Editor’s note: Go online to view the Journal Club questions in the Supplemental Digital Content: seehttp://links.lww.com/ESSR/A8
Cancer is a group of more than 100 diseases characterized by uncontrolled growth of abnormal cells. More than 1.6 million Americans are expected to be diagnosed with cancer in 2013, and more than 580,000 are expected to die from the disease (1). The lifetime probability of developing cancer in the United States is about 45% for men and 38% for women. The current 5-year relative survival rate is 68%; however, this rate varies dramatically depending on the type of cancer and stage of disease. The high incidence and survival rates have resulted in almost 14 million American cancer survivors (i.e., a person with a previous diagnosis of cancer).
Unfortunately, surviving cancer often requires difficult and prolonged medical treatments that can last from several months to many years. These treatments can include surgery, radiation therapy, chemotherapy, hormone therapy, biologic therapy, and stem cell transplant. Not surprisingly, cancer and its treatments can take a significant toll on the physical and emotional well-being of cancer survivors and produce numerous acute and chronic health problems (2). Physical activity (PA) has been proposed as a strategy to help cancer survivors manage symptoms, improve quality of life, and possibly even extend survival (27).
PA and cancer survivorship is the field that studies the links between cancer variables and PA in people who have been diagnosed as having cancer. For the field of PA and cancer survivorship to flourish, researchers must demonstrate compelling links between cancer variables and PA. In this article, I propose a simple framework that describes how cancer variables may be linked with PA. The framework highlights four key propositions that essentially define the field of PA and cancer survivorship. I then draw on some of my own research to provide examples of how each of these four propositions may be tested. My purpose here is to demonstrate the utility of the framework for organizing and conceptualizing PA and cancer survivorship research rather than providing a comprehensive review or analysis of each of the propositions.
A FRAMEWORK FOR PHYSICAL ACTIVITY AND CANCER SURVIVORSHIP RESEARCH
Cancer is a complex disease with many different treatment options, resulting in a large number of “cancer variables” (Table 1). In simplest terms, cancer variables can be divided into disease and treatment variables. Disease variables describe the cancer itself, including the type, biology, spread, transformation, recurrence, and progression of disease. Treatment variables describe the various treatments used to cure or control the disease, including the type, dose, administration, scheduling, combination, and sequencing of treatments. It is these “cancer variables” that make the field of PA and cancer survivorship unique within exercise science. How these cancer variables interact with PA is a conceptual and empirical matter.
PA is a complex behavior with many attributes. One critical attribute is the intensity of PA that varies along a continuum from rest to maximal exertion (Fig. 1). Arbitrary cut points in terms of energy expenditure are used to divide activities into various intensity categories, including sedentary, light, moderate, and vigorous. Other important attributes of PA include type, frequency, duration, program length, progression, program variability (periodization), and context (physical and social environment). Any or all of these PA attributes may be linked with cancer variables.
The proposed framework highlights four primary associations between PA and cancer variables (Fig. 2). Cancer variables may be outcomes of PA, determinants of PA, moderators of PA outcomes, and/or moderators of PA determinants. These four associations give rise to four propositions that essentially define the field of PA and cancer survivorship (Table 2). Support for any of these propositions strengthens the case for the uniqueness and importance of the field of PA and cancer survivorship. If cancer variables are not outcomes of PA, if they do not moderate other outcomes of PA, if they are not determinants of PA, or if they do not moderate other determinants of PA, then the case for the field of PA and cancer survivorship is weak. We simply would apply exercise research from other populations to cancer survivors without any concern for cancer variables. In the following sections, I expand on each of these four propositions and review examples of my own research that have tested each of these propositions. I draw mostly from my own research program while acknowledging the larger body of research that has tested these propositions.
