This 2-armed, parallel-design pilot RCT evaluated the effects of an 8-week, remotely delivered combined (aerobic + resistance training) exercise program compared to an active control group. Participants were randomized in a 1:1 allocation ratio. Reporting of the trial outcomes followed the CONSORT-eHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth; Multimedia Appendix 1) 2011 guidelines [45,46].

Participants were recruited from cancer care organizations, support groups, university listserves, an existing database of research individuals who participated in prior research studies in the laboratory, and social media advertisements (ie, Facebook [Meta Platforms, Inc], Instagram [Meta Platforms, Inc], and X [X Corp]) across Canada between February and July 2023. Eligibility criteria included (1) 40-65 years of age, (2) a history of stage I-III BC (nonmetastatic), (3) a history of receiving chemotherapy within 48 months prior to enrollment, (4) mild cognitive impairment as determined by the Modified Telephone Interview of Cognitive Status (TICS-M) [scores between 21 and 24 to separate individuals with mild cognitive impairment and normal cognition (>24)] [47], (5) self-reported low-activity, defined as <3 days of exercise (<20 min/d) per week in the previous 6 months [48], (6) physician clearance to participate in exercise if required, (7) no previous invasive cancer, (8) no neurological or musculoskeletal comorbidity inhibiting exercise, (9) access to a webcam and internet for videoconferencing, and (10) fluency in English. Participants who were interested in the study contacted the study team to schedule a screening call to further assess their eligibility. Additionally, participants were asked to provide their tentative availability for attending 2 live, remotely delivered classes each week.

Eligible participants were randomized in a 1:1 ratio to either the combined exercise group or the active control group using REDCap (Research Electronic Data Capture; Vanderbilt University) [49,50] by a graduate research assistant (NC). Randomization was conducted upon completion of the informed consent form, all baseline measures (ie, PA) assessment, and a battery of questionnaires. Participants were blinded to the study hypotheses and were only informed of their assigned group. Outcome assessors were kept blinded to the allocation.

Participants in both groups participated in an 8-week, remotely delivered, supervised exercise program conducted via videoconferencing (Zoom [Zoom Communications, Inc]). The program consisted of two 30-minute live sessions and one 30-minute recorded session per week (Figure 1). Recorded sessions aligned with live classes to maintain consistency and allowed for individualized adjustments. All live sessions were led by Qualified Exercise Professionals (QEPs) certified as Registered Kinesiologists or through the American College of Sports Medicine (ACSM) or Canadian Society of Exercise Physiologists (CSEP). Participants were encouraged to keep their cameras on during live sessions to facilitate safety monitoring and provide real-time exercise modifications.

Participants received resistance bands, a Fitbit Inspire 2 heart rate (HR) monitor, and an exercise log. The program included 30 minutes of unsupervised aerobic exercise (eg, walking) 3 times per week, gradually increasing intensity and duration from 40% to 59% of the maximum HR reserve initially to 60%-70% by the end of the program. The program also included 30-minute supervised resistance training sessions 2 times per week. Resistance training focused on major muscle groups using 2-3 sets of 8-12 repetitions at a Rating of Perceived Exertion (RPE) of 12-15, progressively adjusted with stronger resistance bands. A 30-60 second rest was provided between sets and 60 seconds between exercises, during which the QEP demonstrated the next exercise and offered modifications. Participants recorded their HR and RPE in exercise logs after each session. Participants also attended 4 remotely delivered, biweekly, 30-minute behavioral counseling sessions based on the multi-process action control (M-PAC) framework, emphasizing goal-setting, habit formation, and behavioral control to support long-term PA (Table 1 [51-54]).

The active control group participated in a low-intensity, whole-body stretching and toning program designed to improve balance and flexibility. Exercises targeted major muscle groups potentially affected by cancer treatments, such as steroid use, radiation therapy, or surgery [55]. Flexibility exercises, incorporating both static and dynamic movements, were introduced weekly to maintain engagement and progress using resistance bands. Intensity was maintained at a light level (RPE 9-10) [56]. Participants recorded their HR (using Fitbit devices) and RPE in exercise logs after each session.

