Chris Pyke

RACQ International Women's Day Fun Run

Mater, QUT, Wesley, QIMR and UWA Research shows: Exercise can improve Survival in Breast Cancer

Breast Cancer Res Treat (2018) 167:505–514 https://doi.org/10.1007/s10549-017-4541-9

CLINICAL TRIAL

Exercise following breast cancer: exploratory survival analyses of two randomised, controlled trials

S. C. Hayes1 · M. L. Steele1 · R. R. Spence1 · L. Gordon2 · D. Battistutta1 · J. Bashford3 · C. Pyke4 · C. Saunders5 · E. Eakin6

Received: 11 October 2017 / Accepted: 14 October 2017 / Published online: 23 October 2017 © Springer Science+Business Media, LLC 2017

Abstract

Purpose The Exercise for Health trials were randomised, controlled trials designed to evaluate an 8-month pragmatic exercise intervention, commencing 6 weeks post-surgery for women with newly diagnosed breast cancer residing in urban or rural/regional Australia. For these exploratory analyses, the primary and secondary outcomes were overall survival (OS) and disease-free survival (DFS), respectively. Methods Consenting urban- (n = 194) and rural/regional- residing women (n = 143) were randomised to exercise (intervention delivered face-to-face or by telephone) or usual care. Cox proportional hazards models were used to esti- mate hazard ratios (HRs) and 95% confidence intervals (CI) for survival outcomes (exercise group, n = 207, 65% urban women; usual care group, n = 130, 46% urban women). Results After a median follow-up of 8.3 years, there were 11 (5.3%) deaths in the exercise group compared with 15 (11.5%) deaths in the usual care group (OS HR for the

exercise group: 0.45, 95% CI 0.20–0.96; p = 0.04). DFS events for the exercise versus usual care group were 25 (12.1%) and 23 (17.7%), respectively (HR: 0.66, 95% CI 0.38–1.17; p = 0.16). HRs for OS favoured exercise irrespec- tive of age, body mass index, stage of disease, intervention compliance, and physical activity levels at 12 months post- diagnosis, although were stronger (p < 0.05) for younger women, women with stage II + disease, women with 1 + comorbidity at time of diagnosis, higher intervention compliance and for those who met national physical activity guidelines at 12 months post-diagnosis.

Conclusion An exercise intervention delivered during and beyond treatment for breast cancer, and that was designed to cater for all women irrespective of place of residence and access to health services, has clear potential to benefit sur- vival. Trial numbers: ACT RN: 012606000233527; ACT RN: 12609000809235.

Keywords Breast cancer · Exercise · Physical activity · Morbidity · Survival

Introduction

Exercise is increasingly recognised as a central element in the management of breast cancer, with strong evidence for the role of exercise in reducing number and severity of treatment-related symptoms (e.g. fatigue, pain, lymphoe- dema, cognitive impairment, sleep disturbance), as well as improving physical and psychological function during and after treatment [5, 16]. Over the past decade, epide- miological evidence has also demonstrated a consistent and positive relationship between physical activity and survival, with significant risk reductions for all-cause and breast cancer-related death found among those who met national

Electronic supplementary material The online version of this article (doi:10.1007/s10549-017-4541-9) contains supplementary material, which is available to authorised users.

* S. C. Hayes sc.hayes@qut.edu.au

1

3 4 5

Wesley Hospital, Brisbane, Australia
Mater Public and Private Hospital, Brisbane, Australia University of Western Australia, Perth, Australia

Institute of Health and Biomedical Innovation, School
of Public Health and Social Work, Queensland University of Technology, Brisbane, Australia

2
Australia

QIMR Berghofer Medical Research Institute, Brisbane,

6 School of Public Health, Cancer Prevention Research Centre, The University of Queensland, Brisbane, Australia

Vol.:(0123456789)

506

recommendations for physical activity compared with those who did not (hazard ratio [HR]: 0.54, 95% confidence inter- val [CI] 0.38–0.76, p < 0.01; and 0.67, 95% CI 0.50–0.90, p < 0.01, respectively) [12, 13]. While the epidemiological evidence in support of physical activity is provocative, these findings need to be confirmed by prospective, randomised, controlled trials to establish cause and effect.

