Monitoring cancer-related fatigue and quality of life in breast and prostate cancer patients after primary treatment: a study protocol for the REBECCA trials in Norway.

Background
Clinical research is undergoing a revolutionary change. The use of electronic health records (EHR), digital registries, e-health and m-health services, smartphones and the widespread use of portable devices, including wearables, have led to an increasing amount of health, lifestyle and online data that can be continuously collected for each patient. Such data are referred to as “real-world data” (RWD) [1] and are routinely collected from heterogeneous data sources outside any controlled experimental conditions. RWD can capture rich information about the patient’s daily life, including physiological and behavioural measurements from wearables (e.g., heart rate, respiration, physical activity, stress, exercise sessions), mobility patterns inferred from smartphone location signals (e.g., preferred transportation method, time spent at home or at work), and patient-reported outcomes collected regularly through mobile applications (e.g., with weekly or even daily frequency). RWD offers great opportunities for advancing clinical research [2] but so far this opportunity has not been exploited to a great degree.
Randomized controlled trials (RCTs) are currently the gold standard for studying underlying disease mechanisms and assessing the efficacy of different treatments. Although the RCT study design is very valuable and necessary, it also has some limitations. Among these limitations are the gap between the RCT and clinical practice [3], the combinatorial explosion due to large number of factors leading to reduced statistical power [4, 5], and questionable feasibility since uncontrolled progression of comorbid conditions may reduce the sample size. Accordingly, researchers and health authorities around the world (e.g., EMA and FDA) now recognize the opportunities offered by the emergence of RWD and call for “Real-World Evidence”. Despite this important development, recent studies show that current practice is far from exploiting the potential of RWD to gain a better understanding of complex chronic diseases or assessing the safety and efficacy of various treatments [6]. In the case of cancers with a high survival rate, such as breast cancer, linking clinical research with the clinical practice through RWD for improving quality of life (QoL) is highly warranted [7].
In recent years, there has been increased awareness of the late side effects following cancer development and cancer treatment. Breast and prostate cancer are among the diseases where patients experience early, late and chronic side effects such as cancer-related fatigue (CRF), caused in part by the cancer treatment [8]. CRF is defined, by the National Comprehensive Cancer Network Fatigue Guidelines Committee [9], as “a distressing, persistent, subjective sense of physical, emotional and/or cognitive tiredness, related to cancer development or cancer treatment, that is not proportional to recent activity and interferes with usual functioning”, and this tiredness is not relieved by rest or sleep. Moreover, CRF can persist for years after completion of treatment in otherwise healthy cancer survivors [10], and as a result many patients have difficulties in resuming their personal and professional commitments after the cancer [11]. Therefore, surviving cancer becomes a life-changing event for some patients, significantly affecting both their QoL and functional capacity [12]; [13]. In this regard, the QoL of cancer patients has been an active research topic [14], but the results have not yet been translated into improved guidelines for long-term patient management. As of today, Patient Reported Outcome Measures (PROMs) are the classical method used for assessing QoL in cancer patients. However, the use of RWD, collected using wearables and apps, in clinical trials gives an opportunity to objectively monitor patients’ QoL in real time. Through this approach we will be able to characterize the challenges the patients face in everyday life in terms of work participation, physical activity, social activity, sleep, and mental health. This insight will, in turn, make it possible to provide patients with a more personalised follow-up, as well as provide knowledge that could potentially help to formulate future guidelines for the rehabilitation of cancer patients.
In this protocol, we present the REBECCA project, which is an abbreviation for “REsearch on BrEast Cancer induced chronic conditions supported by Causal Analysis of multi-source data”. The rationale behind REBECCA is to contribute new clinical knowledge about patients’ QoL and treatment, and to test a more comprehensive and personalised follow-up of breast cancer patients based on the collection of multi-source RWD. In addition, we collect biological samples to search for biomarkers that can provide new biological knowledge related to CRF as the biological mechanisms associated with the development of CRF are still unknown.

Methods

Ethical approvals
All clinical trials conducted at Stavanger University Hospital (SUH) in this project have been approved by the Regional Ethics Committee of the Western Health Authority (REK Vest) under the IDs 225,855 (REBECCA-1), 242,088 (REBECCA-2) and 619,903 (REBECCA-3). All trials have also been registered at https://www.clinicaltrials.gov/ under NCT05587777, NCT06120595 and NCT06435091.

Consent
Patients are informed about the study at the Department of Surgery, SUH at the time of diagnosis, and written consent is obtained from those who wish to participate. A trained research nurse certified in good clinical practice receives the consent from the patients and stores it according to ethical guidelines.

