Physical activity is known to be associated with lower risk of cardiovascular disease (CVD). Compared with sedentary individuals, persons reporting regular moderate activity have a 20% lower risk of CVD; at higher amounts or with greater intensity, CVD risk reduction is even greater (25).
The effects of physical activity on risk of myocardial infarction (MI) are well documented and may be mediated in part through beneficial changes in blood lipids, inflammatory markers, and insulin sensitivity (7,13,19,25). Strong dose-response associations between exercise intensity and blood lipids-especially HDL-C and triglycerides-have been reported in both observational studies and randomized trials (14). In a randomized trial conducted by Kraus et al. (13), exercise at a caloric equivalent of 20 miles·wk−1 and an intensity equivalent to that of jogging at a moderate pace significantly increased HDL-C by 9% and decreased triglycerides by 13%, although these had no effect on total LDL cholesterol (LDL-C). In the National Health and Nutrition Examination Survey (NHANES), C-reactive protein (CRP) was lower with greater levels of leisure time activity (6). After adjusting for CVD risk factors, the odds ratio for elevated CRP concentration (dichotomized at the ≥85th percentile of the sex-specific distribution) was 0.53 (95% confidence interval (CI) = 0.40-0.71) for participants who engaged in vigorous physical activity three or more times per week compared with sedentary participants. Finally, a meta-analysis of 14 trials of physical activity interventions lasting ≥8 wk found that exercise training reduced hemoglobin A1c (HbA1c) levels by 0.66% on average among middle-aged persons with diabetes (3).
The purpose of this study was to determine the extent to which these and other potential mediating novel biomarkers explain the inverse association between physical activity and risk of CHD among men.
METHODS
Study population.
The Health Professionals Follow-up Study (HPFS) began in 1986, when 51,529 male health professionals age 40-75 yr completed the baseline six-page questionnaire. On biennial, self-administered questionnaires, we collected detailed information on lifestyle characteristics and incident disease.
Between January 1, 1993, and December 31, 1995, a blood sample was requested from all surviving cohort participants; 18,225 provided samples. The men who provided samples were somewhat younger but were otherwise similar with respect to smoking, physical activity, and diet. Among men without CVD or cancer before blood draw, we identified 454 men with incident nonfatal MI (n = 352) or fatal CHD (n = 102) between the date of blood collection and January 31, 2004. Using risk set sampling (24), we selected controls (2:1) matched on age (in 5-yr increments), smoking (in five categories), and month of blood return. For this study, we excluded men (n = 103) who indicated at blood draw that they had a physical impairment. In addition, we excluded men missing physical activity (n = 1) or traditional lipid markers (n = 22). The final analysis included 412 cases and 827 controls. The study was approved by the Committee on the Use of Human Subjects in Research at the Brigham and Women's Hospital. All subjects gave informed consent to participate.
Assessment of biomarkers.
Participants underwent local phlebotomy. Blood samples were collected and returned via overnight courier to the HPFS blood storage and processing facility as described elsewhere (7). The biomarkers in this sample were measured using standard methods (7) and generally showed excellent stability and reproducibility during simulated transport and storage (20). Candidate biomarkers assessed in all men included HDL cholesterol (HDL-C), LDL-C, triglycerides, total cholesterol, apolipoprotein B (Apo B), HbA1c, high-sensitivity CRP (21), fatty acid-binding protein 4, lipoprotein-associated phospholipase A2, and 25-hydroxyvitamin D (25(OH)D). We used the triglyceride/HDL-C ratio as a measure of insulin resistance (17). A subset of biomarkers, which included soluble tumor necrosis factor receptors 1 and 2, interleukin-6, soluble intercellular adhesion molecule 1, soluble vascular cell adhesion molecule 1, lipoprotein(a), fibrinogen, adiponectin (23), and oxidized LDL, were only assessed among the cases and controls accrued from 1994 to 2000 (n = 731).
Assessment of physical activity.
