|Home | About | Journals | Submit | Contact Us | Français|
Dietary fat intake may influence the rate of cognitive change among those at high risk due to vascular disease or risk factors.
Women’s Antioxidant Cardiovascular Study began in 1995-96 as a randomized trial of antioxidants and B vitamin supplementation for secondary prevention in women with cardiovascular disease or ≥ 3 coronary risk factors. From 1998-99, eligible participants aged ≥ 65 years were administered a telephone cognitive battery including five tests of general cognition, memory and category fluency (n=2 551). Tests were administered four times over 5.4 years. The primary outcome was a global composite score averaging z-scores of all tests. Multivariable generalized linear models for repeated measures were used to evaluate the difference in cognitive decline rates across tertiles of total fat and various types of fat.
Total fat intake or different types of fat were not related to cognitive decline. However, older age significantly modified the association: among the oldest participants, higher intakes of mono- and poly-unsaturated fat were inversely related to cognitive decline (p-interaction: 0.06 and 0.04, respectively), and the rate differences between the highest and lowest tertiles were cognitively equivalent to the rate differences observed with being 4-6 years younger.
In women at high risk of cognitive decline due to vascular disease or risk factors, dietary fat intake was not associated with 5-year cognitive change. However, a possible protective relation of unsaturated fats with cognitive decline in the oldest women warrants further study.
A large body of research implicates cardiovascular risk factors (Biessels et al., 2006, Qiu et al., 2005, Shobab et al., 2005) and disease (CVD) (de la Torre, 2006) in the development of age-related cognitive impairment. Affected cognitive domains include executive function, which is traditionally associated with cerebrovascular health, but also general cognition and episodic memory (Breteler et al 1994, Launer 2002, Luchsinger and Mayeux 2004, Korczyn 2005). Given the aging population and improved medical treatment of CVD, the number of people living with CVD has dramatically increased (National Center for Chronic Disease Prevention and Health Prevention, 2009). However, little is known regarding strategies that could reduce the increased risk of cognitive impairment among those with preexisting cardio- or cerebrovascular conditions; thus, identifying preventive factors for such populations is a major public health challenge.
As “heart healthy” fats are related to reductions in vascular risk factors (Hooper et al., 2001, Luchsinger et al., 2007) and disease progression (Leren, 1970), and possibly to cognitive decline in healthy populations (Morris et al., 2004, Solfrizzi et al., 2006), they may also slow cognitive aging in those with vascular risk factors or disease. Thus, we evaluated the association between fat intake and subsequent 5-year cognitive decline in 2 551 community-dwelling older women at high vascular risk.
The Women’s Antioxidant Cardiovascular Study (WACS) began in 1995-96 among 8 171 women, as a 2×2×2 randomized placebo-controlled trial of vitamin E, vitamin C, and beta-carotene supplementation for the secondary prevention of CVD (Bassuk et al., 2004). Eligible participants were female health professionals, ≥40 years, with prevalent CVD or ≥3 coronary risk factors, including a parental history of premature myocardial infarction (MI), diabetes, hypertension, hyperlipidemia, and body mass index ≥30 kg/m2. CVD included MI, stroke, revascularization procedures (percutaneous transluminal angioplasty, coronary artery bypass graft, carotid endarterectomy, or peripheral artery surgery), symptomatic angina pectoris or transient cerebral ischemia. Participants were 94.0% Caucasian and 3.3% African-American. In 1998, a fourth arm for B vitamin (combined folic acid, vitamin B6, and vitamin B12) supplementation was added among 5 442 women.
Until the scheduled end in 2005, participants completed annual mailed questionnaires on compliance, side effects, health and lifestyle characteristics and clinical endpoints. For the primary trial outcome, neither antioxidant supplementation (Cook et al., 2007) nor B vitamin supplementation (Albert et al., 2008) was found to protect against CVD.