The ‘Exercise for Health’ (EfH) trials were two ran- domised, controlled, effectiveness trials comparing the impact of an eight-month, pragmatic, exercise intervention on function, treatment-related side effects and quality of life following breast cancer, compared with usual care. The design and implementation of the trials were coordinated, with one trial involving urban residents [10], while the other included women from rural/regional areas of Queensland, Australia [8]. Main outcomes from the EfH trials have been published previously, showing significant improvements for fitness, fatigue and quality of life [8, 11], and more recently the interventions were shown to be cost-effective [9]. While these trials were neither designed nor powered for analyses of survival outcomes, together they provide an ideal and well-designed opportunity to explore survival outcomes fol- lowing participation in an exercise intervention with poten- tial for wide-scale translation.

Methods

Study design and sample

Details regarding trial registration (ACT RN: 012606000233527; ACT RN: 12609000809235), ethical approval, consent process and trial protocol have been pub- lished previously [8, 10]. Figure 1 presents a CONSORT flow diagram for the two trials; 194 and 143 women repre- sentative of the urban and rural/regional breast cancer popu- lation, respectively, were eligible and agreed to participate. Randomisation involved a computer-generated, unblocked sequence of random numbers (3 groups for urban trial: n = 67 for face-to-face exercise intervention, n = 67 for telephone-delivered exercise intervention, n = 60 for usual care; 2 groups for rural trial: n = 73 for telephone delivery of exercise intervention, n = 70 for usual care).

Exercise intervention

The target weekly exercise dose for women in the inter- vention was 180 + min of aerobic-based and resistance- based, moderate-intensity exercise, to be accumulated on at least 4 days per week. Commencing at 6 weeks post- surgery, women in the intervention had 16 scheduled ses- sions with an Exercise Physiologist. Thus, the majority of the intervention was unsupervised, irrespective of mode

Breast Cancer Res Treat (2018) 167:505–514

of contact (face-to-face or telephone-delivered). Weekly exercise prescription (covering type, duration, frequency and intensity) accommodated the target weekly exercise dose, as well as patient preferences and circumstances. Because outcomes were similar among urban women in the face-to-face and telephone-delivered intervention groups, these were combined as one exercise group for analyses purposes.

Outcomes of interest

The primary and secondary outcomes of interest for the current exploratory analyses were overall survival (OS) and disease-free survival (DFS), respectively. A search of the Queensland Cancer Registry was undertaken to determine vital status and breast cancer experience of all 337 partici- pants (Fig. 1). Data abstracted included death, date of death, cause of death, breast cancer recurrence, date of recurrence, and diagnosis and date of a new primary breast cancer, up to 31 December, 2015. OS was calculated as the time from first breast cancer surgery until death from any cause and DFS was censored at death, breast cancer recurrence, or diagnosis of a new primary breast cancer.

Demographic, disease and treatment data

Baseline data (collected prospectively via participant-admin- istered questionnaires and medical chart review) included age, body mass index, presence of comorbidities, cancer stage, lymph node status, oestrogen and progesterone recep- tor status, human epidermal growth factor receptor-2 status, extent and laterality of surgery.

Statistical analysis

For all analyses, we adhered to the intention-to-treat principle. T-tests and Chi-squared tests for independ- ent samples were used to assess the statistical associa- tion between random group allocation (usual care versus exercise intervention) and baseline demographic, clinical and treatment characteristics. For time-to-event analy- ses, patients were censored if still alive at 31 December, 2015. Median follow-up time was calculated by the reverse Kaplan–Meier approach [15]. OS and DFS were estimated by the Kaplan–Meier method (since more than 50% of the cohort was alive at follow-up, median survival is not inform- ative and was therefore not calculated). A Cox proportional hazards model was used to estimate HRs and 95% CIs for survival outcomes. Considering the coordinated design of the urban and rural EfH trials, the relatively small sample sizes available, the fairly balanced nature of the groups [8, 10] and the exploratory nature of these analyses, unadjusted statistical tests using pooled data from the two trials were

