Daeun Shin, Sungjoo Lee, Jun Pyo Kim, Hyemin Jang, Jihwan Yun, Min Young Chun, Jehyun Ahn, Seongmi Kim, Kyoungmin Kim, Kyoungmin Kim, Soyeon Yoon, Hee Jin Kim, Heekyoung Kang, S.F. Yim, Hee Kyung Park, Seonghyeon Kim, Duk L. Na, Henrik Zetterberg, Kaj Blennow, Fernando González‐Ortiz, Nicholas J. Ashton, Michael W. Weiner, Sang Won Seo, Kyunga Kim, Kyunga Kim
Accurately predicting disease progression remains a major challenge in Alzheimer's disease (AD). Here we show that a biomarker-integrated prognostic staging system can stratify progression risk across the disease course by jointly incorporating cognitive status, established risk factors, plasma biomarkers, and neuroimaging measures. In the K-ROAD cohort (N = 1,263), the dominant prognostic contributors varied by clinical context-GFAP in cognitively unimpaired individuals, hippocampal volume in mild cognitive impairment, and age in dementia-while plasma phosphorylated tau-217 provided consistent secondary prognostic information across stages. Outcome-specific staging captured clinically meaningful gradients of progression risk and informed construction of a unified six-stage framework (Stage 0-IVB) with distinct inflection points of accelerated decline. External validation in the ADNI cohort (N = 290) demonstrated consistent patterns of worsening prognosis, particularly in early and intermediate stages. This system provides a clinically interpretable approach to risk stratification and may serve as an exploratory framework for biomarker-integrated prognostic stratification in AD.