Prospective Cohort Study

B - Click here for explanation of classification scheme.

POSITIVE: See Quality Criteria Checklist below.

The goal of this study was to examine the relationship between EPA and DHA plasma levels and cognitive decline over seven years in a cohort of elderly community dwellers. Cognitive decline was assessed by performance on tests of global cognition, verbal fluency, working memory and executive function, while considering depressive symptoms and ApoE-ε4 status.

• Non-institutionalized
• Aged 65 years and older
• Lived in the French city of Bordeaux
• All participants signed informed consent.

Not described

Recruitment

Not described

Design

This study was part of the three-city study, a prospective cohort study following vascular risk factors for dementia in 9,294 community dwellers in three French cities. The present study examines the Bordeaux sample of the three-city study (N=2,104). Baseline and two, four, and seven year follow-up assessments were completed on this cohort. Participants were administered a variety of tests as a measure of cognitive decline over seven years. Associations between EPA and DHA levels and cognitive decline were examined, taking into account depressive symptoms and ApoE-ε4 carrier status.

Statistical Analysis

• Two-sided Student's tests: Compared mean plasma EPA and DHA and change in cognitive scores (by ApoE-ε4 carrier status and depressive symptoms)
• Linear mixed models: Used to analyze changes in cognition over time from baseline
• Associations of covariates with cognitive evolution from baseline were evaluated for each cognitive test both on the baseline mean score level and slope of cognitive change over time. 

Covariates

• Plasma EPA or DHA proportions transformed into Z scored
• Depressive symptoms (high vs. low).

Adjustment factors

• Age at baseline
• Gender
• Educational level
• Marital status
• ApoE-ε4
• BMI
• Weight loss more than 3kg
• Smoking
• Alcohol consumption
• History of cardiovascular disease
• Diabetes
• Medication use
• Anti-depressant use
• Plasma triglycerides
• Plasma vitamin E.

Timing of Measurements

• Baseline: 1999-2000
• Follow-up examinations were performed two, four and seven years after baseline
• At baseline, sociodemographic information, lifestyle characteristics, medical symptoms, complaints, and conditions, neuropsychological testing, physical examination and blood sampling were performed
• Cognitive assessments (MMSE, IST, BVRT, and TMT-A and TMT-b) were performed at baseline and at least two follow-up visits
• Depressive symptoms measured at baseline
• Plasma fatty acids assessed at baseline.

Dependent Variables

• Cognitive assessments:
  • MMSE (index of global cognitive performance)
  • Isaacs Set Test (IST; assessment of semantic verbal fluency)
  • Benton Visual Retention Test (BVRT; assessment of working memory)
  • Trail-Making Test A and B (TMT-A and TMT-B; evaluation of executive functioning)
• Assessment of depressive symptoms: CES-D (20-item scale to measure depressive symptoms in elderly; scores zero to 60; score of 17 or more in men and 23 or more in women was considered high depressive symptoms).
• Other variables:
  • ApoE-ε4 allelle carrier status:
    • Measured genetic risk for dementia;
      • Carrier: Presence of at least one allele
      • Non-carrier: No allele
• Visual impairment: Assessed at baseline with Parinaud test
• Plasma vitamin E concentration: Measured by HPLC; marker for oxidative stress
• Plasma triglycerides: Used as a proxy for overall lipidic status.

Independent Variables

• Assessment of plasma fatty acids:
  • Percent EPA and percent DHA: Calculated as percentage of total fatty acids
  • Fasting blood samples collected at baseline visits.

Control Variables

• Age at baseline
• Gender
• Educational level
• Marital status
• ApoE-ε4 status
• BMI
• Weight loss more than 3kg
• Smoking
• Alcohol consumption
• History of cardiovascular disease
• Diabetes
• Medication use
• Anti-depressant use (yes vs. no)
• Plasma triglycerides
• Plasma vitamin E.

• Initial N: 1,343  
• Attrition (final N): 1,228 (those non-demented at baseline and completed the MMSE at least once at follow-up; percent men=38.7%
• Age: Mean age: 74.7 years (4.9 SD)
• Ethnicity: French residents of Bordeaux
• Other relevant demographics: Non-demented subjects
  • ApoE-ε4 carriers: N=235, 19.3%
• Anthropometrics: Mean BMI=26.4kg/m² (4.2 SD)
• Location: Bordeaux, France.

