Design of a randomized controlled trial of physical training and cancer (Phys-Can) – the impact of exercise intensity on cancer related fatigue, quality of life and disease outcome

The design is a 2×2 factorial design where participants are randomized to one of the following four groups (Fig. 1):

The RCT is stratified and within each stratum randomization is carried out following a permuted block design with 8 participants per block. The defined strata are the three hospitals (Linköping, Lund and Uppsala) and three cancer sites yielding 9 strata. When all baseline measurements are finalized and reported in the project’s electronic case report form, the randomization is generated automatically in a web-portal (described below) of participants to an exercise group according to predefined strata (see above). The personnel in charge of recruitment then contact the participant with information on study condition and the first visit to the gym is scheduled.

All participants will exercise at least twice weekly for a period of 6 months under the guidance of trained coaches. Training intensity is 40–50% (LM) or 80–90% (H) of each individual’s heart rate reserve (Heart rate reserve = HRpeak-HRrest) and maximal muscle strength.

The calculated sample size was based on detecting the main factorial effects of each factor on Multidimensional Fatigue Inventory (MFI) Physical Fatigue (PF) after completed intervention [32‐34]. The target main effect size was determined to be 2, which according to Purcell et al. [35] is the minimal clinically important difference for the MFI-PF. We assumed that the standard deviation (SD) of PF was 5 in all four treatment arms, which is the SD reported in Purcell et al. [35], Hagelin et al. [36] and in an ongoing study within our research group (manuscript in preparation). Because three tests (two main and one interaction effect) were planned, the 5% significance level was adjusted using Bonferroni correction. In order to have 80% power to detect a main factorial effect of 2 under the null hypothesis of no effect, 67 individuals in each of the four treatment arms are needed, with a total sample size of 268 individuals.

Further adjustments of the sample size were necessary for several reasons. First, the sample size calculation was based on the assumption of no interaction effects. However, Montgomery et al. [29] argue that, if anticipating quantitative interactions, the trial should be powered to detect those effects. Interaction effects often are much smaller in size compared to the main effects. In clinical trials, it is therefore rare to have 80% power to detect the interaction effect. For instance, given the calculated sample size in the present study, the trial would have a power of only 22% to detect an interaction effect of 1. However, in the present study a possible interaction between the factors is of clinical interest. Thus, the original sample size was doubled, increasing the power to detect the interaction between to approximately 50%. Second, to account for missing data and drop-outs, the original sample size was increased further by 30%. Third, given that we include a baseline assessment, and assuming a moderate correlation (0.3) between the MFI-PF at baseline and follow-up assessments, the sample size can be reduced by approximately 10% [37]. Combining the above, we concluded that including approximately 150 individuals in each treatment arm, 600 individuals in total, should meet the statistical requirements of the trial.

All participants are offered guided exercise for 6 months. This is equal to the most extensive adjuvant treatment period with the exception for endocrine therapy which may continue up to 10 years (for persons with breast cancer). It is also an appropriate period of time to achieve physical exercise effects and to establish a physically active lifestyle [22, 38]. Trained coaches will guide both resistance and endurance exercise.

The resistance training will be performed at public gyms during two supervised sessions per week. During the first six weeks after inclusion the participants become familiar with the exercises and tests as well as how to use the Omni-scale for self-reported perceived exertion [39] included in the resistance program. During these six weeks, there will be a progression in sets and power output as well as a reduction in the number of repetitions, resulting in each participant having an individualized program. The following exercises are included in the program and performed on machines; seated leg press, chest press, leg extension, seated row, seated leg curl, and seated overhead press using dumbbells. In addition, participants are instructed to do the following core exercises on a regular basis: sit-ups, the plank, bird-dog and pelvic floor exercises.

The low-to-moderate intensity group exercises at 50% of six- (the first weekly session) and ten repetitions maximum (the second weekly session) corresponding to 12 and 20 repetitions in three sets (reporting 5–7 on the Omni-scale for perceived exertion (39)). Rest periods between sets are two and one minute for the two sessions, respectively. The high intensity group exercises at six- (the first weekly session) and ten repetitions maximum (the second weekly session) corresponding to six and 10 repetitions in three sets (reporting 9–10 on the Omni-scale for perceived exertion (39), with the last set continuing to failure. Rest periods between sets are two and one minute for the two sessions, respectively. Relative exercise intensity is adjusted over the remaining intervention period according to repeated measures of six- and ten repetitions maximum.

