| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH COMMUNICATION |
1 From the Division of Health and Family Studies, Institute for Families in Society, University of South Carolina, Columbia (JZ); the Center for Clinical Pharmacology, School of Medicine, University of Pittsburgh (MFM); and the Department of Epidemiology and Biostatistics, the Arnold School of Public Health, University of South Carolina (REM)
2 Supported by the academic institutions with which the 3 authors are affiliated. 3 Reprints not available. Address correspondence to J Zhang, 4770 Buford Highway, Mailstop K24, Atlanta, GA 30341. E-mail: jzhang02{at}gwm.sc.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: The objective was to examine the relation between serum cholesterol concentrations and performance in immediate memory, visuomotor speed, and coding speed tests.
Design: The participants were 4110 adults aged 20–59 y who completed a set of neurobehavioral tests and had blood specimens collected as a part of the third National Health and Nutrition Examination Survey, 1988–1994.
Results: After adjustment for sociodemographic variables, serum trace elements and vitamins, dietary energy intake, and risk factors for cardiovascular disease, we found inverse linear associations of serum total cholesterol and non-HDL cholesterol with visuomotor speed in men. The least-squares mean (± SE) visuomotor speeds were 231.6 ± 2.6, 224.0 ± 2.2, and 218.9 ± 2.5 ms, respectively, for men with serum total cholesterol concentrations below the 25th, between the 25th and the 75th, and at or above the 75th percentile (P for trend < 0.001) and were 231.7 ± 2.7, 225.8 ± 2.4, and 214.1 ± 2.3 ms, respectively, for men with a non-HDL-cholesterol concentration below the 25th, between the 25th and the 75th, and at or above the 75th percentile (P for trend < 0.001). No significant associations were observed between memory or coding speed and the selected serum cholesterol measures in men, and the scores of the 3 neurobehavioral tests were unrelated to serum cholesterol in women.
Conclusion: Low serum total cholesterol and non-HDL cholesterol are associated with slow visuomotor speed in young and middle-aged men.
Key Words: Serum cholesterol • HDL cholesterol • LDL cholesterol • visuomotor speed • coding speed • memory • third National Health and Nutrition Examination Survey • NHANES III • cognitive function
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The participants selected for the neurobehavioral tests were a systematically selected half-sample of persons aged 20–59 y who were tested at a mobile examination center. Those with odd-numbered survey identification numbers participated in neurobehavioral tests, and those with even numbers participated in an allergy test. Persons who could speak neither English nor Spanish or who were legally blind were not given the test. Of those participating in the neurobehavioral tests, 534 did not complete the SRT (the simplest test) or completed the test but had an outlier, and 417 had no hematologic information. A total of 470 participants were excluded from the current analysis because they were pregnant (n = 113), had a history of cancer (including skin cancer, n = 144), or were participating in a cholesterol-lowering intervention (n = 473). An additional 30 participants were excluded because they had a history of a stroke or experienced a recent stroke that could compromise cognitive performance. Data from the remaining 2152 women and 1958 men were used for the current study.
Cognitive function assessment
NHANES III involved 3 neurobehavioral tests selected from a computerized test battery called the Neurobehavioral Evaluation System (17, 20, 21), which was designed to measure neurobehavioral performance in populations occupationally or environmentally exposed to chemicals. This system has been used to investigate the neurotoxic effects of organic solvents, organophosphate pesticides, nitrous oxide, lead, and mercury (19). The neurobehavioral tests administered in NHANES III were simple reaction time (SRT), symbol-digit substitution (SDS), and serial digit learning (SDL) test.
Visuomotor speed
Visuomotor speed was measured by the SRT test. The participant rested an index finger on a push button and pressed the button as quickly as possible whenever a solid square (4 x 4 cm) was displayed in the center of a blank computer screen. The square remained on the screen until the participant pressed the button, after which it was cleared from the screen. If there was no response, the square was cleared from the screen after an elapsed time of 1000 ms. The measured response was the latency (in ms) between the time of appearance of the square on the screen and the time when the participant pressed the button. Therefore, a better performance was associated with a shorter latency. A total of 50 trials were administered to each participant. The interval between trials varied randomly according to a uniform distribution ranging from 2.5 to 5.0 s to limit anticipatory responses. The mean reaction time of trials 11–50 was calculated. Values 
50 or 
750 ms were considered outliers and were not included in the calculation (19).
