Parkinson's Disease: The Competition Rule Test -- When Mechanism Quality Meets Institutional Endpoint Inertia

What this experiment was designed to test

Every previous M_MED experiment was either a healthy chain, a circular chain, or a new structural class. Parkinson's disease is different -- it is a competition rule test.

Two structural forces are in direct conflict:

Force 1: Endpoint Theory-Derivation (ETD) -- potentially high.

The dopamine depletion hypothesis is one of the best-confirmed mechanisms in clinical neurology:
• Ehringer and Hornykiewicz identified striatal dopamine deficiency from post-mortem brain tissue in 1960 -- BEFORE L-DOPA treatment existed (Cotzias demonstrated L-DOPA efficacy in 1967). The temporal sequence matters: theory preceded therapy, the opposite of the antidepressant and opioid cases.
• L-DOPA response is the strongest pharmacological validation of any neurological disease mechanism. Restore dopamine precursor, motor symptoms dramatically improve.
• DAT-SPECT imaging directly measures dopamine transporter on presynaptic nigrostriatal terminals -- measuring it IS measuring the mechanism.
• Four independent genetic lines (SNCA, LRRK2, PINK1, PRKN) all converge on nigrostriatal dopaminergic vulnerability, providing mechanism confirmation that does not rely on therapeutic response.

Force 2: Institutional endpoint inertia -- also high.

UPDRS (Unified Parkinson's Disease Rating Scale, 1987) is structurally analogous to PANSS in schizophrenia:
• A composite symptom scale developed by clinical committee, not derived from mechanism.
• Entrenched by FDA approval requirements (all PD drug approvals since the 1990s require UPDRS Part III).
• Specified as standard primary endpoint in MDS (Movement Disorder Society) guidelines.
• The specific thresholds, subscale weightings, and 0-4 scoring convention have no mechanistic derivation from the dopamine depletion hypothesis.

Schizophrenia (M_MED4) showed that PANSS institutional inertia produced r(justifies, selects_endpoints) = -0.13. Statins showed that strong mechanism produces r = +0.68. Parkinson's is the test case where both forces are present at near-equal strength.

Pre-registration

All four hypotheses committed before analysis. Hash: b24767c8. Timestamp: 2026-06-03T22:55:44Z. Records: github.com/vladi160/preregistrations, experiment M_MED8_v1.

Results: 3/4 confirmed, 1 partial

h2 CONFIRMED -- dopamine_depletion_hypothesis is rank #1 in the justifies layer (degree = 5, highest). It directly justifies the nigrostriatal pathway (the anatomical locus), DAT-SPECT (the mechanistic biomarker), the UPDRS motor domain (the clinical endpoint domain), and L-DOPA response (the therapeutic prediction). The patterns engine flagged it as the topological keystone of the entire graph: removing it may disconnect the network. This is the strongest founding hypothesis hub concentration in the M_MED series -- the mechanism is structurally load-bearing for the entire evidence chain.

h3 CONFIRMED -- dopamine_depletion_hypothesis is rank #1 in cites_as_support (degree = 3, tied with RCTs and treatment guidelines). Three independent citation lines (L-DOPA pharmacological response, genetic evidence, DAT-SPECT imaging) each independently support the hypothesis. This confirms high EM (Epistemic Monopoly): the founding dopamine hypothesis dominates both the justification layer and the citation layer -- the same structural position as the circular chains in antidepressants, opioids, and schizophrenia. EM = 0.92 (the rank-1 justifies hub is also rank-1 in cites).

h4 CONFIRMED -- dat_spect_biomarker is rank #3 in selects_endpoints (degree = 2). Both the mechanistic biomarker endpoint (DAT-SPECT, purely theory-derived: the mechanism specifies measuring dopamine transporter) and the institutional endpoint (UPDRS, rank #1 with degree = 3) are structurally present in the endpoint selection layer. ETD is contributing -- the mechanistic biomarker is not structurally absent despite institutional UPDRS dominance.

h1 PARTIAL -- r(justifies, selects_endpoints): Pearson = +0.276, Spearman = +0.076, p = 0.475.

Direction is confirmed. Magnitude exceeds the pre-registered 0.20 threshold. But the Pearson and Spearman values diverge sharply, in opposite directions from Alzheimer's disease (M_MED5). This divergence is the finding.

