Improving CDK4/6 inhibitors adherence in breast cancer using the CONnected CUstomized Treatment Platform (CONCURxP) mobile health intervention: Study protocol for ECOG-ACRIN EAQ221CD.

1.
Introduction
Cyclin-dependent kinase 4/6 inhibitors (CDKIs), palbociclib, ribociclib, and abemaciclib, are novel oral targeted therapies that have transformed breast cancer treatment over the past decade. In hormone receptor-positive (HR+), human epidermal growth factor receptor 2–negative (HER2−) metastatic breast cancer (MBC), adding a CDKI to endocrine therapy have nearly doubled progression-free survival from 14.5 to 24.8 months in both first- [1–5] and second-line [6–9] settings. CDKIs have also improved 5-year overall survival from 16.8% to 23.2%.[10–14] In high-risk HR+, HER2− early breast cancer, adjuvant abemaciclib or ribociclib in combination with endocrine therapy significantly reduces recurrence risk vs. endocrine therapy alone.[15–17] Both medications are FDA-approved in this setting.[18, 19]
In studies showing survival benefits of CDKI in MBC, CDKI relative dose intensity (i.e., dose administered over the total time receiving chemotherapy divided by the protocol-specified standard dose intensity) ranged from 87% to 94%.[1–5] Other oral breast cancer therapies also linked relative dose intensity ≥85% with improved survival.[20, 21] While oral therapies reduce clinic visits and infusion needs,[22] they raise concerns about safety, adverse events monitoring, and adherence.[23] CDKI adherence at 6-months ranged from 57.0% to 87%,[24–26] with 7.5% discontinuing by 10 months,[24] mainly due to disease progression or adverse events.[2, 3] Complex dosing (e.g., 3 weeks on/1 week off for palbociclib and ribociclib, or twice daily for abemaciclib) [27] and high out-of-pocket costs (up to $12,056/month)[24] may contribute to non-adherence. One study linked non-adherence to worse progression-free and overall survival.[26]
The rise of mobile health (mHealth) technologies has created new platforms for promoting adherence to oral anticancer therapies[28, 29], though results have been mixed.[29–34] Most mHealth interventions have addressed patient-related barriers (e.g., reminders for forgetfulness and education for knowledge gaps), treatment-related barriers (e.g., symptom monitoring), and healthcare-related barriers (e.g., provider communication).[29, 35] However, no prior studies have specifically evaluated mHealth interventions for CDKI adherence.
We previously pilot tested the feasibility of a personalized mHealth intervention, the CONnected CUstomized Treatment Platform (CONCURxP), to improve Palbociclib adherence in breast cancer. [36] In that study, adherence was monitored using Wisebag (Wisepill Technologies, South Africa), a real-time electronically monitored smart pouch that was used by 72.4% of participants.[36] Wisepill Technologies was also previously used for monitoring endocrine therapies.[37] Building on our previous work,[36] we are now evaluating CONCURxP in a larger trial to improve CDKI adherence in breast cancer. CONCURxP key components include: (a) Wisebag for adherence monitoring; (b) personalized, interactive bidirectional text messages notifying patients of missed or extra doses, with text surveys capturing reasons; (c) closed-loop communication with oncology provider alerts when adherence drops or exceeds a pre-determined threshold; (d) referral to Patient Advocate Foundation (PAF) [38] for financial navigation and assistance with cost concerns; and (e) a study web-portal allowing participants’ real-time CDKI use tracking.
Our framework is informed by the Anderson and Aday behavioral model of health services utilization,[39] and our preliminary studies.[31, 40, 41] Figure 1 outlines how CONCURxP and its components are designed to address key barriers to non-adherence and improve patient outcomes.
We describe the protocol for a randomized controlled trial (RCT) (EAQ221CD: CONCURxP Study), funded by National Cancer Institute (NCI), coordinated by ECOG-ACRIN cancer research group (EA), and conducted in partnership with the NCI Community Oncology Research Program (NCORP). The trial evaluates the effectiveness of CONCURxP vs. enhanced usual care (EUC) in improving CDKI adherence.

