Audit: Submission-Program Divergence Blindspots

Date: 2026-04-17 Triggered by: Rubinsztein (#46) case study -- submission scored 60/100 as a strategic meta-argument about methodology, but his lab's BIN1-ESCRT-III autophagosome closure work is directly load-bearing for Bioenergetic Collapse.

Completion update (2026-05-22): All Tier 1 entries (Chini, Stevens, Gate, Deczkowska, Höglinger) and both Priority-2 entries (Tsai re-eval, Rubinsztein already done 2026-04-21) now have DeepResearch reviews. Integration recommendations for the ADC website are in research/collapse-trilogy/INTEGRATION_MEMO_2026-05-22.md. Tier 2 (Sekhar, Eckert, Grochowska, Golde, Boche, Chini-Schwartz cross-ref) and Tier 3 (Ramsden, Paul, Tsai re-eval) still pending automated mechanism-tag audit (§4 Priority 3) and synthesis-relevance frontmatter rollout (§4 Priority 4).

---

1. The Problem

The ONS scoring pipeline evaluates what entrants submitted (their written hypothesis/framework), not what their research programs actually produce. This creates a systematic blindspot: researchers who submitted broad, strategic, or meta-level arguments get modest scores, even when their active lab work generates mechanisms that are directly load-bearing for the evolving Collapse trilogy.

The original CSC framework defined six convergence nodes. The synthesis has since expanded to include Homeostatic Microglial Collapse and Bioenergetic Collapse, introducing new load-bearing mechanisms (autophagy-lysosomal machinery, NLRP3 inflammasome, NAD+ decline, PINK1/Parkin mitophagy, complement-mediated pruning, PV+ interneuron vulnerability, perineuronal nets) that were not primary scoring targets when entrants were first evaluated.

---

2. Flagged Entrants

TIER 1: HIGH-PRIORITY BLINDSPOTS (Immediate DeepResearch review recommended)

#127 Eduardo Chini -- NAD+ / CD38 / Inflammaging — ✅ REVIEWED 2026-05-22 (bioenergetic/ONS_Chini_NAD_Review.md)

• Submission said: NAD metabolic dysfunction drives AD through inflammaging, CD38+ senescent cells consume NAD+ and trigger amyloid accumulation as sterile inflammatory response.
• Scores: Relevancy 59.2, TKQ 57.0
• What his program actually contains: Chini is one of the world's leading authorities on CD38-mediated NAD+ depletion in aging. His lab's work on CD38 as an immunometabolic checkpoint is directly load-bearing for the Bioenergetic Collapse thesis -- specifically the NAD+ decline / sirtuin signaling axis and the metabolic reprogramming node. His submission adequately describes the mechanism, but the scoring pipeline underweighted it because the CSC framework had no dedicated node for bioenergetic/metabolic collapse.
• Connects to: Bioenergetic Collapse (NAD+ decline, sirtuin signaling, metabolic reprogramming), Homeostatic Microglial Collapse (CD38 on microglia, immunometabolic checkpoints)
• Divergence type: Framework gap -- the submission was accurate but the scoring framework lacked the nodes to capture it.

#158 Beth Stevens -- Complement Pruning / BAMs / SORL1-Lysosomal — ✅ REVIEWED 2026-05-22 (homeostatic-microglial/ONS_Stevens_Complement_Review.md)

• Submission said: Neurovascular dysfunction hypothesis, framed as "dance battle between human and germ," with BBB breakdown and antimicrobial peptide framing.
• Scores: Relevancy 65.3, TKQ 58.0
• What her program actually contains: Stevens' lab is the definitive source for complement-mediated synaptic pruning (C1q, C3, CR3) -- the exact mechanism that is load-bearing for both Homeostatic Microglial Collapse and Convergent Synaptic Collapse. Her recent work on border-associated macrophages with circadian regulation (Bmal1 deletion worsening plaque burden) and SORL1's regulation of lysosomal function in microglia directly feed both HMC and Bioenergetic Collapse. The submission framing as an infection/neurovascular hypothesis dramatically undersells the actual program.
• Connects to: Homeostatic Microglial Collapse (complement-mediated synaptic pruning -- central mechanism), Convergent Synaptic Collapse (pruning-mediated synapse loss), Bioenergetic Collapse (SORL1-lysosomal pathway, BAM function)
• Divergence type: Submission framing mismatch -- submitted an infection-oriented hypothesis but the lab's mechanistic output is complement/microglial/lysosomal.

