¶ Experimental: CAAR-T Cell Therapy for Autoantibody-Mediated Neurotoxicity in AD
¶ Hypothesis
Chimeric Autoantibody Receptor (CAAR) T cells can selectively eliminate B cells producing pathogenic autoantibodies (anti-BACE1, anti-AQP4, anti-neuronal) in Alzheimer's disease patients, reducing autoantibody-mediated synaptic dysfunction and neuronal loss while preserving protective anti-Aβ antibody-producing B cells.
¶ Background
¶ Autoantibodies in AD
Alzheimer's disease patients harbor multiple autoantibody populations with context-dependent effects[@autoantibodies_review; @autoantibodies_pathogenesis]:
Pathogenic autoantibodies:
- Anti-BACE1 — inhibit β-secretase, disrupting APP processing and causing synaptic dysfunction[1]
- Anti-AQP4 — impair glymphatic clearance through astrocyte water channel disruption[2]
- Anti-HuD — target neuronal RNA-binding proteins, causing direct neuronal toxicity
Potentially protective autoantibodies:
- Natural anti-Aβ antibodies — facilitate Aβ clearance[3]
- Anti-4-HNE antibodies — neutralize lipid peroxidation products
¶ The CAAR-T Approach
CAAR-T cells are engineered T cells expressing a chimeric receptor that:
- Contains the target autoantigen (e.g., BACE1 extracellular domain) fused to T cell signaling domains
- When the B cell's surface immunoglobulin binds the autoantigen, the CAAR-T cell kills the B cell
- Preserves normal B cells (that don't bind the autoantigen) and antibody-producing cells against other targets
This approach has been validated in:
- Myasthenia gravis (anti-AChR CAAR-T cells)[4]
- Anti-NMDA receptor encephalitis (preclinical)
- Neuromyelitis optica (anti-AQP4 CAAR-T cells)
¶ Key Challenge: Selectivity
The critical therapeutic challenge is discriminating pathogenic from protective autoantibody-producing B cells. Anti-BACE1 and anti-AQP4 B cells are clearly pathogenic targets. However, anti-Aβ antibody-producing B cells are potentially protective — CAAR-T cells should NOT target these.
Solution: Use a multi-target CAAR-T approach that specifically targets BACE1-reactive and AQP4-reactive B cell clones while preserving anti-Aβ B cell populations. Alternatively, a combinatorial approach with selective depletion of pathogenic clones before introducing anti-Aβ immunotherapy.
¶ Experimental Design
¶ Phase I/IIa Clinical Trial
Title: CAAR-BACE1/CAAR-AQP4 T cell therapy for autoantibody-positive Alzheimer's disease
Design: Open-label, dose-escalation, single-arm
Population:
- N = 24 patients with probable AD (NIA-AA criteria)
- MMSE 18-26 (mild-to-moderate AD)
- Seropositive for either anti-BACE1 or anti-AQP4 autoantibodies (screening)
- Age 55-80 years
Primary Exclusions:
- Active autoimmune disease requiring immunosuppression
- Prior B cell depletion therapy within 12 months
- Active malignancy or history of lymphoma
- Anti-Aβ therapeutic within 6 months
¶ Intervention
Autologous CAAR-T cell infusion (single dose):
| Cohort | Cell Dose | Target |
|---|---|---|
| Cohort 1 (n=6) | 1×10^6 CAAR-T cells/kg | BACE1 or AQP4 |
| Cohort 2 (n=6) | 5×10^6 CAAR-T cells/kg | BACE1 or AQP4 |
| Cohort 3 (n=6) | 1×10^7 CAAR-T cells/kg | BACE1 or AQP4 |
| Cohort 4 (n=6) | Combinatorial (BACE1 + AQP4) | Both |
Manufacturing:
- Leukapheresis to collect patient PBMCs
- T cell transduction with lentiviral vector encoding BACE1-CAAR or AQP4-CAAR
- Ex vivo expansion for 10-14 days
- QC release testing (cell count, viability, transduction efficiency)
- Fresh infusion (not cryopreserved)
Conditioning: Optional low-dose cyclophosphamide (300 mg/m²) 3 days prior to infusion for lymphodepletion (cohorts 3-4 only)
¶ Endpoints
Primary Endpoint (Month 6):
- Change in autoantibody levels (anti-BACE1 or anti-AQP4 titer) in serum and CSF
- Safety: incidence of grade ≥3 adverse events, cytokine release syndrome (CRS)
Secondary Endpoints (Months 3, 6, 12):
- Cognitive: ADAS-Cog13, MMSE, CDR-SB
- Biomarker: CSF Aβ42/40 ratio, p-tau181, t-tau, NfL
- Brain imaging: Amyloid PET (Centiloid), MRI volumetry (hippocampal volume)
- B cell reconstitution: CD19+ B cell recovery, immunoglobulin levels
- Autoantibody panel: comprehensive serum autoantibody profiling (AI-omics approach)[5]
Exploratory Endpoints:
- TCR repertoire analysis of residual B cells (assess for compensatory clones)
- Treg function recovery
- Correlation of baseline autoantibody signatures with treatment response
¶ Monitoring
| Timepoint | Assessment |
|---|---|
| Baseline | Full workup, leukapheresis |
| Week 1 | Safety monitoring, cytokine panel (CRS grading) |
| Week 2 | B cell depletion confirmation |
| Month 1 | Safety, cognitive, CSF collection |
| Month 3 | Interim safety and biomarker assessment |
| Month 6 | Primary endpoint, full biomarker panel, PET imaging |
| Month 12 | Long-term follow-up, B cell reconstitution |
¶ Sample Size Justification
Based on:
- Expected autoantibody titer reduction of ≥50% (primary biomarker endpoint)
- Historical variability in autoantibody levels: SD = 30%
- One-sample t-test with α = 0.05, power = 80%
- Expected dropout rate: 15%
N = 24 provides 80% power to detect a mean 50% reduction in autoantibody titer from baseline, with two-sided α = 0.05.
