Importance  Impairment of methylation status (ie, methionine to homocysteine ratio) may be a modifiable risk factor for structural brain changes and incident dementia.

Objective  To investigate the association of serum markers of methylation status and sulfur amino acids with risk of incident dementia, Alzheimer disease (AD), and the rate of total brain tissue volume loss during 6 years.

Design, Setting, and Participants  This population-based longitudinal study was performed from March 21, 2001, to October 10, 2010, in a sample of 2570 individuals aged 60 to 102 years from the Swedish Study on Aging and Care in Kungsholmen who were dementia free at baseline and underwent comprehensive examinations and structural brain magnetic resonance imaging (MRI) on 2 to 3 occasions during 6 years. Data analysis was performed from March 1, 2018, to October 1, 2018.

Main Outcomes and Measures  Incident dementia, AD, and the rate of total brain volume loss.

Results  This study included 2570 individuals (mean [SD] age, 73.1 [10.4] years; 1331 [56.5%] female). The methionine to homocysteine ratio was higher in individuals who consumed vitamin supplements (median, 1.9; interquartile range [IQR], 1.5–2.6) compared with those who did not (median, 1.8; IQR, 1.3–2.3; P < .001) and increased per each quartile increase of vitamin B₁₂ or folate. In the multiadjusted model, an elevated baseline serum total homocysteine level was associated with an increased risk of dementia and AD during 6 years: for the highest homocysteine quartile compared with the lowest, the hazard ratios (HRs) were 1.60 (95% CI, 1.01-2.55) for dementia and 2.33 (95% CI, 1.26-4.30) for AD. In contrast, elevated concentrations of methionine were associated with a decreased risk of dementia (HR, 0.54; 95% CI, 0.36-0.81) for the highest quartile compared with the lowest. Higher values of the methionine to homocysteine ratio were significantly associated with lower risk of dementia and AD: for the fourth methionine-homocysteine quartile compared with the first quartile, the HR was 0.44 (95% CI, 0.27-0.71) for incident dementia and 0.43 (95% CI, 0.23-0.80) for AD. In the multiadjusted linear mixed models, a higher methionine to homocysteine ratio was associated with a decreased rate of total brain tissue volume loss during the study period (β [SE] per 1-SD increase, 0.038 [0.014]; P = .007).

Conclusions and Relevance  The methionine to homocysteine status was associated with dementia development and structural brain changes during the 6-year study period, suggesting that a higher methionine to homocysteine ratio may be important in reducing the rate of brain atrophy and decreasing the risk of dementia in older adults.