Association of Body Fat Distribution Patterns at MRI with Brain Structure, Cognition, and Neurologic Diseases.

Latent profile analysis of MRI-derived fat metrics identified six distinct body fat distribution profiles in both male and female participants.

• Study included 25,997 participants from the UK Biobank (mean age 55 years ± 7.4 SD; 13,536 female participants).
• LPA was based on eight BMI-adjusted MRI-derived fat quantification metrics from brain, heart, and abdominal MRI scans.
• Six profiles were identified in both sexes, including two lean and four high-adiposity patterns.
• Analyses were stratified by sex.
• This was a secondary analysis of prospective UK Biobank data.

The pancreatic-predominant profile (profile 1) was characterized by elevated proton density fat fraction compared to other profiles.

• Mean BMI-adjusted z score for proton density fat fraction was 2.38 ± 0.74 for male participants and 3.01 ± 1.08 for female participants.
• P < .001 for a difference across profiles for both sexes.
• This profile was one of four high-adiposity patterns identified.

The skinny-fat profile (profile 3) was characterized by the highest adiposity burden in the majority of fat depots despite moderate BMI.

• Elevated adiposity was observed in six of eight depots for male participants and five of eight depots for female participants.
• P < .001 for a difference across profiles for each depot for both sexes.
• This profile was notable because high adiposity burden occurred despite moderate BMI levels.

Pancreatic-predominant and skinny-fat profiles were associated with extensive gray matter atrophy compared with the lean profile.

• Profile 1 (pancreatic-predominant): Cohen d = -0.63 for male participants and -0.58 for female participants.
• Profile 3 (skinny-fat): Cohen d = -0.56 for male participants and -0.12 for female participants.
• P < .001 for a difference across profiles for both sexes.
• Comparisons were made relative to a benchmark lean profile.

Pancreatic-predominant and skinny-fat profiles were associated with elevated white matter hyperintensity load compared with the lean profile.

• Profile 1 (pancreatic-predominant): Cohen d = 0.47 for male participants and 0.42 for female participants.
• Profile 3 (skinny-fat): Cohen d = 0.42 for male participants and 0.20 for female participants.
• P < .001 for a difference across profiles for both sexes.

Pancreatic-predominant and skinny-fat profiles were associated with accelerated brain aging compared with the lean profile.

• Cohen d = 0.25 for male participants with profile 1 (pancreatic-predominant).
• Cohen d = 0.32 for male participants with profile 3 (skinny-fat).
• P < .001 for a difference across profiles.
• Brain aging acceleration was a statistically significant finding specific to male participants in this reported comparison.

Pancreatic-predominant and skinny-fat profiles were associated with cognitive decline and increased risk of neurologic disease compared with the lean profile.

• Both profile 1 and profile 3 showed associations with cognitive decline.
• Both profiles showed increased risk of neurologic disorders.
• Group differences were examined using analysis of covariance (ANCOVA) or rank-based ANCOVA.
• Analyses were stratified by sex and conducted relative to a benchmark lean profile.

BMI-adjusted MRI-derived fat quantification metrics revealed heterogeneity in adiposity patterns that are not captured by BMI alone.

• Eight BMI-adjusted MRI-derived fat quantification metrics were used as inputs for LPA.
• The skinny-fat profile demonstrated that individuals with moderate BMI could still carry high adiposity burden across multiple fat depots.
• Fat distribution patterns showed differential associations with neurologic outcomes, indicating that the distribution of fat—not just total adiposity—matters for brain health.