Data analyses were performed using IBM SPSS Statistics (version 29). As this was a feasibility study, an a priori power calculation was not performed [68,69]. Descriptive statistics (ie, means and SDs) characterized the sample for demographic, clinical, and feasibility outcomes. z-scores were used to identify any outliers in the field. Analyses of covariance (ANCOVA) were used to assess both primary and secondary outcomes of the study. The dependent variable was the mean difference of each intervention outcome (ie, the difference between post- and preintervention scores), while the independent variable was the group (ie, combined exercise or active control). All analyses were conducted while controlling for the baseline values of each outcome. All analyses were conducted on an intention-to-treat basis. Given that the purpose of this feasibility trial was to inform a larger RCT, outcomes were interpreted for potential clinical significance based on the direction and magnitude of numerical differences. Partial eta squared (ηp²) values were reported to describe the observed effect sizes.

The trial protocol was approved by the Research Ethics Board at the University of Toronto (number 43675), and all participants provided written informed consent. Participant privacy and confidentiality were strictly maintained throughout the study; all data were deidentified and stored on secure, password-protected servers accessible only to authorized research personnel, in accordance with institutional and ethical guidelines. Participants received a total of $20 CAD compensation, plus a Fitbit Inspire 2 for their time and participation, consistent with the approved ethics protocol.

Demographic and clinical characteristics of the participants are summarized in Table 2. Overall, all participants were women (21/21, 100%), predominantly White (16/21, 76%), with an average age of 51.9 (SD 7.2) years. The majority were married (17/21, 81%), some had completed university or college (10/21, 48%), and many were employed full-time (13/21, 62%). All participants had chemotherapy and were, on average, 22.9 (SD 13.9) months post diagnosis and 11.8 (SD 12.9) months post treatment. Regarding weekly exercise minutes, participants engaged in 149.4 (SD 96.1) moderate-to-vigorous intensity exercise minutes per week.

Participant flowchart is presented in Figure 2. Feasibility outcomes are summarized in Table 3. Of the 41 women who responded to the invitation, 22 were eligible, and 21 consented to be randomized into either the combined exercise group (aerobic+resistance training; n=10) or the active control group (balance and flexibility; n=11), representing a 51.2% enrollment rate. Participants were divided into two cohorts: (1) the first cohort (n=11) began in May 2023, and (2) the second cohort (n=10) began in July 2023. The sample size aligns with previous research on feasibility and group cohesion for counseling programs for patients with BC [70,71]. Overall, 18 participants completed the 8-week intervention. Three participants withdrew after randomization, but before starting the intervention for reasons, including a change in BC treatment plan (n=1), an issue with the timing of the programming and their schedule (n=1), and a change of interest in participating (n=1). An intention-to-treat analysis was conducted, and therefore adherence was calculated based on attendance and included the dropouts (n=3), since they were enrolled and randomized at the start of the intervention. Consequently, the attrition rate was 14.3%. For the ANCOVA analyses, a complete case analysis was applied (n=18).

Attendance adherence (Table 3) rate for the remotely delivered, combined exercise classes was 70.8% (17/24) and 77.5% (3/4) for behavioral counseling sessions based on the M-PAC that was delivered only to those in the combined exercise group (n=10). Attendance adherence for the active control group was 70.8% (17/24). Participants in the combined exercise group who completed the intervention and postintervention measures (n=9) reported an average RPE of 12.8 and a HR of 111.8 during the resistance training classes. This corresponds to a moderate-intensity RPE and an HR within the light-moderate target HR zone. In addition, participants in the combined exercise intervention (n=9) reported a mean RPE of 12.2 and a mean HR of 119.7 during the unsupervised aerobic (ie, walking) sessions, which corresponds to moderate-intensity RPE and a HR within the moderate-intensity target HR zone. Participants in the active control group who completed the intervention and postintervention measures (ie, including the exercise log; n=8) reported a mean RPE of 11.1 and a mean HR of 85.5 during their balance and flexibility program, which corresponds to a light-moderate intensity RPE and light-intensity target HR zone. There were no adverse events during the intervention in either group. The measurement completion rates for the study were as follows: 85.7% (18/21) for the participant satisfaction questionnaire and 81% (17/21) for the Working Alliance Inventory. For objectively measured cognitive function, completion rates were 85.7% (18/21) for the memory test and 81% (17/21) for the executive function test. Additionally, accelerometer wear compliance was achieved by 66.7% (14/21) of participants.

