Prostate cancer is the second leading cause of death from cancer among men in the U.S.; during 1993, prostate cancer accounted for 13% of all male cancer deaths (14). An estimated 165,000 cases are being diagnosed in the U.S. each year, making prostate cancer the most frequently diagnosed cancer among men (14). Blacks in the U.S. have the highest incidence rates of prostate cancer, followed by whites from North America (53). The etiology of prostate cancer is unknown. Studies have been conducted examining potential risk factors such as diet, venereal diseases, sexual habits, smoking, occupational exposures, and physical activity; however, the results have been inconsistent(21,53).
It has been suggested that physical activity may protect against the development of prostate cancer through reduction of levels of endogenous testosterone (24,44). Male athletes have been shown to have lower levels of circulating testosterone than nonathletes, and men who exercise have reduced testosterone levels immediately after an exercise session(1,20,26,27,37,49,51). The physical activity-prostate cancer hypothesis is compatible with the fact that the standard treatment for prostate cancer is antitestosterone therapy in which the circulating androgen, testosterone, is essentially reduced to levels observed after castration (24). Furthermore, some studies have shown that men with prostate cancer have higher levels of endogenous testosterone compared with nondiseased men, and cancerous tissue has been reported to have higher levels of testosterone compared with normal tissue (2,22,25,28,31).
Despite these hypothesized mechanisms, previous studies have provided inconclusive findings on the relationship between physical activity and prostate cancer(4,17,33,34,40,42,47,50,52). An inverse association between physical activity and prostate cancer was found in six studies(4,17,33,47,50,52), while three studies (34,40,42) revealed a positive association. No data exist on the relationship between cardiorespiratory fitness and prostate cancer. Fitness level is likely to represent the end result of physical activity and may be considered a good indicator of long-term physical activity participation(6). No gold standard exists for the measurement of physical activity; however, cardiorespiratory fitness can be measured validly using a maximal exercise treadmill test to provide estimates of maximal oxygen uptake (10,13).
The purpose of this paper is to report our prospective observations of the association between cardiorespiratory fitness and incident prostate cancer.
The total population at the Cooper Clinic is composed of 28,072 subjects. The population for analysis was composed of 12,975 primarily white (>99%), married, college-educated men, aged 20-80 yr, with no history of prostate cancer, who received a voluntary, preventive medical examination at the Cooper Clinic in Dallas, TX, during 1971-1989 and who provided information on cardiorespiratory fitness and prostate cancer. Fifty percent of the men in this cohort were over the age of 42 yr, with 25% over the age of 50. This baseline examination included a physical examination: a questionnaire on demographic characteristics, physical activity, and health habits; a personal and family health history; anthropometric measures; resting and exercise electrocardiography; blood chemistry tests; blood pressure measurements; and a maximal exercise treadmill test to determine cardiorespiratory fitness. Written informed consent was obtained from the participants. Examination methods and procedures followed a standard manual of operations and have been previously described (10,11,13).
Cardiorespiratory fitness was assessed at the baseline examination with a maximal exercise treadmill test using the Balke protocol(7). The treadmill speed was 88 m·min-1 for the first 25 min. During this time the grade was 0% for the first minute, 2% the second minute, and increased 1% each minute until 25 min had elapsed. After 25 min, the grade remained constant while the speed increased 5.4 m·min-1 each minute until test termination(10). The outcome of interest was total treadmill time in minutes. The duration of the maximal exercise treadmill test is highly correlated with measured maximal oxygen uptake in men (r = 0.92), an accepted measure of cardiorespiratory fitness (43). Fitness levels were created by categorizing total treadmill time in minutes by quartiles.
