PERSEVERE: longitudinal assessment of quality of life and treatment toxicities in Chinese breast cancer patients - a multicentre cohort study protocol.

Introduction
Breast cancer remains the most frequently diagnosed cancer and the leading cause of cancer-related death among women worldwide. In 2022, an estimated 2.3 million new cases and 670 000 deaths were reported globally, with incidence and mortality projected to rise by 38% and 68%, respectively, by 2050—particularly affecting women in low-Human Development Index countries.1 In 2022, there were an estimated 357 000 new breast cancer cases in China (approximately 15.6% of all female cancers).2 Advances in early detection and multimodal treatment have significantly improved survival for patients with localised disease. Consequently, an increasing number of Chinese women are surviving breast cancer and living for many years post-treatment.3 These epidemiological shifts have shifted the clinical and research focus from cure alone to long-term survivorship, emphasising the importance of quality of life (QoL) and management of treatment-related toxicities.
However, treatments for early-stage breast cancer—including surgery, radiotherapy and systemic therapies (chemotherapy, endocrine therapy, targeted agents)—can cause a range of acute and chronic adverse effects.4 These adverse effects include pain, lymphoedema, cancer-related fatigue (CRF), psychological distress such as anxiety and depression, sleep disturbances, cognitive changes and cardiotoxicity, all of which profoundly impact patients’ QoL and their ability to resume normal life activities.58 For example, CRF is a particularly debilitating symptom that can persist for years post treatment,9 10 while breast cancer-related lymphoedema (BCRL) is a chronic, disabling condition with a reported incidence as high as 10%–30%.11
The relationship between QoL and survival outcomes is bidirectional. A decline in QoL can negatively affect treatment adherence, which is directly linked to poorer prognoses, including higher rates of recurrence and mortality. Conversely, studies have shown that a higher QoL is associated with extended survival, suggesting that interventions to improve QoL may also enhance survival outcomes.12 13
To fully understand the patient experience, a multimodal approach is essential. Patient-reported outcomes (PROs) are crucial for directly capturing the patient’s perspective on their health status and the impact of treatment, providing insights often missed by clinician-reported measures alone.14 15 Integrating PROs with clinical data, objective physiological measures and biological markers is key to a holistic understanding of treatment toxicities and recovery trajectories. Despite this need, most research on breast cancer survivorship has been conducted in Western populations. For example, large prospective cohorts in Europe and North America have delineated long-term toxicities and QoL trajectories after breast cancer treatment.1618 However, analogous data from Chinese patients are scarce. Cultural, socioeconomic, healthcare system and possibly biological factors unique to the Chinese context may influence toxicity profiles and rehabilitation needs, underscoring the need for indigenous data.
The PERSEVERE study was therefore designed to address this critical gap by establishing a comprehensive cohort to investigate these issues within the Chinese context, with the ultimate goal of informing the development of targeted interventions to improve survivorship care for breast cancer patients in China.

Methods/Design

Study aim, design and setting
PERSEVERE is a prospective, multicentre observational cohort study. It will function as a patient registry of Chinese women treated for localised breast cancer. The study is coordinated by the Cancer Hospital, Chinese Academy of Medical Sciences (CAMS) in Beijing, with collaboration from multiple leading cancer centres in diverse regions (eg, Shanxi Cancer Hospital in the north, CAMS Shenzhen in the south), and is planned to expand to other sites nationwide. All participating centres follow a common protocol and data collection standard operating procedure (SOP). The study is registered on ClinicalTrials.gov (NCT07010939) and adheres to international guidelines for observational research.19 The study is planned to run from April 2025 to December 2030, encompassing participant recruitment, longitudinal follow-up and data collection. Participant recruitment is expected to continue until approximately 2028, with follow-up extending for up to 5 years after enrolment.

Participants

Inclusion criteria
Women meeting all the following criteria are eligible:
Age ≥18 years.

Histologically confirmed invasive breast cancer, clinical stage I–III.

Planned to undergo curative-intent local therapy (surgery), with or without neoadjuvant or adjuvant systemic therapy (chemotherapy, radiotherapy, endocrine therapy, targeted therapy).

