Design
This study was a proof-of-concept sub-study, which took part within the context of a two-arm, mixed-methods pilot RCT designed to test the effects of a 12-week PA intervention on physical and psychological outcomes among AYAs after cancer treatment [38]. The protocol for the pilot RCT was registered in the ClinicalTrials.gov database (NCT03016728) and was approved by local Research Ethics Boards. The Consolidated Standards of Reporting Trials (CONSORT) 2010 checklist of information to include when reporting a pilot or feasibility trial [39] was adhered to in the preparation of this manuscript (see additional file 1).
Participants
AYAs in the two-arm, mixed-methods pilot RCT were recruited through healthcare provider referral and self-referral over a 12-month period starting in September 2017. To participate in the pilot RCT, AYAs had to: (1) have been diagnosed with cancer between the ages of 15–39 years; (2) have completed cancer treatment within the past 5 years ; (3) have no evidence of progressive or recurrent disease or of secondary or second cancers; (4) be inactive/insufficiently active as assessed using a single-item screening question to which individuals had to respond negatively (“Are you currently engaging in moderate PA, that is activity that increases your heart rate and causes you to sweat, more than three days/week?”); (5) be medically cleared to participate in PA (as determined by a PA readiness questionnaire and in some cases a member of their healthcare team); and, (6) be able to read, understand, and provide informed consent in English. AYAs were not eligible if they self-reported having a physical impairment precluding participation in PA.
Following enrollment into the pilot RCT, participants who met the following criteria were invited to take part in this proof-of-concept sub-study: (1) self-reported being right-handed (e.g., writing and using a computer mouse with the right hand to increase the likelihood of recruiting a sample with left-language lateralization); (2) had no metal implants (e.g., pacemaker) or metal dental work (aside from fillings) that would preclude scanning; (3) were comfortable in small spaces (i.e., not claustrophobic); (4) had eyesight (correctable with contact lenses) that would enable them to view stimuli presented in the scanner; (5) would be able to lay relatively still for 1 h; and, (6) had not been diagnosed with a substance use disorder as assessed by a single-item screening question (“Have you been told, in the last 5 years, by your healthcare provider that you have a substance use disorder?”), to which they had to respond negatively. Nine out of the 16 participants enrolled into the pilot RCT were eligible and enrolled into this sub-study.
Sample size
A power calculation was not performed given the objectives of this proof-of-concept sub-study. Rather, recruitment remained open and was tracked over a period of 12 months to assess the feasibility of year-round recruitment and data collection.
Procedures
After providing written informed consent and being enrolled into the two-arm, mixed-methods pilot RCT by the first author, all participants completed a baseline assessment (week 0) at a location of their choosing (i.e., private room at the University of Ottawa, participants’ home, local cancer support organization) that included behavioral (PA behavior; assessed via self-report and accelerometry), physical (i.e., body composition, musculoskeletal strength, muscular endurance, resting blood pressure, aerobic capacity), and psychological (i.e., self-efficacy for PA, physical self-perceptions, physical self-esteem, global self-esteem) assessments and a qualitative interview. Once baseline assessments were completed, participants were informed by the first author whether they had been randomly assigned to the intervention group or the wait-list control group. Randomization was performed by an independent researcher using a random number generator (without an established allocation ratio) and sequentially labelled envelopes. All participants then completed a mid-intervention/waiting period assessment (week 6; behavioral, physical, and psychological assessments) and a post-intervention/waiting period assessment (week 12; behavioral, physical, and psychological assessments and a qualitative interview). Throughout the two-arm, mixed-methods pilot RCT, feasibility (i.e., recruitment metrics, retention, missing data) and adverse events were tracked (and are reported elsewhere; [38]). At study cessation, all participants were entered into a draw to win a $250 CAD gift card.
Participants who were eligible and enrolled into this proof-of-concept sub-study completed the above procedures, in addition to completing fMRI scans with EF tasks at the Royal Ottawa Mental Health Centre. fMRI scans were conducted concurrent with the baseline assessment (week 0), post-intervention/waiting period assessment (week 12), and 12-week post-intervention/waiting period assessment (week 24). Six of the nine participants enrolled into this sub-study completed all scheduled fMRI scans (i.e., adherence to the scheduled fMRI scans). In addition, enrollment into this proof-of-concept sub-study, outliers and missing data on sub-study assessments, and performance on EF tasks were tracked (see Results).
