When CNS DHA Target Engagement Does Not Translate into Clinical Benefit

When CNS DHA Target Engagement Does Not Translate into Clinical Benefit A Prospective Failure-Attribution and Response-Heterogeneity Framework for PreventE4 A post-primary, pre-secondary-data HISTAMOS adjudication derived from a pre-existing public mechanistic hypothesis HISTAMOS Prospective Adjudication P2 (R0) Document type: Prospective secondary/mechanistic adjudication | Evidence cutoff: 30 August 2026 Author Mohammed El Amin Bouchelit Affiliation HISTAMOS Independent Research / Independent Open-Science Project Correspondence contact@histamos.com - www.histamos.com Version R0 — freeze candidate for public deposition Pre-existing public hypothesis Zenodo Record 21854146 — DOI 10.5281/zenodo.21854146 — published 8 August 2026 Clinical authority None. No participant-level or unpublished PreventE4 analyses are used or claimed. INDEPENDENCE AND INFORMATION-BOUNDARY STATEMENT The central lipid–mitochondrial execution hypothesis underlying this analysis was publicly deposited by HISTAMOS on 8 August 2026, before preparation of the present R0 adjudication. Limited scientific correspondence subsequently occurred with an investigator working in the relevant field. No participant-level PreventE4 data, unpublished analyses, confidential study materials, internal biomarker plans, non-public results, or privileged sponsor documents were used in this work. Scientific ideas communicated privately that were not independently present in the public record or demonstrably present in the pre-existing HISTAMOS publication were excluded. Private correspondence is neither quoted nor used as evidence. No collaboration, data-sharing agreement, consultancy, sponsorship, financial relationship, licensing arrangement, or institutional affiliation with the PreventE4 investigators or the University of Southern California existed at the freeze date. This manuscript was not reviewed, approved, or endorsed by the PreventE4 team or USC and makes no claim of priority over unpublished work by other groups. Abstract Background: PreventE4 demonstrated that 2 g/day docosahexaenoic acid (DHA) robustly increased the cerebrospinal-fluid (CSF) DHA-to-arachidonic-acid (AA) ratio at 6 months, independent of APOE ε4 status, yet no group-level benefit was detected in cognition or brain volumes over 24 months. This result weakens a general CNS-delivery explanation at the measured CSF level but does not identify the downstream layer at which biochemical exposure failed to become functional benefit. Objective: To freeze, before access to non-public participant-level secondary analyses, a causal failure-attribution framework and falsifiable human response-heterogeneity predictions derived from a HISTAMOS mechanistic hypothesis publicly deposited on 8 August 2026. Framework: We separate central biochemical exposure from neuronal membrane utilization, lipid-stress handling, mitochondrial execution capacity, synaptic function, recoverability, and clinical/structural outcome. The HISTAMOS Failure Attribution Standard (HFAS) is used to separate attribution from action and to distinguish program failure from target failure and rescue opportunity. Prospective forecast: Achieved CSF DHA target engagement alone is not expected to be a sufficient clinical-response selector; APOE ε4 alone is not expected to resolve the exposure-to-benefit dissociation; and, if reproducible treatment heterogeneity exists, pre-specified downstream state should add treatment-interaction information beyond central exposure alone. General mitochondrial vulnerability is retained as the strongest competitor to a DHA/lipid-stress-specific execution gate. Prognostic associations without randomized treatment interaction receive no selector credit. Adjudication: The framework pre-specifies HIT, PARTIAL, MISS, and UNRESOLVED outcomes. A null downstream-state interaction with adequate data rejects actionable selectivity rather than triggering post-hoc subgroup expansion; missing or non-identifying measurements require abstention rather than mechanistic storytelling. Conclusion: PreventE4 establishes successful biochemical CNS engagement without detected group-level clinical or structural benefit. The next causal question is not whether more DHA can be delivered, but whether any downstream state reproducibly modifies the translation of delivered DHA into biological or clinical benefit. Keywords: Alzheimer disease; docosahexaenoic acid; DHA; PreventE4; APOE ε4; target engagement; failure attribution; treatment heterogeneity; ACSL6; cPLA2; mitochondrial reserve; precision medicine; HFAS 1. Rationale and public evidence boundary PreventE4 was a phase IIa, 24-month, randomized, double-blind, placebo-controlled trial conducted at the University of Southern California in cognitively unimpaired adults aged 55–80 years with low dietary