From Hypothesis to Panel
The Anatomy of Autism (AofA) lays out the Immune-Derived Autism (IDA) framework as a set of testable predictions — specific, falsifiable claims about what should be measurable in a child's blood, urine, or platelets if the model is correct. The testing protocol below is the direct operational extension of that work: a panel built to test those predictions in practice, organized around the same six biological axes the framework identifies as independently contributing to the IDA cascade.
The organizing principle throughout is the same one that runs through AofA itself: autism is not one biological event, and a single marker rarely tells the whole story. Different children may be driven by different combinations of these six systems — which is why the panel is built in axes, not as one pass/fail test.
Six Axes, Twenty-One Markers
Table 1 summarizes the panel. Each marker's evidence tier follows AofA's own four-level grading (Level 1: multiple independent human ASD cohorts; Level 2: established molecular biology with ASD application; Level 3: in vitro/indirect evidence; Level 4: mechanistically plausible, not yet prospectively tested), plus two additional statuses for markers not yet at Level 1: testable prediction (specific, model-generated, not yet run) and hypothesis-stage (proposed but not yet validated in this population).
| Axis | What it looks for | Sample | Representative markers |
|---|---|---|---|
| Immune / Kynurenine | Immune activation driving a specific pathway (IDO1/kynurenine) known to silence a key inhibitory brain signal | Blood | K:T ratio, cytokine panel, autoantibody panel, QUIN/KYNA ratio (Level 1) |
| Mitochondrial / Energetic | Whether cellular energy production is under strain, compounding immune effects | Blood | Lactate:pyruvate ratio, plasma NAD⁺ (Level 2–3) |
| Methylation / Catecholamine | Gut-derived compounds interfering with dopamine/norepinephrine production, and the body's capacity to clear them | Blood, urine | SAM/SAH ratio, p-cresol, p-cresyl sulfate, PST-P activity (Level 1); free p-cresol:PCS ratio (testable prediction) |
| Astrocyte / Neurotrophic | Whether brain support cells have shifted into a reactive, less protective state | Blood, CSF | BDNF, hevin/SPARCL1 expression, A1 astrocyte markers (Level 2) |
| Gut Hormone / Appetite | Whether distinct appetite patterns reflect two different underlying immune subtypes | Blood | Whole-blood serotonin, fasting acyl ghrelin, fasting GLP-1 (hypothesis-stage) |
| GI Enzymatic | Whether a specific enzyme deficiency is allowing food-derived peptides to build up and act on the brain | Blood | DPP-IV activity (Level 1) |
Table 1. The IDA testing panel across six biological axes. Evidence tiers follow AofA's four-level grading (see AofA, Limitations and Scope). K:T, kynurenine-to-tryptophan ratio; PCS, p-cresyl sulfate; PST-P, phenol sulfotransferase (phenol-sulfating isoform); QUIN/KYNA, quinolinic acid/kynurenic acid ratio.
Each axis identifies a distinct, independently actionable finding — a child can test positive on one axis and not another, and the model predicts that different combinations call for different treatment emphasis, not a single uniform protocol.
A Worked Example: Why "Elevated" Isn't the End of the Question
The Methylation/Catecholamine axis illustrates a principle that applies across the whole panel, and it's worth walking through in detail because it's the clearest case where a single number can actively mislead.
p-Cresol is produced when gut bacteria ferment the amino acid tyrosine that escaped digestion. The body converts most of it into p-cresyl sulfate (PCS) for excretion — in fact, free p-cresol is barely measurable in urine at all; the detectable pool is almost entirely the converted form. Both forms reach the brain and directly interfere with the enzymes that produce dopamine and norepinephrine.
This is not an isolated case. The same pepsin-failure mechanism that lets undigested tyrosine reach the colon also lets undigested phenylalanine reach it — and gut bacteria degrade each along its own path: tyrosine to p-cresol and p-cresyl sulfate, phenylalanine to phenylacetic acid and phenylacetylglutamine. AofA's precursor-starvation mechanism already names both amino acids as jointly depleted; the metabolite side of the picture follows the same shared substrate. Phenylalanine-derived metabolites are not yet part of the formal panel above, but the same upstream/downstream question the p-cresol work raises should apply to them as well, and we expect to extend the Methylation/Catecholamine axis to include them as that work develops.
