{"ranked_hypotheses":[{"title":"H3: Disease-Associated Microglia (DAM) Phase Boundary as Intervention Window","description":"The therapeutic switch from TREM2 activation to inhibition should occur at the precise point when microglia complete the Stage 1→Stage 2 DAM transition, marked by Apoe expression and Lpl induction. Early TREM2 agonism accelerates beneficial Stage 1 function; inhibition after Stage 2 completion prevents maladaptive lipid accumulation. This represents the most mechanistically grounded hypothesis with clear operationalizable markers (TREM2+/CX3CR1+ homeostatic to APOE+/LPL+ lipid metabolism signature transition). While the 'phase boundary' requires precise molecular definition, existing single-cell transcriptomics datasets from prodromal AD cohorts provide immediate testing capability. The AL002 Phase 2 trial represents the most relevant existing dataset for validation.","target_gene":"TREM2, APOE, LPL, CX3CR1","composite_score":0.62,"evidence_for":[{"claim":"TREM2 acts as molecular gatekeeper for homeostatic to DAM transition in 5xFAD mice","pmid":"28678778"},{"claim":"Two distinct DAM stages require TREM2 only for Stage 1→2 transition","pmid":"28678778"},{"claim":"TREM2 agonists (AL002) in Phase 2 trials show biomarker stratification potential","pmid":"NCT05135042"}],"evidence_against":[{"claim":"DAM-like signatures appear in normal aging, challenging pathological framing","pmid":"30662123"},{"claim":"'Phase boundary' not operationally defined; single-cell studies define states, not transitions","pmid":"NA"},{"claim":"TREM2-independent DAM-like cells exist in later disease stages","pmid":"32015132"}]},{"title":"H2: Microglial TREM2 Surface Density as Phase Transition Biomarker","description":"The therapeutic switch should occur when microglial TREM2 surface expression density drops below functional threshold. CSF-soluble TREM2 (sTREM2) serves as an accessible proxy for surface expression. This hypothesis benefits from existing biomarker infrastructure (sTREM2 measured in ADNI, DIAN, and other major cohorts) and the practical feasibility of retrospective analysis within existing TREM2 trials. The 1,000-2,000 receptor threshold from Nasu-Hakola disease data requires AD-specific validation, and the relationship between sTREM2 and functional surface density is complex due to multiple shedding regulation mechanisms.","target_gene":"TREM2 (full-length surface), sTREM2 (cleavage product), ADAM10","composite_score":0.61,"evidence_for":[{"claim":"CSF sTREM2 correlates with brain TREM2 expression in human cohorts","pmid":"27477021"},{"claim":"TREM2 undergoes ADAM10-mediated ectodomain shedding; sTREM2 is detectable in CSF","pmid":"26594862"},{"claim":"sTREM2 is already measured in major AD cohorts enabling retrospective validation","pmid":"ADNI database"}],"evidence_against":[{"claim":"R47H carriers with ~50% surface expression still develop AD, suggesting continuous rather than binary risk","pmid":"29263222"},{"claim":"sTREM2 kinetics are complex, reflecting shedding, neuronal death, and microglial burden independently","pmid":"34240521"},{"claim":"Nasu-Hakola threshold derived from different disease context","pmid":"21459767"}]},{"title":"H4: APOE Isoform-Specific Temporal Windows","description":"Optimal TREM2 modulation timing is determined by APOE isoform-dependent microglial metabolic rewiring. APOE4 carriers exhibit accelerated dysfunction requiring earlier phase transitions; APOE2 carriers show delayed impairment extending activation windows. This framework is implementable through standard APOE genotyping but requires mechanistic validation of the APOE-TREM2 dynamic relationship. The genetic interaction (multiplicative risk) is established, but specific timing predictions are speculative and not derived from prospective data.","target_gene":"APOE (ε2, ε3, ε4), TREM2, ABCA1","composite_score":0.52,"evidence_for":[{"claim":"APOE genotype modifies TREM2 AD risk; R47H effect strongest in APOE4 carriers","pmid":"29263222"},{"claim":"APOE4 impairs microglial cholesterol efflux and reduces lipid-binding capacity","pmid":"29691480"},{"claim":"APOE genotyping is standard of care enabling easy trial stratification","pmid":"Clinical standard"}],"evidence_against":[{"claim":"APOE-TREM2 interaction is