Case-Control Study of Lifetime Total Physical Activity and Prostate Cancer Risk

Transcription

1 American Journal of Epidemiology Copyright 2004 by the Johns Hopkins Bloomberg School of Public Health All rights reserved Vol. 159, No. 8 Printed in U.S.A. DOI: /aje/kwh106 Case-Control Study of Lifetime Total Physical Activity and Prostate Cancer Risk C. M. Friedenreich 1, S. E. McGregor 1, K. S. Courneya 2, S. J. Angyalfi 3, and F. G. Elliott 4 1 Division of Population Health and Information, Alberta Cancer Board, Calgary, Alberta, Canada. 2 Faculty of Physical Education, University of Alberta, Edmonton, Alberta, Canada. 3 Tom Baker Cancer Centre, Calgary, Alberta, Canada. 4 Calgary Health Region, Calgary, Alberta, Canada. Received for publication July 16, 2003; accepted for publication November 13, A population-based case-control study of physical activity and prostate cancer risk was conducted in Alberta, Canada, between 1997 and A total of 988 incident, histologically confirmed cases of stage T2 or greater prostate cancer were frequency matched to 1,063 population controls. The Lifetime Total Physical Activity Questionnaire was used to measure occupational, household, and recreational activity levels from childhood until diagnosis. Multivariable logistic regression analyses were conducted. No association for total lifetime physical activity and prostate cancer risk was found (odds ratio (OR) for 203 vs. <115 metabolic equivalent-hours/week/ year = 0.87, 95% confidence interval (CI): 0.65, 1.17). By type of activity, the risks were decreased for occupational (OR = 0.90, 95% CI: 0.66, 1.22) and recreational (OR = 0.80, 95% CI: 0.61, 1.05) activity but were increased for household (OR = 1.36, 95% CI: 1.05, 1.76) activity when comparing the highest and lowest quartiles. For activity performed at different age periods throughout life, activity done during the first 18 years of life (OR = 0.78, 95% CI: 0.59, 1.04) decreased risk. When activity was examined by intensity of activity (i.e., low, <3; moderate, 3 6; and vigorous, >6 metabolic equivalents), vigorous activity decreased prostate cancer risk (OR = 0.70, 95% CI: 0.54, 0.92). This study provides inconsistent evidence for the association between physical activity and prostate cancer risk. case-control studies; motor activity; prostatic neoplasms; risk factors Abbreviations: CI, confidence interval; MET, metabolic equivalent; NPHS, National Population Health Survey; OR, odds ratio. Physical activity may have an inverse association with prostate cancer risk; however, the epidemiologic evidence is currently inconsistent and the magnitude of the risk reduction observed is small (1). Among 30 previously conducted studies, 14 studies (2 15) suggested an inverse association of physical activity with prostate cancer; however, no overall association was found in 12 studies (16 27), and an increased risk of prostate cancer was observed among the most physically active men in four other studies (28 31). These inconsistent results could be attributable, in part, to methodological issues in these studies, including variations in latent disease detection and possible outcome misclassification, crude assessments of physical activity, inadequate control for confounding, and incomplete examination of effect modification. This study was conducted to address gaps in previous studies. Specifically, all types and parameters of physical activity throughout lifetimes were assessed, and a full examination of confounding factors and potential effect modifiers was undertaken. MATERIALS AND METHODS We conducted a population-based, case-control study in Alberta, Canada, between November 1997 and December This study received ethical approval, and all study subjects provided written informed consent before the interview. Incident, histologically confirmed, invasive cases of stage T2 or greater prostate cancer were identified directly from the Alberta Cancer Registry, a population-based cancer registry that has an estimated 95 percent case ascertainment rate (32). Cases were eligible for the study if they were Alberta residents, aged less than 80 years, English speaking, Reprint requests to Dr. C. M. Friedenreich, Division of Population Health and Information, Alberta Cancer Board, Calgary, Alberta T2N 4N2, Canada ( chrisf@cancerboard.ab.ca). 740

2 Physical Activity and Prostate Cancer 741 and able to complete an interview and if they did not have another previous cancer except nonmelanoma skin cancer. Cases were restricted to stage T2 or greater to eliminate clinically unapparent cases, since controls were not screened for prostate cancer. The referring urologist was sent a staging form to complete for all cases identified as potentially eligible. If the urologist staged the case as T2 or greater, the cases were eligible for the study. Permission to contact patients was sought through their referring physician. Once permission was obtained, in-person interviews were conducted. A total of 1,965 cases were identified as potentially eligible; of these men, 626 could not be contacted for an interview because of physician refusal (n = 24), language barrier (n = 25), moved or telephone disconnected (n = 42), more than 6 months from diagnosis to case ascertainment for the study (n = 511), or deceased (n = 24). Of the remaining 1,339 eligible men, 1,014 (75.7 percent) completed an inperson interview, 265 (19.8 percent) refused, two had an incomplete interview, 34 (2.5 percent) could not be contacted, and 24 were unsuitable for an interview (1.8 percent). Male