No evidence of distant metastasis.

Capable of completing questionnaires and follow-up assessments.

Provided written informed consent prior to any study-specific procedures.

Exclusion criteria
Women meeting any of the following are excluded:

Metastatic (stage IV) or recurrent breast cancer at enrolment.

Prior receipt of curative-intent treatment for current breast cancer before study entry (eg, previous surgery or systemic therapy).

Currently pregnant or breastfeeding.

History of another invasive cancer within 5 years (except non-melanoma skin cancer or in-situ cervical cancer).

Any condition precluding participation (eg, severe mental incapacity, inability to consent).

Recruitment and consent
Participants will be consecutively recruited from the breast surgery or oncology outpatient clinics of participating centres. During initial diagnostic evaluation or treatment planning, eligible patients will be preliminarily identified by attending physicians. Trained research coordinators will then conduct further eligibility screening and provide comprehensive written and verbal information about the study. Written informed consent will be obtained prior to enrolment. The entire process will adhere to ethical standards, ensuring voluntary participation and informed decision-making.
Approximately 3000 patients will be recruited to meet the analytical requirements for the primary and key secondary outcomes. This sample size is determined based on both feasibility and statistical precision and will also support subgroup analyses and multivariable modelling. The final number of participants may be adjusted depending on follow-up completeness and data accrual.

Exposure Definition
In this cohort, we define ‘exposure’ primarily as the type and intensity of breast cancer treatment received, including surgery (breast-conserving vs mastectomy with or without reconstruction), systemic therapy (eg, chemotherapy, targeted therapy, endocrine therapy) and radiotherapy. Each treatment modality will be classified in detail based on clinical records, including timing (neoadjuvant, adjuvant), regimen type, duration and cumulative dose where applicable. In addition to treatment exposures, patient-related factors (eg, baseline sociodemographic characteristics, comorbidities and functional status) and tumour-related features (eg, pathological subtype, staging, molecular subtype) will also be characterised as potential modifiers or interacting exposures. This comprehensive exposure framework allows for investigation of how specific treatments, either alone or in combination, contribute to the development of acute and long-term toxicities, QoL changes and survivorship outcomes.

Data collection

Schedule of assessments
Participants undergo assessments at defined timepoints (table 1). Baseline (T0) occurs after diagnosis but prior to the initiation of surgery or neoadjuvant therapy. For patients undergoing upfront surgery, follow-up assessments are planned at postoperative week 1 (T1), month 1 (T2) and month 3 (T3), followed by continued follow-ups every 6–12 months (T4, T5…) for up to 5 years. For patients receiving neoadjuvant therapy, additional assessments will be conducted at baseline (T0) and after the second, fourth and sixth cycles of treatment to capture the dynamic changes during systemic therapy. These patients will then transition into the postoperative follow-up schedule as described above.

Baseline assessments (T0)
At enrolment, comprehensive baseline data are collected via patient interview, self-report and medical record review. All data are entered into a secure, web-based Research Electronic Data Capture (REDCap) database. Collected information includes:
Sociodemographic factors: age, marital status, education, income, employment, health insurance.

Lifestyle factors: smoking, alcohol use, physical activity.

Anthropometry: height, weight, body mass index, waist/hip circumference.

Medical history: comorbidities, prior treatments, family cancer history.

Reproductive history: menstrual status, parity, breastfeeding history.

Tumour characteristics: clinical and pathological tumour–node–metastasis (TNM) stage, histologic subtype, grade, hormone receptor (oestrogen receptor/progesterone receptor (ER/PR)), human epidermal growth factor receptor 2 (HER2) status and Ki-67 proliferation index.

Planned treatment details: type of surgery and planned neoadjuvant/adjuvant therapies (regimen, timing).

Physical examination: vital signs and standard exam.

Baseline clinical tests: echocardiography (for all patients slated for potentially cardiotoxic therapy to assess left ventricular ejection fraction (LVEF)) and pulmonary function tests (for patients ≥70 years or clinically indicated). Routine labs (CBC, metabolic panel, liver/kidney function, lipid profile, inflammatory markers) are also performed.