13

Breast Cancer Res Treat (2018) 167:505–514 507

Fig. 1 Consort flow diagram of the urban and rural Exercise for Health trials

chosen for the primary evaluations because results would reflect a more representative sample of women with breast cancer. To ensure that this approach was sufficiently con- servative, secondary analyses involved Cox proportional hazards models: (1) using pooled data, adjusting for trial (urban/rural), age (continuous), body mass index (continu- ous), presence of comorbidities at baseline (continuous) and

stage of disease (stage I or II +), (2) including an interaction term for trial (urban/rural) by study group (exercise interven- tion/usual care), and comparing results of this model with those of the unadjusted pooled model to determine the pres- ence of interaction, and (3) stratifying analyses for urban and rural women. The proportionality of hazards assumption was checked for all models through numerical and graphical

13

508

inspection of Schoenfeld residuals. Further exploratory analyses were conducted for defined subgroups, identified a priori for prognostic characteristics including age, body mass index, presence of baseline comorbidities and dis- ease stage, as well as for intervention engagement variables including intervention compliance, physical activity levels at 12 months post-diagnosis and mode of intervention deliv- ery (face-to-face versus telephone-delivered, using data from the urban EfH trial only), with OS and DFS. Intervention compliance was categorised according to whether women received > 75% of scheduled intervention contacts with their allocated exercise physiologist (yes/no). Levels of weekly physical activity at 12 months post-diagnosis were catego- rised as 1. 0, < 150 min and 150 + min; and 2. 0, < 150 min (which may or may not have included vigorous intensity activity); 150 + min of moderate-intensity activity (that is, no vigorous activity undertaken) and 150 + min activity, which included at least 10 min of vigorous intensity activ- ity. Findings from subgroup analyses are presented as for- est plots. p values are two-sided and considered statistically significant if less than 0.05. All analyses were performed in R version 3.3.1, using the survival package [17].

Results

Patient characteristics, intervention adherence and follow‐up

Table 1 presents participant characteristics, pooled and separately for urban- and rural/regional-residing women. Age of those in the usual care group was on average 2 years older than the intervention group. There were also higher proportions of women with oestrogen/progesterone-positive receptor status tumours (p = 0.09) and women receiving chemotherapy (p = 0.08) in the pooled exercise group com- pared with the pooled usual care group. Other characteristics were generally well balanced between the groups. Overall median follow-up time was 8.3 years (inter-quartile range, IQR = 8.0–8.7 years), and was similar for urban and rural intervention and usual care women.

Primary outcome—overall survival

By 31 December 2015, 26 patients (7.7% of the sample) had died from all causes, with 10 women in each group dying from breast cancer (Table 2). Pooled OS was 94.7 and 88.5% for the exercise and usual care groups, respectively (when censored at 7-year follow-up, OS was 96.1 and 89.2%, respectively). Randomisation to the exercise intervention more than halved the hazard of dying during the follow-up period; this result was statistically significant (unadjusted HR: 0.45, 95% CI 0.20–0.97; p = 0.04; Fig. 2a). Similar

Breast Cancer Res Treat (2018) 167:505–514

results were observed following adjustment for study, age, body mass index, presence of comorbidities and disease stage (adjusted HR: 0.44, 95% CI 0.19–0.98; p = 0.046). A formal evaluation of the interaction between study and group assignment for OS was not statistically significant (p = 0.43). Nevertheless, stratified analyses showed that for the urban women, OS findings were somewhat better than the pooled results (96.3% versus 88.3% for exercise versus usual care; unadjusted HR = 0.31, 95% CI = 0.10–0.97; p = 0.04; Fig. 2b), while findings for the rural women remained in favour of exercise but were attenuated and no longer statistically significant (91.8% vs.88.6%; HR = 0.69, 95% CI = 0.24–2.00; p = 0.49; Fig. 2c).