controls were identified through random digit dialing using a pool of telephone numbers randomly generated from available prefixes for the province of Alberta. Controls were frequency matched to cases on age (±5 years) and place of residence (urban/rural) and were free of any cancer diagnosis excluding nonmelanoma skin cancer. The telephone recruiters identified 2,014 men as potentially eligible and 1,276 men (63.3 percent) agreed to receive a study package, but 49 men (3.8 percent) could not be interviewed for the following reasons: moved or telephone disconnected (n = 8), not eligible (n = 5), previous cancer diagnosis (n = 5), language barrier (n = 1), deceased (n = 1), too ill (n = 1) or unsuitable to interview (n = 5), unable to contact (n = 5), or not available during the study period (n = 18). Interviews were completed with 1,082 of the 1,227 (88.2 percent) of the eligible and available men who received the study package. The overall response rate for the controls was 53.7 percent (1,082/2,014). The final data set for analysis included 988 cases and 1,063 controls, since 26 cases and 19 controls had to be removed because their dietary data had too many missing values (n = 29), the interviews were rated as unsatisfactory or questionable by the interviewers (n = 10), or the cases were ineligible because the cancer stage was less than T2 (n = 6). Data collection Respondents reported their personal health history, prostate cancer-screening history, prostate conditions, history of surgery, family history of cancer, lifetime physical activity patterns, dietary intake during the reference year, lifetime alcohol consumption history, smoking habits, demographic characteristics, and usual adult height and weight at each decade of age from 20 to 60 years. All data were up to the date of diagnosis for the cases, and a comparable date for the controls was determined by frequency matching the controls to the cases on time since diagnosis. Diet during the reference year was assessed with the National Cancer Institute s Block food frequency questionnaire (33). The average time interval between diagnosis and interview was 137 days (range, days). One recall calendar focusing on major life events, including residence, education, occupation, and physical activities performed throughout the lifetime, was mailed to the participants before the interview. This calendar, designed specifically for this study, was pilot tested with the questionnaire and used as a means to improve respondent recall of lifetime physical activity. The interviewers used cognitive interviewing methods (34) to assist the respondents in answering the questions. Several quality control measures of the interviewers methods were incorporated into the study design, for example, regular meetings and monitoring of the interviewers methods. Lifetime physical activity was estimated using a questionnaire that had been tested for reliability (35), with a reliability correlation of 0.74 for total lifetime activity, and that had been used in a case-control study of breast cancer (36). This questionnaire assessed occupational, household, and recreational activities separately throughout a respondent s lifetime (from childhood to the reference date, i.e., the diagnosis date for cases and a comparable date for controls), using a table format rather than specific questions. The patterns of physical activity are recorded by the interviewer, including the age started, aged ended, number of months per year, weeks per month, days per week, and hours per day that each activity was performed, so that the frequency and duration of these activities can be determined. The participants also reported the intensity of their activity as sedentary (used for occupational activities only), light, moderate, or heavy (i.e., self-reported intensity). For occupational activity, we obtained the job title as well as up to three descriptors of the actual activity performed at paid or volunteer occupational activities (e.g., standing, sitting, walking). Definitions for each intensity level by type of activity were provided with examples for the study participants. In addition to self-reported intensities, a specific metabolic equivalent (MET) value based on the description of the activity was assigned to each reported activity. The MET values used were abstracted from the Compendium of Physical Activities (37). A MET value is defined as the ratio of the associated metabolic rate for a specific activity to the resting metabolic rate (38). One MET is the average seated resting energy cost of an adult and is set at 3.5 ml of oxygen per kg per minute. The variables estimated in this analysis were the average MET-hours per week per year spent in occupational, household, and recreational activities over the respondent s lifetime (36). Total lifetime physical activity was estimated as the sum of occupational, household, and recreational activities. At the end of the interview, interviewers used standardized methods to measure the study participants current height, weight, and waist and hip circumferences. Statistical analysis Descriptive statistics were performed to characterize the study population and to examine case-control differences. Dietary, alcohol, and physical activity variables were categorized into quartiles according to the distribution of the variables among the controls. Odds ratios for prostate cancer

