Prevailing models of neurodegeneration center on amyloid-$\beta$ and tau, yet metabolic dysfunction precedes proteinopathy by decades. We propose compartment-specific citrate uncoupling: opposing disturbances of citrate in neurons and astrocytes as a proteinopathy-independent driver of neurodegeneration. In neurons, a Warburg-like shift depletes cytosolic citrate, which starves ATP-citrate lyase (ACLY) and erodes the H3K27ac/H3K9ac marks that sustain synaptic gene expression, while loss of citrate-mediated PFK1 inhibition collapses the SLC13A5-dependent glycolytic gate. In surviving astrocytes, loss of ACO1, with SLC13A3 a weaker and less specific accompaniment, traps newly synthesized citrate intracellularly; ACLY then hyper-activates and STAT3 hyperacetylation drives an A1 identity. The intercellular citrate gradient inverts. We reanalyzed GSE5281 (161 postmortem samples across six brain regions). Core TCA-cycle enzymes were coherently downregulated in Alzheimer’s disease in five of six regions, more strongly in women, and not in the disease-spared visual cortex. The SLC13 transporter family instead shifted in opposite directions across regions. A cell-type-resolved reanalysis of the SEA-AD multiregion atlas placed the astrocytic arm on ACO1, which fell with pseudo-progression in all nine regions tested; SLC13A3 fell across most cortical cell types and is not compartment-specific. A four-proposition framework then combines PDK1 induction, Cryo-NanoSIMS, MALDI-TIMS-MSI, and spatial transcriptomics. If the model holds, neurodegeneration reads as a routing failure of a central metabolic hub.

Background. Donepezil, galantamine and rivastigmine are dispensed free of charge to people with mild to moderate Alzheimer's disease in the Brazilian Unified Health System (SUS), but their comparative long-term effectiveness in routine care is unknown. Under the national clinical protocol, memantine is added to a cholinesterase inhibitor (ChEI) when the disease reaches the moderate stage, so memantine add-on can serve as a marker of clinical progression. Methods. Nationwide new-user cohort study emulating a target trial in which adults with Alzheimer's disease eligible for a ChEI would be randomly assigned to start donepezil, galantamine or rivastigmine and followed for memantine add-on. We used outpatient medication dispensing authorizations (APAC records of the Outpatient Information System, SIA/SUS) from all 27 federative units, January 2018 to December 2025. We included patients who started a single ChEI between October 2019, after memantine was first recorded nationally, and December 2025, after at least 12 months without anti-dementia dispensing. The outcome was the first memantine dispensing after the index month. Follow-up ended at add-on, recorded death, loss to follow-up (more than 3 observed months without dispensing) or end of data. Deaths were rarely recorded and were treated as censoring. One pseudonymized health card number shared by many patients was excluded, and follow-up ended earlier in the two affected states. We estimated the risk of add-on at 12 to 48 months had all patients started each drug and remained under observation, using stabilized inverse probability of treatment and censoring weights, with 95% confidence intervals (CI) from 200 bootstrap replicates. The analysis assumed no unmeasured confounding or informative censoring given the measured covariates. Results. The cohort included 142 758 new users (donepezil 74 165; galantamine 30 691; rivastigmine 37 902; mean age 77.3 years; 63.0% women), followed for a median of 14 months; 9 812 memantine add-on events occurred. After weighting, all standardized mean differences were below 0.05. The 24-month weighted cumulative incidence of memantine add-on was 9.7% (95% CI 9.4-10.0) with donepezil, 9.0% (8.6-9.4) with galantamine and 5.6% (5.3-5.9) with rivastigmine. Compared with donepezil, the 24-month risk difference was -0.7 percentage points (pp) (-1.1 to -0.3) for galantamine and -4.1 pp (-4.5 to -3.7) for rivastigmine (risk ratio 0.58, 0.55-0.61). Estimates were stable across sensitivity analyses, including a composite outcome that counted unrecorded deaths and dropouts as events (24-month risk difference -4.1 pp, -4.8 to -3.3) and analyses within age groups (risk ratio 0.64 below 75 years). Deaths were almost never recorded in the dispensing data (at most 0.47% of treated patients per year). Conclusions. In routine SUS care, initiation with rivastigmine, mostly as a transdermal patch, was associated with a lower risk of memantine add-on than initiation with donepezil or galantamine. Because disease severity and death are not recorded in the dispensing data, this difference may reflect prescribing patterns or residual confounding by severity rather than slower progression; additional analyses made differential mortality a less likely explanation. It should be confirmed with linked clinical and mortality data.