Abstract
Background
The effects of exercise on cancer outcomes may differ depending on its positioning within different cancer treatment combinations. We examined whether the associations between physical activity (PA) and cancer outcomes varied by cancer treatment modality or timing of PA.
Methods
We conducted a secondary analysis of the Prostate Cancer Cohort Study consisting of 830 men in Alberta, Canada with newly diagnosed prostate cancer. Lifetime prediagnosis PA was assessed by an in-person interview shortly after diagnosis whereas postdiagnosis PA was assessed at 2–3 year intervals by an in-person interview (first follow-up) or self-report (second and third follow-ups). Cox proportional hazards regression models were used to test interactions between PA and treatment modalities for disease-free survival, overall survival, and prostate cancer-specific disease-free survival.
Results
Postdiagnosis vigorous PA significantly interacted with surgery (p < 0.001) and radiotherapy (p = 0.003). Specifically, patients who had surgery experienced a 61% lower likelihood of a disease-free survival event if they engaged in any versus no postdiagnosis vigorous PA (HR = 0.39, 95% CI = 0.27–0.57). Conversely, patients who received radiotherapy did not experience any benefit from postdiagnosis vigorous PA (HR = 1.14, 95% CI = 0.88–1.47).
Conclusions
The role of PA as a treatment for prostate cancer may depend on its combination and sequencing with other treatments.
Subject terms: Prostate cancer, Targeted therapies
Introduction
Observational studies may provide important insights into the role of physical activity (PA) as a possible cancer treatment; however, few studies have been designed, analyzed, or interpreted from a clinical oncology perspective [1]. The Exercise as Cancer Treatment (EXACT) Framework [2] and the Exercise Across the Postdiagnosis Cancer Continuum (EPiCC) Framework [3] were proposed to facilitate a more systematic approach to the study of exercise as a cancer treatment across a wide range of cancers and treatment protocols. One key proposition of the EXACT and EPiCC Frameworks is that the effects of exercise on cancer outcomes may vary depending on its positioning within different cancer treatment combinations. That is, previous cancer treatments may alter the biology, tumor microenvironment, and/or location of any remaining cancer [4, 5] and, thereby, modify the effects of exercise on cancer outcomes. Conversely, exercise may also alter the biology, tumor microenvironment, and/or location of any remaining cancer [6, 7] and possibly modify the efficacy of any subsequent cancer treatments. Treatment combination and sequencing is a critical issue in clinical oncology [8, 9] that has been understudied in exercise oncology research [1–3].
Most observational studies of PA and cancer outcomes have not addressed the issue of how PA may interact with different treatment modalities and combinations. Rather, most studies have typically analyzed cancer treatments as possible confounders rather than as potential effect modifiers [1]. As a result, the clinical utility of these studies has been limited. One recommendation to improve the clinical utility of these observational studies is to conduct subgroup analyses based on different cancer treatment modalities and combinations [1]. Such analyses may provide insights into the optimal combination and sequencing of PA in relation to other cancer treatments and provide clues about possible mechanisms.
Here, we report a secondary analysis from a cohort study of PA and prostate cancer survival that collected data on cancer treatments [10]. The key finding of the main study was that postdiagnosis recreational PA was associated with a significantly lower risk of death from prostate cancer (HR: 0.56; 95% CI, 0.35–0.90). The primary purpose of the present paper was to explore whether any of the associations of PA with cancer outcomes reported in the primary analysis [10] were altered by the cancer treatment modalities received (surgery, radiation therapy, and hormone therapy) or the timing of PA in relation to those cancer treatments (before or after treatments).
Methods
Study design and participants
The main methods from the cohort study have been reported elsewhere [10]. Briefly, a population-based case-control study was conducted between 1997 and 2000 in Alberta, Canada that included 988 men with histologically confirmed, clinically significant stage T2 or greater invasive prostate cancer. Men were identified through the Alberta Cancer Registry, a North American Association of Central Cancer Registries gold-certified population-based registry that has over 95% case ascertainment. In 2000, for the cohort study, we recontacted all men from the case-control study who were still alive for consent to participate in a prospective cohort study and 830 were able to complete the first follow-up assessment. Participants were followed up for postdiagnosis measurements and mortality outcomes to 2014 through record linkages done by the Department of Cancer Surveillance and Reporting, Cancer Care Alberta, Alberta Health Services.
Eligibility criteria for the cases included being Alberta residents, English speaking, under 80 years of age, able to complete an interview, and no prior cancer diagnosis except non-melanoma skin cancers. Referring physicians were contacted for permission to contact patients. Ethics approval was obtained from the former Alberta Cancer Board (ACB) and the University of Calgary for the case-control study, and from the Alberta Cancer Research Ethics Board and the Conjoint Health Research Ethics Board at the University of Calgary for the cohort follow-up study. All participants provided written informed consent for the cohort study.
Data collection
Physical activity
Prediagnosis PA was assessed by an in-person interview using the Lifetime Total Physical Activity Questionnaire (LTPAQ) [11] within 6 months of diagnosis (mean 4.3 months; standard deviation 1.3). The LTPAQ measures the frequency, intensity, and duration of occupational, household, and recreational physical activities from childhood to diagnosis using cognitive interviewing methods and a recall calendar.