2.
Objectives
The primary objective is to compare CDKI adherence at 12 months, measured by Wisebag, between EUC and CONCURxP arms. Secondary objectives include comparing 12-months patient-reported outcomes (PROs) between arms, including symptom burden, quality of life, patient-provider communication, self-efficacy in symptom management, and financial worry. Exploratory objectives include comparing longitudinal changes in patient-reported outcomes (self-reported adherence, symptom burden, quality of life, and financial worry) from randomization to 12 months, as well as healthcare utilization at 12 months between the arms. We also use mixed methods to describe patients’ experience with the CONCURxP intervention.

3.
Materials and Methods
3.1
Trial Design
EAQ221CD is a multi-site, parallel, 2-arm RCT with patient-level randomization comparing CONCURxP vs. EUC (clinicaltrials.gov: NCT06112613) (Figure 2). Participants and study coordinators are not blinded to arm allocation. This protocol follows SPIRIT guidelines [42] and is based on the March 5, 2025 protocol version, which was reviewed by NCI Division of Cancer Prevention and approved by the Central Institutional Review Board (CIRB). Study progress and patient safety are overseen by the EA Data Safety Monitoring Board (DSMB).

3.2
NCORP Sites
EA includes nearly 1,300 academic and community-based cancer centers and hospitals across the United States and internationally. The EA NCORP Research Base serves as a scientific hub for research conducted through the NCORP, which consists of 45 community sites encompassing over 1,000 practices and health care systems.[43] This trial is open to all NCORP sites, as well as the Winship Cancer Institute of Emory University and Chao Family Comprehensive Cancer Center of University of California Irvine. As of October 2025, 81 practices have accrued patients to the study.
Each NCORP affiliate or sub-affiliate enrolling patients to EAQ221CD is required to complete a site survey within 180 days of their first patient accrual. The survey captures site characteristics, including annual breast cancer patient volume, racial, ethnic and payer mix, zip code distribution, and geographic locations. Sites are asked to update this information within 30 days of the one-year anniversary of their first patient accrual during the study period.

3.3
Eligibility Criteria
Eligibility criteria include age ≥ 18 years, fluency in written and spoken English or Spanish, diagnosis of a new or established breast cancer, initiation of a CDKI within 60 days prior to consent or planned initiation within 30 days after consent (as confirmed via patients’ self-report), access to a personal smartphone or a non-smartphone with an email address and willingness to receive text messages, commitment to receive care and monitoring at a participating site, and ability to understand and sign informed consent.
Exclusion criteria include previous treatment with any CDKI, enrollment in another therapeutic or symptom science trial that monitors or intervenes on CDKI-related symptoms, or offers financial assistance, ECOG performance status ≥ 3, or medical inability to participate.

3.4
Recruitment, Enrollment, and Randomization
Patient screening is conducted according to each site’s local policies and procedures. Patients who meet eligibility criteria are approached by a clinical research associate (CRA). Upon providing informed consent, they are registered in the NCI Oncology Patient Enrollment Network (OPEN) database. Consented patients complete a 30-minute baseline survey either on paper or online via the ECOG-ACRIN Systems for Easy Entry of Patient Reported Outcomes (EASEE-PRO) website within 30 days of registration. A $10 gift card is provided upon survey completion. After confirming CDKI initiation based on self-report and completion of baseline survey, patients are randomized in a 1:1 ratio to one of the two study arms within 45 days of registration, balanced within each NCORP site (i.e., equal number of EUC vs. CONCURxP patients).

