LymphSens study: the enigma of subjective lymphoedema - how often and why do patients report lymphoedema after breast cancer treatment without an objective measurable swelling? The role of lymphatic and sensory processing problems: a protocol for a multicentre prospective longitudinal study
LymphSens study: the enigma of subjective lymphoedema – how often and why do patients report lymphoedema after breast cancer treatment without an objective measurable swelling? The role of lymphatic and sensory processing problems: a protocol for a multicentre prospective longitudinal study
Breast cancer-related lymphoedema (BCRL) at the arm and/or trunk/breast is a highly feared complication following breast cancer treatment and can be objectified using the state-of-the-art criteria based on volume, extracellular water ratio or skin thickness measurements. Although the incidence of objective BCRL is decreasing due to advances in breast cancer treatment, many patients report a sensation of swelling without the presence of objective BCRL, referred to as subjective BCRL. As little is known about the prevalence and the transitions between different BCRL states (no-subjective-objective) over time, as well as about the underlying mechanisms and contributing factors of subjective BCRL, this will be investigated in the LymphSens study.
230 patients with a new diagnosis of unilateral breast cancer will be included in a multicentre longitudinal study. Measurements are performed from presurgery to 12 months postsurgery. The primary objective (aim 1) is to determine the prevalence rate of subjective and objective BCRL at 1, 6 and 12 month(s) postsurgery, as well as transitions between BCRL states (no-subjective-objective BCRL) by a multinomial logistic regression model with generalised estimating equations and transition matrices, respectively. A second objective (aim 2) is to determine factors related to four potential underlying mechanisms (lymphatic, nociceptive, neuropathic and central sensory processing problems) that contribute to the occurrence of subjective BCRL in comparison with no self-reported swelling and objective BCRL. As a third objective (aim 3), within the group of patients with subjective BCRL, we will determine factors related to these four underlying mechanisms that contribute to the severity of subjective BCRL. The analyses for aim 2 and aim 3 will be conducted both at specific time points, that is, 1, 6 and 12 month(s) postsurgery using exploratory analysis and across all time points collectively using multivariable binary logistic regression models or multivariable longitudinal models for repeated measures.
The LymphSens study protocol received approval from the Ethics Committee of UZ Leuven (S68133) and UZ Antwerp/University of Antwerp (5676-003252). The results of the LymphSens study will be presented at conferences and published in peer-reviewed journals.
http://creativecommons.org/licenses/by-nc/4.0/
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In 2020, WHO registered 2.3 million new breast cancer cases.1 If current trends continue, breast cancer will result in over 3 million new cases annually by 2040, primarily due to population growth and ageing.2 Breast cancer-related lymphoedema (BCRL) is one of the major and highly feared sequelae of breast cancer treatment and it is linked to substantial upper extremity morbidity, reduced quality of life and increased costs for both patients and the healthcare system.3
With BCRL, in addition to the accumulation of fluid in the interstitial tissues of the arm, trunk or breast, significant cellular changes such as inflammation, adipose tissue deposition and dermal thickening can be observed.4 The complexity of the detection of BCRL is related to these changes. Objective BCRL at the arm can be detected using state-of-the-art criteria as follows: (1) ≥5% increase of the volume difference between arms compared with the presurgical difference,5 (2) >6.5 change in the L-Dex score from the presurgical score by bioimpedance spectroscopy6 and/or (3) a positive extracellular water (ECW) ratio of >0.3850 determined by bioimpedance analysis (BIA).7 Objective BCRL at the trunk and/or breast can be identified (1) by a percentage of water content (PWC) ratio, ≥1.32 and 1.40, respectively, using8 9 and/or (2) through ultrasonography, showing a difference of >2 SD from the skin thickness compared with the non-affected side or reference value measured at the presurgery.10 When the objective measurements above fail to confirm the presence of BCRL despite self-reported swelling, the condition is referred to as subjective BCRL, which has been defined based on the patient’s sensation of a difference in size of the arms and/or trunk/breast without any measurable swelling.
