Abstract
Patient stratification is critical for the sensitivity of clinical trials at early stages of neurodegenerative disorders. In Huntington's disease (HD), genetic tests make cognitive, motor and brain imaging measurements possible before symptom manifestation (pre-HD). We evaluated pre-HD stratification models based on single visit resting-state functional MRI (rs-fMRI) data that assess observed longitudinal motor and cognitive change rates from the multisite Track-On HD cohort (74 pre-HD, 79 control participants). We computed longitudinal performance change on 10 tasks (including visits from the preceding TRACK-HD study when available), as well as functional connectivity density (FCD) maps in single rs-fMRI visits, which showed high test-retest reliability. We assigned pre-HD subjects to subgroups of fast, intermediate, and slow change along single tasks or combinations of them, correcting for expectations based on aging; and trained FCD-based classifiers to distinguish fast- from slow-progressing individuals. For robustness, models were validated across imaging sites. Stratification models distinguished fast- from slow-changing participants and provided continuous assessments of decline applicable to the whole pre-HD population, relying on previously-neglected white matter functional signals. These results suggest novel correlates of early deterioration and a robust stratification strategy where a single MRI measurement provides an estimate of multiple ongoing longitudinal changes.
Introduction
Huntington’s Disease (HD) is a genetic neurodegenerative disorder where "premanifest" (pre-HD) individuals can be identified decades before clinical diagnosis. However, the rate of disease progression varies significantly between individuals, posing a major challenge for clinical trials. To evaluate potential treatments effectively, researchers must identify "fast progressors" who are actively declining. Currently, this requires years of longitudinal behavioral testing. This study investigates whether a single, non-invasive resting-state functional MRI (rs-fMRI) scan can replace years of testing by accurately stratifying patients based on their rate of cognitive and motor decline.
Methodology
The study utilized the Track-On HD cohort, comprising 74 pre-HD participants and 79 healthy controls across four international sites. Longitudinal behavioral data, spanning up to seven years, was integrated from the preceding TRACK-HD study to calculate the rate of change (slopes) for 10 cognitive and motor tasks.
As shown in the figure above, the researchers employed a multi-step analytical pipeline:
- Functional Connectivity Density (FCD): Voxel-wise correlation matrices were generated from rs-fMRI data to create FCD maps, measuring the number of strong functional links for each voxel.
- Machine Learning Stratification: Logistic Regression with Elastic Net regularization was used to train classifiers to distinguish "fast" from "slow" progressors based on FCD maps.
- Robustness and Validation: To ensure generalizability, Leave-One-Site-Out Cross-Validation (LOSO-CV) was employed, training the model on three sites and testing it on a fourth.
- White Matter Inclusion: Functional signals from white matter, typically discarded as noise, were included to capture early degenerative changes in connectivity.
Key Results and Insights
- Predictive Power of White Matter: Functional signals in white matter, particularly in tracts connecting the striatum to the motor cortex, were highly predictive of cognitive decline on the Symbol Digit Modalities Test (SDMT).
- Classification Accuracy: The model successfully distinguished fast from slow progressors on the SDMT task with an Area Under the Curve (AUC) of 0.73 (p=0.0023).
- Continuous Assessment: Beyond binary classification, the "distance to the decision boundary" served as a continuous score that correlated significantly with actual longitudinal decline rates (Spearman’s rho = 0.41 for SDMT).
- Independence from Atrophy: The predictive value of FCD maps remained significant even after controlling for physical brain atrophy (measured via Voxel-Based Morphometry), suggesting that functional changes may precede or occur independently of structural loss.
- Reliability: FCD maps demonstrated high stability, acting as a reliable "fingerprint" for individual subjects over time.
Discussion and Future Directions
While this study demonstrates that a single rs-fMRI scan can detect ongoing decline, the data used was retrospective. Future research is needed to determine if this approach can prospectively forecast decline years in advance. Additionally, while LOSO-CV provides robust validation, testing the model on a completely independent external cohort is a necessary step before clinical implementation. The inclusion of functional white matter signals represents a significant paradigm shift, highlighting previously overlooked correlates of early neurodegeneration in Huntington's disease.
Summary updated