P3–118: Conversion to MCI in healthy individuals with abnormal CSF Aβ42 levels is associated with specific longitudinal morphological changes of the brain

Marco Lorenzi, Martina Bocchetta, Nicholas Ayache, Xavier Pennec, Giovanni B. Frisoni

Abstract

Abnormal levels of amyloid-beta (Aβ42) in the cerebrospinal fluid (CSF) and progressive brain atrophy are hallmark indicators of Alzheimer’s Disease (AD). However, the precise longitudinal relationship between these factors during the preclinical phase remains poorly understood. This study presents a dynamic model of the association between CSF Aβ42 levels and structural brain changes in healthy elderly individuals. By analyzing longitudinal T1 MR images over three years from the ADNI cohort, we identified that healthy Aβ+ individuals exhibit an accelerated atrophy pattern—specifically in the temporal areas and hippocampi—that predicts subsequent conversion to Mild Cognitive Impairment (MCI). Our findings suggest that clinical conversion in the presence of amyloid is driven by a stereotypical AD-like pathological pattern, identifiable through specific trajectories of brain changes.

Introduction

Alzheimer’s Disease is a progressive condition that begins decades before clinical symptoms manifest. A significant challenge in current research is identifying which cognitively normal elderly individuals are on a trajectory toward Mild Cognitive Impairment (MCI) and eventual AD. While abnormal amyloid-beta levels and brain shrinkage are known biomarkers, many "healthy" individuals harbor amyloid pathology without showing immediate cognitive decline. This research addresses the need for a dynamic model that tracks how brain structure evolves over time in relation to amyloid status, aiming to define the "structural signature" of early-stage neurodegeneration.

Methodology

The study utilized a longitudinal dataset from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), tracking 86 healthy elderly subjects over a three-year period. The cohort was divided into two groups based on baseline CSF Aβ42 levels: 49 subjects with normal levels (Aβ-) and 37 with abnormal levels (Aβ+).

To quantify subtle structural changes, the researchers employed an optimized non-rigid registration pipeline on baseline and follow-up T1-weighted MR images. This approach allowed for voxel-wise statistical comparisons of longitudinal atrophy. Two primary metrics were developed to characterize these changes:

  1. Atrophy Score: A localized measure focusing on the percentage rate of hippocampal shrinkage and the expansion of the temporal horns.
  2. Aβ-induced Score: A global, whole-brain metric designed to capture the specific longitudinal structural deformations statistically associated with abnormal Aβ42 levels.

Key Results

The Amyloid-Specific Atrophy Pattern

Healthy individuals with abnormal amyloid levels (Aβ+) demonstrated a significant acceleration in brain atrophy compared to their Aβ- counterparts. This matter loss was primarily localized in the temporal poles and hippocampi, accompanied by a corresponding expansion of the ventricles and temporal horns.

Predictors of MCI Conversion

The study found a stark divergence in brain trajectories among those who eventually progressed to MCI:

  • Aβ+ Converters: Of the 37 Aβ+ subjects, 9 (24%) converted to MCI within three years. These individuals exhibited a pronounced, stereotypical AD-like atrophy pattern.
  • Aβ- Converters: In contrast, the 6 subjects (12%) from the Aβ- group who converted to MCI showed negligible or inconsistent patterns of longitudinal temporal atrophy.

Clinical Correlations

The interaction between the whole-brain Aβ-induced score and the localized atrophy score served as a powerful predictor of clinical decline. Higher scores on these indices were strongly associated with worsening performance on the Alzheimer's Disease Assessment Scale (ADAS11) and the Clinical Dementia Rating Sum of Boxes (CDRSB), as well as the eventual transition to an MCI diagnosis.

Discussion

The findings suggest that "clinical conversion" is not a monolithic event. When amyloid is present, conversion appears to be the result of a specific, identifiable pathological pathway—likely early-stage AD. Conversely, conversion in the absence of amyloid (Aβ-) may be driven by different biological processes that do not follow the same structural trajectory.