The pooled prevalence rate of objective BCRL is reported in the literature to be 16.6%, ranging from 11.8% to 53.5% for axillary lymph node dissection and from 0% to 15.8% for sentinel lymph node procedures.11 12 In contrast, the prevalence rate of subjective BCRL is much higher than that of objective BCRL.13 Longitudinal studies indicate that 50%–80% of patients experience subjective arm BCRL-related symptoms such as feelings of swelling, heaviness and/or tightness within 12 months postsurgery.14 15 One of these studies also found that although 61% of the patients did not have a measurable relative arm volume change, the majority (80%) of them still reported symptoms of subjective arm BCRL during follow-up.14 Additionally, Hayes et al16 reported that the prevalence of subjective arm BCRL-related symptoms indicated at least mild severity ranged from 30% to 55% within 9 months postsurgery.
In addition, it remains unclear why so many patients report swelling without evidence of objective arm and/or trunk/breast BCRL, as well as what characterises the different types of BCRL (ie, no self-reported swelling, subjective BCRL and objective BCRL). To unravel possible underlying mechanisms for different types of BCRL groups and to determine the contributing factors to the severity of subjective BCRL, we hypothesise four potential mechanisms. These mechanisms include lymphatic problems (1) and sensory processing problems such as nociceptive (2), neuropathic (3) and/or central sensory problems (4).
The first potential mechanism is lymphatic problems such as the amount of swelling and the severity of dermal rerouting at the upper limb and trunk/breast. The increased amount of swelling at the upper limb or trunk/breast can create tension on the skin, causing nociceptive symptoms such as heaviness, firmness and/or tightness, even before BCRL becomes clinically present.13 A previous prospective longitudinal study revealed that most patients reported symptoms of subjective BCRL such as swelling, tightness or heaviness following breast cancer treatment, despite relative arm volume differences of only 0% to <3%.14 However, in this study, the timing of symptom reporting was not clearly outlined. In addition, dermal rerouting or backward flow from the lymph collector to the lymph capillaries may activate the mechanoreceptors and result in self-reported swelling.17 18 Near-infrared fluorescent imaging or lymphofluoroscopy is an innovative technique that can be used to assess the changes in lymphatic architecture.19 On a lymphofluoroscopy, a functioning lymph collector is seen as a linear pattern, while dermal rerouting is recognised by splash, stardust or diffuse patterns.20
Subjective BCRL may also be explained by altered sensory processing, which refers to how the nervous system interprets sensory messages and translates them into responses. These alterations may include nociceptive, neuropathic, central sensory processing problems.21–23 Nociceptive input from the upper limb and trunk, resulting from various cancer treatments, may contribute to reporting the feeling of swelling. Breast cancer surgery and radiotherapy can cause scar tissue formation and fibrosis in the surgical and radiotherapy areas. Increased severity of myofascial adhesions at the level of scars, local muscle stiffness and tenderness in the upper limb including the shoulder region may lead to symptoms like tightness, firmness and pain.24 25 In a systematic review, the average prevalence of tenderness was found to be 36% in patients without objective BCRL but with either no BCRL and/or subjective BCRL,13 making it the most common symptom in this group. Patients often interpret these symptoms as signs of BCRL, a condition that many feel anxious about.15 Additionally, an impaired range of motion (ROM) of the arm is common after surgery and radiotherapy and may contribute to the experience of symptoms like pain and feeling of heaviness, leading to the misinterpretation of BCRL.26
