Abstract Details

Title Smoking Is Associated with Accelerated Brain Aging: A Large, Diverse Population-based Analysis

Topic General Neurology

Presentation(s) S43 - General Neurology 2 (1:24 PM-1:36 PM)

Poster/Presentation
Number 003

Background Machine learning has made significant success in accurately predicting individuals' ages from brain MRI scans and enabled analysis of large datasets. Research has shown higher predicted brain ages are linked to accelerated brain aging and brain abnormalities. However, studies on smoking’s impact on brain aging are scarce, particularly in a large and diverse cohort.

Objective To examine the relationship between cigarette smoking and brain health in a diverse, large population.

Design/Methods We used a deep learning model for brain age estimation, initially pretrained on T1-weighted scans of general population (N=53,542; https://doi.org/10.1016/j.neuroimage.2022.119210) and further fine-tuned and validated on our in-house T1 scans (N=1,559). Brain ages were estimated for 2,853 ever-smokers (past and current smokers) and 6,455 never-smokers, aged 30-70. We examined associations between smoking variables (pack-years, packs-per-day, total smoking years) and brain age using multiple linear regression adjusted for age, sex, ethnicity, daily alcohol consumption, stage-2 hypertension, diabetes, history of cerebro/vascular events and respiratory illness. Additional analysis explored ever-smokers' smoking habits on brain volume.

Results Pack-years had a significant effect on brain age (β=0.037;p=2.33e-13). In subsequent analyses, we found the number of packs-per-day (β=0.62;p=1.80e-05) had the second-largest impact on accelerated brain aging, followed by hypertension (β=0.68;p=3.52e-06). Diabetes (β=0.54;p=1.03e-04), alcohol consumption (β=0.48;p=3.47e-34), and smoking years (β=0.013;p=0.015) were also significantly associated with increased brain age. No ethnicity-specific effects of smoking were found in our study cohort (62.0% Caucasians/16.8% Mixed/16.4% Asians/2.6% Arab and Africans/2.2% Others). For a subset of 1,426 ever-smokers with available brain volume data, packs-per-day had a substantial effect (β=-7.6;p=4.66e-03), leading to more than three times greater brain volume reduction compared to age (β=-2.2;p=1.78e-54). Smoking years, smoking start age, and years of smoking cessation did not show significant associations with brain volume.

Conclusions Our results highlight packs-per-day had a greater impact on accelerated brain age than total smoking years resulting in notably reduced brain volumes.

Authors/Disclosures
Soojin Lee (Prenuvo) PRESENTER	No disclosure on file
Saurabh Garg (Prenuvo)	No disclosure on file
Madhurima Datta (Prenuvo)	No disclosure on file
Thanh Nguyen (Prenuvo)	No disclosure on file
Nasrin Akbari (Prenuvo)	No disclosure on file
Arun Rajendran (Prenuvo)	No disclosure on file
Sam Hashemi (Prenuvo)	Mr. Hashemi has received personal compensation for serving as an employee of Prenuvo.
Yosef Chodakiewitz (Prenuvo)	No disclosure on file
Raj Attariwala, MD, PhD	Dr. Attariwala has received personal compensation in the range of $50,000-$99,999 for serving as a Consultant for Prenuvo. Dr. Attariwala has stock in Prenuvo. Dr. Attariwala has received research support from Voxelwise Imaging Technology. Dr. Attariwala has received personal compensation in the range of $100,000-$499,999 for serving as a Radiologist reading cases and helping build in house software tools with Prenuvo.