Prostate Cancer Measurement
Questionnaires were mailed to the men in 1982 and 1990 for the purposes of periodic health status updates and monitoring. A series of case finding questions were included on each questionnaire. Cases of incident prostate cancer were ascertained by asking each individual whether he had ever been diagnosed with prostate cancer. If the answer was affirmative, the year of diagnosis was also obtained. Fatal cases of prostate cancer were ascertained by mortality surveillance using Social Security Administration files, Departments of Motor Vehicles, nationwide credit bureau network, National Death Index, and local area telephone directories. Further details of this mortality surveillance have been described elsewhere(10,13). Vital status through 1989 is 93% complete. However, only six prostate cancer cases were fatal due to the limited duration of follow-up. Therefore, all analyses included only the incident prostate cancer cases. All participants with existing prostate cancer at their first visit were eliminated. The remaining participants were followed from their baseline examination to the outcome of interest, newly diagnosed prostatic cancer, death from other causes, loss to follow-up due to administrative difficulties or operational limitations, or until the end of the study period. Vigorous attempts to reach all participants could not be implemented before the undertaking of the present analysis; therefore, we did not have the ability to separate true nonresponders from those who never received the mail survey and thus did not have the opportunity to respond. Information was obtained on cancer occurrence from 47% (N = 12,975) of the participants.
Physical Activity Measurement
Physical activity was assessed in a subsample of the total Cooper Clinic population (N = 28,072) at baseline and in the 1982 mail survey by inquiring about participation in various sports and activities. The subsample available for analysis consisted of men who provided information on physical activity, prostate cancer, age, body mass index, and smoking status(N = 7570). A self-administered questionnaire, which included a checklist of aerobic activities (Appendix) was provided to the subject during the baseline examination and in the 1982 mail survey. Each person was asked “For the last 3 months which of the following activities have you performed regularly?” Individuals provided information about participation, frequency per week, the intensity level, and duration per workout for the sports and activities on the checklist. This information was then used to create a physical activity index to estimate the total energy in kilocalories (kcal) expended in sport or activity each week, by multiplying the MET score (3,5) by the duration of the workout, body weight in kilograms, and frequency per week engaged in the sport or activity. Total weekly energy expenditure was calculated by summing kcal·wk-1 expended for each sport or activity. Physical activity levels were created by categorizing total weekly energy expenditure as follows: <1000, 1000-<2000, 2000-<3000, and ≥3000 kcal·wk-1. The cutpoints were defined a priori based upon published work (33).
Descriptive analyses were performed to characterize the study population. Subsequently, the association between cardiorespiratory fitness and incident prostate cancer was assessed using proportional hazards regression(18,19). This technique allows estimation of the incidence rate ratio while controlling for available potential confounding variables. Baseline values of age, body mass index, and smoking status were potential confounding variables adjusted in the multivariate analyses. The Statistical Analysis System (SAS) was used for all analyses(46).
The duration of the maximal exercise treadmill test in minutes was used to define quartiles of baseline cardiorespiratory fitness, <13.7, 13.7-<17.0, 17.0-<21.0, and ≥21.0 min on treadmill test. These categories were entered into the multivariate model as indicator variables, with the lowest level representing the referent category. Person-time was assigned to the appropriate fitness categories according to the baseline fitness level.
We also evaluated physical activity as a risk factor for incident prostate cancer, although information on physical activity was available for only a subsample of the study cohort, men who provided fitness and prostate cancer information. The two physical activity measures (assessed at baseline and 1982 survey) were used to create an average physical activity variable with four levels of energy expenditure (N = 12,098). If an individual only provided one measure of physical activity at the initial visit or in the 1982 survey, then the single measure was used as the average measure. For those participants whose single 1982 physical activity measure was used as the average measure, we eliminated them if there was any prior evidence of prostate cancer. Follow-up information on prostate cancer was available for 63% (N = 7,570) of the participants who had at least one measure of physical activity.
To assess confounding, we included each potential confounder in the model and compared its effect on the incidence rate ratio with that of the model without the potential confounder. The final model was selected based on the change in the estimate of effect produced by including the potential confounder in the model and the a priori biological importance of the potential confounder (45). A 10% change was the minimum criterion deemed necessary for confounding variable inclusion. The potential confounders were categorized and entered into the model as indicator variables. We examined age categorized in groups: 60-64, 65-69, and 70-80 yr old. Because prostate cancer rarely occurs in younger men, the age group composed of 20- to 59-yr-olds was the referent category. The effect of age was also examined by stratifying young (<60 yr old) versus old (≥60 yr old) and assessing the association between fitness and cancer. Body mass index was dichotomized at the median, and we compared those above the median to those below the median. To control for smoking status, we classified participants as current, past, or never smokers based upon responses to questions asked at the baseline examination. The category composed of never smokers was the referent group.