Coding speed
Coding speed was measured by the SDS test. On the upper half of the computer screen, the subject was presented with a grid that paired 1 of 9 different symbols with 1 of the digits from 1 to 9. A similar grid was displayed on the bottom half of the screen with the same symbols presented in a scrambled order and the spaces for the corresponding digits left blank. The participant was asked to enter, as quickly as possible, the matching digit for each symbol. A total of 5 trials were conducted with a different pairing of digits and symbols on each trial. The amount of time (in s) required to enter each digit was recorded, and the total latency of the last 4 trials was calculated. A quick coding speed was associated with a shorter latency. The SDS test score was modestly correlated with that of the SRT test (Pearson r = 0.35).
Learning and immediate memory
Learning and immediate memory were measured by the SDL test. The participant was presented with a series of digits displayed one at a time on the computer screen. Each digit was displayed on the screen for 0.6 s with intervals of 0.6 s between digits. After all the digits were displayed, the participant was asked to press the numeric keys to enter the entire sequence of numbers in the order in which they were presented. All the trials contained the same 8-digit sequence. Testing continued until the subject responded correctly on 2 consecutive trials or until the subject attempted 8 trials. For each trial, 2 points were credited when <6 of the 8 digits were in the correct position, 1 point when either 6 or 7 digits were in the correct position, and 0 points when all 8 digits were reported correctly. The score used in the current study represents the sum of the error scores for each trial. A lower score reflects better learning and immediate memory. The SDL test score was modestly correlated with the SDS score (Pearson r = 0.49) and was weakly correlated with the SRT score (Pearson r = 0.27).
Measurements of serum cholesterol
NHANES III participants were randomly assigned to a morning or a nonmorning group. Most participants from the nonmorning group did not fast 12 h before a blood sample was collected. During the health examination, blood samples were collected from examinees across a range of fasting statuses and were processed according to a standard protocol (16). We selected 3 cholesterol measures: serum total cholesterol (TC), HDL cholesterol, and non-HDL cholesterol (NHDLC). LDL cholesterol was calculated by using the Friedewald equation, and a substantial percentage of participants had missing LDL-cholesterol values because of limitations with the Friedewald equation (22) or participants’ fasting status. Therefore, the current study used NHDLC (TC – HDL cholesterol) rather than LDL cholesterol. All cholesterol measures were categorized into 3 sex-specific levels: high, at or above the 75th percentile (men: TC 224 mg/dL, HDL cholesterol 52 mg/dL, and NHDLC 178 mg/dL; woman: TC 215 mg/dL, HDL cholesterol 63 mg/dL, and NHDLC 166 mg/dL); low, below the 25th percentile (men: TC <169 mg/dL, HDL cholesterol <36 mg/dL, and NHDLC <120 mg/dL; woman: TC <169 mg/dL, HDL cholesterol <44 mg/dL, and NHDLC <111 mg/dL); and intermediate (the remaining subjects).
Definition of covariates
Covariates were selected from the literature(19, 23-25) and included demographic variables (eg, age, income, and education), questionnaire variables [eg, previous night’s night sleep (more than usual, not more than usual); energy level (tired, not tired); consumption of coffee, tea, or cola in the past 3 h (yes, no); and consumption of alcohol in the past 3 h (yes, no)]. The risk factors for cardiovascular diseases included current cigarette smoker (yes, no), current alcohol drinker (yes, no), high blood pressure (yes, no), waist-to-hip ratio [high (>1.00 for men and >0.80 for women) and low], and leisure-time physical activity. Smokers were defined as those who had smoked cigarettes in the past 5 d, and drinkers were those who had consumed 12 drinks in the previous 12 mo. Hypertension was defined in 2 ways: uncontrolled hypertension (a systolic blood pressure 140 mm Hg or a diastolic blood pressure 90 mm Hg) and hypertension that necessitated medication. Serum trace elements; vitamins C, E, and A; carotenoids (
-carotene, ß-carotene); calcium; iron; and lead were measured. Vitamin E and ß-carotene are fat-soluble and correlated with TC; these variables are reported per unit of cholesterol. We also measured serum total protein and serum albumin concentrations and determined daily dietary energy intakes from a 24-h dietary recall as indicators of malnutrition or energy deprivation. Serum glucose was used to control for acute effects of low energy intake. For the women, we were concerned about possible confounding from menopausal status and created an indicator variable for menopausal status by dichotomizing the continuous age variable by using 51 y (the average age when menopause starts) as the cutoff. Depression has been identified as a risk factor for poor cognitive performance (26, 27) and has been linked with low serum cholesterol concentrations as well. Because a diagnosis of depression was available only for adults aged 19–39 y in NHANES III, we controlled for depression in the secondary analysis.