CANCER VARIABLES MAY BE OUTCOMES OF PHYSICAL ACTIVITY
Perhaps the primary proposition of the field of PA and cancer survivorship is that cancer variables may be outcomes of PA. By definition, knowledge about the effects of exercise on cancer outcomes can be obtained only from research with cancer survivors. This proposition may be the most important for the field of PA and cancer survivorship because it highlights the outcomes that are unique to this field and of most significance to cancer survivors and oncologists. If exercise were to have important effects on cancer outcomes, it is likely that cancer survivors would be motivated to exercise, oncologists would be motivated to prescribe exercise, cancer care organizations would be motivated to offer exercise programs, and insurance companies would be motivated to cover the costs of such programs. As outcomes of PA, cancer variables also may be mechanisms (i.e., intermediate outcomes) of other outcomes. For example, exercise may lower the risk of disease recurrence, which may lead to improved quality of life.
Limited research has examined the effects of exercise on cancer outcomes. We conducted one of the early studies to examine the effects of exercise on a cancer treatment variable. In the Supervised Trial of Aerobic versus Resistance Training (START), we compared aerobic and resistance exercise with usual care in 242 breast cancer patients receiving chemotherapy (12). We closely monitored chemotherapy completion rate to ensure that vigorous-intensity exercise did not interfere with breast cancer patients’ ability to complete their treatments. Unexpectedly, the study showed that the weight training group completed more of its chemotherapy than the usual care group (89.8% vs 84.1%; P = 0.033). Given that this finding was unexpected and the potential mechanisms are unclear, it needs to be replicated before it can be considered reliable. Nevertheless, the START Trial provides an example of how exercise might be examined for its effects on treatment completion.
We also conducted one of the first studies to report cancer treatment response as an outcome. In the Healthy Exercise for Lymphoma Patients (HELP) Trial, we compared aerobic exercise with usual care in 122 lymphoma patients receiving chemotherapy or no treatments (15). In the 54 patients who were receiving chemotherapy, we tracked their response to the chemotherapy to ensure no adverse effect. Treatment response was evaluated as either progressive disease (disease gets worse), stable disease (no meaningful change), partial response (some reduction in disease), or complete response (no evidence of disease). We found that 46.4% of lymphoma patients in the exercise group had a complete response to their chemotherapy compared with just 30.8% in the usual care group (P = 0.24). Although the finding was not statistically significant and the mechanisms are unclear, the HELP Trial provides an example of how exercise can be examined for its impact on cancer treatment response.
Perhaps the most compelling cancer outcomes for patients and oncologists are the disease outcomes. A growing number of observational studies have shown links between higher levels of PA and a lower risk of recurrence, cancer-specific mortality, and all-cause mortality (3). We recently reported on one of the first randomized trials to examine the effects of exercise on disease outcomes in any cancer patient group (13). In the previously mentioned START Trial, we followed up study patients for disease-free survival (DFS), which includes recurrences, second cancers, and deaths from any cause. The two exercise arms (aerobic and resistance) were combined for analysis. Eight-year DFS was 82.7% for the exercise groups compared with 75.6% for the control group (hazard ratio [HR] = 0.68; 95% confidence interval, 0.37–1.24; log-rank P = 0.21). Although not statistically significant, the magnitude of this effect is similar to that observed in many drug trials. This exploratory follow-up provides some of the first randomized data to suggest that adding exercise to standard chemotherapy for breast cancer may improve disease outcomes.
Although some exercise studies have reported cancer variables as outcomes, very few have targeted cancer variables as their primary outcome, perhaps because of the large sample sizes required for cancer outcomes compared with fitness outcomes or patient-reported outcomes. To address this issue, we are conducting two large exercise studies with disease outcomes as the primary end points. In the Colon Health And Life-Long Exercise ChaNGE (CHALLENGE) Trial (7), we are examining the effects of a structured PA intervention on DFS in 962 colon cancer survivors with stage II or III colon cancer who have completed adjuvant therapy within the past 2 to 6 months. To our knowledge, the CHALLENGE Trial is the first randomized exercise trial in any cancer survivor population with a disease outcome as the primary end point.