3 742 Friedenreich et al. were estimated using unconditional logistic regression, and a full assessment of confounding and effect modification was done. Univariate modeling was initially used to identify potential risk factors for prostate cancer in these data. Variables considered as confounders were age, region of residence (urban/rural), education, marital status, ethnic or cultural group, frequency of regular check-ups, number of digital rectal examinations and prostate-specific antigen tests, ever diagnosed with heart disease, hypertension, other chronic diseases, first degree family history of prostate cancer, first degree family history of other cancers, ever cigarette smoking, ever smoking both cigarettes and pipes, total pack-years of smoking, ever alcohol consumption and total lifetime alcohol intake, ever had a vasectomy, ever diagnosed with benign prostatic hyperplasia, body mass index (weight (kg)/height (m) 2 ), waist/hip ratio, weight gain from age 20 years to the reference year, total caloric intake, daily dietary fat, and carbohydrate, protein, and lycopene intakes. Only those variables that were confounders in this data set or for which there was a biologic rationale were included in the final model. Determination of confounding was done by assessing both the association between the confounder and outcome and a change in risk effect. Final models were adjusted for age, region, family history of prostate cancer, total caloric intake, body mass index, waist/hip ratio, total lifetime alcohol intake, education, and frequency of digital rectal examinations and prostate-specific antigen tests (excluding the last test performed in cases during the year prior to the interview that may have been a diagnostic tool and not a screening test). Models of each type of physical activity were adjusted for the other types of activity. Possible effect modification was assessed in this study by stratifying all models by age (<65 years and 65 years), family history of prostate cancer, body mass index (<25, 25 30, >30), and alcohol intake (ever vs. never). Tests for linear trend were performed for all models of categorized data by including the continuous, rather than the categorized, variable for each of the variables being modeled, and all p values reported are two sided. The association between physical activity and prostate cancer risk was examined in three sets of analyses. The first considered how risk was related to the type of physical activity, the second to the time period in life, and the third to the dose (specifically, the intensity, duration, and frequency) of activity. RESULTS The cases and controls had similar sociodemographic characteristics (table 1) with the exception that a higher percentage of controls had completed college or university education. Cases were clearly more likely to have ever visited a urologist and to have experienced a higher incidence of benign prostatic hyperplasia, prostatitis (data not shown), and other prostate conditions (data not shown). These benign conditions, while not risk factors for prostate cancer, may have increased the likelihood of screening for prostate cancer and subsequent diagnosis. Cases and controls had comparable medical histories with the exception of family histories of prostate cancer or other cancers that were more prevalent among cases than controls. Smoking and TABLE 1. Descriptive statistics for the study population (n = 2,051), Alberta, Canada, Risk factor * Numbers in parentheses, standard deviation. Cases (n = 988) Controls (n = 1,063) Mean age (years) 67.4 (7.4)* 65.5 (8.4) Currently married or common-law (%) Highest education (%) Less than high school High school diploma Trade certificate College University degree Ethnic group (%) British Western European Eastern European French Northern European Other Frequency of medical check-ups over lifetime (%) Yearly Every few years Occasionally Ever had benign prostatic hyperplasia or prostatic hypertrophy (%) Ever had vasectomy (%) Any immediate family had cancer (%) First degree family history of prostate cancer (%) Type of smoker (%) Occasional Former occasional smoker Former smoker Current smoker Mean total pack-years of smoking among smokers 25.2 (21.0) 25.4 (21.8) Ever drunk alcoholic beverages (%) Mean body mass index (weight (kg)/height (m) 2 ) 28.0 (3.9) 28.2 (4.2) Mean total lifetime physical activity (metabolic equivalent-hours/week/year) (71.1) (68.0) Mean total occupational activity (metabolic equivalent-hours/week/year) (68.6) (67.7) Mean total household activity (metabolic equivalent-hours/week/year) 21.4 (17.0) 19.2 (15.1) Mean total recreational activity (metabolic equivalent-hours/week/year) 17.6 (16.2) 19.5 (16.6) Mean total lifetime physical activity (hours/ week/year) 44.4 (10.9) 44.2 (10.4) alcohol consumption histories were also similar between the two groups as were their anthropometric characteristics. The average amount of total lifetime physical activity completed by the two groups was also very similar. We then examined the univariable associations of individual risk factors and prostate cancer risk (table 2). A

