Effect of home-based online training and activity feedback on oxygen uptake in patients after surgical cancer therapy: a randomized controlled trial

CRBP-TS (ColoRectal, Breast, and Prostate Cancer—Telemonitoring and Self-management) was a randomized, multicenter trial involving an intervention (IG) and control group (CG) that assessed supervised and home-based post-surgery online training to strengthen physical performance and patient empowerment via automated activity feedback information after cancer surgery. A detailed description of the study design has been published [38]. The study was approved by the Ethics Committee of the Medical Faculty, University of Leipzig (reference number 056/20-ek), and at all participating sites. The patients were screened, informed, and enrolled at University Hospitals in Dresden, Hannover, and Leipzig. All participants provided written informed consent. Cancer patients with International Classification of Diseases codes C18/19/20 (colorectal cancer), C50 (breast cancer), and C61 (prostate cancer) who underwent curative (R0) surgery at stages T1N0M0 to T3N3M0 (Tumor, Nodes, Metastases; including M1 with achieved R0 resection); ECOG (Eastern Cooperative Oncology Group) < 1; and aged between 18 to 75 years were eligible to participate in this trial. In total, 148 patients were recruited and randomized (Fig. 1 and Table 1).

An online-based system was used to assign patients in a 1:1 allocation to IG or CG. Randomization was stratified by study site and cancer entity using block sizes of two and four blocks.

Home-based and body-weight online training consisted of entity-specific, individually performance-adapted, and heart rate-limited strength-endurance training using video presentations. Individual entry and follow-up levels based on patients’ individual physical performance level were used. The training was scheduled in accordance with exercise guidelines [8, 28] for two (at least) or preferably three times or more per week for 30 min per session (5-min warm-up; four rounds with five different body-weight upper and lower body exercises with 40 s loading and 20 s recovery period per set; 5-min cool down). The strength endurance exercises mainly done with the patient’s own body weight included for example stepping exercises, squats, rowing, upper body push and pull exercises, jumps, and core exercises. For the included entities, partially adapted training videos were designed (i.e., breast cancer patients avoided shoulder exercises, colorectal cancer patients avoided exercises in prone position, and prostate cancer patients avoided jumping exercises). The target training intensity was determined by the perceived exertion (target 5–8; CR10 scale) [39‐41], and adjustable after each training month’s period, because strength endurance training (interval exercise with predominantly peripheral fatigue) cannot be adequately controlled by heart rate zones. Nevertheless, a heart rate sensor was used to document exercise intensity during exercise sessions and to enable immediate feedback to patients. For forensic reasons, an individual maximum heart rate (75% heart rate max or symptom-limited heart rate) was defined using a cardiopulmonary exercise test) at baseline and was individually adjustable during a repeated cardiopulmonary exercise test after 3 months (study visit 2).

At baseline, all patients were given a wearable (Vivo active 4; Garmin, Olathe, Kansas, US) for activity tracking (steps per day, active minutes per day with an intensity > 3 MET) connected to a tablet (Lenovo Tab M10 TB-X606X; Lenovo, Hongkong, China) via Bluetooth with the CRBP-TS application. The wearable unit was to be worn 24 h a day during the entire study period. An automatic data transfer from patient to an electronic patient file (case report form) was provided via internet access (LTE, Deutsche Telekom AG, Germany) in the tablet device. The CRBP-TS application was meant to be regularly used by the patients to visualize the training video presentations, to record their heart rate (chest belt) and to receive their heart-rate feedback during the training sessions, to complete questionnaires, and to receive physical activity feedback (steps per day, activity time in minutes > 3 MET per week) during the intervention period. Structured information on general health improvement, disease prevention, and lifestyle changes (diet, training, and self-perception) was also provided by the study team (physician, sports scientist, and a study nurse) via the CRBP-TS application. In case of unscheduled exercise breaks, patients were encouraged via short message service as reminders to increase their adherence. CG patients received general information about lifestyle changes and physical activity according to guidelines and, as described in the IG, the wearable and a tablet with the CRBP-TS application to enable the receipt of activity feedback information (without video presentations).