Participant responses regarding program satisfaction are summarized in Tables 4 and 5. Overall, participants reported high satisfaction with the intervention. The majority of participants (17/18, 94.4%) agreed or strongly agreed that the program was rewarding and appreciated the virtual delivery format. Most participants (15/18, 83.3%) agreed or strongly agreed that the intervention helped them increase their PA levels, and 77.7% (14/18) indicated that they would recommend the program to other patients with BC. Most participants found the intervention beneficial, and participants 62.5%) agreed or strongly agreed that the behavioral counseling sessions were personally useful. Similarly, 61.1% of participants agreed or strongly agreed that the intervention was not burdensome.

Participant responses on the WAI-SR are summarized in Table 6. Overall, participants reported positive perceptions of their alliance with the QEP. Most (16/18, 88.2%) indicated that they very often or always felt mutual respect, and a substantial majority (14/18, 76.5%) felt appreciated and cared for.

Changes in objectively measured cognitive function scores from baseline to post intervention for the combined exercise group vs the active control group are presented in Table 7. There were no statistical differences in objectively measured cognitive function in the combined exercise group compared to the active control group. However, adjusted group mean differences trending toward the combined exercise group were observed in the scores for the Picture Sequence Memory Task (mean difference +5.33, 95% CI –12.5 to 23.2; ηp²=0.03), List Sorting Working Memory Test (mean difference +8.17, 95% CI –4.2 to 20.6; ηp²=0.12), and Auditory Verbal Learning (mean difference +3.22, 95% CI –2.0 to 8.5; ηp²=0.12). Table 7 also shows the 4 cognition outcomes from the PsyToolkit: repeat trial time (ms), switch trial time (ms), repeat trials correct (%), and switch trials correct (%). No significant differences were found between the combined exercise group compared to the active control group. However, a reduction in repeat trial time trended toward the combined exercise group (mean difference –146.14, 95% CI 420.87 to128.59 ms; ηp²=0.04). Likewise, a greater decrease in the switch trial time trended toward the combined exercise group (mean difference –394.35, 95% CI –1035.67 to 246.96 ms; ηp²=0.07) compared to the control group (mean difference –124.39 95% CI –522.06 to 273.27 ms; ηp²=0.07). For accuracy, improvements in repeat trials correct (mean difference +4.86%, 95% CI –0.61 to 10.36; ηp²=0.02) and switch trials correct (mean difference +19.70%, –0.30 to 39.70; ηp²=0.02) also trended towards the combined exercise group, while the control group exhibited decreases in these measures.

Table 8 presents the changes in device-based exercise levels among patients with BC. There were no statistical differences between the two groups. However, an increase in total moderate-to-vigorous intensity exercise minutes was observed from pre to post intervention, trending toward participants in the active control group compared to those in the combined exercise group (mean difference in change 58.8, 95% CI 105.8.8-223.6 minutes; ηp²=0.06).