Statistical methods
As recommended by the National Center of Health Statistics, we used SUDAAN software (28) (SAS-callable version 7.5; SAS Institute, Research Triangle Park, NC) with appropriate weighting and nesting variables (16-19). In all analyses that related cognitive function to serum cholesterol or covariates, we controlled for age (continuous). No data transformation was performed. Because this study evaluated results from 3 neurobehavioral tests, a Bonferroni adjustment of the level to P < 0.017 (0.05/3) was considered when examining the association between any domain of cognitive function and serum cholesterol.
Preliminary analyses indicated that a sex difference was prominent; therefore, the formal analyses were performed separately by sex. Before multivariable modeling, we examined the association between each domain of cognitive function and each covariate. If a covariate in continuous form was shown to not be related with poor performance on the neurobehavioral tests, further examination was made by using percentiles or clinically recommended cutoffs, if available, to identify a possible nonlinear association. For illustration purposes only, all serum elements—including 3 cholesterol measures—were categorized into 3 sex-specific levels: at or above the 75th percentile, between the 25th and the 75th percentile, and below the 25th percentile. Daily dietary energy intake was categorized in the same way. However, in the multivariable regression, the continuous variables that were categorized for reporting remained in continuous form if the R2 value of the regression model was better than that obtained with the use of categorized variables. Multivariable models were generated in a stepwise fashion for each neurobehavioral test. In the first step, the main effect and covariates, for which the overall P values of age-adjusted means were <0.20, were entered into the model. In the second step, the covariates with a P value > 0.10 were dropped, and the interactions between main effect and vitamins E and A per cholesterol unit, dichotomized age variables, and ethnicity were tested. We kept the covariates with P values of regression coefficients <0.05 in multivariable models. However, to avoid overadjustment and to increase the precision of estimates, we removed any covariate whose addition to the multivariable model resulted in a change of <10% in the mean score of cognitive measures. The main effects were always retained. The least-squares means of each test score were calculated for each level of main effects from the finalized regression models. Regression models with cholesterol in continuous form were also run to assess the linear relations. Finally, we conducted a secondary analysis of data on participants aged 19–39 y to examine whether the inclusion of depression altered the findings from the overall sample.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The average age of the subjects was 
37 y, and 89% of the men and 87% of the women were non-African Americans. There were significant differences between men and women in most of the characteristics presented in this table. For example, compared with men, women had a more favorable serum cholesterol profile (ie, lower NHDLC and higher HDL-cholesterol concentrations) and they performed more slowly in the visuomotor speed test but more quickly in the coding speed test.
|
, the performances on the neurobehavioral tests were associated with many factors. The average scores for all 3 tests increased with age and decreased as the education level increased. Serum antioxidants (eg, vitamin E, ß-carotene, and selenium) were associated with memory and coding speed in both sexes. The concentrations of serum albumin, total protein, and serum glucose were associated with performance on different tests at P values <0.017. The adults with regular leisure-time physical activity performed significantly better on all 3 tests than did the adults who had no leisure-time physical activity. Low dietary energy intake resulted in a poor immediate memory and slow coding speed for both men and women.