Leisure time physical activity was assessed by questionnaire every 2 yr since baseline through a series of questions on average total time per week spent on each of 10 activities during the previous year. Walking pace, categorized as casual (<2 miles·h−1), normal (2-2.9 miles·h−1), brisk (3-3.9 miles·h−1), or striding (≥4 miles·h−1), was also assessed. A MET score was assigned to each activity based on its energy cost (1). Vigorous activities (MET values ≥ 6) were jogging, running, bicycling, swimming, tennis, squash/racquetball, and calisthenics/rowing. Moderate activities (3 ≤ METs < 6) were brisk walking, heavy outdoor work, and weightlifting. The validity and reproducibility of the physical activity measurement have been reported in detail elsewhere (4). Briefly, correlations between four 1-wk diaries and the questionnaire for total activity and vigorous activity were 0.65 and 0.58, respectively.
We chose to focus on vigorous activity because of its stronger associations with CHD risk (14,25). To represent long-term levels of physical activity more accurately and to reduce measurement error (12), we calculated the average number of hours of vigorous activity per week from the 1990, 1992, and 1994 questionnaires. We categorized participants into quintiles based on the distribution in controls.
Assessment of nonfatal MI and fatal CHD.
Self-reported myocardial infarctions were confirmed according to the World Health Organization criteria that included symptoms plus either diagnostic ECG changes or elevated cardiac enzymes (26). Fatal CHD was confirmed by hospital records or an autopsy or by CHD listed as the cause of death on the death certificate, if it is listed as an underlying and most plausible cause of death and if evidence of previous CHD was available. We included sudden cardiac deaths, defined as death within 1 h of symptom onset with no previous serious illness and no other plausible cause.
Statistical analysis.
All analyses were performed using SAS statistical software, version 9.1 (SAS Institute, Inc., Cary, NC). We calculated odd ratios as unbiased estimates of incidence rate ratios from conditional logistic regression models, stratifying on the matching factors. Because some of the matched sets of cases and controls were broken after exclusions, we additionally used unconditional logistic regression for some analyses to maximize power. Similar results were obtained for both, thus conditional logistic regression results are presented. In multivariable analyses, we adjusted for moderate activity (five categories), batch (case before or after 2002), fasting status (<8 h, ≥8 h), parental history of MI at or before age 60 yr (yes/no), alcohol consumption (five categories), aspirin use (yes/no), vitamin E supplement use (yes/no), energy-adjusted intake of polyunsaturated fat, trans-fatty acids, omega-3 fatty acids, and fiber (each in quintiles), as well as history of diabetes, hypertension, and hypercholesterolemia diagnosed in 1992 or earlier (to avoid adjusting for an intermediate in the causal pathway). We used covariate information primarily from 1994, the time of blood draw. Information from previous questionnaires was used when covariate data were missing (<6% was missing). Tests for linear trend were performed using the median value for each category of physical activity.
To examine the extent to which various CVD biomarkers contributed to the risk reduction associated with vigorous-intensity physical activity, we initially assigned the median of each physical activity category to construct a continuous variable. We then calculated the rate ratio associated with a 3-h·wk−1 increase in vigorous activity, which corresponded to the difference between the highest and the lowest quintile. We considered the magnitude of change in the regression coefficient for vigorous-intensity activity with and without adjustment for each potential mediator. The percent of physical activity-CHD association explained by the intermediate variable was computed as follows: (1 − (βmediator-adjusted model / βmultivariable model)) × 100%. A positive change in the regression coefficient indicates a change in the rate ratio toward the null. The SEE was calculated using equation 5 from Lin et al. (16).
Because a subset of the biomarkers was only assessed among the cases and controls accrued from 1994 to 2000, we restricted analyses examining these potential mediators to this group (n = 731).
RESULTS
Table 1 shows the characteristics of case and control participants in the HPFS. As expected, traditional CVD risk factors were more common among cases compared with controls. Cases had a lower HDL-C and higher LDL-C and were also more likely to have a diagnosis of hypertension, hypercholesterolemia, and diabetes.
;)
TABLE 1: Baseline characteristics of 412 incident MI cases and 827 matched controls in the HPFS.
