Table 3.

Summary of longitudinal studies on physical activity and white matter hyperintensities published in English through March 18, 2014.

Citation	Design	Sample	Physical activity	White matter hyperintensities (WMH)	Statistical analysis	Results
Carmelli et al. (1999)	Longitudinal	U.S. World War II veterans, monozygotic white male twins, n = 74, age 68–79 at baseline.	Interviewer-administered questionnaire at baseline.	Measured over 25 years after physical activity on scanners at four study sites. Image evaluation based on a semi-automated segmentation analysis involving operator-guided removal of non-brain elements.	Univariate correlation	No association
Gow et al. (2012)	Longitudinal	Lothian, U.K., n = 691, 53% male, mean age 70 (SD = 0.8) at baseline.	Self-report questionnaire measured at baseline. Physical activity rated on a 6-point scale, from “moving only in connection with necessary (household) chores” to “keep-fit/heavy exercise or competitive sport several times per week.”	Measured 3 years after physical activity. WMH calculated semi-automatically fusing 2 previously aligned structural MRI sequences. Stroke lesions were extracted from total WMH, which were normalized by intracranial volume. WMH were also rated using FLAIR & T2-weighted sequences and the Fazekas scale coded for periventricular and deep WMH separately in right and left hemispheres, and subsequently combined into total WMH ranging from 0 to 6 with ↑ score = ↑WMH.	Linear regression, adjusted for age, sex, social class, IQ, dementia risk & disease history (hypertension, cardiovascular disease and stroke).	A higher level of physical activity predicted lower WMH (standardized β = 0.09, non-standardized β = 0.09, p = 0.029), when WMH was measured semi-automatically.
Podewils et al. (2007)	Longitudinal	U.S., population-based, n = 179; 38% male, mean age 77 (SD = 6), 59 had Alzheimer's disease, 60 had mild cognitive impairment, and 60 cognitively stable, 3 group frequency matched by 5-year age category and sex.	Modified Minnesota Leisure Time Questionnaire, covered 2 weeks prior to baseline interview. Total energy expenditure in kilocalories/week was categorized as 271–759, 760–1874 and > 1875.	Measured 6–7 years after physical activity. Periventricular and deep WMH measured with standardized semi-quantitative rating scale by a single reader blinded to physical activity and covariates.	Multiple linear regression adjusting for age, sex, ethnicity, years of education, APOE & corresponding region-specific baseline WMH score.	Physical activity > 1875 kcal/week was significantly associated with less periventricular (β = 0.85, CI = 0.32, 1.4) and deep (β = 0.69, CI = 0.01, 1.4) WMH in the cognitively stable group only.
Rovio et al. (2010)	Longitudinal	North Karelia & Kuopio, eastern provinces of Finland, random population-based sample, n = 75, 32 active, 43 sedentary; 21 with dementia, 23 with mild cognitive impairment & 31 normal controls; 29.7% male, mean age at midlife was 50.6 (SD = 6.0) years, and 71.6 (SD = 4.1) at re-examination.	Self-administered questionnaire at midlife: “How often do you participate in leisure time physical activity that lasts at least 20–30 min and causes breathlessness and sweating?” Active persons participated in leisure time physical activity ≥ 2 ×/week; sedentary persons participated in leisure time physical activity ≤ 1 ×/week.	Measured with a mean duration follow-up of 20.9 (SD = 4.9) years. A semi-quantitative visual rating scale was used by a single trained rater blinded to clinical data. The total score was used to address the participants belonging to the upper quintile of the distribution to one group (severe WMH) and persons belonging to all other quintiles to another group (no or mild WMH).	Logistic regression unadjusted (crude model) and after that adjusted for age, sex, diagnosis of dementia or mild cognitive impairment, years of education and follow-up time, systolic blood pressure, total serum cholesterol, BMI, APOE 4 allele carrier status, and smoking. Furthermore, all analyses were additionally adjusted for white matter volume.	Crude analysis significant (OR = 4.65, p = 0.03). No longer significant when control for socio-demographic and vascular factors (OR = 4.20, p = 0.32).
Willey et al. (2011)	Longitudinal	Northeast U.S., community-based cohort, n = 1238, 40% male, 65% Hispanic, mean age 70 ± 9 SD at baseline Exclusion criteria: history of stroke.	Self-report duration and frequency of various leisure time/recreational activities for the 2 weeks prior to interview categorized by quartiles of the metabolic equivalent (MET) score; VO2 max.	Measured a mean of 6 ± 3 years after physical activity assessment. WMH was measured semi-automatic, corrected for total cranial volume, and log-transformed. Analyses were performed blind to participant identifying information.	Linear regression unadjusted and adjusted models with demographics (age, sex, race–ethnicity, and education) and vascular risk factors (systolic blood pressure, diastolic blood pressure, glomerular filtration rate, diabetes mellitus, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, moderate alcohol use, and smoking) were constructed.	No association