🧬 biology

Subcortical Shape Variations and Their Associations with Cognition Across the 8th Decade of Life. A Study in the Lothian Birth Cohort 1936

This study of the Lothian Birth Cohort 1936 reveals that heterogeneous subcortical shape changes, particularly vertex displacements in the hippocampus and ventral diencephalon, are associated with cognitive aging across the eighth decade of life, offering insights beyond what gross volumetry can provide.

Original authors: Maria del C. Valdes-Hernandez, Wonjung Park, Joanna Moodie, Susana Muñoz Maniega, Janie Corley, Fraser N. Sneden, Mark E. Bastin, Joanna M. Wardlaw, Simon R. Cox, Jinah Park

Published 2026-05-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Maria del C. Valdes-Hernandez, Wonjung Park, Joanna Moodie, Susana Muñoz Maniega, Janie Corley, Fraser N. Sneden, Mark E. Bastin, Joanna M. Wardlaw, Simon R. Cox, Jinah Park

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Picture: Watching the Brain's "Sculpture" Age

Imagine your brain isn't just a solid block of tissue, but a complex, living sculpture made of eight different clay figures sitting deep inside your head. These figures are the subcortical structures (like the hippocampus, thalamus, and others).

For a long time, scientists have tried to understand how these figures change as we get older by simply weighing them or measuring their total size (volumetry). It's like trying to understand how a person is aging by only looking at their weight on a scale. You know they might be losing weight, but you don't know where they are losing it, or if they are shrinking in one spot while bulging in another.

This study, looking at people in their 70s and 80s (the "Lothian Birth Cohort 1936"), decided to look at the shape of these clay figures instead. They wanted to see if the way these brain structures warp, shrink, or bulge is connected to how well people think and remember things.

The Cast of Characters (The Brain Structures)

Think of these eight brain structures as different characters in a play, each with a specific job:

The Thalamus: The busy switchboard operator, connecting different parts of the brain.
The Hippocampus: The librarian, responsible for filing away memories.
The Caudate & Putamen: The coaches, helping with learning new skills and controlling movement.
The Globus Pallidus: The traffic controller, managing the flow of information.
The Amygdala: The emotional alarm system.
The Nucleus Accumbens: The reward center, linking feelings to actions.
The Ventral Diencephalon: The attention manager.

What They Did: The "Time-Lapse" Camera

The researchers took MRI scans of these people at four different points over nine years (from age 72 to 82). They didn't just measure the total volume; they used a special computer model to map the surface of each brain structure like a 3D topographic map.

Imagine taking a photo of a balloon every few years. A simple measurement would tell you the balloon is getting smaller. But this study looked at how the rubber stretched and wrinkled. Did it shrink evenly? Did it get pinched in the middle? Did it bulge out in one spot?

The Main Findings: It's Not a Uniform Shrinkage

1. The "Shrinkage" is Messy and Uneven
The study found that these brain structures don't just shrink like a deflating balloon. The changes are heterogeneous (mixed up).

The "Pinch" vs. The "Bulge": In some areas, the brain tissue was shrinking inward (like a deflating balloon). In other tiny spots, it was actually bulging outward.
Left vs. Right: The left and right sides of the brain didn't always change in the same way. For example, the hippocampus (the memory librarian) changed shape differently on the left side compared to the right side. However, the thalamus (the switchboard) and the globus pallidus (traffic controller) shrank more evenly and symmetrically on both sides.

2. The Connection to Thinking Skills
The researchers asked: Does the way these shapes change have anything to do with how well people think?

The Answer: Yes. They found that changes in general thinking ability were linked to specific "pushes" and "pulls" on the surface of these brain structures.
The Direction Matters: When thinking skills went down, it was often linked to the brain tissue shrinking inward (atrophy) in specific spots. However, for the caudate (the coach), the pattern was different; it showed some outward bulging, which was actually linked to the data in a unique way.