Follow-up studies of clinical diseases with potentially long latency periods must account for the frequently uncertain temporality between the exposure and the outcome of interest. Latent or subclinical disease could be responsible for a level of the hypothesized exposure. To address the concern of bias that could be due to latent disease, we analyzed the data with 1, 3, and 5 yr of person-time follow-up removed after the time of the initial questionnaire.
The results of all the regression analyses are presented in the form of mutually adjusted incidence rate ratios and 95% confidence intervals.
Demographic and anthropometric data of the study population (N = 12,975) by fitness level are detailed in Table 1. Age, weight, and body mass index were lower in men with higher levels of fitness. Men with higher levels of fitness were more likely to have never smoked compared with those men with a low level of fitness, who were more likely to be current smokers.
There were 14,849 participants lost to follow-up from the initial cohort of 28,072. We examined loss to follow-up by levels of fitness and found virtually no difference between the groups. Subjects lost to follow-up had similar baseline characteristics compared with subjects with follow-up information(Table 2). Of the subjects successfully followed, 24% included in the analysis for fitness were followed for the entire study period and there were 4,719 (36%) followed from baseline through 1982.
Because the results comparing the estimates among analyses with 1, 3, or 5 yr of person-time removed were similar, we excluded only 1 yr of person-time from the final analyses to maximize the amount of person-time available. Exclusion of first year cases is done to avoid confusion by subtle changes of exposure variables in response to incipient disease(33). In addition to the baseline exclusions, subjects with less than 1 yr of follow-up were excluded from the analyses (N= 248).
The data in Table 3 show results on cardiorespiratory fitness. Higher cardiorespiratory fitness levels were inversely associated with the probability of development of prostate cancer. Relative to the least fit group, crude incidence rate ratios changed from 0.76 to 0.52 to 0.18 across levels of fitness (P for trend = 0.0001). The adjusted point estimates changed from about 1 in the first two categories to 0.73 and 0.26 in the categories of the more fit individuals (P for trend = 0.0036). When the data were stratified by young (<60 yr old) versus old (≥60 yr old), the protective effect of fitness on prostate cancer was limited to the younger group. The adjusted estimates ranged from 0.75 to 0.11 for increasing levels of fitness in the group <60 yr of age. These results should be interpreted with caution due to the small number of cases in the older group,N = 33. There were no significant interactions for body mass index or smoking status.
Incidence rate ratios for prostate cancer and activity level in the subcohort with information on physical activity and prostate cancer are presented in Table 4. Crude and adjusted estimates show an inverse association between physical activity and prostate cancer, comparing increased levels of physical activity to the referent. The data do not support a linear trend; however, the results are based on a small number of cases that may lead to imprecise estimates. The results inTable 4 were not altered when the cutpoints for the physical activity levels were changed to <1000, 1000-<2500, 2500-<3000, and ≥3000 kcal·wk-1.
The validity of the physical activity measure and the consistency of fitness and activity levels over time were assessed by comparing total weekly energy expenditure in sports or activities to the results of the maximal treadmill exercise test. Participants who have high levels of total energy expenditure are expected to have better fitness. A significant correlation between self-reported physical activity levels and maximal exercise treadmill time was observed. The correlation coefficient was r = 0.41 for baseline physical activity level and maximal exercise treadmill time and for the correlation between 1982 activity level and treadmill time, r = 0.32. The correlation between baseline physical activity and 1982 physical activity levels was r = 0.35. In other analyses we found that the incidence of prostate cancer was similar in individuals who did not provide physical activity information as in those who did. We also established that the basic assumption of the proportional hazards model (e.g., hazards are proportional over time) was appropriate for our data.