Intervention group
Intervention group participants received a 12-week PA program, which was individualized using their baseline assessment results. Participants also received a yoga mat, water bottle, sweat towel, and socks (which they could keep) and were lent hand weights, resistance bands, and a Polar A300 monitor and heart rate strap (which they had to return post-intervention). Briefly, the 12-week PA intervention consisted of four weekly PA sessions, which lasted 25–45 min. The volume and intensity of each session was modified and progressed on an individual basis. Two sessions per week focused on strength activities (e.g., squats, lunges, shoulder presses) performed for 1–3 sets of 6–12 repetitions; these sessions were supervised by the first authorFootnote 1 for the first 6 weeks (at a location of participants’ choosing; i.e., private room at the University of Ottawa, participants’ home, local cancer support organization) and then were unsupervised for the remaining 6 weeks. Two sessions per week focused on aerobic activities (e.g., walking, rowing, indoor/outdoor bicycling, jogging) performed at 40–75 % of participants’ heart rate reserve. Aerobic sessions were unsupervised throughout. Participants were asked to self−monitor intensity using the Polar A300 monitor with a heart rate strap and/or a 10−point Perceived Exertion Scale.
Wait-list control group
Wait-list control group participants were asked to continue with their usual routine for 12 weeks. No restrictions were placed on their PA. After the 12-week intervention period, the wait-list control group participants received a 12-week individualized PA program in the same way as the intervention group.
Data collection
As described in Procedures, multiple assessments were completed to collect data for the two-arm, mixed-methods pilot RCT. Henceforth, only measures and methods related to the objectives of this proof-of-concept sub-study are presented. Further details related to main pilot RCT objectives are published elsewhere [38].
Sociodemographic, medical, and leisure time PA information
At baseline, participants self-reported their sex, age, age at cancer diagnosis, cancer type and treatments, education, and work status. In addition, they completed a modified version of the Leisure Time Exercise Questionnaire [40], wherein they reported the frequency and duration of leisure-time PA (i.e., PA performed during one’s free time) at mild (i.e., minimal effort; e.g., yoga, bowling, golf, easy walking), moderate (i.e., not exhausting; e.g., fast walking, tennis, easy bicycling, easy swimming), and vigorous intensities (i.e., heart beats rapidly; e.g., running, jogging, hockey, football). This information was collected to describe the sample.
Feasibility
Enrollment to the proof-of-concept sub-study, adherence to scheduled fMRI scans, outliers, and missing data
To assess the feasibility of neuroimaging with EF tasks among AYAs, the number of participants from the pilot RCT who enrolled into this proof-of-concept sub-study and reasons for declining were recorded. As well, adherence to scheduled fMRI scans, outliers, and missing data on sub-study assessments were tracked.
EF task performance data
To examine whether the EF tasks worked as intended during the fMRI scanning sessions, participants’ performance (i.e., errors and reaction times) on EF tasks during the fMRI scans was documented.
Preliminary evidence for the effect of PA on neural activity
Participants completed fMRI scans on a 3 Tesla Siemens Biograph Magnetom MR-PET scanner (Siemens, Erlangen, Germany) equipped with a 12-channel head coil. Whole brain echo planar fMRI was performed using a gradient echo pulse sequence (TR/TE 3000/34 ms, FA 90°, FOV 200 × 200 mm2, voxel size 1.6 mm × 1.6 mm × 3 mm, 48 axial slices, slice thickness 3 mm, band-width 2894 Hz). The total time for the scan was 1 h. At each assessment, the protocol was comprised of two EF fMRI tasks (described below; Letter n-back and Go/No Go), diffusion tensor imaging (DTI), and resting state fMRI. DTI and resting state fMRI results are presented elsewhereFootnote 2.