Neuropathic problems arise from a lesion or disease affecting the somatosensory nervous system. Altered somatosensory functions, including touch, temperature and proprioception, have been extensively studied and documented in patients with breast cancer.27–29 Axillary surgery and/or radiotherapy may cause an injury of a sensory nerve such as the intercostobrachial, thoracodorsal and/or medial cutaneous nerve.30 This can result in symptoms like neuropathic pain, numbness or sensory loss, altered cold and heat detection sensitivity and altered sensory acuity, which patients may misinterpret as swelling as well.31 As proof of concept for mechanisms of alterations in sensory processing in subjective arm BCRL,29 we compared features of sensory processing problems between patients with objective arm BCRL, subjective arm BCRL and patients with no self-reported swelling at arm. The results show trends towards higher frequency and/or severity of neuropathic problems such as increased severity of sensory loss and altered sensory acuity in patients with subjective arm BCRL versus those with no self-reported swelling at arm. There were no differences in neuropathic problems between subjective and objective arm BCRL.29
Lastly, patients with subjective BCRL may show alterations in central sensory processing. In the breast cancer population, the presence of central sensitisation is evidenced by increased pain sensitivity due to widespread hyperalgesia, impaired efficacy of inhibitory pain mechanisms and dysfunctional facilitatory mechanisms of the central nervous system.32 33 These mechanisms may contribute to altered perception and the persistence of signs and symptoms related to nociceptive and neuropathic problems, beyond tissue healing.34 35 Additionally, disturbances in body perception due to brain neuroplasticity may also lead to the feeling of swelling at arm and/or trunk/breast, due to the perception of a larger size than the actual size.36 In our proof-of-concept study, we have also found that more severe central sensory processing problems such as enhanced pain facilitation and altered body perception were present in patients with subjective arm BCRL vs patients with no self-reported swelling at arm.29
In conclusion, subjective arm or trunk/breast BCRL seems to be a complex and multifactorial problem. Previous studies have not clearly established the prevalence rates of subjective BCRL at various time points after breast cancer treatment, particularly for the trunk/breast region, nor the transitions between the three different states of BCRL (no BCRL, subjective BCRL or objective BCRL).13–16 Additionally, prior studies lacked clear definitions for identifying subjective BCRL and baseline evaluations for detecting objective arm BCRL were also missing.13–16 In these studies, detection of objective BCRL was often based on a single predefined criterion derived from volume measurements taken after surgery.13–16 Inappropriate definitions of subjective and objective BCRL may have resulted in missing some patients with objective BCRL and leading to the underestimation or overestimation of the prevalence rates of objective and subjective BCRL. Moreover, previous research aiming at characterising different BCRL groups did not address potential underlying mechanisms like lymphatic, nociceptive, neuropathic or central sensory processing problems.29 37–40 Finally, to our knowledge, no studies have investigated the extent to which specific factors contribute to the severity of subjective arm or trunk/breast BCRL at various time points postsurgery.
The aims of the study, separately addressing arm and trunk/breast BCRL, are presented in table 1. The primary objective (=aim 1) is to determine the prevalence rates of subjective BCRL and objective BCRL at 1, 6 and 12 month(s) postsurgery, at the arm and at the trunk/breast, respectively. In addition, the transitions between different types of BCRL (ie, no, subjective or objective BCRL) on a group level and within a subject over time will be explored, again for the arm and trunk/breast region separately.
Table 1
Objectives and statistical analyses of the LymphSens study
A second objective (=aim 2) is to determine factors related to four potential underlying mechanisms (ie, lymphatic, nociceptive, neuropathic and central sensory processing problems) that contribute to the occurrence of subjective BCRL in comparison with no self-reported swelling and objective BCRL, separately again for the arm and breast/trunk. These factors will be determined both at specific points, that is, 1, 6 and 12 month(s) postsurgery and across all time points together.
As a third objective (=aim 3), within the group of patients with subjective arm and trunk/breast BCRL, we will determine factors related to these four potential underlying mechanisms that contribute to the severity of subjective arm and trunk/breast BCRL. These factors will be determined both at specific points, that is, 1, 6 and 12 month(s) postsurgery and across all time points together.