The competition rule answer: balanced, with a structural signature

Neither force wins cleanly. But they are distinguishable at different levels of analysis:

ETD wins at the magnitude level (Pearson r = +0.276). The dopamine hypothesis is genuinely prominent in both justifies (rank #1, degree = 5) and selects_endpoints (rank #2, degree = 2). In absolute degree terms, the mechanistic grounding contributes to endpoint selection. The theory is not structurally invisible.

Institutional inertia wins at the rank-order level (Spearman r = +0.076). UPDRS has the largest rank inversion in the M_MED series: rank #11 in justifies (degree = 1, only one theory edge) but rank #1 in selects_endpoints (degree = 3, two of three edges from FDA approval and treatment guidelines without theoretical content). The rank gap = 10. This inversion dominates the rank-order (Spearman) correlation, producing near-zero Spearman despite positive Pearson.

A new structural diagnostic: the Pearson-Spearman gap direction

Compare the two cases where Pearson and Spearman diverge:

| Case | Pearson | Spearman | Gap direction | Mechanism | |---|---|---|---|---| | Alzheimer's (M_MED5) | +0.09 | +0.43 | Spearman > Pearson | RPA (Rank-Preserving Amplification) | | Parkinson's (M_MED8) | +0.28 | +0.08 | Pearson > Spearman | MPI (Magnitude-Preserving Inversion) |

These are structurally opposite:

Institutional amplification (Alzheimer's): the endpoint has genuine rank-order theory coupling -- the amyloid cascade does specify amyloid biomarkers as the right thing to measure. But FDA accelerated approval and guideline embedding give the endpoint high absolute degree beyond what theory alone produces. Spearman captures the rank alignment; Pearson is diluted by the degree inflation.

Institutional rank inversion (Parkinson's): the theory contributes in absolute magnitudes -- the dopamine hypothesis is genuinely high-degree in both layers. But UPDRS acquires 2/3 of its selects_endpoints edges from institutional sources (FDA and MDS guidelines), achieving rank #1 despite weak theoretical grounding. Pearson captures the magnitude contribution; Spearman is disrupted by the rank inversion.

The direction of the Pearson-Spearman gap is a diagnostic of HOW institutional inertia operates:
• Amplification (Spearman > Pearson): institution increases endpoint degree while theory maintains rank alignment
• Inversion (Pearson > Spearman): institution elevates endpoint rank beyond its theoretical standing while theory maintains magnitude contribution

Why Parkinson's is not schizophrenia

Schizophrenia (M_MED4) also showed a severe PANSS rank inversion and produced r = -0.13. Parkinson's shows an even larger UPDRS rank inversion (gap = 10 vs gap ~ 7 for PANSS) but produces r = +0.276 rather than negative.

The difference is not the quality of the institutional endpoint. Both UPDRS and PANSS are committee-designed composite symptom scales without mechanistic threshold derivation. The difference is the mechanism quality: the dopamine depletion hypothesis has a degree = 5 in the justifies layer (4 direct out-edges), providing enough magnitude to counterbalance the UPDRS inversion. The D2 blockade hypothesis in schizophrenia was weaker in the justification layer, allowing the PANSS inversion to push the correlation negative.

Mechanism quality shifts the correlation from negative to positive while institutional inertia prevents it from reaching healthy values. The difference between Parkinson's and schizophrenia is not the endpoint -- it is the mechanism.

What graph structure reveals about Parkinson's vs schizophrenia

graph_connected = True. Alpha-synuclein pathology (Lewy body staging, Braak 2003) creates a justification bridge to the dopamine hypothesis: alpha-synuclein aggregation in SNc neurons at Braak stage 3-4 explains WHY dopaminergic neurons die. This is the same integrated pluralism structure as Alzheimer's disease: one framework predicts the other as its cellular mechanism rather than competing with it.

Schizophrenia (M_MED4) showed the opposite -- dopamine and NMDA hypotheses had zero structural contact across all three layers. Alzheimer's and Parkinson's both show integration rather than isolation, suggesting that neurodegeneration diseases may be structurally more integrated than psychiatric conditions in the M_MED series.

Two new citation shadows

The M_MED8 output contains the two largest citation shadows found so far:

Nigrostriatal pathway (rank #3 in justifies, rank #11 in cites, gap = 8). The specific anatomical pathway that the mechanism specifies -- the dopaminergic projection from substantia nigra to striatum -- is mechanistically central but virtually absent from clinical citations. Clinicians cite the hypothesis and the trials; they do not cite the pathway anatomy.