A second ongoing study is the Alberta Moving BEyond bReast cancer (AMBER) Study (18), which is a prospective cohort study designed specifically to examine the role of PA and health-related fitness in breast cancer survivorship. The AMBER Study will enroll 1500 newly diagnosed stage I to IIIC breast cancer survivors in Alberta, Canada. Assessments will be made at baseline (within 90 d of surgery), 1 year, 3 years, and 5 years consisting of objective and self-reported measurements of PA, health-related fitness, biomarkers, lymphedema, and patient-reported outcomes. Participants will be followed for disease outcomes, including recurrence, second cancers, and deaths. To our knowledge, AMBER is the first cohort study in any cancer survivor group to have PA and health-related fitness as the primary exposures of interest.
Determining the link between PA and cancer outcomes may be the most important undertaking for the field of PA and cancer survivorship. Some evidence is available to suggest that exercise may improve treatment outcomes, but the data are only suggestive. Moreover, evidence is available to suggest that PA is associated with a lower risk of recurrence and cancer-specific mortality in several cancer survivor groups; however, most studies to date are observational. Experimental research on the mechanisms of PA and cancer outcomes would also strengthen the case for a causal inference between PA and cancer outcomes.
CANCER VARIABLES MAY BE MODERATORS OF PHYSICAL ACTIVITY OUTCOMES
Another key proposition of the field of PA and cancer survivorship is that cancer variables may be moderators of PA outcomes. That is, cancer variables may alter the typical exercise response observed in other populations in terms of both safety and efficacy outcomes. The exercise response might be amplified or nullified depending on the cancer variable. This proposition is fundamental to the field of PA and cancer survivorship because it suggests that research on other populations may not generalize to cancer survivors. Moreover, it suggests that even research on one specific cancer variable (i.e., a disease or treatment characteristic) may not generalize to another specific cancer variable (i.e., disease or treatment characteristic). Cancer variables may be general moderators (e.g., comparing cancer survivors with healthy controls) or specific moderators (e.g., comparing cancer patients with the same disease and stage receiving two different chemotherapy regimens) of the exercise response (Table 3). Cancer variables may moderate any exercise outcome, including standard physiological outcomes, psychosocial outcomes, symptoms, disease outcomes, or any of the mechanisms of these outcomes. If this proposition is supported by research, it suggests that the exercise prescription to improve outcomes in cancer survivors may need to take into account cancer variables.
Most of the research in the field of PA and cancer survivorship has focused on this second proposition by targeting outcomes, such as physical functioning, fatigue, symptoms, and quality of life (25,26). Much of this research, however, has treated cancer variables as “implied” moderators of the exercise response. That is, the rationale for the study implies a potentially different exercise response based on some cancer variable, but the moderator counterpart is not actually included in the study. For example, a study might examine the effects of exercise on fatigue in colon cancer patients receiving chemotherapy. In the introduction, the authors may imply that the fatigue response to exercise may be different because the patients have colon cancer and/or are receiving chemotherapy; yet patients with the implied moderator counterpart are not included in the study (e.g., other cancer patients receiving chemotherapy or colon cancer patients not receiving chemotherapy). Nevertheless, the authors typically compare their results with previous studies of the implied moderator in the discussion section. In essence, the primary focus of these studies is on quantifying the magnitude of exercise effect in the presence of the particular cancer variable(s) rather than on directly testing the implied moderator effect. Such a study design provides important information on the effects of exercise in the cancer context, but it provides the weakest evidence of a true moderating effect of cancer variables.
There are a number of important conceptual and methodological issues to consider when examining cancer moderators in randomized exercise trials (6,22,29). Some of the most important design and analytical features are summarized in Table 4. Ideally, researchers should include the moderator counterpart in the study, stratify the sampling frame based on the moderator (if necessary), stratify randomization based on the moderator, randomize both moderator counterparts to intervention versus comparison, conduct a statistical test of the interaction (including adjustment for important covariates), and adequately power the study to test for an interaction effect (including accounting for multiple testing). Few PA and cancer survivorship studies have included all of these design features to test cancer moderators.
