The association between lipids and physical activity may partly explain the reduced risk of cardiovascular disease associated with lifelong physical activity(12,16,20,21,36,42,80) although it has been suggested (66) that adjustment for covariates may remove the observed association. In a meta-analysis of 66 training studies (69), the average exercising subject was found to have a reduction in total cholesterol, triglyceride, and LDL cholesterol with an increase in HDL cholesterol. Some trials have tried to isolate the effects of weight loss and compare weight loss alone with exercise alone (75,78) as this may explain part of the published differences between runners and sedentary men(73).
The strength of association between physical activity and lipoproteins, especially HDL cholesterol, seen in cross-sectional, case control, and to a lesser extent in intervention studies, has not been seen in population studies(5,9,14,15,22,49,51,64), and the differences in lipid levels seen across populations are much less than the differences seen between trained athletes and controls.
Most studies have been of middle aged males although one study(57) found that HDL cholesterol was also increased in older active males and females (aged 50-89 yr). Interestingly, while in this study there was a doseresponse relationship in males, HDL cholesterol levels were slightly lower in females who reported heavy exercise.
Apolipoproteins are the peptide components of the various lipoproteins and act as membrane receptor sites and cofactors for enzymes in lipoprotein metabolism. Those with coronary artery disease have significantly lower levels of apoAI than those without coronary heart disease (44) and apoAI and apoAII are useful discriminators in cerebral atherosclerosis(34). Measurement of apoAI may be a better discriminant than HDL cholesterol for coronary heart disease (6) and in predicting risk of coronary heart disease in young men(3). ApoB is the apolipoprotein associated with LDL cholesterol; and apoB, together with LDL cholesterol, has been directly related to the extent and severity of coronary atheroma(19). Measurement of apoB concentration may improve the ability to detect individuals at risk of CHD (6), and the ratio of apoB/apoAI and apoB/apoAII help predict the presence or severity of coronary heart disease in men (54). Lipoprotein(a) is a low-density lipoprotein-like particle that has been associated with increased coronary heart disease(25,45,62,70) and to the incidence of coronary heart disease in a prospective study (59). Lipoprotein(a) level is also predictive of angiographically determined coronary artery disease independent of LDL (10) and has been shown to be an independent risk factor in cerebrovascular atherosclerosis(84).
In comparisons of active and less active people(24,37,40,47,53,68) and in longitudinal training studies(31,32,63,65,74), a relationship between physical activity and apoAI has been shown, although not all have demonstrated this consistently(27,67,81). In one remarkable study, apoAI concentrations were increased in 13 male mountaineers during a strenuous climb of 8-wk duration (52) while in contrast a marathon run has been shown not to result in an increase (60). A major recent review highlighted the inconsistencies in the relationship between physical activity and apolipoproteins apoAI, apoAII, and apoB(13). There is little work relating Lp(a) to physical activity, although in one study vigorous aerobic exercise lowered plasma Lp(a) levels by (26%) in a small group (N = 16) of fit men who participated in 8 d of cross-country skiing (23). In contrast, others have found no significant difference in Lp(a) levels between middle aged males runners and controls (26), nor a direct association between Lp(a) and cardiorespiratory fitness(28).
The aim of this study was to explore the relationship between physical activity, lipids, apolipoproteins, and Lp(a) in a population with a high incidence of cardiovascular disease through the Northern Ireland Health and Activity Survey (43), a cross-sectional population survey of physical activity, and other lifestyle parameters and cardiovascular risk factors.
MATERIALS AND METHODS
Fieldwork for this study was carried out from February to November 1992 and comprised three parts: a computer assisted interview, a physical appraisal, and a fasting blood sample. It was designed to yield a representative sample of adults aged 16 yr or over in Northern Ireland and employed two-stage probability sampling. The first stage was a sample of 1600 addresses taken from the Northern Ireland Rating and Valuation lists, stratified by region to ensure proportional sampling across the province. The second stage was carried out by the interviewers so that one person was selected in each household using a Kish grid random selection procedure (33). The sample was weighted to take account of household size and the appropriate adjustment made to the method of calculating the standard error. Because of this weighting the numbers in each subgroup are not whole numbers; thus, subtotals may not add up to the total number of participants.
Physical activity was classified under three headings: a) the highest level of activity in the 4 wk before the interview (none, light, moderate, vigorous); b) the frequency and intensity of bursts of physical activity lasting 20 min in the 4 wk before the interview, which is a more accurate reflection of habitual activity (6 categories: level 0-level 5); and c) the proportion of life active in sport and exercise since age 14 (0, <0.25, 0.25-0.49, 0.50-0.74, and 0.75+).
The method of classification of physical activity was as follows: The interview included a comprehensive record of all aspects of physical activity in the previous 4 wk:i) physical activity in the home: housework, gardening, DIY; ii) physical activities outside the home: walking and cycling, occupation; iii) sporting activity, current and historical; iv) diet and lifestyle; v) self-assessment of activity, health, and fitness; and vi) health.
Activities were grouped according to energy costs(46,48,56,79) so that light activities were those with an energy cost of 2-5 kcal·min-1, moderate activities had an energy cost of 5-7.5 kcal·min-1, and vigorous activities had an energy cost of 7.5 kcal·min-1 or greater. These thresholds correspond to three levels of energy expenditure, i.e., less than 40%, 40%-60%, and more than 60% ˙VO2max for an average middle-aged man of 65 kg body weight. Respondents were assigned to these categories independent of age, sex, or aerobic fitness. Respondents were asked the duration of each activity on the most recent occasion (or usual time if the most recent occasion was different from usual). Activities were graded as either greater or less than 20 min and frequency of activity graded into four categories. Respondents were allocated to one of six groups based on activity (of at least 20-min duration). Those who participated on 12 or more occasions during the 4 wk were further subdivided into three groups based on a combination of vigorous, mixed, and nonvigorous activities. There were three other groups: those who participated in activity on 5-11 occasions, those who participated on 1-4 occasions, and those who had no vigorous or moderate activity of 20 min duration in the 4-wk period. Using information gained on intensity and duration, respondents were allocated to one of the following groups-level 0: no moderate or vigorous activity lasting 20 min; level 1: 1-4 occasions of at least moderate activity; level 2: 5-11 occasions of at least moderate activity; level 3: 12 or more occasions of moderate (no vigorous activity); level 4: 12 or more occasions mixed between moderate and vigorous activity; and level 5: 12 or more occasions of vigorous activity.