To our knowledge, this is the first study to pilot a remotely delivered, supervised combined exercise program for patients with BC following chemotherapy, exhibiting mild cognitive impairment. The findings of this study will be used to inform larger, definitive RCTs by examining which aspects of the trial were feasible, as well as further refinements that need to be made. Specifically, there are few exercise training studies that target BC treatment-related cognitive impairment as a primary outcome. Recent reviews examined evidence for the use of exercise as an intervention for CRCI [25,31,72]. However, these studies did not consistently measure cognitive function as a primary outcome, and less than half assessed cognitive function with well-normed objective measures. The varied results within neuropsychological testing and the recent emphasis on individualized care all highlight the critical need to identify patients with BC who are especially at risk for cognitive decline. Our pilot is noteworthy, as it addressed some of these conceptual and methodological issues that may advance progress in the field. These include having an active control condition to match contact time, examining different types of exercise, such as combined exercise, including computer-based objectively measured cognitive function, as well as theory-based, behavioral counseling for exercise uptake and adherence.

This study was feasible and exceeded the a priori feasibility indicators of adherence rate ≥70%, attrition rate <30%, and no adverse events, and was well received by patients with BC. Adherence rates in our remotely delivered trial were consistent (63%-100% [28,30,73]) or lower (88% [72]; 87.9% [26]; 80% [74]; 99.9% [75]) than previous exercise trials targeting CRCI. However, it is important to note that these exercise trials were supervised (in person) sessions that included different exercise modalities, such as high-intensity interval training [28,73], combined exercise with Nordic or power walking [30], home-based walking [71], and aerobic exercise [26,72,75]. Nevertheless, our adherence rates are encouraging when compared to other virtually supervised exercise interventions delivered via Zoom in a group format, with various combinations of aerobic and resistance exercises, where adherence ranged from 78% to 100% [76]. Our trial also had low attrition rates, which were consistent with prior research with remotely delivered exercise interventions (0%-29%; [76]).

Recruitment rates for virtually supervised exercise interventions vary widely, with reported rates between 8% and 60% [76] likely due to the heterogeneity of recruitment methods. While our recruitment rate (51.2%) was consistent with previous research, our study relied mainly on self-referrals, which tended to attract highly motivated cancer survivors who were well-educated, White, and in better overall health and physical functioning. Homogeneous samples (eg, White, young, and highly educated) are a common limitation across distance-based exercise oncology trials [77]. Recruitment strategies and partnerships with organizations that serve racial and ethnic populations of cancer survivors should be strengthened at the onset of the intervention, which may assist with recruitment efforts [78].

In our secondary outcomes, no statistical differences were observed in objectively measured cognitive function. Small to medium effect size improvements were observed in episodic memory (Picture Sequence Memory Task), working memory, and executive function updating (List Sorting Working Memory Test), and immediate memory and verbal learning (Auditory Verbal Learning), trending toward the combined exercise group. On the other hand, large effect size improvements were observed in crystallized abilities (Oral Reading Recognition Test) plus language and vocabulary (Picture Vocabulary), slightly trending toward the active control group. For executive function shifting, greater accuracy and reductions in reaction time (small to medium effect sizes) were observed for the switch trials in the combined exercise group. To date, few studies have tested the effects of exercise on objectively measured cognitive function outcomes in patients with BC following chemotherapy, with mixed results [40,72,75]. Preliminary results from these trials support cognitive changes following an exercise intervention, which is consistent with a study that examined the effects of a 3-month PA intervention compared to a waitlist control arm on cognitive functioning in patients with BC [43]. The intervention improved processing speed only among patients with BC diagnosed within the past 2 years. One proof-of-concept aerobic exercise intervention examined cognitive function using objective neuropsychological testing and functional magnetic resonance imaging in patients with BC. This study found improvements in cognitive symptoms with notable patterns of neural activation. However, the changes either did not reach statistical significance or had small effect sizes [72]. Finally, one of the first large-scale studies to examine the effects of aerobic exercise on neurocognitive function tested the effects of 6 months of moderate-intensity aerobic exercise on neurocognitive function in patients with BC receiving endocrine therapy. The trial focused on women who initiate exercise within 2 years of completing primary therapy (surgery +/− chemotherapy) [75]. Findings show improved performance in processing speed and verbal memory in the exercise group and no change in the usual care control group. It is important to note that these studies were either home-based aerobic programs or in person supervised aerobic activity. They also used a wide range of cognitive measures that were collected in person, whereas our study used a cognitive battery that was assessed remotely.