For example, an early study by MacVicar and Winningham (24) examined the most general cancer moderator (i.e., patients with cancer vs people without cancer). Specifically, these researchers examined the effects of 10 wk of aerobic exercise training on aerobic fitness and mood disturbance in six stage II breast cancer patients receiving chemotherapy compared with six healthy women. Results showed a similar response to exercise in both groups for both outcomes. The results are important because they demonstrate the combined impact (or lack of impact) of several cancer variables on PA response (e.g., a specific combination of cancer site, disease stage, and treatment regimen). Nevertheless, the design of the study was limited because the moderator counterparts were not both randomized to intervention versus comparison groups.
Our research has examined more specific cancer moderators of the exercise response such as disease stage and cancer treatments and included additional design features that strengthened the test of the cancer moderator. For example, in our HELP Trial, we sampled approximately equal numbers of lymphoma patients receiving chemotherapy versus no treatments, stratified on treatment status before randomization, randomized both groups of patients to exercise versus usual care, and conducted a statistical test of the interaction. We found that aerobic exercise improved health-related fitness and patient-reported outcomes overall and that there was no statistical interaction between group assignment and treatment status (14). That is, patients on chemotherapy had the same positive response to exercise as patients not on treatment. In further subgroup analyses, we did find that improvements in lean body mass were moderated by disease stage (14). Specifically, we found that aerobic exercise training resulted in a significant increase in lean body mass for lymphoma patients with stage III/IV disease but not for patients with stage I/II disease or no evidence of disease. Nevertheless, the study was not powered to test for moderator effects, and we did not adjust for multiple testing.
Similarly, in our START Trial, we stratified breast cancer patients on chemotherapy regimen (taxane-based vs non–taxane-based chemotherapy), randomized both groups of patients to exercise versus usual care, and conducted a statistical test of the interaction (10). We found that chemotherapy regimen moderated the effects of exercise training on muscular strength. Specifically, weight training resulted in a large improvement in muscular strength in breast cancer patients who were receiving non–taxane-based chemotherapies compared with a more modest improvement for breast cancer patients receiving taxane-based chemotherapies. To our knowledge, these data were the first to suggest that the exercise response may vary based on the chemotherapy regimen. In further analyses of cancer moderators, we reported a similar moderating effect of disease stage on body composition changes as in the HELP Trial (10). Similar to our HELP Trial, however, the START Trial was not powered to test for moderator effects and we did not adjust for multiple testing.
Finally, in our follow-up of disease outcomes in the START Trial (13), we conducted subgroup analyses based on cancer variables and found a suggestion of stronger effects of exercise on DFS for women who had stage II/III cancer, estrogen receptor–positive tumors, human epidermal receptor 2–positive tumors, received taxane-based chemotherapies, and received at least 85% of their intended chemotherapy dose intensity. These findings demonstrate how some cancer variables may moderate the effects of exercise on other cancer variables (i.e., cancer moderators of cancer outcomes). Nevertheless, we did not conduct statistical tests of the interactions because our sample size was not adequate; consequently, these results are only suggestive of possible moderating effects.
Observational studies also can examine cancer variables as moderators of PA outcomes, with the typical advantage of having larger sample sizes and greater power to detect moderator effects. For example, in a survey of 588 young adult cancer survivors (YACS), we found significant interactions between PA participation and past chemotherapy for physical functioning, self-esteem, stress, and depression (4). The general pattern of the interactions was that the association between PA and the outcomes was stronger and more dose dependent for YACS who had previously been treated with chemotherapy, whereas it was more modest and exhibited a threshold association for YACS who had not received chemotherapy. These data suggest that YACS who were previously treated with chemotherapy may stand to benefit even more from exercise.