The LymphSens study is designed as a multicentre prospective longitudinal study and is implemented and reported in line with the Strengthening the Reporting of Observational Studies in Epidemiology statement for observational study.41
This study is conducted at the Multidisciplinary Breast Centre and Department of Physical Medicine and Rehabilitation of the University Hospital Leuven (UZ Leuven) and Breast Clinic of the University Hospital of Antwerp (UZA), Belgium.
Patients who are being diagnosed with primary unilateral non-metastatic or oligometastatic breast cancer and scheduled for breast surgery in combination with axillary lymph node dissection or sentinel node biopsy are included and followed from presurgery up to 12 months postsurgery. Therefore, the maximum duration of the study for a single patient is 12 months (±1 month).
Inclusion criteria are as follows: (1) adults (men and women) who are ≥18 years of age, (2) diagnosed with primary unilateral non-metastatic or oligometastatic breast cancer, (3) scheduled for surgery (mastectomy or breast conserving surgery; in combination with axillary lymph node dissection or sentinel node biopsy), (4) able to read, understand and speak Dutch and (5) have voluntary written informed consent of the patient. Patients are excluded if they: (1) have distant metastases, (2) planned bilateral lymph node surgery, (3) have oedema of the arm from another cause, (4) have an allergy to iodine or indocyanine green or (5) are physically or mentally unable to participate throughout the entire duration of the study. Inclusion and exclusion criteria represent the known risk factors for objective arm and trunk/breast BCRL and other side effects at the upper limb region possibly related to subjective arm and trunk/breast BCRL.5 42–44
A total of 230 patients will be recruited in the UZ Leuven and UZA together. Patient recruitment started on 26 March 2024, with an estimated inclusion rate of eight patients per month. This means that the estimated end of enrolment (‘last patient in’) is expected by May 2026, with the ‘last patient out’ anticipated by May 2027.
For each participating centre, the pre-operative lists are screened for patients scheduled to undergo breast surgery in combination with a lymph node procedure for unilateral breast cancer. This screening is conducted by a researcher of the study team (CG, KV, NV or LS). If the inclusion criteria are met during the initial screening process, the researcher provides a detailed explanation of the study, including its purpose, assessments and contact points, using study recruitment materials such as a flyer and a brochure. If the patient expresses interest in participating in the study, the researcher further explains the informed consent form (ICF).
After obtaining informed consent, a baseline assessment (T0) is performed a maximum of 1 week before surgery to establish reference values for oedema, nociceptive, neuropathic and central sensory processing. In addition to the baseline assessment, all patients are assessed at 1 month (T1), 6 months (T2) and 12 months (T3) postsurgery. Since BCRL and other comorbidities often develop within this timeframe, this follow-up period is most relevant for the project’s aims.11 12 Each clinical assessment lasts approximately 1 hour and 40 min at T0 and 2 hours at T1, T2 and T3. At the start of each clinical measurement, the assessor asks whether the patients have taken pain medication in the previous 24 hours, and if so, which type of pain medication they have taken, as this may influence sensory testing.45
Assessments of patient-related factors
Demographics and patient-related factors such as age, gender, education level, civil status and operated side are collected by interview at baseline assessment (T0). Additionally, as a patient-related factor, characteristics of pain (presence, intensity, pain-related functionality, pain localisation) and psychological symptoms such as depression, anxiety and stress are assessed at all assessment points (at T0, T1, T2 and T3) by the Brief Pain Inventory (BPI), the Margolis Pain Diagram and the Depression, Anxiety and Stress Scale (DASS-21), respectively. The BPI consists of questions about pain intensity, medication use, functionality during various activities of daily living and response to treatment.46 47 The localisation of pain is assessed by the Margolis Pain Diagram,48 including 45 different and predefined body areas. Patients are required to identify the regions where they experienced pain for at least 24 hours during the preceding 4 weeks. The DASS-21 evaluates the severity of depression, anxiety and stress over the past week.49 Higher scores on the DASS-21 indicate a higher level of depression, anxiety or stress. In addition, as descriptive variables, body length by a stadiometer and body weight by the InBody (InBody, Seoul, South Korea) device are measured.