#139 David Gate -- NLRP3 / Complement / Microglial Dysfunction — ✅ REVIEWED 2026-05-22 (homeostatic-microglial/ONS_Gate_NLRP3_Review.md)

• Submission said: Antiviral neuroinflammation as primary AD driver, with viral infections exploiting immunosenescence.
• Scores: Relevancy 62.8, TKQ 56.0
• What his program actually contains: Gate's submission explicitly names NLRP3 inflammasome activation, complement system activation, and TREM2 -- three of the most load-bearing molecules for the Collapse trilogy. His key molecules list includes NLRP3, IL-1beta, Complement C5a, and TREM2. The scoring penalized the viral framing as somewhat generic, but the actual mechanistic content maps almost perfectly to both Homeostatic Microglial Collapse (TREM2 receptor mechanics, complement pruning) and Bioenergetic Collapse (NLRP3 inflammasome / mitochondrial DAMPs).
• Connects to: Homeostatic Microglial Collapse (TREM2, complement system), Bioenergetic Collapse (NLRP3 inflammasome, mitochondrial DAMPs)
• Divergence type: Framing penalty -- correct mechanisms buried under a viral etiology narrative that was scored as less novel.

#70 Aleksandra Deczkowska -- Microglial Senescence / Complement / MMP-2/9 — ✅ REVIEWED 2026-05-22 (homeostatic-microglial/ONS_Deczkowska_DAM_Review.md)

• Submission said: AD as accelerated aging of Abeta management, with inflamm-aging, microglial senescence, mitochondrial dysfunction.
• Scores: Relevancy 61.3, TKQ 60.0
• What her program actually contains: Deczkowska is a pioneer in disease-associated microglia (DAM) biology -- she co-authored the original DAM signature paper. Her submission explicitly lists complement-mediated synapse elimination and MMP-2/MMP-9 in the key molecules. These are precisely the mechanisms for perineuronal net degradation (HMC) and complement-mediated pruning (CSC). Her broader research program includes TREM2 receptor mechanics and DAM/LDAM state transitions.
• Connects to: Homeostatic Microglial Collapse (DAM/LDAM states, complement elimination, MMP-2/MMP-9 -- perineuronal net degradation), Convergent Synaptic Collapse (complement-mediated pruning)
• Divergence type: Program undersell -- the submission is a broad aging-framework paper, but her research program is specifically about the microglial state transitions that are central to HMC.

#140 Gunter Hoglinger -- Mitochondrial Complex I / ATP Depletion — ✅ REVIEWED 2026-05-22 (bioenergetic/ONS_Hoglinger_ComplexI_Review.md)

• Submission said: Environmental toxins (annonacin, rotenone) as mitochondrial complex I inhibitors cause ATP depletion and tau redistribution.
• Scores: Relevancy 61.2, TKQ 50.0
• What his program actually contains: Hoglinger's work on mitochondrial complex I inhibition and ATP depletion is directly load-bearing for the Bioenergetic Collapse thesis -- specifically the OXPHOS dysfunction and metabolic reprogramming nodes. His demonstration that CxI inhibition causes dose-dependent tau pathology provides one of the clearest mechanistic links between bioenergetic failure and tauopathy. The low TKQ score (50) reflects that the submission framed this narrowly as an environmental toxin story, but the underlying mechanism (mitochondrial OXPHOS failure -> ATP depletion -> tau mislocalization) is a core Bioenergetic Collapse pathway.
• Connects to: Bioenergetic Collapse (OXPHOS dysfunction, mitochondrial quality control, metabolic reprogramming)
• Divergence type: Framing penalty -- environmental toxin framing obscured the generalizable bioenergetic mechanism.