Lifetime participation since the age of 14 yr was calculated by counting the number of years in which each respondent participated in vigorous or moderate activities on a regular basis (at least once per week for a few months of the year or more). This included walks of over 2 miles and cycling. Dividing the number of years of participation by total years from age 14 to the present, created an index of lifetime participation. Thus, a person who participated in sport or active recreation every year since age 14 would have a score of 1, a person who participated in half those years would have a score of 0.5, etc.
The physical appraisal was performed in a specially designed mobile laboratory at 14 hospital sites throughout the province. Height, weight, and blood pressure were measured using standard protocols(2). Height was measured using a Holtain stadiometer and weight was measured to the nearest 0.1 kg using a Phillips Electronic scales(HP 5320). Body mass index was used as an index of obesity(72). Blood pressure was measured using a Hawksley random zero sphygmomanometer. Habitual dietary preferences were assessed by questionnaire.
A venous blood sample (30 ml) was taken either on a morning during the week preceding the physical appraisal or not less than 2 d following the physical appraisal at the respondent's home, place of work, or physical appraisal laboratory. Samples were centrifuged at 3,000 rpm at room temperature, aliquoted, and stored in duplicate at -45°C. Total cholesterol was measured as an automated process on the Technicon Dax as part of the full lipid profile using an enzymatic cholesterol method.
Quality control was Armtrol normal (lot no.1147) and elevated (lot no. 1148) [Diamed UK Ltd. Co Antrim] made up daily by the addition of 10-ml distilled water to each vial using a grade A bulb pipette. Quality control racks were run every 40 samples. Calibration was with CDC reference laboratory, Glasgow. The reference range was 3.6-6 mmol·l-1(139.3-232.2 mg.dl-1). Method is linear from 0 to 12.9 mmol·l-1 (0.-499.2 mg·dl-1). Between batch imprecision (CV) was 3.5% at a mean level of 4.5 mmol·l-1 (174.1 mg·dl-1) and 3.3% at a mean of 7.3 mmol·l-1 (282.5 mg·dl-1).
Triglycerides were also measured as an automated process on the Technicon Dax as part of the full lipid profile. Reference range was 0.34-2.26 mmol·l-1 (30.1-200.0 mg·dl-1). Method is linear from 0 to 5.8 mmol·l-1 (0-513.3 mg·dl-1). Between-batch imprecision (CV) was 1.7% at a level of 1.09 mmol·l-1 (96.5 mg·dl-1) and 2.1% at a level of 3.24 mmol·l-1 (286.7 mg·dl-1).
HDL cholesterol and HDL cholesterol subfractions (HDL2/HDL3) were measured by enzymatic photometry (Quantolip, Immuno) (35). Precision and accuracy were checked using Quantolip norm control (Immuno R, no. 8253015, Immuno Ltd., Kent, U.K.). Between batch imprecision for HDL (CV) was 5.4% at a mean level of 0.92 mmol·l-1 (35.6 mg·dl-1d for HDL3 was 7.4% at a mean of 0.80 mmol·l-1 (31.0 mg·dl-1).
The formula used to calculate LDL was as follows: LDL cholesterol = total cholesterol - (triglyceride/5 + HDL cholesterol). Apolipoprotein levels were determined using immunoturbimetry. Between batch imprecision for apoAI (CV) was 3.9% at a mean level of 95 mg·dl-1, for apoAII was 4.6% at 35 mg·dl-1, and for Apo B was 5.6% at a mean of 61 mg·dl-1.
Measurement of Lp(a) was by a one-step sandwich ELISA (Immuno). Cross-reactions with LDL and plasminogen are not detectable. Between batch imprecision (CV) at a mean level of 17.1 mg·dl-1 was 10.9% and at a mean level of 35.9 mg·dl-1 was 10.4%.
Analysis of variance was used for comparison of means between activity groups and two-way analysis of variance was used to adjust for age. LP(a) and triglyceride showed a skewed distribution and were log transformed prior to analysis. The relationship between lipids and apolipoproteins and activity was examined using multiple regression to adjust for the influence of age, body mass index, alcohol intake (none, 1-9, 10+ units per week), smoking (none, cigars, or <10 cigarettes, 10+ cigarettes per day), education (tertiary, secondary, primary), social class (nonmanual, manual), mean blood pressure, diet (good, bad), and coffee drinking (yes, no).
The survey was approved by the Research Ethical Committee of the Faculty of Medicine, The Queens University of Belfast and all participants gave informed consent.
There was 1456 valid addresses and interviews were achieved at 1020, giving a 70% response. Of nonresponders, it was not possible to make contact with 145(10%), 285 (20%) refused interview, and there were six missing questionnaires. Of those who were interviewed, 62% performed the physical appraisal. The final sample interviewed was comparable to the population of the 1991 Northern Ireland Census by sex and age, and those who attended the physical appraisal were representative of those who completed the questionnaire in respect of age, height, weight, physical activity, health, and obesity, although they were more likely to be male and nonsmokers (P ≤ 0.05). The numbers given in the tables are the weighted figures.
Physical Activity, Lipids, and Apolipoproteins
Lipid and lipoprotein distributions by age are shown inFigure 1. Relationships between physical activity, lipids, and lipoproteins were examined by comparison of means before and after adjustment for age (Tables 1-3).
There was a lower mean total cholesterol in those who were most physically active, but only the relationship between cholesterol and habitual activity in females (P ≤ 0.05) and past participation in males (P≤ 0.01) remained significant after age adjustment. There was a significantly higher HDL cholesterol in habitually active males (P≤ 0.05), which remained significant when adjusted for age (P ≤ 0.05). There was a lower LDL cholesterol in males and females who were active when categorized by the highest recorded activity but the relationship in females only (P ≤ 0.05) persisted after age adjustment. Log triglyceride was lower in those who were habitually active, but only the relationship in females (P ≤ 0.05) persisted after age adjustment.
The relationships between activity, defined by habitual activity, the highest recorded activity, past participation, and lipids and lipoproteins were examined using multiple regression with adjustment for age, body mass index, alcohol intake, smoking, education, social class, mean blood pressure, diet, and coffee drinking (Table 4a,b).
The relationship with physical activity is complex. In males, before adjustment for possible confounders, there were lower total cholesterol(P ≤ 0.05) and log triglyceride (P ≤ 0.05) with increasing habitual activity but, after adjustment, these relationships were not sustained. In contrast, those who were most active in the past had higher total cholesterol (P ≤ 0.05) and LDL cholesterol (P≤ 0.05) after adjustment for possible confounders.