Secondary outcome—disease‐free survival

During the total follow-up period, 20 women were diagnosed with recurrent breast cancer (5.9%) and 13 women with a new primary breast cancer (3.9%; Table 2). Eleven of these women also died during the follow-up period. The propor- tion of women dying from any cause or experiencing subse- quent breast cancer was lower for the exercise group (12.1%) than the usual care group (17.7%), although this difference was not statistically significant (unadjusted HR: 0.66, 95% CI 0.38–1.17, p = 0.16; Fig. 3a; adjusted HR: 0.65, 95% CI 0.36–1.17; p = 0.15). When data were censored at 7-year follow-up, DFS was 89.9% and 83.1%, for exercise and usual care groups, respectively. Differences between the pooled unadjusted results and results from the model including an interaction between study and group assignment for DFS were marginally significant (p = 0.07). For women in the urban trial, DFS was significantly higher for the exercise group (91.8%) than the usual care group (80.0%) (unadjusted HR 0.39, 95% CI: 0.17–0.89, p = 0.03; Fig. 3b). For women in the rural trial, there was no effect of exercise throughout the duration of follow-up; DFS was 84.9% in the exercise group compared with 85.7% in the usual care group (unad- justed HR: 1.22, 95% CI 0.55–2.68, p = 0.63; Fig. 3c).

Subgroup analyses

Subgroup analyses for prognostic characteristics including age, body mass index, presence of comorbidities and stage of disease at diagnosis, as well as for intervention engage- ment variables including intervention compliance, weekly physical activity levels at 12 months post-diagnosis and mode of intervention delivery, for OS are shown in Fig. 4a and b. All subgroups show improved survival with exercise, as reflected by HRs below 1, although confidence intervals are generally wide and individual results are not necessar- ily statistically significant. The benefits of exercise on OS were statistically significant (p < 0.05) for younger women (< 55 years), those with stage II + disease, and those with

13

Breast Cancer Res Treat (2018) 167:505–514
Table 1 Baseline characteristics of study participants and cancer treatment by study arm and group allocation

509

Pa 0.103

0.424 0.211

0.565

0.376 0.584 0.658 0.372

0.116 0.761 0.588 0.375

Baseline characteristics

Age (years) Mean ± SD

Body mass index (kg/m2) Mean ± SD

Pooled
Usual care n = 130

53.9 ± 8.3

Urban
Pa Usual care

n = 60

0.021 53.7 ± 7.9

Rural Usual care

n = 70

0.122 54.1 ± 8.7

Exercise n = 207

51.7 ± 8.8

Exercise Pa n = 134

51.7 ± 8.7

Exercise n = 73

51.7 ± 9.0

Comorbiditiesb
0 12.3 16.4 1 20.0 23.7 2 23.8 28.0 3+ 43.8 31.9

Stage
Stage 0/I 40.0 34.8 Stage II/III 53.1 58.5 Unknown 6.9 6.8

Lymph node status
Negative 57.7 51.2

ER/PR status
Positive 53.9 64.7

HER2 status
Negative 64.6 71.0

Most extensive surgery
Lumpectomy 53.1 58.5 Mastectomy/MRM 43.1 38.7 Unknown 3.9 2.9

Chemotherapy
Yes 59.2 69.1

Radiotherapy
Yes 64.6 66.2

Hormone therapy
Yes 60.8 63.8

Herceptin
Yes 12.3 15.9

15.0 20.9

20.0 26.1 0.168 26.7 23.1 38.3 29.9

35.0 34.3 0.606 61.7 61.9

3.3 3.7 0.541 55.0 53.7 0.094 53.3 67.2 0.467 56.7 73.1 0.601 56.7 69.4

43.3 30.6 0.0 0.0

0.083 66.7 70.1 0.859 68.3 70.9 0.661 63.3 63.4 0.446 15.0 15.7

27.2 ± 7.0 %%%%%%

26.7 ± 5.1

0.502 26.5 ± 5.8

26.6 ± 4.9

26.8

0.470 10.0 8.2 20.0 19.2 21.4 37.0 48.6 35.6

0.988 44.3 35.6 45.7 52.1 10.0 12.3

0.399 60.0 46.6 0.098 54.3 60.3 0.065 71.4 67.1 0.119 50.0 38.4

42.9 53.4 7.1 8.2

0.751 52.9 67.1 0.849 61.4 57.5 1.000 58.6 64.4 1.000 10.0 16.4

± 5.4

0.903 27.7 ± 7.8

ER/PR estrogen receptor/progesterone receptor; HER2 epidermal growth factor receptor 2

ap values were calculated using unpaired t tests for continuous variables and Chi-squared tests for categorical variables

bComorbidities included heart attack, angina, high blood pressure, high cholesterol, other heart conditions, stroke, diabetes, asthma, chronic bronchitis, emphysema, ulcer, migraine headaches, osteoporosis, arthritis, rheumatoid arthritis, depression, other cancers, other illness and other conditions