primarily genetic, not clearly mechanistic","pmid":"NA"},{"claim":"APOE2 is protective despite different microglial lipid responses","pmid":"25411441"},{"claim":"Specific '3-5 years before MCI onset' timing for APOE4 not derived from data","pmid":"NA"}]},{"title":"H6: Metabolic State Transition as Phase Switch Criterion","description":"The optimal therapeutic window is defined by microglial metabolic reprogramming from oxidative phosphorylation (OxPHOS) to glycolysis (Warburg-like shift). TREM2 agonism should precede the OxPHOS→glycolysis transition; inhibition should follow it. Lactate accumulation and succinate dehydrogenase activity decline in CSF could serve as biomarkers. However, this hypothesis requires development of HIF1α inhibitors for CNS use and validation that TREM2 manipulation can alter metabolic trajectory at all.","target_gene":"TREM2, HIF1α, lactate, succinate dehydrogenase, PGC-1α","composite_score":0.45,"evidence_for":[{"claim":"DAM microglia show glycolytic signature with lactate dehydrogenase B upregulation","pmid":"28973389"},{"claim":"TREM2 deficiency impairs mitochondrial complex IV function","pmid":"28842538"},{"claim":"Glycolytic microglia show reduced phagocytic capacity","pmid":"28973389"}],"evidence_against":[{"claim":"HIF1α inhibitors for CNS use do not exist, requiring novel compound development","pmid":"NA"},{"claim":"Unclear whether TREM2 agonism can alter metabolic trajectory","pmid":"NA"},{"claim":"Longest timeline (10-15 years) and highest development cost ($300-500M)","pmid":"NA"}]},{"title":"H5: Neurodegeneration-Onset Timing Based on TREM2-Dependent Pyroptosis Threshold","description":"The phase transition should be triggered when microglia reach NLRP3 inflammasome activation threshold. While mechanistically attractive (TREM2→SYK→NLRP3 axis), this hypothesis inverts the relationship demonstrated by Zhang et al. (2022), which showed TREM2 negatively regulates NLRP3. The correct framing would be timing intervention to NLRP3 activation onset rather than TREM2 hyperactivation. MCC950 and dapansutrile provide existing NLRP3 inhibitor options for alternative implementation.","target_gene":"TREM2, NLRP3 inflammasome, CASP1, GSDMD, SYK","composite_score":0.36,"evidence_for":[{"claim":"GSDMD pores detected in AD microglia","pmid":"33239777"},{"claim":"NLRP3 inhibitors (MCC950) show efficacy in AD models","pmid":"27974631"},{"claim":"TREM2-DAP12-SYK signaling axis exists in microglia","pmid":"28842538"}],"evidence_against":[{"claim":"Zhang et al. (2022) demonstrated TREM2 negatively regulates NLRP3; hypothesis inverts this","pmid":"35292473"},{"claim":"TREM2 activation primarily activates anti-inflammatory/pro-survival pathways","pmid":"28842538"},{"claim":"If TREM2 activation were pro-pyroptotic, TREM2 loss-of-function should be protective","pmid":"NA"}]},{"title":"H1: Amyloid Phospholipid Composition Ratio as Phase Switch Trigger","description":"Optimal phase transition from TREM2 activation to inhibition occurs when amyloid plaques undergo compositional shift when oxidized phospholipid to native lipid ratio crosses a critical threshold. This hypothesis is mechanistically plausible (TREM2 binds lipid ligands) but fundamentally limited by absence of validated in vivo biomarkers for oxidized phospholipid composition. No PET ligand, CSF assay, or imaging approach exists for measuring this ratio in living subjects.","target_gene":"TREM2, oxidized phospholipids (oxPL), SYK pathway","composite_score":0.28,"evidence_for":[{"claim":"TREM2 preferentially binds lipid ligands with distinct affinities for native versus oxidized species","pmid":"32302526"},{"claim":"SYK hyperactivation occurs in late-stage microglia","pmid":"36697000"}],"evidence_against":[{"claim":"No validated in vivo imaging agent or CSF biomarker exists for oxPL composition","pmid":"NA"},{"claim":"'Critical threshold ratio' entirely invented with no empirical basis","pmid":"NA"},{"claim":"Oxidized lipids accumulate with normal aging, not specifically in AD","pmid":"29891877"},{"claim":"Development cost >$500M with high probability of failure","pmid":"NA"}]},{"title":"H7: Network-Level Synchronization Threshold in Microglial Clusters","description":"Individual microglial TREM2 states are less critical than emergent network