Men reporting greater vigorous-intensity physical activity had a lower body mass index (BMI) and better biomarker profiles compared with men reporting lesser amounts of activity (Table 2). HDL-C and vitamin D were higher, whereas triglycerides, CRP, Apo B, and HbA1c were lower in men reporting more vigorous activity compared with less activity. An exception was LDL-C, which was fairly constant across categories of physical activity. Although the Ptrend for Apo B was not statistically significant in the age-adjusted analysis, when additionally adjusted for other potential confounders, vigorous activity was significantly associated with lower levels of Apo B (Ptrend = 0.02; data not shown).
TABLE 2: Age-adjusted mean biomarker levels by quintiles of average vigorous activity among 827 male health professionals, HPFS.
There was a significant inverse association between average vigorous activity and risk of MI (Ptrend = 0.03; Table 3). In multivariable-adjusted models additionally adjusted for preexisting hypertension, hypercholesterolemia, and diabetes, the relative risk (RR) of MI was 0.69 (95% CI = 0.45-1.05) for men who reported ≥3 h of vigorous activity per week compared with men reporting no vigorous activity. When treated as a continuous variable, the RR for a 3-h·wk−1 increase in vigorous activity was 0.78 (95% CI = 0.61-0.98). The RR (95% CI) for the association between moderate physical activity (in quintiles) and risk of MI, adjusted for the same covariates as well as vigorous activity, were 1.13 (0.77-1.64), 0.96 (0.65-1.43), 0.76 (0.51-1.14), and 0.84 (0.56-1.26; Ptrend = 0.17; data not shown). Although the inverse association between moderate activity and CHD was not statistically significant within this nested case-control study, we have found a statistically significant inverse association within the whole HPFS cohort (RR for a 1-h·wk−1 increase in moderate activity: 0.97 (95% CI = 0.95-0.98)).
TABLE 3: Rate ratios for incident MI according to categories of average vigorous-intensity physical activity in 1990-1994.
We added potential mediating biomarkers individually into multivariable models to assess the magnitude to which each biomarker changed the association of vigorous-intensity physical activity with risk of MI (Table 4). When compared with the multivariate RR of MI of 0.78 (95% CI = 0.61-0.98) for each 3-h·wk−1 increase in vigorous activity, adjustment for HDL-C alone attenuated the RR to 0.86 (95% CI = 0.67-1.09). As noted in Table 4, this suggests that approximately 38% (95% CI = 3%-74%) of the inverse association between vigorous-intensity physical activity and CHD is mediated through HDL-C. Similar analyses for triglycerides (22%), vitamin D (21%), Apo B (19%), HbA1c (14%), and CRP (9%) also attenuated the RR. In addition, in the subset of 731 men in which it was measured, adjustment for adiponectin attenuated the RR by 52% (95% CI = −16% to 121%). The smaller sample size in this subgroup of men resulted in wide CI.
TABLE 4: Comparison of rate ratios for incident MI for a 3-h·wk−1 increase in vigorous-intensity PA after adjustment for mediating biomarkersa among men (n = 1239).
In multivariable analyses, we also adjusted for the combination of biomarkers where the P value for percent change was 0.15 or less when modeled separately. We excluded adiponectin because it was not measured in all men. Triglycerides and HDL were individually included rather than their ratio. When we adjusted for the combination of HDL-C, triglycerides, vitamin D, Apo B, HbA1c, and CRP, the regression coefficient for the association between vigorous activity and CHD was attenuated by 70% (95% CI = 12%-128%; Table 4). In this model, triglycerides and CRP were no longer significantly associated with risk of CHD; thus, the most parsimonious model included HDL-C, vitamin D, Apo B, and HbA1c. This final model attenuated the RR of CHD associated with vigorous-intensity activity by 70% (95% CI = 12%-127%; Fig. 1).