Breast imaging: breast imaging data (mammography, ultrasound and MRI when clinically indicated) will be collected at baseline and 3 months postoperatively. All imaging results will be interpreted according to the Breast Imaging Reporting and Data System classification, providing standardised measures of tumour characteristics and early treatment response.

PROs: participants complete validated questionnaires (via REDCap or paper-to-digital entry) to capture baseline QoL and symptoms (see below).

Follow-up assessments (T1–Tn)
At each follow-up visit, data collection focuses on PROs and clinical events. The following are obtained:
PRO questionnaires: the same instruments used at baseline (see below) are administered at each scheduled follow-up, primarily via electronic collection (REDCap ePRO). Automated reminders and on-site support ensure completion. Telephone outreach is used for missed visits, with >95% follow-up anticipated. The timing of questionnaire administration in relation to scheduled clinical visits and follow-up imaging assessments will be documented to allow appropriate contextual interpretation of patient-reported psychological outcomes.

Clinical outcomes: updates on treatments received, treatment adherence and clinical events are abstracted from medical records. Surgical complications and re-interventions, including re-excision for positive margins, axillary re-operations, postoperative seroma requiring intervention and delays in adjuvant treatment (eg, delayed initiation of radiotherapy), are systematically captured during follow-up.

Objective assessments: no routine repeat echocardiograms or pulmonary tests are mandated for research purposes after baseline. Any clinically indicated tests are captured. Lymphoedema is screened clinically at follow-up visits.

Performance/function: selected measures of physical function or rehabilitation may be documented (eg, grip strength, return-to-work status) as part of rehabilitation objectives.

PRO measures
The following validated instruments are used:
European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire-Core 30 (QLQ-C30, version 3.0): general cancer QoL (global health, functional scales, symptom scales).

EORTC Quality of Life Questionnaire-Breast Cancer module (QLQ-BR23): breast cancer-specific QoL module.

EORTC Quality of Life Questionnaire-Fatigue module (QLQ-FA12): focuses on CRF (physical, emotional, cognitive domains).

Hospital Anxiety and Depression Scale (HADS): screens for anxiety and depression.

Pittsburgh Sleep Quality Index (PSQI): assesses sleep quality and disturbances.

Brief Pain Inventory—Short Form (BPI-SF): measures pain intensity and interference.

Perceived Stress Scale (PSS-10): measures perceived stress levels.

Social Support Rating Scale (SSRS): evaluates perceived social support (Chinese version).

International Physical Activity Questionnaire: assessing physical activity levels across various domains of daily life.

These instruments have validated Chinese versions.2024 Trained staff facilitate completion at in-person visits; remote electronic follow-up is used when needed.

Biospecimen collection
Timing and types: biological samples will be collected only at baseline, prior to surgery or during the surgical procedure and prior to the initiation of neoadjuvant therapy when applicable. Standardised SOPs ensure consistency across centres. Collected specimens include:
Peripheral blood (10 mL): drawn into EDTA tubes; processed to isolate plasma and peripheral blood mononuclear cells (PBMCs). Plasma and PBMC aliquots are stored for future analyses.

Tumour tissue: fresh-frozen tumour specimens will be collected from surgical samples at baseline. For patients receiving neoadjuvant therapy, core needle biopsy specimens obtained prior to treatment initiation will also be preserved. In addition, tumour tissues will be collected at the time of definitive surgery. These specimens will be used for future translational research on treatment response and mechanisms of resistance.

Adjacent normal breast tissue: paired peritumoral normal tissue (when available) is collected during surgery.

Processing and storage: blood samples are processed within a few hours of collection; plasma is stored at −80 °C and PBMCs are cryopreserved in liquid nitrogen. Fresh tumour and adjacent normal tissues are immersed in RNAlater overnight to preserve RNA integrity, then snap-frozen and stored at −80 °C. All biospecimens are barcoded and stored under standardised quality control procedures in a centralised biobank.
Intended use: the biobanked samples will support future translational research, including genomics, transcriptomics, proteomics, metabolomics and biomarker discovery related to treatment toxicities and recovery.

