertion.
- Pre-analytical tolerance data — freeze-thaw cycles, time to processing, haemolysis index, anticoagulant. LBD cohorts recruit from memory clinics and movement disorder clinics with heterogeneous sample handling.
- Multiplex capacity from a single draw. An LBD panel needs α-synuclein, a phosphorylated tau species and a neurodegeneration marker from the same aliquot. Sequential single-analyte assays on split samples introduce volume constraints and cross-assay variance the trial does not need.
Building the panel: three decision points, three requirements
An LBD blood biomarker strategy is easier to specify if it is written against decisions rather than analytes. Three decisions carry most of the risk in a phase 1/2 program.
Decision 1 — Who enters the trial
Two filters, not one. Confirm synucleinopathy, and separately measure Alzheimer’s co-pathology burden so it can be excluded or covaried. A single categorical synuclein result does not distinguish a patient with isolated Lewy pathology from one carrying intermediate-to-high Alzheimer’s neuropathologic change alongside it, and those two patients have different expected trajectories [1][11]. This is the filter that changed the phase 2b design discussed above [13].
Decision 2 — How the cohort is stratified
Pre-specify strata on continuous biomarker values rather than post hoc subgroups. The phase 2a analysis that produced the useful signal was post hoc [10]; the sponsor then had to run a second trial to confirm it prospectively [13]. That is an expensive way to learn a stratification rule. A multiplex panel measured at screening and baseline gives a program the option value of defining strata before unblinding.
Decision 3 — What moves during dosing
This is where provenance becomes non-negotiable. A pharmacodynamic or target engagement readout has to change with treatment and be attributable to brain tissue. Neither a categorical seeding result nor a mixed-origin plasma concentration satisfies both conditions. A brain-enriched, quantitative, repeatable measurement satisfies both in principle — and its analytical validation package is the thing to scrutinize before a program depends on it.
Glial fibrillary acidic protein (GFAP) and neurofilament light (NfL) belong in the panel as monitoring markers, not as primary readouts. Both associated with cognitive symptoms in the DLB consortium cohort [11], and both are useful for tracking neurodegeneration and neuroinflammation over time. Neither is specific to Lewy pathology, and neither will tell a sponsor whether a synuclein-targeted agent hit its target.
Where this is heading
The field is converging on a biological rather than syndromic definition of Lewy body disease, and the biomarker infrastructure is the rate-limiting step. Seed amplification assays established that antemortem molecular confirmation of synucleinopathy is achievable; standardization work now underway across laboratories will determine how reliably it transfers between sites [6]. In parallel, plasma phosphorylated tau has moved from research finding to enrollment criterion in a DLB trial inside three years [10][13]. The remaining gap is a quantitative, brain-specific, repeatable α-synuclein measurement — one that can be drawn monthly and interpreted as a statement about brain tissue rather than about blood. Enrichment for the neuron-derived vesicle compartment is currently the most direct route to closing it, and the programs that validate it thoroughly now will be the ones able to make target engagement claims in the next generation of LBD trials.
NeuroDex builds blood-based biomarker programs around this constraint, using ExoSORT™ to enrich plasma for the neuron-derived vesicle compartment before multiplex measurement. Talk to our team about biomarker strategy for an LBD program.