FIGURE 1: Main mediators of the association between physical activity and CHD: percent of physical activity-CHD association explained by risk factors. The rate ratios are conditioned on the matching factors of age, smoking status, and date of blood sample return and adjusted for moderate activity, batch, fasting status, parental history of MI at or before age 60 yr, aspirin, vitamin E supplement use, intake of polyunsaturated fat, trans-fat, EPA + DHA, fiber, alcohol intake, and preexisting hypertension, diabetes, or hypercholesterolemia. The percent of physical activity effect explained by the intermediate variable was computed as follows: (1 − (β mediator-adjusted model / multivariable model)) × 100%.
We found that BMI only explained 8% (95% CI = −4% to 20%) of the inverse association between physical activity and CHD. As an alternative to BMI, we examined waist circumference as a potential mediator of this association. Adjusting for waist circumference attenuated the RR to a similar extent as BMI (11%); however, we did not include waist circumference in our final results because it was only measured in a subset of individuals.
We performed a sensitivity analysis only including men that were fasting at least 8 h before blood draw (n = 725). The percent explained was attenuated, but HDL-C, vitamin D, Apo B, and HbA1c still explained 46% (95% CI = 8%-85%) of the inverse association. We also performed the analysis after adjusting for antihypertensive and cholesterol-lowering medications and obtained similar results to those presented. Lastly, we performed the analysis excluding cases that occurred in the first 2 yr of follow-up and found a modestly attenuated total percent explained (54%; 95% CI = 17%-89%).
DISCUSSION
In this prospective analysis of middle-aged men, vigorous exercise was associated with a lower risk of CHD. A 3-h·wk−1 increase in vigorous-intensity physical activity was associated with a 22% (95% CI = 2%-39%) lower risk of MI. In multivariable models including preexisting CVD-related conditions, potential biological mediators of effect-HDL-C, vitamin D, Apo B, and HbA1c-attenuated the RR by 70% (95% CI = 12%-127%).
Other studies have investigated the potential mediators between physical activity and risk CHD, finding that biomarkers explain 22%-40% of the inverse association (2,11,18,28). Inflammatory factors have been identified as mediators in all studies, explaining 13%-28% of the association between exercise and CHD. Hamer and Stamatakis (11) found that CRP and fibrinogen explained 13.2% of the association between physical activity and total CVD, which is similar to what we found. In the Women's Health Study (18), inflammatory markers explained 20.9%, novel lipids 11.1%, traditional lipids 13.4%, HbA1c 5%, and BMI 6.8%. All risk factors combined explained 35.5% of the risk reduction. A larger proportion of the inverse association between physical activity and CHD risk was explained by the potential mediators we investigated when compared with previous studies. Potential explanations for this difference may be better exercise assessment in our cohort with the repeated measures of activity over 4 yr and the greater variability in moderate and vigorous physical activity in this cohort of men.
We also found that a modest proportion of the physical activity-CHD association was explained by serum 25(OH)D, which was not accounted for in previous studies. Although not a traditional cardiovascular risk factor, low levels of serum 25(OH)D have previously been found to be associated with increased risk of MI (8). In addition, physical activity has been associated with increased levels of serum 25(OH)D likely due to increased sun exposure, a strong determinant of serum 25(OH)D (9).
There are several possible mechanisms by which physical activity decreases risk of CHD. One pathway is through antiatherogenic effects resulting from improved inflammation, lipid profile, insulin sensitivity, and body fat (3,6,7,13,19,22,25). In the current analysis, we found these pathways explain approximately 45%-70% of the decreased risk of CHD associated with physical activity. In addition, physical activity may induce favorable changes in blood viscosity, blood pressure, and hemostatic factors that decrease risk of thrombogenesis and thrombosis (5,19). Physical activity also decreases myocardial O2 demand and improves coronary blood flow and endothelial function, which decrease risk of myocardial ischemia. A decreased risk of ventricular arrhythmias results from increased vagal tone and decreased adrenergic activity (10,15). Future studies are necessary to determine which of the above mechanisms account for the remainder of risk reduction.
Strengths of our study include the prospective design, detailed information on physical activity and covariates, a wide range of cardiovascular biomarkers, long duration of follow-up for the end point of MI, and strict criteria for defining coronary events.