3. The "Snapshot" vs. The "Movie"
The study compared two ways of looking at the data:

The Snapshot (Cross-sectional): Looking at everyone at one specific age to see who has better shapes and better thinking.
The Movie (Longitudinal): Watching the same people change over 9 years to see how their shape changes match their thinking changes.
The Result: For some structures (like the thalamus), the "snapshot" and the "movie" told very similar stories. For others (like the hippocampus and amygdala), the stories were quite different. This suggests that for some brain parts, the way they age over time is a different story than the differences we see between people at a single moment.

The "Why" and the "How"

The researchers were careful to rule out other factors. They checked if things like high blood pressure, diabetes, or head size changed the results. They found that the link between brain shape and thinking remained strong even after accounting for these factors.

They also tested their computer models to make sure the results weren't just a glitch in the software. They tried different "templates" (different starting maps) and found the results stayed the same. This is like checking a sculpture from different angles and with different lighting to make sure the shape you see is real, not an illusion.

The Takeaway

This study tells us that as we age into our 80s, our deep brain structures don't just get smaller in a uniform way. They undergo a complex, uneven dance of shrinking and bulging.

Crucially, the specific way these structures warp is connected to how well we think. Some structures, like the switchboard (thalamus), shrink in a very predictable, symmetrical way that matches our thinking decline. Others, like the memory librarian (hippocampus), change in a more chaotic, uneven way that is harder to predict just by looking at the whole picture.

By looking at the shape rather than just the size, scientists are getting a much clearer, more detailed map of how our brains age and how that aging affects our minds.

Technical Summary: Subcortical Shape Variations and Their Associations with Cognition Across the 8th Decade of Life

Problem Statement
While the study of brain morphology is crucial for understanding development, aging, and disease, traditional volumetric analyses often fail to capture functionally relevant aspects of brain aging. Gross volumetry may obscure nuanced regional changes, such as specific surface deformations or localized atrophy patterns that are not uniform across a structure. Although subcortical structures (e.g., thalamus, hippocampus, basal ganglia) are known to play critical roles in cognition, memory, and executive function, the specific trajectories of their morphological shape changes in cognitively normal elderly individuals—and how these changes associate with general cognitive decline—have not been fully documented. This study addresses the gap in understanding how subcortical shape trajectories evolve during the 8th decade of life and their relationship with general cognitive ability ( $g$ ).

Methodology
The study utilized data from the Lothian Birth Cohort 1936 (LBC1936), a longitudinal study of community-dwelling individuals born in 1936. The analysis covered four waves of data collection (Wave 2 to Wave 5), corresponding to mean ages of approximately 72.5, 76.2, 79.3, and 82.0 years.

Data Acquisition: Participants underwent T1-weighted MRI scans and comprehensive cognitive testing. The cognitive battery included 13 tests (e.g., Block Design, Matrix Reasoning, Verbal Fluency) used to derive a general cognitive factor ( $g$ ) via a latent growth curve model. Childhood intelligence was controlled for using the Moray House Test No. 12 scores from age 11.
Image Processing: Subcortical structures (thalamus, caudate, putamen, globus pallidus, hippocampus, amygdala, nucleus accumbens, and ventral diencephalon) were segmented automatically using SynthSeg, a deep learning method.
Shape Modeling:
- Template Construction: Templates for each structure were generated by rigidly aligning binary masks from 100 randomly selected subjects using the Iterative Closest Point (ICP) algorithm. A 30% occupancy threshold defined the template mesh, which was then smoothed and remeshed to an average edge length of ~1.5 mm.
- Longitudinal Reconstruction: Individual meshes were reconstructed by deforming the template mesh to fit each subject's MRI at each time point. A subsequent AI-based deformable mesh optimization framework (using a PointNet architecture) ensured smooth, anatomically plausible reconstructions while maintaining vertex-to-vertex correspondence across time points and subjects.
- Deformation Metrics: Local deformity was calculated as a signed Euclidean distance between the template and the individual mesh. Inward deformations (atrophy) were represented as positive values relative to the template, while outward deformations were negative.
Statistical Analysis:
- Cross-sectional: General Linear Models (GLM) assessed the association between vertex-wise shape variations and $g$ at each wave, adjusting for childhood IQ.
- Longitudinal: Linear Mixed-Effect Models (LMM) and ANCOVA were used to analyze the association between shape changes (between time points) and cognitive changes over 3-, 6-, and 9-year intervals.
- Sensitivity: Analyses were repeated using templates with different vertex counts (1490 vs. 2588 for hippocampus) and different random subject samples to ensure robustness. Covariates included vascular risk factors, Brain Health Index (BHI), and Intracranial Volume (ICV).