Postdiagnosis PA was assessed at three follow-up time points: 2000–2002 (average 2.5 years), 2002-2004 (average 4.7 years), and 2004–2007 (average 6.8 years). At the first follow-up, PA was assessed by in-person interview using the modified LTPAQ, which measured PA since diagnosis. At the second and third follow-ups, PA since the last follow-up was assessed using a mailed self-reported version of LTPAQ that we developed and validated [12]. The LTPAQ has shown acceptable construct validity, with moderate to strong correlations (r = 0.41–0.71) with established instruments [13], and convergent validity based on the associations between LTPAQ-derived moderate-to-vigorous physical activity and physical and mental health outcomes [14] in adult populations.
Vital status
Vital Statistics Alberta provided monthly information on vital status with cause of death from Statistics Canada to the Alberta Cancer Registry (ACR) between 2000 and 2014. The time between actual death and reporting to ACR was 3 months on average. If a participant moved to another province in Canada, the information was collected by vital status linkage with other provinces. If a participant’s vital status was unknown, annual linkage with the Alberta Health registration file allowed determining vital status at the time of the last residence in Alberta. If a participant left Alberta and his vital status was unknown by the end of the study, censoring time was the date of leaving Alberta.
Participant characteristics
At the baseline in-person interview, participants reported demographic (e.g., marital status, education, and ethnicity), personal health, prostate cancer-related (e.g., screening history, cancer conditions, surgery, and family history), and health behavior information (e.g., smoking, alcohol consumption, and diet). Anthropometric measurements were also directly assessed by the interviewers using standardized methods and calibrated scales.
Information on staging, treatments, vital status, comorbidities, new primary cancer diagnoses, and recurrences/progressions were collected by medical chart reviews. The Alberta Cancer Board treatment centres provided medical charts and, if necessary, physicians also provided charts. Highly trained Health Record Technicians from the Alberta Cancer Registry, with substantial experience, participated in designing the chart review form and coding manual, and conducted data abstractions throughout three time points (2002, 2008, and 2014). Interrater reliability for data abstraction was high, with substantial agreement for the outcomes. The technicians reviewed the entire medical chart and determined event status based on the definition in the coding manual. To obtain missing information, personalized letters were sent to physicians, or a health record technician visited physicians’ offices. In the current study, recurrence was defined as further disease, identified through prostate-specific antigen (PSA) changes and secondary treatments, following a significant disease-free period. When a patient with localized cancer undergoes therapy and subsequently maintains undetectable PSA levels for a period of time, then has an increase in PSA level, or when a doctor starts a second line treatment on a patient, these are both indicative of recurrence. The time of recurrence was stipulated as the time of starting second-line treatment or deciding not to treat. Progression was defined as when PSA did not drop to undetectable levels following radical prostatectomy, when a participant’s condition became progressively worse, when treatment did not produce positive results, or when distant metastases at diagnosis became clinically stable and then progressed again. Moreover, any patients who have metastatic prostate cancer are considered to have progressive disease. Progression is often confirmed by bone scan. The time of progression was stipulated as when increased/abnormal PSA was recorded or other diagnostic testing found progression.
Statistical analysis
Pre- and postdiagnosis recreational, vigorous, and total PA were used as exposures. Vigorous PA was defined by a metabolic equivalent value ≥ 6, and included occupational and recreational vigorous PA. We did not include prediagnosis vigorous PA in the analyses because there were only eight participants in the “No vigorous PA” group because of high levels of vigorous occupational PA before diagnosis. Total PA included non-sedentary occupational, household, and recreational activities. To create categorical PA variables, we used the median for recreational and total PA, and ‘any vs none’ for vigorous PA for adequate distribution.
For the postdiagnosis PA variables, we calculated a weighted average of PA using all available follow up assessments for each participant, taking into account the duration of the interval between each data collection time point. If a participant was censored or missed a PA assessment at a particular time point, that missing data point was excluded from the calculation. Similarly, if a participant experienced recurrence or progression, any PA assessments after the event were excluded from the calculation.
We analyzed three survival outcomes: disease-free survival, overall survival, and prostate cancer-specific disease-free survival. Disease-free survival is survival without prostate cancer recurrence or progression, second cancer, or death from any cause; prostate cancer-specific disease-free survival is survival without prostate cancer recurrence or progression, or death from prostate cancer; and overall survival is defined as survival without death from any cause.
Cox regression was used to examine the interaction of PA and treatments on survival by including interaction terms in the models between each PA variable and treatment modality. Cox regression was also used to estimate hazard ratios for each of the cancer outcomes by PA levels and treatment modalities. For those statistically significant interactions from Cox regression analyses, we created Kaplan-Meier survival curves and performed Log rank tests to compare differences in survival between PA levels by treatments.
Covariates included age at diagnosis, region, overall stage, Gleason score, number of times a PSA test was done, postdiagnosis comorbidities, PSA level, other treatments, total pack/years of smoking at diagnosis, postdiagnosis total pack/years of smoking, and time to any first prostate cancer recurrence or progression. Covariates were selected based on a combination of literature review and data-driven assessment. We initially identified potential confounders based on literature review and our previous studies, then we conducted preliminary analyses to evaluate associations between each candidate covariate and the outcomes. Nonetheless, several key clinical variables, such as cancer stage, Gleason score, PSA level, were forced into the models regardless of statistical significance based on their strong theoretical and clinical relevance. These covariates were consistent with the primary paper from the study [10] and were forced into the Cox regression models. The covariate ‘time to any first recurrence/progression of prostate cancer’ was excluded from the analyses for disease-free and prostate-specific disease-free survival to prevent multicollinearity. All statistical analyses were performed using SPSS 29 (IBM Corp, Armonk, NY, US).