at least one comorbidity at baseline. Findings also suggested a greater effect on survival outcomes among those in the exercise group who reported higher intervention compliance and for those meeting national physical activity targets at 12 months of follow-up (minimum of 150 min of weekly physical activity, p < 0.05), especially when at least 10 min of vigorous intensity activity was undertaken each week. Similar findings were observed for subgroup analysis for DFS (Supplementary Fig. 1), with the impact of exercise

on DFS being significantly improved for urban women who received the intervention through face-to-face contact with an exercise physiologist (p < 0.05).

Discussion

These findings support a favourable effect of exercise on OS and likely positive impact on DFS, irrespective of age,

13

510 Breast Cancer Res Treat (2018) 167:505–514

13

Table 2 Overall, breast cancer-specific and disease-free survival events by study arm and group allocation Pooled

Urban Usual care

Rural Usual care

Overall survival events Breast cancer-specific deaths

26 (7.7)

15 (11.5)

11 (5.3)

7 (11.7)

5 (3.7)

8 (11.4)

6 (8.2)

Non-breast cancer-related deaths Disease-free survival eventsa

20 (5.9) 6 (1.8) 48 (14.2)

10 (7.7) 5 (3.8) 23 (17.7)

10 (4.8) 1 (0.5) 25 (12.1)

5 (8.3)
2 (3.3) 12 (20.0)

5 (1.5) 0 (0.0) 11 (8.2)

5 (7.1)
3 (4.3) 11 (15.7)

5 (6.8)
1 (1.4) 14 (19.2)

Recurrence of breast cancer Alive at follow-up Deceased by follow-up

20 (5.9) 9 (2.7) 11 (3.3) 13 (3.9) 13 (3.9) 0 (0.0)

8 (6.2) 3 (2.3) 5 (3.1) 5 (3.8) 5 (3.8) 0 (0.0)

12 (5.8) 6 (4.6) 6 (4.6) 8 (3.9) 8 (6.2) 0 (0.0)

4 (6.7) 2 (3.3) 2 (3.3) 3 (5.0) 3 (5.0) 0 (0.0)

5 (3.7) 3 (2.2) 2 (1.5) 3 (2.2) 3 (2.2) 0 (0.0)

4 (5.7) 1 (1.4) 3 (4.3) 2 (2.9) 2 (2.9) 0 (0.0)

7 (9.6) 3 (4.1) 4 (5.5) 5 (6.8) 5 (6.8) 0 (0.0)

New primary breast cancer Alive at follow-up Deceased by follow-up

All patients n = 337
n (%)

Usual care n = 130
n (%)

Exercise n = 207 n (%)

n = 60 n (%)

Exercise n = 134 n (%)

n = 70 n (%)

Exercise n = 73 n (%)

aDisease-free survival events include recurrence of breast cancer, new primary breast cancers and all-cause deaths. Note that women who had recurrence and died before follow-up only count once towards disease-free survival events

Breast Cancer Res Treat (2018) 167:505–514 511

Fig. 2 Cumulative incidence of death from any cause of pooled (a), urban (b) and rural (c) breast cancer patients randomised to the control arm or the exercise intervention arm

Fig. 3 Cumulative incidence of breast cancer recurrence, new primary breast cancer or death from any cause of pooled (a), urban (b) and rural (c) breast cancer patients randomised to the control arm or the exercise intervention arm

stage of disease, presence of other comorbidities or body mass index. Due to the exploratory nature of these analyses, certainty around the magnitude of effect is limited. Nonethe- less, our observed improvements of over 50% for OS and over 30% for DFS (HRs: 0.45 and 0.66, respectively) are consistent with results from other available studies. Follow- ing a meta-analysis of 6 cohort studies, involving > 12,000 women with breast cancer, a hazard ratio of 0.59 (95% CI: 0.53–0.65, p < 0.01) for OS was reported for those engag- ing in sufficient levels of physical activity post-breast cancer diagnosis compared with those who were sedentary [12]. Further, exploratory analyses conducted by Courneya and colleagues following a randomised, controlled trial evalu- ating the role of exercise during chemotherapy for breast cancer reported a HR of 0.60 (95% CI: 0.27–1.33) for those in the exercise group compared with those in the usual care group [6]. Additionally, our point estimate for DFS is also

similar to that reported by Courneya and colleagues (HR: 0.68; 95% CI 0.37–1.24; n = 242), with both trials having similar sample sizes [6].

