DOI: 10.1002/alz.080628 ISSN: 1552-5260

Atrophy in brainstem nuclei with early Alzheimer’s Disease pathology, a biomarker for identification of at risk individuals?

Susanne G Mueller
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Geriatrics and Gerontology
  • Neurology (clinical)
  • Developmental Neuroscience
  • Health Policy
  • Epidemiology



Autopsy studies suggest that AD‐related tau pathology may affect the brainstem already in adolescence/young adulthood suggesting that it could be a biomarker for the identification of at risk individuals in the preclinical stage. The overall goal was to investigate a. if a newly developed method for the segmentation of internal brainstem structures from T1 and T2 weighted MRIs can be used to identify subjects with volume losses in brainstem nuclei known to be affected by tau pathology, and b. if these brainstem volume losses affect cognition.


The age‐adjusted Total Cognition Composite Score (CCS) and T1 and T2 weighted images were obtained from 675 participants in the Human Connectome Project of typical Aging. The T1 and T2 image intensities were recalibrated and a T1/T2 image calculated. The brainstem was extracted from these 3 images, and k‐means clustering used to identify five brainstem tissue types (Figure 1). The five tissue maps were warped onto a brainstem tissue template on which 48 structures were labeled. Jacobian determinant (JD) maps were calculated and the mean JD from these rois extracted. Linear regression with the mean roi JD as dependent and age as independent variable were used to generate a 675×48 matrix where residuals >2 SD above the fit line were set to 1 (larger than age‐average), those >2 SD below to ‐1 (smaller than age‐average) and all others to 0. Hierarchical clustering analysis with the 675×48 matrix as input was used to identify different brainstem atrophy patterns.


The cluster analysis identified 11 clusters. 31 individuals (mean(SD) age: 56.2(16.2), range:36‐100) were assigned to Cluster 8. This cluster had the most rois (n = 17) with significant volume loss (Figure 2 for detailed results) and the 31 individuals the lowest CCS score compared to those assigned to other clusters (98.2(19.5) vs.109.6(15.2), p = 0.003).


The new brainstem segmentation identified a subgroup with lower cognitive performance and widespread volume loss exceeding that in their age cohort in brainstem nuclei known to be affected by AD. Additional studies will be necessary to determine if individuals showing this specific brainstem atrophy pattern are indeed at risk to develop AD in their later life.

More from our Archive