3.5
Study Interventions (Figure 2)
3.5.1
Wisebag Activation
Patients in both arms are mailed an introductory package containing the Wisebag and a study guide with instructions on storing their CDKI blister pack in the Wisebag, and how to use and charge the Wisebag. Patients in the CONCURxP arm also receive login instructions for the portal to view their adherence history. After receiving the Wisebag, patients are contacted by study team to set up their medication schedule and review the study guide. The Wisebag uses the evriMED1000 Medication Dispenser technology (Wisepill Technologies, South Africa)[44], which is 4G LTE enabled for remote, real-time adherence monitoring. It employs the same electronic monitoring as the widely used Wisepill pillbox, recording each container opening. The Wisebag was customized to fit CDKI blister packs, offering flexibility for patients to store and transport their medication in its original packaging. At each opening, the device uploads the data to the server via mobile network. Participants in both arms exclusively use the Wisebag for their prescribed CDKI, from Wisebag activation through 12 months post-randomization.

3.5.2
EUC (Arm A)
Every 30 days, participants receive educational materials about their CDKI and are asked via text or email (if no smartphone) to confirm Wisebag use and report any changes in their CDKI prescription or schedule. Reported changes are verified with the site CRA. EUC patients do not have portal access to view dose history, do not receive reminders, and their providers do not have access to adherence data.

3.5.3
CONCURxP arm (Arm B)
Patients in the CONCURxP arm receive the followings:
A) Missed Dose Reminders
Patients get automated text messages (Table 4) if they miss their scheduled dose, such as not opening the Wisebag within 2 hours of the dose time. For palbociclib and ribociclib, no reminders are sent during the off-cycle week. If the Wisebag remains unopened within 11 hours (palbociclib or ribociclib) or 5 hours (abemaciclib) after the scheduled dose, a missed dose message will be sent via text or email (if no smartphone) including a link to a multiple-choice question about reasons for non-adherence. This question is not sent if one was sent in the prior 3 days. If a patient reports missing doses due to a provider-advised pause or termination, reminders are paused automatically until notified by the patient or CRA to restart.

B) Over-Adherence Confirmation
If the Wisebag is opened more than once in 24 hours (palbociclib and ribociclib) or more than once in 12 hours (abemaciclib), or during an off-cycle, patients receive an automated text message (Table 4) followed by a single question to confirm the event.

C) Closing the Patient-Provider Communication Loop
When a patient reports over-adherence or missed doses due to side effects, the study team alerts the CRA to notify the oncology provider for follow-up. The timing and nature of follow-up are at the providers’ discretion, as CONCURxP is not designed for emergency response. For other missed dose reasons, if a patient misses ≥3 doses of palbociclib or ribociclib, or ≥8 doses of abemaciclib within last 28 days, the CRA is alerted to notify the oncology provider. Alerts to the CRA and oncology team are limited to one per 28 days to prevent alert fatigue, though patients may receive daily reminders.

D) Address Financial Barriers to Adherence
If cost is reported as a reason for missed doses or a barrier to adherence in follow-up surveys, the study team informs the CRA to offer the patient a referral to the Patient Advocate Foundation (PAF) or support from onsite financial counselors or navigators. This ensures that sites without financial navigation services can still provide support to CONCURxP participants. PAF assists with medication costs and other financial needs, including housing, utilities, and transportation.

E) View Adherence History
Participants can view their CDKI dose history at any time via a customized study web portal, based on the Wisepill platform. However, no automated reminders prompt participants to check their adherence history.

3.6
Intervention Follow-up Procedures
Patients complete a 30-minute follow-up survey, on paper or online, at 3-, 6-, and 12-months post-randomization, and receive $20, $30, and $40 gift cards, respectively. The intervention continues until consent withdrawal, death, 12-month follow-up completion, provider-advised permanent CDKI discontinuation (with continued survey completion), or medical ineligibility. Site CRAs will complete monthly forms documenting CDKI changes, pauses or discontinuation (supplementary Material 1), quarterly forms on receipt of internal financial counseling, and forms at 3, 6, and 12 months to document hospitalizations.