TIER 2: MODERATE-PRIORITY BLINDSPOTS (Should be re-evaluated)

#141 Rajagopal Sekhar -- Glutathione / PINK1 / Impaired Autophagy

• Submission said: GSH deficiency causes energy metabolism failure, driving mitochondrial dysfunction, oxidative stress, impaired autophagy, senescence.
• Scores: Relevancy 53.7, TKQ 51.0
• What his program actually contains: Sekhar's key molecules list includes PGC1-alpha, PINK1, and LC3A/B -- three directly load-bearing molecules for Bioenergetic Collapse. PGC-1alpha is the master regulator of mitohormesis/AMPK signaling, PINK1 is the sentinel kinase for mitophagy, and LC3 is the autophagosome marker. His GlyNAC supplementation work shows correction of mitochondrial dysfunction, impaired autophagy, and cellular senescence simultaneously. The CSC scoring gave him near-zero on most nodes because the framework had no bioenergetic/autophagy node.
• Connects to: Bioenergetic Collapse (PINK1/Parkin mitophagy, PGC-1alpha, autophagy-lysosomal machinery, NAD+ via GSH metabolism)
• Divergence type: Framework gap -- correct mechanisms, no matching scoring node.

#95 Anne Eckert -- Mitochondrial OxPhos / Bioenergetic Shift

• Submission said: Menopause-driven bioenergetic shift compromises mitochondrial function and redox homeostasis, creating a vicious cycle with Abeta/tau pathology.
• Scores: Relevancy 53.0, TKQ 56.0
• What her program actually contains: Eckert's work on mitochondrial OxPhos dysfunction and the bioenergetic shift at menopause maps directly to the Bioenergetic Collapse thesis. Her focus on sex-specific vulnerability (estradiol loss compromising OxPhos) adds a critical dimension the trilogy currently underweights. The framework gap is the same as Sekhar and Hoglinger -- no bioenergetic node in CSC.
• Connects to: Bioenergetic Collapse (OXPHOS, metabolic reprogramming, mitochondrial quality control)
• Divergence type: Framework gap.

#93 Katarzyna Grochowska -- Endolysosomal / Autophagy / NLRP3

• Submission said: Toxic Abeta species, neuroinflammation, disruption of proteostasis in neurons and glia, NLRP3 inflammasome activation by internalized Abeta.
• Scores: Relevancy 57.3, TKQ 55.0
• What her profile contains: Her submission already mentions endolysosomal disruption, autophagic pathway impairment, and NLRP3 inflammasome activation. Key molecules include TREM2, SORL1, BIN1, and PICALM -- four major AD risk genes that converge on the autophagy-lysosomal pathway. This is a case where the scoring partially captured the mechanisms (endosomal-nexus: 8/10) but missed the new significance these have for Bioenergetic Collapse.
• Connects to: Bioenergetic Collapse (autophagy-lysosomal machinery, NLRP3), Homeostatic Microglial Collapse (TREM2)
• Divergence type: Partial capture -- well-scored on CSC endosomal node but newly relevant to Bioenergetic and HMC.

#131 Todd Golde -- PERK / Translational Regulation / Tau Toxicity

• Submission said: Broad defense of amyloid cascade with nuance about brain reserve and prodromal phase.
• Scores: Relevancy 56.3, TKQ 64.0
• What his program actually contains: Golde's recent work on the PERK-B/DLX1 axis -- where a PERK haplotype selectively upregulates DLX1 translation to promote tau toxicity -- connects to the metabolic stress response (PERK is an ER stress sensor in the unfolded protein response). This is adjacent to the Bioenergetic Collapse thesis's ER stress / autophagy failure pathway. His submission was scored as low-novelty (2/10) because it defended the standard amyloid cascade, but his lab's actual work is on translational regulation under cellular stress.
• Connects to: Bioenergetic Collapse (ER stress, UPR, metabolic reprogramming)
• Divergence type: Submission-program mismatch -- submitted a conventional amyloid argument, lab produces novel UPR/translational regulation work.