Study population
SUH serves as the only hospital for a population of approximately 360 000 inhabitants in the South-Western part of Norway. All breast cancer patients that are treated at SUH and meet the inclusion criteria, are informed about the REBECCA studies and invited to participate.

Inclusion and exclusion criteria
Inclusion criteria for participation in the REBECCA-1 and − 2 study are women between 19 and 80 years of age who are diagnosed with histologically detectable M0 breast cancer (stage 0-III) requiring neoadjuvant or adjuvant endocrine therapy, chemotherapy and/or radiotherapy and who can understand the protocol and participate in the follow-up plan.
Exclusion criteria are male breast cancer patients, patients with a previous cancer diagnosis (except skin cancer treated only by surgery), patients who have previously been treated with chemotherapy/radiotherapy, and patients that do not have a mobile smartphone. Non-native speakers with limited Norwegian proficiency are also excluded from participation.
The inclusion criteria that apply for participation in the REBECCA-3 study are males under 80 years with histologically detectable M0 prostate cancer requiring primary surgery or primary radiotherapy followed by hormonal therapy and who can understand the protocol and participate in the follow-up plan.
Exclusion criteria for the REBECCA-3 study include patients with a previous cancer diagnosis (except skin cancer treated only by surgery), patients who have previously been treated with chemotherapy/radiotherapy, non-native speakers with limited Norwegian proficiency, and patients without a mobile smartphone.

The REBECCA platform
In REBECCA we monitor the QoL of breast cancer patients who have undergone primary treatment through mobile and wearable applications, as well as through a web interface and browser plugin [15]. Participants are trained in the use of the wearable device (smartwatch), the REBECCA mobile apps and the web browser plugin before use. The system is unobtrusive, recording the patients’ behavior and certain online interactions. The system also has a lightweight mechanism for patients to report their symptoms including fatigue, pain, tiredness and sleep, as well as emotional symptoms related to depression, anxiety, and fear of recurrence. Patients may also authorize companions to provide data about their physical and emotional state, such as family members and friends, through a mobile companion app. The companion app is intended to provide the medical doctor/researchers with additional information about the patients’ health, with a particular focus on their fatigue status. Use of the companion app is voluntary. Internally, these apps collect raw data from multiple sources and (after local processing) upload them to the REBECCA server located in Germany for data storage and further processing. Sensitive information such as social media interactions and location data is processed locally (at the patients’ edge devices) to extract aggregated indicators, for protecting the privacy of individuals. Clinical data including diagnoses, clinical examination data, and treatment schedules are registered in REDCap and subsequently synchronized to the REBECCA server via the REDCap API for inclusion in the big data analyses. Data about the local urban environment is also collected in REBECCA, as well as the socioeconomic context of the individual through external sources, such as maps and GIS data, aerial photography, street photography, geotagged social media as well as statistical authority data. A conceptual overview of the REBECCA platform is illustrated in Fig. 1.

The system uses the collected data in conjunction with large volumes of retrospective datasets to produce explainable causal models that i.) assess the safety and effectiveness of therapies, ii.) provide real-world evidence about the causal mechanisms between factors and outcomes and about the interaction of comorbidities [16]. Exported data and analysis results are provided to the medical doctor to support improved patient management through data-driven clinical decision support including i.) visual analytics of the patient’s lifestyle and behaviour and ii.) identification of changes in functional and emotional status that can potentially indicate underlying medical conditions, like CRF. By providing access to detailed visualizations and patient summary reports, the REBECCA platform will help clinicians develop personalized and improved care plans through i.) adjustments in medication and dosage, ii.) context-specific and actionable recommendations for physical activity and eating, iii.) measuring how well patients adhere to prescribed lifestyle guidelines (and identifying potential barriers towards compliance).

User involvement
Members of the Swedish Breast Cancer Society, Amazona (www.amazona.se), have contributed as patient user representatives in this project. Amazona members have been actively involved in REBECCA since the beginning of the project, both in the decision of which RWD to collect, as well as in the development and testing of the REBECCA platform. During the system co-creation, an online user survey was sent out via email to approximately 3,500 breast cancer patients and survivors, which were members of Amazona and sister organisations. Out of them, 544 answers were received. The survey included questions about challenges that breast cancer patients face during and after treatment, about the use of technology (e.g., smartwatches, installation of apps etc.) among patients, and which RWD they consider as acceptable and important to collect in relation to their QoL in the REBECCA project. The REBECCA platform was developed based on the input from these user surveys so that the RWD collection would not be perceived as intrusive and burdensome for the patients participating in the clinical studies. User representatives from Amazona have also tested the smartwatch used in the studies, and have contributed with data for testing of the overall REBECCA platform.

