Preference-Based Exercise RCT for Men With PC on ADT

BACKGROUND: The backbone of scientific evidence to address most scientific questions, including behavioural interventions, is randomized controlled trials (RCTs). Prostate cancer is the most common malignancy in men, affecting 1 in 7 men. Almost 50% of these men will receive androgen deprivation therapy (ADT) at some point after diagnosis to reduce disease progression and prolong survival. ADT is associated with multiple adverse effects, including diminished quality of life (QOL), fatigue, reduced muscle mass, and metabolic effects. Numerous RCTs in the setting of men on ADT have demonstrated that exercise programs, particularly supervised 1:1 in-centre programs, are clinically effective. However, they are resource-intensive and have low scalability. Alternative delivery models such as group-based supervised exercise or home-based exercise are promising alternatives but require rigorous efficacy data. While conducting a two-centre exercise trial comparing different exercise delivery models, fewer than 40% of eligible men agreed to participate. Two common reasons why men declined participation were distance to exercise centres and unwillingness to be randomized. Novel clinical trial designs, such as preference-based trials, may help answer effectiveness questions and enhance generalizability by recruiting a more representative pool of men. Increased efficiencies in recruitment, however, may come at a cost of bias with non-randomized designs as well as differential effects of exercise and varying adherence. In parallel with a multi-centre phase III RCT funded by CIHR, we propose to conduct a pilot preference-based trial, examining 3 key issues. First, can we successfully enroll men who are unwilling to be randomized due either to distance from centre or a strong preference for one exercise delivery model? Second, are these men different from men who are randomized in terms of baseline characteristics, study retention, and adherence to the intervention? Third, are the benefits of an exercise program similar to men who are randomized to the same arm?

OBJECTIVES:

To determine, in men with PC on ADT who are otherwise eligible to participate in a RCT of different exercise delivery models but decline to be randomized due to distance from the study centre or strong preference for one exercise delivery model:

1. What proportion of these men is willing to be enrolled in a preference-based trial.
2. How comparable are baseline characteristics of men who agree to participate in the intervention arm of their choice and all those in the randomized study?
3. How comparable are study retention and adherence to the intervention for men in the corresponding arms of the preference based trial and randomized trial?
4. How comparable are the benefits in terms of quality of life (QOL) and physical fitness outcomes in those with a preference for a particular exercise and those without a preference?

HYPOTHESES

1. Systematic differences exist in baseline characteristics, specifically distance to the study centre, age, education, prior experience with participating in exercise programs, and fatigue severity among participants willing to be randomized versus not randomized to an exercise intervention.
2. Adherence will be greater among those selecting treatment by preference vs being randomized.
3. Benefits (QOL and fitness) will be greater among those selecting treatment versus being randomized.

METHODS: Patients aged 18 or older on ADT for high-risk/locally advanced, biochemically relapsed, or asymptomatic metastatic prostate cancer will be eligible if they are otherwise eligible for an ongoing phase III RCT of exercise but decline to participate due to either distance from study centre or strong preference for one exercise delivery model. The main phase III RCT is examining a 6-month individualized, progressive, mixed-modality exercise program including aerobic, resistance, and flexibility components delivered using one of two delivery models: supervised in-centre group exercise 3 times weekly or home-based exercise supported by a Fitbit, smart phone app, and remote health coach.

To meet the objectives of this pilot preference trial, we will enroll 50 men per preference arm at three experienced study sites (Princess Margaret Cancer Centre, Toronto, Scarborough and Rouge Hospital, Scarborough and the University of Calgary, Calgary and). Aim 1. We will report the proportion of men who are eligible for the preference-based trial who agree to participate, overall and by treatment arm and reason for refusal to be randomized.

Aim 2/Hypotheses 1 and 2. We will compare the participants in the preference-based trial to those in the RCT. Comparisons of baseline characteristics will be between the total number of participants in the RCT compared separately to each arm of the preference trial. ANOVA will be used for continuous variables and chi-square analysis for categorical variables. Variables to be compared include distance from participants' home to the study centre, age, education, prior participation in an exercise program, and symptom severity. Differences will be presented with 95% CIs.

Aim 3/Hypothesis 3. Within-group change scores will be assessed for the preference-based groups, and the mean changes along with the 95%CIs will be reported. Next, we will compare these changes for the co-primary and secondary outcomes between the preference-based groups and the phase III RCT (e.g., combined preference arms to combined RCT arm) to obtain the so-called selection-effect. Finally, although it is subject to confounding by unknown preference effects in the RCT participants, we will estimate the differences between outcomes in those randomized to and choosing each specific exercise delivery method, using constrained linear mixed effects model adjusting for the baseline value and ADT duration. Skewed data will be transformed as appropriate.

SIGNIFICANCE: Behavioural interventions require significant patient commitment and although most RCTs to date have shown evidence of benefit on various outcomes, recruitment rates have been low and many otherwise eligible men are not willing to be randomized for different reasons. Designing trials that allow inclusion of such men is obviously going to improve generalizability but whether the results of interventions are similarly effective and understanding the feasibility and potential biases of enrolling such men are fundamental unresolved issues. This pilot study will take advantage of an existing trial to begin to answer these questions and determine whether a larger preference-based trial is worth undertaking. Our findings have potential widespread implications for behavioural trials across many health care settings.