DHA intake (<200 mg/day) and at least one dementia risk factor. Of 739 screened individuals, 365 were randomized; 181 entered the lumbar-puncture arm and 184 the no-lumbar-puncture arm. Participants received 2 g/day DHA or placebo, stratified by APOE ε4 status [1,2]. The 2026 primary report is already public and therefore cannot be counted as a prospective HISTAMOS prediction. At 6 months, DHA increased the CSF DHA/AA ratio by 0.17 (95% CI 0.15–0.18) versus −0.02 (95% CI −0.04 to −0.0004) with placebo, for a between-group difference of 0.19 (95% CI 0.16–0.21; p<0.0001). The biochemical treatment effect did not differ by APOE ε4 status (interaction p=0.71). No treatment differences were detected in brain volumes or cognitive performance over 24 months. Overall dropout was 38%, largely during the COVID-19 period [1]. These observations establish a strong central biochemical target-engagement signal at the measured CSF level. They do not directly establish productive neuronal membrane incorporation, a specific intracellular lipid flux, normalization of arachidonic-acid signaling, competent mitochondrial/synaptic execution, or the reversibility of pre-existing structural damage. The public result therefore localizes the unresolved problem downstream of simple oral exposure and makes a generalized APOE ε4-dependent CSF delivery failure insufficient at the tested regimen, but it does not identify a single downstream mechanism. Layer Public evidence Current attribution Claim boundary Oral/systemic exposure Randomized 2 g/day DHA; biochemical response observed. SUPPORTED Does not establish an optimal dose for downstream function. CSF DHA/AA engagement Large 6-month treatment effect; no APOE ε4 interaction. SUPPORTED Central compartment engagement ≠ direct neuronal membrane utilization. APOE ε4-dependent CSF delivery failure No interaction on the primary biochemical endpoint. NOT SUPPORTED at CSF level Does not eliminate APOE-linked biology downstream of delivery. Neuronal membrane utilization Not directly identified by the primary trial readout. UNRESOLVED CSF concentration/ratio ≠ membrane incorporation or lipid-species distribution. Lipid-stress handling / cPLA2 Mechanistically plausible from independent literature; not localized by PreventE4. UNRESOLVED Pathway plausibility ≠ treatment-effect modifier. Mitochondrial / synaptic execution Pre-existing hypothesis; not directly tested by the public trial result. UNRESOLVED General vulnerability must be separated from DHA-specific modification. Clinical / structural benefit No detected group-level difference over 24 months. OBSERVED MISS of broad benefit claim Null average effect does not prove a target failure or an actionable subgroup. CNS TARGET ENGAGEMENT ≠ NEURONAL UTILIZATION ≠ FUNCTIONAL BENEFIT PROGRAM MISS ≠ TARGET INVALIDITY ≠ DELIVERY FAILURE ≠ RESCUE OPPORTUNITY 2. Scope, authority, and question under adjudication This R0 record is not a forecast of the already known primary PreventE4 result. It is a prospective secondary/mechanistic adjudication frozen after publication of the group-level trial outcome and before access to any non-public participant-level secondary analyses. Its object is narrower: to state in advance which causal interpretations remain admissible, which heterogeneity patterns would support or refute the pre-existing downstream hypothesis, and when the evidence should remain unresolved. The primary adjudication question is: after successful central biochemical engagement, does any pre-treatment or early downstream state reproducibly modify the randomized effect of DHA on biological, cognitive, or structural outcomes, and can that state be localized to neuronal lipid utilization, lipid-stress handling, general execution vulnerability, a DHA/lipid-stress-specific execution gate, or measurement/recoverability limits? Authority boundary: all claims below are HYPOTHESIS / PROSPECTIVE SECONDARY-ANALYSIS FORECAST. This paper adds no human biological validation, no treatment recommendation, and no patient-level selector. Mechanistic literature provides plausibility and discriminators; it does not retroactively convert the PreventE4 null group-level result into a rescue mechanism. 3. Pre-existing public HISTAMOS hypothesis and causal architecture On 8 August 2026, HISTAMOS publicly deposited “Stress-Specific Mitochondrial Reserve as a Candidate Modifier of Lipid-Driven Synaptic Failure in Alzheimer Disease” on Zenodo (DOI 10.5281/zenodo.21854146) [3]. That article explicitly treated the target-engagement/benefit dissociation as a downstream problem and proposed a causal architecture linking DHA handling, an ACSL6-related membrane prestate, cPLA2/AA-eicosanoid stress, mitochondrial execution reserve, and synaptic function. The earlier article did not claim that this chain is established in human Alzheimer disease. It retained general mitochondrial vu