Our study also has limitations that should be considered. In a mediator analysis, we must be aware of potential confounders for the main effect association and the association between mediators and CHD. We adjusted for age, smoking, diet, alcohol use, and preexisting disease. Unmeasured confounding by other variables, particularly biomarkers or genes that may be associated with both mediator and CHD, remains a potential limitation.
Our study population, consisting of male health professionals, is not representative of the general population. Therefore, we cannot necessarily generalize our results to other populations with different educational levels, incomes, or distributions of race and ethnicity. Nonetheless, the relative homogeneity of the cohort in socioeconomic status may actually enhance the internal validity of this study.
We measured biomarkers at a single point in time. One measurement may not reflect true levels over time. To examine the potential effect of measurement error, we used data from men in a reproducibility study to correct for random within-person measurement error in HDL-C (27). After correcting for measurement error, we found that HDL-C explained 50% of the physical activity-CHD association compared with the uncorrected estimate of 38%.
The mediators determined to explain the greatest proportion of association may be statistical intermediates but may not be in the causal pathway. The pathway between physical activity and CHD is likely quite complex; thus, the most significant mediators may simply be markers of various pathways and not causal components. In addition, as we fit the parsimonious model by selecting mediators with the most significant percent change when modeled separately, the 70% attenuation of the physical activity-CHD association may be optimistic.
Measurement error in physical activity is unlikely to bias our results because data were assessed prospectively; thus, errors in physical activity are likely to be nondifferential with respect to subsequent MI status. In addition, reporting error is unlikely to be associated with measurement of biomarkers. Random misclassifications would be expected to lead to underestimation of the true association.
Engaging in regular vigorous-intensity activity is associated with a lower risk of MI among men. This inverse association can be partially explained by the beneficial effects of physical activity on HDL-C, vitamin D, Apo B, and HbA1c. Although the inverse association attributable to these biomarkers is substantial, future research should explore benefits of exercise beyond these biomarkers of risk.
This study was supported National Institutes of Health grants AA11181 and HL35464. A. K. Chomistek was supported by an institutional training grant (HL07575) from the National Heart, Lung, and Blood Institute.
The authors thank the participants of the HPFS for their continued cooperation and participation, Dr. Walter Willett for his comments and suggestions in the preparation of this article, Lydia Liu and Mathew Pazaris for their helpful statistical assistance, and Kathy Bush and Janine Neville-Golden for their laboratory support.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
REFERENCES
1. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities: classification of energy costs of human physical activities.
Med Sci Sports Exerc. 1993;25(1):71-80.
2. Boekholdt SM, Sandhu MS, Day NE, et al. Physical activity, C-reactiveprotein levels and the risk of future coronary artery disease in apparently healthy men and women: the EPIC-Norfolk prospective population study.
Eur J Cardiovasc Prev Rehabil. 2006;13:970-6.
3. Boule NG, Haddad E, Kenny GP, Wells GA, Sigal RJ. Effects of exercise on glycemic control and body mass in type 2 diabetes mellitus: a meta-analysis of controlled clinical trials.
JAMA. 2001;286:1218-27.
4. Chasan-Taber S, Rimm EB, Stampfer MJ, et al. Reproducibility and validity of a self administered physical activity questionnaire for male health professionals.
Epidemiology. 1996;7:81-6.
5. Cornelissen VA, Fagard RH. Effects of endurance training on blood pressure, blood pressure-regulating mechanisms, and cardiovascular risk factors.
Hypertension. 2005;46(4):667-75.
6. Ford ES. Does exercise reduce inflammation? Physical activity and C-reactive protein among U.S. adults.
Epidemiology. 2002;13:561-8.
7. Fung TT, Hu FB, Yu J, et al. Leisure-time physical activity, television watching, and plasma biomarkers of obesity and cardiovascular disease risk.
Am J Epidemiol. 2000;152:1171-8.
8. Giovannucci E, Liu Y, Hollis BW, Rimm EB. 25-Hydroxyvitamin D and risk of myocardial infarction in men.
Arch Intern Med. 2008;168(11):1174-80.
