HAI Book 2025 - Flipbook - Page 387
Gallego-Rudolf, Jonathan
93
Tracking biomarkers accuracy for predicting progression to mild
cognitive impairment
Jonathan Gallego-Rudolf1,2, Alex Wiesman2, Sylvain Baillet2, Sylvia Villeneuve1
1
Douglas Research Centre, Montreal, QC, CA
Montreal Neurological Institute, Montreal, QC, CA
2
The Alzheimer's disease (AD) continuum includes a wide preclinical phase in which neuropathological events take
place decades before symptom onset. Identifying early biomarkers for predicting the risk of progression from the
asymptomatic to the mild cognitive impairment (MCI) stage is crucial for improving patient prognosis, as
therapeutic interventions might be more promising at this stage.
We assessed spectral power features from task-free magnetoencephalographic (MEG) recordings, magnetic
resonance imaging (MRI) derived hippocampal volume, plasma biomarkers, and positron emission tomography
(PET) measures of A´ and tau deposition in a group of cognitively unimpaired older adults with a family history of
AD (N=99). From this sample, 30 individuals developed MCI based on a multidisciplinary consensus who had access
to longitudinal neuropsychological assessments and were blind to AD biomarkers. We benchmark these
biomarkers using a series of logistic regression models to assess the temporal evolution of their accuracy for
predicting MCI progression, in combination with clinical information (Figure 1).
Neurophysiological activity features were as good as A´ and tau PET for predicting progression to MCI when
acquired close to symptom onset but showed a steeper decrease in accuracy at larger intervals compared to PET
and plasma biomarkers, which remained the most accurate over time (Figure 2). When incorporating all features
into the same model, A´-PET proved to be sensitive for predicting MCI progression up to 6 years before clinical
onset, while MEG and tau-PET (along with age and sex) were good predictors 3-4 years prior to MCI progression
(Figure 3).
Our results delineate a timeline evolution for the accuracy of distinct biomarkers for predicting progression from
the asymptomatic to the prodromal stage of the AD continuum. These results highlight the dynamic sensitivity of
different biomarkers for predicting symptom onset and might serve as a reference for future work in clinical
settings and trials.
HAI2025 - 387