Results
Participant characteristics are presented in Table 1. The median age at diagnosis was 64 years in the surgery group and 69 years in the radiation therapy group. The median BMI was 27.7 kg/m2 in both groups. Most participants had stage II disease in both surgery (80.8%) and radiation therapy group (80.3%). In the surgery group, 2.9% and 35.9% underwent radiation and hormone therapy, respectively. In the radiation therapy group, 1.9% and 75.4% underwent surgery and hormone therapy, respectively. The median postdiagnosis vigorous PA was 0.4 h/wk/year in the surgery group and 0.0 h/wk/year in the radiation therapy group.
Table 1.
Participant characteristics in the Alberta Prostate Cancer Cohort Study by Primary Treatment and Postdiagnosis Vigorous Physical Activity, 2002-2014.
| Characteristics | Surgery | Radiation therapy | ||||
|---|---|---|---|---|---|---|
| Total (N = 245) | No postdiagnosis vigorous PA (N = 100) | Any postdiagnosis vigorous PA (N = 145) | Total (N = 366) | No postdiagnosis vigorous PA (N = 214) | Any postdiagnosis vigorous PA (N = 152) | |
| Race (n, %), White | 235 (95.9) | 95 (95.0) | 140 (96.6) | 347 (94.8) | 201 (93.9) | 146 (96.1) |
| Married (n, %) | 217 (88.6) | 89 (89.0) | 128 (88.3) | 308 (84.2) | 177 (82.7) | 131 (86.2) |
| Education (n, %), high school or less | 86 (35.0) | 38 (38.0) | 48 (33.1) | 156 (42.7) | 100 (46.7) | 56 (36.8) |
| Region (n, %), urban | 151 (61.6) | 59 (59.0) | 92 (63.4) | 222 (60.7) | 127 (59.3) | 95 (62.5) |
| Overall stage (n, %) | ||||||
| II (T1/T2, N0, M0) | 198 (80.8) | 80 (80.0) | 118 (81.4) | 294 (80.3) | 170 (79.4) | 124 (81.6) |
| III (T3, N0, M0) | 35 (14.3) | 14 (14.0) | 21 (14.5) | 21 (5.7) | 11 (5.1) | 10 (6.6) |
| III or IV (T3, NX, MX) | 5 (2.0) | 3 (3.0) | 2 (1.4) | 37 (10.1) | 24 (11.2) | 13 (8.6) |
| IV | 7 (2.9) | 3 (3.0) | 4 (2.8) | 14 (3.8) | 9 (4.2) | 5 (3.3) |
| Gleason score (n, %) | ||||||
| <7 | 102 (41.6) | 45 (45.0) | 57 (39.3) | 131 (35.8) | 79 (36.9) | 52 (34.2) |
| 7 | 114 (46.5) | 40 (40.0) | 74 (51.0) | 162 (44.3) | 85 (39.7) | 77 (50.7) |
| >7 | 29 (11.8) | 15 (15.0) | 14 (9.7) | 73 (19.9) | 50 (23.4) | 23 (15.1) |
| Treatment received (n, %)a | ||||||
| Surgery | 245 (100.0) | 100 (100.0) | 145 (100.0) | 7 (1.9) | 5 (2.3) | 2 (1.3) |
| Radiation therapy | 7 (2.9) | 5 (5.0) | 2 (1.4) | 366 (100.0) | 214 (100.0) | 152 (100.0) |
| Hormone therapyb | 88 (35.9) | 34 (34.0) | 54 (37.2) | 276 (75.4) | 157 (73.4) | 119 (78.3) |
| First-degree relative history of prostate cancer (n,%) | 60 (24.5) | 26 (26.0) | 34 (23.4) | 70 (19.1) | 47 (22.0) | 23 (15.1) |
| Heart disease (n, %) | 28 (11.4) | 10 (10.0) | 18 (12.4) | 71 (19.4) | 43 (20.1) | 28 (18.4) |
| Hypertension (n, %) | 70 (28.6) | 29 (29.0) | 41 (28.3) | 113 (30.9) | 69 (32.2) | 44 (28.9) |
| Other chronic disease (n, %) | 98 (40.0) | 39 (39.0) | 59 (40.7) | 178 (48.6) | 112 (52.3) | 66 (43.4) |
| Smoking status (n, %), current smoker | 30 (12.2) | 17 (17.0) | 13 (9.0) | 54 (14.8) | 35 (16.4) | 19 (12.5) |
| Median (Q1, Q3) | Median (Q1, Q3) | |||||
| Age at diagnosis, yr | 64 (58, 68) | 64 (59, 68) | 64 (58, 68) | 69 (64, 73) | 70 (66, 74) | 66 (61, 72) |
| Body mass index, kg/m2 | 27.7 (25.3, 30.1) | 27.6 (25.2, 30.8) | 27.7 (25.5, 29.8) | 27.7 (25.5, 30.4) | 27.6 (25.5, 30.3) | 27.9 (25.7, 30.5) |
| Dietary caloric intake, kcal/d | 2060 (1719, 2495) | 2061 (1648, 2606) | 2060 (1751, 2483) | 2033 (1658, 2532) | 2016 (1652, 2539) | 2055 (1671, 2530) |
| Total lifetime alcohol intake, g/wk | 1984 (638, 4172) | 1795 (535, 4367) | 2174 (694, 4166) | 1748 (532, 4830) | 1733 (478, 4797) | 1835 (695, 5072) |
| Prediagnosis PA | ||||||
| Total PA, MET-h/wk/yr | 137.2 (98.7, 183.6) | 136.7 (97.7, 183.2) | 138.6 (99.5, 184.8) | 141.3 (97.1, 197.3) | 143.0 (100.8, 198.4) | 137.5 (89.3, 195.5) |
| Recreational PA, MET-h/wk/yr | 13.8 (7.0, 23.4) | 9.6 (5.4, 18.7) | 17.3 (9.7, 30.7) | 12.2 (6.7, 22.5) | 11.0 (5.7, 18.1) | 15.9 (8.5, 28.0) |