Outcomes

Primary outcome
The primary outcome is the change in EORTC QLQ-C30 global health status/QoL score from baseline to 12 months after treatment initiation. The primary analysis will focus on the mean change at 12 months. Longitudinal mixed-effects models will additionally assess trajectories of the QLQ-C30 global score over the full 5-year follow-up.

Secondary outcomes
Trajectories and risk factors for CRF, measured by the EORTC QLQ-FA12.

Trajectories of other PROs, including functional and symptom scales of the EORTC QLQ-C30 and QLQ-BR23, anxiety/depression (HADS), sleep quality (PSQI) and pain (BPI-SF).

Incidence and severity of BCRL, defined as an increase in arm circumference ≥2 cm compared with the contralateral arm or a volume difference >10%, confirmed by clinical examination.

Incidence of cardiotoxicity, defined as a LVEF decline ≥10 percentage points to a value <50% or a new abnormal global longitudinal strain, in line with European Society of Cardiology (ESC) and American Society of Clinical Oncology (ASCO) guidelines.25

Identification of demographic, clinical, psychosocial and biological predictors of QoL and toxicity trajectories.

Functional rehabilitation outcomes (eg, time to return to work or normal activities).

Exploration of multiomics signatures associated with toxicities using baseline biospecimens.

Stress (PSS-10) and social support (SSRS) measures over time.

Clinical outcomes including treatment adherence, disease recurrence/progression and overall survival.

Clinically meaningful changes will be interpreted using established thresholds (eg, 5–10 point change on QLQ-C30 is considered minimally important).

Sample size and power
This study aims to recruit approximately 3000 patients. The sample size is primarily designed to ensure adequate precision in estimating the primary and key secondary outcomes. Based on prior data, the SD of the global health status score on the EORTC QLQ-C30 is estimated to be around 30. Assuming a two-sided significance level of 0.05 and targeting a margin of error of 5% for the estimation of this SD, the required sample size was calculated to be about 2500 participants using standard precision-based formulas. To account for an anticipated 20% dropout over 5 years, the final enrolment target has been increased to 3000 patients, which will also ensure sufficient power for subgroup analyses, trajectory modelling and exploratory multivariable analyses.

Data management and quality control
All study data are managed centrally using the REDCap platform. REDCap provides secure web-based data entry with user permissions, audit trails and automated logic checks to enhance data integrity.26 A detailed data dictionary and standardised Case Report Forms are used across sites. Electronic PRO data collection is integrated into REDCap.
Key quality control measures include:
SOPs: comprehensive SOPs govern all aspects of the study (patient screening, consent, data entry, PRO administration, biospecimen handling, adverse event recording, etc). All personnel are trained on these SOPs.

Training: investigators and coordinators at each centre receive protocol training and REDCap training sessions.

Data monitoring: a dedicated data management team conducts real-time monitoring. REDCap’s data quality module and custom reports are used to track missing or out-of-range data. Routine data audits and source-data verification (on a sample of records) are performed by the coordinating centre to ensure accuracy.

Query resolution: procedures are in place for generating and resolving data queries in REDCap when inconsistencies or missing values are identified.

Missing data: missingness will be tracked (REDCap reports on follow-up completion). We will employ appropriate methods (eg, multiple imputation, sensitivity analyses) to address missing PRO data in analyses.

A Data and Safety Monitoring Plan oversees study conduct and data integrity. Although this is an observational study, periodic reviews ensure protocol adherence and ethical conduct.

Statistical analysis plan
Descriptive statistics and data preprocessing: baseline characteristics, sociodemographic factors and treatment data will be summarised using means (SD), medians (IQR) or proportions for the entire cohort and specific subgroups, such as those defined by surgery type or therapy received. To ensure data integrity, missing patient-reported data will be tracked through the REDCap system, and we will employ multiple imputation or sensitivity analyses to address missingness under the Missing At Random assumption. Potential transient effects of follow-up imaging and the receipt of imaging results on anxiety-related outcomes will be considered in sensitivity analyses where appropriate.

