Skip to content

H08 — Gut-Lumen Sink Platform Thesis (Stub)

Stub status. Committed at stub-level on 2026-05-15 to register the platform's load-bearing scientific claim in the falsification-card directory and force the "what would kill this thesis" framing onto the canonical gut-lumen-sink.md and cross-validation.md Claim 1. Full population (assumption stack, killshot menu, pre-committed thresholds, kill switches, failure-mode coverage map) is queued as Phase 2 — see "Open Follow-Ups" at the bottom of this file.

Identified as the platform's #1 riskiest assumption in the 2026-05-09 sweep (synthesis/done/2026-05-09-riskiest-assumption-1) with Pass 3 verdict "Confirmed, prioritize." The pre-registration note on H01 (H01-ward-dual-cassette.md §Pre-registration) does not apply until this stub is upgraded to a full card. When the upgrade happens, the upgraded version is what gets pre-registered; the stub is informational scaffolding only.

Claim (provisional, stub-level)

Placing active uricase in the intestinal lumen of typical (non-CKD) gout patients produces a sustained serum urate reduction of −0.5 to −1.0 mg/dL at the comp-019 mid-dose (25 mg/day delivered functional uricase), with magnitudes stratified by ABCG2 genotype as predicted by the flux model:

Scenario comp-019 Predicted ΔSUA at 25 mg/day (Monte Carlo median, 90% CI)
WT/WT, male gout −0.83 mg/dL (−1.13 to −0.57)
Q141K heterozygous, male gout −0.67 mg/dL (−0.91 to −0.45)
Q141K homozygous, male gout −0.50 mg/dL (−0.68 to −0.34)
Severe ABCG2 dysfunction (Q126*+Q141K compound) −0.28 mg/dL (−0.39 to −0.20)

The claim has multiple sub-components that the full card will decompose:

  1. The in-silico flux model magnitudes translate from Monte Carlo prediction to human RCT outcomes within a band that distinguishes clinical efficacy from placebo (mean ΔSUA ≥ 0.5 mg/dL with p<0.05 in a powered RCT).
  2. ABCG2-mediated intestinal urate secretion is rate-limiting enough in non-CKD patients that adding luminal uricase materially changes the serum-urate equilibrium (vs. renal compensation fully offsetting the gut delta).
  3. ALLN-346 Study 202's null result in the broader cohort reflects fixable formulation / dose / cohort-selection problems, not a fundamental ceiling on the gut-lumen sink mechanism in typical gout.
  4. The OE engineered strain achieves GI-survival sufficient to deliver therapeutic luminal uricase to the secretion-rich ileum/colon at therapeutic densities (Claim 2 binding constraint at 5/10 in cross-validation.md).
  5. The mechanism is genotype-robust per comp-019 — non-Q141K patients show the largest per-patient response, not the smallest — so the platform demographic does not narrow to Q141K carriers.

Why this is the riskiest assumption

The biological logic is sound: ABCG2 secretes ~33% of daily urate elimination into the gut lumen, and a luminal uricase sink amplifies whatever ABCG2 delivers. comp-019's flux model is internally consistent (substrate-limited regime at all dose scenarios tested, capacity ratios 32–1300×, anchored against Miyazaki 2025's direct human in-vivo jejunal urate secretion measurements).

But the clinical-translation link — from modelled intestinal urate flux to a measured, patient-meaningful SUA change in typical gout — is entirely unvalidated in vivo:

  • ALLN-346 Phase 2a Study 201 showed signal in CKD patients.
  • ALLN-346 Phase 2a Study 202 (broader cohort, ~typical gout) showed 0–5% SUA reduction, no significance vs. placebo, and the program terminated with 19/200 patients enrolled.
  • Zero uricase trials (oral or systemic — ALLN-346, PRX-115, rasburicase, pegloticase) have stratified results by ABCG2 Q141K genotype. The Q141K × allopurinol response literature is rich; the Q141K × uricase response literature is empty.
  • The comp-019 flux model is prospective and unvalidated.