The study design of the clinical REBECCA trials
For the REBECCA clinical trials at SUH the patient follow-up will end on February 28th, 2026. The end date for all studies is December 31, 2038.

REBECCA-1: characterization of quality of life and late effects in breast cancer patients
REBECCA-1 is an observational study where the aim is to monitor the QoL of breast cancer patients to advance our understanding of the factors that affect their QoL, as well as the functional limitations of patients suffering from CRF during and after breast cancer treatment.
In the REBECCA-1 study, breast cancer patients are included into the study at the time of diagnosis. After end of primary treatment (surgery, chemotherapy given neoadjuvant or adjuvant and radiation) the patients are assigned to one of two groups:

The CRF group (n = 49): Breast cancer patients with a Visual Analog Scale-fatigue (fVAS) score ≥ 50, and/or do have a change in fVAS > 25 from the time of inclusion to the end of treatment.

The mild CRF group (n = 49): Breast cancer patients with a fVAS score < 50.

All participants undergo patient visits at the hospital every 6 months, and the study lasts for 18 months, with an optional participation for another 6 months. In addition to the two patient groups, we also recruit a control group of healthy women (n = 49). The control group is followed for 12 months with the option to extend their participation in the study by 6 months. All patients in the REBECCA-1 observational study receive the REBECCA system after the end of treatment for objective RWD collection for 12–18 months. The healthy control women receive the REBECCA-system at the time of inclusion. Figure 2 shows an overview of the REBECCA-1 study design. The participants are trained to use the system, and the patients also have the option to assign a companion that can contribute information about their QoL in the REBECCA companion app.
During the study, all patients meet with the study nurse at the time of diagnosis, at the end of treatment, and at 6 and 12 months after the completion of treatment for collection of PROMs and biological material. For the healthy women in the control group, this is done at the time of inclusion, and after 6 and 12 months.

REBECCA-2: An intervention study based on real-world data to improve quality of life after breast cancer treatment.

REBECCA-2 is an intervention study where the aim is to use multi-source RWD to monitor the QoL of breast cancer patients affected by CRF during and after the cancer treatment, and to provide them with a more personalized follow-up to improve their QoL.
In the REBECCA-2 study the patients are included into the study at the time of diagnosis. After end of primary treatment (surgery, chemotherapy given neoadjuvant or adjuvant and radiation) the patients are randomized into two groups:

i) Control group (n = 55): Standard follow-up

The patients who participate in the control arm receive standard follow-up according to national guidelines in Norway [17]. This includes control visits 6, 12 and 18 months after inclusion into the project, as well as three consultations by phone during the project period.

ii) Experimental group (n = 55): Standard follow-up + REBECCA assisted follow-up

The patients participating in the experimental group receive a commercial wearable device (smartwatch), which they wear for 12 months. In addition, they install the REBECCA patient app on their personal mobile phone for us to obtain objective RWD related to the participants’ QoL and lifestyle over the next 12 months. Patients with severe fatigue (fVAS score > 50 and/ fVAS > 25 from the last patient visit) during follow-up are referred to personalized training with a physiotherapist at Pusterommet, SUH. After the 12-month follow-up, participants may choose to remain in the study for an additional 6 months, for a total participation period of up to 24 months. Figure 3 shows an overview of the REBECCA-2 study design.
During the study period the patients receive follow-up visits at the time of diagnosis, after the end of treatment as well as 6, 12 and 18 months after primary treatment. The patients also have the opportunity for three additional medical consultations if the REBECCA platform detects signs of deterioration in their QoL based on changes in the patient’s lifestyle, or emotional health.
The attending physician has access to the collected RWD through the REBECCA Clinical Dashboard in order to actively use the information to personalize the patients’ follow-up (Fig. 4). If the RWD analysis shows signs of a deterioration in QoL based on the early warning system, the patients receive extra consultations. The intervention includes changes in prescribed medications, dietary advice, advice on exercise, referral to training with a physiotherapist or consultations with a psychologist/psychiatrist.
Patients in the experimental group may recruit a companion who will be able to contribute information about the patient’s QoL in the REBECCA companion app.
For a patient with severe CRF that is referred to training with a physiotherapist, the patient’s functional level is assessed at their first visit using a submaximal treadmill walking test, as well as tests for balance and muscle strength [18]. Based on the results of this test, a tailored training program is developed which is adapted to the patient’s functional level and physical condition. Each patient starts with a low amount of exercise before this is gradually increased to expand the CRF patient’s tolerance for activity. A physiotherapist follows the patient closely and the exercise program is adapted to the patient’s physical condition throughout the follow-up period.

