Predicting White Matter Pathology using Cerebrovascular Risk Factors in a UK Biobank Sample
Poster No:
1774
Submission Type:
Abstract Submission
Authors:
Katie Moran1, Daniela Montaldi1, Nils Muhlert1
Institutions:
1The University of Manchester, Manchester, Greater Manchester
First Author:
Co-Author(s):
Introduction:
Cerebrovascular burden is proposed to impact cognitive function through disruption of white matter networks (Hilal et al., 2021). As such, developing predictive models to assess current pathological states and forecast future outcomes is crucial for identifying individuals at high risk of white matter pathology and subsequent cognitive decline later in life. However, vascular risk factors are highly comorbid (Kesarwani et al., 2009). Understanding the interactions between risk factors and their collective contribution to pathology will provide insight into mechanisms and help identify and target key contributors. Here, we adopted partial least squares regression (PLSR) models to assess how these risk factors cluster to predict current white matter pathology, future progression and the rate of change between two time points. This approach aims to improve our understanding of the treatment needs of individuals most at risk for developing white matter pathology.
Methods:
We used a sample of 4123 (M:46%, F:54%, age: 46-81) with complete neuroimaging data across 2 time points. Vascular risk factors were recorded at time point 1, including systolic & diastolic blood pressure, cholesterol, smoking status, waist to hip ratio, BMI and diabetes status. We also quantified risk factor durations where available (smoking, diabetes, hypertension). Risk factors were loaded as predictor variables, alongside sex and age. Outcome variables were total volume of white matter hyperintensity (WMH), and 2 latent constructs of anterior white matter integrity, separately for FA & MD, measured at time point 1 and time point 2 (approx. 3 years apart). Latent constructs were quantified using confirmatory factor analysis (all tracts - p < .001). We also calculated the rate of change between time points for each neuroimaging measure, which were loaded into separate PLSR models. Data were divided into 70% train (with cross validation) and 30% test sets.
Results:
Principal components were selected using cross-validation. For WMH & FA, 2 principal components were derived: 'PC1:Cardiometabolic & Anthropometric (high)' and PC2: 'Anthropometric (low) & Lifestyle' (Figure 1b & 1c). The MD models and time point 2 FA model had 3 latent constructs, which were divided into 'PC1: Cardiometabolic,' 'PC2: Anthropometric (low)' and 'PC3: Anthropometric (high) & Lifestyle' (Figure 1a). The 'Cardiometabolic & Anthropometric (high)' latent construct was loaded positively with risk factors such as systolic blood pressure, hypertension duration, cholesterol, as well waist to hip ratio and BMI. The 'Anthropometric' constructs were either loaded positively (high) or negatively (low) with BMI, Waist to hip ratio & diastolic blood pressure and 'Lifestyle' added smoking risk and number of pack years. Age loaded highly across all constructs.
The results highlight that the models performed reasonably well when predicting current white matter pathology at time point 1 ((R², RMSE) FA: 17%, .93; MD: 20%, .88; WMH: 17%, 1.01) and white matter pathology at time point 2 ((R², RMSE) FA: 20%, .92; MD: 25%, .87; WMH: 18%, .79). In contrast, the model was poor at predicting rate of white matter change between the two time points ((R², RMSE) FA: 2.4%, 1.23; MD: 2.6%, .99; WMH: < 1%, .89).
The results highlight that the models performed reasonably well when predicting current white matter pathology at time point 1 ((R², RMSE) FA: 17%, .93; MD: 20%, .88; WMH: 17%, 1.01) and white matter pathology at time point 2 ((R², RMSE) FA: 20%, .92; MD: 25%, .87; WMH: 18%, .79). In contrast, the model was poor at predicting rate of white matter change between the two time points ((R², RMSE) FA: 2.4%, 1.23; MD: 2.6%, .99; WMH: < 1%, .89).
Conclusions:
We demonstrate that vascular risk factors and demographic variables are reliable predictors of white matter pathology - both at macroscale (WMH) and at microscale (DTI metrics). Specifically, high levels of cardiometabolic and anthropometric risk factors pose the greatest risk for the development of white matter pathology, but low levels of anthropometric factors are also evidenced to contribute substantially. This is consistent across the two time points, but not the rate of change, for all neuroimaging measures. Future research should focus on establishing the optimal parameters for risk factor intervention across the lifespan, with a specific focus on high cardiometabolic and anthropometric risk factors.
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Connectivity (eg. functional, effective, structural)
Diffusion MRI Modeling and Analysis
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
White Matter Anatomy, Fiber Pathways and Connectivity 1
Keywords:
Cerebrovascular Disease
Modeling
White Matter
Other - UK Biobank
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
Kesarwani, M., Perez, A., Lopez, V. A., Wong, N. D., & Franklin, S. S. (2009). Cardiovascular comorbidities and blood pressure control in stroke survivors. Journal of hypertension, 27(5), 1056-1063.