In females there were highly significantly lower cholesterol (P≤ 0.01), LDL cholesterol (P ≤ 0.01), and log triglyceride(P ≤ 0.01) in those most habitually active. However, after adjustment for possible confounders, only the relationship with log triglyceride persisted and this applied only when those who were inactive were compared to the least active group. Total cholesterol (P ≤ 0.01), LDL cholesterol (P ≤ 0.05), log triglyceride (P ≤ 0.01), and HDL2 (P ≤ 0.01) were lower in those most active when categorized by the highest recorded activity but after adjustment there was a positive relationship with total cholesterol so that cholesterol was greater by 1.33 mmol·l-1 (51.5 mg·dl-1) in those who had performed light activity compared with those who recorded no activity. This relationship was not present, however, with moderate or vigorous activity. The relationship with LDL cholesterol was present when light activity was compared with inactivity (P ≤ 0.01), and this association was also positive. There were no significant relationships with past participation.
The relationship between activity and Chol:HDL ratio was also examined and there was a lower Chol:HDL ratio in those habitually active males (P≤ 0.05), but only at the highest activity level, and this relationship was not sustained after adjustment for other factors. In females, there was lower Chol:HDL ratio (P ≤ 0.05) in those habitually active but not after adjustment.
The age distributions of apolipoproteins and Lp(a) are shown inTable 5. The relationships between the apolipoproteins and highest recorded activity (Table 6), habitual activity (Table 7), and past activity(Table 8) were explored using analysis of variance. There was a significant relationship between apo AI and activity in males employing both habitual activity (P ≤ 0.01) and the highest recorded activity (P ≤ 0.01). When adjusted for age the relationship with activity remained significant using both habitual activity (P ≤ 0.01) and the highest recorded activity (P ≤ 0.01). Apo AI in the least habitually active group was 151.8 mg·dl-1 compared to 154.8 mg·dl-1 in the most active group. The increase in apo AI with activity was not linear and apo AI was greatest at level 4 where it was 169.4 mg·dl-1. Only in this group of males did the mean apo AI exceed the overall mean apo AI level among females of 166.1 mg·dl-1. There was no significant association between apo AI and physical activity using either habitual activity or the highest recorded activity in females, nor any association with past participation in males or females.
There was a significant increase in apo AII with habitual activity(P ≤ 0.05) in males but this was not evident when adjusted for age. Apo AII in the least active group was 57.4 mg·dl-1 and 61.9mg·dl-1 in the most active group. The increase with activity was not linear and the highest level of apo AII was at level 4 (68.0 mg·dl-1). There was no relationship with the highest recorded activity in males and no association between habitual activity or the highest recorded activity in females. There was no association with past activity in males and females.
Apo B was significantly lower with increasing activity in males defined by the highest recorded activity (P ≤ 0.01), although this relationship was no longer significant when adjusted for age. Apo B was lower with activity in females using both habitual activity (P ≤ 0.01) and the highest recorded activity (P ≤ 0.001). When adjusted for age, the relationship remained significant in females using both habitual activity (P ≤ 0.05) and the highest recorded activity(P ≤ 0.05). Apo B in the least habitually active group was 131.7 mg·dl-1 compared with 107.8 mg·dl-1 in the most active group. There was an association between apo B and past participation in males after adjustment for age (P ≤ 0.05) but not in females. In summary, there was a relationship between activity and apo B in females after adjustment for age.
In neither males nor females was there any association between activity and Lp(a) with the exception of past participation in females (P ≤ 0.05), which was not significant after adjustment for age.
The relationships between activity and apo AI:apo B ratio and HDL:apo AI ratio were also examined. There was a significant relationship between the highest recorded activity and apo AI:apo B ratio in males (P ≤ 0.001) and females (P ≤ 0.001), and with habitual activity and apo AI:apo B ratio in females (P ≤ 0.05) but the relationships were not sustained after age adjustment.
The relationships between activity, defined by habitual activity, the highest recorded activity, and past participation, and apolipoproteins were examined using multiple regression with adjustment for age, body mass index, alcohol intake, smoking, education, social class, mean blood pressure, diet, and coffee drinking (Table 912a,b).
In males there was a highly significant increase with age in apo B(P ≤ 0.001) and a highly significant increase in apo B(P ≤ 0.001) and a decrease in apo AI (P ≤ 0.001) with increasing body mass index. There was a significant inverse relationship between the highest recorded activity and apo AI (P ≤ 0.05) and apo B (P ≤ 0.05). After adjustment for the influence of possible confounders only the relationship between the highest recorded activity and apo AI (P ≤ 0.01) was sustained but the nature and direction of the relationship was unexpected as a positive relationship with apo AI might have been anticipated. Because of the small numbers in the lowest activity categories, the no activity and light activity categories were aggregated and the analyses repeated, and after aggregation the relationship was no longer present. There was a significant reduction in apo AI: apo B (P ≤ 0.05) with past activity after adjustment.
In females there was a highly significant increase with age in apo AI(P ≤ 0.001), apo AII (P ≤ 0.001), apo B (P≤ 0.001), and Lp(a) (P ≤ 0.05) and a highly significantly greater apo B (P ≤ 0.001) with increasing body mass index.
There was a significant inverse relationship between habitual activity and apo B so that those who were most active had more favorable levels. However, after adjustment for possible confounding factors, this relationship no longer persisted. Using the highest recorded activity as the index of physical activity, there was a decrease in apo B (P ≤ 0.01) of 29 mg·dl-1 for the vigorous activity compared to the no activity group. After adjustment for the influence of possible confounding factors, this relationship was sustained (P ≤ 0.05) with a greater apo B of 27 mg·dl-1 between no activity and light activity. The cell sizes, however, in both these groups were small.
In summary, more favourable lipid and apolipoprotein profiles were observed with greater physical activity in males and females, but there was an age related trend that confounded the relationship with activity. When age, body mass index, alcohol intake, smoking, education, social class, mean blood pressure, diet, and coffee drinking were included in multiple regression, the significance level of most relationships was changed, although not all of these factors had a significant independent effect. The main findings were of relationships with activity, both current and past, but these were not sustained after adjustment and the direction of the some of the relationships was unexpected.
