Associations between cognitive performance and Mediterranean dietary pattern in patients with type 1 or type 2 diabetes mellitus

The study addresses whether adherence to a Mediterranean diet (MedD) is associated with better cognitive performance—specifically verbal memory—in individuals with diabetes mellitus, and whether such associations differ by diabetes type (type 1 vs. type 2) and by disease duration (recently diagnosed vs. ≥5 years). Prior evidence suggests MedD benefits cognitive function, largely in healthy older adults or those with elevated cardiovascular risk. However, findings among patients with type 2 diabetes are inconsistent, and it is unclear whether any differential effects exist by diabetes type or early in the disease course. The study aims to clarify these associations using well-characterized cohorts from the German Diabetes Study.

Systematic reviews report that MedD is linked to better global cognition and specific domains like memory and fluency in predominantly healthy elderly or high cardiovascular risk populations. Randomized trials (e.g., PREDIMED substudies) showed memory benefits with MedD supplemented with olive oil or nuts. Evidence in type 2 diabetes-specific cohorts is mixed: some report reduced cognitive impairment risk with higher MedD adherence, while others find benefits primarily among non-diabetic individuals. Mechanistic hypotheses include antioxidant, anti-inflammatory, and vascular benefits, potentially mediated by n-3 fatty acids, though findings on n-3 FA are inconsistent across studies. There is a knowledge gap concerning differences by diabetes type and disease duration.

Design: Cross-sectional analysis of participants in the German Diabetes Study (GDS). Groups included: metabolically healthy individuals (n=41), individuals with type 1 diabetes (total n≈119; analyzed as recently diagnosed n=75 and ≥5 years n=44), and individuals with type 2 diabetes (total n≈180; analyzed as recently diagnosed n=118 and ≥5 years n=61/62). Inclusion required completion of cognition tests, the Multiple Choice Word Test (MWT-B), and a validated food frequency questionnaire (FFQ). Ethics approval was obtained and all participants provided written informed consent. Assessments: - Cognitive testing: Standardized “Brief Assessment of Cognition in Schizophrenia” (BACS) plus additional tests covering multiple domains; verbal memory was the primary domain of interest. - Dietary assessment: Self-administered semi-quantitative 148-item FFQ from EPIC-Potsdam, assessing habitual intake over the prior 12 months. Derived mean total energy intake (TEI; MJ/day) and food intake (g/day). - Mediterranean diet adherence: Modified Mediterranean Diet Scale (MMDS; range 0–9). Using sex- and cohort-specific medians, assigned 1 point for higher intake of beneficial components (vegetables, legumes, fruits, cereals, fish), 0 for higher intake of detrimental components (meat, dairy); 1 point for moderate ethanol intake; 1 point for lipid ratio (unsaturated:saturated fatty acids) at or above median. - Clinical/metabolic data: HbA1c, BMI, fasting C-peptide, lipids, etc. Statistical analysis: - Software: SAS 9.4. Two-sided P<0.05 considered significant. Normally distributed variables as mean±SD, non-normal as median (25th; 75th). - Primary analysis: Multiple linear regression with MMDS (continuous) as predictor and cognitive test outcomes as dependent variables. Model 1 unadjusted; Model 2 adjusted for age, sex, TEI; Model 3 additionally adjusted for crystallized intelligence (MWT-B). Fasting C-peptide was also included in sensitivity analyses (added to Model 3) to adjust for insulin secretion. - Effect illustration: Calculated adjusted means of outcomes by MMDS tertiles (T1–T3). - Interaction: Tested MMDS × diabetes type/duration interaction on verbal memory using multiple linear regression adjusted per Model 3. Group characteristics: Metabolically healthy individuals and type 2 diabetes groups were nominally older and had higher BMI than type 1 diabetes groups. Glycemic control in diabetes groups was generally within guideline-recommended ranges. Mean MMDS across participants was about 4.5. TEI approximated 10 MJ/d in healthy individuals and type 1 diabetes groups, and ~8.6 MJ/d in type 2 diabetes groups.

- Primary outcome: In type 2 diabetes with known duration ≥5 years (n≈61–62), higher adherence to the MMDS was associated with better verbal memory scores after adjustment for age, sex, TEI, and MWT-B (Model 3: P=0.043). Unadjusted and partially adjusted models were also significant (Model 1: P=0.010; Model 2: P=0.026). - Sensitivity analysis adding fasting C-peptide to Model 3 confirmed the association in long-standing type 2 diabetes, with adjusted means by MMDS tertiles: T1: -0.66 (95% CI -1.22; -0.11), T2: -0.87 (-1.27; -0.46), T3: -0.07 (-0.56; 0.41), P=0.042, indicating relatively higher (less negative) verbal memory performance with greater MMDS adherence. - No significant associations between MMDS and verbal memory were found in: recently diagnosed type 2 diabetes (P=0.275), type 1 diabetes (recent: P=0.215; ≥5 years: P=0.626), or metabolically healthy individuals (P=0.666). - Interaction analyses indicated that the positive association between MMDS and verbal memory was specific to patients with type 2 diabetes duration ≥5 years (interaction P=0.024). - Other cognitive domains: No associations were observed between MMDS adherence and other cognitive performance measures in any group.

The findings suggest that closer adherence to a Mediterranean dietary pattern is linked to better verbal memory specifically among individuals with longer-standing type 2 diabetes, but not among those with recent-onset type 2 diabetes, type 1 diabetes at any duration, or metabolically healthy individuals. This pattern aligns with prior evidence that MedD benefits cognition predominantly in populations with elevated cardiometabolic risk and that cognitive changes in type 2 diabetes tend to emerge over time. Potential mechanisms include the antioxidant and anti-inflammatory properties of the MedD, as well as beneficial effects of n-3 fatty acids on neuroinflammation, oxidative stress, and thrombosis, although literature on n-3 fatty acids remains mixed. The selective association with verbal memory mirrors studies showing memory improvements with MedD, potentially reflecting hippocampal sensitivity to diet-related factors. The absence of associations in type 1 diabetes is consistent with earlier observations that verbal memory impairment is more characteristic of type 2 diabetes after several years of disease progression.

Higher adherence to a Mediterranean diet is associated with better verbal memory performance in patients with type 2 diabetes of at least five years’ duration, but not in patients with type 1 diabetes, recently diagnosed type 2 diabetes, or metabolically healthy individuals. These results underscore the potential cognitive benefits of MedD in the later course of type 2 diabetes. Future research should include longitudinal and interventional studies to clarify causality, delineate temporal dynamics across the diabetes trajectory, and explore underlying biological mechanisms.

- Cross-sectional analysis with separate recent-onset and ≥5-year duration groups; lacks within-subject longitudinal follow-up to assess change over time. - Dietary intake measured by FFQ, subject to recall bias and limited granularity; the EPIC-FFQ does not allow assessment of glycemic load for adjustment. - Analysis at the dietary pattern/food group level limits nutrient-specific inference. - Relatively small sample sizes within subgroups, precluding matching for factors like age, sex, and socioeconomic status; residual confounding possible despite adjustments. - Heterogeneity in methods across literature complicates comparability; however, a standard MMDS scoring was used to enhance comparability. - Generalizability may be limited to metabolically well-controlled patients and similar settings.