#121 Delphine Boche -- Vessel-Associated Microglia / ECM / Microglial Heterogeneity

• Submission said: Lipoprotein particle entrapment in aging ECM disrupts cholesterol delivery to neurons.
• Scores: Relevancy 64.5, TKQ 61.0
• What her program actually contains: Boche's research on vessel-associated microglia and microglial heterogeneity is directly relevant to HMC's border-associated macrophages and DAM state classification. Her ECM focus connects to the PNN axis. However, her submission was appropriately scored for CSC. The divergence here is moderate -- her ECM/cholesterol framing mapped well to ApoE4-hub, but the microglial heterogeneity dimension is more important for HMC than the scoring reflects.
• Connects to: Homeostatic Microglial Collapse (microglial heterogeneity, vessel-associated microglia as BAM analogue)
• Divergence type: Evolutionary relevance -- her work becomes more important as HMC thesis develops.

#127 + #84 Chini-Schwartz Connection: CD38 Immunometabolic Checkpoint

• Note: Michal Schwartz (#84) is already well-scored (neuroimmune-interface: 10/10), but her recent CD38 work (anti-CD38 treatment abrogating meningeal Th17 immunity and improving cognition) directly connects to Chini's CD38/NAD+ program. These two researchers form a CD38 immunometabolic axis that bridges HMC and Bioenergetic Collapse. Schwartz's score reflects her neuroimmune relevance, but the CD38-specific mechanism connecting to NAD+ decline is not captured.

TIER 3: LOWER-PRIORITY BUT WORTH NOTING

#167 Christopher Ramsden -- DAM Phenotype / O-GlcNAcylation / Autophagy

• Scores: Relevancy 57.8, TKQ 52.0
• His submission mentions DAM phenotype, autophagy, mitochondria, and NLRP3 in key molecules. Moderate overlap with Bioenergetic and HMC but submission is broadly framed.
• Connects to: HMC (DAM), Bioenergetic (autophagy, mitochondria)

#156 Bindu Paul -- Mitochondrial Dysfunction / Redox / H2S

• Scores: Relevancy 51.0, TKQ 48.0
• Her work on mitochondrial dysfunction and transsulfuration pathway connects to Bioenergetic Collapse's redox-metabolism nodes, but at lower specificity than Sekhar or Eckert.
• Connects to: Bioenergetic Collapse (mitochondrial dysfunction, oxidative stress)

#116 Li-Huei Tsai -- PV+ Interneurons / Gamma / Microglia / V-ATPase

• Scores: Relevancy 59.6, TKQ 49.0
• Already has a DeepResearch review. Her work is directly load-bearing for CSC (PV+ interneuron vulnerability, gamma oscillation disruption) and touches Bioenergetic (V-ATPase, bioenergetic resuscitation). The low TKQ (49) reflects the "therapeutic-rescue" classification, but her mechanistic findings on PV+ interneurons are more fundamental than that label suggests.
• Connects to: Convergent Synaptic Collapse (PV+ interneurons, gamma oscillations -- core mechanisms), Bioenergetic Collapse (V-ATPase)
• Status: Already reviewed but should be flagged for re-evaluation against CSC thesis.