Key Results

Heterogeneous Atrophy Patterns: Subcortical shape changes were heterogeneous. While overall volume contraction was the general trend, specific regions exhibited varied atrophy patterns. The hippocampus and ventral diencephalon showed distinct morphological deformations that differed between left and right hemispheres. In contrast, the thalami and globus pallidi exhibited more uniform, nearly symmetrical volume contractions across the 9-year period.
Caudate Anomaly: Unlike other structures, the caudate showed a positive volumetric slope (apparent enlargement) in some regions. Detailed inspection revealed this was likely an artifact of automatic segmentation extending into white matter hyperintensities and perivascular spaces, rather than true tissue growth.
Cognitive Associations:
- General Trend: Changes in general cognition were primarily associated with inward vertex displacements (atrophy).
- Structure-Specific Findings:
  - Thalamus: Significant associations were found in the antero-medial region, extending to latero-dorsal and pulvinar regions, particularly around age 79.
  - Hippocampus: Dorsal sections of CA1 and CA2 showed consistent negative associations with $g$ . Associations in CA3, CA4, and the dentate gyrus were significant at ages 76 and 79 but disappeared by age 82.
  - Striatum (Caudate/Putamen/Globus Pallidus): Cross-sectional associations were observed at borders with the internal/external capsules. However, longitudinal shape changes in the putamen and globus pallidus were generally not associated with cognitive changes over the 9-year period.
  - Amygdala: Structural changes were limited to shrinkage. Association patterns differed significantly between hemispheres, with the right amygdala showing broader involvement (lateral and basolateral regions) compared to the left.
Cross-sectional vs. Longitudinal Concordance: There was moderate spatial concordance ( $r \approx 0.49–0.53$ ) between cross-sectional and longitudinal association maps. This agreement was strongest in the thalamus, caudate, putamen, and globus pallidus ( $r \approx 0.60–0.71$ ) but weaker in the hippocampus and amygdala ( $r \approx 0.24–0.34$ ). This suggests that while some subcortical systems show stable spatial organization of cognitive vulnerability, others (like the hippocampus) may exhibit greater divergence between inter-individual baseline differences and within-person aging-related changes.
Robustness: Sensitivity analyses confirmed that the results were robust to variations in template mesh density and subject selection. Adjusting for ICV, vascular risk factors, or BHI did not significantly alter the strength or direction of the associations.

Significance and Claims
The paper claims to be the first to document how subcortical brain structures experience dynamic, non-uniform morphological changes associated with fluctuations in general cognitive performance across the 8th and early 9th decades of life in a community-dwelling cohort.

The authors emphasize that:

Subcortical "Aging" is Not Uniform: Different structures age at different speeds and with different patterns; some regions may appear enlarged due to segmentation artifacts or specific pathology, while others shrink.
Shape Offers Nuance: Studying shape deformations provides a more nuanced understanding of the anatomy of normal aging trajectories than volumetry alone, capturing specific regional vulnerabilities.
Methodological Validation: The study validates a shape modeling framework that is robust to segmentation inconsistencies and template choices, demonstrating that vertex-wise correspondence can effectively capture subtle, continuous changes in aging brains.
Clinical Relevance: The findings suggest that the spatial organization of cognition-related subcortical associations is not uniform across structures. The divergence between cross-sectional and longitudinal patterns in the hippocampus and amygdala may reflect the specific temporal and pathological sensitivities of medial temporal lobe structures in aging and neurodegeneration.

The study concludes that these shape changes are representative of healthy aging patterns in the 8th decade, though it notes that participants who dropped out of the study tended to have worse cerebrovascular disease and cognition, suggesting the findings may represent a "healthier" subset of the elderly population. All data, templates, and code are made available to support future research in cognitive aging and epidemiology.