There were lower total cholesterol, LDL cholesterol, and triglyceride levels with increasing physical activity. However, these relationships were inconsistent and less than may have been expected from previous cross-section and intervention studies. Age, body mass index, and other covariates had an important influence on lipids and lipoproteins so that after adjustment for their effects, a number of associations were no longer significant.
Previous studies have not shown a consistent association between physical activity and total cholesterol, but in this study there was a significant reduction in cholesterol with increasing physical activity among males and females. There appeared to be a statistical association between activity and cholesterol of a magnitude that would be clinically significant, but after adjustment for confounding factors using multiple regression, this was no longer statistically significant.
While most cross-sectional studies have shown an increase in HDL cholesterol with activity, evidence from population studies has been inconclusive. In this study there was a relationship between physical activity and HDL cholesterol among males only when activity was classified using habitual activity. The difference between the most active group and the least active group was 0.04 mmol·l-1 (1.55 mg·dl-1)(3%). However, the trend was inconsistent and, among those at level 4, HDL cholesterol was 1.42 mmol·l-1 (55.0 mg·dl-1), which represents a difference of 0.2 mmol·l-1 (7.74 mg·dl-1) (16%) from level 0.
The difference in HDL cholesterol with physical activity before age adjustment was less than the difference noted by Haskell(21) in cross-sectional studies between athletes and controls: 0.26 mmol·l-1 (10.1 mg·dl-1) versus 0.78 mmol·l-1 (30.2 mg·dl-1). In a meta-analysis of 66 training studies, mostly short-term high-intensity training programs in young people, Tran and coworkers (69) found that HDL cholesterol increased by only 0.03 mmol·l-1 (1.16 mg·dl-1). The difference between the least active and most active groups in our study was in keeping with the small differences noted in MRFIT (41) of 0.02 mmol·l-1 (0.77 mg·dl-1) and in the Lipid Research Clinics Program Prevalence Study (22) of 0.075 mmol·l-1 (2.90 mg·dl-1) among males, only in the age group 30-49 yr, where a significant relationship in males was not found in all age groups. Shaper and Wannamethee (64) found a similar lack of consistency between activity and HDL cholesterol in a British study where the difference between the most active and the least active subgroup was 0.4 mmol·l-1 (15.5 mg·dl-1) (3%), with a difference only apparent in the most active group.
There was no statistically significant relationship found between activity and HDL cholesterol among females. Previous cross-sectional studies have not always shown a relationship, and the absence of a relationship between activity and HDL cholesterol among females is in keeping with the different response seen in training studies between males and females by Brownell and his colleagues (7), who suggested that males and females differ in their lipid and lipoprotein response to exercise. However, an increase of 0.093 mmol·l-1 (3.60 mg·dl-1) was seen in the most active females in the Lipid Research Clinics Program Prevalence Study (22). Reaven's group (57) also found an inconsistent relationship between exercise and lipoproteins in older males and females such that, while HDL cholesterol increased with moderate leisure time exercise in both males and females, there was an unexpected decrease in HDL cholesterol with heavy exercise. It is interesting to note a similar pattern in this study. In conclusion, the relationship between activity and HDL cholesterol was inconsistent and where the relationship was significant, in males using habitual activity, the margin of difference between the least active and most active groups was small at 0.04 mmol·l-1 (1.55 mg·dl-1).
There was a highly significant lower LDL cholesterol, with the highest recorded activity in both males and females. The difference in LDL cholesterol between the least active (3.52 mmol·l-1 (136.2 mg·dl-1)) and most active males (3.00 mmol·l-1(116.1 mg·dl-1)) was 15% and between the least active (3.51 mmol·l-1 (135.8 mg·dl-1)) and most active (2.63 mmol·l-1 (101.8 mg·dl-1)) females was 25%. There was no significant relationship with habitual activity. Using multiple regression there was no significant relationship seen in males, but there was lower LDL cholesterol in habitually active females and with the highest recorded activity before adjustment for possible confounders. In contrast, after adjustment for these factors, there appeared to be a higher LDL cholesterol with physical activity using the highest recorded activity for light activity compared with inactivity, which was unexpected. The relationship between LDL cholesterol and activity in both cross-sectional and intervention studies has been inconsistent and usually small(12,17). When LDL cholesterol has been found to be lower in endurance athletes, the difference is usually in the range 7%-12% and is seen mostly in very lean runners (21). In the Northern Ireland Health and Activity Survey there was a relationship between activity and LDL cholesterol using the highest recorded activity that was statistically and clinically significant, but it is puzzling why this relationship was only present using the highest recorded activity that represents the maximum activity in the previous 4 wk and not with habitual activity that reflects a more consistent activity pattern.
Those who were most active had lower (log) triglyceride levels. The difference between the least active group and most active group of males using highest recorded activity was 0.22 mmol·l-1 (19.5 mg·dl-1) (16%) while the corresponding difference in females was 0.69 mmol·l-1 (61.1 mg·dl-1) (40%). Using multiple regression, there was a consistent dose response relationship in females but in males a difference was seen only in the most active group. After adjustment for possible confounders, only the difference between inactivity and light activity in females remained significant. Brownell and his coworkers(7) found a different response to training among males and females with a 9% decrease with training in males but a 14% increase in females. Goldberg and colleagues (17) also found a different response to training among males and females but in the opposite direction, with a significant reduction in triglycerides in females but no significant change in males. Cross-sectional and training studies show an inconsistent relationship between activity and triglycerides. Many show decreased triglyceride levels with activity, but most studies are of males so little is known of the response in females. Brownell's group emphasized that males and females differ in their response to exercise and that combining data for males and females may obscure important differences. Wood and Stephanick(82) suggested that triglyceride levels are lower in endurance athletes because trained muscles use fat more efficiently as a fuel.
There was no association seen between activity and HDL2 and HDL3 subfractions, with the exception of a positive association between past activity and HDL3 in males and an inverse association between highest recorded activity and HDL2 in females before adjustment for possible confounders. A stronger association might have been forecast as cross-sectional and training studies have shown a relationship. Williams and his colleagues(76) demonstrated significantly higher HDL2a, HDL2b, and less dense HDL3 among distance runners compared with sedentary controls and difference have been shown in training studies(75,81,83) and in trained patients after myocardial infarction (4). Reports on HDL3 have been variable: Ballantyne and colleagues (4) found no change in HDL3, Wood and coworkers (83) found an increase in HDL3 with training, while Nye and colleagues (55) found an initial increase but subsequent decrease with training, mirroring changes in HDL2. HDL2 and HDL3 are metabolically related: HDL3 becomes HDL2 with the addition of lipids so an increase in HDL2 and a decrease in triglycerides would be consistent with this relationship. However HDL2 and HDL3 subfractions were not significantly related to activity in this study.