---

3. Proposed Fix: Synthesis-Relevance Rescore System

3.1 Architecture

Add a fourth scoring layer to the pipeline:

Current Pipeline:
  1. CSC Framework Score -> relevancy_score
  2. Ten Key Questions -> tkq_composite
  3. PubMed Enrichment -> tags, recent_publications

Proposed Pipeline:
  1. CSC Framework Score -> relevancy_score
  2. Ten Key Questions -> tkq_composite
  3. PubMed Enrichment -> tags, recent_publications
  4. Synthesis-Relevance Rescore -> trilogy_relevance (NEW)

3.2 Synthesis-Relevance Rescore (SRR)

The SRR evaluates each entrant against the current state of the synthesis, not the original CSC framework. It consists of three components:

A. Mechanism Tag Audit (automated)

Cross-reference each entrant's tags and key molecules against a maintained registry of load-bearing variables for each thesis:

{
  "bioenergetic_collapse": {
    "tier1_mechanisms": [
      "autophagy", "autophagosome", "lysosomal-acidification", "v-ATPase",
      "PINK1", "Parkin", "mitophagy", "mitochondrial-quality-control",
      "NLRP3", "inflammasome", "mitochondrial-DAMPs",
      "PI3K-AKT-mTOR", "OXPHOS", "glycolytic-shift", "metabolic-reprogramming",
      "NAD+", "sirtuin", "SIRT1", "CD38",
      "AMPK", "PGC-1alpha", "mitohormesis",
      "TGF-beta", "SMAD"
    ],
    "tier2_molecules": [
      "BIN1", "ESCRT-III", "LC3", "Beclin-1", "ATG5", "ATG7",
      "mTORC1", "ULK1", "TFEB", "cathepsin",
      "Complex-I", "ATP-depletion", "ER-stress", "UPR", "PERK",
      "glutathione", "GSH", "GlyNAC"
    ]
  },
  "homeostatic_microglial_collapse": {
    "tier1_mechanisms": [
      "P2ry12", "Tmem119", "Sall1", "homeostatic-microglia",
      "DAM", "LDAM", "dystrophic-microglia", "microglial-senescence",
      "TREM2", "lipid-sensing",
      "C1q", "C3", "CR3", "complement-pruning",
      "MMP-2", "MMP-9", "ADAMTS", "cathepsin-S", "perineuronal-net",
      "Tyro3", "Axl", "Mer", "TAM-receptor",
      "border-associated-macrophages", "BAM"
    ],
    "tier2_molecules": [
      "CD38", "CD33", "SORL1", "PICALM",
      "vessel-associated-microglia", "microglial-heterogeneity",
      "Bmal1", "circadian-neuroinflammation",
      "inflamm-aging", "immunosenescence"
    ]
  },
  "convergent_synaptic_collapse": {
    "tier1_mechanisms": [
      "PV-interneuron", "parvalbumin", "fast-spiking-interneuron",
      "perineuronal-net", "aggrecan", "tenascin-R",
      "gamma-oscillation", "40Hz", "neural-synchrony",
      "excitation-inhibition-balance", "E-I-balance",
      "cognitive-resilience"
    ],
    "tier2_molecules": [
      "GAD67", "GABA", "synaptic-pruning",
      "hippocampal-hyperexcitability", "neural-entrainment"
    ]
  }
}

For each entrant, compute a mechanism overlap score per thesis:

• Count tier1 hits (weight: 3 points each)
• Count tier2 hits (weight: 1 point each)
• Normalize to 0-10 scale

B. Submission-Program Divergence Flag (manual/semi-automated)

For entrants where the PubMed enrichment reveals mechanisms not present in the original submission tags, flag the divergence:

divergence_flags:
  - type: "framework-gap"        # Submission was accurate but scoring framework lacked nodes
  - type: "framing-mismatch"     # Submission framing obscured relevant mechanisms  
  - type: "program-evolution"    # Researcher's program has evolved since submission
  - type: "partial-capture"      # Some mechanisms scored, new ones now relevant

Add to researcher profile frontmatter:

synthesis_relevance:
  bioenergetic_collapse: 8    # 0-10
  homeostatic_microglial_collapse: 3
  convergent_synaptic_collapse: 1
  divergence_flag: "framework-gap"
  divergence_note: "Submission accurately described CD38/NAD+ mechanism but CSC had no bioenergetic node"
  last_rescored: 2026-04-17