By exploring the patient’s medical file, cancer and cancer treatment-related information such as the tumour node metastases staging system (tumour size, lymph node stage, stage of cancer), type of cancer, type of breast surgery (mastectomy or breast conserving surgery), lymph node surgery (axillary lymph node dissection or sentinel node biopsy), chemotherapy, radiotherapy and hormone therapy is collected. A self-developed questionnaire is used to assess physiotherapy care for the treatment of lymphoedema and shoulder complaints, including questions on the timing, frequency, duration and content of physiotherapy. Physiotherapy care will be categorised as no physiotherapy, optimal physiotherapy (if all required components of physiotherapy are provided) or suboptimal physiotherapy (if not all required components of physiotherapy are provided). We will describe the study population based on the demographics, patient-related factors, medical information and physiotherapy care. Furthermore, some variables such as age, body weight, psychological symptoms and cancer-related and treatment-related factors, and content of physiotherapy will be included in the analyses as a potential factor for the presence/severity of subjective BCRL (see item statistical analysis for aim 2 and aim 3). For more detailed information on these patient-related and treatment-related factors, please see online supplemental table 1.
Assessments of the different types of BCRL
Table 2 gives an overview and more detailed description of different types of BCRL. To identify the presence of subjective BCRL, first the presence of self-reported swelling at arm and trunk/breast is assessed using the Norman Questionnaire,50 which questions the differences in the sizes of both hands, lower arms, upper arms, trunk and/or breast region. For the presence of self-reporting swelling at the arm, if the patient responds ‘yes’ to the item self-reported difference in the sizes of at least one of three sites (hand, lower arm and upper arm), the presence of ‘self-reported swelling at arm’ is recorded as ‘yes’. Similarly, for ‘self-reporting swelling at trunk/breast’, if the response is ‘yes’ to the item self-reported difference in the sizes of at least one of these two sites (trunk or breast region), the presence of ‘self-reported swelling in the trunk/breast’ is recorded as ‘yes’. If the patient has self-reported swelling at arm or trunk/breast but no objective arm or trunk/breast BCRL, then the presence of subjective arm or trunk/breast BCRL is recorded as ‘yes’.
Table 2
Overview of the definitions and assessment methods of the different types of BCRL
The presence of objective arm BCRL is detected by a volume measurement of the hands, circumference measurements of the arms and by determining extracellular fluid by BIA and BIS. The presence of objective trunk/breast BCRL is detected by the tissue dielectric measurements and ultrasonography. The presence of objective arm or trunk/breast BCRL is recorded as ‘yes’ if one of the predetermined criteria is present. Lastly, the patients who do not have subjective and objective BCRL at arm or trunk/breast will be categorised as no arm or trunk/breast BCRL.
Besides being the primary outcome of aim 1, this outcome will also be used as a dependent variable in aim 2.
Assessments of factors related to four potential underlying mechanisms
In table 3, possible factors related to four potential underlying mechanisms are listed. A total of 15 potential independent factors, including lymphatic problems, nociceptive, neuropathic and central sensory processing problems, were proposed. In short, as lymphatic problems, the amount of swelling and the severity of dermal rerouting are assessed. The amount of swelling at the arm is recorded as the relative upper limb volume difference (in %) based on the arm and hand volume measurements, while the amount of swelling at the trunk and/or breast is recorded using the PWC ratio. The severity of dermal rerouting is evaluated using near-infrared fluorescence imaging or lymphofluoroscopy. To assess nociceptive problems, the severity of myofascial adhesions at scar tissue, local muscle stiffness, local muscle tenderness and impaired shoulder ROM are measured. The variables for neuropathic problems include the presence of neuropathic pain, the severity of sensory loss, the severity of altered cold detection sensitivity, the severity of altered heat detection sensitivity and the severity of altered sensory acuity. Finally, central sensory problems are evaluated by examining the severity of widespread hyperalgesia, efficacy of inhibitory pain mechanism, the degree of pain facilitation and the severity of altered body perception.