From 1998-2000, a mean 3.5 years from antioxidant randomization, a cognitive substudy using telephone interviews was initiated. Among the 8 171 WACS participants, we used the following inclusion criteria: 1) active participants as of 1998-2000 and 2) aged ≥65 years. After excluding 5 001 non-eligible participants who were not active or aged <65 years, 3 170 WACS participants remained. Of the 3 170 eligible women for cognitive interview, 190 were unreachable, 156 declined participation, and 2 824 (95% of contacted women) completed the initial telephone cognitive assessment. Participants received three follow-up cognitive assessments at approximately two-year intervals. High follow-up rates were maintained: 93% completed at least one follow-up assessment, and 81% completed at least three assessments. In the fourth assessment, 24% of participants were not contacted for their interview as only a short interval had passed between their third interview and the end of the trial. Participation rates in each of the four cognitive interviews were virtually identical across the diverse treatment and placebo groups. For this study on diet and cognitive decline, we excluded 273 participants with incomplete dietary information. Thus, the analysis sample for the present study was comprised of 2 551 women. The project was approved by the institutional review board of Brigham and Women’s Hospital, Boston.
Previous studies of the effect of the main agents on cognitive change found that neither the antioxidant supplementation (Kang et al., 2009) nor the B vitamin supplementation (Kang et al., 2008) was significantly associated with a slowing down of cognitive change in women with preexisting cardiovascular disease or risk factors.
The Willett semi-quantitative food frequency questionnaire (SFFQ) (Willett, 1998) was administered at WACS enrollment. The SFFQ asked about the usual consumption (nine possible response categories) during the past year of 116 food items with a specified portion size. Detailed information was collected on type of fats or oils used in food preparation, as well as frequency of consuming fried foods away from home. For each food, the reported consumption frequency was multiplied by its nutrient content from the database of the US Department of Agriculture (U.S. Department of Agriculture, 2009). Fat intakes were computed as a sum across all food sources, taking into account types of margarine and fats used in cooking and baking. Finally, dietary intakes were energy-adjusted using the residual method and categorized into tertiles for analysis. We considered total, saturated, monounsaturated, and polyunsaturated fat, and the ratio of polyunsaturated to saturated fat.
The Willett SFFQ has been extensively studied for validity. In a validation study among 173 women (Willett et al., 1985), dietary fat intake as measured with the SFFQ correlated with dietary fat as measured with four 1-week dietary records collected over 1 year, with Pearson correlations ranging from 0.5 to 0.7 for the various fat types. Furthermore, polyunsaturated fat intake was significantly correlated with polyunsaturated fatty acids in adipose tissue (Spearman correlation = 0.4) (London et al., 1991).
Cognitive function was assessed by telephone, using five cognitive tests administered by trained interviewers: Telephone Interview of Cognitive Status (TICS), TICS 10-word list delayed recall, East Boston Memory Test immediate recall, East Boston Memory Test delayed recall, and animal naming test. General cognition was evaluated with the TICS (Brandt et al., 1988), a telephone adaptation of the Mini-Mental State Examination (range is 0 to 41 points). Verbal memory was assessed with the TICS 10-word list immediate recall test, TICS 10-word list delayed recall test, East Boston Memory Test immediate recall, and East Boston Memory Test delayed recall (Scherr et al., 1988); the East Boston Memory Test involves recalling 12 elements from a story. A test of category fluency (Morris et al., 1989), which partially measures executive function, was also administered: women were asked to name as many animals as possible in one minute.
The primary outcome was the rate of change from the first to the last assessment in the global composite score, computed as an average of all cognitive tests made into z-scores. Also, as decline in verbal memory is strongly associated with risk of Alzheimer disease (Small et al., 2000), a secondary outcome was the change in the verbal memory composite score, which was calculated by averaging the z-scores of the verbal memory tests. To derive the two composite scores for participants who did not complete all tests (only 0.5% for both the composite scores), we used the means of the z-scores from the available relevant tests.