4 Physical Activity and Prostate Cancer 743 TABLE 2. Age-adjusted odds ratios for risk factors for prostate cancer (n = 2,051), Alberta, Canada, Risk factor Cases (no.) Controls (no.) OR* 95% CI* Ever married , 1.18 Highest education attained Less than high school High school diploma , 1.10 Trade certificate , 1.08 College , 1.02 University , 0.71 First degree family history of prostate cancer , 2.75 First degree family history of other cancer besides prostate , 1.52 Ever had benign prostatic hyperplasia , 2.08 Ever had digital rectal examination , 1.00 No. of times had digital rectal examination ( 2 vs. 0) 753 vs vs , 1.00 Ever had prostate-specific antigen test , 1.64 No. of times had prostate-specific antigen test ( 2 vs. 0) 246 vs vs , 1.99 Ever had a vasectomy , 1.09 Ever alcohol drinker , 2.22 Total lifetime alcohol drinking (g/year) Nondrinkers <1, , ,173 <3, , , 2.46 p trend 0.09 Ever cigarette smoker ,1.34 Total pack-years of smoking Nonsmokers < , < , , 1.44 p trend 0.78 Total caloric intake (kcal/day) <1, ,633 <1, , ,933 <2, , , , 1.84 p trend 0.01 * OR, odds ratio; CI, confidence interval. The last digital rectal examination and prostate-specific antigen test performed in the cases in the year prior to the interview date were excluded from the analysis to eliminate any that were actually performed for diagnostic rather than screening purposes. decreased risk of prostate cancer was found for higher education, with a clear trend in decreasing risk with increasing level of education (odds ratio (OR) = 0.55, 95 percent confidence interval (CI): 0.42, 0.71) found for university education versus no secondary school completion. Prostate cancer risk was also associated with ever having a benign prostatic hyperplasia diagnosis (OR = 1.73, 95 percent CI: 1.43, 2.08). A doubling in risk of prostate cancer was observed for a first-degree family history of prostate cancer (OR = 2.14, 95 percent CI: 1.66, 2.75), while any family history of cancer besides prostate also increased risk (OR = 1.28, 95 percent CI: 1.07, 1.52). Ever having consumed alcohol and any category of alcohol consumption above that of nondrinkers were associated with a statistically significant increased risk of 67 percent. Smoking increased prostate cancer risk only marginally. Statistically significant increased prostate cancer risks were also observed for the highest quartile of total caloric intake. Subsequently, we examined the influence of total lifetime physical activity and each type of activity (table 3). Overall, there was a 13 percent, nonstatistically significant, decreased risk of prostate cancer for those men who were the most physically active in their lifetimes. Occupational and recreational activities decreased risk while household activity increased risk. The greatest risk reduction was for recreational activity, for which the risk was 0.80 (95 percent CI:

5 744 Friedenreich et al. TABLE 3. Odds ratios for lifetime physical activity by type of activity (n = 2,051), Alberta, Canada, Type of physical activity quartile cutpoints (metabolic equivalent-hours/week/year) Lifetime total physical activity Cases (no.) (n = 988) * Adjusted for age, region (four urban vs. rural areas in Alberta), education (highest level of education achieved), body mass index, waist/hip ratio, total caloric intake, average lifetime total alcohol intake, first degree family history of prostate cancer, number of times had prostate-specific antigen test done, and number of digital rectal examinations. Models of each type of activity were adjusted for the other types of activity. OR, odds ratio; CI, confidence interval. 0.61, 1.05). An increased risk for household activity of 1.36 (95 percent CI: 1.05, 1.76) was found in the highest quartile of activity only, and a dose-response effect for increasing risk with increasing household activity was also observed. The only effect modification in these data was for men with a family history of prostate cancer. The risk in the highest quartile of lifetime activity was 0.48 (95 percent CI: 0.21, 1.11) (p trend = 0.03), while men without this family history had a risk of 1.00 (95 percent CI: 0.73, 1.36). We then examined for which time period(s) in life physical activity was associated with prostate cancer risk (table 4). A decreased risk of prostate cancer was observed for activity during childhood and adolescence (up to age 18 years) and from age 65 years or more. Early life activity decreased risk by 22 percent, although this decrease was not statistically significant. Activity later in life ( 65 years) decreased risk in the second and third quartiles but not in the upper quartile. No associations with activity performed at any level between ages 18 and 65 years were found. Furthermore, there was Controls (no.) (n = 1,063) Age adjusted Multivariable adjusted* OR 95% CI OR 95% CI 0 < < , , < , , , , 1.17 p trend Lifetime occupational activity 0 < < , , < , , , , 1.22 p trend Lifetime household activity 0 < < , , < , , , , 1.76 p trend Lifetime recreational activity 0 < < , , < , , , , 1.04 p trend little evidence for any dose-response effect with increasing levels of activity and prostate cancer risk. The question of how the dose of activity was associated with risk was examined in the final two analyses. The first examined risk by light, moderate, or vigorous levels of intensity (<3, 3 6, and >6 METs) (table 5). A decreased risk of prostate cancer was restricted to men who performed at least 4.83 hours per week of vigorous intensity activity. For this group, the risk was 0.70 (95 percent CI: 0.54, 0.92). Light intensity activities decreased risk somewhat, albeit not statistically significantly, and an increased risk was found for moderate intensity activities in the second and third quartiles. In the final analysis (table 6), only the frequency and duration of the activities were considered. A 27 percent statistically significant decreased risk of prostate cancer was found among men who had 40 or more hours of occupational activity per week. Prostate cancer risk increased for men who did 8.3 hours per week of household activity but appeared to decrease for men who had between 3 and 4.7