Pre-planned subgroup analyses undertaken as part of this work suggest that the favourable survival response to exercise following a breast cancer diagnosis was particularly evident for those residing in an urban area, of younger age (< 55 years), having stage II + disease, or for those with at least one comorbidity at the time of their breast cancer diagnosis. While exploratory in nature, Courneya and col- leagues also reported similar hypothesis-generating findings with respect to stage of disease, but found that older women had a potentially greater survival response to exercise than younger women [6]. Population-based five-year survival rates for women with stage I disease versus stage II + dis- ease are 98 and 83%, respectively [18]. Clearly, the capacity to demonstrate a beneficial survival response in women with

13

512

Breast Cancer Res Treat (2018) 167:505–514

Fig. 4 Forest plots showing unadjusted hazard ratios (HRs) and 95% confidence intervals (CI) for the impact of randomisation to the exercise group on overall survival by subgroup and for all patients. n number of patients per subgroup. aDisease stage was missing for 23 women. bComorbidities included heart attack, angina, high blood

less invasive disease is limited. Also, our findings suggest- ing that younger women may show a stronger response are likely confounded to some degree by stage and/or type of breast cancer; e.g. younger women were more likely to be diagnosed with later stage and/or more aggressive disease than older women in our study and elsewhere [3, 18]. There is an extensive evidence-base that demonstrates the benefits of exercise with respect to the management of comorbidi- ties such as cardiovascular disease, osteoporosis and type 2 diabetes [14], and as such, the favourable exercise response for those with at least one comorbidity at diagnosis of breast

pressure, high cholesterol, other heart conditions, stroke, diabetes, asthma, chronic bronchitis, emphysema, ulcer, migraine headaches, osteoporosis, arthritis, rheumatoid arthritis, depression, other cancer, other illness and other conditions. cPhysical activity minutes were missing for 16 women

cancer is not surprising. Since more than 80% of our sample presented with 1 or more comorbidities at baseline, our find- ings accentuate the importance of integrating exercise into standard breast cancer care.

In this study, rural residence reflects regional or remote location, including both town and country settings, but is non-metropolitan in nature. In Australia, women residing in such rural environments have estimated 5-year survival rates from breast cancer of 84% compared with 90% for urban women [2]. As such, one might hypothesise a greater poten- tial for the positive effects of exercise to be demonstrated in

13

Breast Cancer Res Treat (2018) 167:505–514

rural-residing women. However, further examination of the urban/rural differences in survival in this study reveals that urban women experienced benefits for all causes of mortality and breast cancer morbidity, whereas rural women showed improvements only for mortality from other causes. Whether this represents sampling variability and random fluctuation between the two trials or a lack of exercise benefit for DFS for rural women is unclear. We cannot discount the possibil- ity that the exercise intervention evaluated in the EfH trials was unable to overcome the adverse impact of differences in other characteristics known to exist between rural and urban women, such as education, income and access to care [1, 7].

The EfH trials were effectiveness trials, and together with the survival findings from Courneya and colleagues’ efficacy trial [6], provide mounting evidence suggesting that exercise can influence survival. Courneya et al. evaluated the effect of participating in a highly supervised, moderate-intensity, aerobic-only or resistance-only intervention throughout the duration of chemotherapy. Our trial involved the prescription of a mixed-type, moderate-intensity intervention that was largely unsupervised but was supported by regular face-to- face or telephone-based contact with an Exercise Physiolo- gist during and beyond adjuvant treatment. The majority of the women in our trial met exercise targets by walking at moderate intensity, with or without bouts of vigorous- intensity or resistance-based exercise. Consequently, when the intervention protocols of the EfH trials and Courneya and colleagues trial are considered together, findings sug- gest that at least with respect to survival outcomes, type of exercise may be less important; rather, they support the notion that doing any exercise, irrespective of type, may be what matters most. Our findings also suggest that mainte- nance of some activity at 12 months post-diagnosis, is better than none and that more exercise is better than less, at least up to internationally recommended levels of exercise dose (that is, 150 + min of moderate-intensity physical activity per week). However, incorporating some vigorous-intensity exercise into weekly physical activity levels may provide additional survival benefit.