3.7
Patient Interviews
Starting 15 months after the first patient is randomized, 1 to 2 Arm B participants who completed the 12-month follow-up and consented to the optional interview are selected quarterly. Within 6 months of follow-up completion, selected participants complete a 30–40-minute interview. Selection is based on their race (white vs. Black) and CDKI adherence (<85% and ≥85%). A total of 30 patients will be interviewed.

3.8
Study Outcomes
3.8.1
Primary Outcome
The primary endpoint is 12-month CDKI adherence, measured via Wisebag. Adherence is defined as the proportion of scheduled doses taken within the accepted time window on days the patient was expected to take CDKI. Days off-cycle, during provider-advised pauses, or when hospitalized (i.e., medication is not self-managed) are excluded from the denominator.
For palbociclib and ribociclib (which are once per day medications), the accepted window is defined as +12 to +36 hours from a prior consecutive dose. A dose taken between +2 to +12 hours after a prior consecutive dose is considered a double dose. A dose taken more than +36 hours after a prior consecutive dose implies a missed dose and will be counted as the next dose.
For abemaciclib (which is a twice-daily medication), the accepted window is defined as +8 to +16 hours from the prior consecutive dose. A dose taken between +2 to +8 hours after the prior consecutive dose is considered a double dose. A dose taken more than +16 hours after a prior consecutive dose is considered a missed dose and should be counted as the next dose (Supplementary Table 1).
We will also evaluate over-adherence, calculated as the number of times that the Wisebag recorded a dose taken in any of these scenarios: (1) Off-cycle week or during a pause or termination advised by the provider; (2) a double dose (Supplementary Table 1).

3.8.2
Secondary Outcomes
The secondary endpoints include electronically-monitored CDKI persistence and patient-reported outcomes. CDKI persistence is defined as the number of days from CDKI initiation to discontinuation of medication, measured as the number of days from initiation until the first day of a gap that is 30 days or longer.
The remaining endpoints are based on other patient-reported outcomes, including self-reported CDKI adherence, symptom burden, quality of life, patient-provider communication, self-efficacy for managing symptoms and financial worry, which will be assessed with the participant surveys. See Table 1 for the specific study measures and their time points of distribution.

3.8.3
Exploratory Outcome
The exploratory endpoints include healthcare utilization and longitudinal changes in patient-reported endpoints over time. All instances of Emergency Department (ED) visits and hospital stays (including the reason for visit) will be abstracted from patients’ medical records by site CRAs at 3-, 6-, and 12-months and categorized as anticipated (for chemotherapy or planned procedure) versus unanticipated (complication or treatment side effect). We will also assess self-reported healthcare utilization using patient surveys at 3-, 6- and 12-months using validated items from the National Health Interview Survey (NHIS).

3.8.4
Mixed Methods Outcomes
Table 2 outlines the mixed methods outcomes that will be assessed from CONCURxP participants only via surveys, qualitative interviews, portal activity log data, and administrative records. These include portal usability, patient satisfaction and experience with the intervention, and triggered notifications to the oncology team or PAF.

3.8.5
Moderators
We will also collect measures that may modify the effectiveness of the CONCURxP intervention on CDKI adherence via surveys and complement them with electronic medical record (EMR) data at baseline, 3, 6, and 12 months after randomization. These factors include sociodemographic information, insurance, health literacy, cancer variables, endocrine therapy adherence, medical history, including information about patients’ treatment other than CDKI, prior systemic treatments, ECOG performance status, NCI morbidity index, and participation in clinical trials. Patients will also be asked about their self-reported adherence to endocrine therapy using an adapted version of the PROMIS Medication Adherence Scale [45].