6 Physical Activity and Prostate Cancer 745 TABLE 4. Odds ratios for lifetime physical activity by age periods in life (n = 2,051), Alberta, Canada, hours per week of recreational activity. However, no trend of decreasing risk with increasing levels of recreational activity was found. DISCUSSION Age period (metabolic equivalent-hours/week/year) Cases (no.) (n = 988) Controls (no.) (n = 1,063) * Adjusted for age, region (four urban vs. rural areas in Alberta), education (highest level of education achieved), body mass index, waist/hip ratio, total caloric intake, average lifetime total alcohol intake, first degree family history of prostate cancer, number of times had prostate-specific antigen test done, and number of digital rectal examinations. OR, odds ratio; CI, confidence interval. Overall, no association between total lifetime physical activity and prostate cancer risk was observed. However, risk reductions were found for recreational activity, for activity done early and possibly also later in life, for vigorous intensity activity, and for occupational activity when only duration and frequency were examined. There was also an increased risk observed for household activity when considered by total volume of activity (including intensity) and when only frequency and duration were examined. One explanation for the risk reductions found for vigorous and recreational activity is that these are salient events more readily recalled than household activity or lower intensity activities (39). It is not clear why an increased risk was found for household activity. This increased risk may be Age adjusted Multivariable adjusted* OR 95% CI OR 95% CI 0 18 years 0 < < , , < , , , , 1.04 p trend years 0 < < , , < , , , , 1.30 p trend years 0 < < , , < , , , , 1.40 p trend years 0 < < , , < , , , , 1.36 p trend real or may be attributable to a higher degree of measurement error associated with this particular type of activity. This study s results can be compared with the 30 previously conducted studies (2 31). Of these 30 studies, three were early cohort studies that used prostate cancer mortality as an outcome rather than incidence and broad classifications of physical activity as either usual lifetime occupational activity (14, 15) or college athletic participation (31). The two occupational cohort studies found decreased prostate cancer mortality rates among men with the highest levels of occupational activity, while the college athlete cohort study found higher mortality rates among major athletes. The remaining 27 incident studies were divided into 16 cohort studies (2 8, 12, 16 22, 28) and 11 case-control studies (9 11, 13, 23 27, 29, 30). A wide range of methods for physical activity assessment was used in these investigations, and the definition of the highest category of activity in each study differed across studies. Of these 27 studies, 10 found a borderline or fully statistically significant decreased prostate cancer risk for men who were the most active compared with

7 746 Friedenreich et al. TABLE 5. Odds ratios for lifetime physical activity by intensity of activity (n = 2,051), Alberta, Canada, Hours/week/year performed at activity intensity level Cases (no.) (n = 988) * Adjusted for age, region (four urban vs. rural areas in Alberta), education (highest level of education achieved), body mass index, waist/hip ratio, total caloric intake, average lifetime total alcohol intake, first degree family history of prostate cancer, number of times had prostatespecific antigen test done, and number of digital rectal examinations. Models for each level of intensity of activity were adjusted for the other levels of intensity of activity. OR, odds ratio; CI, confidence interval. those least active (2 11); two found a nonsignificant decreased risk (12, 13); 12 observed no effect (16 27); and three found increased risks overall for prostate cancer among those who were the most physically active (28 30). When risk reductions are observed in these studies, the magnitude of the decrease is modest, ranging from 10 to 30 percent. This risk reduction is notably less than what has been observed for colon, breast, and endometrial cancers for which reductions of percent have been found (1). Some of the studies examined risks by subgroups within their populations and, for three studies (18, 23, 26), the subgroup analyses yielded different results from those found for all cases combined. Giovannucci et al. (18) observed a decreased risk associated with the highest level of physical activity only for men with metastatic prostate cancer. Le Marchand et al. (23) stratified the study population into men aged less than 70 years and 70 or more years and found no effect of physical activity in the younger men but a 60 percent borderline significant increased risk of prostate cancer in the older men. Villeneuve et al. (26) found a statistically significant decreased prostate cancer risk for early life strenuous activity for one province (Ontario) in their Canada-wide case-control study. This decreased effect was not found in other provinces or for other age periods. Controls (no.) (n = 1,063) Age adjusted Multivariable adjusted* OR 95% CI OR 95% CI Low intensity activity (<3 metabolic equivalenthours/week/year) 0 < < , , < , , , , 1.18 p trend Moderate intensity activity (3 6 metabolic equivalent-hours/week/year) 0 < < , , < , , , , 1.26 p trend Vigorous intensity activity (>6 metabolic equivalent-hours/week/year) 0 < < , , < , , , , 