Primary analysis: the primary analysis will focus on the mean change in the EORTC QLQ-C30 global health status/QoL score from baseline to the 12 month post-treatment initiation milestone. We will employ linear mixed-effects models (LMM) or generalised estimating equations to assess longitudinal changes in QoL while accounting for within-subject correlation of repeated measures, adjusting for key baseline covariates including age, cancer stage and treatment modality.

Secondary analysis: secondary analyses will evaluate longitudinal trajectories of other PROs, including CRF (EORTC QLQ-FA12), anxiety and depression (HADS), sleep quality (PSQI), pain (BPI-SF) and functional and symptom scales of the EORTC QLQ-C30 and QLQ-BR23. Similar longitudinal modelling approaches to those used in the primary analysis will be applied, including LMMs and latent class growth modelling, to identify distinct subgroups of patients with heterogeneous recovery or decline patterns over time. Incident clinical outcomes, including BCRL and cardiotoxicity, will be analysed using time-to-event methods such as Kaplan-Meier estimates and Cox proportional hazards regression. Multivariable regression models will be used to identify independent demographic, clinical, treatment-related and psychosocial predictors of these outcomes. Key clinical events occurring during follow-up—such as surgical re-operations, postoperative complications and delays in adjuvant treatment—will be recorded and considered as potential confounders or time-varying covariates in longitudinal and time-to-event analyses where appropriate.

Exploratory analysis: exploratory multi-omics integration will use established bioinformatic pipelines to process genomic, transcriptomic, proteomic and metabolomic data from baseline biospecimens, with the aim of identifying biological signatures associated with symptom persistence. In addition, machine learning approaches, such as random forests or neural networks, may be applied to develop and internally validate risk prediction models for adverse outcomes, with model performance evaluated using metrics including discrimination (eg, area under the curve) and calibration. This exploratory approach is intended to enhance the clinical interpretability of observed QoL trajectories in a real-world treatment setting.

Clinical interpretation and software: all clinically meaningful changes will be interpreted using established minimally important differences, specifically the 5–10 point threshold for the QLQ-C30 scales. Analyses will be performed using standard statistical software (R, SAS or Python) and specialised bioinformatics tools, following a detailed SAP finalised before database lock in strict adherence to Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Ethics and dissemination
Ethics approval and consent: the study protocol (ID: NCC25/629-5575) has been approved by the Independent Ethics Committee of the National Cancer Center/Cancer Hospital, CAMS. Approval from the local institutional review board at each participating centre will be obtained prior to the initiation of participant recruitment at that site. Details of participating centres and the status of ethics approval at each site are summarised in table 2.
Written informed consent will be obtained from all participants by trained research staff before enrolment. Consent covers the collection of clinical data and PROs, as well as the long-term storage of de-identified biospecimens for future ancillary research, including genetic analyses. Participants will be informed of their right to withdraw from the study at any time without any impact on their clinical care.
Confidentiality and data protection: participant confidentiality will be strictly maintained. Data collected via the REDCap system will be coded using unique study identification numbers, with personal identifiers stored separately in secure, access-restricted and encrypted files. Only authorised study personnel will have access to the full dataset, and all research data and biospecimens will be de-identified prior to analysis.
The study will comply with China’s Personal Information Protection Law and regulations of the Human Genetic Resource Administration of China, ensuring that the collection, storage, use and sharing of sensitive personal information and biospecimens meet all relevant legal and ethical requirements.
Adverse events and monitoring: as this is an observational study, no study-related therapeutic interventions are administered. Any unexpected adverse events related to study procedures, such as complications associated with blood collection, will be monitored and documented. Each participating centre will report procedure-related serious events or complications, if any, to the coordinating centre in accordance with local institutional and regulatory requirements. Study conduct, protocol adherence and data integrity will be overseen through predefined monitoring procedures throughout the planned follow-up period.
Dissemination and data sharing: study findings will be disseminated through publications in peer-reviewed open-access journals and presentations at national and international scientific conferences. De-identified individual participant data underlying published results will be considered for sharing on reasonable written request, from 6 months to 5 years following publication. Requests will be evaluated based on scientific merit, and access will be granted following approval and execution of a data access agreement. Shared data will be limited to that required for independent verification of published findings.