These results provide support for the hypothesis that cardiorespiratory fitness is inversely associated with risk of incident prostate cancer. The upper two quartiles of fitness levels were associated with reduced risks of prostate cancer compared with the lowest quartile of fitness. An age interaction was evident with the protective effect limited to the young age group. Energy expenditure ≥1000 kcal·wk-1, was consistently associated with reduced risks of prostate cancer compared with energy expenditure levels less than 1000 kcal·wk-1. Adjusting for available potential confounders did not substantially alter this association. The point estimates for the 2000-<3000 kcal·wk-1 category were higher than the adjusted estimates for the other activity categories. This is likely due to chance because of the small number of cases.
This is the first study reporting on fitness level and risk of prostate cancer. Fitness is clearly correlated with habitual physical activity and represents a characteristic that may be relevant to the occurrence of chronic disease, compared with short-term physical activity that may vary depending on transient motivation and opportunity or inaccurate reporting. These data on fitness concern the totality of the population enrolled in the study, and although 53% of the subjects eventually did not provide information on prostate cancer outcome, there was no evidence in these data to suggest that nonresponse was jointly associated with fitness and prostate cancer(a precondition for bias). This is further supported by the similarity of the proportion of subjects with prostate cancer information in those who have only fitness data and in those with both fitness and physical activity data and by the similar incidence of prostate cancer among those with and those without physical activity information. Furthermore, the data on physical activity are essentially compatible with the fitness results and they appear to be in agreement with the weight of the existing evidence.
Observational studies of this nature have the potential for bias. The maximal exercise treadmill test, used to assess fitness, is an accepted and validated measure of cardiorespiratory fitness (43). Any misclassification of this exposure would be random since prostate cancer outcome was not known at the time of treadmill testing. Our results are not subject to recall bias because of the prospective nature of the study.
Cases of prostate cancer were identified using self-reported data. No validation of self-reported prostate cancer has been accomplished; however, self-report of other diseases such as myocardial infarction, stroke, and hypertension were validated in this population (13). Self-reported hypertension had a sensitivity of 98% and a specificity of 99% in a study conducted in this population by Blair et al.(13). In a study conducted by Giovannucci et al.(23), in a different population, self-reported prostate cancer was validated by review of medical and pathology records and a diagnosis of adenocarcinoma was confirmed in 99.4% of the men.
There is potential for bias in this study if there was differential diagnosis of prostate cancer between fitness levels. However, subjects who are fit are more health conscious. There is a body literature indicating that physically active subjects are more likely to have contact with physicians and the health care system than subjects who are not physically active (8,9,32,35,48). This difference would create an increased likelihood of diagnosis of prostate cancer in those men with increased levels of fitness and thus would reduce the strength of an inverse association between fitness and prostate cancer. Therefore, if diagnostic bias occurred in this study, the observed estimates are likely to be an underestimate of the true strength of the association between fitness and prostate cancer. It would be important to have information on the stage of diagnosis of the tumor to predict what implications fitness may have on the future progression of the tumor, but such data are not available.
Loss to follow-up was defined as a nonresponse to both the 1982 and 1990 questionnaires. Loss to follow-up could have generated biased results only if nonresponse was correlated with both fitness level and prostate cancer. However, a substantial amount of bias would need to exist to explain the strong inverse associations observed in these data. We examined loss to follow-up by levels of fitness and found virtually no difference between the groups. Subjects lost to follow-up had similar baseline characteristics compared with subjects with follow-up information. We believe the high percentage of loss to follow-up was due in part to the mobility of the population, and our inability to maintain current addresses on some of the men. Thus, many nonresponders did not have the opportunity to answer the mail survey because they did not receive it.
We were unable to control for the potential confounding effects of diet in our analyses. A positive association between fat intake and prostate cancer has been observed in some case-control and cohort studies(39,41). For diet to be confounding the observed association between fitness and prostate cancer, active men would have to consume a diet composed of less fat. There are no data available on the association between diet and fitness levels, and no association has been observed between physical activity level and dietary composition(9,12,15,36). The results of Lee et al. (33) showed increased caloric intake in highly active men compared with those who were less active, but the proportion of calories from fat consumed was the same across inactive, moderately active, and highly active groups.