The cross-validation feasibility rating is 6.5/10 (cross-validation.md §Claim 1): mechanism validated preclinical, clinical signal exists but is marginal and inconsistent, the 3.5 remaining points await Phase 2b RCT validation. If the real-world typical-gout effect is <0.5 mg/dL, the platform's urate-lowering value proposition collapses from "core mechanism" to "mild adjunct" — reshaping the commercial framing, the self-experiment framing, and the priority of downstream peer tracks (LBP chassis, siRNA / URAT1, medicinal-mushroom complement track, compounding pharmacy delivery where it exists).

Assumption Stack (placeholder — to be populated in Phase 2)

Anticipated load-bearing assumptions, to be confirmed and weighted in the full card:

  1. comp-019's Monte Carlo flux model predicts human RCT outcomes within a band of ±0.3 mg/dL of the mean (i.e., the modelled −0.83 mg/dL WT/WT prediction translates to a measured −0.5 to −1.1 mg/dL human RCT mean).
  2. Renal compensation in patients with normal eGFR does not fully offset the gut delta. comp-019's renal-compensation parameter sensitivity spans 0–50%; this card commits the assumption that ≤30% offset preserves clinical efficacy.
  3. ALLN-346 Study 202's null result is explained by either (a) insufficient effective dose at the secretion-rich ileum/colon, (b) insufficient cohort enrichment for likely responders, or © insufficient duration to reach steady-state — NOT by a fundamental mechanism failure.
  4. Comparator candidates currently in clinical development (PRX-115, any post-ALLN-346 oral-uricase program) do not preempt or invalidate the OE-specific delivery thesis.
  5. The OE strain achieves >10⁸ functional uricase units delivered post-stomach to the ileum/colon at a per-dose cost compatible with typical-gout commercial pricing (or, for the open-source/self-experiment surface, compatible with the koji-strain-library distribution model).
  6. Q141K trafficking rescue by ABCG2 modulators (butyrate via HDAC, see abcg2-modulators.md) does not invalidate the H08 mechanism — it complements it.

Killshot Menu (placeholder — to be populated in Phase 2)

The full menu will follow H01's score = (kill_pr × info_weight) / (cost × time_penalty) ranking, with each killshot tagged to specific assumptions and failure modes per linter-design.md §4–5.

Anticipated highest-priority killshots, ranked roughly by leverage:

  • Lit scan: post-ALLN-346 oral-uricase Phase 2 typical-gout efficacy. Has any program reported a typical-gout Phase 2 readout since ALLN-346 termination? If yes, that's a strong external prior on H08. Cheapest possible upstream move.
  • Re-analysis of ALLN-346 Study 202 publicly available cohort-level data. No published uricase trial has stratified by ABCG2 Q141K. A retrospective genotype-stratified re-analysis of Study 202 (if obtainable via FOIA / sponsor request / published supplementary data) would be a low-cost, high-info killshot — does the WT/WT subset show the largest effect, as comp-019 predicts? If the genotype-stratified pattern matches comp-019, H08 strengthens dramatically; if it doesn't, H08's flux-model translation is wrong.
  • n=1 self-experiment. See self-experiment-protocol.md and the FEUA protocol in memory/project_feua_at_ua_retest.md. Brian's own serum UA + spot urinary FEUA at baseline / mid / post a koji-engineered uricase strain delivery, decomposing renal (URAT1) vs. intestinal (ABCG2) compartments. Tier 4 in the validation-experiments hierarchy; gives directional signal without a Phase 2b cohort.
  • Recombinant rasburicase oral-delivery N=1 pilot. Lower-cost upstream test — does any oral-delivered uricase produce the predicted SUA drop in a typical-gout subject? Decouples the OE strain delivery question from the mechanism question. If the mechanism fails with an off-the-shelf enzyme, H08 dies before strain engineering matters.
  • Phase 2b RCT, typical-gout cohort with Q141K + Q126* stratification. The gold standard. comp-019's RCT design recommendation: typical-gout cohort, 25 mg/day, stratification not enrichment, single-dose, pre-stratify by CKD stage. Definitive killshot; highest cost.
  • In vivo ileal/colonic urate concentration measurement post-OE-strain delivery (humanized microbiome model or human pilot). Tests whether the strain actually delivers functional uricase to the secretion-rich segment. If luminal urate doesn't drop measurably, the platform thesis fails upstream of SUA outcomes.
  • Renal compensation magnitude measurement in a human pilot. If renal reabsorption compensates ≥50% of the gut delta, comp-019's prediction band is wrong on the high side — H08 needs to retract to a smaller predicted ΔSUA, possibly below the 0.5 mg/dL clinical threshold.