Table 3
Overview of the definitions and assessments of factors related to four potential underlying mechanisms
Assessment of the severity of subjective BCRL
For aim 3, the dependent factor is the severity of subjective BCRL at arm and trunk/breast, as determined by the ‘difference in size’ subscale of the Norman Questionnaire.50 The severity of subjective arm BCRL is based on the perceived size difference at three sites (hand, lower arm and upper arm) is measured with two items: (1) frequency, rated from ‘0: not applicable to 4: almost always’ and (2) severity, rated from ‘0: not applicable to 3: very noticeable’. The score is between 0 and 21. The severity of subjective trunk/breast BCRL is based on the perceived size difference in the trunk and breast or breast region with the same two items: (1) frequency, rated from ‘’: not applicable’ to ‘4: almost always’ and (2) severity, rated from ‘0: not applicable’ to ‘3: very noticeable’. The total score ranges from 0 to 14.
The sample size calculation is based on the detection of prevalence rates between the periods (aim 1). Expected prevalence rates based on literature and expert opinion of subjective BCRL are 60%–70% at 1 month postsurgery, 50%–60% at 6 months postsurgery and 40% at 12 months postsurgery.13–15 40 51 The sample size has been determined to have at least 80% power to detect—based on a two-sided McNemar test with alpha equal to 0.05—a difference of 10 percentage points in prevalence between two periods, assuming a moderate correlation of 0.40–0.50 between two periods. Based on a correlation of 0.45, 218 patients are required in total. The assumed correlation corresponds to percentages of discordant patients equal to 9% and 19%. Specifically, 9% of patients who initially did not report subjective BCRL went on to develop it, while 19% of those with subjective BCRL transitioned to having no subjective BCRL. Anticipating a dropout of 5%, a total of 230 patients are required (218×1/0.95=230). With this number of patients, an expected prevalence rate of 60% can be estimated with a precision of 6.5% (halfwidth of the 95% CI) and a rate of patients who transition from one to another state equal to 10% with a precision of 4%.
Note that the study is not explicitly powered for aims 2 and 3, but with the planned sample size, effect sizes as small as 0.5 (Cohen’s d) can be detected at a single time point when 60% have subjective arm or trunk/breast BCRL and 20% have objective arm or trunk/breast BCRL (based on a two-sided t-test with alpha equal to 0.05).
SPSS (IBM) and SAS software will be used to conduct the statistical analyses.
Statistical analyses for aim 1
The prevalence rate of subjective BCRL and objective BCRL will be compared between three time points (at T1, T2 and T3) using a multinomial logistic regression model with generalised estimating equations (GEE) to deal with the within-patient correlation. 95% two-sided CIs for the prevalence rates of no BCRL, subjective BCRL and objective BCRL will be reported at each time point. Within-patient changes will be described with transition matrices between consecutive time points. Intermittent missing information on BCRL and dropout will be handled with multiple imputation using the fully conditional specification approach.52 The imputation model is described in the imputation section. Note that the analyses will be performed separately for arm and trunk/breast BCRL.
Statistical analyses for aim 2
For aim 2a, patients who report swelling at the arm but have no objective arm BCRL will be assigned to the subjective arm BCRL group (see table 2, item 1.1). Patients who meet at least one of four predefined criteria for objective arm BCRL (see table 2, item 2.1) will be assigned to the objective arm BCRL group. Patients who do not report any size differences in the hand, lower arm or upper arm according to the Norman Questionnaire (see table 2, item 1.1) will be assigned to the no self-reported swelling at arm group. For aim 2b, the assignment of patients to the study groups will be performed using a similar approach as aim 2a (see table 2, items 1.2 and 2.2).