The primary limitation of this work is the exploratory nature of the analyses, since the trials were not originally designed to evaluate the impact of exercise on survival. As a consequence, statistical power was limited and preci- sion around the effect sizes reported is lacking. Nonethe- less, strengths include long follow-up, complete follow-up including recurrence data and adherence to intention- to-treat principles in analyses. Also, pooling of the two EfH trials likely provides for more generalizable results. However, since the resulting sample overrepresents non- metropolitan women (the rural trial represents 43% of our sample in contrast to 20–30% of the general population who live outside a major metropolitan area in Queensland

513

[4]), the pooled results may be biased in a conservative direction. Moreover, these results were derived from a pragmatic trial, evaluating an intervention that addressed logistical issues around implementation and reflected real- world constraints around the ability to tightly control exer- cise dose undertaken by participants.

Had we restricted this report to the EfH urban trial only, the results would have been unambiguous: statis- tically significant, clinically important improvements in both OS and DFS 8 years following breast cancer diag- nosis among those in the exercise group. Our decision to include the rural trial, while enhancing generalizability of results, has introduced some uncertainty about the extent and magnitude of the benefit. Further, given the explora- tory nature of the analyses, caution in over-interpreting findings is needed. Nonetheless, while we await results from adequately powered randomised, controlled exercise trials, findings presented here represent important contri- butions towards understanding the effect of exercise on survival following breast cancer. Specifically, from previ- ously published reports by ourselves [8, 9, 11] and others [5, 16], trial evidence is now accruing to establish that it is safe, feasible and cost-effective for women with breast cancer to engage in regular physical exercise during and after adjuvant treatment, and that doing so will improve fatigue, fitness, quality of life and likely survival.

Acknowledgements Study investigators would like to acknowledge and thank those who agreed to participate in the Exercise for Health tri- als. Their contributions to improving future breast cancer care are sig- nificant. Thank you also to the breast care nurses, exercise physiologists and research staff, in particular Sheree Rye and Tracey DiSipio, for their efforts in study recruitment, implementation and analysis, and to Emeritus Professor Beth Newman, who has provided invaluable review of this manuscript and mentorship throughout all phases of these trials.

Funding The EfH urban and rural/regional trials were supported by two National Breast Cancer Foundation (NBCF) project grants. During the conduct of the EfH trials, SH and EE were supported by NBCF and National Health and Medical Research Council (NHMRC) research fel- lowships, respectively. SH is currently supported by a Cancer Council Queensland Fellowship.

Compliance with ethical standards

Conflict of interest

of interest.

Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

The authors declare that they have no conflict

Informed consent was obtained from all indi- vidual participants included in the study.

Informed consent

13

514

References

  1. Australian Institute of Health and Welfare (2016) Australia’s health 2016. In: Australia’s Health No. 15. AIHW, Canberra

  2. Australian Institute of Health and Welfare (2012) Breast can-

    cer in Australia: an overview. In: Cancer Series No. 71. AIHW,

    Canberra

  3. Australian Institute of Health and Welfare (2015) Breast cancer

    in young women: key facts about breast cancer in women in their

    20s and 30s. In: Cancer Series No. 96. AIHW, Canberra

  4. Australian Institute of Health and Welfare (2004) Rural, regional and remote health: a guide to remoteness classifications. In: Rural

    Health Series No. 4. AIHW, Canberra

  5. Cormie P, Zopf EM, Zhang X, Schmitz KH (2017) The impact

    of exercise on cancer mortality, recurrence, and treatment-related adverse effects. Epidemiol Rev 39:71–92. doi:10.1093/epirev/ mxx007

  6. Courneya KS, Segal RJ, McKenzie DC, Dong H, Gelmon K, Friedenreich CM, Yasui Y, Reid RD, Crawford JJ, Mackey JR (2014) Effects of exercise during adjuvant chemotherapy on breast cancer outcomes. Med Sci Sports Exerc 46:1744–1751. doi:10.1249/MSS.0000000000000297