0.92 p trend None of the previous studies had a comprehensive measure of lifetime total activity. Seven studies measured usual lifetime occupational activity (11, 15, 20, 24) or some aspect of recreational activity over the lifetime (9, 26, 27) rather than using current activity as a surrogate for lifetime activity. Le Marchand et al. (23) did record a complete lifetime occupational history but did not measure household or recreational activity. Hence, the results of this study cannot be compared directly with any previous investigation. Nonetheless, for these studies that measured some aspect of lifetime activity, decreased risks for occupational activity were found in two studies (11, 15) and for recreational activity in another study (40); however, no effect was found in the remaining studies for occupational and recreational activity (20, 23, 24, 27, 41). Only one of these studies suggested an increased risk associated with occupational activity within a subgroup of the population (23). Few investigators have specifically examined and reported the association between prostate cancer risk and intensity of physical activity (22, 41), and only one study found a decreased prostate cancer risk for strenuous occupational activity done in early adulthood within a subgroup of the study population (41). Hence, additional examination of the effect of intensity of activity on prostate cancer risk is

8 Physical Activity and Prostate Cancer 747 TABLE 6. Odds ratios for lifetime total physical activity by type of activity (frequency and duration only) (n = 2,051), Alberta, Canada, Type of activity (hours/week/year) Cases (no.) Controls (no.) Age adjusted Multivariable adjusted* (n = 988) (n = 1,063) OR 95% CI OR 95% CI Lifetime total activity 0 < < , , < , , , , 1.11 p trend Lifetime occupational activity 0 < < , , < , , , , 0.96 p trend Lifetime household activity 0 < < , , < , , , , 1.71 p trend Lifetime recreational activity 0 < < , , < , , , , 1.17 p trend * Adjusted for age, region (four urban vs. rural areas in Alberta), education (highest level of education achieved), body mass index, waist/hip ratio, total caloric intake, average lifetime total alcohol intake, first degree family history of prostate cancer, number of times had prostate-specific antigen test done, and number of digital rectal examinations. Models of each type of activity were adjusted for the other types of activity. OR, odds ratio; CI, confidence interval. needed to confirm the finding of our study that vigorous intensity activity decreased risk by 30 percent. Clearly, there is considerable inconsistency in the literature on physical activity and prostate cancer risk. Several reasons for the inconsistency across studies are possible. First, the true effect of physical activity on prostate cancer risk may be less than what appears to exist with colon, breast, and endometrial cancers and, hence, the ability to detect these modest associations may be compromised by the measurement methods used in previous studies. Second, the effect of physical activity may be restricted to subgroups within the population. In our study, it appeared that the effect was more evident among younger men and for intense, recreational activity. Giovannucci et al. (18) found a risk reduction for men with metastatic prostate cancer. No other studies have examined effect modification by stage of disease. Hence, physical activity may have a differential effect on prostate cancer risk dependent on numerous factors including age, stage of disease, and type and intensity of activity. There is an underlying biologic rationale to support an etiologic role for physical activity in prostate carcinogenesis. At least four plausible mechanisms exist that might explain how activity influences prostate cancer risk (42). These mechanisms include alterations in endogenous hormones, energy balance, immune function, and antioxidant defense mechanisms. Men who are physically active may have lower endogenous androgen levels, decreased body fat, enhanced immune function, and better antioxidant defense mechanisms than inactive men (42). While evidence exists to support these hypotheses, the association remains weak and unclear. Hence, further epidemiologic studies are needed that also obtain biologic samples so that the biologic mechanisms can be examined simultaneously with the epidemiologic data. The limitations of this study also need to be addressed. To begin, although this study was population based, we had a lower response rate among the controls than among the cases, which may have introduced a selection bias arising from a possible healthy volunteer bias into our study results.

9 748 Friedenreich et al. To address this concern, we compared cases and controls on several lifestyle, medical, and health variables. We found that control study subjects were not uniformly more health conscious or healthy than the cases. We also compared the controls with a sample of men surveyed in Alberta as part of Statistics Canada s National Population Health Survey (NPHS) (43). The comparison was made with men of the same age and also living in Alberta. Results were weighted (and examined) by age category. Our study controls were similar to NPHS participants in terms of marital status; however, controls were more likely to have a university education, and younger members of the cohort were more likely to be White. Controls were less likely to be current smokers, while NPHS participants were more likely to have never smoked. Control participants were more