Pre-Committed Thresholds (placeholder — to be populated in Phase 2)

To be defined when the killshot menu is populated. Anticipated structure follows H01: declared Alive / Killed / Pending thresholds for each load-bearing claim, plus kill switches independent of the scientific thresholds.

Provisional anchor (to be confirmed in full card):

  • Alive: any future Phase 2b-class data showing mean ΔSUA ≤ −0.5 mg/dL in typical (non-CKD) gout with p < 0.05.
  • Killed: any Phase 2b-class data showing mean ΔSUA > −0.3 mg/dL in typical gout with adequate dose and adherence (i.e., the predicted band of −0.5 to −1.0 is materially missed on the high side, not on the noise side).
  • Pending: anything in between, including n=1 self-experiment signal that is directionally positive but not powered.

Failure Modes Probed (placeholder — to be populated in Phase 2)

To be populated. Anticipated relevant failure modes from linter-design.md §5:

  • species-gap-translation (mouse ABCG2 vs. human; mouse intestinal anatomy vs. human ileal-colonic axis)
  • dose-translation scaling (predicted 25 mg/day functional uricase vs. delivered active enzyme at the secretion segment)
  • published-literature-gap (zero Q141K-stratified uricase trials in the entire clinical corpus)
  • renal-compensation offset (model assumes ≤30%; could be higher in practice)
  • formulation gap (ALLN-346's formulation choices may have under-delivered active enzyme to the secretion segment)
  • trial-design selection (Study 202's broader cohort may have included responders the design failed to detect)

Status

Stub. No killshot executed. No assumption stack pre-registered. Full hypothesis card is queued as Phase 2 — see "Open Follow-Ups" below.

Survival count: 0.

Survival score: 0.0 (undefined until full card and first survived killshot).

Open Follow-Ups (Phase 2)

ID Item Type Status
P2-1 Lit scan: any post-ALLN-346 (2024+) oral or gut-targeted uricase Phase 2 typical-gout readout Opus subagent literature scan Queued
P2-2 Re-analysis attempt: ALLN-346 Study 202 cohort-level genotype data accessibility (FOIA / sponsor request / supplementary data grep) Foreground subagent + procedural Queued
P2-3 Populate full assumption stack, weights, and confidence tiers per H01 template Inline upgrade Queued
P2-4 Populate full killshot menu with score-ranking per linter-design.md §4–5 Inline upgrade Queued
P2-5 Populate pre-committed thresholds + kill switches Inline upgrade (requires P2-1, P2-2 results to anchor magnitudes) Queued
P2-6 Failure-mode coverage map (lint cross-check against linter-design.md §5) Inline upgrade Queued
P2-7 n=1 self-experiment design integration with FEUA protocol per memory/project_feua_at_ua_retest.md Cross-link to self-experiment-protocol.md Queued

Tracked across the 6 follow-up surfaces per the walk-synthesis skill §5: 1. This page's "Open Follow-Ups" section (above) 2. open-questions.md — to be added in same commit 3. computational-experiments.md — Planned Analyses table addition (cross-link to ALLN-346 Study 202 re-analysis attempt) 4. This H08 falsification card stub (current) 5. index.md — one-line surfacing of H08 as the platform's riskiest-assumption anchor 6. synthesis/done/2026-05-09-riskiest-assumption-1*.md — closure annotation

Cross-References