Patient and public involvement
Patients and the public were consulted during the preparatory phase of the study to inform the refinement of the research focus and the selection and acceptability of PRO measures. Feedback from breast cancer patients helped to ensure the relevance, clarity and feasibility of questionnaire content, follow-up procedures and the perceived burden and time required for participation.
Patients were not involved as partners in the formal development of the research question, final study design or conduct of the study. However, patient perspectives will continue to be incorporated during the dissemination and knowledge translation phase. Study findings will be shared with participants and relevant patient advocacy groups through newsletters, educational seminars and lay summaries. Feedback obtained during these activities will be considered to inform future refinements of the cohort and the design of subsequent patient-centred research.

Discussion
The PERSEVERE study is a landmark effort to prospectively and comprehensively investigate the multifaceted impact of local treatments on early-stage breast cancer patients in China, addressing a critical knowledge gap. By establishing one of the first large-scale, multicentre cohorts of its kind in China, PERSEVERE will generate robust, culturally specific evidence to inform and improve survivorship care. The study’s strength lies in its multimodal dataset, which integrates detailed clinical data, extensive PROs captured via REDCap and a well-curated biobank. This approach is crucial for moving beyond documenting symptoms to identifying their complex determinants—including demographic, clinical, psychosocial and biological factors. In contrast to previous research (often single-centre or cross-sectional), PERSEVERE emphasises high-quality longitudinal measurement. For example, rigorous data management (centralised REDCap, SOPs, dedicated training) and proactive follow-up are expected to result in minimal missing data and high PRO completion rates. An independent steering committee oversees study conduct, and the protocol is prospectively registered, reflecting adherence to best practices for observational research.
Methodologically, this study leverages three interlinked advances. First, the frequent and comprehensive PRO assessments (EORTC questionnaires, symptom scales, etc) combined with objective baseline measures will enable dynamic profiling of QoL and symptom trajectories. Second, the biobanking of baseline specimens permits future omics analyses to uncover biological mechanisms underpinning persistent toxicities (eg, molecular pathways of fatigue or cardiotoxicity). Third, advanced analytics—including machine learning and risk modelling—will be applied to integrate these data into predictive tools. In this way, PERSEVERE will not only identify correlates of poor outcomes but also strive to produce actionable risk stratification models that could guide personalised survivorship interventions.
An important strength of this cohort is the inclusion of patients receiving neoadjuvant therapy, which enables real-time investigation of dynamic biological changes throughout systemic treatment. In this subcohort, longitudinal biospecimens—including tumour tissue (pre-treatment core needle biopsy and post-treatment surgical samples), PBMCs and plasma—are systematically collected at key treatment milestones. This design facilitates in-depth translational research on treatment response and resistance mechanisms at both the tumour and systemic levels. By integrating multiomics profiling of tissue and blood-derived components, the study allows for comprehensive exploration of molecular alterations, immune modulation and circulating biomarkers across the treatment continuum.
Despite its strengths, some limitations must be acknowledged. The cohort is restricted to women with stage I–III disease in China; therefore, findings may not generalise to metastatic patients, to other ethnic populations or to countries with different healthcare systems. Additionally, for patients undergoing primary surgery, biospecimens are collected only at baseline, which means that we cannot directly study intratreatment biological changes. On-treatment functional or physiological assessments (beyond PROs) are also not routinely captured, potentially missing subclinical effects. Like all long-term PRO studies, PERSEVERE may face missing data if participants drop out. We also acknowledge that anxiety and psychological distress may be transiently influenced by the timing of follow-up imaging and the receipt of imaging results, including false-positive findings. However, our follow-up protocols and planned statistical methods aim to minimise and account for this bias.27 Finally, as an observational study, causal inferences will be limited; we will rely on multivariable and sensitivity analyses to adjust for confounding.
In conclusion, the PERSEVERE cohort represents a major opportunity to advance understanding of treatment-related toxicities and QoL in Chinese breast cancer survivors. By generating high-quality, multifaceted data, this study will redefine supportive care paradigms for this population. The insights gained will not only help to improve the long-term health and well-being of millions of breast cancer survivors in China but will also contribute to the global understanding of breast cancer survivorship.