Exploratory analyses at each time point separately will be based on Mann-Whitney U test and χ² test to compare factors related to four potential underlying mechanisms (ie, lymphatic, nociceptive, neuropathic and central sensory problems) between ‘patients with subjective BCRL and patients with no self-reported swelling’ and ‘patients with subjective BCRL and patients with objective BCRL’. Since the variable ‘amount of swelling’ partially defines group membership, this outcome will only be compared between patients with subjective BCRL and patients with no self-reported swelling. When comparing patients with objective BCRL and subjective BCRL, ‘amount of swelling’ will only be reported descriptively. In addition, differences in patient-related (ie, age, a relative change in weight vs a previous time point, psychosocial characteristics (ie, level of depression, stress and anxiety based on the DASS-21 score), cancer-related and treatment-related factors (ie, type of surgery (mastectomy or breast conserving surgery), chemotherapy (yes/no), taxane-based chemotherapy (yes/no), radiotherapy (yes/no), hormone therapy (yes/no) and content of physiotherapy (no physiotherapy/suboptimal physiotherapy/optimal physiotherapy)) will be verified.
Multivariable binary logistic regression models will be used to determine factors which contribute to the occurrence of subjective BCRL (compared with objective BCRL and not reporting swelling, respectively). The analysis will thus be performed on subsets of the data (‘subjective and objective BCRL’ and ‘subjective and no self-reported swelling’, respectively), all three time points combined. By adding interaction terms, it can be verified if the relation with a specific factor differs between the time points. GEE and multiple imputation (see ‘Imputation’ section) will be used to handle the within-patient correlation and missing information, respectively. Apart from possible factors (lymphatic, nociceptive, neuropathic and central sensory problems), the model will contain the mentioned patient-related, cancer-related and treatment-related factors and variables referring to the physiotherapy characteristics. Note that some of the considered factors are time-constant (ie, type of surgery), but most of them are time-varying. To evaluate the impact of a specific therapy, binary variables such as chemotherapy and radiotherapy will take the value one as soon as the therapy is started. Analyses will be performed separately for arm and trunk/breast BCRL. Lagged information (ie, amount of swelling or group membership at a previous time point) will only be used in additional, exploratory analyses.
Statistical analyses for aim 3
The factors for the severity of the subjective BCRL will be analysed on a subgroup of observations, that is, on the patients who report at the specific time-point subjective BCRL. Hence, the number of measurements per patient will depend on the number of times the patient experiences subjective BCRL. The same set of factors as for aim 2 will be verified as possible contributing factors (independent factors). Exploratory analyses at each time point separately will be based on Spearman’s correlation and Mann-Whitney U test to determine relations of factors (continuous and binary factors) with the severity of subjective BCRL, respectively at each time point, that is, 1, 6 and 12 month(s) postsurgery.
A multivariable longitudinal model for repeated measures (with an unstructured covariance matrix for the time points) will be constructed to determine the factors that contribute to the severity of subjective BCRL. The model will thus be fitted on all three time points, that is, 1, 6 and 12 month(s) combined. By adding interaction terms, it can be verified if the relation with a specific factor differs between the time points. The model will include the same factors as in aim 2. Multiple imputation will be used to handle missing severity and/or factor information. Analyses will be performed separately for arm and trunk/breast BCRL.
Imputation
Missingness can occur for the information required to define group membership, as well as for the severity of BCRL and all considered factors. Therefore, multivariate imputation will be performed using the fully conditional specification approach.52 In this approach, for each of the variables with a missing value, a regression model is specified using all other predictors and all outcome variables as covariates. The process is iterated (one iteration consists of one cycle through all variables) until convergence to the multivariate distribution is obtained. 20 imputed datasets will be created and results will be combined using Rubin’s rule.53 These datasets will be used for all aims except for the exploratory analyses, which will be based on observed information.