likely to have undergone a medical check-up and to have hypertension, heart disease, and (among younger members of the cohort) a higher body mass index but less likely ever to have had a prostate-specific antigen test. While the assessment of physical activity level in the present study is not comparable with NPHS data, control participants were more likely to be totally sedentary than were NPHS participants. Based on these comparisons, no major selection bias is evident for our control population. The other sources of bias possible in this study are recall and misclassification bias. Since this study was retrospective and because we were interviewing men a few months after their cancer diagnosis, it is possible that they would have ruminated on their diagnosis and biased their responses to these questions. The possibility that such a bias was introduced into this study was reduced by including questions on numerous exposures and by not placing any particular emphasis on physical activity in this questionnaire. In addition, there had been limited public awareness of any association between physical activity and prostate cancer at the time of data collection for this study. Misclassification bias could have occurred in this study since the respondents reported detailed lifetime physical activity patterns. The effect of misclassification of physical activity would have been to decrease the ability of the study to demonstrate an effect of physical activity on prostate cancer risk (nondifferential misclassification bias). Since we observed relatively few associations between physical activity and prostate cancer risk, this bias is a recognized possible limitation of these data. This study used a comprehensive measure of lifetime physical activity that required several analyses to examine the associations thoroughly. In so doing, there were multiple comparisons made and, consequently, some of the decreased and increased statistically significant associations observed could be attributable to chance alone. It is equally important to recognize the strengths of our investigation. These include the restriction of cases to T2 or greater stage, the comprehensive measure of lifetime total activity previously shown to be reliable, the large sample size, the complete assessment of confounding and effect modification, and the population-based study design. Given that prostate cancer incidence is increasing in Canada (44) and worldwide (45) and because few modifiable lifestyle risk factors have been elucidated for this second most common cancer mortality among men, a clear rationale exists for further investigations of lifetime physical activity as a means of prostate cancer risk reduction. These studies would ideally include detailed assessments of lifetime activity in other ethnic and racial groups. Our Alberta population was primarily Caucasian and relatively homogeneous with respect to socioeconomic status and underlying risks for prostate cancer. The possibility of effect modification by family history of prostate cancer, as found in this study, should be explored further in future studies. ACKNOWLEDGMENTS This study was funded by a grant from the Medical Research Council of Canada and the Alberta Cancer Board Research Initiative Program. A National Health Scholar Award from Health Canada, a New Investigator Award from the Canadian Institutes of Health Research, and a Health Scholar Award from the Alberta Heritage Foundation for Medical Research funded C. M. Friedenreich. A Canadian Institutes of Health Research Investigator Award funded K. S. Courneya. A Population Health Investigator Award from the Alberta Heritage Foundation for Medical Research funded S. E. McGregor. A Research Team Grant from the National Cancer Institute of Canada, with funds from the Canadian Cancer Society, supports the research program of C. M. Friedenreich and K. S. Courneya. The authors thank Kathleen Douglas-England, Valerie Hudson, Melanie Keats, and Aleata Rhyorchuk who did study coordination; Lisa Alexander, Michelle Clark, Barry Clattenburg, Pearl Cooke, Linda Davison, Marilyn Dickson, Doreen Mandziuk, Jodi Parrotta, and Nicole Slot who did interviews and recruitment; and Dorothy Allred, Sarah Forrester, and Vicki Wannop who did data entry. They also thank Xuechao Chen who conducted the data analysis and Marla Orenstein who assisted with manuscript preparation. REFERENCES 1. Friedenreich CM, Orenstein MR. Physical activity and cancer prevention: etiologic evidence and biological mechanisms. J Nutr 2002;132(suppl):3456S 64S. 2. Thune I, Lund E. Physical activity and the risk of prostate and testicular cancer: a cohort study of 53,000 Norwegian men. Cancer Causes Control 1994;5: Oliveria SA, Kohl HW, Trichopoulos D, et al. The association between cardiorespiratory fitness and prostate cancer. Med Sci Sports Exerc 1996;28: Hartman TJ, Albanes D, Rautalahti M, et al. Physical activity and prostate cancer in the Alpha-Tocopherol, Beta-Carotene (ATBC) Cancer Prevention Study (Finland). Cancer Causes Control 1998;9: Lund Nilsen TI, Johnsen R, Vatten LJ. Socio-economic and lifestyle factors associated with the risk of prostate cancer. Br J Cancer 2000;82: Clarke G, Whittemore AS. Prostate cancer risk in relation to anthropometry and physical activity: the National Health and Nutrition Examination Survey I Epidemiological Follow-Up Study. Cancer Epidemiol Biomarkers Prev 2000;9:

