Older adults with elevated blood biomarkers of Alzheimer disease pathology or broader neurobiological risk who more closely adhered to a diet with lower inflammatory potential had a lower risk of dementia in a longitudinal cohort study, although the strength of the findings varied by biomarker and analysis.
The findings, published in JAMA Network Open, suggest that blood biomarkers may help identify patients for future diet-based dementia prevention studies. The study was observational, used biomarker thresholds derived from a prior analysis of the same cohort, and does not establish that dietary changes reduced dementia risk.
Researchers analyzed data from 1,865 patients aged 60 years or older without dementia who were enrolled in the Swedish National Study on Aging and Care in Kungsholmen. The cohort had a mean age of 71 years, and 60% of patients were female.
At baseline, researchers measured serum biomarkers reflecting Alzheimer disease pathology and broader neurobiological risk. Phosphorylated tau at threonine 217 was used as a marker of Alzheimer disease–related pathology. Neurofilament light chain and glial fibrillary acidic protein were used as nonspecific markers of neuronal injury, neurodegeneration, and astrocyte activation or damage.
Diet quality was assessed using a validated food frequency questionnaire and repeated dietary data collected over 6 years. Researchers evaluated 3 dietary patterns: the Alternate Mediterranean Diet, the Alternative Healthy Eating Index, and the reversed Empirical Dietary Inflammatory Index. Higher scores indicated healthier diet quality, including lower inflammatory potential for the reversed Empirical Dietary Inflammatory Index.
The primary outcome was incident all-cause dementia, identified through clinical evaluations, medical records, and death certificates. Alzheimer disease–related dementia was a secondary outcome. Patients were followed for up to 15.9 years, with a mean follow-up of 8.4 years. During follow-up, 240 patients developed dementia.
In adjusted Cox regression models, each 1–z-score increase in reversed Empirical Dietary Inflammatory Index adherence was associated with a lower hazard of dementia among patients with elevated biomarker levels. The association was strongest among patients with elevated phosphorylated tau at threonine 217, among whom higher adherence was associated with a 29% lower hazard of dementia (hazard ratio, 0.71; 95% CI, 0.58-0.88). The corresponding estimates were 21% lower among patients with elevated neurofilament light chain (hazard ratio, 0.79; 95% CI, 0.66-0.95) and 27% lower among those with elevated glial fibrillary acidic protein (hazard ratio, 0.73; 95% CI, 0.60-0.89).
The researchers reported that the reversed Empirical Dietary Inflammatory Index showed the most consistent inverse associations among patients with elevated biomarker levels. The interaction findings, however, were less uniform than the hazard estimates alone suggest. For this dietary pattern, evidence that the diet-dementia association differed by biomarker level was statistically significant only for phosphorylated tau at threonine 217; the interaction tests for neurofilament light chain and glial fibrillary acidic protein did not reach significance.
The pattern also differed by dietary score. Higher adherence to the Alternate Mediterranean Diet and the Alternative Healthy Eating Index was generally associated with lower dementia risk among patients with lower biomarker levels rather than among those with elevated biomarker concentrations. The researchers reported similar patterns for Alzheimer disease–related dementia.
The researchers cautioned that the study evaluated multiple dietary patterns across multiple biomarkers. They reported both nominal and Bonferroni-adjusted significance thresholds, and not all findings met the stricter threshold. The authors noted that the Alternative Healthy Eating Index findings were less consistent than the reversed Empirical Dietary Inflammatory Index findings, since the main Cox model and spline analyses for the Alternative Healthy Eating Index were not fully aligned.
Ancillary analyses added another layer of nuance. When researchers evaluated 10-year dementia outcomes while accounting for the competing risk of death, healthier dietary patterns were associated with less dementia-free time lost among some patients with elevated biomarker levels. These analyses were consistent with an association in higher-risk patients, especially those with elevated phosphorylated tau at threonine 217, but they did not fully mirror the primary Cox model pattern, in which the Alternate Mediterranean Diet and Alternative Healthy Eating Index associations were generally stronger in lower-biomarker groups.
Sensitivity analyses excluding patients with likely misreported dietary intake, cognitive impairment without dementia, or early dementia onset produced comparable findings. Additional analyses using alternative biomarker cutoffs and an alternative dietary inflammation index also yielded similar results.
Subgroup analyses suggested that some associations differed by age, apolipoprotein E epsilon 4 carrier status, and sex. The researchers described these findings as exploratory and cautioned that they should be interpreted carefully given the number of comparisons.
The researchers noted several limitations. Dietary intake and several covariates were self-reported, raising the possibility of measurement error and residual confounding. Biomarkers were measured only at baseline, and dietary data were unavailable beyond the 6-year assessment period. The biomarker cutoffs were developed in a previous study of the same cohort, meaning the "elevated" biomarker groups should not be interpreted as equivalent to clinically validated Alzheimer disease biomarker-positive groups across assays or practice settings.
Missing data may also have affected the findings. From the initial cohort of 3,363 patients, researchers excluded 778 patients because of missing serum biomarker data and 383 because of missing baseline dietary information. Patients excluded from the analysis were generally older, less educated, and had more comorbidities than those included.
Generalizability may be limited. The cohort consisted of community-dwelling, urban, relatively highly educated and affluent older adults from Stockholm, Sweden. Race and ethnicity data were not collected, although the researchers noted that most patients were likely White.
In an accompanying invited commentary, Sokratis Charisis, MD, of Massachusetts General Hospital, and Nikolaos Scarmeas, MD, of the National and Kapodistrian University of Athens Medical School, wrote that the study may help inform future dietary intervention trials in patients with underlying neurobiological risk.
They noted that although observational studies have linked Mediterranean and anti-inflammatory diets with cognitive benefits, randomized clinical trials of dietary interventions for cognitive outcomes have largely reported null findings. One possible explanation, they wrote, is that prior trials may have enrolled generally healthy patients with little prevalent neuropathology, making measurable cognitive benefits harder to detect during typical follow-up periods.
"This finding highlights the potential relevance of healthy dietary habits as secondary prevention strategies," Dr. Charisis and Dr. Scarmeas wrote. The study, they added, "represent[s] an important step in this direction, laying the groundwork for future clinical trials" of dietary interventions for cognitive decline and dementia prevention.
"These results reinforce the importance of dietary interventions for dementia, not only for the general population but also for individuals already at elevated disease risk, and support the development of precision public health strategies and personalized dietary recommendations in clinical practice," wrote lead study author Anja Mrhar, MSc, of Karolinska Institutet and the University of Ljubljana, and colleagues.
Disclosures: The study researchers reported no conflicts of interest. Dr. Scarmeas reported receiving institutional grants from Novo Nordisk and Bristol Myers Squibb and serving on a data safety monitoring board for PRInnovation GmbH PRimus AD outside the submitted work. No other commentary disclosures were reported.
Source: JAMA Network Open