While the relationships between total cholesterol, HDL cholesterol, LDL cholesterol, triglyceride, and physical activity were inconsistent, there was an overall trend toward a less atherogenic lipid profile before adjustment for possible confounders. The difference between more active groups and less active groups was small, but relatively small changes in absolute terms may influence population cardiovascular disease: Law and his colleagues(39) estimated that a difference in total cholesterol of 0.6 mmol·l-1 (23.2 mg·dl-1) was associated with a 27% difference in mortality from coronary heart disease in cohort studies and a slightly larger difference in international comparisons. Rifkind(58) found that a 0.026- 0.052 mmol·l-1(1.01-2.01 mg·dl-1) increment in HDL cholesterol was associated with a 2%-4% reduction in coronary heart disease risk, Gordon and his colleagues (18) in estimates based on four prospective U.S. studies suggested that a 1-mg·dl-1 (0.026 mmol·l-1) increment in HDL cholesterol was associated with a 2%-3% decrease in risk of ischemic heart disease, and Salonen's group(61) believed that a similar increase in HDL cholesterol would reduce risk by 4%.
Mean concentrations of apo AI and apo B in males and females in Northern Ireland are greater than those reported in the large Swedish study(29) and those reported in the five different populations described by Alaupovic et al. (1). In the absence of an internationally accepted secondary reference standard it is difficult to compare results between laboratories. Alaupovic and colleagues(1) found that in general females have higher concentrations of apo AI, apo AII and lower apo B than males. They describe the increase with age in apo B in both sexes and find that the increase in apo AI is more evident in males.
There was an association between activity and apolipoproteins apo AI and AII, the protein constituents of HDL in males, but not in females. Apo AI and apo AII are protein constituents of HDL cholesterol and apo B is a protein constituent of LDL cholesterol; thus, changes in apolipoproteins with exercise would be consistent with expected changes in HDL cholesterol and LDL cholesterol. When the relationships were examined in age categories, significant relationships were found in males between activity and apo AI which support the overall results (data not shown).
When these relationships were examined further using multiple regression, the relationship in males was unexpected. Apo AI was lower in those who were most active using the highest recorded activity, after adjustment for age, body mass index, alcohol intake, smoking, education, social class, mean blood pressure, diet, and coffee drinking.
A relationship between activity and apolipoproteins has previously been shown in cross-sectional studies between male athletes and controls(24,37,68). Thompson et al.(68) studied male distance runners with high HDL cholesterol (1.71 mmol·l-1 (66.2 mg·dl-1)), and there was a significant difference between runners and controls in terms of apo AI (170 vs 124 mg·dl-1) and apo AII (73.9 vs 61.7 mg·dl-1). Herbert and his colleagues (24) studied distance runners with a mean HDL cholesterol of 65 mg·dl-1 (1.69 mmol·l-1), compared with controls, and found a significant difference in apo AI (167 vs 139 mg·dl-1). This association was also seen at the other end of the activity spectrum, where Nikkila and his coworkers (53) had shown lower apo AI in immobilized spinal injury patients than controls(males: 117.1 vs 150.8 mg·dl-1, respectively; females: 118.2 vs 178.4 mg·dl-1). In this survey there was a difference in males between the least active group using the highest recorded activity (169.4 mg·dl-1) and the most active group (157.7 mg·dl-1) of 11.7 mg·dl-1, but the cell sizes in the two groups of least activity were very small and when aggregated the difference between the combined least active groups (156.3 mg·dl-1) and the most active group (157.7 mg·dl-1) was 1.4 mg·dl-1 and in the opposite direction. This was not significant before, but was significant after age adjustment (P ≤ 0.05). Using habitual activity, the difference between the least active group (151.8 mg·dl-1) and the most active group (154.8 mg·dl-1) was 3.0 mg·dl-1. The differences in apolipoprotein levels between the most active and least active groups in this study are much less than those seen in case control studies, but there is a much greater gradient in level of physical activity in the case control studies where well-trained distance athletes or those completely immobilized were compared with controls.
Training studies have shown variable results. Kiens and his colleagues(31) found that apo AI was increased by 10% following training in sedentary middle aged males in the absence of confounding factors such as changes in body weight, adiposity, smoking, alcohol, or diet. Thompson and his coworkers (67) found that, even with a 14% increase in HDL cholesterol after a 4-month training program, apo AI and apo AII levels were no different although the biological half life of apo AI and apo AII was greater at the end of the study. They concluded that HDL cholesterol concentration increases were due to prolonged HDL cholesterol survival but contrasted the intensity and duration of training among endurance athletes with the modest training program in their own study, and suggested that the intensity and duration of training had been inadequate to increase apo AI and apo AII.
Apo B was only significantly reduced in females after adjustment for age. The association was strongest using the highest recorded activity as a measure of activity. Little data are available from cross-sectional or training studies on the relationship between apo B and activity but, in one study(81) there was a decrease in apo B with exercise which was significantly inversely correlated with miles run per week. However, there was a decrease in both the exercise group and the control group with no significant difference between them.
In this survey, the difference in mean apo B between the least active group(131.7 mg·dl-1) and the most active group (107.8 mg·dl-1) of females was 23.9 mg·dl-1 (18%) using habitual activity which would be consistent with the reduction in LDL cholesterol with activity. Using multiple regression, the relationship with physical activity was such that those who were most active had lower apo B levels-a dose response relationship. Apo B is associated with coronary heart disease; thus, the lower apo B with activity among females would be consistent with the many epidemiological studies that have shown a reduction in coronary heart disease with activity.
Lp(a) is recognized as an important risk factor for coronary heart disease and most interest has focused on the genetic determinants of Lp(a) levels. There was no change in Lp(a) with activity except with past participation among females, which was not significant after age adjustment, which suggests little change in Lp(a) derived from an increase in physical activity.
Kiens and his colleagues (31) suggested that the influence of physical training on apolipoproteins may be one mechanism by which physical activity offers a protective influence on coronary artery disease. No longitudinal studies have shown a cardiovascular benefit associated with an elevation in apo AI or decrease in apo B, but the metabolic relationship with HDL cholesterol, LDL cholesterol, and case control studies suggest a potential benefit. In this study the decrease in apo B with increasing physical activity is consistent with this potential benefit, but the relationship with apo AI does not support a beneficial effect. In general the differences are much less than those seen in case control and training studies and it may be that normal population variability in physical activity is insufficient to greatly modify apolipoprotein levels.