The telephone cognitive test battery has high reliability and validity. Among 35 high-functioning, educated women, the correlation between TICS scores administered twice 31 days apart was 0.7 (p<0.01). In a validation study conducted in a sample of 61 women, the global composite score from the brief telephone-administered assessment correlated strongly with the overall score from an extensive in-person interview (Pearson correlation=0.8) (Weuve et al., 2004). Moreover, in a clinical validation study among 88 older female health professionals, who were similar demographically to WACS participants, poor performance in the TICS and in the verbal memory composite score were both associated with significant 8 and 12 fold increases, respectively, in subsequent dementia diagnoses.
We considered several potential confounding factors plausibly linked with both cognitive decline and fat intake. Basic models included as covariates, age at initial cognitive assessment, education and total energy intake. In the main multivariable models, further adjustment was made on WACS randomization assignments, as well as numerous other potential confounders related to lifestyle and heath status: marital status (married, divorced, widowed, single), alcohol intake (abstinence, 0.1-0.9g/day, ≥10g/day), physical activity (quartiles of weekly calories expended from exercise and climbing the stairs), use of multivitamin supplements (no, yes), smoking status (never, past, current), postmenopausal hormone therapy use (never, past, current), body mass index (quartiles), aspirin use exceeding 10 days in the previous month or not, non-steroidal anti-inflammatory drug use exceeding 10 days in the previous month or not, history of depression, cardiovascular profile at baseline (myocardial infarction, stroke, revascularization procedures, symptomatic angina pectoris, transient cerebral ischemia, none of such diseases), diabetes (no, yes on treatment, yes without treatment), hypertension (no, yes on treatment, yes without treatment), hyperlipidemia (no, yes on treatment, yes without treatment), food intakes of vitamin C (tertiles), vitamin E (tertiles), carotene (tertiles) and fiber (tertiles).
The nutrient database did not include data on trans-fat intake. Thus, for analyses of mono- and poly-unsaturated fats, we further adjusted for intake of food items that are major contributors of trans-fats in the US diet: margarine, fried foods consumed away from home, and processed baked foods such as crackers, cookies and pastries.
First, we compared participants’ age and age-standardized characteristics at dietary assessment across tertiles of total fat intake. Then, we used general linear models for repeated measures with random intercepts and slopes, to estimate the association of types of fat intake with the annual rate of change (expressed in standard units (SU)) over the four cognitive assessments, incorporating the longitudinal correlation within study subjects using unstructured covariance structures. For each type of fat, we evaluated two models, the basic-adjusted model and the multivariable-adjusted model. We tested for linear trends across tertiles of fat intake by modeling a continuous variable in which all participants in a given tertile of intake were assigned the median value. We used Wald tests for statistical testing. All models were fitted by maximum likelihood method using SAS software (SAS release 9.1, SAS Institute Inc., Cary, NC).
We examined potential effect modification by key risk factors for cognitive change: age at cognitive assessment, baseline scores, level of education, and cardiovascular profile at randomization. We conducted stratified analyses, and the tests of effect modification were performed by evaluating a three-way interaction term of time, the fat of interest, and the potential effect modifier. Additionally, we examined associations between types of fats and cognitive decline only among women who declared that their diet at randomization was almost the same as their usual diet over the past five years.
We also evaluated a “carbohydrate substitution” model. In this isocaloric model, we included multivariable covariates and total energy as well as all sources of energy except for carbohydrates (i.e., protein, individual fats and alcohol). Each energy source was represented as a continuous nutrient density, expressed as the energy from the individual source as a percentage of total energy. This model can be interpreted as the effect of replacing 1% of energy from carbohydrates with an equivalent percentage of energy from the fat. In this model, for simplicity of interpretation, the fats were grouped into “good” fats (the sum of mono- and polyunsaturated fats) and “bad” fat (saturated fat).
The average time from dietary assessment to initial cognitive assessment was 3.5 years (range 3.1–4.7), and the average time from the initial to the last cognitive assessment was 5.4 years (range 4.1–6.1). The global composite score as of the first assessment ranged from -3.94 to 2.03 (mean value= -0.003, SD=0.65), with higher scores indicating better cognitive function. The median TICS score was 35 (range 7–41).