| Vigorous PA, h/wk/year | 4.5 (2.0, 11.0) | 4.1 (1.6, 10.0) | 5.2 (2.2, 11.7) | 5.1 (1.7, 10.9) | 5.8 (1.7, 11.7) | 3.8 (1.7, 10.1) |
| Postdiagnosis PA | ||||||
| Total PA, MET-h/wk/yr | 89.6 (55.8, 135.4) | 64.1 (42.2, 96.4) | 116.6 (69.5, 164.3) | 72.6 (38.5, 116.0) | 57.1 (31.6, 98.2) | 94.4 (61.4, 147.7) |
| Recreational PA, MET-h/wk/yr | 16.3 (6.0,. 32.4) | 7.7 (2.6, 17.1) | 22.8 (12.7, 43.6) | 12.7 (3.3, 24.9) | 7.8 (2.0, 18.5) | 19.9 (9.1, 36.8) |
| Vigorous PA, h/wk/year | 0.4 (0.0, 3.4) | 0.0 (0.0, 0.0) | 2.6 (0.9, 6.1) | 0.0 (0.0, 1.5) | 0.0 (0.0, 0.0) | 2.1 (0.7, 4.9) |
Values are n (%) or median (quartile 1, quartile 3).
MET metabolic equivalent, PCa prostate cancer, Q quartile, PA physical activity.
aParticipants could have more than one treatment.
bIncluded bilateral orchiectomy, luteinising hormone-releasing hormone agonists, nonsteroidal antiandrogens, steroidal antiandrogens.
Associations between different PA variables and disease-free survival by treatment modality are reported in Table 2. There was a statistically significant interaction between postdiagnosis vigorous PA with surgery (p < 0.001) and radiation therapy (p = 0.003). Among participants who underwent surgery (n = 245), those who engaged in any postdiagnosis vigorous PA were approximately 60% less likely to have a recurrence, progression, second cancer, or death compared to those who did not engage in any postdiagnosis vigorous PA (HR = 0.39, 95% CI = 0.27–0.57). Among participants who did not undergo surgery (n = 585), there was no statistically significant difference in disease-free survival based on postdiagnosis vigorous PA (HR = 1.00, 95% CI = 0.82–1.22). Conversely, in the group who did not receive radiation therapy (n = 464), participants who engaged in any postdiagnosis vigorous PA were approximately 35% less likely to have a recurrence, progression, second cancer, or death compared to those who did not engage in any postdiagnosis vigorous PA (HR = 0.64, 95% CI = 0.50–0.83). In the group who received radiation therapy (n = 366), there was no statistically significant difference in disease-free survival based on vigorous PA (HR = 1.14, 95% CI = 0.88–1.47).
Table 2.
Associations of prediagnosis and postdiagnosis physical activity with disease-free survival in prostate cancer patients by treatment modality.
| Surgery HR (95% CI) | Hormone therapy HR (95% CI) | Radiation therapy HR (95% CI) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| P for Interaction | No (N = 585) | Yes (N = 245) | P for Interaction | No (N = 303) | Yes (N = 527) | P for Interaction | No (N = 464) | Yes (N = 366) | |
| Prediagnosis recreational PA (≥13 vs <13 MET-hrs/wk/yr) | 0.74 | 0.90 (0.74–1.09) | 0.80 (0.55–1.18) | 0.78 | 0.93 (0.67–1.30) | 0.90 (0.73–1.11) | 0.45 | 0.80 (0.63–1.02) | 0.98 (0.76–1.26) |
| Prediagnosis total PA (≥145 vs <145 MET-hrs/wk/yr) | 0.88 | 1.02 (0.84–1.24) | 0.89 (0.61–1.30) | 0.99 | 1.17 (0.85–1.60) | 1.00 (0.82–1.23) | 0.37 | 1.11 (0.88–1.41) | 0.89 (0.68–1.14) |
| Postdiagnosis recreational PA (≥13 vs <13 MET-hrs/wk/yr) | 0.21 | 0.84 (0.69–1.02) | 0.62 (0.43–0.90) | 0.46 | 0.67 (0.49–0.92) | 0.83 (0.68–1.03) | 0.36 | 0.73 (0.57–0.93) | 0.87 (0.67–1.11) |
| Postdiagnosis vigorous PA (Any vs none) | <0.001 | 1.00 (0.82–1.22); p = 0.99 | 0.39 (0.27–0.57); p<0.001 | 0.08 | 0.61 (0.44-0.85) | 0.88 (0.71–1.09) | 0.003 | 0.64 (0.50-0.83); p = 0.001 | 1.14 (0.88–1.47); p = 0.31 |
| Postdiagnosis total PA (≥73 vs <73 MET-hrs/wk/yr) | 0.49 | 0.83 (0.68–1.02) | 0.64 (0.43-0.95) | 0.99 | 0.96 (0.70–1.31) | 0.76 (0.61–0.95) | 0.29 | 0.76 (0.59–0.98) | 0.88 (0.68–1.13) |
Adjusted for age at diagnosis, stage, Gleason score, PSA level, region, number of times had PSA test done, total pack-years of smoking at diagnosis, postdiagnosis total pack-years of smoking, postdiagnosis comorbidity, and the other two treatments.