Also, we were unable to control for some other potential risk factors: venereal disease, sexual habits, and occupational exposures. The results of previous studies examining these factors have been inconsistent(21,39,53). It is unlikely that one or more of these factors is strongly associated with both fitness and prostate cancer incidence, and could cause substantial confounding.
Varying amounts of person-time after completion of the baseline questionnaire were eliminated to assess the effect of latent disease, which may have affected fitness levels in diseased individuals. If participants with existing disease had altered fitness results, then the association between fitness and prostate cancer could be spurious. Eliminating 1, 3, or 5 yr of person-time follow-up, however, did not substantially alter the results.
The mechanism behind the protective effect of fitness on prostate cancer may be related to the sex hormone testosterone. The prostate is a secondary sex gland that is affected by hormonal stimulation. The growth and development of the organ relies on the presence of sex hormones. Testosterone was singled out as a possible cause of prostate cancer as early as 1941 when androgen deprivation was used as therapy in advanced prostatic cancer(30). Currently, metastatic prostate cancer is treated with antitestosterone therapy or adrenal androgen blockage(24). Animal studies have shown that prostate cancer can be induced in rats by the administration of exogenous testosterone(16,29,38). In studies involving humans, prostate cancer cases appear to have higher levels of endogenous testosterone compared with nondiseased men (2,22,31). This evidence seems to implicate testosterone in the development of prostate cancer. Furthermore, athletes have been shown to have lower levels of testosterone, and individuals who exercise may have a temporary decrease in post-exercise levels of testosterone(1,20,26,27,37,49,51). If fit individuals have lower levels of endogenous testosterone and lower levels of testosterone are associated with decreased incidence of prostate cancer, it should be expected that men who are more physically fit or active would be at a lower risk of developing prostate cancer compared with men who are inactive.
It should be noted highly trained athletes have been observed to have lower basal circulating testosterone, whereas men who exercise experience a more acute and temporary drop in testosterone levels. It would seem that a continued decrease of testosterone levels would be more important in influencing a decreased prostate cancer risk compared with a transient change due to an exercise session. We observed a protective effect on prostate cancer across all levels of increased fitness and physical activity.
There are no previously reported data concerning cardiorespiratory fitness and prostate cancer. Epidemiological studies on the association between prostate cancer and physical activity are not entirely consistent. Increased exercise in the form of occupational activity, recreational exercise, or household work has been observed to be protective for prostate cancer in six studies(4,17,33,47,50,52). Conversely, three studies have shown an increased risk of prostate cancer with increasing levels of physical activity(34,40,42). In two of the studies, participation in college athletics was the exposure of interest(40,42). In the third study, an inverse association was found between risk of prostate cancer and the proportion of life spent in occupations involving only sedentary or light work(34). It may be that college athletics participation is too remote in time to be etiologically relevant for prostate cancer or may not be related to lifetime fitness and that the proportion of time spent in sedentary occupation is a poor indicator of overall fitness.
These results suggest that moderate to high levels of cardiorespiratory fitness may protect against the incidence of prostate cancer. This study supports the physical activity-prostate cancer hypothesis and provides evidence that cardiorespiratory fitness may be a better predictor of prostate cancer risk than physical activity.
1. Aakvaag, A., T. Sand, P. K. Opstad, and F. Fonnum. Hormonal changes in serum in young men during prolonged physical strain.Eur. J. Appl. Physiol.
2. Ahluwalia, B., M. A. Jackson, G. W. Jones, A. O. Williams, M. S. Rao, and S. Rajguru. Blood hormone profiles in prostate cancer patients in high-risk and low-risk populations. Cancer
3. Ainsworth, B. E., D. S. Jacobs, Jr., A. S. Leon, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med. Sci. Sports Exerc.
4. Albanes, D., A. Blair, and P. R. Taylor. Physical activity and risk of cancer in the NHANES I population. Am. J. Public Health
5. American College of Sports Medicine. Guidelines for Exercise Testing and Prescription
. Philadelphia: Lea and Febiger, 1986, pp. 158-161.
6. American College of Sports Medicine. The recommended quantity and quality of exercise for developing and maintaining cardiorespiratory and muscular fitness in healthy adults. Med. Sci. Sports Exerc.