During the study, PROMs and biological material are collected from all participants at each study visit.

REBECCA-3: REBECCA monitoring of prostate cancer patients
REBECCA-3 is a feasibility study where the aim is to use multi-source RWD to monitor the QoL of prostate cancer patients affected by CRF during and after cancer treatment to investigate whether the REBECCA-system is accepted by male patients and can be used in other cancer types.
In the REBECCA-3 study the patients are included at the time of diagnosis. After the end of primary treatment (surgery or primary radiation therapy followed by hormonal treatment) the patients receive the REBECCA-system for objective RWD collection for four months. Figure 5 shows an overview of the REBECCA-3 study design. The participants are trained in using the system, and the patients are also able to recruit a companion that can contribute information about their QoL in REBECCA’s companion app.
During the study, all participants meet with the study nurse at the time of diagnosis, following treatment and four months after completion of treatment (study end) for collection of PROMs and biological material.

Data collection

Patient reported outcome measures (PROMs)
PROMs related to lifestyle, QoL, working conditions and sleep are collected in the three REBECCA trials. For this purpose, we collect PROM forms (ΕORTC-QLQ-BR23 [19] SF-36 [20], ΕORTC-QLQ-C30 [21], HADS [22], FSS [23], FQ/Chalder fatigue scale [24], EPIC-26 [25], Bergen Insomnia Scale [26] and VAS-fatigue [27, 28]) as well as a self-reporting form. In the REBECCA-3 trial, the EORTC-QLQ-BR23 questionnaire was replaced with EPIC-26 which is specific for prostate cancer. The self-report form collects information about the patient’s working life, income, level of education and family structure. In addition, we also ask patients to share information about their use of medications.
In all three REBECCA studies, the PROM data are collected at the time of diagnosis and after the end of primary treatment. For the two breast cancer studies, PROMs are further collected every six months throughout the study period, while in the prostate cancer study PROMs are only collected again at the end of the study, four months after termination of the primary treatment. Study data were collected and managed using REDCap electronic data capture tools hosted at Helse Vest IKT [29, 30].

Real-world data
In REBECCA, objective RWD is collected to gain insight into the patients’ QoL in real-time. This includes data on the patient’s lifestyle (activity including step counts, sleep, eating habits), movement (GPS position data), stress level (which can be estimated using heart rate and heart rate variation taken from the smartwatch), local environment and meals (photos in the REBECCA app uploaded by the patient) and any concerns (internet usage, internet history) (Fig. 6). These types of data were selected given the potential that changes in any of these factors may give us an early warning of the development of CRF and/or have a direct impact on patients’ QoL in terms of physical and emotional functioning.
In the three REBECCA studies the RWD are collected following primary treatment and throughout the study period. For this data collection not to be intrusive, patients can decide for themselves whether they want to pause the GPS tracking or monitoring from the web browser plug-in. Patients are also able to decide the date after which the browser history should be registered in REBECCA.
All RWD are directly captured by the patient’s smartphone or uploaded via Bluetooth to the smartphone from the smartwatch. The REBECCA patient app is responsible for both types of data capture as well as for transferring the collected data to the REBECCA server in Germany for long-term storage, and analysis.

Clinical data
Clinical data related to breast -or prostate cancer and their cancer treatments are collected from the patient’s electronic health records. The data are stored in a REDCap database located at Microsoft Azure [29–33].

Biological samples
Biological material is collected from all participants in the clinical REBECCA studies to investigate biomarkers related to CRF. The biological material includes collection of blood (60 mL), urine (20 mL) and feces samples (20 mL), as well as archival tissue samples collected during surgery. After collection, blood samples are processed within 2 h for further isolation of serum, plasma (cold and room-tempered), platelets and peripheral blood mononuclear cells (PBMCs) according to the description in Table 1. All samples are then aliquoted before storage at -80 °C. Aliquots of whole blood (EDTA-K3), urine and feces are also stored at -80 °C (Table 1).

All samples are registered in a regional Laboratory information system for biobank samples, LabVantage Biobank. Parameters related to the sampling including time of sampling and freezing, number of aliquots, volume, and any discrepancies are registered.

