Aging related cognitive changes associated with Alzheimer's disease in Down syndrome

Nicholas C. Firth, Carla M. Startin, Rosalyn Hithersay, Sarah Hamburg, P. A. Wijeratne, Kin Y. Mok, John Hardy, Daniel C. Alexander, André StrydomAnnals 2018 · DOI

TL;DR

Individuals with Down syndrome (DS) have a significantly higher risk of developing Alzheimer's disease (AD). This study uses an event-based model (EBM) to map the sequence of cognitive decline in DS, revealing that memory and attention tests are key early indicators, while informant reports become more useful in later stages. This work provides a framework for earlier diagnosis and better selection of outcome measures for clinical trials targeting AD in DS.

What Problem Does This Paper Address?

Alzheimer's disease (AD) neuropathology is nearly universal in adults with Down syndrome (DS) by their fourth decade, leading to a greatly increased risk of dementia. However, the specific trajectory of cognitive decline associated with AD in DS, particularly in the preclinical and prodromal stages, is not well-defined. This is a significant challenge because individuals with DS often have pre-existing intellectual disability (ID) and varying baseline cognitive abilities, making it difficult to distinguish normal aging from AD-related decline and to accurately diagnose dementia. Existing studies often use cross-sectional data or short-term longitudinal measurements, which can be insufficient for understanding long-term disease progression. The authors aim to address this by developing a data-driven model to characterize the sequence of cognitive deterioration in DS and assign individuals to distinct disease stages.

Methods

The core of this study is the application of an Event-Based Model (EBM). This is a data-driven, semisupervised machine learning technique that can estimate the order of events (in this case, cognitive decline markers) from cross-sectional data. The EBM works by identifying patterns in how different cognitive measures change over time, effectively creating a "staging" system for disease progression.

Here's a breakdown of their approach:

  • Data Input: They used cognitive test data and informant-rated questionnaire data from 283 adults with DS.
  • Biomarker Selection: A battery of cognitive tests was chosen based on their psychometric properties and relevance to cognitive domains affected by AD and DS. These included tests for general cognitive abilities, visuospatial memory, object memory, orientation, executive function (rule learning, planning), verbal fluency, attention, motor coordination, and hand-eye coordination. Informant measures included adaptive behavior, dementia-specific behaviors, and memory.
  • Handling Floor/Ceiling Effects: To address issues where individuals might score at the lowest or highest possible score on a test (making it impossible to detect change), they used imputation techniques for missing data and allocated zero scores for tasks that were clearly unmanageable.
  • EBM Application: The EBM was fit using two groups: "Younger Adults" (YA, 16-35 years) as a "predecline" control group, and "Older Adults" (OA, 36+ years) as a "pathological" group expected to have varying degrees of AD neuropathology. The model estimates the probability of a biomarker being abnormal (indicating disease progression) and then determines the most likely sequence of these "events."
  • Staging: Based on the estimated event sequence, each participant was assigned a disease stage.
  • Validation: The model's validity was tested by comparing disease stages between individuals with and without a clinical dementia diagnosis, and by examining the influence of APOE genotype (specifically the e4 allele, known to increase AD risk) on disease staging.

Datasets

The study utilized data from the LonDownS (London Down Syndrome) Consortium. This dataset comprises cognitive test results and informant-rated questionnaire data from 283 adults with Down syndrome. The paper does not introduce a new benchmark dataset but rather leverages existing data collected for longitudinal studies of cognitive ability in DS.

Key Insights and Ideas

  • Early Warning Signs: The EBM identified that declines in visuospatial associate memory, hand-eye coordination, and semantic verbal fluency are among the earliest cognitive changes associated with AD progression in DS. This is significant because these tests are relatively specific and may be more sensitive to early, subtle changes than broader cognitive assessments.
  • Informant Reports Lag Behind: Informant-rated measures of everyday function and cognitive ability were found to be "late events" in the sequence, suggesting they are more useful for tracking decline in later stages of AD, closer to or after a dementia diagnosis. This highlights the potential for direct cognitive testing to detect changes before they are noticeable to caregivers.
  • Model Validity: The EBM's ability to stage individuals with clinical dementia diagnoses later than those without, and the association of the APOE e4 genotype with later staging in older adults, provides strong validation for the model's accuracy in reflecting AD progression.
  • Implications for Clinical Trials: The identified sequence of cognitive decline offers a roadmap for selecting appropriate outcome measures in clinical trials. Early-stage tests could be used to identify individuals in preclinical or prodromal stages for interventions aimed at delaying or preventing dementia onset.

Key Results

  • The EBM identified a sequence of 16 "events" or cognitive markers that represent the progression of AD-related cognitive decline in DS.
  • Early events (occurring earliest in the sequence) included visuospatial associate memory (CANTAB PAL), hand-eye coordination (NEPSY-II visuomotor precision), and verbal fluency.
  • Mid-sequence events involved tests of object memory and planning/rule learning (Tower of London, IED task).
  • Late events were primarily informant-rated measures (DLD, Short ABS, OMQ revised).
  • Individuals with a clinical diagnosis of dementia were significantly more likely to be staged later in the disease progression (p < 0.001).
  • Older adults with the APOE e4 genotype were significantly more likely to be staged later compared to those with APOE2 and APOE3 genotypes (p < 0.002 and p < 0.001, respectively).
  • The Younger Adult group was significantly more likely to be at earlier stages of the disease compared to the Older Adult group (p < 0.001).

Limitations and Future Work

  • Cross-Sectional Data: While the EBM is powerful for inferring progression from cross-sectional data, it is acknowledged that longitudinal data would provide a more definitive confirmation of the event sequence and staging trajectories. The authors mention that future work could involve staging patient follow-up scores and checking the monotonicity of EBM stage trajectories.
  • Test Sensitivity: For the object memory task, a significant proportion of younger adults scored at the ceiling, suggesting this test may not be sensitive enough for measuring ability changes in individuals with strong premorbid abilities. The authors suggest increasing the number of objects used in future studies to improve sensitivity.
  • Verbal Memory: The study acknowledges a lack of tasks specifically focusing on verbal memory, which is often an early domain affected in AD.
  • Future Directions: The authors suggest that this EBM methodology could be valuable for:
  • Distinguishing AD-related decline from reversible decline due to comorbidities.
  • Selecting appropriate participants for clinical trials targeting prodromal AD in DS.
  • Analyzing clinical trial data by comparing treatment effects on progression along the EBM stages.
  • Improving dementia diagnosis and informing care planning and support.

Summary updated

Abstract

Our results identify tests of memory and sustained attention may be particularly useful measures to track decline in the preclinical/prodromal stages of AD in DS whereas informant-measures may be useful in later stages (i.e. during conversion into dementia, or postdiagnosis). These results have implications for the selection of outcome measures of treatment trials to delay or prevent cognitive decline due to AD in DS. As clinical diagnoses are generally made late into AD progression, early assessment is essential.
BibTeX
@article{firth2018aging,
  author = {Nicholas C. Firth and Carla M. Startin and Rosalyn Hithersay and Sarah Hamburg and P. A. Wijeratne and Kin Y. Mok and John Hardy and Daniel C. Alexander and André Strydom},
  title = {{Aging related cognitive changes associated with Alzheimer's disease in Down syndrome}},
  journal = {Annals},
  year = {2018},
  doi = {10.1002/acn3.571},
}
Figure