7. Balke, B. and R. W. Ware. An experimental study of physical fitness in Air Force personnel. U.S. Armed Forces Med. J.
8. Blair, S. N. and H. W. Kohl. Measurement and evaluation of health behaviors and attitudes in relationship to physical fitness and physical activity patterns. In: Assessment of Physical Fitness and Physical Activity in Population Based Surveys
, T. Drury (Ed.). Washington, DC: U.S. Government Printing Office, pp. 527-545.
9. Blair, S. N., D. R. Jacobs, Jr., and K. E. Powell. Relationships between exercise or physical activity and other health behaviors. Public Health Rep.
10. Blair, S. N., H. W. Kohl, III, R. S. Paffenbarger, Jr., D. G. Clark, K. H. Cooper, and L. W. Gibbons. Physical fitness and all-cause mortality: a prospective study
of healthy men and women. J.A.M.A.
11. Blair, S. N., K. H. Cooper, L. W. Gibbons, L. R. Gettman, S. Lewis, and N. Goodyear. Changes in coronary heart disease risk factors associated with increased treadmill time in 753 men. Am. J. Epidemiol.
12. Blair, S. N., N. M. Ellsworth, W. L. Haskell, M. P. Stern, J. W. Farquhar, and P. D. Wood. Comparison of nutrient intake in middle-aged men and women runners and controls. Med. Sci. Sports Exerc.
13. Blair, S. N., N. N. Goodyear, L. W. Gibbons, and K. H. Cooper. Physical fitness and incidence of hypertension in healthy normotensive men and women. J.A.M.A.
14. Boring, C. C., T. S. Squires, and T. Tong. Cancer statistics, 1992. Cancer
15. Bouchard, C., R. J. Shephard, T. Stephens, J. R. Sutton, and B. D. McPherson (Eds.). Proceedings of the International Conference on Exercise, Fitness, and Health
. Champaign, IL: Human Kinetics, 1990.
16. Brown, C. E., S. Warren, R. N. Chute, K. J. Ryan, and R. B. Todd. Hormonally induced tumors of the reproductive system of parabiosed male rats. Cancer Res.
17. Brownson, R. C., J. C. Chang, J. R. Davis, and C. A. Smith. Physical activity on the job and cancer in Missouri. Am. J. Public Health
18. Cox, D. R. Partial likelihood. Biometrika
19. Cox, D. R. Regression models and life-tables. J. R. Stat. Soc. [Ser A]
20. Dessypris, A., K. Kuoppasalmi, and H. Adlercreutz. Plasma cortisol, testosterone, androstenedione and luteinizing hormone (LH) in a non-competitive marathon run. J. Steroid Biochem.
21. Devesa, S. S., D. T. Silverman, J. L. Young Jr., et al. Cancer incidence and mortality trends among whites in the United States, 1947-84. J. Natl. Cancer Inst.
22. Ghanadian, R., C. M. Puah, and E. P. N. O'Donoghue. Serum testosterone and dihydrotestosterone in carcinoma of the prostate.Br. J. Cancer
23. Giovannucci, E., A. Ascherio, E. B. Rimm, G. A. Colditz, M. J. Stampfer, and W. C. Willett. A prospective cohort study of vasectomy and prostate cancer in US men. J.A.M.A.
24. Gittes, R. F. Carcinoma of the prostate. N. Engl. J. Med.
25. Habib, F. K., I. R. Lee, S. R. Stitch, and P. H. Smith. Androgen levels in the plasma and prostatic tissues of patients with benign hypertrophy and carcinoma of the prostate. J. Endocrinol.
26. Hackney, A. C., W. E. Sinning, and B. C. Bruot. Reproductive hormonal profiles of endurance-trained and untrained males.Med. Sci. Sports Exerc.
27. Hackney, C., W. E. Sinning, and B. C. Bruot. Hypothalamicpituitary-testicular axis function in endurance-trained males.Int. J. Sports Med.
28. Hammond, G. L. Endogenous steroid levels in the human prostate from birth to old age: a comparison of normal and diseased tissues.J. Endocrinol.