PA physical activity, HR hazard ratio.
Associations between different PA variables and prostate cancer specific disease-free survival by treatment modality are reported in Table 3. Postdiagnosis recreational (p = 0.040) and vigorous PA (p < 0.001) demonstrated statistically significant interactions with surgery. In participants who received surgery, those who engaged in more than 13 MET-hrs/wk/yr of postdiagnosis recreational PA or any vigorous PA were 54% (HR = 0.46, 95% CI = 0.29–0.73) and 62% (HR = 0.38, 95% CI = 0.23–0.61) less likely to have a prostate cancer-related event, respectively. No statistically significant differences based on PA levels was found in the no surgery group. Postdiagnosis vigorous PA also demonstrated a statistically significant interaction with radiation therapy (p = 0.012). Participants who received radiation therapy experienced no difference in prostate cancer-specific disease-free survival based on vigorous PA level. Conversely, in the group who did not receive radiation therapy, engaging in postdiagnosis vigorous PA was associated with a 29% lower likelihood of having a prostate cancer-related event (HR = 0.71, 95% CI = 0.52–0.97).
Table 3.
Associations of prediagnosis and postdiagnosis physical activity with prostate cancer-specific disease-free survival in prostate cancer patients by treatment modality.
| Surgery HR (95% CI) | Hormone therapy HR (95% CI) | Radiation therapy HR (95% CI) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| P for Interaction | No (N = 585) | Yes (N = 245) | P for Interaction | No (N = 303) | Yes (N = 527) | P for Interaction | No (N = 464) | Yes (N = 366) | |
| Prediagnosis recreational PA (≥13 vs <13 MET-hrs/wk/yr) | 0.81 | 0.84 (0.65–1.08) | 0.74 (0.46–1.21) | 0.98 | 0.96 (0.60–1.55) | 0.86 (0.67–1.11) | 0.39 | 0.75 (0.56–1.02) | 0.98 (0.70–1.36) |
| Prediagnosis total PA (≥145 vs <145 MET-hrs/wk/yr) | 0.52 | 0.95 (0.74–1.23) | 0.92 (0.57–1.49) | 0.88 | 1.15 (0.75–1.79) | 0.96 (0.74–1.25) | 0.47 | 1.11 (0.83–1.50) | 0.81 (0.58–1.15) |
| Postdiagnosis recreational PA (≥13 vs <13 MET-hrs/wk/yr) | 0.040 | 0.88 (0.69–1.14); p=0.34 | 0.46 (0.29–0.73); p=0.001 | 0.49 | 0.72 (0.46–1.12) | 0.79 (0.61–1.03) | 0.12 | 0.65 (0.48–0.89) | 0.95 (0.68–1.31) |
| Postdiagnosis vigorous PA (Any vs none) | <0.001 | 1.12 (0.87–1.44); p = 0.39 | 0.38 (0.23–0.61); p<0.001 | 0.09 | 0.66 (0.43–1.03) | 0.94 (0.72–1.23) | 0.012 | 0.71 (0.52–0.97); p=0.031 | 1.32 (0.95–1.84); p = 0.10 |
| Postdiagnosis total PA (≥73 vs <73 MET-hrs/wk/yr) | 0.13 | 0.97 (0.75–1.26) | 0.48 (0.29–0.81) | 0.31 | 0.92 (0.58–1.44) | 0.89 (0.68–1.16) | 0.23 | 0.79 (0.57–1.08) | 1.03 (0.73–1.43) |
Adjusted for age at diagnosis, stage, Gleason score, PSA level, region, number of times had PSA test done, total pack-years of smoking at diagnosis, postdiagnosis total pack-years of smoking, postdiagnosis comorbidity, and the other two treatments.
PA physical activity, HR hazard ratio.
Associations between different PA variables and overall survival by treatment modality are reported in Supplementary Table 1. Prediagnosis total PA had a borderline statistically significant interaction with hormone therapy (p = 0.049), however, no other significant associations were identified.
Kaplan-Meier survival curves illustrating these significant interactions are shown in Fig. 1. Disease-free survival was significantly different between any versus no postdiagnosis vigorous PA groups among participants who received surgery (57.9% vs 32.0%, p < 0.001) but not among participants who did not receive surgery (23.2% vs 17.1%, p = 0.07) (Fig. 1). Conversely, disease-free survival was significantly different by postdiagnosis vigorous PA in the no radiation therapy group (46.2% vs 17.7%, p < 0.001) but not in the radiation therapy group (23.0% vs 23.4%, p = 0.80) (Fig. 1).
Fig. 1. Unadjusted associations between postdiagnosis vigorous physical activity and disease-free survival in prostate cancer patients by treatment modality.
Panel a no sugery, b sugery, c no radiation therapy, and (d) radiation therapy.
For prostate cancer specific disease-free survival, there were significant differences by postdiagnosis recreational (73.2% vs 58.3%, p = 0.010) and postdiagnosis vigorous PA (73.8% vs 57.0%, p = 0.002) in the surgery group but not in the no surgery group (Figs. 2 and 3, respectively). Conversely, no difference was identified in the radiation therapy group (50.7% vs 58.4%, p = 0.31) but there was a significant difference by postdiagnosis vigorous PA in the no radiation therapy group (63.3% vs 50.6%, p < 0.001) (Fig. 3).
