Brain Age as Accurate Biomarker of Preclinical Cognitive Decline

Strategic Insights -

Research from the Academic College of Tel Aviv-Yaffo has led to significant findings on the use of brain age as a biomarker for early risk stratification and monitoring of cognitive aging trajectories in individuals with Alzheimer's disease (AD). The study, which was published in the Journal of Neurology, analyzed longitudinal T1-weighted MRI data from 224 cognitively unimpaired older adults over a 12-year period. The results showed that brain-predicted age difference (brain-PAD) was the strongest baseline association with future conversion to AD and had the highest classification accuracy among the examined biomarkers.

Key Takeaways:

  • The study involved 224 cognitively unimpaired older adults, with 112 converters to mild cognitive impairment (HC-converters) and 112 stable controls (HC-stable).
  • The groups were matched at baseline for age, education, and cognitive scores, with a mean age of approximately 74-75 years and 16.4 years of education.
  • Brain-PAD showed the strongest baseline association with future conversion to AD, with a b-value of 1.25 and t-value of 3.52 (p = 0.0009) and had the highest classification accuracy (AUC = 0.66, sensitivity = 62%, and specificity = 67%).
  • Longitudinal mixed-effects models revealed a significant positive slope in brain-PAD for converters (b = 0.0079, p = 0.003) and a non-significant trend in stable controls (b = 0.0047, p = 0.075).
  • Hippocampal volume and AD-cortical signature declined similarly in both groups, while mean cortical thickness had limited discriminative or dynamic utility.
  • The findings support brain-PAD, derived from routine T1-weighted MRI using machine learning, as a sensitive, performance-independent biomarker for early risk stratification and monitoring of cognitive aging trajectories.

Statistics:

  • 224 participants were involved in the study, with 112 converters to mild cognitive impairment and 112 stable controls.
  • The groups were matched at baseline for age, with a mean age of approximately 74-75 years.
  • Brain-PAD showed a b-value of 1.25 and t-value of 3.52 (p = 0.0009).
  • The classification accuracy of brain-PAD was AUC = 0.66, with sensitivity = 62%, and specificity = 67%.
  • Longitudinal mixed-effects models revealed a significant positive slope in brain-PAD for converters (b = 0.0079, p = 0.003) and a non-significant trend in stable controls (b = 0.0047, p = 0.075).
  • Hippocampal volume and AD-cortical signature declined similarly in both groups, with a mean decline of 0.01 and 0.05 units per year, respectively.
  • Mean cortical thickness had limited discriminative or dynamic utility, with a decline of 0.005 units per year.

Sources:

  • Brain age as an accurate biomarker of preclinical cognitive decline: evidence from a 12-year longitudinal study. Journal of Neurology, 2025;272(10):672.
  • Springer Heidelberg, Tiergartenstrasse 17, D-69121 Heidelberg, Germany (www.springer.com; www.springerlink.com/content/0340-5354/)
  • Odelia Elkana, Behavioral Sciences, Academic College of Tel Aviv-Yaffo, Tel Aviv, Israel (news reporters, October 17, 2025)