3.9
Statistical Analysis
3.9.1
Power Calculation for the Primary Objective
The 6-month CDKIs adherence in MBC in 2,906 patients ranged between 67.3% and 81% (mean of 79%)[24]. For CDKIs to have survival benefits, the relative dose intensity is expected to be more than 85% [20, 21]. In the current study, we estimate that the mean adherence rate in enhanced usual care at 12 months will be 79% (SD, 0.19), and in the CONCURxP arm at 12 months will be 86% (SD, 0.19). Recruiting 234 patients, the study will have 80% power to detect a 7% increase in adherence rate in the CONCURxP arm at 12 months (two-sided alpha =0.05, Table 3). We estimate that 40% of patients will drop out by 1 year due to death, stopping CDKI, limited WiseBag use or malfunction, or lost to follow-up.[44, 46, 47] Thus, 390 (i.e., 234/0.60) participants need to be recruited into the study and randomized.

3.9.2
Plans for Handling Missing Data
We will compare the baseline characteristics between the two arms to assess whether randomization distributed covariates evenly. Outcome analyses will be intent-to-treat. We will monitor patient attrition and determine whether the dropout is informative or not. If the data are missing randomly, we will implement a method such as multiple imputations to impute missing values. If missing data are not random, we will implement various adjustment methods in our analysis (e.g., Bayesian modeling with informative priors). [48]
For Wisebag data, the device only works if it is used, has a signal, and has battery life. If patients report not using the device for the prior month, or if a signal is not detected, then data for the period will be excluded from the analyses.

3.9.3
Analysis Plan
3.9.3.1
Primary Outcome
For each arm, mean and SD CDKI adherence rates for all patients will be calculated. A two-sample t-test will be used to compare CDKI adherence between the two arms at 12 months. As a randomized study, it is hoped that the two groups will be balanced between potential confounders. Appropriate comparisons between the groups will be performed (chi-squared tests, t-tests, etc.) to identify any potential confounders that were not balanced. To complement the primary between-arm comparison of 12-month adherence, we will conduct exploratory sensitivity analyses of repeated adherence from randomization through 12 months using generalized estimating equations (GEE) with an identity link, robust (sandwich) variance, and a working correlation structure (exchangeable by default; AR(1) in sensitivity).
Sociodemographic, insurance, cancer variables (including stage), endocrine therapy adherence, type of CDKI, health literacy and health beliefs, medical history, and NCORP practice characteristics may modify the association between study arms and CDKI adherence. These effects will be modeled through linear regressions by assessing interactions between these variables in their categorical form and the arm status to determine if they serve as moderators of the association seen in the above model.
Due to the large number of potential variables in the model, we will consider use of penalized regression methods, e.g., LASSO or Elastic Net to facilitate variable selection, which are also capable of properly dealing with highly correlated variables (i.e., multicollinearity).[49] Principal component analysis may also be performed to reduce multicollinearity by creating “new” variables from linear combination of the tested ones in the regression model. [50]

3.9.3.2
Secondary Outcomes and Exploratory Outcome
For each arm, mean days of persistence, and the proportion of patients who discontinue the medication earlier than 12 months will be calculated and compared between the two arms using the two-sample t-test and chi-squared test, respectively. Additionally, sensitivity analyses will be performed jointly modeling adherence and treatment discontinuation.
For all patient-reported outcomes (Table 1), we will compare mean scores and changes in mean scores (from baseline) at each time point between the arms using two-sample t-tests. A GEE approach [51] will be used to assess the longitudinal change over time, incorporating follow-up data for patient-reported outcomes. This will consider the repeated measures structure of observations within the same individuals over time and allow for analysis of incomplete data across time.
The impact of the intervention on health utilization will be explored. The difference in healthcare utilization between two arms will first be evaluated using the two-sample t-test for the continuous variables (e.g., length of stay or number of ED visits) and then be modeled through the regressions to allow for the control of other covariates if needed.