When we examined characteristics of participants by total fat intake (Table 1), we found that women in the top tertile of total fat intake had less education and were engaged in less physical activity than women in the bottom tertile. Also, women with higher intakes of total fat had a higher body mass index and had more diabetes and hypertension, but not other cardiovascular conditions, such as MI, angina, and hyperlipidemia. Women with higher intakes of total fat had lower dietary intakes of alcohol, vitamin C, vitamin E, carotene and fiber.
In both the basic-adjusted and multivariable-adjusted models, we did not observe any significant differences across various fat tertiles in the mean annual rate of change for the global composite score (Table 2). We also did not find relations for the verbal composite score, TICS or category fluency test (data not shown).
Results were not substantially altered when we controlled for additional dietary factors known to be main contributors of trans-fat intake. We observed almost no change for mono-unsaturated fat and a somewhat stronger (but still non-significant) inverse association between cognitive decline and polyunsaturated fat.
When we investigated whether the associations with types of fats on cognitive decline may differ by age, level of education, baseline scores and cardiovascular profile at randomization, we found significant interactions with being older (i.e., being above or below 72 years, the median age). Among women aged 65-72 years, fat intake was not associated with cognitive decline (Table 3); however, among older women aged 73-91 years, higher intakes of mono-unsaturated fat and polyunsaturated fat were associated with slower decline in the global cognitive score (p for interaction =0.06 and 0.04, respectively) (Table 3). In the oldest stratum, as compared with women in the lowest tertile, the rate of decline in the global cognitive score of women in the highest tertile for monounsaturated fat was slower by 0.02 SU (95% CI = -0.01, 0.05; p for trend =0.13) and for polyunsaturated fat intake, the rate of decline was slower by 0.03 SU (95% CI = 0.00, 0.05; p for trend =0.07). As the rate of decline is slower by 0.005 SU per year of being 1 age younger (p<0.01), these differences in rates between the highest and lowest tertiles were equivalent to the differences in rates observed with being 4 years younger for monounsaturated fat and 6 years for polyunsaturated fat.
With decline in the verbal memory as the outcome, in these oldest women, the associations with both fats were similarly protective (p for trend =0.10 for polyunsaturated fat and p for trend =0.05 for monounsaturated fat).
Substitution models yielded results that were consistent with the main results, where the replacements of energy from carbohydrates with the same percentage of energy from “good” fats (mono- and polyunsaturated fatty acids) or “bad” fat (saturated fat) were not significantly related to cognitive decline in the whole sample (data not shown in table). However, when we examined these models by age, we observed that among older women aged 73-91 years, a 5% isocaloric replacement of energy from carbohydrates with energy from “good” fats corresponded to a reduction of mean annual rate of global composite score by 0.02 SU (95% CI = 0.00, 0.04; p-value = 0.04).
We considered the effect of excluding persons who reported that their current diet was different from their usual diet over the past five years to include only those with stable long-term diets. In the restricted sample of 859 women (34% of the whole sample) with a stable diet as of baseline, the results were similar, with no significant associations between fat intake and overall cognitive decline.
In this large prospective study of older women at high vascular risk, we observed no overall association between dietary fat intake and cognitive decline over 5-years follow-up. However, older age significantly modified the association: in older women (>72 y), substitution of carbohydrates with mono- or polyunsaturated fats was associated with a significantly slower rate of cognitive decline, which was cognitively equivalent to delaying aging by about 4-6 years.
