Treating the Base, Not the Top: A Friction-Theoretic Hypothesis for Cofactor-Support Below the Diagnostic Threshold

Paper 8D · Pødenphant Lund, T. (2026u) · Hypothesis paper · Live on Zenodo

A clinical diagnosis is an outcome-cluster: a shared symptom-picture similar enough to receive a common label. The substrate that produces it (the configuration of genes, cofactor-status, and enzyme-capacity) is the more fundamental level, and it is not identical with the diagnosis. If that is right, a large population carries a vulnerable substrate without ever crossing a diagnostic threshold. This paper asks what, if anything, can be done for that sub-threshold population, and proposes one falsifiable answer: cofactor-support matched to a measured substrate-profile, tested as a preventive intervention class. It is a hypothesis and an invitation to test it. It is not clinical advice, and it recommends nothing at the individual level.

DOI (concept)	10.5281/zenodo.20277290
Type	Hypothesis paper (theoretical, with framework-prediction synthesis and falsification criteria)
Cite-letter	2026u
Worked exemplar	One-carbon / folate–B12–MTHFR cycle
Author	Tomas Pødenphant Lund [ORCID]

The construct: outcome-cluster versus route-identity

The framework-term substrate here means the configuration of genes (heritable parameter-settings), cofactor-status (a modifiable, rate-limiting node-occupancy), enzyme-capacity, and the history of past race-resolutions. The substrate is what computes; the behaviour, the symptom-picture, and the diagnosis are the computation's output-clusters. Diagnostic systems cluster on the outcome, not on the route by which an individual arrived; multiple substrate-routes converge on overlapping outcomes, so the inverse mapping (outcome back to route) is ill-posed at the individual level. The paper positions this as a layer on top of, not a rival to, the Research Domain Criteria programme (RDoC): what it adds is narrow, the route-identity claim plus the rate-limiting-cofactor reading that follows from it.

The mechanism: rate-limiting nodes and the prevention-privilege

Where the bottleneck sits. Catecholamine synthesis is rate-limited at tyrosine hydroxylase, which needs BH4, iron, and oxygen; downstream steps need vitamin B6 (as PLP), copper, and ascorbate. BH4 is a four-pathway hub (it also serves serotonin synthesis, phenylalanine conversion, and nitric-oxide synthase), so a shortfall strikes several transmitter systems at once. The illustration the title points at: a stimulant that blocks reuptake raises the synaptic concentration of whatever transmitter was produced, so if production is itself cofactor-rate-limited, reuptake-inhibition acts downstream of the bottleneck. The chemistry claim is deliberately modest: the one-carbon axis touches catecholamine synthesis only at a secondary BH4-salvage node and, more robustly, through SAMe-dependent methylation, enough to make MTHFR a plausibly broad-reach variant without claiming folate drives BH4 recycling.

Why the one-carbon axis carries the weight. The folate–B12–B6–methionine–SAMe cycle is rate-limiting for two distinct processes at once: neurotransmitter synthesis, and (because SAMe is the universal methyl-donor) epigenetic trace-deposition. The same cofactor-intervention can therefore act at two levels, the framework's reading of why cofactor-response is heterogeneous in clinical populations. The key variants, MTHFR C677T and A1298C, are common and carry documented gene × folate-status interactions, but the published meta-analyses stratify by ethnicity and age rather than by measured folate-status, so the interaction itself remains a prediction to test, not an established result.

The prevention-privilege. The framework's architectural constraint is that one can add to the substrate but not cleanly subtract: trace-deposition runs at rate α, trace-shallowing at rate γ, with α much greater than γ. This is a falsifiable kinetic prediction, not a no-erase law. The paper grants that biology has genuine degradative operations (autophagy, receptor endocytosis, synaptic pruning, demethylation) but argues each is evoked by an added effector and runs on slower kinetics than acquisition, consistent with the documented acquisition-versus-extinction asymmetry. The consequence: prevention operates before trace-deposition, where additive intervention (cofactor-support among them) is architecturally possible, whereas cure must compete against deposited trace that cannot be subtracted. That makes prevention mechanically privileged, not merely gentler, a conditional about architectural favourability rather than an efficacy result.

The sub-threshold population, empirically

The population is defined negatively by the diagnostic system: present in every sample, invisible to recruitment that filters on diagnosis. Per-condition meta-analytic prevalences make the scale concrete, and in every condition examined the sub-threshold band is larger than the diagnosed one.