Uncovering the most important cancer moderators of the exercise response is a critical undertaking for the field of PA and cancer survivorship. Based on current research, there are some suggestions that cancer type, disease stage, and treatments may moderate the exercise response, but the relative importance of these moderators is unknown and they may depend on the PA intervention and outcome. Moreover, designing exercise trials to provide strong tests of cancer moderators is a challenge because of the more sophisticated design features and substantially larger sample sizes required. An effort to combine individual studies of PA and cancer survivorship for meta-analyses may provide additional power to examine cancer moderators (5).
CANCER VARIABLES MAY BE DETERMINANTS OF PHYSICAL ACTIVITY
The third proposition of the field of PA and cancer survivorship is that cancer variables may be determinants of PA. That is, cancer variables may influence a person’s ability and/or willingness to exercise. In one sense, cancer determinants research addresses the issue of the feasibility of exercise in cancer survivors (i.e., the impact of cancer variables on exercise adherence or PA participation). Consequently, testing this proposition is an important undertaking for the field. Cancer variables may be determinants of any aspect of PA noted earlier, including the type, frequency, intensity, duration, progression, periodization, and context of PA. By extension, cancer variables also may be determinants of any mediators of PA, such as attitude, confidence, social support, benefits, and barriers. Moreover, cancer variables may themselves be mediators of other determinants of PA. For example, age may be associated with PA in cancer survivors because it influences the types of treatments offered and/or side effects experienced.
A growing number of studies have examined the link between cancer variables and exercise adherence or PA participation (9). It is still unclear, however, if cancer survivors as a group have lower exercise levels than the general population (23). Nevertheless, it seems that more specific cancer variables such as cancer type, disease stage, and treatments may influence PA behavior (9). For example, in our START Trial, we found that adherence to supervised exercise during chemotherapy was predicted by disease stage, with better adherence, paradoxically, for patients with more advanced disease (11). Conversely, at 6-month follow-up of the START Trial, we found that type of surgery predicted exercise levels, with survivors who had received breast-conserving surgery (i.e., lumpectomy) being more likely to exercise than survivors treated with mastectomy (8).
As a second example, in our HELP Trial, we found lower adherence to supervised exercise in lymphoma patients previously treated with radiation therapy (17), whereas cancer type predicted meeting the exercise guidelines at 6-month follow-up (16). Specifically, Hodgkin lymphoma patients were more likely to be exercising than non-Hodgkin lymphoma patients. As a final example, in a cross-sectional survey of 359 ovarian cancer survivors (28), we found that survivors were more likely to be meeting PA guidelines if they were more than 5 years postdiagnosis, had early-stage disease, and were currently free of disease.
Given that some cancer variables may be determinants of PA, it is important to shed light on the possible mediators of these associations. For example, in a prospective study using the theory of planned behavior to predict exercise in 397 bladder cancer survivors (21), we found that bladder cancer survivors who received adjuvant therapy and had invasive disease had lower levels of exercise. Consistent with these associations, we also found that bladder cancer survivors who had received adjuvant therapy had less favorable affective attitudes toward exercise (i.e., felt that exercise would be less enjoyable) compared with those survivors who had not received adjuvant therapy. Moreover, bladder cancer survivors who had invasive disease compared with superficial disease had less favorable affective attitudes, instrumental attitudes (i.e., felt exercise would be less beneficial for them), and descriptive norms (i.e., felt that fewer people in their social network exercised). Consequently, it is likely that cancer variables have their impact on PA through social cognitive variables; however, very few studies actually have examined social cognitive variables as mediators of cancer variables.
CANCER VARIABLES MAY BE MODERATORS OF PHYSICAL ACTIVITY DETERMINANTS
The fourth proposition of the field of PA and cancer survivorship is that cancer variables may be moderators of PA determinants. That is, cancer variables may alter the standard associations between exercise and other determinants typically observed in other populations. The same cancer variables that moderate the exercise response (outcomes) might also moderate exercise determinants. For example, cancer variables may moderate the associations of PA with demographic variables (e.g., age, sex, education), medical variables (e.g., body mass index, comorbidities), environmental variables (e.g., neighborhood walkability), and even other cancer variables. Moreover, cancer variables may moderate the associations between social cognitive mediators and PA (e.g., attitudes, self-efficacy, social support, intention). Finally, cancer variables may moderate the effectiveness of PA behavior change interventions (e.g., telephone counseling, Internet-delivered interventions, print materials, physician recommendation). Research on this proposition is critical to the field of PA and cancer survivorship because it suggests that behavior change interventions to promote PA in cancer survivors may need to take into account cancer variables.
