Measurement fidelity and the liver–brain axis in MASLD: reappraising the evidence linking hepatic fibrosis to incident dementia

Background Reviews of the liver–brain axis in metabolic dysfunction–associated steatotic liver disease (MASLD) increasingly present hepatic fibrosis as a supported risk factor for dementia. The underlying evidence is heterogeneous in how the exposure is measured, from registry codes and serum or proxy scores to structural measurements of very different fidelity. Methods This narrative review reorders the evidence by the fidelity of the exposure measurement to true hepatic fibrosis, across five tiers from registry coding to structural measurement, and fixes the outcome as clinically or registry-adjudicated incident dementia. Results Reordered this way, the most precise positive estimates come from measures that embed dementia risk factors, chiefly the age-containing fibrosis-4 index (FIB-4) and, more weakly, severity-based registry coding. Proxy scores are unstable in direction, and the structural measures closest to liver tissue show no observed association or are absent. The available structural estimates are imprecise: their confidence intervals include the FIB-4 hazard ratios of 1.18–1.52 from the two largest cohorts, and neither differs statistically from the UK Biobank FIB-4 estimate, so they cannot be read as evidence against an association. No prospective liver magnetic resonance elastography (MRE) cohort for incident dementia was identified within the scope of this review. Conclusion Because measurement fidelity is closely coupled with attainable sample size and cost, three explanations remain largely inseparable (Mendelian randomization aside): construct contamination in embedded measures, confounding by measurement cost and statistical power, and a real fibrosis–brain effect. This review does not adjudicate among them but makes the inseparability explicit and specifies the study that could resolve it: a prospective cohort pairing a high-fidelity structural exposure with clinically adjudicated incident dementia.