9. Giovannucci E, Liu Y, Rimm EB, et al. Prospective study of predictors of vitamin D status and cancer incidence and mortality in men.
J Natl Cancer Inst. 2006;98:451-9.
10. Hambrecht R, Wolf A, Gielen S, et al. Effect of exercise on coronary endothelial function in patients with coronary artery disease.
N Engl J Med. 2000;342:454-60.
11. Hamer M, Stamatakis E. Physical activity and risk of cardiovascular disease events: inflammatory and metabolic mechanisms.
Med Sci Sports Exerc. 2009;41(6):1206-11.
12. Hu FB, Stampfer MJ, Colditz GA, et al. Physical activity and risk of stroke in women.
JAMA. 2000;283:2961-7.
13. Kraus WE, Houmard JA, Duscha BD, et al. Effects of the amount and intensity of exercise on plasma lipoproteins.
N Engl J Med. 2002;347:1483-92.
14. Lee IM.
Epidemiologic Methods in Physical Activity Studies. Oxford (UK): Oxford University Press; 2009. p. 328.
15. Leung FP, Yung LM, Laher I, Yao X, Chen ZY, Huang Y. Exercise, vascular wall and cardiovascular diseases: an update (part 1).
Sports Med. 2008;38(12):1009-24.
16. Lin DY, Fleming TR, De Gruttola V. Estimating the proportion of treatment effect explained by a surrogate marker.
Stat Med. 1997;16:1515-27.
17. McLaughlin T, Abbasi F, Cheal K, Chu J, Lamendola C, Reaven G. Use of metabolic markers to identify overweight individuals who are insulin resistant.
Ann Intern Med. 2003;139:802-9.
18. Mora S, Cook N, Buring JE, Ridker PM, Lee IM. Physical activity and reduced risk of cardiovascular events: potential mediating mechanisms.
Circulation. 2007;116:2110-8.
19. Mora S, Lee IM, Buring JE, Ridker PM. Association of physical activity and body mass index with novel and traditional cardiovascular biomarkers in women.
JAMA. 2006;295:1412-9.
20. Pai JK, Curhan GC, Cannuscio CC, Rifai N, Ridker PM, Rimm EB. Stability of novel plasma markers associated with cardiovascular disease: processing within 36 hours of specimen collection.
Clinical Chemistry. 2002;48:1781-4.
21. Pai JK, Pischon T, Ma J, et al. Inflammatory markers and the risk of coronary heart disease in men and women.
N Engl J Med. 2004;351:2599-610.
22. Pischon T, Hankinson SE, Hotamisligil GS, Rifai N, Rimm EB. Leisure-time physical activity and reduced plasma levels of obesity-related inflammatory markers.
Obesity Res. 2003;11(9):1055-64.
23. Pischon T, Hotamisligil GS, Rimm EB. Adiponectin: stability in plasma over 36 hours and within-person variation over 1 year.
Clin Chem. 2003;49:650-2.
24. Prentice RL, Breslow NE. Retrospective studies and failure time models.
Biometrika. 1978;65:153-8.
25. Physical Activity Guidelines Advisory Committee.
Physical Activity Guidelines Advisory Committee Report, 2008. Washington (DC): US Department of Health and Human Services; 2008. p. 683.
26. Rose GA, Blackburn H, Gillum R, Prineas RJ.
Cardiovascular Survey Methods: WHO Monograph Series No. 56. Geneva (Switzerland): World Health Organization; 1982. pp. 162-5.
27. Rosner B, Spiegelman D, Willett WC. Correction of logistic regression relative risk estimates and confidence intervals for random within-person measurement error.
Am J Epidemiol. 1992;136(11):1400-13.
28. Wennberg P, Wensley F, Johansson L, et al. Reduced risk of myocardial infarction related to active commuting: inflammatory and haemostatic effects are potential major mediating mechanisms.
Eur J Cardiovasc Prev Rehabil. 2010;17:56-62.