Fig. 2. Unadjusted associations between postdiagnosis recreational physical activity and prostate cancer-specific disease-free survival by surgery.
Panel a no surgery and b surgery.
Fig. 3. Unadjusted associations between postdiagnosis vigorous physical activity and prostate cancer-specific disease-free survival by treatment modality.
Panel a no surgery, b surgery, c no radiation therapy, and d radiation therapy.
Discussion
Few observational studies have examined whether the association of PA with cancer outcomes is altered by previous, current, and/or subsequent cancer treatments. In our secondary analysis of a prospective cohort study, we observed several associations of postdiagnosis PA with prostate cancer outcomes that differed depending on the primary treatment modality. Specifically, we found that postdiagnosis PA, especially vigorous PA, was associated with improved prostate cancer outcomes in patients treated with surgery but not with radiation therapy. Given that our postdiagnosis PA measure largely covered the time after primary treatment, our results suggest that vigorous PA may serve as an adjuvant monotherapy or adjuvant concurrent therapy with androgen deprivation therapy (ADT) after surgical resection for prostate cancer.
Preclinical rodent models have shown that PA reduces the growth and spread of primary or metastatic cancer [6] including prostate cancer [15]. Few preclinical studies, however, have examined the role of PA after surgical excision or irradiation of the primary tumor with or without concurrent ADT. One recent study suggested that exercise significantly increased metastasis-free survival in a breast cancer model but only in mice treated with surgery [16], a finding consistent with our study. The authors noted that surgeries elicit a stress response and cause complications that promote postoperative metastatic spread and/or disease recurrence by activating dormant micrometastases or residual cancer cells at the surgical site. Consequently, exercise may play a larger role in improving cancer outcomes in patients exposed to surgery. Nevertheless, it is unknown how exercise may interact with standard prostate cancer treatments in preclinical models. Future preclinical studies should examine if the sequencing and combination of exercise in relation to other prostate cancer treatments alters its anti-cancer effects.
In terms of observational studies, Friedenreich et al. [17] performed a systematic review and meta-analysis of studies examining PA and cancer outcomes that included a separate meta-analyses for prostate cancer studies. The results of nine prostate cancer studies showed that higher prediagnosis PA was not statistically significantly associated with a lower risk of cancer-specific mortality (HR = 0.90; 95% CI = 0.75–1.08). Conversely, the results of four studies showed that higher postdiagnosis PA was associated with a lower risk of prostate cancer-specific mortality (HR = 0.70; 95% CI = 0.55–0.90). Across all studies, subgroup analyses were reported by demographic/health variables (e.g., sex, body mass index, menopausal status) but not by cancer treatments.
Observational studies specifically focused on prostate cancer [10, 18–21] have also rarely examined whether the association between postdiagnosis PA and cancer outcomes varies by cancer treatment modality. In a prospective cohort study among 2,705 men diagnosed with nonmetastatic prostate cancer who survived at least 4 years, Kenfield et al. [19] reported that there were no significant interactions between PA and primary treatment for mortality outcomes. Few details of the analyses were provided, however, and it is unclear how primary treatments were categorized and whether all PA exposures were analyzed for possible interactions.
In one of the few studies to perform detailed subgroup analyses based on cancer treatments in any cancer type, Lee et al. [22] examined the associations between PA performed after surgery only, after surgery plus chemotherapy and/or radiotherapy, and after chemotherapy with or without radiotherapy (and no surgery) with cancer outcomes in 43,596 colorectal cancer survivors from the Korean National Health Insurance Service database. In stratified analyses by treatment group, PA after treatments was associated with a lower risk of mortality in colon cancer patients who had surgery only (HR = 0.75; 95% CI, 0.65–0.87) or surgery plus chemotherapy and/or radiotherapy (HR = 0.84; 95% CI, 0.73–0.97). There was a lower risk of mortality in patients who did not receive surgery but received chemotherapy with or without radiotherapy (HR = 0.74; 95% CI, 0.46–1.19). Although formal tests of interactions were not reported, these data suggest that PA may serve as an adjuvant monotherapy or concurrent adjuvant therapy after surgical resection for colorectal cancer, similar to our findings.
There are several possible explanations for why PA may be associated with a lower risk of prostate cancer events after surgery but not after radiation therapy. Radical prostatectomy involves removal of the entire prostate gland, seminal vesicles, nearby tissue, and sometimes nearby lymph nodes depending on risk features [23]. Radical prostatectomy is appropriate for any patient whose cancer appears clinically localized to the prostate and is generally reserved for younger and healthier patients whose life expectancy is 10 years or more [23]. Conversely, radiation therapy involves external beams of radiation focused on the prostate only (gland and seminal vesicles) or the whole pelvic region which includes the prostate gland, seminal vesicles, and the pelvic lymph nodes (bilateral common iliac, external iliac, internal iliac, presacral, and obturator) [24]. Radiation therapy tends to be offered to older patients or those with significant comorbidities who may not be ideal candidates for surgery. In our cohort, the mean age of participants who underwent radical prostatectomy was younger than those who received radiation therapy. This baseline difference in age might influence the higher rate of events in the radiation therapy group.