While the relationship between physical activity and lipid subfractions were inconsistent and differed in males and females, this may be due to the nature of the study. The differences in risk factors within such a homogenous group are modest and may require a larger sample size so that small differences can be detected as significant. Failure to demonstrate significance may thus be due to Type II error.
There are also limitations to the measurement of activity using questionnaires (50,71). Previous epidemiological studies have been criticized because of poor questionnaires and inadequate fitness measurement (30) with the suggestion that questionnaires may have failed to measure physical activity accurately because they did not include all work and leisure activity associated with daily living. Although the Northern Ireland Health and Activity Survey was designed to include the physical activity components of all aspects of daily living, it is possible that the questionnaire used was neither a precise nor sensitive enough measuring instrument to detect subtle variations in physical activity within a relatively homogenous population. In contrast, studies which have in the past shown a strong relationship using a questionnaire may be confounded by other factors.
The relatively small difference in lipoproteins between the most active and least active groups was in contrast to previous case control studies of athletes and controls and some training studies. Differences have been greatest in case control studies, but other differences between the athletes and controls may confound the relationship. Athletes may have constitutional advantages that allow them to train more easily, or athletes may be inherently different with different metabolisms. Williams and his colleagues(77), in a year-long prospective study, found the increase in HDL cholesterol in a group of runners to be directly related to the running mileage: those who ran most miles had the highest initial HDL cholesterol concentrations. They suggested that those who choose to become endurance athletes were different from those who were usually sedentary and that selection bias may be a factor. Thus, HDL cholesterol concentration may influence self-selection of subjects for active and inactive lifestyle, and individuals who choose to become endurance athletes may be metabolically different from sedentary subjects, a difference that is possibly in part genetic (11). In this context, it is interesting to note that in the meta-analysis by Tran and coworkers (69), initial levels of total cholesterol, triglyceride, and HDL cholesterol were strongly correlated with the changes as a result of training. Laporte and colleagues (38) described this phenomenon as “a circular enhancing effect” between innate physiological make-up and selection of lifestyle, i.e., there are two different mechanisms which are related. Beneficial changes may be due to activity, but those who are most fit may find it easier to be active. This is supported by a recent observation(8) that HDL particles obtained before and after acute exercise from runners and nonrunners show different abilities to facilities cholesterol efflux from macrophages.
Case control studies may be confounded by other factors. Those who are physically active may be learner and more muscular, consume a different diet, and be less likely to smoke. Changes in lipoproteins with activity in training studies have been less than in case control studies, which suggests that confounding may have been present. However, although prospective studies may control for confounding in the methodology, it is difficult to eliminate completely even in well-designed prospective studies and bias may occur, for example, through self selection in the recruitment phase.
Few of the training studies had control groups and, as shown by Wood and his coworkers (83), changes may occur in both the active participants and controls, with both groups improving lipoprotein profile but without significant differences between intervention group and controls. Weight loss has been shown to have a marked influence on lipoproteins and may have a large confounding effect in training studies. Williams and his co-workers (75) found that weight loss and exercise had comparable influence on lipoprotein subfractions. In addition, most training studies were relatively short-term interventions and the results may not be applicable to habitual physical activity.
The study demonstrated an association between physical activity and lipids and apolipoprotein subfractions with an overall trend toward a less atherogenic profile before adjustment for age. The associations, however, were not consistent across all subgroups: males and females differed, and the relationship was not always consistent using the two methods of categorizing physical activity. The significance level of most relationships was changed after adjustment using multiple regression, although not all of these factors had a significant independent effect. After adjustment for possible confounding factors, many of the apparently beneficial relationships were not sustained and some fractions were unexpectedly higher. This may suggest that the reduction of cardiovascular disease risk with physical activity may be mediated through factors other than lipids, lipoproteins, or apolipoproteins.
1. Alaupovic, P., W. J. McConathy, J. Fesmire, M. Tavella, and J. M. Bard. Profiles of apolipoproteins and apolipoprotein B containing lipoprotein particles in dyslipoproteinaemia. Clin. Chem.
2. Allied Dunbar National Fitness Survey. Activity and Health Research
. London: Health Education Authority and The Sports Council, 1992.
3. Assmann, G. Genes and dyslipoproteinaemias. Eur. Heart J.
11: (Suppl. H)4-8, 1990.
4. Ballantyne, F. C., R. S. Clark, H. S. Simpson, et al. The effect of moderate physical exercise on the plasma lipoprotein subfractions of male survivors of myocardial infarction. Circulation
5. Blair, S. N., K. H. Cooper, L. W. Gibbons, et al. Changes in coronary heart disease risk factors associated with increased treadmill time in 753 men. Am. J. Epidemiol.
6. Breckenridge, W. C. The role of lipoproteins
and apolipoproteins in prediction of coronary heart disease risk. Clin. Invest. Med.
7. Brownell, K. D., P. S. Bachorik, and R. S. Ayerle. Changes in plasma lipid and lipoprotein levels in men after a program of moderate exercise. Circulation
8. Campaigne, B. N., R. N. Fontaine, C. P. Myung-Sook, and Z. Rymaszewski. Reverse cholesterol transport with acute exercise Med. Sci. Sports Exerc.
9. Cooper, K. H., M. L. Pollock, R. P. Martin, S. R. White, A. C. Linnerud, and A. Jackson. Physical fitness levels vs selected coronary risk factors. J.A.M.A.
10. Dahlen, G. H., J. R. Guyton, M. Attar, J. A. Farmer, J. A. Kautz, and A. M. Gotto, Jr. Association of levels of Lp(a), plasma lipids and other plasma lipoproteins
with coronary artery disease documented by angiography. Circulation
11. Despres, J. P., S. Moorjani, A. Tremblay, et al. Heredity and changes in plasma lipids and lipoproteins
after short term exercise training in males. Arteriosclerosis
12. Dufaux, B., G. Assmann, and W. Hollman. Plasma lipoproteins
and physical activity: a review. Int. J. Sports Med.
13. Durstine, J. L. and W. L. Haskell. Effects of exercise training on plasma lipids and lipoproteins
. In: Exercise and Sport Science Reviews
, J. O. Holloszy (Ed.). Baltimore: Williams and Wilkins, 1994, pp. 477-521.