Protocols for analyses of the biological samples
The biological mechanisms associated with the development of CRF are still unknown, but several studies support the hypothesis that inflammatory processes contribute to fatigue during, and especially after, cancer treatment [8]. Various approaches for analyses of the biological material collected in REBECCA will be explored to search for new biomarkers related to the development of CRF.

Immunological biomarkers
Inflammation has been shown to influence fatigue by initiating or maintaining elevated inflammatory activity [8]. In the context of cancer, inflammation can be initiated, or maintained, due to the tumour itself causing the production of proinflammatory cytokines [34], or due to cytokines produced during the cancer treatment in response to tissue damage from radiation or chemotherapy [35]. To search for new CRF-related biomarkers, we will evaluate the expression levels of several inflammatory biomarker candidates including Interleukin (IL)-1ß, IL-6, IL-8, IL-10, IL-1 receptor antagonist (IL-1RA) and soluble IL-1 receptor II (sIL-RII) [36, 37] in plasma from patients with severe CRF, mild CRF and healthy controls, using ELISA-based immunoassays. The biomarker data will be correlated to CRF status obtained by RWD, and QoL PROMs.

DNA methylation patterns
Chemotherapy is associated with DNA methylation patterns that may explain persistent inflammation and/or CRF in women treated for breast cancer [38]. Preliminary data has also shown reduced methylation at eight CpG sites in PBMCs of chemotherapy-treated patients, associated with increased levels of IL-6 and sTNFR2, which correlate with fatigue [38]. To investigate this further, we will isolate DNA from PBMCs and analyze methylation patterns using the Infinium MethylationEPIC v2 kit, and quantitative PCR for analysis of specific methylation sites. We will compare methylation levels between patients with severe and mild CRF versus healthy controls and monitor biomarker levels longitudinally as well as to compare them with changes in QoL based on the collected questionnaires (PROMs).

Gut Microbiome
A study investigating the gut microbiota in CRF reported distinct microbial signatures between individuals with mild and severe fatigue. Patients with mild CRF harbored higher levels of short-chain fatty acid (SCFA)-producing bacteria with known anti-inflammatory properties, whereas those with severe CRF exhibited an enrichment of pro-inflammatory taxa [39]. These findings were supported by a systematic review, which found consistent associations between increased abundance of pro-inflammatory bacteria, such as Enterobacteriaceae, and higher fatigue scores, while SCFA-producing or anti-inflammatory bacteria were linked to lower fatigue severity in individuals with cancer [40]. To investigate these associations in our cohort, fecal DNA will be extracted using the DNeasy PowerSoil Kit (Qiagen), following the manufacturer’s protocol. Metagenomic libraries will be prepared using the VAHTS Universal Plus DNA library Prep Kit and sequenced on the Illumina NovaSeq X plus platform, generating approximately 6 GB of data per sample, corresponding to about 20 million 150 bp paired-end reads.
Raw sequencing reads will be demultiplexed and processed using BBtools suits for quality control [41]. This includes trimming of adapter sequences (using a standard adapter reference), removal of low-quality reads (average Phred score < 20, read length < 75 nucleotides, or containing > 2 ambiguous bases), and filtering of PhiX spike-in control sequences. Host-derived reads will be removed by aligning sequences to the human reference genome (GRCh38) using Bowtie2 with default parameters [42]. Low-complexity reads were filtered using BBDuk based on entropy-based criteria.
Taxonomic profiling at the species level will be performed using MetaPhlAn (v4.1.1 and the database mpa_vJun23_CHOCOPhlAnSGB_202403) with default parameters [43]. Functional profiling will be conducted using HUMAnN (v3.0), which maps quality-filtered reads to the UniRef90 protein database to quantify gene family abundances and reconstructed microbial metabolic pathways based on MetaCyc annotations [44].

Endpoints, power calculations and statistical analysis plan
REBECCA at SUH consists of three independent studies including an observational study (REBECCA-1), a randomized clinical trial (REBECCA-2) and a feasibility study (REBECCA-3).

