The 1984 NINCDS–ADRDA criteria required dementia and post-mortem confirmation. They could no longer account for the unprecedented growth in biomarker science. An international group convened in 2005 to propose a new framework.
For the first time, AD could be diagnosed in vivo at the prodromal (pre-dementia) stage. The framework required a core clinical criterion — an episodic memory deficit of the hippocampal type (impaired free recall not corrected by cueing) — supported by at least one biomarker: MRI, FDG-PET, CSF, or amyloid PET.
The 2007 framework created a conceptual ambiguity: did "Alzheimer's disease" refer to the clinical entity or the underlying neuropathology? A shared vocabulary was needed for both research and clinical communities.
Alzheimer's disease was redefined as a purely clinical entity — symptomatic individuals with a specific phenotype and in vivo biological evidence. Alzheimer's pathology was separated as the neuropathological process, which may exist without clinical expression. This distinction would become the cornerstone of the 2021 and 2024 IWG position against purely biological diagnosis.
Eight years of accumulated evidence on biomarkers and clinical phenotyping called for a refined framework. The IWG-2 incorporated new data on CSF, amyloid PET, tau PET, and neuroimaging, and extended the criteria to atypical forms and preclinical states.
Biomarkers were reclassified: pathophysiological markers (CSF Aβ42, tau, p-tau; amyloid PET) directly indicate Alzheimer's pathology and anchor the diagnosis; topographical markers (MRI, FDG-PET) reflect downstream neurodegeneration and are better suited to measuring disease progression. The FCSRT was confirmed as the gold-standard memory test, with ~92% specificity for identifying prodromal AD.
Issued from a joint IWG / Alzheimer's Association meeting (Washington DC, July 2015). As biomarker accessibility expanded, a rigorous and unified definition of the preclinical stage had become urgently needed — along with a frank discussion of its ethical implications.
The paper demonstrated that the majority of cognitively normal amyloid-positive individuals will not develop symptoms in their lifetime. Binary biomarker thresholds are inherently artificial. Individual risk stratification is not yet clinically actionable. This landmark paper directly anticipated the arguments of the 2021 and 2024 IWG papers against purely biological diagnosis.
In 2018, the NIA–AA proposed a purely biological definition of AD based on biomarker status alone, with no clinical requirement. The IWG responded with a detailed critique of this approach and concrete recommendations for clinical practice.
The IWG demonstrated three critical limitations of a purely biological definition: (1) the lifetime risk of AD dementia in an amyloid-positive 65-year-old man is only ~22% — the majority will remain asymptomatic; (2) copathologies (Lewy bodies, TDP-43, vascular) are the rule, not the exception, confounding biomarker interpretation; (3) binary biomarker thresholds are arbitrary and do not map onto disease reality. The diagnosis of AD must remain restricted to symptomatic individuals.
The revised AA 2024 criteria further entrenched the biological definition. Societal concerns about labeling cognitively normal individuals with AD, the emergence of blood-based biomarkers, and the approval of disease-modifying therapies made an updated IWG response essential.
The IWG 2024 paper formalises three mutually exclusive categories: asymptomatic at-risk (amyloid positive, uncertain progression), presymptomatic AD (near-deterministic biomarker profiles: monogenic mutations, Down syndrome, APOE ε4 homozygous with SORL1, amyloid+ tau+ in neocortical regions), and Alzheimer disease (symptomatic, clinical phenotype + pathophysiological biomarkers). Core 1 biomarkers alone are insufficient to diagnose AD in cognitively normal individuals.