|
). The difference in visuomotor speed between subjects with high and low TC concentrations was 12.7 ms (231.6 – 218.9), and the difference in visuomotor speed between subjects with high and low NHDLC concentrations was as much as 17.5 ms (231.7 – 214.2). A series of multiple linear regression analyses with TC in continuous form showed a decrement of 5.0 ± 1.2 ms (P < 0.001) in visuomotor speed for each 1-mmol/L (39-mg/dL) increase in serum TC concentration. In the secondary analysis, in which a history of depression was an additional covariate, all of the estimates remained unchanged relative to the overall sample (data not shown).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The main findings of the current study are consistent with those of previous studies. In a naturalistic cross-sectional study of 177 healthy adults aged 25–60 y of both sexes, Muldoon et al (9) observed an association between poor performance on the Block Design Test and low serum TC. Similarly, Benton (10) reported a relation between low TC and slow mental speed from the results of a choice reaction time test in college students. These observations from cross-sectional studies were reinforced by findings from a clinical trial (13), which found decreased cognitive performance on tests for attention and psychomotor speed in hypercholesterolemic persons after treatment with lovastatin, and by findings from a twin study, which found that low serum TC concentrations predicted a decline in SDS performance (11). The principal similarity in cognitive demand between the Block Design, Choice Reaction Time, and SDS tests and the SRT in the current study is the speed of information processing. All of these tests require visuoconstructional abilities and rapid manipulation of information. A sex difference was salient in the current study, which may have been due to sex hormones or to other unmeasured sex-related factors. We observed that men were quicker than were women on the visuomotor speed test; the effect of low serum cholesterol on visuomotor speed might be more salient in a population with a rapid basal speed.
The origins of these relations between TC, NHDLC, and measures of visuomotor speed are not clear. The current study provides additional evidence to rule out some proposed explanations. Muldoon et al (13) speculated that lovastatin penetrates the central nervous system and, conceivably, cognitive effects may be the result of neuropharmacologic actions that are independent of cholesterol lowering. Statin drugs lower circulating concentrations of vitamin E and ubiquinone and may affect the synthesis of polyunsaturated fatty acids, which are integral to neuronal membranes (27). The association we observed may be independent of any neuropharmacologic actions because, in the current study, the participants were not taking any cholesterol-lowering medications. Cholesterol is a component of circulating complex lipoprotein particles that carry phospholipids, triacylglycerol, and many micronutrients such as ß-carotene and vitamin E (9). Vitamin E has been shown to be related to cognitive function because of its antioxidant properties (29). Researchers, therefore, speculate that low concentrations of one or more components of lipoprotein particles circulating in the bloodstream may produce subtle but measurable impairments of mental processes by influencing the supply of fat-soluble micronutrients, specifically, vitamin E, ß-carotene, and vitamin A (9). This hypothesis was not supported by the findings of the current study. Adjustment for ß-carotene and vitamin E did not weaken the association. Alternatively, other investigators suggested that any intellectual impairment might be related to some illness or environmental variable (including dieting) that resulted in energy deprivation or weight loss and not to low serum cholesterol concentrations (30, 31). Our results suggest that this is not the case because inclusion of serum total protein, serum albumin, glucose, and daily dietary energy intake did not materially change the estimates of the main effects.
Compared with the typical findings that occur in intoxicated or brain-injured patients, the association between low serum cholesterol and visuomotor speed observed in the current study might be considered small. Its clinical relevance, however, would be well demonstrated when compared with the aging-related declines in visuomotor speed. After adjustment for potential confounders, we observed that visuomotor speed increased by 0.09 ± 0.15 ms for each 1-y age increase (P > 0.10; data not shown in detail). Therefore, from the ages of 30–59 y, the cumulative increase in mean visuomotor speed would be 3 ms, which was only 25% of the difference (12.7 ms) between the men with a TC concentration at or above the 75th percentile and men with a TC concentration below the 25th percentile. In other words, the contribution of the difference in TC to the variance in visuomotor speed was several fold greater than that from aging from 30 to 59 y. The major concern of clinical relevance was that if the association observed in the current study and in previous studies proves to be robust, the effects could be widespread and substantial because more and more people are encouraged to lower their serum cholesterol concentrations (32). Equally important, if the difference widens over time, it may have an effect on quality of life and even survival, because relative small impairments in cognitive function may shorten survival in later life (26). For example, the impairment could affect automobile driving, a situation encountered in everyday life and a task that requires the integration of a broad array of cognitive abilities: sustained attention and speed and accuracy of psychomotor performance (13). The findings of the current study might be helpful in explaining the high incidence of deaths from accidents in persons with low serum cholesterol concentrations (33-38).