REBECCA-1

Objective, primary and secondary endpoints
The primary endpoint of the REBECCA-1 observational study is the difference in health-related QoL of women in the CRF, mild-CRF and the healthy control group, as measured through scores from the SF-36 [20] and EORTC-QLQ-C30 [21] questionnaires, at the evaluation of visit 3, which is 12 months post initiation, i.e., at the time of breast cancer diagnosis and approx. 6 months after the end of primary cancer treatment.
Secondary endpoints include (i) the assessment of differences in terms of the physical, emotional and social functioning, role limitations as well as general health, measured through SF-36 and EORTC-QLQ-C30, as above, but also through the Hospital Anxiety and Depression Scale (HADS) [22], (ii) the assessment of RWD indicators (measured via the wearable device and the patient app) as predictors of physical functioning and overall QoL of the individuals, as well as the use of longitudinal data to measure the effect of fatigue on physical functioning, including measurements from all visits. Specifically, as part of the REBECCA project, a Functional Index (FI) will be developed using the RWD collected from the wearable and patient app. This experimental index will be designed to quantify aspects of the participant’s physical functioning based on physical activity and physiological measures from the wearable device, as well as the movement patterns extracted from the location signal collected via the patient app. Novel Machine Learning regression models will map RWD indicators captured within a sliding time window of approximately 14 days, to FI. The latter will be assessed with respect to its ability to approximate the physical functioning status of patients by examining its association with the physical component summary of SF-36 [20] and the physical functioning scale score of EORTC-QLQ-C30 [21] acquired at the end of the 14 days window and (iii) assessment of fatigue status measured by the VAS-fatigue questionnaire at the patient visits [28, 45].

Statistical analysis plan and power calculations

Cross-sectional Group Comparisons

To test for potential differences in health-related QoL in visit 3, 12 months post-initiation and 6 months after the completion of the cancer treatment, we will apply one-way analysis of variance (ANOVA) across the three groups (CRF, mild CRF, and healthy controls). Additionally, in case the test suggests that there are statistically significant differences between the groups we will perform post-hoc pairwise comparisons with correction for multiple testing (e.g., Tukey’s Honest Significant Difference) for identifying and quantifying the differences between groups.
Bibliography suggests a strong effect (d > 0.8) of fatigue on the QoL of breast cancer survivors [46]. For ANOVA, assuming a conservative effect size of at least f = 0.3, we can calculate that n = 37 subjects are needed per group, to detect differences with power of 0.8 at 0.05 significance level. Assuming further a 20% dropout rate, at least 44 patients will need to be recruited in this study. To further be able to detect differences among the CRF and mild-CRF groups using Tukey’s HSD test, this number of patients allows the detection of an effect of d > 0.67, which is consistent with the relevant literature.

Longitudinal Analysis

Furthermore, for evaluating longitudinal changes in the QoL of the patients across the six-monthly hospital visits we will use Linear Mixed Effects (LME) models to capture the variations due to individual differences. These models will include fixed effects for the group in which the participant belongs (mild CRF vs. CRF), the visit (as a four-level factor) and their interaction. Next, random effects will be specified for subject ids of the participants to account for intra-individual correlations across repeated measures. This modeling strategy handles unbalanced data resulting from missed visits, assuming the missingness follows a missing at random (MAR) pattern. Moreover, the repeated measures which will be used for these models will include the records of participants from all four visits, i.e., at the time of diagnosis, at the end of the treatment, and at 6 and 12 months after treatment follow-up visits.
Simulations carried out using the simr [47] and lme4 [48] R packages using estimates derived from indicate that there is significant increase in power (close to 100%) in estimating the effect of the group in visit 3, using the same number of patients as in the cross-sectional analysis (i.e., 44 patients per group, also considering dropout).

Additional analyses

The data will also be used to inform the development of methods for data analysis using causal models, including Structural Causal Models [49] and the estimation of treatment effect using machine learning models [50].
For assessing agreement between self-reported data and the objectively measured REBECCA RWD indicators (Sect. Real -world data), only the data from the 14 days preceding each patient visit will be considered (see Sect. REBECCA-1 Objective, primary and secondary endpoints). Subjects with insufficient data to reliably compute indicators will be excluded from the analysis. Then, estimates of agreement between the estimated Functional Index and the reported physical functioning dimension of EORTC-QLQ-C30 will be assessed, after standardization, both in terms of their mean absolute error, as well as using Bland-Altman plots.

Software

All analyses will be performed using a combination of Python and R, via the rpy2 package. Mixed effects models will be estimated using the lme4 R package [48].

REBECCA-2

Objective, primary and secondary endpoints
In the REBECCA-2 study the primary endpoint is health-related QoL measured by the SF36-questionnaire [20] at the 18 months post initiation evaluation (patient visit 3), where an increase of ≥ 10% indicates an improvement in QoL, as a result of using the REBECCA intervention Similar to the REBECCA-1 observational study, secondary endpoints are related to differences observed in terms of the objectively measured RWD, i.e., the REBECCA indicators between the experimental and control arm, as well the assessment of differences in terms of the physical, emotional and social functioning, role limitations as well as general health, measured through SF-36 [20] and EORTC-QLQ-C30 [21], but also through the HAD [22].