  7. Dasgupta P, Baade PD, Aitken JF, Turrell G (2012) Multi- level determinants of breast cancer survival: association with geographic remoteness and area-level socioeconomic disad- vantage. Breast Cancer Res Treat 132:701–710. doi:10.1007/ s10549-011-1899-y

  8. Eakin EG, Lawler SP, Winkler EA, Hayes SC (2012) A rand- omized trial of a telephone-delivered exercise intervention for non-urban dwelling women newly diagnosed with breast cancer: exercise for health. Ann Behav Med 43:229–238. doi:10.1007/ s12160-011-9324-7

  9. Gordon LG, DiSipio T, Battistutta D, Yates P, Bashford J, Pyke C, Eakin E, Hayes SC (2017) Cost-effectiveness of a pragmatic exercise intervention for women with breast cancer: results from a randomized controlled trial. Psychooncology 26:649–655. doi:10.1002/pon.4201

10. 11.

12. 13.

14.

15. 16.

17. 18.

Breast Cancer Res Treat (2018) 167:505–514

Hayes S, Rye S, Battistutta D, Yates P, Pyke C, Bashford J, Eakin E (2011) Design and implementation of the Exercise for Health trial—a pragmatic exercise intervention for women with breast cancer. Contemp Clin Trials 32:577–585. doi:10.1016/j. cct.2011.03.015

Hayes SC, Rye S, Disipio T, Yates P, Bashford J, Pyke C, Saun- ders C, Battistutta D, Eakin E (2013) Exercise for health: a ran- domized, controlled trial evaluating the impact of a pragmatic, translational exercise intervention on the quality of life, function and treatment-related side effects following breast cancer. Breast Cancer Res Treat 137:175–186. doi:10.1007/s10549-012-2331-y Ibrahim EM, Al-Homaidh A (2011) Physical activity and survival after breast cancer diagnosis: meta-analysis of published studies. Med Oncol 28:753–765. doi:10.1007/s12032-010-9536-x Lahart IM, Metsios GS, Nevill AM, Carmichael AR (2015) Physical activity, risk of death and recurrence in breast cancer survivors: a systematic review and meta-analysis of epidemio- logical studies. Acta Oncol 54:635–654. doi:10.3109/02841 86X.2014.998275

Pedersen BK, Saltin B (2015) Exercise as medicine—evidence for prescribing exercise as therapy in 26 different chronic diseases. Scand J Med Sci Sports 25(Suppl 3):1–72. doi:10.1111/sms.12581 Schemper M, Smith TL (1996) A note on quantifying follow-up in studies of failure time. Control Clin Trials 17:343–346 Schmitz KH, Courneya KS, Matthews C, Demark-Wahnefried W, Galvao DA, Pinto BM, Irwin ML, Wolin KY, Segal RJ, Lucia A, Schneider CM, von Gruenigen VE, Schwartz AL (2010) Ameri- can College of Sports Medicine roundtable on exercise guide- lines for cancer survivors. Med Sci Sports Exerc 42:1409–1426. doi:10.1249/MSS.0b013e3181e0c112

Therneau T (2015) A Package for Survival Analysis in S Version 2.38. https://CRAN.R-project.org/package=survival
Youlden D, Cramb S, Baade P (2009) Current status of female breast cancer in Queensland: Brisbane 1982 to 2006. In: Viertel Centre for Research in Cancer Control, Cancer Council Queens- land, Brisbane

13

I’ve signed up to the RACQ International Women’s Day Fun Run!

I'm raising funds to support women with breast cancer, providing things like counselling, wigs and mastectomy bras. As little as $58 can make a profound impact and take some of the stress off women with breast cancer and their families. Will you help me?

My Achievements

Thank you to my Sponsors

$113.36

Chris Pyke

Its a good cause. We are a great team, working to be part of many great teams

$3.16

Chris Pyke

$113.36

Kris Kerr

Great work! 🏃‍♂️ 💨

$1,000

Justin Perron

Good luck with the run!!

$113.36

Richard Loong

Good on you Chris.

$500

Anonymous

I will have the heat packs ready Chris 😜

$113.36

Peter And Therese Borzi

Go Chris!!

$60.32

Gerard Keating

Go Hard or Go Home 👍