• Sub-threshold depression — 11.02% prevalence (95% CI 9.78–12.33; n = 1,129,969 across 113 studies; Zhang et al. 2023), roughly three times the point-prevalence of current major depression; 17.62% convert to MDD over follow-up versus 6.08% of asymptomatic controls (IRR 2.95).
• Other conditions — sub-threshold childhood ADHD 17.7% (Ogundele & Morton 2025); broader autism phenotype 14–23% of parents of autistic children (Sasson et al. 2013; Ruzich et al. 2015); sub-threshold OCD 8.3% versus 0.7% diagnosed (Adam et al. 2012); psychotic experiences 7.2% lifetime, 3.5× conversion with persistence (Linscott & van Os 2013); bipolar-spectrum 5.1% (Merikangas et al. 2007).

Genetic-substrate carriage tells the same story: schizophrenia top-decile polygenic-risk absolute risk is about 2% against a 1% base-rate, so most high-risk carriers do not carry the diagnosis, and monozygotic concordance for ADHD is well below unity despite about 74% heritability (Faraone & Larsson 2019). The paper flags openly that these prevalence figures aggregate across heterogeneous operational definitions of "sub-threshold", so the unification is framework-layered, not a claim that the sources share a definition.

Sub-threshold is not healthy. Substrate-markers track the trajectory: the Rotterdam Study related hyperhomocysteinemia, B12 deficiency, and folate deficiency to depression including sub-threshold presentations (Tiemeier et al. 2002); Whitehall II linked baseline CRP and IL-6 to the cognitive symptoms of depression over 12 years (Gimeno et al. 2009). The cost of leaving it untreated is not stasis: the framework reads it as daily race-resolution depositing hysteresis on whichever compensating route circumstances make reinforcement-available, producing delayed-onset comorbidity (the ADHD-to-substance-use trajectory being the clearest instance) that the genetic-co-heritability account alone does not fully explain.

The methodology: the cofactor-map, run backwards

Paper 8's cofactor-mapping methodology starts from a clinical state and works forward to cofactor candidates. This paper runs the same chain backwards: the entry-point is the measured substrate-profile, not a clinical state. Same biochemistry, same cofactor-chains, different starting point. The five steps: (1) measure the substrate-profile by blood-panel plus genetic test; (2) map each measured marker onto the biochemical chains; (3) identify rate-limiting nodes (a variant alone is not a bottleneck; a variant plus a low measured cofactor at that node is); (4) stratify candidate support by implementation-friction, always as candidates to test rather than recommendations; (5) iterate against re-measured markers, not against symptom-impression. The discipline-critical step is the first one: profiling is by measurement, never inferred from symptoms or family history, and compound profiles with contraindications (the worked illustration is a postmenopausal woman whose breast-cancer history closes the hormonal route) are explicitly medical, with the decision belonging to the treating clinician.

Predictions and falsification criteria

The paper specifies nine predictions, each with a falsification criterion. P1 and P5 are pivotal: combined failure would require abandoning the core thesis. Three (P5, P7, P8) are framework-distinctive, generated by this framework specifically rather than by competing biomarker-stratified-intervention or genetic-co-heritability models. Several are testable on existing public-access cohort data without new collection.

• P1 (pivotal) — carrier × cofactor-status interaction. Among risk-variant carriers, those with sufficient measured cofactor-status show a sub-threshold trait-profile; those with insufficient status show the manifest phenotype. The effect is an interaction, not a main-effect of either alone. Pre-specified primary test: MTHFR C677T TT-versus-CC × measured baseline folate on incident depression, in a cohort that actually assayed one-carbon status (the Rotterdam Study), since UK Biobank did not assay folate, B12, or homocysteine. Falsified if manifestation is predicted by variant-load alone with no cofactor-status moderation.
• P5 (pivotal, framework-distinctive, a negative prediction) — cofactor-support is null where there is no bottleneck. Framed as forward-looking to keep it falsifiable: if the MoodFOOD trial's existing data (Bot et al. 2019, unstratified OR 1.06, non-significant) were re-analysed stratified by MTHFR-genotype and baseline folate/B12/homocysteine, the rate-limited subgroup is predicted to show a positive effect the unstratified analysis dilutes toward null. Committed to a single pre-named dataset; falsified by one adequately-powered result whose genotype × folate interaction CI includes no-interaction, with no transfer-on-null and no "needs more replications" retreat.
• P7 (framework-distinctive) — Category-B durability after withdrawal. After sustained profile-matched support, planned withdrawal produces a durability-profile that overshoots pharmacokinetic clearance in a sub-population, signalling consolidation of a new substrate-state rather than continuous-maintenance perturbation. Falsified if functional decay is fully predicted by marker decay across the population (Category C only).
• P8 (framework-distinctive) — untreated sub-threshold load accumulates hysteresis on compensating routes; treatment-duration and treatment-onset-age moderate downstream comorbidity-risk. Made environment-stratified and risky on each stratum (substance-route in high-availability environments; over-control/burnout in high-control low-substance environments), with strata assigned a priori from population-level indices. Falsified if treatment-duration shows no moderation after controlling for polygenic-risk loading and environment.
• P2, P3, P4, P6, P9 (sharpening, not pivotal). Family-divergence tracks cofactor-status not only variant-load (P2); substrate-markers detect the vulnerability before the threshold is crossed (P3); cofactor-support effect-size is stratified by rate-limiting status, a who-benefits hypothesis testable by re-analysing existing trials (P4); profile-matched support in identified sub-threshold individuals improves markers and function versus controls (P6); and behavioural classification carries lower-information but non-zero evidence about the substrate-vector, with direct measurement dominating individual-level decisions (P9). Each carries its own falsification criterion.