Because several types of fats, cognitive outcomes and subgroup analyses were examined, the effect modification by older age may have been due to chance and thus, these results should be interpreted with caution. However, as older participants have a faster rate of cognitive decline than younger participants, differences in rates are easier to detect in the oldest strata. This fact may underlie our observation of protective associations with higher mono- and poly-unsaturated fat intake only among the subset of older women. This is consistent with other studies of types of fat intake on cognitive aging among populations at high vascular risk (Beydoun et al., 2008, Devore et al., 2009). Notably, in a study by Devore et al., which focused on women ≥70 years with type 2 diabetes in the Nurses’ Health Study (mean follow-up: 1.8 years), protective associations with higher intake of unsaturated fats was observed in relation to cognitive decline (Devore et al., 2009).
The modulation of vascular processes is likely a key-element to understand how dietary fat composition could indirectly act on the brain in persons with preexisting CVD. Secondary prevention trials to increase polyunsaturated fat and/or reduce saturated fats have found lower recurrence of cardiovascular events (Leren, 1966, Tuttle et al., 2008), conditions that have been closely associated with higher cognitive risk (de la Torre, 2006). Increased dietary intake of polyunsaturated fatty acids may also act on the brain in enhancing neurotransmission, neuroprotection, and neurogenesis, partly via antioxidant and anti-inflammatory effects (Lindqvist et al., 2006, Yehuda et al., 2005). Interestingly, in the oldest women, we observed that higher intakes of mono-and poly-unsaturated fats at baseline was associated with lower subsequent risks of decline in both the global composite score and verbal composite score. However, we did not detect any such associations with the category fluency test. This specificity of the association to different outcomes, each representing different cognitive domains, underscores the complexity of the potential etiologic pathway between dietary fat intake and cognitive decline. Although a number of animal studies (de Wilde et al., 2002, Winocur and Greenwood, 2005) support a neuroprotective effect of unsaturated fats, plausible mechanisms have not been fully elucidated, and results from human studies remain inconclusive (Fotuhi et al., 2009). In particular, epidemiologic data are scarce for populations at high vascular risk.
Our study had several strengths. The data from the WACS cognitive cohort included a detailed dietary assessment with information on various fat intakes and four repeated cognitive assessments. We adjusted for a wide array of potential confounders, including various dietary, health, medication and lifestyle factors.
Nonetheless, some methodological issues should be considered. It is possible that a single assessment of fat consumption in late adulthood does not reflect the long-term dietary intakes of participants, which may be more etiologically relevant. This may be particularly true in our population of women at high cardiovascular risk, who were probably encouraged to change their diet. To address this, we carefully adjusted the analyses on participants’ cardiovascular profiles at enrollment (which may provide an indicator of the severity of the condition) and conducted subgroup analyses only among women who reported at enrollment that their diet changed very little in the past five years. Overall, our main “null result” was robust, since we did not observe that the findings were different from the main analyses. A limitation was that we lacked updated dietary data after baseline, however, because of the possibility of reverse causation (i.e., participants who experience slight cognitive change may actively change their diet) such analyses may have inherent biases, and the analyses with baseline diet would be preferred. Second, a telephone cognitive assessment might lack validity. However, both reliability and validity studies of our telephone instrument have provided convincing evidence of its utility to evaluate cognitive function in an epidemiologic study. In addition, using a similar telephone battery, significant associations with cognitive aging have been found with a large number of risk factors, including dietary variables (Kang et al., 2005, Lee et al., 2006). Finally, participants were mostly Caucasian, which precludes extending our findings to other ethnic minorities.
In a large sample of community-dwelling women aged ≥65 with pre-existing cardiovascular disease or risk factors, no associations were observed between intakes of various fats with 5-year cognitive decline. However, a trend toward a possible beneficial effect of unsaturated fats for preserving cognitive function in oldest women warrants further study.
We are grateful to the investigators, staff (especially Martin Van Denburgh), and participants of the WACS cognitive substudy. This work was supported by the National Institutes of Health (HL046959, AG15933) and the American Heart Association. MNV is supported by the Fondation Bettencourt-Schueller for her postdoctoral fellowship; she performed data analysis and manuscript writing. JHK and FG both procured funding for this research, and contributed to the study design, data analysis, and manuscript writing.
CONFLICT OF INTEREST The authors declare no conflict of interest.