Dissociable contributions of cortical thickness and surface area to cognitive ageing: evidence from multiple longitudinal cohorts.

This multi-cohort longitudinal study reveals that cortical thickness, which declines more steeply with age, is a more sensitive marker of dynamic cognitive ageing processes, whereas cortical area remains relatively stable and primarily reflects trait-like cognitive ability.

The Gist

The Big Question: Why Do Some Brains Age Better Than Others?

Imagine your brain is like a house. For a long time, scientists have been measuring the "size" of the house (the total volume) to see how well it's holding up as it gets older. They knew that as people get older, the house tends to shrink. But they didn't know how it was shrinking.

This study asks a crucial question: Is the house shrinking because the walls are getting thinner, or because the floor plan is getting smaller?

In brain science terms:

• Cortical Thickness = The thickness of the walls (the grey matter).
• Cortical Surface Area = The size of the floor plan (how much space the brain covers).

The researchers found that these two things are actually completely different stories when it comes to aging and memory.

---

The Two Characters: The "Wall" vs. The "Blueprint"

To understand the findings, let's think of the brain as a building with two distinct features:

1. The Walls (Cortical Thickness)

Think of the walls as the active, living part of the house. They are made of bricks and mortar that are constantly being used, repaired, and sometimes worn down by the weather (aging).

• What the study found: As we get older, these walls get thinner quite steadily.
• The Connection: When the walls get thinner, the house's ability to function (your thinking skills) drops noticeably.
• The Metaphor: If the walls of your house start crumbling, the rooms become drafty and cold. This directly affects how comfortable you are living there. The study shows that thinning walls are the main reason our thinking skills decline as we age.

2. The Floor Plan (Surface Area)

Think of the floor plan as the original blueprint of the house. It was drawn up when the house was built (during childhood and early development) based on the genetic "architect."

• What the study found: This floor plan is surprisingly stable. Even as the house gets very old, the size of the floor doesn't change much until very late in life.
• The Connection: The size of the floor plan is mostly determined by your genes (your DNA). It predicts how smart you were when you were young, but it doesn't tell us much about how your thinking will change as you get older.
• The Metaphor: Imagine a mansion with a huge floor plan. Even if the house is old and the walls are thin, the size of the rooms hasn't changed. The floor plan tells you about the house's potential when it was new, but it doesn't explain why the house is falling apart today.

---

The "Detective Work": How They Figured It Out

The researchers didn't just look at one group of people; they looked at three different groups of people from different places (Cambridge, the US, and a diverse community study) over many years. They used a "time machine" approach:

1. The Cross-Sectional Clue: They looked at people of all ages at one moment in time. They saw that older people had thinner walls, and this matched up with lower test scores.
2. The Longitudinal Clue (The Time Travel): They followed the same people over 10+ years. They watched the walls get thinner year by year.
   • The Result: In the people whose walls got thinner the fastest, their thinking skills dropped the fastest.
   • The Twist: In the people whose floor plans changed, their thinking skills didn't really change at all. The floor plan was just a "trait" (something you are born with), while the wall thickness was a "state" (something that changes with time).

The "Genetic Blueprint" vs. The "Aging Process"

The study also looked at Polygenic Scores (a score based on your DNA that predicts intelligence).

• The Blueprint (Area): Your DNA strongly influences the size of your floor plan. If you have "smart genes," you likely have a larger floor plan, which helps your baseline intelligence.
• The Aging (Thickness): Your DNA doesn't really control how fast your walls thin out. That is driven by the aging process itself.

The Analogy:

• Surface Area is like the engine size of a car you bought new. A big engine (large area) means you started fast.
• Thickness is like the wear and tear on the engine parts. Over time, the parts get worn down.
• The Finding: It doesn't matter if you started with a V8 engine (large area); if the engine parts wear down quickly (thin walls), the car will eventually slow down. The wear and tear is what predicts the slowdown, not the original engine size.

Why Does This Matter?

For a long time, scientists just measured the "total volume" of the brain. It's like measuring the total square footage of a house without checking if the walls are crumbling or the roof is leaking.

This study tells us:

1. Stop looking at just the total size. We need to measure the walls (thickness) and the floor (area) separately.
2. Thickness is the warning sign. If you want to know if someone's brain is aging poorly, look at how fast their walls are thinning.
3. Area is the history book. The floor plan tells you about the person's genetic potential, but it doesn't tell you about their current health.

The Bottom Line

As we age, our brains don't just "shrink" in a generic way.

• The floor plan (Area) stays mostly the same; it's our genetic legacy.
• The walls (Thickness) are the ones that wear down, and that is what causes our thinking to slow down.

By understanding this difference, doctors and scientists can better predict who is at risk for cognitive decline and perhaps find ways to keep the "walls" thick and strong for longer.

Technical Summary

1. Problem Statement

Cortical volume is a standard biomarker for brain ageing and cognitive decline, but it is a composite measure conflating two distinct morphometric features: cortical thickness and cortical surface area. While these features are genetically independent and follow divergent developmental trajectories, their specific, dissociable contributions to cognitive ageing remain unclear.

• The Gap: Existing literature is inconsistent. Some studies suggest area drives decline, others suggest thickness, and many rely on cross-sectional designs or aggregate volume measures, which obscure distinct mechanisms.
• The Hypothesis: The authors propose a causal framework where cortical thickness reflects dynamic, age-related degenerative processes (mediating the link between age and cognitive decline), while surface area reflects stable, genetically determined neurodevelopmental architecture (mediating the link between polygenic scores and baseline cognitive ability).

