## Hypothesis 1: MOGAD FcγR-Mediated Demyelination
### Weak Links
- **Overreliance on EAE models**: Takai et al. used C57BL/6 EAE—a fundamentally different disease model than human MOGAD, which rarely produces spontaneous relapse in the same pattern
- **Mechanistic exclusivity**: The hypothesis claims this mechanism "predominates over complement," but Peschl et al. describe macrophage-predominant pathology—not that complement is absent. Multiple MOGAD cases show complement deposition (Fischer et al., 2020)
- **CSF1R/IRAK4 as loose ends**: These pathways are included without mechanistic justification for their specific role in MOG-FcγR axis
### Counter-Evidence
- K旁的尸体解剖 data (Takeshita et al., 2017) show complement C9 deposition in MOGAD lesions
- Human MOGAD peripheral blood shows Th17-skewed responses (Wang et al., 2020), suggesting T-cell primacy
- Fostamatinib failed in a Phase 2 NMOSD trial (NCT02369354) where AQP4-seropositive patients dominated; no data for MOGAD specifically
### Falsifying Experiments
1. **Required**: Demonstrate that FcγRIII knockout mice are fully protected from passive MOG-IgG transfer (current data show partial protection only)
2. **Test**: Deplete macrophages via CSF1R blockade (PLX3397) in transfer model—demyelination should persist if antibody-complement axis operates independently
3. **Human validation**: Quantify FcγR genotype polymorphisms (FCGR2A/3A) in MOGAD vs. controls—association would strengthen causality
### Revised Confidence: 0.58 (down from 0.72)
---
## Hypothesis 2: Paraneoplastic CD8+ T Cell Neuronal Injury
### Weak Links
- **Mechanistic extrapolation**: The cited Pittock/Lennon paper concerns Yo antibody cerebellar degeneration—not spinal cord; CD8 cytotoxicity in the original paper operates through different effector mechanisms
- **RIPK3/MLKL in motor neurons**: Necroptosis is well-characterized in immune cells and some epithelial cells, but direct evidence of RIPK3-mediated death in spinal motor neurons is absent (Kay et al., 2016 showed they may be resistant)
- **Viral antigen specificity**: EBV/HHV-6 association with paraneoplastic myelopathy specifically is not established—this is extrapolated from MS literature
### Counter-Evidence
- Paraneoplastic myelopathies associated with well-characterized antibodies (anti-Hu, Yo, CRMP5) show extensive neuronal loss but **variable** T cell infiltration patterns (Graus et al., 2021)
- Anti-Hu syndrome involves CDR2 antigen presentation via dendritic cells—not direct CD8 killing
- Some paraneoplastic myelopathies improve with IVIG/rituximab, suggesting antibody-mediated or reversible components
### Falsifying Experiments
1. **Required**: Perform TCR sequencing of spinal cord-infiltrating CD8+ T cells in paraneoplastic myelopathy autopsy tissue—show antigen-specific expansion vs. bystander activation
2. **Test**: RIPK3 knockout mice infected with EBV and engrafted with HLA-A*02:01 human neurons—do motor neurons undergo necroptosis?
3. **Negative prediction**: If you eliminate CD8+ T cells (CD8β knockout) in a paraneoplastic mouse model (e.g., Yo antigen vaccination), myelopathy should not develop—does it?
### Revised Confidence: 0.41 (down from unstated—estimated 0.55)
---
## Cross-Cutting Issues
| Issue | Hypothesis 1 | Hypothesis 2 |
|-------|--------------|--------------|
| Human tissue evidence | Weak (3 autopsy cases) | Absent (no spinal cord TCR data) |
| Therapeutic prediction specificity | SYK inhibitors lack MOGAD efficacy data | No testable therapy proposed |
| Mechanistic exclusivity | Claims predominance without ruling out complement | Claims irreversibility from necroptosis without ruling out other death pathways |
**Overarching recommendation**: Both hypotheses would benefit from integration—MOGAD may involve both FcγR-dependent demyelination and subsequent CD8+ responses to released antigens, while paraneoplastic myelopathy likely involves simultaneous antibody and T-cell mechanisms. The paper's focus on "uncommon" myelopathies with poorly characterized mechanisms would benefit from more cautious framing given these evidentiary gaps.