References
- Sekiya H, Matsubara T, DeTure MA, Dickson DW. Neuropathology of Lewy body dementia: Lewy-related pathology, α-synuclein oligomers, and comorbid pathologies. Molecular Neurodegeneration. 2025;20(1):117. https://doi.org/10.1186/s13024-025-00900-6
- Rizzo G, Arcuti S, Copetti M, Alessandria M, Savica R, Fontana A, Liguori R, Logroscino G. Accuracy of clinical diagnosis of dementia with Lewy bodies: a systematic review and meta-analysis. Journal of Neurology, Neurosurgery & Psychiatry. 2018;89(4):358–366. https://doi.org/10.1136/jnnp-2017-316844
- McKeith IG, Boeve BF, Dickson DW, et al. Diagnosis and management of dementia with Lewy bodies: fourth consensus report of the DLB Consortium. Neurology. 2017;89(1):88–100. https://doi.org/10.1212/WNL.0000000000004058
- Siderowf A, Concha-Marambio L, Lafontant DE, et al. Assessment of heterogeneity among participants in the Parkinson’s Progression Markers Initiative cohort using α-synuclein seed amplification: a cross-sectional study. The Lancet Neurology. 2023;22(5):407–417. https://doi.org/10.1016/S1474-4422(23)00109-6
- Bellomo G, Toja A, Paolini Paoletti F, et al. Investigating alpha-synuclein co-pathology in Alzheimer’s disease by means of cerebrospinal fluid alpha-synuclein seed amplification assay. Alzheimer’s & Dementia. 2024;20(4):2444–2452. https://doi.org/10.1002/alz.13658
- Kumar R, Gravett S, Jelic V, et al. Diagnostic performance of the α-synuclein seed amplification assay for dementia with Lewy bodies: a comparison across 4 laboratories. Neurology. 2026;106(4):e214614. https://doi.org/10.1212/WNL.0000000000214614
- Levin J, Baiardi S, Quadalti C, et al. α-Synuclein seed amplification assay detects Lewy body co-pathology in autosomal dominant Alzheimer’s disease late in the disease course and dependent on Lewy pathology burden. Alzheimer’s & Dementia. 2024;20(6):4351–4365. https://doi.org/10.1002/alz.13818
- Dutta S, Hornung S, Kruayatidee A, et al. α-Synuclein in blood exosomes immunoprecipitated using neuronal and oligodendroglial markers distinguishes Parkinson’s disease from multiple system atrophy. Acta Neuropathologica. 2021;142(3):495–511. https://doi.org/10.1007/s00401-021-02324-0
- Jiang C, Hopfner F, Katsikoudi A, et al. Serum neuronal exosomes predict and differentiate Parkinson’s disease from atypical parkinsonism. Journal of Neurology, Neurosurgery & Psychiatry. 2020;91(7):720–729. https://doi.org/10.1136/jnnp-2019-322588
- Alam JJ, Maruff P, Doctrow SR, Chu HM, Conway J, Gomperts SN, Teunissen C. Association of plasma phosphorylated tau with the response to neflamapimod treatment in patients with dementia with Lewy bodies. Neurology. 2023;101(17):e1708–e1717. https://doi.org/10.1212/WNL.0000000000207755
- Bolsewig K, van Unnik AAJM, Blujdea ER, et al. Association of plasma amyloid, P-tau, GFAP, and NfL with CSF, clinical, and cognitive features in patients with dementia with Lewy bodies. Neurology. 2024;102(12):e209418. https://doi.org/10.1212/WNL.0000000000209418
- Gonzalez MC, Ashton NJ, Gomes BF, et al. Association of plasma p-tau181 and p-tau231 concentrations with cognitive decline in patients with probable dementia with Lewy bodies. JAMA Neurology. 2022;79(1):32–37. https://doi.org/10.1001/jamaneurol.2021.4222
- Prins ND, de Haan W, Gardner A, Blackburn K, Chu HM, Galvin JE, Alam JJ. Phase 2A learnings incorporated into RewinD-LB, a phase 2B clinical trial of neflamapimod in dementia with Lewy bodies. The Journal of Prevention of Alzheimer’s Disease. 2024;11(3):549–557. https://doi.org/10.14283/jpad.2024.36
- Welsh JA, Goberdhan DCI, O’Driscoll L, et al. Minimal information for studies of extracellular vesicles (MISEV2023): from basic to advanced approaches. Journal of Extracellular Vesicles. 2024;13(2):e12404. https://doi.org/10.1002/jev2.12404
- Uemura MT, Robinson JL, Cousins KAQ, et al. Distinct characteristics of limbic-predominant age-related TDP-43 encephalopathy in Lewy body disease. Acta Neuropathologica. 2022;143(1):15–31. https://doi.org/10.1007/s00401-021-02383-3
Leave a Reply