What this paper is, and is not

What it is: a hypothesis and an invitation to test it; a re-indexing of an existing cofactor-mapping methodology to start from measured substrate-markers; a synthesis of established meta-analytic and cohort evidence about a real, large, under-sampled population; and a set of nine falsifiable predictions, several runnable on existing cohorts (Rotterdam Study, Whitehall II, UK Biobank, ABCD, MIDUS, Swedish national-register data) without new collection.

What it is not: it is not a diagnostic manual, and it does not tell any individual what to do. It makes no patient-level recommendations, specifies no doses, and names no products. It does not claim cofactor-support treats manifest, diagnosed conditions (that scope belongs to Paper 8); its claims are preventive and sub-threshold. It is not a call to diagnose more people; the argument is for decoupling substrate-support from the diagnostic gate, not for expanding diagnostic categories. The measurement-discipline is also a safety-discipline: the paper names concrete hazards (folate masking B12 deficiency while neurological damage progresses; iron in an iron-replete person; supplement–drug interactions; over-supplementation toxicity; mood-switch risk in bipolar-spectrum individuals; equity and genetic-data governance), each a reason the testing must be clinician-mediated and panel-based rather than a reason against testing.

Limitations (stated plainly in the manuscript)

The biological claims here are predictions, not proven facts. The reliability of the paper's claims is layered explicitly: the biochemical chains are established textbook material; the gene × cofactor associations are real published findings but with modest effect-sizes and substantial heterogeneity, so they are candidate factors to test rather than settled causal chains for an individual; and the framework-level claims (the substrate as the fundamental level, the additive-only asymmetry, the mechanical prevention-privilege) are hypothesis, which is what the paper invites adjudication of. The pivotal cofactor-stratified tests are constrained by data: the largest genotyped biobank did not assay one-carbon cofactors, while the cohorts that did are smaller, so the gene × cofactor interaction may need consortium-scale pooling to reach power. No new data is collected here; the paper unifies existing findings and specifies tests. And the author is an independent researcher without institutional cohort-access or clinical credentials, and without a clinical collaborator currently in place; the empirical realisation of the predictions therefore requires partnership not yet established, which the manuscript states as a structural limitation rather than presenting the programme as resourced. The mathematics of the rate-limited-cascade reading is a deliberate simplification of a multi-node biochemistry.

Connections to other papers in the series

• Paper 1 (Friction Theory) — the substrate-universal framework: RACE architecture, commit-pressure, and the substrate whose route-resolution this paper reads at the biological level.
• Paper 8 (Pressure, Hysteresis, and Experience) — the diagnosis-indexed clinical-intervention paper. It supplies the outcome-cluster-versus-route-identity distinction (§6.3), the cofactor-mapping methodology this paper runs backwards (§5.1), and the additive-only prevention asymmetry (§4.1b). The scope-boundary is explicit: Paper 8 is diagnosis-indexed treatment, Paper 8D is diagnosis-independent sub-threshold support.
• Paper 8B (Compound Race Pathology) — the multi-scale framework for treatment-resistant conditions; the compound-profile reasoning here is the sub-threshold counterpart.
• Paper 8C (Translational Research-Program) — the trial-design companion; its biomarker-panel template can operationalise this paper's prospective preventive prediction (P6).
• Learning — the learning-and-memory thread, where the hysteresis and consolidation machinery this paper borrows (the prevention-privilege, the Category-B durability prediction) is developed.

Read the paper

The full paper is on Zenodo (concept DOI 10.5281/zenodo.20277290):

Read on Zenodo → · Plain English version · Dansk version

This page summarises a hypothesis paper. It is not medical advice, not a diagnostic manual, and does not tell any individual what to do. The biological claims are predictions to be tested, not proven facts. Any decision about supplements, testing, or treatment belongs with a qualified clinician.