Research on cancer moderators of PA determinants is limited. Moreover, similar to the research on cancer moderators of PA outcomes, much of the research on PA determinants has treated cancer variables as “implied” moderators. For example, we conducted a study that examined the effects of an oncologist’s recommendation to exercise in newly diagnosed breast cancer survivors attending their treatment consultation (20). Our implication was that the effectiveness of an exercise recommendation from an oncologist to breast cancer patients about to start adjuvant treatment may be different from a recommendation from a general physician to an otherwise healthy adult not facing a life-threatening condition. Our findings suggested that an oncologist recommendation may increase exercise behavior in newly diagnosed breast cancer survivors — similar to the effects of physician-based counseling in the general primary care population.
Nevertheless, some PA determinants research has directly tested cancer moderators. For example, in a study of 741 breast, prostate, and colorectal cancer survivors, we examined the invariance of the theory of planned behavior across cancer types using structural equation modeling (19). In terms of the invariance analyses, we found that the intention to planning relationship was stronger for breast versus prostate cancer survivors; the affective attitude to intention relationship was stronger for colorectal versus prostate cancer survivors; and the planning to PA relationship was stronger for prostate versus colorectal cancer survivors. These data suggest that the relative strength of the associations among the theory of planned behavior and PA may vary as a function of cancer type. In additional analyses, we found that cancer treatment variables were strong correlates of PA in prostate cancer survivors but not for breast or colorectal (i.e., a cancer moderator of cancer determinants), and body mass index was strongly associated with PA in breast cancer survivors but not prostate or colorectal. These data suggest that cancer type also may moderate the associations between standard medical/health variables and PA.
As another example, in our study of bladder cancer survivors discussed earlier (21), we examined whether the associations among the theory of planned behavior and PA varied by receipt of adjuvant therapy. We found that receiving adjuvant therapy moderated the association between perceived behavioral control, intention, and PA based on a statistical test of the interaction. Specifically, for bladder cancer survivors who had received adjuvant therapy, intention was the only independent predictor of exercise; whereas for survivors who had not received adjuvant therapy, perceived behavioral control was the only independent predictor of exercise. Research on cancer moderators of PA determinants will help shed light on whether interventions to promote PA in cancer survivors need to take into account cancer variables.
In this article, I have proposed a simple framework that highlights four primary associations between cancer variables and PA. I have summarized these links in the form of four general propositions for the field of PA and cancer survivorship. I then reviewed some of my own research that has tested each of these propositions. A more complex framework could have been proposed that included mediated moderation, moderated mediation, and moderated moderators, but the additional complexity may have obscured the four most basic propositions. It is hoped that this simple organizational framework will stimulate new and more systematic research in the field of PA and cancer survivorship.
The field of PA and cancer survivorship is making progress in our understanding of how a cancer diagnosis and its treatments interact with exercise. Nevertheless, the proposed framework highlights that much more research remains to be done. Cancer variables may be outcomes of PA, but few randomized trials have demonstrated causal effects or targeted cancer variables as their primary outcome. Cancer variables may be moderators of PA outcomes, but much of the evidence is implied rather than demonstrated. Cancer variables may be determinants of PA, but the mediators of these associations are unclear. Cancer variables may be moderators of PA determinants (and interventions) but, again, much of the evidence is implied rather than demonstrated. The field of PA and cancer survivorship will be strengthened further by demonstrating important links between PA and cancer variables anywhere within the proposed framework. Research questions are virtually limitless given the large number of cancer variables, PA variables, PA outcomes, and PA determinants. The challenge for this field is to identify which cancer variables are the most important outcomes, determinants, and moderators for informing clinical practice and public health policy regarding PA and cancer survivorship.