Perhaps the greatest limitation of the current study was its cross-sectional design. The inherent limitation of cross-sectional data prevented us from investigating causality. Both serum cholesterol and cognition are strongly influenced by genetic factors, and any observed covariation could have been due to the concomitant expression of genotype without cholesterol and cognitive function being causally related to one another (9). Although the richness of the NHANES III data allowed us to delineate the relation between serum cholesterol and cognition after adjustment for a wide array of potential confounding factors, we were unable to control for the fasting or feeding status or the use of medications other than cholesterol-lowering drugs (eg, psychotropic medication, ß-blockers, oral contraceptives, and estrogen). All these factors can affect lipid concentrations and cognitive function. Also, it has been documented that cognitive performance fluctuates across menstrual rhythms (39). Serum samples were collected only once, which may misrepresent true cholesterol concentrations. All these factors could lead to measurement errors with ensuing loss of precision, which would likely result in the attenuation of the true relations rather than in the production of spurious ones. The absence of menstrual rhythm data or other sex-related factors might be one reason for our failure to identify an association in women.
The current study has unique strengths as well. An increasing number of studies indicate that hormones play an important role in cognitive function (39-41) and that lipid metabolism differs by sex (42). The relatively large sample size of each sex in the current study allowed us to sufficiently control for the potential confounding or modifying effects from sex by analyzing the sexes separately. In addition, the NHANES III participants had been randomly selected from the community-dwelling US population; therefore, our results should be fairly generalizable to adults in the United States.
In conclusion, our findings indicate that a low concentration of serum TC or NHDLC is associated with relatively slower reaction times in men, and this association is independent of sociodemographic variables, risk factors for cerebrovascular diseases, and serum trace elements and vitamins. The clinical significance of this association is unclear but may be inferred from comparison with the age-related decline in visuomotor speed. Given the consistency of the findings, this association deserves more detailed and sustained study in large prospective investigations.
|
ACKNOWLEDGMENTS |
|---|
All authors contributed to the study design and data analyses and participated in critically revising the manuscript for important intellectual content. MFM and REM supervised the study. JZ obtained the data, performed the data analysis, and drafted the manuscript. None of the authors had any financial or personal interest, including advisory board affiliations, in any company or organization within the food or pharmaceutical industries.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. M. Mielke, Q.-L. Xue, J. Zhou, P. H. M. Chaves, L. P. Fried, and M. C. Carlson Baseline Serum Cholesterol Is Selectively Associated With Motor Speed and Not Rates of Cognitive Decline: The Women's Health and Aging Study II J. Gerontol. A Biol. Sci. Med. Sci., June 1, 2008; 63(6): 619 - 624. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. de Frias, D. Bunce, A. Wahlin, R. Adolfsson, K. Sleegers, M. Cruts, C. Van Broeckhoven, and L.-G. Nilsson Cholesterol and Triglycerides Moderate the Effect of Apolipoprotein E on Memory Functioning in Older Adults J. Gerontol. B. Psychol. Sci. Soc. Sci., March 1, 2007; 62(2): P112 - P118. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. L. Van, V. K. Bakalov, and C. A. Bondy Monosomy for the X-Chromosome Is Associated with an Atherogenic Lipid Profile J. Clin. Endocrinol. Metab., August 1, 2006; 91(8): 2867 - 2870. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Zhang, J. R. Hebert, and M. F. Muldoon Dietary Fat Intake Is Associated with Psychosocial and Cognitive Functioning of School-Aged Children in the United States J. Nutr., August 1, 2005; 135(8): 1967 - 1973. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Zhang, M. F. Muldoon, R. E. McKeown, and S. P. Cuffe Association of Serum Cholesterol and History of School Suspension among School-age Children and Adolescents in the United States Am. J. Epidemiol., April 1, 2005; 161(7): 691 - 699. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. F Muldoon, J. Zhang, and R. E McKeown Reply to SL Black Am. J. Clinical Nutrition, February 1, 2005; 81(2): 538 - 538. [Full Text] [PDF]
|
|
|
|
|
|
S. L Black Serum cholesterol and visuomotor speed: inverse or direct association? Am. J. Clinical Nutrition, February 1, 2005; 81(2): 537 - 538. [Full Text] [PDF]
|
|
|
|
 |
|
 Minerva BMJ, August 7, 2004; 329(7461): 358 - 358. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