Additionally, if whole pelvic radiation therapy was provided to higher risk participants in our study, this treatment may have reduced the local recurrence rates in these patients since up to 30% of disease recurrences are pelvic rather than systemic [25]. It is possible that the higher potential for micrometastases in the pelvic region after surgical excision may have provided a greater opportunity for PA to reduce local recurrences and distant micrometastases, which are also more common after surgery [26]. Conversely, it is also possible that radiation therapy altered the biology of any remaining local cancer cells making them more resistant to exercise. Unfortunately, we did not collect detailed data on whether surgery included pelvic lymph node dissection or whether the radiation therapy included whole pelvic irradiation. Future studies should collect more detailed data on treatments received (e.g., type, dose, field, schedule).
Another possible explanation for these findings is that patients treated with surgery were less likely to receive ADT compared to patients treated with radiation therapy (35% versus 75%; see Table 1). The lower rate of ADT treatment in surgical patients may have resulted in higher rates of untreated micrometastases that were amenable to PA effects. In fact, there were borderline statistically significant interactions (p values < 0.10) suggesting that patients not treated with ADT benefited from postdiagnosis vigorous PA more than patients treated with ADT for both disease-free survival (Table 2) and prostate cancer-specific disease-free survival (Table 3). Although we adjusted for other treatments in all our analyses, it is possible that radiation therapy and ADT combined provided a synergistic benefit [9] that limited the opportunity for PA to demonstrate any treatment effects.
Our secondary analysis has several important limitations and strengths. One limitation is that we were not able to analyze treatment combinations because of a modest sample size. A second limitation is that our postdiagnosis PA measures covered the time from diagnosis to follow-up and, therefore, we were not able to disentangle PA performed immediately before treatment, during or between treatments (e.g., radiation therapy, androgen deprivation therapy), or immediately after treatments. Consequently, we were not able to identify the optimal timing of PA in relation to surgery or radiation therapy beyond the general assessment of prediagnosis and postdiagnosis. Although more challenging, future studies should assess PA in relation to clinically meaningful cancer treatment-related time periods (e.g., before, during, between, and after) rather than in relation to diagnosis (e.g., time points before diagnosis or since diagnosis) or arbitrary time periods (e.g., past 6 months, past 2 years). Finally, our cohort was diagnosed and treated in the late 1990s and does not reflect current management of prostate cancer.
In conclusion, our secondary analysis found that postdiagnosis PA, especially vigorous PA, was associated with improved prostate cancer outcomes in patients treated with surgery but not with radiation therapy. Further preclinical and observational studies are needed to confirm this finding and identify possible mechanisms and explanations. If confirmed, our findings may provide a more precise clinical target for definitive trials of PA as a prostate cancer treatment as well as for PA promotion efforts in newly diagnosed prostate cancer patients.
Supplementary information
Author contributions
Conceptualization: KYA, JYJ, CMF, and KSC, Data curation: QW, Formal analysis: KYA, JYJ, QW, and KSC, Funding acquisition: CMF and KSC, Investigation: KYA, FZA, QW, CMF, and KSC, Methodology: KYA, JYJ, QW, CMF and KSC, Supervision: CMF and KSC, Writing-original draft: KYA, FZA, and KSC, Writing-review and editing: KYA, JYJ, FZA, QW, CMF, and KSC.
Funding
This work was supported by the Canadian Institutes for Cancer Research (MOP-67217); the National Cancer Institute of Canada with funds from the Canadian Cancer Society (011004); and the Alberta Cancer Board-Research Initiative Program (4570). Christine M. Friedenreich was supported by career awards from Alberta Innovates-Health Solutions and the Alberta Cancer Foundation. Kerry S. Courneya was supported by the Canada Research Chair Program and a Foundation Grant from the Canadian Institutes of Health Research.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
Competing interests
The authors declare no competing interests.
Ethics approval and consent to participate
Ethics approval was obtained from the former Alberta Cancer Board (ACB) and the University of Calgary for the case-control study, and from the Alberta Cancer Research Ethics Board and the Conjoint Health Research Ethics Board at the University of Calgary for the cohort follow-up study. All participants provided written informed consent for the cohort study. This study was performed in accordance with the Declaration of Helsinki.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
The online version contains supplementary material available at 10.1038/s41416-025-03123-0.