Finally, the simple framework I propose here also might have utility for PA researchers working with other chronic disease groups (e.g., heart disease, diabetes, pulmonary disease). It may be as simple as developing a Table 1 for the particular disease. The framework also may have utility for PA researchers interested in other population groups based on demographics (e.g., age, sex, ethnicity). Of course, such variables will not be outcomes of PA, but the other three propositions would seem to be of interest. Finally, the framework may even have utility for PA researchers interested in particular physiological and psychosocial outcomes (e.g., physical functioning, depression, anxiety, fatigue). For example, depression is known to be both an outcome and determinant of PA, but perhaps it is also a moderator of outcomes and determinants.
The author is supported by the Canada Research Chairs Program. The author acknowledges the many colleagues, students, staff, and participants who have contributed to his research program.
The author declares no conflicts of interest.
1. American Cancer Society. Cancer Facts & Figures 2013
. Atlanta (GA): American Cancer Society; 2013 [cited 2013 Nov]. Available from: http://www.cancer.org/research/cancerfactsfigures/cancerfactsfigures/cancer-facts-figures-2013
2. Aziz NM. Cancer survivorship research: State of knowledge, challenges and opportunities. Acta Oncol
. 2007; 46 (4): 417–32.
3. Ballard-Barbash R, Friedenreich CM, Courneya KS, Siddiqi SM, McTiernan A, Alfano CM. Physical activity, biomarkers, and disease outcomes in cancer survivors: A systematic review. J. Natl. Cancer Inst
. 2012; 104 (11): 815–40.
4. Bélanger LJ, Plotnikoff RC, Clark A, Courneya KS. Physical activity and health-related quality of life in young adult cancer survivors: A Canadian provincial survey. J. Cancer Surviv
. 2011; 5 (1): 44–53.
5. Buffart LM, Kalter J, Chinapaw MJ, et al. Predicting OptimaL cAncer RehabIlitation and Supportive care (POLARIS): Rationale and design for meta-analyses of individual patient data of randomized controlled trials that evaluate the effect of physical activity and psychosocial interventions on health-related quality of life in cancer survivors. Syst. Rev
. 2013; 2 (1): 75.
6. Cook DI, Gebski VJ, Keech AC. Subgroup analysis in clinical trials. Med. J. Aust
. 2004; 180 (6): 289–91.
7. Courneya KS, Booth CM, Gill S, et al. The colon health and life-long exercise change trial: A randomized trial of the National Cancer Institute of Canada Clinical Trials Group. Current Oncology
. 2008; 15 (6): 8–16.
8. Courneya KS, Friedenreich CM, Reid RD, et al. Predictors of follow-up exercise behavior 6 months after a randomized trial of exercise training during breast cancer chemotherapy. Breast Canc. Res. Treat.
2009; 114 (1): 179–87.
9. Courneya KS, Karvinen KH, Vallance JKH. Exercise motivation and behavior change. In: Feuerstein M, editor. Handbook of Cancer Survivorship
. New York (NY): Springer Science+Business Media, LLC; 2007, p. 113–32.
10. Courneya KS, McKenzie DC, Mackey JR, et al. Moderators of the effects of exercise training in breast cancer patients receiving chemotherapy: A randomized controlled trial. Cancer
. 2008; 112 (8): 1845–53.
11. Courneya KS, Segal RJ, Gelmon K, et al. Predictors of supervised exercise adherence during breast cancer chemotherapy. Med. Sci. Sports Exerc
. 2008; 40 (6): 1180–7.
12. Courneya KS, Segal RJ, Mackey JR, et al. Effects of aerobic and resistance exercise in breast cancer patients receiving adjuvant chemotherapy: A multicenter randomized controlled trial. J. Clin. Oncol
. 2007; 25 (28): 4396–404.
