Risk factors and cognitive correlates of white matter hyperintensities in ethnically diverse populations without dementia: The COSMIC consortium

K.n. Lin, Wei Wen, Darren M. Lipnicki, Louise Mewton, Rory Chen, Jing Du, Dadong Wang, Ingmar Skoog, Therese Rydberg Sterner, Jenna Najar, Ki Woong Kim, Ji Won Han, Jun Sung Kim, Tze Pin Ng, Roger Ho, Denise Qian Ling Chua, Kaarin J. Anstey, Nicolas Cherbuin, Moyra E. Mortby, Henry Brodaty, Nicole A. Kochan, Perminder S. Sachdev, Jiyang Jiang

TL;DR

By pooling data from five international cohorts, researchers found that while vascular issues like hypertension and smoking universally increase white matter hyperintensities (WMH), their impact varies significantly by ethnicity. Most notably, hypertension and stroke are more strongly associated with brain aging in White populations than in Asians, suggesting that "one-size-fits-all" neurological risk profiles may be incomplete.

What Problem Does This Paper Address?

As we age, MRI scans often reveal "bright spots" in the brain’s white matter, known as White Matter Hyperintensities (WMH). These are key markers of cerebral small vessel disease and are linked to cognitive decline and stroke.

While we know that factors like high blood pressure and diabetes contribute to these lesions, most research has focused on single, often homogenous, populations. This creates a significant gap: we don’t fully understand if the "biological price" of a risk factor—like smoking or hypertension—is the same across different ethnicities and geographical regions. The COSMIC consortium (Cohort Studies of Memory in an International Consortium) set out to harmonize data from around the globe to see if our ethnic background changes how our lifestyle affects our brain health.

Methods

The researchers analyzed MRI and clinical data from 1,946 non-demented older individuals across four countries. To make this "apples-to-apples" comparison possible, they employed several sophisticated techniques:

  • UBO Detector: An automated pipeline that uses a k-nearest neighbor classifier to segment and quantify WMH volumes from FLAIR and T1 MRI scans.
  • ComBat Harmonization: Since the data came from seven different MRI scanners (ranging from 1.5T to 3T), they used the ComBat algorithm to remove "batch effects"—statistical noise caused by different hardware rather than biological differences.
  • Structural Equation Modeling (SEM): This allowed the team to model the complex relationships between vascular risk factors, WMH volumes (categorized into periventricular and deep white matter), and cognitive domains simultaneously.

Datasets

The study utilized five major population-based cohorts:

  1. KLOSCAD (South Korea)
  2. H70-study (Sweden)
  3. SLAS-I & II (Singapore)
  4. MAS (Sydney, Australia)
  5. PATH (Canberra/Sutherland, Australia)

This mix provided a robust comparison between White (Swedish and Australian) and Asian (Korean and Singaporean) subsamples.

Key Insights and Ideas

  • The Ethnicity Interaction: The most striking finding was that hypertension and stroke had a significantly stronger association with WMH volumes in White participants than in Asians. This suggests that Asian populations might have different compensatory mechanisms or that the "threshold" for vascular damage differs.
  • The Cholesterol Paradox: In the Asian subsample, hypercholesterolemia was actually associated with lower WMH volumes. The authors hypothesize this isn't because high cholesterol is "good," but rather a "statin effect." Asians often show a stronger response to cholesterol-lowering medications at lower doses due to different pharmacokinetics, which may provide a neuroprotective side effect.
  • The Physical Activity "Sweet Spot": You don't need to be an elite athlete to protect your brain. The data showed that "occasional" exercisers had significantly less WMH than sedentary individuals, but there was no major additional benefit found in the "frequent" exercise group.
  • Location Matters: Deep white matter hyperintensities (DWMH) were more strongly tied to cognitive deficits in processing speed and executive function than those located near the brain's ventricles.

Key Results

  • Universal Risks: Diabetes, hypertension, stroke, current smoking, high BMI, and high alcohol intake were all confirmed as drivers of larger WMH volumes across the entire sample.
  • Cognitive Impact: Higher volumes of deep WMH were associated with slower processing speed ($\beta = -0.067, p = 0.026$) and poorer executive function ($\beta = -0.063, p = 0.037$).
  • Quantitative Ethnic Split: In White participants, the association between hypertension and whole-brain WMH was significant ($\beta = 0.074, p = 0.001$), whereas, in the Asian subsample, the association was much weaker and not statistically significant ($\beta = 0.016, p = 0.115$).

Limitations and Future Work

The authors acknowledge several hurdles. First, the study is largely cross-sectional, meaning they can't definitively say these risk factors caused the WMH progression over time. Second, harmonizing self-reported data (like physical activity levels) across different cultures is notoriously difficult.

Future research needs to include low- and middle-income countries, as the Asian cohorts in this study (Singapore and South Korea) are from high-income regions. Additionally, more granular data on statin use and specific types of cholesterol (HDL vs. LDL) will be required to solve the "cholesterol paradox" observed in this study.

Summary updated

Abstract

The current study highlights ethnic differences in the contributions of vascular risk factors to WMHs.
BibTeX
@inproceedings{lin2024risk,
  author = {K.N. Lin and Wei Wen and Darren M. Lipnicki and Louise Mewton and Rory Chen and Jing Du and Dadong Wang and Ingmar Skoog and Therese Rydberg Sterner and Jenna Najar and Ki Woong Kim and Ji Won Han and Jun Sung Kim and Tze Pin Ng and Roger Ho and Denise Qian Ling Chua and Kaarin J. Anstey and Nicolas Cherbuin and Moyra E. Mortby and Henry Brodaty and Nicole A. Kochan and Perminder S. Sachdev and Jiyang Jiang},
  title = {{Risk factors and cognitive correlates of white matter hyperintensities in ethnically diverse populations without dementia: The COSMIC consortium}},
  booktitle = {Alzheimer},
  year = {2024},
  doi = {10.1002/dad2.12567},
}
Figure