14. Evans, A. E., E. E. McCrum, and C. C. Patterson. Change of Heart Baseline Clinical Survey. Report to the Health Promotion Agency (NI), The Queen's University of Belfast, 1990.
15. Folsom, A. R., C. J. Caspersen, H. L. Taylor, D. R. Jacobs, R. V. Luepker, and O. Gomez-Marin. Leisure time physical activity and its relationship to coronary risk factors in a population based sample: the Minnesota Heart Survey. Am. J. Epidemiol.
16. Goldberg, L. and D. L. Elliot. The effect of exercise on lipid metabolism in men and women. Sports Med.
17. Goldberg, L., D. L. Elliott, R. W. Schutz, and F. E. Kloster. Changes in lipid and lipoprotein level after weight training.J.A.M.A.
18. Gordon, D. J., J. L. Probstfield, R. J. Carrison, et al. High density lipoprotein cholesterol and cardiovascular disease: four prospective American studies. Circulation
19. Hamsten, A., G. Walldius, A. Szamosi, G. Dahlen, and U. de Faire. Relationship of angiographically defined coronary disease to serum lipoproteins
and apolipoproteins in young survivors of myocardial infarction.Circulation
20. Hartung, H. Diet and exercise in the regulation of plasma lipids and lipoproteins
in patients at risk of coronary heart disease.Sports Med.
21. Haskell, W. L. The influence of exercise training on plasma lipids and lipoproteins
in health and disease. Acta Med. Scand.
22. Haskell, W. L., H. L. Taylor, P. D. Wood, H. Schrott, and G. Heiss. Strenuous physical activity, treadmill exercise test performance and plasma high-density lipoprotein cholesterol. The Lipid Research Clinics Program Prevalence Study. Circulation
62(Suppl. IV):IV53-IV61, 1980.
23. Hellsten, G., K. Boman, G. Hallmas, and G. Dahlen. Lipids and endurance physical activity. Atherosclerosis
24. Herbert, P. N., D. M. Bernier, E. M. Cullinane, L. Edelstein, M. A. Kantor, and P. D. Thompson. High density lipoprotein metabolism in runners and sedentary men. J.A.M.A.
25. Hoefler, G., F. Harnoncourt, E. Paschke, W. Mirtl, K. H. Pfeiffer, and G. M. Kostner. Lipoprotein Lp(a). A risk factor for myocardial infarction. Arteriosclerosis
26. Hubinger, L., L. T. Mackinnon, and F. Lepre. Lipoprotein (a) [Lp(a)] levels in middle-aged male runners and sedentary controls. Med. Sci. Sports Exerc.
27. Huttunen, J. K., E. Lansimes, E. Voutilainen, et al. Effect of moderate physical exercise on serum lipoproteins
. A controlled clinical trial with special reference to serum high density lipoproteins
28. Israel, R. G., M. J. Sullivan, R. H. L. Marks, R. S. Cayton, and T. C. Chenier. Relationship between cardiorespiratory fitness and lipoprotein (a) in men and women. Med. Sci. Sports Exerc.
29. Jungner, I., G. Walldius, I. Holme, W. Kolar, E. Steiner. Apolipoprotein B and A1 in relation to serum cholesterol and triglycerides in 43,000 Swedish males and females. Clin. Lab. Res.
30. Kannel, W. B., P. Wilson, and S. N. Blair. Epidemiological assessment of the role of physical activity and fitness in development of cardiovascular disease. Am. Heart J.
31. Kiens, B., I. Jorgensen, S. Lewis, et al. Increased plasma HDL-cholesterol and apo A-1 in sedentary middle-aged men after physical conditioning. Eur. J. Clin. Invest.
32. Kiens, B., H. Lithell, and B. Vessby. Further increase in high density lipoprotein in trained males after enhanced training.Eur. J. Appl. Physiol.
33. Kish, L. Survey Sampling
. New York: Wiley, 1965, pp. 398-401.
34. Kostner, G. M., E. Marth, K. P. Pfeiffer, and H. Wege. Apolipoproteins A1, A11 and HDL phospholipids but not Apo B
are risk indicators for occlusive cerebrovascular disease. Eur. Neurol.
35. Kostner, G. M., E. Molinari, and P. Pichler. Evaluation of a new HDL2/HDL3 quantitation method based on precipitation with polyethylene glycol. Clin. Chem. Acta
36. Krauss, RM. Exercise, lipoproteins
and coronary artery disease. Circulation
37. Krauss, R. M., F. T. Lindgren, P. D. Wood, P. D. Haskell, J. J. Albers, and M. C. Cheung. Differential increases in plasma high density lipoprotein subfractions and apolipoproteins in runners.Circulation
38. Laporte, R. E., S. Dearwater, J. A. Cauley, C. Slemenda, and T. Cook. Physical activity or cardiovascular fitness: which is more important for health? Phys. Sportsmed.
39. Law, M. R., N. J. Wald, and S. G. Thompson. By how much and how quickly does reduction in serum cholesterol concentration lower risk of ischaemic heart disease? Br. Med. J.
40. Lehtonen, A., J. Viikari, and C. Ehnholm. The effect of exercise on high density lipoprotein apoproteins. Acta Physiol. Scand.
41. Leon, A. S., J. Connett, D. R. Jacobs, and R. Rauramaa. Leisure time physical activity and the risk of coronary heart disease and death: MRFIT. J.A.M.A.
42. Lewis, B. Diet and exercise as regulators of lipid risk factors. Drugs
40(Suppl. 1):19-25, 1990.
43. MacAuley, D., E. E. McCrum, G. Stott, et al. The Northern Ireland Health and Activity Survey Report. Belfast: HMSO, 1994.
44. Maciejko, J. J., D. R. Holmes, B. A. Kottke, et al. Apolipoprotein A-1 as a marker for angiographically assessed coronary artery disease. N. Engl. J. Med.
45. Mbewu, A. D. and P. N. Durrington. Lipoprotein (a) structure, properties, and possible involvement in thrombogenesis and atherogenesis. Atherosclerosis
46. McArdle, W. D., F. L. Katch, and V. L. Katch.Exercise Physiology, Energy, Nutrition and Human Performance
. Philadelphia: Lea and Febiger, 1986.
