Disentangling the Normal Aging from the Pathological Alzheimer’s Disease Progression on Cross-sectional Structural MR Images.

Marco Lorenzi, Xavier Pennec, Nicholas Ayache, Giovanni B. Frisoni2013

TL;DR

This paper introduces a novel method to disentangle the effects of normal brain aging from Alzheimer's disease (AD) progression using structural MRI. By separating these two biological processes, the authors can better identify AD-specific changes and show that advanced AD stages are characterized by both accelerated aging and distinct pathological atrophy.

What Problem Does This Paper Address?

Understanding Alzheimer's disease (AD) progression is crucial for developing effective treatments and diagnostic tools. However, brain morphology changes due to AD are often intertwined with those caused by normal aging. This makes it challenging to pinpoint the specific structural changes that are truly indicative of AD, especially in its early, pre-symptomatic, and prodromal stages. The ability to differentiate between age-related brain changes and AD-specific pathology is fundamental for accurately assessing disease progression and identifying reliable biomarkers.

Methods

The core of the proposed method is a non-rigid registration technique that models the observed brain anatomy as a combination of two components:

  1. Normal Aging Component (w_age): This component captures the typical changes in brain structure associated with healthy aging. The authors model this as a stationary and linearly evolving process in a reference anatomical space, defining a "virtual age" for each subject.
  2. Specific Component (w_specific): This component represents the structural changes that are not attributable to normal aging and are therefore hypothesized to be AD-specific.

The method works by first estimating a "virtual age" for each subject by projecting their brain anatomy onto a model of healthy aging. The difference between the subject's actual anatomy and this projected "aging" anatomy then yields the specific component. This specific component is further analyzed by quantifying the divergence of the deformation field, which represents local compression (atrophy) or expansion (e.g., of CSF spaces).

Datasets

The study utilizes data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. This is a well-established, publicly available dataset for AD research. The authors specifically used structural MRI data from:

  • 57 healthy subjects with normal CSF Aβ42 levels (Group Aβ-)
  • 41 healthy subjects with abnormal CSF Aβ42 levels (Group Aβ+)
  • 86 subjects with mild cognitive impairment who converted to AD (Group MCIconv)
  • 110 MCI subjects who remained stable (Group MCIstable)
  • 134 AD patients (Group AD)

Key Insights and Ideas

  • AD is not just accelerated aging: The study strongly suggests that AD is not simply a faster version of normal aging. Instead, it involves both accelerated aging and a distinct, complementary pathological atrophy process.
  • "Virtual Age" as a biomarker: The concept of "virtual age" derived from the aging model provides a quantitative measure of how much a subject's brain appears older than their biological age due to AD-related processes. This "virtual aging" increases as the disease progresses.
  • Specific atrophy patterns emerge early: The specific component of atrophy is detectable even in healthy individuals with elevated Aβ42 levels (Aβ+ group) and becomes more pronounced in MCI converters and AD patients, highlighting its potential as an early marker.
  • Divergence as a measure of atrophy: The divergence of the specific deformation field effectively quantifies regional brain matter loss, providing a sensitive measure for differentiating between clinical groups.

Key Results

  • Virtual Age Increases with AD Progression: The study found that subjects in the Aβ+, MCIconv, and AD groups exhibited significantly higher "virtual ages" compared to the healthy Aβ- group. MCI converters were also found to be "virtually older" than stable MCI subjects, suggesting an accelerated aging process linked to the impending conversion to AD.
  • Discriminative Power of Specific Component: The specific component, particularly its divergence maps, demonstrated good classification accuracy. For instance, it achieved 91% sensitivity and 84% specificity in distinguishing AD patients from healthy controls. While lower, the classification for MCI converters versus stable MCI subjects (67% sensitivity, 63% specificity) still significantly outperformed chance.
  • Regional Atrophy Patterns: The specific atrophy patterns were found to be most pronounced in MCI converters and AD patients, affecting areas like the ventricles, temporal poles, entorhinal cortex, and hippocampi.

Limitations and Future Work

  • Assumption of Global Aging: The method assumes a global aging process for estimating the "virtual age." The authors acknowledge that this might lead to misinterpreting regional neurodegeneration as global accelerated aging if not carefully handled.
  • Interaction between Aging and AD: The paper notes the plausibility of a "feedback" process where AD-specific changes might accelerate global aging. While the model attempts to disentangle these, the complex interplay remains an area for further investigation.
  • Future Directions: The authors suggest that their framework can lead to new insights into AD dynamics and promote the development of novel diagnostic techniques for early disease detection. Further work could explore the application of this method for drug development, specifically targeting disease-modifying drugs that do not interfere with natural aging.

Summary updated

Abstract

The morphology observed in the brain of patients affected by Alzheimer’s disease (AD) is the contribution of different biological processes such as the normal aging and the AD-specific pathological matter loss. The ability to differentiate these complementary biological factors is fundamental in order to reliably evaluate the pathological AD-related structural changes, especially at the earliest phase of the disease, at prodromal and pre-clinical stages. We propose a method based on non rigid-registration to estimate the different contributions of these comple-mentary factors, and to identify the brain structural changes which are specific for the pathological component. The experimental results pro-vide a description of the anatomical changes observed across the AD time span: normal aging, normal aging at risk, conversion to MCI and latest AD stages. More advanced AD stages are associated to “virtually older” brains, and to increased specific morphological changes that are not re-lated to the normal aging. These results provide new insights that can? Data used in preparation of this article were obtained from the Alzheimers Disease Neuroimaging Initiative (ADNI) database (www.loni.ucla.edu/ADNI). A complete listing of ADNI investigators can be found at: www.loni.ucla.edu/ADNI/Collaboration/ADNI Authorship list.pdf ha l-0
BibTeX
@misc{lorenzi2013disentangling,
  author = {Marco Lorenzi and Xavier Pennec and Nicholas Ayache and Giovanni B. Frisoni},
  title = {{Disentangling the Normal Aging from the Pathological Alzheimer’s Disease Progression on Cross-sectional Structural MR Images.}},
  year = {2013},
}
Figure