Statistical analysis plan and power calculations

Cross-sectional analysis at visit 4 (18 months post-initiation)

The planned statistical analysis will follow traditional principles of randomized control trial case-study analysis, with two parallel, independent participant groups. Specifically, the primary study hypothesis will be tested through independent group comparisons (T-test) of the primary outcome measure (SF-36 score) at the 18 months follow-up evaluation (Fig. 3). For power calculations, we target modest improvement treatment effects (15% QoL improvement in the experimental vs. control group). Thus, for the primary analysis (0.90 power; alpha: 0.05; equal group allocation and average 20% standard deviation across individual primary outcome measurements), each study arm will require the participation of 39 patients per group. Assuming a conservatively high dropout of 30%, 55 patients will need to be recruited per group, for a total of 110 patients completing the study, at least until the 12-month post treatment evaluation.

Longitudinal analysis

Similarly to the observational study, LME models will be used in this case to take into account repeated measures, leading to increased study power. Simulations indicate that the numbers selected in the cross-sectional analysis (n = 39, after dropout) can detect even a small effect of 8% improvement in the experimental group with 0.95 power, when considering measurements from the four visits.

Additional analyses

The data will also here be used to inform the development of methods for data analysis using causal models, including Structural Causal Models [49]and the estimation of treatment effect using machine learning models [50].
For assessing agreement between self-reported data and the objectively measured REBECCA RWD indicators (Sect. Real-world data), only the data from the 14 days preceding each patient visit will be considered (see Sect. REBECCA-1 Objective, primary and secondary endpoints ). Subjects with insufficient data to reliably compute indicators will be excluded from the analysis. Then, estimates of agreement between the estimated functional index and the reported physical functioning dimension of EORTC-QLQ-C30 [21] will be assessed, after standardization, both in terms of their mean absolute error, as well as using Bland-Altman plots.

Software

All analyses will be performed using a combination of Python and R, via the rpy2 package. Mixed effects models will be estimated using the lme4 R package [48].

REBECCA-3

Objective, primary and secondary endpoints
The primary endpoint of the REBECCA-3 feasibility study is the evaluation of the average REBECCA system usage rate throughout the monitoring period of 4 months (Fig. 5) measured by the number of active data contribution days. The following criteria will be used for defining an active data contribution day:

Physical activity indicators - At least 10 h of physical activity monitoring (e.g., steps).

Mobility indicators - At least 10 h of coverage.

Sleep indicators - At least one sleep session lasting over 2 h.

If any of the above criteria is met, then the corresponding day will be considered an active data contribution day.
Secondary endpoints include the agreement between the Functional Index and the physical functioning dimension of the EORTC-QLQ-C30 [21] questionnaire, measured via the mean absolute error as well as Bland-Altman plots, both applied on standardized measures. Finally, the System Usability Scale questionnaires will also be used to assess the acceptability of the REBECCA system by male patients [51]. These results will provide indications on the transferability of the REBECCA approach to other forms of cancer and other patient groups. As the REBECCA-3 study is a feasibility study, these descriptive analyses will be performed across the whole sample group.

Statistical analysis plan and power calculations
The main purpose of the REBECCA-3 study is to assess the level of transferability of the REBECCA intervention to patients with other forms of cancer. The primary quantitative comparison concerns the mean number of active data contribution days per patient over a four-month monitoring period, evaluated against the corresponding metric from the REBECCA-2 study. To determine whether data contributions in REBECCA-3 are statistically equivalent to those observed in the REBECCA-2, the Two One-Sided Tests (TOST) procedure for equivalence testing of independent means will be applied. Assuming 55 patients for the intervention arm of REBECCA-2 study (as estimated in Sect. Statistical analysis plan and power calculations), a common standard deviation of 20 days for both groups and an equivalence margin of +/-15 days over a period of 4 months, the power analysis indicates that a minimum of 22 patients using the system in REBECCA-3 would provide 80% power at the 0.05 significance level to conclude equivalence.