References
- 1.Courneya KS, Friedenreich CM. Designing, analyzing, and interpreting observational studies of physical activity and cancer outcomes from a clinical oncology perspective. Front Oncol. 2023;13:1098278. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Courneya KS, Booth, CM. Exercise as a cancer treatment: a clinical oncology framework for exercise oncology research. Front Oncol. 2022;12:957135. [DOI] [PMC free article] [PubMed]
- 3.Courneya KS, McNeely ML, Booth CM, Friedenreich CM. An integrated framework for the study of exercise across the postdiagnosis cancer continuum. Front Oncol. 2024;14:1432899. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Masuda H, Harano K, Miura S, Wang Y, Hirota Y, Harada O, et al. Changes in triple-negative breast cancer molecular subtypes in patients without pathologic complete response after neoadjuvant systemic chemotherapy. JCO Precis Oncol. 2022;6:e2000368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Topham JT, O’Callaghan CJ, Feilotter H, Kennecke HF, Lee YS, Li W, et al. Circulating tumor DNA identifies diverse landscape of acquired resistance to anti-epidermal growth factor receptor therapy in metastatic colorectal cancer. J Clin Oncol. 2023;41:485–96. JCO2200364. [DOI] [PMC free article] [PubMed]
- 6.Ashcraft KA, Peace RM, Betof AS, Dewhirst MW, Jones LW. Efficacy and mechanisms of aerobic exercise on cancer initiation, progression, and metastasis: a critical systematic review of in vivo preclinical data. Cancer Res. 2016;76:4032–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yang L, Morielli AR, Heer E, Kirkham AA, Cheung WY, Usmani N, et al. Effects of exercise on cancer treatment efficacy: a systematic review of preclinical and clinical studies. Cancer Res. 2021;81:4889–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Francis S, Orton A, Stoddard G, Tao R, Hitchcock YJ, Akerley W, et al. Sequencing of postoperative radiotherapy and chemotherapy for locally advanced or incompletely resected non-small-cell lung cancer. J Clin Oncol. 2018;36:333–41. [DOI] [PubMed] [Google Scholar]
- 9.Ma TM, Sun Y, Malone S, Roach M, 3rd, Dearnaley D, Pisansky TM, et al. Sequencing of androgen-deprivation therapy of short duration with radiotherapy for nonmetastatic prostate cancer (SANDSTORM): a pooled analysis of 12 randomized trials. J Clin Oncol. 2023;41:881–92. JCO2200970. [DOI] [PMC free article] [PubMed]
- 10.Friedenreich CM, Wang Q, Neilson HK, Kopciuk KA, McGregor SE, Courneya KS. Physical activity and survival after prostate cancer. Eur Urol. 2016;70:576–85. [DOI] [PubMed] [Google Scholar]
- 11.Friedenreich CM, Courneya KS, Bryant HE. The lifetime total physical activity questionnaire: development and reliability. Med Sci Sports Exerc. 1998;30:266–74. [DOI] [PubMed] [Google Scholar]
- 12.Friedenreich CM, Courneya KS, Neilson HK, Matthews CE, Willis G, Irwin M, et al. Reliability and validity of the past year total physical activity questionnaire. Am J Epidemiol. 2006;163:959–70. [DOI] [PubMed] [Google Scholar]
- 13.De Vera MA, Ratzlaff C, Doerfling P, Kopec J. Reliability and validity of an internet-based questionnaire measuring lifetime physical activity. Am J Epidemiol. 2010;172:1190–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Torres ER. Validation of the lifetime total physical activity questionnaire (LTPAQ) in midlife and older adults with a history of late-onset depression. Arch Psychiatr Nurs. 2018;32:580–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Webber M, Li M, Cai C, Zou K. Effects of exercise training on prostate cancer: current evidence and potential molecular mechanisms. Adv Exercise and Health Sci. 2024;3:160–169.
- 16.Stagaard R, Jensen A, Schauer T, Bay ML, Tavanez AR, Wielsoe S, et al. Exercise boost after surgery improves survival in model of metastatic breast cancer. Front Immunol. 2025;16:1533798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Friedenreich CM, Stone CR, Cheung WY, Hayes SC. Physical activity and mortality in cancer survivors: a systematic review and meta-analysis. JNCI Cancer Spectr. 2020;4:pkz080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bonn SE, Sjolander A, Lagerros YT, Wiklund F, Stattin P, Holmberg E, et al. Physical activity and survival among men diagnosed with prostate cancer. Cancer Epidemiol Biomark Prev. 2015;24:57–64. [DOI] [PubMed] [Google Scholar]
- 19.Kenfield SA, Stampfer MJ, Giovannucci E, Chan JM. Physical activity and survival after prostate cancer diagnosis in the health professionals follow-up study. J Clin Oncol. 2011;29:726–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Richman EL, Kenfield SA, Stampfer MJ, Paciorek A, Carroll PR, Chan JM. Physical activity after diagnosis and risk of prostate cancer progression: data from the Cancer of the Prostate Strategic Urologic Research Endeavour. Cancer Res. 2011;71:3889–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wang Y, Jacobs EJ, Gapstur SM, Maliniak ML, Gansler T, McCullough ML, et al. Recreational physical activity in relation to prostate cancer-specific mortality among men with nonmetastatic prostate cancer. Eur Urol. 2017;72:931–9. [DOI] [PubMed] [Google Scholar]
- 22.Lee M, Lee Y, Jang D, Shin A. Physical activity after colorectal cancer diagnosis and mortality in a nationwide retrospective cohort study. Cancers. 2021;13:4804. [DOI] [PMC free article] [PubMed]
- 23.Schaeffer EM, Srinivas S, Adra N, An Y, Bitting R, Chapin B, et al. NCCN clinical practice guidelines in oncology: prostate cancer. Version 4. 2024. Plymouth Meeting (PA): National Comprehensive Cancer Network; 2024. [cited 2024 Nov 25]. Available from: https://www.nccn.org/guidelines/guidelines-detail?category=1&id=1459
- 24.Rodrigues M, Wong WW, Boorjian SA. Clinically localized prostate cancer: AUA/ASTRO guideline 2022. Linthicum (MD): American Urological Association; 2022. Available from: https://www.auanet.org/guidelines-and-quality/guidelines/clinically-localized-prostate-cancer-aua/astro-guideline-2022
- 25.Nini A, Gandaglia G, Fossati N, Suardi N, Cucchiara V, Dell’Oglio P, et al. Patterns of clinical recurrence of node-positive prostate cancer and impact on long-term survival. Eur Urol. 2015;68:777–84. [DOI] [PubMed] [Google Scholar]
- 26.Roy S, Sun Y, Eastham JA, Gleave M, Beltran H, Kishan AU, et al. Radical prostatectomy (RP) versus radiotherapy (RT) in high-risk prostate cancer (HR-PCa): emulated randomized comparison with individual patient data (IPD) from two phase III randomized trials (RCTs). Am Soc Clin Oncol. 2025;43:309
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article will be shared on reasonable request to the corresponding author.