Association of Long-Term Vascular Risk Factor Variability from Childhood and Midlife Cognitive Function: The Bogalusa Heart Study

Technical Summary: Association of Long-Term Vascular Risk Factor Variability from Childhood and Midlife Cognitive Function

Problem Statement
While midlife cognitive function (CF) is a critical predictor of late-life dementia, the mechanisms driving heterogeneity in cognitive trajectories during this period remain incompletely understood. Previous research has established that variability in vascular risk factors (VRFs) during adulthood is associated with increased dementia risk, independent of mean levels. However, it is unclear whether long-term variability in VRFs beginning in childhood influences midlife CF. The "two-hit" vascular hypothesis suggests that early-life exposure to high VRF levels causes endothelial dysfunction (first hit), leading to amyloid and tau accumulation (second hit) by midlife. Yet, the specific role of variability (fluctuation) in these risk factors from childhood through midlife, distinct from their mean levels, has not been extensively evaluated.

Methodology
This study utilized data from the Bogalusa Heart Study (BHS), a longitudinal community-based cohort initiated in 1973. The analytic sample consisted of 1,010 participants (mean age 48 years at cognitive assessment; 61% women; 34% Black) who had at least three VRF measurements from childhood to midlife and a complete neuropsychological assessment at midlife.

• Exposures: The study evaluated long-term variability in four VRFs: systolic blood pressure (SBP), fasting glucose, non-high-density lipoprotein cholesterol (non-HDL), and body mass index (BMI). Variability was quantified using four distinct indices to capture different aspects of fluctuation:
  • Standard Deviation (SD) and Coefficient of Variation (CV): Measuring variability around the mean.
  • Deviation from age-predicted values (DEV) and Residual Standard Deviation (RSD): Measuring variability around age-adjusted slopes, accounting for non-linear trends over time using linear mixed-effect models.
  • Each index was categorized into tertiles (lowest, middle, highest).
• Outcomes: Cognitive function was assessed using a Global Cognitive Score (GCS), derived from eight standardized neuropsychological tests. Three specific cognitive domains were also analyzed: executive function, episodic memory, and attention/processing speed. Scores were standardized by age, race, and sex.
• Statistical Analysis: Multivariable linear regression models were employed, with the lowest tertile of variability serving as the reference. Models were progressively adjusted for covariates including smoking, comorbidities (hypertension, diabetes, dyslipidemia, obesity), life-course mean VRF levels, socioeconomic status (education parameter, employment, depressive symptoms), and APOE ε4 status. Sensitivity analyses excluded participants on VRF-lowering medications.

Key Results
The study found that long-term VRF variability from childhood is associated with midlife cognitive function, independent of mean VRF levels and relevant covariates.

• SBP and Glucose Variability: Higher variability in SBP and glucose was consistently associated with poorer cognitive outcomes.
  • Participants in the highest tertile of SBP variability (measured by DEV) and glucose variability (measured by SD and CV) exhibited significantly lower GCS compared to the lowest tertile.
  • A graded pattern was observed: middle and highest tertiles of SBP and glucose variability were associated with poorer episodic memory and executive function, with effect sizes increasing as variability increased.
• BMI Variability: The association between BMI variability and CF was bidirectional and non-linear.
  • Moderate (middle tertile) BMI variability was associated with better GCS and specific domain performance (executive function, attention/processing speed).
  • High (highest tertile) BMI variability was associated with poorer executive function, though this association was attenuated after adjusting for life-course mean BMI, socioeconomic status, and APOE ε4 status.
• Non-HDL Variability: The association was non-linear. Only the middle tertile of non-HDL variability (measured by CV) was associated with poorer executive function. The highest tertile showed no significant association with cognitive domains.
• Robustness: Sensitivity analyses excluding participants on medication yielded similar results, supporting the robustness of the findings.

Significance and Claims
The authors claim that this study extends the exposure window for VRF variability research beyond adulthood, demonstrating that variability patterns established in childhood persist and influence midlife cognitive trajectories.

• Novelty: Unlike prior studies focusing on older adults or single time points, this research utilizes a unique dataset with repeated measures from childhood to evaluate how disruptions in VRF homeostasis over the life course shape midlife cognition.
• Mechanistic Insight: The findings suggest that maintaining stable VRF levels from early life may be crucial for brain health. The bidirectional and non-linear associations observed for BMI and non-HDL suggest the existence of threshold effects, where moderate variability may be benign or even protective (potentially reflecting healthy lifestyle changes), while extreme variability becomes detrimental.
• Clinical Implication: The study posits that targeting the variability of vascular risk factors, rather than solely focusing on mean levels, starting from early life, could be a viable strategy for preventing late-life cognitive dysfunction. The authors emphasize that monitoring variability alongside mean levels may help detect residual cognitive risk before dyslipidemia or other conditions become clinically apparent.

The paper concludes modestly, noting that as an observational study, causal inference is limited, and further research is needed to define optimal intervention windows and thresholds, particularly for BMI and non-HDL variability.