I treat “confidence” here as confidence in the **therapeutic causal hypothesis**, not merely confidence that the pathway is biologically involved.
## 1. TDP-43 LLPS Dysregulation in ALS/FTD
**Weak links:** TDP-43 condensates are real, but LLPS may be an upstream stress response or assay artifact rather than the central toxic transition in vivo. Patient inclusions are end-stage, heterogeneous, and often reflect loss of nuclear TDP-43 function, RNA splicing defects, impaired nucleocytoplasmic transport, and stress-granule biology all at once. Chaperone overexpression could reduce visible aggregates while worsening proteostasis or masking toxicity.
**Counter-evidence / confounds:** TDP-43 pathology is common across ALS/FTD, but familial TARDBP mutations are a minority of ALS. Stress granule colocalization does not prove stress granules seed disease. FRAP after arsenite stress may model acute cellular injury more than chronic motor neuron degeneration. Recent reviews still frame LLPS-to-aggregation as plausible but not therapeutically settled. Source: [Carey & Guo 2022](https://pmc.ncbi.nlm.nih.gov/articles/PMC8847598/), [Song 2024](https://pubmed.ncbi.nlm.nih.gov/38029395/).
**Falsifying experiments:**
Show that restoring droplet dynamics normalizes TDP-43 nuclear function, cryptic exon suppression, axonal transport, and motor neuron survival **without** merely dissolving inclusions. In vivo, conditional neuron-specific modulation of TDP-43 condensate-driving domains should alter disease progression in a direction predicted by LLPS dynamics.
**Revised confidence:** **0.58**.
## 2. GBA1 Loss-of-Function Exacerbates Alpha-Synuclein via Lysosomal Impairment
**Weak links:** This is genetically strong, but the therapeutic version is less certain. GBA1 mutations increase PD risk, yet penetrance is incomplete, and substrate accumulation, ER stress, lipid remodeling, inflammation, and alpha-synuclein handling may all contribute. TFEB activation is broad and may rescue lysosomal stress without proving the GBA1-alpha-syn feed-forward loop is primary.
**Counter-evidence / confounds:** Reduced GCase activity in sporadic PD substantia nigra supports convergence, but it could be secondary to neurodegeneration or alpha-syn burden rather than causal. Dat-Cre Gba1 knockout may create developmental or cell-type-specific artifacts and may not recapitulate heterozygous human GBA1-PD. Source: [Gegg et al.](https://pubmed.ncbi.nlm.nih.gov/23034917/).
**Falsifying experiments:**
Rescue experiments should compare GCase enzymatic restoration, substrate lowering, TFEB activation, and alpha-syn lowering side by side. If substrate normalization fails to reduce alpha-syn seeding, lysosomal impairment, or dopaminergic vulnerability in aged humanized alpha-syn models, the central feed-forward claim weakens.
**Revised confidence:** **0.74**.
## 3. TREM2-SYK Axis in Microglial State Transitions
**Weak links:** The biology is strong; the therapeutic claim is now less secure. TREM2 can promote plaque-associated microglial responses, but “more TREM2 signaling” is not necessarily better. Microglial activation can compact plaques while increasing inflammatory injury, synapse loss, or tau-related neurodegeneration. Effects are likely disease-stage and pathology-context dependent.
**Counter-evidence / confounds:** TREM2 deficiency changes plaque-associated microglia and neuritic injury in mouse models, but amyloid mouse models poorly predict human cognitive benefit. Most importantly, the 2026 phase 2 AL002 trial in early AD showed target engagement and pharmacodynamic response but did **not** meet the primary clinical endpoint. Source: [TREM2 review](https://pmc.ncbi.nlm.nih.gov/articles/PMC9783481/), [AL002 phase 2 Nature Medicine 2026](https://www.nature.com/articles/s41591-026-04273-1).
**Falsifying experiments:**
A good falsifier is a longitudinal amyloid-plus-tau model where TREM2 agonism improves plaque metrics but worsens tau spread, synapse density, neurofilament, or cognition. In humans, failure to improve downstream neurodegeneration biomarkers despite robust microglial target engagement would argue the axis is not sufficient therapeutically.
**Revised confidence:** **0.60**.
## 4. C9orf72 DPR Toxic Gain-of-Function
**Weak links:** C9orf72 disease is not only DPR toxicity. Repeat RNA foci, antisense transcripts, C9orf72 haploinsufficiency, TDP-43 pathology, immune dysfunction, nucleocytoplasmic transport, and cell-type vulnerability all compete as drivers. Poly-GR/PR toxicity in models can be strong but sometimes depends on supraphysiologic expression.