All patients were assessed at baseline and 3 and 6 months (study visit 2 and 3) after randomization. The CRBP-TS application measured and evaluated the daily activity of patients in the IG and CG. All study subjects underwent clinical examinations according to standard operating procedures; this included medical history, anthropometry (BIACORPUS RX 4004 M, MEDI CAL HealthCare GmbH, Germany), flow-mediated dilation, blood analysis (clinical chemistry; tumor and inflammation marker panel) cardiopulmonary exercise testing, and completing the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QoLQ-C30). Subjects also underwent cardiopulmonary exercise testing to determine primary and secondary end points (custo, BT300 electrocardiogram, custo GmbH, Germany; PhysioFlow impedance cardiography, Manatec Biomedical, France; Dynostics ergo-spirometry, Sicada GmbH, Germany) on an electronically braked semi-recumbent ergometer. The initial load was 30 Watts with 10 Watts/min increments until subjective or objective exhaustion or the occurrence of termination criteria [42].

Staff members conducting the evaluations were not blinded to treatment groups. Cardiopulmonary exercise capacity was tested according to current recommendations [42] and analyzed in blinded manner at the study core laboratory in Leipzig. Maximum oxygen uptake (V̇O_2max) was defined as the highest 30-s average within the last minute of exercise. Blood parameters were analyzed at the central core laboratory (Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University Hospital Leipzig, Leipzig, Germany).

Our primary end point was the change in V̇O_2max after 6 months (study visit 3). Secondary end points included changes from baseline to 6 months for cardiopulmonary exercise testing parameters (cardiac output [CO], rate pressure product [RPP], and peak power output), anthropometric parameters (body mass index [BMI], body cell mass [BCM]), C-reactive protein (CRP), and the quality of life (EORTC QoL-C30). Activity parameters (steps per day, active minutes > 3 MET per week) were recorded during the entire study period in IG and CG. The additional secondary end points of changes in flow-mediated dilatation, blood parameters (inflammation panel, tumor makers, liquid biopsies, and metabolic markers), questionnaires (Patient Health Questionnaire-2; Depression Anxiety Stress Scale; Fatigue Severity Scale and Oral Health Impact Profile), and other anthropometric parameters (fat mass, lean body mass) from baseline to 6 months are not reported here. Serious adverse events (SAE) were documented and categorized at each study site and then evaluated. To identify SAEs, participants were asked to self-report any health problems during the study period and were questioned about events by the site investigator at each study visit. Selection criteria were death, life-threatening, hospitalization, disability, or permanent damage. SAE reports were assessed according to GCP ICH by the study site principal investigator (temporal or exercise association with training; alternative relationship, e.g., accident or new illness) and reported to the head of clinical trial.

The trial protocol defined 13% change in V̇O_2max in the IG compared to the CG as clinical or substantial importance [43, 44]. By an assumed mean difference of 3.6 ml/kg/min between home-based online training and the CG, 80% power and an alpha of 0.05, a sample containing 40 patients per group was needed. Considering probable dropouts and a moderate number of missing values, we aimed to include 100 patients (50 in IG and CG) in this study. To enable a subgroup analysis per entity (breast, prostate and colorectal cancer), we needed to enroll 300 patients.

Data were analyzed using IBM SPSS Statistics (Version 29; IBM, Armonk, New York, USA) and displayed using GraphPad Prism (Version 9; GraphPad Software Inc., California, USA). The Shapiro–Wilk test was used to analyze the sampling distribution. The evaluation was conducted on an intention-to-treat basis, and all randomized participants were included. Available data on participants with missing data were included under the “missing at random” assumption. Per-protocol analyses were conducted, including only participants in the IG who completed all study visits and who had engaged in at least 1.5 training sessions per week.

To evaluate the primary and secondary end points, we applied mixed-effects models with repeated measurement structure (estimated using restricted maximum likelihood). In this model, the measured values (measured at baseline, 3-month, and 6-month follow-up) were treated as the dependent variable. As fixed effects, we have included the randomization arm and categorical time covariate in the model. Interactions were modeled for group and time (categorical). As random effect(s), we had an intercept for subjects. Within the mixed models, we estimated 95% confidence intervals (CI) and p-values for contrasts between groups for the 3- and 6-month periods. In a sensitivity analysis, we included only those patients with complete paired baseline and 6-month follow-ups for time difference within groups (paired t test for dependent samples). Sphericity was checked using Mauchly’s W test, and Greenhouse–Geisser correction was applied when necessary. All analyses were two-sided, and the level of significance was p = 0.05.