Discussion
The introduction of precision medicine for the treatment of cancer is expected to result in an increasing number of cancer survivors in the future. Among breast cancer patients, targeted treatment has resulted in a 5-year relative survival of approximately 90%, which creates a need for more knowledge about life after cancer. Thus, in recent years there has been a growing awareness of both late side effects, and of patients’ QoL after the cancer treatment.
The overall clinical aim of the REBECCA studies is to improve the QoL of breast cancer patients affected by complex chronic diseases, like CRF, during and after treatment. With the planned data collection, we expect to get better insight into the patients’ QoL, both based on the collection of traditional PROMs and through the analysis of novel objective RWD. As of today, patients’ QoL and the CRF burden is measured through questionnaires (PROMs). PROMs are, however, based on subjective opinions from the patients themselves, where the perception of, for instance, CRF can vary considerably. We, therefore, supplement the PROM data with the collection of objective RWD in REBECCA, as we expect this will give us new insights into patient’s life challenges and enable us to better track their progress during and after the cancer treatment. We anticipate, for instance, that a change in the patient’s physical and social lifestyle is strongly linked to the development of CRF, and to the QoL in general. We will therefore: (i) collect GPS position data to gain insight into the patient’s life space and look into changes in daily mobility patterns, (ii) we will collect data on the patients’ physical activity to look into any changes, (iii) we will investigate if there is a change in the patients social activity (cafè and restaurant visits) due to development of CRF and, (iv) we will collect information about the patients sleep patterns and eating habits. The purpose of collecting data about the patients’ internet use, through REBECCAs web browser plug-in, is to gain insight into the patient’s emotional health status as internet use is anticipated to reveal cancer-related fear, post-traumatic stress, anxiety and depression [52]; [53].
During the years, there have been considerable publications on the QoL of breast cancer patients, but the impact to improved outcomes in breast cancer care has been questioned. In 2008 Montazeri published a bibliographic review where the literature related to health-related QoL in breast cancer patients from 1974 to 2007 was presented [14]. This review concludes that breast cancer patients receiving chemotherapy, and hormonal therapy, often experience side-effects and symptoms that significantly affect their overall QoL negatively. Although breast cancer treatment has improved in recent years by becoming more personalized, CRF is still reported to be the most prevalent side-effect, with up to 65% of cancer patients being affected by CRF during and/or after treatment [54]. Of these, over two-thirds of the patients report that they have suffered from severe CRF for at least 6 months, while one-third of patients experienced persistent CRF for several years after treatment [55, 56]. For breast cancer patients, Abrahams and colleagues report in their meta-analysis, involving 12,327 breast cancer survivors, that CRF status is related to the treatment received by the patients [57]. They also observed a relatively large reduction in the incidence of severe fatigue in the first six months following treatment [57]. In addition to this, emotional distress has been reported to be very common among breast cancer patients, even years after treatment [14]. Altogether, these findings highlight the importance of the REBECCA project where we will collect objective, real-time QoL data to map the life challenges of cancer patients and use this information to provide more personalized follow-up with the aim of improved counseling, social support, and advice on tailored exercise during and after cancer treatment.
Several guidelines have been established for recommendations on screening, assessments and management of CRF [54, 58]. According to these guidelines, and other studies, there is increasing evidence that physical exercise has a positive effect both on CRF, and in improving the patient’s functional QoL [59–61]. Hence in the REBECCA-2 intervention study the patients that receive REBECCA-assisted follow-up are advised on physical activity as well as diet, sleep, alcohol and tobacco use, anxiety and distress. Patients with severe fatigue are also referred to a physiotherapist where a tailored exercise program is developed which is adapted to the patient’s functional level and physical condition. However, in addition to having a positive effect on CRF and QoL, physical activity has also been shown to reduce systemic inflammation, which is thought to contribute to chronic CRF [8]. Theories include that the tumour itself may cause this inflammation through the production of pro-inflammatory cytokines [34], or that cytokines may be produced during cancer treatment in response to tissue damage from radiation or chemotherapy [35]. Additionally, changes in DNA methylation patterns and the gut microbiota have also been hypothesised to contribute to inflammatory processes [38, 39, 62, 63]. Thus, in REBECCA, we collect biological material in order to search for CRF related biomarkers (i.e., immunological markers and DNA methylation patterns) in plasma and/or the faecal microbiome, that can give new biological and prognostic information related to the development of CRF in patients with breast and prostate cancer.
Altogether, the REBECCA project aims to examine how RWD can transform research and patient care in breast cancer. By combining multiple data sources from wearable devices, mobile apps, PROMs, biological samples, and online behaviour with causal analysis methods, REBECCA introduces a new approach to studying CRF and overall QoL. Furthermore, the project will examine the use of real-time RWD to map the everyday challenges of cancer patients and use this information to provide more personalized interventions, with improved counseling, social support, and tailored exercise guidance during and after cancer treatment.
.