13. Courneya KS, Segal RJ, McKenzie DC, et al. Effects of exercise during adjuvant chemotherapy on breast cancer outcomes. Med. Sci. Sports Exerc
. (in press).
14. Courneya KS, Sellar CM, Stevinson C, et al. Moderator effects in a randomized controlled trial of exercise training in lymphoma patients. Cancer Epidemiol. Biomarkers Prev
. 2009; 18 (10): 2600–7.
15. Courneya KS, Sellar CM, Stevinson C, et al. Randomized controlled trial of the effects of aerobic exercise on physical functioning and quality of life in lymphoma patients. J. Clin. Oncol
. 2009; 27 (27): 4605–12.
16. Courneya KS, Stevinson C, McNeely ML, et al. Predictors of follow-up exercise behavior 6 months after a randomized trial of supervised exercise training in lymphoma patients. Psychooncology
. 2012; 21 (10): 1124–31.
17. Courneya KS, Stevinson C, McNeely ML, et al. Predictors of adherence to supervised exercise in lymphoma patients participating in a randomized controlled trial. Ann. Behav. Med
. 2010; 40 (1): 30–9.
18. Courneya KS, Vallance JK, Culos-Reed SN, et al. The Alberta moving beyond breast cancer (AMBER) cohort study: A prospective study of physical activity and health-related fitness in breast cancer survivors. BMC Cancer
. 2012; 12 (1): 525–37.
19. Forbes CC, Blanchard CM, Mummery WK, Courneya KS. A comparison of physical activity correlates across breast, prostate and colorectal cancer survivors in Nova Scotia. Supp. Care. Canc
. 22: 891–903.
20. Jones LW, Courneya KS, Fairey AS, Mackey JR. Effects of an oncologist’s recommendation to exercise on self-reported exercise behavior in newly diagnosed breast cancer survivors: A single-blind, randomized controlled trial. Ann. Behav. Med
. 2004; 28 (2): 105–13.
21. Karvinen KH, Courneya KS, Plotnikoff RC, Spence JC, Venner PM, North S. A prospective study of the determinants of exercise in bladder cancer survivors using the theory of planned behavior. Support Care Cancer
. 2009; 17 (2): 171–9.
22. Kraemer HC, Wilson GT, Fairburn CG, Agras WS. Mediators and moderators of treatment effects in randomized clinical trials. Arch. Gen. Psychiatry
. 2002; 59 (10): 877–83.
23. Kwon S, Hou N, Wang M. Comparison of physical activity levels between cancer survivors and non-cancer participants in the 2009 BRFSS. J. Cancer Surviv
. 2012; 6 (1): 54–62.
24. MacVicar MG, Winningham ML. Promoting the functional capacity of cancer patients. Cancer Bullet
. 1986; 38: 235–9.
25. Mishra SI, Scherer RW, Geigle PM, et al. Exercise interventions on health-related quality of life for cancer survivors. Cochrane Database Syst. Rev
. 2012; 8.
26. Mishra SI, Scherer RW, Snyder C, Geigle PM, Berlanstein DR, Topaloglu O. Exercise interventions on health-related quality of life for people with cancer during active treatment. Cochrane Database Syst. Rev
. 2012; 8.
27. Schmitz KH, Courneya KS, Matthews C, et al. American college of sports medicine roundtable on exercise guidelines for cancer survivors. Med. Sci. Sports Exerc
. 2010; 42 (7): 1409–26.
28. Stevinson C, Tonkin K, Capstick V, et al. A population-based study of the determinants of physical activity in ovarian cancer survivors. J. Phys. Activity Health
. 2009; 6 (3): 339–46.
29. Wang R, Lagakos SW, Ware JH, Hunter DJ, Drazen JM. Statistics in medicine — reporting of subgroup analyses in clinical trials. N. Engl. J. Med
. 2007; 357 (21): 2189–94.