Hypertension effects on brain volumes in presence of structural cardiac and renal function mediators

272 

Abstract Submission 

Mohamed Salah Khlif¹, Carolina Restrepo², Sheila Patel³, Leonid Churilov⁴, Louise Burrell³, Amy Brodtmann¹

¹Cognitive Health Initiative, Central Clinical School, Monash University, Melbourne, VIC, ²Cabrini Health, Melbourne, VIC, ³Department of Medicine, University of Melbourne, Austin Health, Heidelberg, VIC, ⁴Melbourne Medical School, University of Melbourne, Melbourne, VIC

Mohamed Salah Khlif  
Cognitive Health Initiative, Central Clinical School, Monash University
Melbourne, VIC

Carolina Restrepo  
Cabrini Health
Melbourne, VIC

Sheila Patel  
Department of Medicine, University of Melbourne, Austin Health
Heidelberg, VIC

Leonid Churilov  
Melbourne Medical School, University of Melbourne
Melbourne, VIC

Louise Burrell  
Department of Medicine, University of Melbourne, Austin Health
Heidelberg, VIC

Amy Brodtmann, MBBS FRACP PhD  
Cognitive Health Initiative, Central Clinical School, Monash University
Melbourne, VIC

Hypertension and type 2 diabetes mellitus are two common, closely interlinked, comorbidities that frequently coexist in the same individual due to shared risk factors such as obesity. Hypertension causes cerebral small vessel disease (cSVD) resulting in ischemia, accelerated brain atrophy, and higher accumulation of white matter hyperintensities (WMH) and is associated with renal and cardiac damage. These disease states are associated with increased risk of cognitive impairment and dementia (Canavan & O'Donnell, 2022).
Left ventricular hypertrophy (LVH), as a marker of hypertension and an independent risk factor for cardiovascular disease, has also been associated with WMH (Nagaraja et al., 2022) and cognitive impairment (Georgakis et al., 2017). Thus, LVH may sit on a causal pathway between hypertension and pathological brain aging and cognitive impairment.
We conducted causal mediation analyses (Fig. 1) to determine the extent to which the total effects of hypertension on brain global and regional atrophy and WMH accumulation may be mediated by cardiac (LVH) and renal (estimated Glomerular Filtration Rate, eGFR) metrics.

Baseline multimodal MRI was completed as part of the Diabetes and Dementia (D2) study (Patel et al., 2017). Hypertension (exposure) for 129 participants was determined on 24- hour blood pressure mean > 140/90 mmHg. All participants had eGFR > 30 mL/min/1.73m2; 36 participants were diagnosed with LVH. Total brain volume (TBV) was estimated based on FreeSurfer (v7.3.2) longitudinal segmentation of T1-weighted scans. Hippocampal and thalamic volumes were estimated using the subregion segmentation functionalities in FreeSurfer (Iglesias et al., 2015; Iglesias et al., 2018). WMH volumes were estimated using manual delineations on FLAIR scans. Age, sex, and obesity (BMI > 30 kg/m2) were included as pre-exposure confounders. Finally, the R package "mediation" (Tingley et al., 2014) was used to perform mediation and sensitivity analyses. We used bootstrap for robust effects and confidence intervals estimations.

We included 129 participants (age: 65.1 ± 7.2 years, men: 72) including 97 who had a history of hypertension); 45 participants had current/untreated hypertension at baseline (64.8 ± 6.3 years, men: 26) on ambulatory BP testing. The distributions of covariates and mediators between the hypertension groups at baseline were similar (see Fig. 1). For both LVH and eGFR mediators, analyses revealed no causal mediation between current hypertension and any of the outcome measures: TBV, WMH, or hippocampal and thalamic volumes (Fig. 2). The highest mediated proportion (6.8%) of total effect was for LVH and WMH. The direct and total effects of hypertension on thalamic and WMH volumes were found to be significant (see Fig. 2). Sensitivity analyses showed that it was highly unlikely for the assumption of sequential ignorability to be violated in these cases (ρThalamus at zero effect = 0.85, ρWMH at zero effect = -0.65). Motivated by reported associations between LVH and deep WMH in acute stroke based on multivariate regression (Nagaraja et al., 2022), we completed a causal mediation analysis with deep WMH as the outcome. Again, insignificant mediation effect was found (only 7.3% of total effect).

In this sample of participants with T2DM, we report significant effects of hypertension on thalamic atrophy and WMH burden without mediation from the other renal and cardiac variables considered. We conclude that prior reported effects of LVH on structural brain measures may have been due to unobserved/unmeasured or imbalanced confounders.

Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) ¹

Other Methods ²

Cerebrovascular Disease

Cognition

Computational Neuroscience

Degenerative Disease

STRUCTURAL MRI

Other - WMH, Hypertension, Causal inference, Mediation