 

Key Findings

Annual change in cognitive scores according to baseline depressive symptoms and ApoE-ε4 status:

• MMSE:
  • High depressive symptoms were significantly associated with lower baseline mean MMSE scores; high depressive symptoms: MMSE score=26.80 (SD=2.00); low depressive symptoms: MMSE score=27.60 (SD=2.00) (P<0.001)
  • High depressive symptoms were significantly associated with higher average annual decline in MMSE; high depressive symptoms: MMSE average annual change = -0.41 (SD=1.07); low depressive symptoms: MMSE average annual change = -0.12 (SD=0.55) (P<0.001)
  • MMSE scores did not differ significantly by ApoE-ε4 status at baseline
  • Higher annual declines in MMSE scores were observed in ApoE-ε4 carriers than non-carriers; high depressive: Average decline = -0.21 (SD=0.64); low depressive = -0.12 (SD=0.60) P<0.05
• IST:
  • High depressive symptoms were significantly associated with lower mean IST scores at baseline; high depressive: IST score=28.10 (SD=6.90); low depressive IST score=29.90 (SD=6.20) (P<0.05)
  • Depressive status was not associated with average annual change in IST
  • IST scores were not associated with ApoE-ε4 status
• BVRT:
  • High depressive symptoms were significantly associated with lower mean BVRT scores at baseline; high depressive BVRT score=10.80 (SD=2.5); low depressive BVRT score=11.40 (SD=2.10) (P<0.05)
  • Depressive status was not associated with average annual change in BVRT scores
  • BVRT scores were not associated with ApoE-ε4 status
• TMT-A and TMT-B:
  • High depressive symptoms were significantly associated  with higher mean performance times on TMT-A and TMT-B scores
    • TMT-A: High depressive score=65.63 (SD=27.36); low depressive score=59.83 (SD=23.66); P<0.05
    • TMT-B: High depressive score=136.46 (SD=65.64); low depressive score=120.89 (SD=53.37); P<0.05
  • Depressive symptoms were not found to be associated with average annual changes in TMT-A and TMT-B scores 
  • Scores were not associated with ApoE-ε4 status.

Effects of plasma DHA on cognitive decline:

• Plasma DHA not significantly associated with evolution of any cognitive test scores over time (adjusted for age, gender and education)
• Plasma DHA not significantly associated with MMSE errors at baseline or decline in MMSE with time
• IST performance: Scores were significantly associated with plasma DHA at baseline; but IST evolution was not found to be associated with DHA
• TMT-A and TMT-B: Not associated with plasma DHA at baseline or over time
• Relationship between plasma DHA and BVRT scores:
  • Significant interaction found between DHA proportion and ApoE-ε4 on the change of BVRT scores over time (P=0.02)
  • In non-carriers of ApoE-ε4, an increase in plasma DHA was not associated with BVRT performances over time 
  • In ApoE-ε4 carriers, a one-SD increase in plasma DHA was related to a slower decline in BVRT performances (β=0.061; SD=0.024; P=0.01); However, after exclusion of incident dementia cases, there was no longer a relationship between ApoE-ε4 status and BVRT change over time. (In elderly that remained dementia-free over the follow-up period, DHA was not associated with BVRT performance over time).

Effects of plasma EPA on cognitive decline:

• Plasma EPA not associated with change in cognitive tests when adjusted for age, gender and education
• Fully adjusted models: Plasma EPA was not associated with MMSE errors at baseline or decline in MMSE over time
• IST scores: Higher plasma significantly associated with higher IST scores at baseline, but not evolution of IST scores
• TMT-A and TMT-B: plasma EPA not associated with these scores
• Relationship between plasma EPA and BVRT scores:
  • Independent interactions between plasma EPA and both depressive symptoms (P=0.05) and ApoE- ε4 (P=0.006) on the change of BVRT score over time were observed
  • In non-carriers with low depressive symptoms, plasma EPA was not related to evolution of BVRT scores (β=0.001; SD=0.012; P=0.93)
  • In ApoE- ε4 carriers with low depressive symptoms, higher EPA was associated with slower change in BVRT scores (β=0.076; SD=0.025; P=0.003)
  • In ApoE- ε4 non-carriers with high depressive symptoms, higher plasma EPA was associated with slower evolution of BVRT scores (β=0.096; SD=0.048; P<0.04)
  • In ApoE- ε4 carriers with high depressive symptoms, higher plasma EPA was associated with slower evolution of BVRT scores (β=0.171; SD=0.049; P<0.001
  • Exclusion of incident dementia cases and adjustment for visual impairment and total energy intake did not change results substantially for the relationship between plasma EPA and BVRT.

Other Findings

• 7.8% of sample were found to have high depressive symptoms (CES-D score ≥17 in men and ≥23 in women
• Lower baseline plasma EPA was significantly associated with higher depressive symptoms (P=0.02)
• Lower baseline plasma DHA was not associated with higher depressive symptoms (P=0.95)
• ApoE-ε4 carrier status:
  • Plasma EPA and DHA were not significantly different in the whole sample between ApoE-ε4 carriers and not carriers
  • Among those with high depressive symptoms, plasma EPA was lower, but not DHA, in carriers (borderline significance P=0.06)
  • Among participants with low depressive symptoms, plasma EPA did not differ by ApoE-ε4 status.

The decline in visual working memory may be delayed in ApoE- ε4 carriers by EPA and DHA. EPA, but not DHA may impact the slowing of cognitive decline in depressed, older people.

Government:	The French National Research Agency; conseil Regional d'Aquitaine
Not-for-profit	Association France Alzheimer

Plasma EPA and DHA levels were only measured at baseline. It is possible that these levels could have changed over the course of the seven-year study.