3.9.3.3
Mixed Methods Outcomes
Mean and SD for portal usability measured via surveys will be calculated. Portal usability will further be reported through the Wisepill portal, including proportion of patients viewing their history of doses taken on the portal or taking the CDKI dose after a smart reminder. We will also report the number of triggered referrals to oncology providers and PAF and their outcomes.
Satisfaction and experience with intervention and barriers encountered with various components of the CONCURxP will be assessed through one-on-one qualitative interviews. The Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) framework will be used to guide the evaluation questions and examine the effectiveness of the intervention and implementation outcomes.[52] Conventional content analysis methods will be used for qualitative methods, previously used by our team. In-depth interviews will be audio-recorded and professionally transcribed verbatim. NVivo, version 12, will be used to manage and analyze all qualitative data. The codebook will be based on the evaluation questions and emergent themes identified through open coding of the first few transcripts and team discussions.[53] After the development of the codebook, each interview will be coded independently by two analysts with discrepancies resolved through discussion. Node reports will then be generated (e.g., text associated with a specific code) using NVivo to facilitate the identification of sub-themes and similarities and differences by group (i.e., level of adherence, race). Patient data will be summarized, with illustrative quotes, into matrices by race and adherence level of participants to further identify patterns.[54] Matrices will be reviewed by two analysts and will provide an audit trail to increase the trustworthiness of the findings.[55]

4.
Discussion
Advances in oral therapies have improved survival in breast cancer and reduced care burden by enabling convenient at-home dosing without the need for infusion. To ensure CDKIs achieve their full clinical benefit, patients need support with adherence to maintain a relative dose intensity above 85%.[1–5] Novel approaches are needed to support adherence and enable timely monitoring between clinic visits.
As healthcare increasingly relies on telemedicine and technology-enabled interventions, now more widely covered by insurers,[56] our study examines CDKI adherence in this evolving landscape. mHealth offers a way for providers to enhance patient engagement and monitoring beyond in-person visits. Our intervention, CONCURxP, builds on prior work from our multidisciplinary team,[31, 36, 57] combines proven components of previously tested mHealth adherence interventions,[29, 58] and is uniquely designed to minimize burden on both patients and providers.
Most prior mHealth interventions for oral anticancer therapy targeted selected adherence barriers, such as forgetfulness or side effects.[59, 60] CONCURxP addresses multiple common barriers, including forgetfulness, side effects, and cost concerns. The bidirectional text messages, unlike unidirectional messaging used in some prior interventions,[59] will remind patients of missed or double doses and actively ask them for reasons for these events. Text messages are automated and triggered by the Wisebag to minimize clinic burden. The intervention will be available in English and Spanish at community oncology practices nationwide to increase accessibility and participant diversity. Approximately 80% of cancer patients receive their care in community oncology practices. NCORPs’ diversity and range of clinical settings make it an ideal environment to evaluate the real-world impact of this approach.
As of October 2025, this study has reached over 50% of its accrual goals and is on track to complete accrual by mid-2026. This trial will provide robust evidence on how an innovative, mHealth-based intervention can improve CDKI adherence and quality of life among patients with breast cancer. We acknowledge study limitations, including a lack of stratified randomization and opportunities to improve the motivational tone of text reminders. Additionally, we are not actively monitoring patient symptoms or providing tailored support for symptom management. Oncology alerts are triggered only when a side effect is cited as the reason for non-adherence. Wisebag captures container opening and not medication ingestion which introduces potential misclassification, a limitation for all adherence monitoring devices. Further, we acknowledge that since both arms use Wisebag, the EUC group may still experience the Hawthorne effect. However, it was not feasible from a practical perspective to have a control arm without intervention. Despite this, we believe our results will inform the development of mHealth interventions to engage with patients outside of clinic visits, monitor patient-reported outcomes, and improve patient communication with their care team in a variety of conditions. A successful mobile technology-driven intervention could be disseminated across systems, conditions, and populations. Our results will provide valuable and actionable evidence for healthcare providers, policymakers, and insurers who continually strive to achieve the “Triple Aim” – to reduce costs while improving health outcomes and the patient experience.

Supplementary Material
1