The first six weeks after inclusion the participants familiarize themselves with the use of the heart rate monitor and perceived exertion using the Borg-scale [40] for monitoring of exercise intensity and perceived exertion. Four endurance sessions are conducted at the gym and supervised by a coach. Thereafter, the endurance exercise is home-based and followed up by a coach. All participants are instructed to warm-up for approximately 10 min at moderate intensity before each endurance session. The participants are instructed to wear the heart rate monitor during every session and report perceived exertion in a diary. Exercise frequency is recommended to be 2–4 times a week. The low-to-moderate intensity group do continuous-based exercise (running, cycling, walking uphill or any other endurance-based activity) in bouts of at least 10 min at an exercise intensity of 40–50% of the heart rate reserve. The main aim is to reach 150 min of moderate intensity per week. The high intensity group conduct high-intensity interval exercise at an exercise intensity of 80–90% of the heart rate reserve (at the end of the 3rd session) with two minutes exercise (running, cycling, walking uphill or any other endurance-based activity) and two minutes rest between sessions. The participants start with five sessions, increasing to six after the six weeks familiarization period, thereafter adding one session every fourth week until 10 sessions are reached as the maximum, corresponding to 75 min of vigorous intensity per week.

Self-regulatory BCTs are provided for the H + BCTs and the LM + BCTs groups. These are strategies to facilitate adherence to the high and low-to-moderate intensity exercise programs, respectively. These support strategies focus on the adherence to the exercise intervention, primarily on the endurance exercise, as it is home-based, and on maintaining exercise according to individual preferences after the completion of the interventions. The BCTs include a) behavioural goal-setting, b) short-term action planning, c) self-monitoring, d) review of behaviour goals, e) problem solving and functional behaviour analysis to identify affect individual-specific determinants of exercise behaviour, and f) long-term coping planning to maintain physical exercise by own choice after the intervention is completed. In addition, motivational aspects are explored by initial interviews based on a) previous experiences with physical exercise, b) persons’ outcome expectations for following the exercise intervention as prescribed, c) persons’ anticipated barriers and facilitators following the exercise intervention, and d) self-efficacy to partake in planned weekly exercise. The BCTs are provided according to protocol, but are individually tailored according to each participant’s need and functional behavioural analyses. Each participant meets their coach on a regular basis during the program, with gradually decreasing frequency in order to enhance self-regulation at the end of the intervention. These coaching sessions take place either in connection with the scheduled exercise sessions or via phone. Participants with access to internet can use electronic and easily accessible self-monitoring of exercise behaviour in a web portal developed for the Phys-Can study (described below).

Coaches (physiotherapists and personal trainers) have been trained to provide the interventions in the four groups. The education consisted of three common course days for all coaches and three additional days for those providing the BCT conditions with home assignments between course days. The education included lectures and seminars on cancer and cancer treatment, exercise physiology, and repeated practice sessions in the gym according to the intervention protocol on high and low/moderate exercise. Additional lectures and practice on exploration of motivation and use of BCTs were given to coaches providing the BCT conditions.

Written intervention protocols have been developed, specifying all intervention components in the four groups. The coaches are to follow the protocol closely, keep logs of attendance at gym sessions, check heart rate during endurance sessions and monitor any adverse advents related to exercise. Adverse events caused by the exercise are registered in a web-portal (described below). Grade 1 (e.g. muscle strain) means that the participant have to terminate the ongoing specific exercise but can continue with the exercise session. Grade 2 (e.g. fall in blood pressure) means that the participant must terminate the exercise session. Any severe adverse events (e.g. fracture) are reported directly to the PI and managed by healthcare.

To address the coaches’ fidelity to the intervention protocol, research staff visit the gyms repeatedly during the intervention, giving feedback on both physical exercise and use of BCTs. Regular project group meetings are held with all project personnel, where issues about delivering the intervention are discussed and solved. In addition, one coach representative from each study site takes part in monthly Skype meetings where common issues relating to the intervention are discussed and coordinated. Coaches’ use of BCTs is monitored by audio recordings of conversations between coaches and participants, and individual feedback is given by one of the research staff.

