Female Alzheimer's Disease as a Coherent Biological Ecology: A Framework for Pipeline Triage

Abstract Alzheimer's disease clinical research conventionally treats sex as a covariate to be adjusted for. We argue this analytical practice is structurally incapable of representing the biology of Alzheimer's disease in the female population, where roughly two-thirds of patients are women and where lifetime risk approximately doubles relative to men. The unexplained biological residual of the female prevalence excess — estimated at 40–60% of the gap once differential survival is accounted for — is substantial, and its characterisation requires sex-stratified analysis at the architectural level rather than covariate adjustment at the parameter level. Female Alzheimer's disease is, on the framework presented here, a distinct biological ecology with its own coupled-system dynamics, its own bifurcation structure, and its own intervention windows, none of which sex-as-covariate analysis can recover. The framework characterises the female ecology as a coupled system of five sequential link perturbations clustered around the perimenopausal transition — L1 metabolic (cerebral bioenergetic decoupling), L2 vascular (cerebrovascular protection withdrawal and amplified consequence-coupling), L3 proteostatic (glymphatic and autophagic clearance disengagement), L4 inflammatory (microglial state transition with FDAMic emergence), L5 synaptic (plasticity and reserve depletion) — together with at least one identified coupling node, CN-Cho (the estrogen-PEMT-phosphatidylcholine maintenance system whose disengagement degrades L2, L4, and L5 simultaneously). The architecture is not new biology. It is a coherent integration of the primary literature of the Brinton, Mosconi, Mishra, Zeisel, and other laboratories whose primary mechanistic work the framework synthesises. The contribution is the architecture itself: a topology that distinguishes sequential links from coupling nodes, maps their bidirectional and forward couplings, and renders the female ecology navigable rather than complex. The five sequential links and the coupling node share a characteristic timing-dependent response to estrogen-based intervention: coherence-supporting before the postmenopausal state consolidates, variable to deleterious after. This pattern, replicated independently across L1, L2, L3, L4, and L5 in the source literature, is the bifurcation signature the framework integrates. The vocabulary of bifurcation, attractor, state transition, and coupling that the framework deploys is not imposed on a biology that does not speak it; it is the vocabulary the source laboratories have independently adopted — Brinton's tipping points, the FDAMic literature's cell-fate trajectories, the synaptic plasticity literature's mode-shifts, the choline biochemistry's bidirectional coupling and vicious-cycle dynamics. The framework unifies language the field has been adopting organically. Formal dynamical-systems specification (state-space, bifurcation parameter, attractor basins) is deferred to a subsequent SIP-CORE methods paper; the present framework's claims rest on structural-architectural reasoning grounded in convergent empirical evidence rather than on formal mathematics. The framework generates eight falsifiable predictions, with operational statistical thresholds for refutation specified in advance. Two — the egg-AD sex differential (P1) and the rs12325817 stratification of the choline-AD effect (P2) — are immediately tractable on existing cohort data and require only analytical work. Three further predictions concern sex-stratified longitudinal imaging trajectories at L5 (synaptic density, P3), L4 (microglial state, P6), and L3 (glymphatic flow, P7) that the framework specifies in advance of cohort assembly currently underway. Two predictions concern super-additive therapeutic effects in combination trials not currently active (P4: L1+L2 spanning; P5: CN-Cho-cholinergic). One prediction (P8) concerns upstream-state mediation of cross-cohort variability in the TREM2-APOE-sex three-way interaction, addressing observed attenuation in the Rotterdam Study and analogous cohorts. The framework specifies, for each prediction, the component-level consequence of refutation and three framework-level falsifiers whose joint occurrence would refute the architecture as a structure rather than refining its components. Applied categorically to the 2026 Cummings Alzheimer's clinical pipeline (Index Date 1 January 2026; 36 Phase 3 agents across eight CADRO categories), the framework identifies three structural observations. The pipeline is overweighted toward downstream damage-mitigative agents — amyloid (28%) and the transmitter-receptor class (39%, dominated by cholinergic-axis agents acting on CN-Cho output) together account for two-thirds of Phase 3 effort, while upstream coherence-restorative classes (metabolism/bioenergetics, upstream-targeting inflammation interventions) together represent under 15%. The three Phase 3 cholinergic-axis trials present an immediate retrospective stratification opportunity through CN-Cho biomarker assessment of stored biospecimens, recovering signal currently lost as within-group variance. The combination-therapy gap is structural: none of the eight Phase 3 combination trials targets the framework's highest-priority pairings (L1+L2 metabolic-vascular spanning; CN-Cho bypass with cholinergic pharmacology). The candidate-by-candidate triage and quantified power-calculation work for these recommendations is the work of the companion paper (Paper 2a). The architecture established here is the foundation on which both Paper 2a and the parallel male-ecology characterisation (Paper 2b) build.