47. Miller, N. E., S. Rao, B. Lewis, G. Bjorsvik, K. Myhre, and O. D. Mjos. High density lipoprotein and physical activity.Lancet
48. Montoye, H. J. Estimation of habitual physical activity by questionnaire and interview. Am. J. Clin. Nutr.
49. Montoye, H. J., W. D. Block, and R. Gayle. Maximal oxygen uptake and blood lipids. J. Chronic Dis.
50. Montoye, H. J. and H. L. Taylor. Measurement of physical activity in population studies: a review. Hum. Biol.
51. MRFIT. Multiple Risk Factor Intervention Trial Research Group. Relationship between baseline risk factors and coronary heart disease and total mortality in the Multiple Risk Factor Intervention Trial.Prev. Med.
52. Nestel, P. J., M. Podkolinski, and N. H. Fidge. Marked increase in high density lipoproteins
53. Nikkila, E., T. Kuusi, and P. Myllynen. High density lipoprotein and apolipoprotein A1 during physical inactivity.Athlerosclerosis
54. Noma, A., T. Yokosuka, and K. Kitamura. Plasma lipids and apolipoprotein as discriminators for presence and severity of angiographically defined coronary artery disease. Atherosclerosis
55. Nye, E., K. Carlson, P. Kirstein, and S. Rossner. Changes in high density lipoprotein subfractions and other lipoproteins
induced by exercise. Clin. Chem. Acta
56. Passmore, R. and J. V. G. A. Durnin. Energy Work and Leisure
. London: Heinemann 1967.
57. Reaven, P. D., J. B. McPhilips, E. L. Barrett-Connor, and M. H. Criqui. Leisure time exercise and lipid and lipoprotein levels in an older population. J. Am. Geriatr. Soc.
58. Rifkind, B. M. Gemfibrozil lipids and coronary risk.N. Engl. J. Med.
59. Rosengren, A., L. Wilhelmsen, E. Eriksson, B. Risberg, and H. Wedel. Lipoprotein (a) and coronary heart disease: a prospective case-control study in a general population sample of middle aged men.Br. Med. J.
60. Sady, S. P., P. D. Thompson, E. M. Cullinane, M. A. Kantor, E. Domagala, and P. N. Herbert. Prolonged exercise augments plasma triglyceride clearance. J.A.M.A.
61. Salonen, J. T., R. Salonen, K. Seppanen, R. Rauramaa, and J. Tuomilehto. HDL, HDL2 and HDL3 subfractions and the risk of acute myocardial infarction. a prospective population study in eastern Finnish men.Circulation
62. Scanu, A. M. and G. M. Fless. Lipoprotein (a) heterogenecity and biological relevance. J. Clin. Invest.
63. Schwartz, R. S. The independent effects of dietary weight loss and aerobic training on high density lipoproteins
and apolipoprotein A1 concentrations in obese men. Metabolism
64. Shaper, A. G. and G. Wannamethee. Physical activity and ischaemic heart disease in middle aged British men. Br. Heart J.
65. Stubbe, I., P. Hansson, A. Gustafson, and P. Nilsson. Plasma lipoproteins
and lipolytic enzyme activities during endurance training in sedentary men: changes in high-density lipoprotein subfractions and composition. Metabolism
66. Superko, H. R. Exercise training, serum lipids, and lipoprotein particles: is there a change threshold? Med. Sci. Sports Exerc.
67. Thompson, P. D., E. M. Cullinane, S. P. Sady, et al. Modest changes in high density liporotein concentration and metabolism with prolonged exercise training. Circulation
68. Thompson, P. D., B. Lazarus, E. Cullinane, et al. Exercise, diet or physical characteristics as determinants of HDL-levels in endurance athletes. Atherosclerosis
69. Tran, Z. V. A. Weltman, G. V. Glass, and D. P. Mood. The effects of exercise on blood lipids and lipoproteins
: a meta-analysis of studies. Med. Sci. Sports Exerc.
70. Uterman, G. The mysteries of lipoprotein (a).Science
71. Washburn, R. A., H. L. Taylor, H. J. Montoye. The assessment of physical activity by questionnaire. Am. J. Epidemiol.
72. Weiner, J. S. and J. A. Lourie. Human Biology, a Guide to Field Methods
. Cambridge, MA: IBP, Blackwell Scientific, 1969.
73. Williams, P. T. Weight set point and high density lipoprotein concentration of long distance runners. Metabolism
74. Williams, P. T., R. M. Krauss, K. M. Vranizan, J. J. Albers, and P. D. Wood. Effects of weight loss by exercise and by diet on apolipoproteins A1 and A11 and the particle size distribution of high density lipoproteins
in men. Metabolism
75. Williams, P. T., R. M. Krauss, K. M. Vranizan, and P. D. S. Wood. Changes in lipoprotein subfractions during diet-induced and exercise induced weight loss in moderately overweight men.Circulation
76. Williams, P. T., R. M. Krauss, P. D. Wood, F. T. Lindgren, C. Giotas, and K. M. Vranizan. Lipoprotein subfractions of runners and sedentary men. Metabolism
77. Williams, P. T., P. D. Wood, W. L. Haskell, et al. The effects of running mileage and duration on plasma lipoprotein levels.J.A.M.A.
78. Williams, P. T., P. D. Wood, R. M. Krauss, et al. Does weight loss effect the exercise-induced increase in plasma high-density lipoproteins
79. Wilson, P. W. F., R. S. Paffenbarger, Jr., J. N. Morris, and R. J. Havlik. Assessment methods for physical activity and physical fitness in population studies: Report of NHLBI workshop. Am. Heart J.
80. Wood, P. D. and W. L. Haskell. The effect of exercise on plasma high density lipoproteins
81. Wood, P. D., W. L. Haskell, S. N. Blair, et al. Increased exercise level and plasma lipoprotein concentrations: a one year randomized study in sedentary middle-aged men. Metabolism
82. Wood, P. D. and M. L. Stephanick. Exercise, fitness and health. a consensus of current knowledge. In: Exercise, Fitness and Health
, C. Bouchard, R. J. Shepherd, T. Stephens, J. Sutton, and B. McPherson (Eds.). Champaign, IL: Human Kinetics, 1988.
83. Wood, P. D., M. L. Stefanick, D. M. Dreon, et al. Changes in plasma lipids and lipoproteins
in overweight men during weight loss through dieting as compared with exercise. N. Engl. J. Med.
84. Zenker, G., P. Koltringer, G. Bone, K. Niederkorn, K. Pfeiffer, and G. Jurgens. Lipoprotein (a) as a strong predictor of cerebrovascular disease. Stroke