29. Hovenanian, M. S. and C. L. Deming. The heterologous growth of cancer of the human prostate. Surg. Gynecol. Obstet.
30. Huggins, C. and C. V. Hodges. Studies on prostatic cancer. I. The effect of castration, of estrogen and of androgen injection on serum phosphatases in metastatic carcinoma of the prostate. Cancer Res.
31. Jackson, M. A., J. Kovi, M. Y. Heshmat, et al. Characterization of prostatic carcinoma among blacks: a comparison between a low-incidence area, Ibadan, Nigeria, and a high-incidence area, Washington, DC. Prostate
32. Langlie, J. K. Interrelationships among preventive health behaviors: a test of competing hypotheses. Public Health Rep.
33. Lee, I., R. S. Paffenbarger, Jr., and C. Hsieh. Physical activity and risk of prostatic cancer among college alumni.Am. J. Epidemiol.
34. Marchand, L. L., L. N. Kolonel, and C. N. Yoshizawa. Lifetime occupational physical activity and prostate cancer risk. Am. J. Epidemiol.
35. Mechanic, D. The stability of health and illness behavior: results from a 16-year follow-up. Am. J. Public Health
36. Moore, C. E., G. H. Hartung, R. E. Mitchell, C. M. Kappus, and J. Hinderlitter. The relationship of exercise and diet on high-density lipoprotein cholesterol levels in women. Metabolism
37. Morville, R., P. C. Pesquies, C. Y. Guezennec, B. D. Serrurier, and M. Guignard. Plasma variations in testicular and adrenal androgens during prolonged physical exercise in man. Ann. Endocrinol.
38. Noble, R. L. The development of prostatic adenocarcinoma in Nb rats following prolonged sex hormone administration.Cancer Res.
39. Nomura, A. M. and L. N. Kolonel. Prostate cancer: a current prospective. Epidemiol. Rev.
40. Paffenbarger, Jr., R. S., R. T. Hyde, and A. L Wing. Physical activity and incidence of cancer in diverse populations: a preliminary report. Am. J. Clin. Nutr.
41. Pienta, K. J. and P. S. Esper. Risk factors for prostate cancer. Ann. Int. Med.
42. Polednak, A. P. College athletics, body size, and cancer mortality. Cancer
43. Pollock, M. L., R. L. Bohannon, K. H. Cooper, et al. A comparative analysis of four protocols for maximal treadmill stress testing.Am. Heart J.
44. Ross, R. K., A. Paganini-Hill, and B. E. Henderson. Epidemiology of prostatic cancer. In: Diagnosis and Management of Genitourinary Cancer
, D. G. Skinner and G. Lieskovsky (Eds.). Philadelphia: W. B. Saunders, 1988, pp. 40-45.
45. Rothman, K. J. Modern Epidemiology
. Boston: Little, Brown and Co., 1986, pp. 178-220.
46. SAS Institute. SAS/STAT Software: The PHREG Procedure, Version 6
. Cary, NC: SAS Institute Inc., 1991, pp. 1-54.
47. Severson, R. K., A. M. Y. Nomura, J. S. Grove, and G. N. Stemmermann. A prospective analysis of physical activity and cancer.Am. J. Epidemiol.
48. Stephens, T. Health practices and health status: evidence from the Canada Health Survey. Am. J. Prev. Med.
49. Strauss, R. H., R. R. Lanese, and W. B. Malarkey. Weight loss in amateur wrestlers and its effect on serum testosterone levels.J.A.M.A.
50. Vena, J. E., S. Graham, M. Zielezny, J. Brasure, and M. K. Swanson. Occupational exercise and risk of cancer. Am. J. Clin. Nutr.
51. Wheeler, G. D., S. R. Wall, A. N. Belcastro, and D. C. Cumming. Reduced serum testosterone and prolactin levels in male distance runners. J.A.M.A.
52. Yu, H., R. E. Harris, and E. L. Wynder. Case-control study of prostate cancer and socioeconomic factors. Prostate
53. Zaridze, D. G. and P. Boyle. Cancer of the prostate: epidemiology and aetiology. Br. J. Urol.