AbstractBackgroundAlzheimer’s disease (AD) has recently been reconceptualized into a dual entity, a biological disease, and a clinical syndrome. Following these advances, the National Institute on Aging – Alzheimer’s Association (NIA‐AA) and the International Working Group (IWG) released four updated diagnostic guidelines for AD diagnosis: NIA‐AA 2011, IWG 2016, NIA‐AA 2018, and IWG 2021. Although many studies still rely solely on a clinical diagnosis of AD, and a consensus on diagnostic criteria is still lacking. Here, we aim to elucidate how the aforementioned AD guidelines affect the individuals’ diagnostic label.MethodClinical, demographic and biomarker data were extracted from ADNI (n = 1215), and used to build classification algorithms according to diagnostic criteria from NIA‐AA 2011, IWG 2016, NIA‐AA 2018, and IWG 2021. Subjects then were labeled as Not AD (NA), At Risk of Developing AD (AR), or AD. Results were compared according to cognitive stage, cognitively unimpaired (CU) or cognitively impaired (CI), and to AD biomarkers profile. Kaplan‐Meier curves were used to evaluate conversion to CI in CU individuals using each diagnostic guideline. All analyses were performed in R (v 4.0.3).ResultIndividuals were diagnosed using criteria from the NIA‐AA 2011 [NA = 36.8%, AR = 23.2%, AD = 40%], IWG 2016 [NA = 37.9%, AR = 28.8%, AD = 33.3%], NIA‐AA 2018 [NA = 40.5%, AR = 11.1%, AD = 48.4%], and the IWG 2021 [NA = 51.3%, AR = 8.7%, AD = 40%] (Figure 1). Diagnoses were discordant between criteria in 39.8% of the individuals, with 69.4% of those being in the A+T‐ or A‐T+ biomarker groups. Conversion to CI of CU individuals AR compared to NA were calculated for the NIA‐AA 2011 [log‐rank 0,001], IWG 2016 [log‐rank 0,000], NIA‐AA 2018 [log‐rank 0,24], and IWG 2021 [log‐rank 0,000].ConclusionOur findings indicated that almost 40% of the studied population presented discordant diagnoses using different AD criteria. Most of these had only one positive biomarker (amyloid or tau). All guidelines except for the NIA‐AA 2018 managed to differentiate asymptomatic individuals at risk of developing cognitive impairment. Our findings have important implications in clinical practice and the design of future clinical trials.