C. Periodic Rescore Trigger

The SRR should be re-run whenever:

1. A new thesis is added to the trilogy (or the trilogy becomes a quartet, etc.)
2. The load-bearing variable registry is updated
3. PubMed enrichment reveals new publications that change the relevance picture
4. A new convergence axis is integrated (e.g., the PNN axis)

3.3 Integration with Existing Scores

The SRR does NOT replace the CSC relevancy score or TKQ composite. Instead, it adds a trilogy_relevance field that can be used to:

1. Prioritize DeepResearch reviews -- entrants with high trilogy_relevance but no review get priority
2. Identify synthesis witnesses -- researchers whose work provides evidence for specific thesis claims
3. Flag re-evaluation candidates -- entrants whose trilogy_relevance exceeds their CSC relevancy by >20 points

The combined score could be:

effective_relevance = 0.4 * csc_relevancy + 0.3 * tkq_composite + 0.3 * trilogy_relevance

Or trilogy_relevance could remain a parallel track used for prioritization without modifying the historical scores.

3.4 Implementation Steps

1. Create the load-bearing variable registry as a JSON file at kb/tools/trilogy_mechanisms.json
2. Write a scoring script that cross-references entrant tags + key molecules against the registry
3. Add synthesis_relevance frontmatter to each researcher profile
4. Run the mechanism tag audit across all 150 entrants (automated)
5. Manually review the top ~20 flagged entrants for divergence classification
6. Generate a ranked priority list for DeepResearch reviews

---

4. Recommended Immediate Actions

Priority 1: DeepResearch Reviews (these entrants need reviews urgently)

1. #127 Eduardo Chini -- NAD+/CD38 program is directly load-bearing for Bioenergetic Collapse. No DeepResearch review exists.
2. #158 Beth Stevens -- Complement pruning, BAMs, SORL1-lysosomal function. No review exists. Her work is arguably the single most important external validation for HMC.
3. #139 David Gate -- NLRP3/complement/TREM2 nexus. No review. Mechanisms map precisely to HMC + Bioenergetic.
4. #70 Aleksandra Deczkowska -- DAM biology pioneer, MMP-2/9, complement pruning. No review. Core HMC mechanisms.
5. #140 Gunter Hoglinger -- Mitochondrial Complex I / ATP depletion / tauopathy link. No review. Core Bioenergetic mechanism.

Priority 2: Re-evaluate Existing Reviews

6. #116 Li-Huei Tsai -- ✅ RE-EVALUATED 2026-05-22 (convergent-synaptic/ONS_Tsai_PV_Gamma_Reevaluation.md). Re-classified from therapeutic-rescue to core-mechanism + therapeutic; +16.4 score gap.
7. #46 David Rubinsztein -- ✅ REVIEWED 2026-04-21 (bioenergetic/ONS_Rubinsztein_Autophagy_Review.md). The case study that triggered this audit. BIN1-ESCRT-III as sporadic genetic anchor of Convergence 1.

Priority 3: Create the Mechanism Registry

8. Build kb/tools/trilogy_mechanisms.json with the load-bearing variable registries defined above.
9. Run the automated mechanism tag audit across all 150 entrants to identify any additional blindspots not caught by this manual review.

Priority 4: Structural Fix

10. Add synthesis_relevance fields to the researcher profile schema.
11. Implement the SRR as a repeatable process tied to thesis evolution.

---

5. Pattern Summary

The blindspots cluster into three categories:

Pattern	Count	Examples
Framework gap -- submission was accurate but CSC lacked nodes to score it	4	Chini, Sekhar, Eckert, Hoglinger
Framing mismatch -- submission narrative obscured relevant mechanisms	3	Stevens, Gate, Golde
Evolutionary relevance -- work becomes more important as synthesis evolves	3	Deczkowska, Boche, Grochowska

The most dangerous pattern is framework gap: these researchers described exactly the right mechanisms, but the scoring system had no place to put the points. This is the pattern that the SRR is specifically designed to catch going forward.