The study monitoring plan has been developed by a clinical research assistant who is not involved in data collecting (HV). All aspects of the design, delivery, quality assurance and data analysis will be checked at least once a year. In addition, a yearly meeting of the steering committee of the LymphSens study, consisting of the independent experts, a patient representative and a statistician in addition to the data management team, will be organised to oversee the study procedures and data collection.
The input of patients with breast cancer contributed to composing the aims of the study. In addition, a patient representative in the steering committee of the LymphSens study provided input on the execution and feasibility of data collection prior to the start of the recruitment.
The LymphSens study will be conducted in accordance with the principles of the Helsinki Declaration and the principles of Good Clinical Practice (GCP). The study protocol has been approved by the main Ethical Committee of UZ/KU Leuven (S68133) and the local Ethical Committee of UZA/UAntwerp (5676-003252). Furthermore, future changes in the protocol, ICF or other relevant documents will be submitted to the ethical committee. The study is registered in Clinical.Trials.gov (NCT06324721).
The research team is dedicated to fully disclosing the results of the study. The results of the study will be disseminated as published articles in peer-reviewed high-impact journals and presentations at international and national scientific meetings.
During this project, the Trial Master File (TMF) and the Investigator Site File (ISF) will be maintained to ensure regulatory compliance and proper documentation. The TMF contains all essential documents related to the conduct of the study, providing evidence that the study is being conducted in line with GCP and regulatory requirements. The ISF is maintained at each investigational site and includes site-specific documents, ensuring proper management and oversight at the local level. Together, these files help ensure transparency, accountability and the integrity of the data of the study.
Data are prospectively collected from the participating patients by the co-investigators (CG, NV, KV, LS) using the electronic case report forms and stored in the Research Electronic Data Capture (REDCap) system. Only the unique study identification number is recorded in the databases. Personal data of each patient are separated into a subject identification log, which is the only link between the unique study identification number and the patient’s identity. This log is password protected and stored on a secure network drive, separate from other study files and is accessible only to the principal investigator and the relevant study team members. The General Data Protection Regulation will be respected regarding pseudonymisation of the data.
All relevant digital study data are maintained by the principal investigator and the co-investigators. Data will be uploaded to a shared network drive folder linked to their University of Leuven (KU Leuven) or University of Antwerp (UAntwerp) account and/or stored in REDCap. In REDCap, patients will be assigned to a Data Access Group (DAG) to ensure data are kept separate for each study site. Investigators will be linked to a DAG based on their site affiliation, ensuring that they can only enter and view data from their own site. This approach safeguards patient confidentiality and ensures compliance with data access rules. The same approach will be applied regarding the storage of other relevant study data in site-specific shared network folders. The Trial Data Manager will perform thorough consistency checks on the received data. Only coded information will be extracted from REDCap and used for downstream research analysis. On project completion, both online and paper-based research data will be preserved on a shared network drive linked to a KU Leuven or UAntwerp account. All data will be stored for 25 years after the project concludes, in accordance with the requirements of the Clinical Trial Centre at UZ Leuven.
In the LymphSens study, the complexity and multifactorial nature of subjective BCRL will be addressed through carefully designed aims and study methodology. This multicentre longitudinal study will determine the prevalence of subjective and objective BCRL and the transitions between the different types of BCRL (no-subjective-objective BCRL) and unravel possible underlying mechanisms for subjective BCRL, objective BCRL and no-self-reported swelling. In addition, the present study will determine the factors that contribute to the severity of subjective arm or trunk/breast BCRL. As a result, the present study will acquire crucial knowledge on who develops different types of BCRL and why patients have subjective arm or trunk/breast BCRL. This advanced understanding will lead to improvements in the monitoring, evaluation and management of all types of BCRL, with particular focus on the large group of patients experiencing subjective arm or trunk/breast BCRL.
Not applicable.