Recent reviews on the use of exercise as a potential intervention for cognitive deficits associated with cancer and its treatment demonstrated notable improvements in the cognitive domains of working and visual memory, attention and concentration, cognitive flexibility, inhibitory control, and verbal fluency [26,79], consistent with our study. Null effects were observed in other RCTs with a walking intervention [74], exercise intervention with supervised aerobic and strength training, as well as Nordic or power walking [30], and interval and continuous aerobic exercise in patients with BC following chemotherapy [73]. While findings from previous trials have been inconclusive, the positive improvements in cognitive function reported in our study could be attributed to recruiting patients with BC who exhibit mild cognitive impairment at the onset of the intervention. This may have provided greater room for improvement post intervention. Also, a range of cognitive assessments measuring cognitive domains that are most impaired by chemotherapy (eg, memory, attention, and executive function) increased the sensitivity to detect impairments post intervention. Finally, patients with BC in the combined exercise intervention were adhering to the prescribed exercise dose (ie, moderate-to-vigorous intensity physical activity [MVPA]), which is critical in achieving greater cognitive benefits and is consistent with reviews demonstrating that sustained MVPA elicits changes in cognitive function in patients with BC [3,25,80]. One of the most important mediators of cognition in patients with BC may be cardiorespiratory fitness [81,82]. Exercise-induced changes in cognitive function can be explained by alterations in brain morphology and functional dynamics (ie, prefrontal cortex and hippocampus) and may be partially mediated by improvements in cardiorespiratory fitness [82]. This underscores the pivotal role of cardiorespiratory fitness in amplifying the neuroprotective effects of MVPA. Given the link between cognition and cardiorespiratory fitness and the established effects of exercise on cardiorespiratory fitness, exercise might also remediate CRCI in patients with BC.

Notable improvements were also observed in the active control group performing light-intensity exercise, but these improvements were not in the cognitive domains most impaired by chemotherapy. The active control group engaged in a light-intensity exercise, which is insufficient to elicit improvements in cardiorespiratory fitness. Previous reviews demonstrated that both aerobic and resistance training that were moderate-to-vigorous intensity may be required to elicit cognitive benefits [3]. Nevertheless, improvements in the Oral Reading Recognition Test and Picture Vocabulary Test trended toward the active control group (although exploratory). These domains are related to language and/or crystallized abilities, and it is possible that increases in cognitive performance likely reflect expectancy or practice effects from repeated exposure to the assessments. Therefore, the remote administration of the NIH Toolbox Cognition Battery may not fully capture subtle CRCI-related changes.

In terms of changes in exercise minutes, improvements in MVPA from pre to post intervention trended toward the active control group (although exploratory). This finding is inconsistent with previous studies, which demonstrated improvements in device-based exercise minutes in patients with BC [30,43]. Given that we had an active control group, both groups were motivated to engage in exercise. Accelerometers do not accurately measure resistance training, which may not have been captured in patients with BC, especially in the combined exercise group, who continued engaging in resistance training. Patients with BC in the combined exercise group may have engaged in less structured exercise outside of the planned classes because they were receiving a moderate-intensity multicomponent exercise program compared to the active control group. Therefore, the reported effect sizes between our groups are possibly smaller given the active comparator group. Also, patients with BC self-reported not meeting exercise guidelines at screening but subsequently showed higher levels of activity with device-based measures at baseline. Future research should consider using device-based measures to screen patients with BC not meeting exercise guidelines to ensure that the inclusion criteria are met.