**Counter-evidence / confounds:** The ASO story is a major caution. BIIB078 reduced DPR biomarkers but did not show clinical benefit and was discontinued; later analyses reported persistent key CNS pathologies despite target engagement. That does not refute DPR toxicity, but it weakens “DPR reduction alone is sufficient.” Sources: [C9orf72 discovery](https://pubmed.ncbi.nlm.nih.gov/21944778/), [BIIB078 trial](https://www.sciencedirect.com/science/article/pii/S1474442224002163), [2025 molecular impact](https://pubmed.ncbi.nlm.nih.gov/40865525/).
**Falsifying experiments:**
If selective DPR suppression in human motor neuron-organoid or animal models leaves TDP-43 mislocalization, neurodegeneration, and functional decline unchanged despite large DPR reductions, DPRs are probably not the dominant therapeutic node. Experiments must separately suppress sense RNA, antisense RNA, and restore C9orf72 protein.
**Revised confidence:** **0.72**.
## 5. Astrocyte Senescence Drives Neuroinflammation in ALS
**Weak links:** Senescence markers are not the same as irreversible senescence. Reactive astrocytes, DNA damage responses, aging, inflammation, oxidative stress, and terminal disease state can all induce p16/p21/SASP-like signatures. Senolytics have broad off-target effects and may remove cells that are compensatory or trophic.
**Counter-evidence / confounds:** SOD1G93A is an aggressive familial ALS model and may overstate astrocyte-autonomous mechanisms. ALS CSF cytokines are not cell-source-specific. Dasatinib/quercetin and navitoclax have major immune, vascular, platelet, and CNS penetration concerns. Young astrocyte rescue supports astrocyte dysfunction, not necessarily senescence as the driver.
**Falsifying experiments:**
Use astrocyte-specific genetic ablation of p16-high or p21-high cells, not systemic senolytics alone. If selective removal of senescent astrocytes reduces SASP but does not preserve motor neurons, NMJ integrity, or survival, the hypothesis fails therapeutically. Single-cell spatial profiling must show senescent astrocytes precede motor neuron loss.
**Revised confidence:** **0.45**.
## 6. Mitophagy Induction in Sporadic Parkinson’s Disease
**Weak links:** Mitochondrial dysfunction is central in PD, but “induce mitophagy” is too broad. Excess mitophagy can deplete mitochondria, impair energetics, or stress already vulnerable dopaminergic neurons. PINK1/Parkin familial PD does not necessarily generalize to sporadic PD, where alpha-synuclein, lysosomes, inflammation, pesticides, aging, and vascular factors interact.
**Counter-evidence / confounds:** PINK1 knockout mice often have weak or late dopaminergic neurodegeneration, so rescue in that model may not translate. Nicotinamide riboside changes NAD metabolism broadly, not specifically mitophagy. Human phase I data show target engagement and short-term signals, not disease modification. Source: [NADPARK phase I](https://www.sciencedirect.com/science/article/pii/S1550413122000456).
**Falsifying experiments:**
In aged sporadic PD-relevant models with alpha-syn pathology, demonstrate that mitophagy flux increases in substantia nigra dopaminergic neurons and that blocking mitophagy genetically abolishes benefit. If NR/urolithin benefits persist when PINK1/Parkin-dependent mitophagy is blocked, the proposed mechanism is wrong.
**Revised confidence:** **0.57**.
## 7. Tau Propagation via Muscarinic Receptor-Mediated BBB Transcytosis
**Weak links:** This is the least secure. LRP1-mediated tau uptake has credible evidence, but the jump to M1/M3 muscarinic receptor-mediated BBB transcytosis and peripheral-to-CNS tau propagation is speculative. Tau spread in AD is more parsimoniously explained by neuronal connectivity, local release/uptake, glial handling, LRP1/heparan sulfate pathways, and regional vulnerability.
**Counter-evidence / confounds:** The proposed experiment injects tau seeds directly into brain, so it does not test BBB transcytosis from periphery into CNS. CHRM1/CHRM3 double knockout has major cholinergic, vascular, cognitive, and BBB physiology confounds. Tau PET in mice is technically difficult and tracer specificity can be problematic. Source: [LRP1 tau uptake/spread](https://pubmed.ncbi.nlm.nih.gov/32296178/).
**Falsifying experiments:**
Use peripheral tau seed administration with endothelial-specific CHRM1/CHRM3 deletion and direct measurement of labeled tau crossing the BBB. If endothelial muscarinic loss does not reduce CNS tau entry while LRP1 or heparan sulfate perturbation does, the muscarinic transcytosis claim should be downgraded sharply.
**Revised confidence:** **0.32**.
**Overall ranking after critique:** strongest mechanistic targets are **GBA1/GCase** and **C9orf72 DPR/RNA biology**, but both need better translation strategy. **TREM2** remains biologically important but clinically weakened by AL002. **Tau-muscarinic BBB transcytosis** and **astrocyte senescence as primary ALS driver** need the most basic causal validation.