Scholarlink

Purpose <p>Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by the accumulation of amyloid-β (Aβ), tau pathology, and neurodegeneration. Although the temporal sequence of these biomarkers has been well described, few studies have leveraged these temporal differences to optimize stage-specific prognosis. This study aimed to construct a spatiotemporal model of AD biomarker progression and to identify stage-dependent optimal biomarkers for predicting time to dementia onset using only cross-sectional data.</p> Methods <p>We developed a novel method, Multimodal Integrated Spatiotemporal Trajectory Estimation (MIST), to model the progression of Aβ PET, tau PET, and structural MRI data. The model was built via a large cohort from the Alzheimer’s Disease Neuroimaging Initiative (ADNI; <i>n</i> = 1475) and validated in an independent cohort from the Open Access Series of Imaging Studies (OASIS-3; <i>n</i> = 876).</p> Results <p>Our results revealed that Aβ deposition occurred first, approximately 18.8 years before dementia onset (95% CI: 16.4–20.6), followed by tau pathology in Braak I–II regions at 6.8 years before onset (95% CI: 5.3–7.7), and hippocampal neurodegeneration at 0.5 years before onset (95% CI: −0.8–1.8). Importantly, these temporal patterns corresponded to biomarker-specific predictive utility. In cognitively normal individuals, Aβ PET combined with cognitive measures yielded the best performance for predicting time to onset, with a mean absolute error (MAE) of 2.23 years (95% CI: 1.56–2.94). In contrast, in the mild cognitive impairment (MCI) stage, structural MRI combined with cognitive measures achieved superior prediction of MAE = 1.09 years (95% CI: 0.81–1.42). The model also showed strong discrimination of imminent onset, achieving an AUC of 0.88 (95% CI: 0.83–0.92) for predicting progression within three years in all preclinical participants.</p> Conclusions <p>This study not only establishes a robust and reproducible spatiotemporal model of AD biomarker progression but also demonstrates that the most informative biomarker for predicting dementia onset varies across disease stages. Our findings highlight the potential of cross-sectional biomarker data for stage-specific prognosis, offering a practical tool for clinical risk stratification and personalized prognostic assessment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction of alzheimer’s disease time to dementia onset using cross-sectional data from spatiotemporal biomarker progression patterns

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • Tianhao Zhang,
  • Binbin Nie,
  • Hua Liu,
  • Guangjuan Mao,
  • Chujun OuYang,
  • Xiaochen Jiang,
  • Baoci Shan

摘要

Purpose

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by the accumulation of amyloid-β (Aβ), tau pathology, and neurodegeneration. Although the temporal sequence of these biomarkers has been well described, few studies have leveraged these temporal differences to optimize stage-specific prognosis. This study aimed to construct a spatiotemporal model of AD biomarker progression and to identify stage-dependent optimal biomarkers for predicting time to dementia onset using only cross-sectional data.

Methods

We developed a novel method, Multimodal Integrated Spatiotemporal Trajectory Estimation (MIST), to model the progression of Aβ PET, tau PET, and structural MRI data. The model was built via a large cohort from the Alzheimer’s Disease Neuroimaging Initiative (ADNI; n = 1475) and validated in an independent cohort from the Open Access Series of Imaging Studies (OASIS-3; n = 876).

Results

Our results revealed that Aβ deposition occurred first, approximately 18.8 years before dementia onset (95% CI: 16.4–20.6), followed by tau pathology in Braak I–II regions at 6.8 years before onset (95% CI: 5.3–7.7), and hippocampal neurodegeneration at 0.5 years before onset (95% CI: −0.8–1.8). Importantly, these temporal patterns corresponded to biomarker-specific predictive utility. In cognitively normal individuals, Aβ PET combined with cognitive measures yielded the best performance for predicting time to onset, with a mean absolute error (MAE) of 2.23 years (95% CI: 1.56–2.94). In contrast, in the mild cognitive impairment (MCI) stage, structural MRI combined with cognitive measures achieved superior prediction of MAE = 1.09 years (95% CI: 0.81–1.42). The model also showed strong discrimination of imminent onset, achieving an AUC of 0.88 (95% CI: 0.83–0.92) for predicting progression within three years in all preclinical participants.

Conclusions

This study not only establishes a robust and reproducible spatiotemporal model of AD biomarker progression but also demonstrates that the most informative biomarker for predicting dementia onset varies across disease stages. Our findings highlight the potential of cross-sectional biomarker data for stage-specific prognosis, offering a practical tool for clinical risk stratification and personalized prognostic assessment.