A web-portal has been designed. The aim of the Phys-Can web portal is to facilitate collection of outcome measures in the multiple study sites and to enable easily accessible electronic self-monitoring of exercise behaviour. The participants sign on to the web portal and complete forms related to outcome assessments and self-monitoring of endurance training included as part of the intervention (those who have no access to internet fill out paper forms). The data base is located at Uppsala University and constructed according to security principles from the Division of Security at Uppsala University. All communication with the data base is encrypted and backups are performed on a regular basis in order to secure data. The web portal automatically sends e-mails or text messages to participants when the scheduled assessment is available, informing them to long on to the portal to complete the questionnaires. Participants also receive reminders by e-mail or text messages. Those who choose to complete paper forms receive the questionnaires and reminders by regular mail. The web portal also notifies research assistants by e-mail when an assessment point is about to open for a specific participant and alerts the research assistants if questionnaires are not completed within a set time frame. This enables precise monitoring of time points for both participants who use the web portal or choose to fill out paper forms. At assessment points when blood samples, fitness tests and recordings of physical activity are included, a research assistant contacts participants by phone to schedule an appointment.

The data will be analysed according to the intention-to-treat (ITT) principle, i.e. all participants will be analyzed as randomized, regardless of whether they complete the intervention as planned. The primary parameters of interests are the three factorial effects (2 main and one interaction) after completed intervention. Each factorial effect will be estimated taking into account the stratified design following Imbens and Rubin [53]. Thus, we will provide estimates for the overall factorial effects, which are the primary parameters of interest, and also stratum-specific factorial effects which could facilitate external validity. Baseline measurements will be included in the analysis model to increase the precision of the estimates.

Drop-outs and missing values will be handled according to the following procedure. The main analysis will be based on data where missingness is handled using multiple imputations by chained equations. The specification of the multiple imputation model (e.g. choice of auxiliary variables) will be decided after comparing the subject characteristics in those with complete and incomplete data. Under the assumption of missing at random (MAR), an analysis based on multiple imputation conforms with the ITT. Due to the assumption of MAR, which could be a strong assumption to make, factorial effects will also be estimated using complete blocks only. This is a result from the permuted block design which allows for unbiased estimation of the factorial effects using the complete blocks only. Sensitivity analysis will conducted by comparing estimates from the two analyses above with estimates based on complete cases only, a pattern-mixture model, and tipping points analyses [54].

Even though the main aim of the present comparative effectiveness study is to compare different exercise intensities on CRF and HRQoL, the lack of a randomized usual care control group may be a limitation. Due to strong evidence that physical exercise is beneficial for persons during cancer treatment we considered it unethical to randomise persons to an untreated control group that will not be offered exercise. To form a historical cohort, an observational study, was initiated and data collected preceded the randomised study. Persons with the same diagnosis and with the same inclusion/exclusion criteria as in the intervention study were included in the cohort study at all three recruiting centres. The inclusion started in September 2014 and ended immediately prior to start of inclusion to the intervention study. A total of 95 participants were included in the observational study and will be followed for 10 years with same outcomes measures as in the intervention study.

The sample size has been calculated to be 600 persons in total, which is a challenge in a multicentre, complex intervention study. However, a feasibility study (manuscript submitted) preceding the randomized study evaluated the inclusion procedures, settings, intervention components, instruments, and tests.

The lesson learned from the feasibility study has contributed in several ways both to the final planning of the Phys-Can trial. All the physical fitness tests, physical activity monitoring, and conducting exercise of low-to-moderate or high intensity, seem feasible to implement in exercise oncology interventions. In addition, exercise including resistance training with arm movements above the head and repetitive muscle contractions of the arm is feasible and may be safe for patients wearing a peripherally inserted central venous catheter (PICC) (a venous access used for chemotherapy administration is a) due to adjuvant chemotherapy. The results highlighted the need for an enhanced learning and reporting regarding endurance exercise and checking of intensity levels. The results also implied that BCT do not have to target the exercise behavior during supervised sessions; rather, the support should target behaviors pertaining to unsupervised or home-based exercise.

In summary, the Phys-can study will contribute to our understanding of the value of exercise and exercise intensity in preventing CRF and maintaining HRQOL during and after treatment and, potentially, clinical outcomes as well. It will also provide insights into possible biological mechanisms through which exercise affects treatment outcomes. The value of BCTs in terms of adherence to, and maintenances in, exercise behaviour in persons with cancer will be evaluated. Implementation of the results into clinical practice will be facilitated by the close collaboration between researchers and clinicians as well as the facts that the study is performed in non-clinical settings (public gyms) which may create a pathway between hospitals to society.