10 Physical Activity and Prostate Cancer Wannamethee SG, Shaper AG, Walker M. Physical activity and risk of cancer in middle-aged men. Br J Cancer 2001;85: Norman A, Moradi T, Gridley G, et al. Occupational physical activity and risk for prostate cancer in a nationwide cohort study in Sweden. Br J Cancer 2002;86: Yu H, Harris RE, Wynder EL. Case-control study of prostate cancer and socioeconomic factors. Prostate 1988;13: Brownson RC, Chang JC, Davis JR, et al. Physical activity on the job and cancer in Missouri. Am J Public Health 1991;81: Bairati I, Larouche R, Meyer F, et al. Lifetime occupational physical activity and incidental prostate cancer (Canada). Cancer Causes Control 2000;11: Steenland K, Nowlin S, Palu S. Cancer incidence in the National Health and Nutrition Survey I. Follow-up data: diabetes, cholesterol, pulse and physical activity. Cancer Epidemiol Biomarkers Prev 1995;4: Andersson SO, Baron J, Wolk A, et al. Early life risk factors for prostate cancer: a population-based case-control study in Sweden. Cancer Epidemiol Biomarkers Prev 1995;4: Paffenbarger RS, Hyde RT, Wing AL. Physical activity and incidence of cancer in diverse populations: a preliminary report. Am J Clin Nutr 1987;45: Vena JE, Graham S, Zielezny M, et al. Occupational exercise and risk of cancer. Am J Clin Nutr 1987;45: Severson RK, Nomura AM, Grove JS, et al. A prospective analysis of physical activity and cancer. Am J Epidemiol 1989;130: Hsing AW, McLaughlin JK, Zheng W, et al. Occupation, physical activity, and risk of prostate cancer in Shanghai, People s Republic of China. Cancer Causes Control 1994;5: Giovannucci E, Leitzmann M, Spiegelman D, et al. A prospective study of physical activity and prostate cancer in male health professionals. Cancer Res 1998;58: Liu S, Lee IM, Linson P, et al. A prospective study of physical activity and risk of prostate cancer in US physicians. Int J Epidemiol 2000;29: Putnam SD, Cerhan JR, Parker AS, et al. Lifestyle and anthropometric risk factors for prostate cancer in a cohort of Iowa men. Ann Epidemiol 2000;10: Pukkala E, Kaprio J, Koskenvuo M, et al. Cancer incidence among Finnish world class male athletes. Int J Sports Med 2000;21: Lee IM, Sesso HD, Paffenbarger RS Jr. A prospective cohort study of physical activity and body size in relation to prostate cancer risk (United States). Cancer Causes Control 2001;12: Le Marchand L, Kolonel LN, Yoshizawa CN. Lifetime occupational physical activity and prostate cancer risk. Am J Epidemiol 1991;133: Dosemeci M, Hayes RB, Vetter R, et al. Occupational physical activity, socioeconomic status, and risks of 15 cancer sites in Turkey. Cancer Causes Control 1993;4: Whittemore AS, Kolonel LN, Wu AH, et al. Prostate cancer in relation to diet, physical activity, and body size in blacks, whites, and Asians in the United States and Canada. J Natl Cancer Inst 1995;87: Villeneuve PJ, Johnson KC, Kreiger N, et al. Risk factors for prostate cancer: results from the Canadian National Enhanced Cancer Surveillance System. The Canadian Cancer Registries Epidemiology Research Group. Cancer Causes Control 1999; 10: Lacey JV Jr, Deng J, Dosemeci M, et al. Prostate cancer, benign prostatic hyperplasia and physical activity in Shanghai, China. Int J Epidemiol 2001;30: Cerhan JR, Torner JC, Lynch CF, et al. Association of smoking, body mass, and physical activity with risk of prostate cancer in the Iowa 65+ Rural Health Study (United States). Cancer Causes Control 1997;8: West DW, Slattery ML, Robison LM, et al. Adult dietary intake and prostate cancer risk in Utah: a case-control study with special emphasis on aggressive tumors. Cancer Causes Control 1991;2: Sung JF, Lin RS, Pu YS, et al. Risk factors for prostate carcinoma in Taiwan: a case-control study in a Chinese population. Cancer 1999;86: Polednak AP. College athletics, body size, and cancer mortality. Cancer 1976;38: Cancer in North America, Vol 1. Incidence. Sacramento, CA: North American Association of Cancer Registries, Block G, Hartman AM, Dresser CM, et al. A data-based approach to diet questionnaire design and testing. Am J Epidemiol 1986;124: Willis GB. Cognitive interviewing and questionnaire design: a training manual. Hyattsville, MD: National Center for Health Statistics, (Cognitive methods working paper series no. 7). 35. Friedenreich CM, Courneya KS, Bryant HE. The Lifetime Total Physical Activity Questionnaire: development and reliability. Med Sci Sports Exerc 1998;30: Friedenreich CM, Bryant HE, Courneya KS. Case-control study of lifetime physical activity and breast cancer risk. Am J Epidemiol 2001;154: Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc 2000;32(suppl):S Anshell MH, Freedson P, Hamill J, et al. Dictionary of the sports and exercise sciences. Champaign, IL: Human Kinetics, Jacobs DR Jr, Ainsworth BE, Hartman TJ, et al. A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med Sci Sports Exerc 1993;25: Yu H, Harris RE, Wynder EL. Case-control study of prostate cancer and socioeconomic factors. Prostate 1988;13: Villeneuve PJ, Johnson KC, Kreiger N, et al. Risk factors for prostate cancer: results from the Canadian National Enhanced Cancer Surveillance System. The Canadian Cancer Registries Epidemiology Research Group. Cancer Causes Control 1999; 10: Friedenreich CM, Thune I. A review of physical activity and prostate cancer risk. Cancer Causes Control 2001;12: Statistics Canada. National Population Health Survey, Ottawa, Ontario, Canada: Statistics Canada, National Cancer Institute of Canada. Canadian cancer statistics Toronto, Ontario, Canada: National Cancer Institute of Canada, Cancer incidence in five continents. Vol VIII. IARC Sci Publ 2002;(155):1 781.