The mixed effect model of repeated measurement of maximal oxygen uptake showed a significant time effect (p < 0.001), a p-value of 0.056 for interaction effect and no group effect (Table 2). After the 6-month intervention, the estimated change in V̇O_2max differed significantly between groups 1.24 [95% CI: 0.23 to 2.25: p = 0.017] mL/kg/min. The IG’s increase in V̇O_2max from baseline to 6 months was significant (mean [SD] for IG: 1.82 [2.7] mL/kg/min), unlike the CG’s (mean [SD] for CG: 0.66 [3.5] mL/kg/min). Figure 2 displays the change in V̇O_2max across the study visits at 3 and 6 months. The estimated change at the 3-month time point showed no significant group difference. Subgroup analysis per entity for the change in oxygen uptake is reported in Table S1 (additional file 1) and 3-month visit data are presented in Table S2 (additional file 1).

The increase in peak power output after 6 months showed a significant time effect in the mixed model (p < 0.001) and within groups without significant differences between IG and CG (3.8 W [95% CI, − 1.9 to 9.5]). However, after 6 months, the change in the rate pressure product (RPP) was significantly lower in the IG than the CG (− 1079 [95% CI, − 2157 to − 1]; Table 2, Fig. 2B) and showed an interaction effect in the mixed model analysis (p = 0.04). Nonsignificant increases during the intervention period in cardiac output (CO) were observed within the IG and in the global mixed model analysis with no group difference (0.4 l/min [95% CI, − 4.2 to 3.4]; Table 2). Change in BMI after 6 months was not significant and showed no group differences (Table 2). The difference between the change in IG und CG was 0.3 [95% CI, − 0.6 to 0.03]. There was no significant difference in either the change in C-reactive protein (CRP) within groups over time or between groups (Table 2, Fig. 2D). Nevertheless, body cell mass (BCM) rose significantly in both groups, although no group differences were evident (Table 2). The change in quality of life (QoL) revealed a significant increase over time in the CG, but this did not differ significantly between groups (Table 2). The activity data (activity in minutes per week > 3 MET) revealed no group difference between IG and CG patients in the (Table 2). The CG showed a decrease in activity behavior over time (− 48 min per week > 3 MET). There was no interaction effect. Patients randomized to IG demonstrated throughout the study period (week 1 to week 25) 128 (SD ± 135) minutes of total activity > 3 MET and 29 (SD ± 32) minutes activity with intensity exceeding 6 MET per week. The CG revealed a mean of 142 (SD ± 122) minutes total activity > 3 MET and 31 (SD ± 37) minutes activity > 6 MET per week during the 6-month study period (week 1 to week 25). App usage in both groups decreased significantly during the study period and differed between groups but showed no interaction effect (Table 2). Figure 3 illustrates the course of activity (steps per day; active minutes > 3 MET) and app usage per week during the entire study period in IG and CG.

One hundred twenty-two patients completed the 6-month study period (62 in IG, 60 in CG). Dropouts (IG: n = 14; CG: n = 12) were mainly for clinical reasons (n = 9; hospitalizations or worsening health) or consent withdrawal (n = 17; motivational problems [n = 6], difficulty handling the devices [n = 6], other reasons [n = 5]). Patients randomized to the IG participated in a mean of 2.1 (SD ± 1.1) training sessions per week in the intervention period.

We documented 18 adverse events in 16 patients (11%) classified as serious adverse events (SAE) that were unrelated to the exercise intervention (no temporal relation to training, other cause such as accident, new disease diagnosis or scheduled surgery; IG 10 [13%] patients; CG 6 patients [8%]; chi-square p = 0.35). Hospitalization or disease progression was the cause for being classified as an SAE in all these cases. Reasons were scar hernia (n = 5; without timely relation to the training), metastases or recurrence (n = 3), accidents involving a trauma to the passive musculoskeletal system (n = 3), additional carcinoma (n = 2), stoma relocation (n = 2), pulmonary embolism (n = 1), peritonitis (n = 1), and urethral stenosis (n = 1).