Older adults with diabetes who are not obese face significantly higher dementia risk than those without the chronic condition, according to new findings presented at the Gerontological Society of America (GSA) 2025 Annual Scientific Meeting in Boston.

The findings suggest that cognitive risk related to diabetes may be less about excess weight and more about the underlying metabolic stress of the disease — raising concerns that a clinically overlooked group may be at a greater risk than previously believed.

“These results really challenge the assumption that obesity is the primary driver of cognitive complications in diabetes,” said Junyu Sui, PhD student in nursing and health innovation at the Edson College of Nursing and Health Innovation at Arizona State University in Phoenix, who led the study. “For clinicians, the key takeaway is that a lower BMI does not mean lower risk. Nonobese, diabetic patients may actually need more proactive cognitive screening and more aggressive management strategies.”

Sui and his team analyzed data from the All of Us Research Program out of the National Institutes of Health. Diabetes and dementia cases were identified using Ninth and 10th Revision of the International Classification of Diseases codes, and obesity was classified using BMI thresholds and waist circumference cutoffs. The associations remained consistent whether obesity was defined by BMI or waist circumference.

The retrospective analysis of more than 268,000 US adults aged over 50 years followed for a median of 10 years; 2216 (0.8%) developed dementia. People who were not obese showed the highest risk of developing dementia.

Individuals with type 1 diabetes who were not obese showed a fourfold increased risk for dementia (hazard ratio [HR], 4.0; 95% CI, 2.86-5.6) compared with adults without diabetes. Type 2 diabetes without obesity was associated with an 82% higher risk (HR, 1.82; 95% CI, 1.45-2.28) after adjusting for demographic, socioeconomic, and health factors.

“This suggests a unique metabolic vulnerability in nonobese diabetes that warrants much closer clinical attention,” Sui said.

People with obesity and either type of diabetes showed a lower risk, including those with type 1 (HR, 1.69; 95% CI, 1.36-2.12) or type 2 (HR, 1.39; 95% CI, 1.24-1.55).

Fadi Ramadan, MD, geriatrician and assistant professor of medicine at Tufts Medical Center in Boston, said the biological pathways linking diabetes to dementia are well established.

“Diabetes is among the vascular risk factors,” said Ramadan, who was not involved with the research. “We also [see] people with heart disease, and they’re not obese; they have dementia. So I think any vascular risk factor can contribute to dementia risk.”

Ramadan said compounds produced by diabetes can accumulate in blood vessels and disrupt normal circulation and are “linked to diabetic dementia because they contribute to neuroinflammation, oxidative stress, [and] protein dysfunction in the brain.”

Sui described similar mechanisms in his presentation, noting that what he called “lean diabetes” may reflect severe metabolic dysfunction despite normal body size.

Ramadan said the results are “very important” because a clinical gap persists in screening and catching these patients. Screening tools — such as the Montreal Cognitive Assessment — would detect cognitive impairment in this population if they were used consistently, he said.

“Endocrinologists don’t really screen for dementia. For primary care physicians, there’s no time,” said Ramadan. “Diabetics fall through the cracks when it comes to dementia.”

The US Preventive Services Task Force does not recommend routine screening for cognitive impairment in asymptomatic adults.

Nancy Schoenborn, MD, associate professor of medicine in the Division of Geriatric Medicine and Gerontology at Johns Hopkins University School of Medicine in Baltimore, described the study as thought-provoking but preliminary.

“I think this is interesting data,” Schoenborn said, “but not to a point that’s mature enough to impact clinical practice.”

She added that the findings should spur further investigation.

“I’d wonder if this would lead to more research to understand what is happening biologically that’s different in obese versus nonobese patients and if we should treat the nonobese diabetes patients differently,” Schoenborn said.

Sui and Ramadan reported having no relevant disclosures.

Lara Salahi is a health journalist based in Boston.