2. Methodology

The study employed a multi-cohort, multi-method approach to test these hypotheses across the adult lifespan.

Data Sources:

• Discovery Cohort: Cam-CAN (Cambridge Centre for Ageing and Neuroscience), a population-based lifespan sample ($N \approx 635$ cross-sectional; $N \approx 133$ longitudinal).
• Replication Cohorts: OASIS-3 (Open Access Series of Imaging Studies) and HABS-HD (Health and Aging Brain Study – Health Disparities). These were pre-registered for replication.
• Sample Characteristics: Combined $N > 4,000$ participants, diverse age ranges (18–95 years), and ethnic backgrounds.

Measures:

• Cognition: Fluid intelligence ($g$), derived from principal component analysis (PCA) of various cognitive batteries (e.g., Cattell, DSST, TMT, Fluency tests).
• Neuroimaging: Global mean cortical thickness (mm) and total cortical surface area (mm²) derived from T1-weighted MRI using FreeSurfer.
• Genetics: Polygenic Scores (PGS) for fluid intelligence (available only in Cam-CAN).

Analytical Strategy:

1. Cross-Sectional Parallel Mediation (SEM): Tested whether thickness and area act as distinct mediators between exogenous variables (Age, PGS) and cognition.
2. Longitudinal Mixed-Effects Models: Estimated age-related trajectories (linear and non-linear) for thickness and area, separating between-person differences from within-person change.
3. Correlated Change Analysis: Regressed individual slopes of cognitive change against slopes of cortical change.
4. Extended Longitudinal Models: Tested if cognitive change is predicted by both cortical change and baseline cortical values (controlling for reverse causality).
5. Bivariate Latent Change Score Models (BLCSM) & Latent Growth Models (LGM): Used to test bidirectional, time-lagged effects (e.g., does baseline brain predict future cognitive change, or vice versa?).
6. Regional Analysis: Repeated analyses on specific Regions of Interest (ROIs) to check for spatial heterogeneity.

3. Key Contributions

• Methodological Rigor: The study moves beyond simple volume measures by explicitly modeling thickness and area as separate pathways. It utilizes a "discovery + pre-registered replication" framework across three distinct cohorts.
• Causal Inference Framework: By combining cross-sectional mediation with longitudinal time-lagged models (BLCSM/LGM), the authors provide stronger evidence for the directionality of effects (e.g., distinguishing trait-like stability from state-like decline).
• Resolution of Inconsistencies: The paper reconciles conflicting prior literature by demonstrating that the relationship between cortical metrics and cognition depends on whether one is looking at baseline traits (area) or dynamic change (thickness).

4. Key Results

A. Cross-Sectional Mediation (Static Associations)

• Age $\to$ Cognition: Thickness was a significantly stronger mediator than area. In Cam-CAN, thickness mediated 14% of the age-cognition effect vs. 9% for area. This pattern replicated in OASIS-3 (29% vs. 13%) and HABS-HD (19% vs. 15%).
• PGS $\to$ Cognition: Area mediated the genetic effect (12% in Cam-CAN), while thickness showed no significant mediation. This supports the hypothesis that area reflects stable, genetically influenced architecture.

B. Longitudinal Trajectories (Dynamic Change)

• Decline Rates: Thickness declined linearly and consistently across cohorts ($\approx -0.004$ to $-0.005$ mm/year). Area decline was more heterogeneous and, in some cohorts, less steep until very old age.
• Change-Change Correlation: The rate of thickness decline was positively associated with the rate of cognitive decline (accounting for ~1–3% of variance). In contrast, area change showed weak or non-significant associations with cognitive change.

C. Baseline Predictors of Future Change (Time-Lagged Effects)

• Baseline Thickness: Significantly predicted future cognitive decline, even after controlling for concurrent cortical change and reverse paths (cognition $\to$ brain). This held true across all cohorts and models.
• Baseline Area: Did not predict future cognitive decline in any cohort or model specification.
• Reverse Causality: Higher baseline cognition predicted slower decline in thickness in some cohorts (OASIS-3, HABS-HD), but baseline area never predicted cognitive change.

D. Regional Specificity

• Analyses of specific ROIs (e.g., rostral middle frontal, inferior parietal) confirmed that thickness-cognition associations were directionally consistent across the cortex.
• Area-cognition associations were highly heterogeneous (varying in sign and significance across regions), suggesting area's influence is not uniform.

5. Significance and Implications

• Refining Biomarkers: The study argues that relying solely on cortical volume obscures critical biological signals. Thickness is identified as a sensitive marker for dynamic neurobiological processes underlying cognitive ageing (e.g., synaptic loss, dendritic atrophy), whereas Area reflects stable, trait-like variation in cognitive potential.
• Clinical Relevance: Distinguishing these features improves the accuracy of predictive models for cognitive vulnerability. Thickness decline may serve as an early indicator of pathological or accelerated ageing, while area may be more relevant for establishing normative baselines.
• Theoretical Advancement: The findings support a dual-process model of brain ageing:
  1. Developmental/Genetic Pathway: Early life factors determine surface area, which sets a baseline for cognitive ability.
  2. Degenerative/Ageing Pathway: Ongoing ageing processes drive cortical thinning, which directly correlates with the rate of cognitive decline.
• Future Directions: The authors highlight the need to include subcortical structures and white matter in future models, as cortical measures alone do not capture the full variance of cognitive ageing.

In summary, this paper provides robust evidence that cortical thickness and surface area are dissociable contributors to cognitive ageing, with thickness serving as the primary driver of age-related cognitive decline and area reflecting stable, genetically determined cognitive traits.