behaviors when >40% of plaque-associated microglia enter coordinated DAM states. Below this threshold, TREM2 activation remains therapeutic; above it, synchronized responses drive collective neurotoxicity. This hypothesis cannot be tested in living subjects due to fundamental imaging limitations; single-cell transcriptomics, CX43 connectivity mapping, and fate-mapping require tissue and are not applicable to human longitudinal studies.","target_gene":"TREM2, C1q, C3, complement system, CX43 (gap junctions)","composite_score":0.25,"evidence_for":[{"claim":"Microglial clustering around plaques shows coordinated gene expression","pmid":"28678778"},{"claim":"Complement deposition on synapses requires activated microglia","pmid":"29849141"}],"evidence_against":[{"claim":"Cannot measure '40% coordinated DAM states' in living subjects","pmid":"NA"},{"claim":"No PET ligand exists for microglial coordination state","pmid":"NA"},{"claim":"Real-time imaging of gap junction connectivity not possible in human cortex","pmid":"NA"},{"claim":"Longitudinal fate-mapping in human subjects not possible","pmid":"NA"}]}],"synthesis_summary":"The seven hypotheses for temporal TREM2 modulation in Alzheimer's disease represent a spectrum from mechanistically grounded (H3: DAM phase boundary) to currently untestable (H7: network synchronization). H3 and H2 emerge as the highest priority for immediate translational development due to their operationalizable biomarker frameworks leveraging existing single-cell transcriptomics datasets (Keren-Shaul et al., 2017) and established CSF sTREM2 measurement platforms in major AD cohorts. H4 provides a critical stratification framework that could explain trial failures in unstratified populations, though specific timing predictions require prospective validation. The remaining hypotheses face fundamental limitations: H1 and H7 lack necessary biomarker/imaging infrastructure, H5 requires mechanistic revision (inverted TREM2-NLRP3 relationship), and H6 requires novel compound development for HIF1α inhibition.\n\nThe critical bottleneck across all hypotheses is not target validation—TREM2 is a proven therapeutic target with active clinical programs including AL002 (Phase 2)—but rather the lack of validated biomarker-based decision algorithms for determining when to switch therapeutic modalities. This suggests a research strategy prioritizing biomarker validation over additional target validation, with particular emphasis on APOE genotype-stratified longitudinal studies combining sTREM2 trajectory analysis with DAM state markers. An adaptive clinical trial design incorporating interim biomarker-driven randomization based on H3/H2 hybrid biomarkers could simultaneously address multiple hypotheses while minimizing development cost and timeline.","knowledge_edges":[{"source_id":"H3","source_type":"hypothesis","target_id":"Keren-Shaul et al. 2017","target_type":"primary_literature","relation":"defines_stages"},{"source_id":"H2","source_type":"hypothesis","target_id":"sTREM2 biomarker","target_type":"clinical_biomarker","relation":"measured_by"},{"source_id":"H3","source_type":"hypothesis","target_id":"H2","target_type":"hypothesis","relation":"complementary_biomarkers"},{"source_id":"H4","source_type":"hypothesis","target_id":"APOE genotype","target_type":"genetic_factor","relation":"stratifies_by"},{"source_id":"H4","source_type":"hypothesis","target_id":"H3","target_type":"hypothesis","relation":"modifies_timing"},{"source_id":"H5","source_type":"hypothesis","target_id":"Zhang et al. 2022","target_type":"primary_literature","relation":"contradicted_by"},{"source_id":"H1","source_type":"hypothesis","target_id":"Wang et al. 2020","target_type":"primary_literature","relation":"evidence_basis"},{"source_id":"H6","source_type":"hypothesis","target_id":"Ulland et al. 2017","target_type":"primary_literature","relation":"evidence_basis"},{"source_id":"H7","source_type":"hypothesis","target_id":"CX43","target_type":"target_protein","relation":"requires_imaging"},{"source_id":"AL002","source_type":"clinical_trial","target_id":"H3","target_type":"hypothesis","relation":"tests"},{"source_id":"AL002","source_type":"clinical_trial","target_id":"H2","target_type":"hypothesis","relation":"retrospective_validation"}]}