Overall, patients with BC were highly satisfied with the intervention. Most patients with BC found the study rewarding, enjoyed the virtual delivery format, would recommend the study to other patients with BC, and did not find the study assessments burdensome. Patients with BC also felt a mutual respect with their QEP and felt appreciated by their QEP, further contributing to their positive experience. This suggests that few adjustments need to be made to the intervention or the study design before implementing a future RCT. In our study, an additional QEP attended the classes to monitor participant safety and to assist the lead QEP in managing the virtual environment. This was beneficial to both the lead QEP and patients with BC, and future remotely delivered interventions should consider a second QEP. Additionally, the QEP’s personality, approachability, and teaching style played an important role in the enjoyment and feasibility for patients with BC. Future studies should consider QEPs with cancer-specific training to offer exercise modifications to patients with BC with limitations or injuries. Finally, our classes included 15 minutes of optional time before or after each class for patients with BC to interact for overall class enjoyment and accountability and to help foster a sense of community. This was beneficial to the overall satisfaction of the trial and should be implemented in future studies.

Our trial should be interpreted within the context of important strengths and limitations. The remotely delivered component of the trial was a strength, as it removed preexisting barriers to exercise (eg, time constraints, access, and hesitancy toward public space). The combined exercise and active control group sessions were delivered by knowledgeable QEPs who were engaging and provided modifications that allowed all patients with BC to participate in exercise, despite any physical limitations. The addition of a recorded session reduced participant burden by providing flexibility and autonomy. Our study was strengthened by the inclusion of behavioral counseling sessions, which were guided by the M-PAC framework to encourage exercise adoption and adherence [51]. The inclusion of a rigorous comparison group was an additional strength since both groups received supervised, remotely delivered exercise programs at differing intensities. Finally, the use of a standardized, well-validated battery of cognitive tests via the NIH Toolbox was another strength and in line with the ICCTF’s recommendations [6].

Although this study had a small sample size and a heterogeneous population, which limits the generalizability of the findings, this feasibility study is crucial for calculating sample sizes and refining methodologies for larger, future trials. In addition, the sample was predominantly White, highly educated, middle- to older-aged, and within the early cancer survivorship period. The focus on early-stage patients with BC within 48 months of completing treatment may not be generalizable to other cancer populations, such as individuals with metastatic disease or patients who are pretreatment, undergoing treatment, or immediately post treatment. Furthermore, the heterogeneity in the time since treatment in our study may preclude identifying the optimal exercise timing to prevent or ameliorate CRCI, which requires future investigation. It is important to note that patients with BC may be more likely to seek out care for cancer rehabilitation and view themselves as research participants compared to patients with BC who are more culturally and socially diverse populations. Future research should consider the recommendations for increasing racial and ethnic diversity in cancer clinical trials by the American Society of Clinical Oncology and Association of Community Cancer Centers Joint Research Statement [78]. Another limitation of this study was not accounting for menopausal status, which could be a confounding variable that may impact cognitive function in patients with BC. Future studies should consider chemotherapy-induced menopause as a possible mechanism for CRCI. Furthermore, our study did not include self-reported cognitive function, which is another important indicator of CRCI, as it taps into day-to-day cognitive performance in relation to cancer and its treatments [6,83]. Therefore, including both self-report and objective cognitive function assessments in future research is warranted.

In conclusion, this feasibility study provides preliminary evidence that a remotely delivered, combined exercise program is feasible and may improve objectively measured cognitive function. There are a few exercise studies that have rigorously evaluated CRCI as a primary outcome in patients with BC. The limited number of studies conducted makes it difficult to draw conclusions about the efficacy of exercise for improving cognitive function in patients with BC. There is a need for rigorous and well-designed RCTs. While the use of shorter exercise protocols like our study provides preliminary evidence for cognitive benefits, long-term exercise is more efficacious in addressing age-related cognitive decline and improving cognition [84]. Future trials should include active control conditions, test multicomponent exercises (ie, aerobic and resistance training), and examine the mechanisms underlying the relationship between exercise and cognitive function. Given that CRCI is an emerging cancer survivorship issue, the next step for this work is to conduct a larger RCT that is adequately powered to examine the effects of a remotely delivered, combined exercise intervention on cognitive function in patients with BC following chemotherapy. If replicated, these findings have the potential to identify effective and modifiable management strategies to enhance cognitive health in patients with BC.