¶ Safety Considerations
¶ Cytokine Release Syndrome (CRS)
- CAAR-T cells target autoantigens on B cells (not T cells or neurons), minimizing on-target/off-tumor toxicity
- Expected CRS rates lower than CD19 CAR-T (which targets all B cells)
- Grade 1-2 CRS expected in up to 40%; Grade 3-4 in <10%
- Tocilizumab and corticosteroid rescue protocols in place
¶ B Cell Aplasia
- Selective depletion of autoantibody-producing B cells may reduce overall immunoglobulin levels
- Monitor IgG, IgM, IgA every 3 months
- IVIG replacement for symptomatic hypogammaglobulinemia
¶ Off-Target Effects
- CAAR constructs use scFv-like autoantigen display; potential cross-reactivity assessed in vitro
- Off-tumor/off-target toxicity monitoring in first-in-human cohorts
¶ Biomarker Companion Study
Integrated autoantibody profiling using AI-omics approach (PMID:40406128):
- Baseline characterization: Comprehensive serum + CSF autoantibody array (500+ antigens)
- Longitudinal tracking: Autoantibody panel at baseline, months 1, 3, 6, 12
- Single-cell B cell profiling: BCR repertoire from peripheral blood, correlate with autoantibody signatures
- Machine learning classifier: Train model to predict treatment response based on baseline autoantibody signature
This creates a precision medicine framework for identifying which AD patients will benefit from CAAR-T therapy.
¶ Statistical Analysis Plan
Primary analysis: Paired t-test (or Wilcoxon signed-rank for non-normal data) comparing baseline to month 6 autoantibody titers.
Secondary analyses:
- Mixed-effects model for repeated cognitive measures (ADAS-Cog13, CDR-SB)
- ANCOVA for biomarker changes adjusting for baseline covariates (APOE4 status, age, baseline MMSE)
- Subgroup analysis by autoantibody type (anti-BACE1 vs anti-AQP4 vs both)
Biomarker correlate analysis: Pearson/Spearman correlation between autoantibody reduction and cognitive/biomarker changes.
Missing data: Multiple imputation under MAR assumption; sensitivity analysis under MNAR.
¶ Success Criteria
Phase I go/no-go criteria (Month 6):
- ✓ Autoantibody titer reduced ≥50% in ≥60% of patients
- ✓ No Grade 4 CRS or unexpected safety signals
- ✓ B cell reconstitution occurring by month 9-12
Phase II expansion criteria:
- ✓ Cognitive stabilization or improvement (ADAS-Cog13 decline
points vs historical) - ✓ Amyloid PET stabilization or reduction (Centiloid change <5)
- ✓ Biomarker improvements consistent with AD modification
¶ Projected Timeline
| Milestone | Timeline |
|---|---|
| IND filing | Month 0 |
| Site activation (3 sites) | Month 6 |
| First patient enrolled | Month 9 |
| Last patient enrolled | Month 24 |
| Primary endpoint (6M FU) | Month 30 |
| Data lock + top-line results | Month 33 |
| Phase II protocol finalization | Month 36 |
¶ Budget Estimate
| Category | Cost (USD) |
|---|---|
| Manufacturing (24 × CAAR-T) | $1.2M |
| Clinical operations (3 sites, 30 months) | $1.8M |
| Biomarker companion studies | $400K |
| Imaging (PET/MRI) | $300K |
| Regulatory/IND | $150K |
| Data management | $200K |
| Total estimated | ~$4.1M |
¶ Related Pages
- Autoimmune Hypothesis in AD — mechanistic foundation
- Hypothesis Rankings — Autoimmune score: 59/100
- CAAR-T cells mechanism — therapeutic approach
¶ References
¶ Additional references
Anti-BACE1 autoantibodies in Alzheimer's disease: prevalence and functional impact. 2023. ↩︎
Aquaporin-4 autoantibodies in Alzheimer's disease: prevalence and clinical associations. 2023. ↩︎
Natural anti-amyloid-beta antibodies in Alzheimer's disease and healthy aging. 2023. ↩︎
Chimeric autoantibody receptor T cells for myasthenia gravis. 2023. ↩︎
Autoantibodies in Alzheimer's disease: Multifaceted roles and therapeutic horizons. 2025. ↩︎