Letter n-back
During participants’ fMRI scan, a letter n-back task (designed by the third author) was presented. This task consisted of black letters presented in the middle of a white screen, one at a time, for 1500 milliseconds (ms) each with a 500 ms interstimulus interval (ISI). The block design task included two conditions: a control condition (‘Press for X’; a button press required for every X presented) and a working memory condition (‘Press for 2-back’; a button press required when the letter presented was the same as the one presented 2 letters prior). Instructions were presented for three sec before each block with ‘Press for X’ or ‘Press for 2-back’, respectively. There were no X’s presented in the ‘Press for 2-back’ blocks. Six blocks of each condition were performed with 16 stimuli presented randomly in each block. Six responses were required within each block. Rest periods were interspersed between blocks for 21 sec with the word ‘Rest’ on the screen.
Go/No Go
After the letter n-back task, while participants were still in the scanner, they completed the Go/No Go task (developed by the third author). The time between tasks was just enough to remind participants of the instructions for the Go/No Go task and to ensure they were comfortable continuing. The Go/No Go task consisted of black letters presented one at a time in the middle of a white screen for 75 ms with an ISI of 952 ms. Twelve stimuli were presented in each block, with four blocks of each condition: ‘Press for X’ (respond with button press for every X presented) and ‘Press for all letters except X’ (respond with button press when all letters other than the X were presented). Instructions were presented on the screen prior to each respective block for three sec and there were five required responses in each block. Fifty percent of the letters were X to build up a prepotent response to the X. Interspersed between the letter blocks were 21 sec rest periods with the word ‘Rest’ on the screen.
Data processing and analysis
No formal hypothesis testing for efficacy was undertaken because the aim of this proof-of-concept sub-study was not to assess efficacy, and it was underpowered for this. Rather, descriptive statistics were computed to describe participants and to report on feasibility outcomes for the enrolled sub-study sample (n = 9) and the analytical sample (n = 5), using IBM SPSS 26 (IBM Corp.). Descriptive statistics included means with standard deviations and frequencies. EF task performance data were exported from E-Prime 2.0 [41] and were visually inspected to explore potential differences in performance (i.e., errors of commission or omission and reaction time for all correct responses occurring within 900 ms of stimulus presentation) from pre- to post-PA intervention on the letter n-back and Go/No Go. In addition, exploratory paired sample t-tests were performed to examine differences in performance from pre- to post-PA intervention. The fMRI data were post-processed and analyzed using Statistical Parametric Mapping (SPM) 12. The fMRI scans for both tasks were motion corrected through realignment, normalized to the standard SPM Montreal Neurological Institute template and spatially smoothed with an 8 mm FWHM Gaussian kernel. The letter n-back task images, for each person at each time point, underwent individual participant analyses with the ‘Press for 2-back’ minus ‘Press for X’ contrast as the working memory contrast of interest. Motion correction was applied as a regressor for all first-level analyses. Baseline fMRI scans (week 0) were treated as the ‘pre-PA intervention’ data for all participants as this was the first time participants saw the scanner and completed the EF tasks. The fMRI scans from week 12 and week 24 were treated as the ‘post-PA intervention’ data for the PA intervention group and wait-list group, respectively. A paired sample t-test comparing pre- and post-PA intervention fMRI scans during working memory processing and response inhibition tasks was conducted to address objective two and ascertain if there was preliminary evidence supporting the possible effect of PA on neural activity (as detected by the BOLD signal). The Go/No Go task images were analyzed in a similar procedure with the contrast of interest for response inhibition: ‘Press for all letters except X’ minus ‘Press for X’. All pre- and post-PA intervention analyses were whole brain investigations and were conducted at a set threshold of puncorr = 0.001, with a cluster-wise correction at pFWE = 0.05 and a set cluster size larger than 10 voxels.
As described above, proof-of-concept studies typically involve small sample sizes. However, this is at the expense of decreased statistical power and potential inability to detect statistically significant effects. Thus, proof-of-concepts studies may need to deviate from the standard significance level of 0.001. To this end, a higher type I error probability was set (i.e., an uncorrected significance level of 0.05) to decrease the risk of missing a potentially beneficial effect of PA.