DAT-SPECT biomarker (rank #4 in justifies, rank #9 in cites, gap = 5). The most mechanistically grounded endpoint in the evidence chain -- measuring dopamine transporter IS measuring the mechanism -- accumulates almost no clinical citations despite its theoretical prominence.

A pattern is emerging across citation shadows in the M_MED series: specific mechanistic elements (pathways, biomarkers, genetic variants) consistently underperform in citations relative to their justification prominence. Clinical citations go to the hypothesis and to trial results -- not to the specific mechanistic details that make the hypothesis scientifically credible. This may reflect how clinical evidence chains are actually used: practitioners cite what they need for treatment decisions, not what grounds the mechanism.

A new structural observable: r(selects_endpoints, cites_as_support)

The Parkinson's output reveals: r(selects_endpoints, cites_as_support) = -0.302.

Endpoint selection and citation accumulation are anti-correlated. UPDRS (rank #1 in selects, degree = 3) has degree = 0 in citations -- it is never cited as support for anything. The major clinical trials (rank #2 in citations) have degree = 0 in endpoint selection.

This negative r(se, cites) is actually a partially healthy structural signal despite the high EM: the measurement layer is not part of the citation loop. In fully circular chains, we would expect r(se, cites) to be positive -- the same founding node would dominate all three layers including endpoint selection. In Parkinson's, the measurement layer (UPDRS) is disconnected from the citation accumulation dynamics (which flow to the hypothesis and trials).

This is the structural reason Parkinson's does not become a full circular chain despite EM = 0.92: the institutional endpoint prevents the justification-citation loop from also capturing the endpoint selection layer.

A structural element absent from circular chains: neuroprotection failure

The neuroprotection_failure node -- encoding multiple failed neuroprotection trials (CoQ10, creatine, ADAGIO ambiguity) -- is present in both the justifies and cites_as_support layers pointing to treatment guidelines. Negative empirical evidence explicitly shapes the institutional output: MDS guidelines state that no neuroprotective therapy has proven efficacy, constraining the guideline scope to symptomatic treatment only.

This structure is absent from every M_MED circular chain examined. In antidepressants, opioids, schizophrenia, and Alzheimer's, negative trials did not appear as structural moderators on the evidence chain. Their absence in circular chains is itself a structural finding: circular evidence chains may systematically exclude or absorb falsifying evidence, while chains with stronger mechanistic grounding incorporate it.

The complete eight-experiment comparative survey

| Chain | Pearson | Spearman | EM | ETD | MSC | Structural class | |---|---|---|---|---|---|---| | M_MED2 vaccines | +0.75 | +0.81 | Low | High | Low | Healthy (ETD-dominated) | | M_MED6 statins | +0.68 | +0.76 | Low | High | Low | Healthy (ETD-dominated) | | M_MED7 epilepsy | +0.49 | +0.59 | Med | Low | High | Multi-framework convergent | | M_MED1 antidepressants | +0.41 | -- | High | Med | Low | Mild circularity | | M_MED8 Parkinson's | +0.28 | +0.08 | High | Low-med | Low | MPI (Institutional rank inversion) | | M_MED3 opioids | +0.14 | -- | High | Low | Low | Strong circularity | | M_MED5 Alzheimer's | +0.09 | +0.43 | High | Med-low | Low | RPA (Institutional amplification) | | M_MED4 schizophrenia | -0.13 | -- | High | Neg | Low | Measurement-layer dominance |

*p < 0.05

What IRDME cannot say

This analysis reads graph structure, not scientific content. It cannot:
• Evaluate whether dopamine replacement therapy is the correct treatment for Parkinson's disease
• Assess whether UPDRS or DAT-SPECT is the better primary endpoint for PD trials
• Determine whether alpha-synuclein-targeting therapies will succeed where dopamine replacement has not
• Make any recommendation about individual patient care

What it can say: the Parkinson's disease evidence chain has high epistemic monopoly (the founding hypothesis is the topological keystone), partial endpoint theory-derivation (the mechanistic biomarker is structurally present but institutionally secondary to UPDRS), and a structural boundary condition for circular collapse: the negative r(selects_endpoints, cites_as_support) prevents the measurement layer from being absorbed into the citation loop despite the founding hypothesis dominating both justification and citation layers.