Here is the part a single "elevated p-cresol" reading cannot resolve:
| Question | Single elevated reading | Full panel (ratio + PST-P) |
|---|---|---|
| Is gut-derived catecholamine risk present? | Yes | Yes |
| Is the cause too much production (substrate/bacterial), or too little clearance? | Cannot tell | Yes — free p-cresol:PCS ratio |
| Is the body's own clearance capacity (PST-P enzyme) working? | Cannot tell | Yes — direct platelet assay |
| Does this map to an actual catecholamine synthesis problem? | Cannot tell | Yes — paired with DBH activity |
Table 2. What a single elevated reading can and cannot resolve, versus the full panel. Phenylalanine's parallel bacterial metabolites — phenylacetic acid and phenylacetylglutamine — raise the same upstream/downstream question and are candidates for future inclusion in this axis.
A child whose p-cresol is elevated because of excess bacterial production, and a child whose p-cresol is elevated because their sulfation capacity can't keep up, will show the identical single number — but they need different treatment. The first points toward diet and gut microbiome composition. The second points toward supporting clearance capacity directly; treating the gut in that child addresses nothing, because the gut was never the bottleneck.
Other Work in This Space
Multi-metabolite urinary screening for autism has advanced meaningfully in the past year. Flynn, Adams, and colleagues at Arizona State University have published both a comprehensive review of p-cresol's effects in autism and a multi-site study proposing a diagnostic approach — the "MDM System™" — that classifies children by counting how many of over twenty gut-derived metabolites exceed a healthy reference range, reporting strong sensitivity and specificity in their cohort. Their own data, notably, found phenylacetylglutamine among the metabolites with the strongest separation between ASD and typically developing children — independent confirmation that this second amino acid pathway matters too. This is genuine, valuable progress toward objective biomarkers in a field that has relied entirely on behavioral diagnosis.
That work and the panel described here were developed independently and answer different questions. The MDM System™ is built to detect whether elevated gut-derived metabolite burden is present. It does not distinguish, for any individual elevated metabolite, whether that elevation reflects excess production or impaired clearance — free p-cresol and p-cresyl sulfate are counted as separate, independent markers rather than interpreted as a ratio.
That distinction matters most at the point of treatment, and it points toward what a complete answer likely looks like: not one intervention, but two prongs matched to two distinct mechanisms. The upstream prong — substrate- or bacteria-driven elevation — already has candidate interventions to build on, including microbiome-directed approaches and a hypothesized inflammation-reduction pathway (see Intervention Logic for current status). The downstream prong — clearance-driven elevation — is where the science is still ahead of the treatment: PST-P activity can now be measured directly, genetic sequencing has already ruled out a coding-variant explanation, and two independent threads point toward why activity might still be reduced. One traces to undermethylation reducing the transsulfuration pathway that supplies sulfate itself — mechanistically plausible from existing human ASD findings, not yet directly tested. The other is a documented, dose-dependent link between PST-P activity and levels of TMAO, a molecule depleted in ASD via hyponatremia/hypoosmolarity; correcting that with urea (a salt-sparing diuretic) restored TMAO levels, biochemical markers, and behavior in a rat model (Launay & Vodovar, iScience, 2026) — not yet tested in humans or specifically for p-cresol clearance. Whether that TMAO/PST-P link runs through folding, stability, or direct catalytic effect isn't established, just that the dependency itself is real. Both threads trace back to the same adenosine-accumulation node active elsewhere in this framework, raising the possibility that downstream and upstream aren't as separate as they first appear — but that possibility is untested. Right now, only the upstream prong has a candidate intervention built out — which is itself a useful, testable way to frame where the field needs to go next.
The Principle Across All Six Axes
This article describes a hypothesis-generating diagnostic framework. Several markers above are Level 1–2 evidence from independent human cohort studies; others are testable predictions or hypothesis-stage proposals not yet validated in this population, as indicated throughout. Educational content — not a substitute for individualized medical advice. Full mechanistic basis, evidence-level rationale, and citations: The Anatomy of Autism.