This "Aβ-induced signature" provides a potential window for early intervention. By identifying the specific brain trajectories associated with amyloid pathology, clinicians may be better equipped to distinguish between benign age-related changes and the early stages of a neurodegenerative disease.

Conclusion

Abnormal CSF Aβ42 levels in healthy elderly individuals are associated with a significant and specific acceleration of longitudinal brain atrophy. The combination of global structural shifts and localized hippocampal loss forms a predictive signature for MCI conversion. These results underscore the importance of longitudinal modeling in understanding the transition from healthy aging to Alzheimer's Disease, pointing toward a future where specific trajectories of brain change can inform early diagnosis and personalized treatment strategies.

Summary updated

Abstract

Abnormal levels of Aβ42 in the CSF and brain atrophy are key factors for the development of Alzheimer's disease(AD), even though their mutual influence in the subsequent development of dementia still needs to be fully understood. The aim of this work is to provide a dynamic longitudinal model of the association between CSF Aβ42 levels and structural changes of the brain in healthy elderly. This model is used to characterize potential trajectories towards AD in subjects subsequently converted to MCI. Baseline to 3-years brain T1 MR images were selected for 49 healthy subjects with normal levels of CSF Aβ42 (Aβ-), and for 37 with abnormal level of Aβ42(Aβ+) of ADNI. The images were processed according to an optimized pipeline for longitudinal data based on non-rigid registration. Group-wise models of the average longitudinal atrophy were estimated from the subject-specific time series, and statistically compared voxel-wise. Subject-specific longitudinal changes were also quantified by 1) the percentage rate of hippocampal atrophy and temporal horn expansion(atrophy score), and 2) a whole brain score of the longitudinal structural changes due to Aβ42(Aβ-induced score). Healthy Aβ42+ subjects showed an accelerated atrophy pattern associated to increased matter loss in temporal areas and hippocampi, as well as to increased ventricles expansion. The pattern was more pronounced for the Aβ+ who subsequently converted to MCI (9/37), while the subjects Aβ- converted to MCI (6/49) showed negligible atrophy patterns (Figure 1). Aβ+ subjects had higher Aβ-induced and atrophy scores, and the interaction of these two indices was strongly associated to longitudinal worsening of ADAS11 and CDRSB, and to conversion to MCI stage (Table 1). Abnormal Aβ42 levels in healthy elderly are associated to 1) significant acceleration of longitudinal brain atrophy in the temporal areas, and 2)global specific longitudinal structural changes. In particular, the Aβ+ subjects with higher specific longitudinal brain trajectories and greater atrophy did subsequently convert to MCI, and were characterized by a typical AD-like atrophy pattern. The present results point out that clinical conversion in presence amyloid is likely associated to early AD, and is already identifiable by a specific trajectory of brain changes and by a stereotypical pathological AD pattern. Voxel-by-voxel comparison of the average group-wise log-jacobian determinant associated to the longitudinal trajectories (p<0.05, permutation test). Blue: significant volume loss, yellow: significant volume gain. A) Subjects positive to Aβ show higher longitudinal brain matter loss when compared to the negative, prevalently localized in temporal poles and hippocampus, and correspondent expansion of the ventricles and temporal horns. B) The pattern becomes more evident if we consider only the subjects subsequently converted to MCI, who already show a specific pattern of AD progression. C) On the contrary, subjects Aβ negative converted to MCI do not show a consistent pattern of longitudinal temporal atrophy. The longitudinal models of the differential progressions between the different groups are available at the link: http://www-sop.inria.fr/members/Macro.Lorenzi/ModelAbeta
BibTeX
@inproceedings{lorenzi2013p,
  author = {Marco Lorenzi and Martina Bocchetta and Nicholas Ayache and Xavier Pennec and Giovanni B. Frisoni},
  title = {{P3–118: Conversion to MCI in healthy individuals with abnormal CSF Aβ42 levels is associated with specific longitudinal morphological changes of the brain}},
  booktitle = {Alzheimer},
  year = {2013},
  doi = {10.1016/j.jalz.2013.05.1189},
}
Figure