Abstract
Background
Arm symptoms commonly endure in post-breast cancer period and persist into long-term survivorship. However, a knowledge gap existed regarding the interactions among these symptoms. This study aimed to construct symptom networks and visualize the interrelationships among arm symptoms in breast cancer survivors (BCS) both with and without lymphedema (LE).
Patients and Methods
We conducted a secondary analysis of 3 cross-sectional studies. All participants underwent arm circumference measurements and symptom assessment. We analyzed 17 symptoms with a prevalence >15%, identifying clusters and covariates through exploratory factor and linear regression analysis. Contemporaneous networks were constructed with centrality indices calculated. Network comparison tests were performed.
Results
1116 cases without missing data were analyzed, revealing a 29.84% prevalence of LE. Axillary lymph node dissection [ALND] (vs sentinel lymph node biopsy [SLNB]), longer post-surgery duration, and radiotherapy significantly impacted overall symptom severity (P < .001). “Lymphatic Stasis,” “Nerve Injury,” and “Movement Limitation” symptom clusters were identified. Core symptoms varied: tightness for total sample network, firmness for non-LE network, and tightness for LE network. LE survivors reported more prevalent and severe arm symptoms with stronger network connections than non-LE group (P = .010). No significant differences were observed among different subgroups of covariates (P > .05). Network structures were significantly different between ALND and SLNB groups.
Conclusion
Our study revealed arm symptoms pattern and interrelationships in BCS. Targeting core symptoms in assessment and intervention might be efficient for arm symptoms management. Future research is warranted to construct dynamic symptom networks in longitudinal data and investigate causal relationships among symptoms.
Keywords: breast neoplasm, arm symptom, lymphedema, symptom network, network analysis
This study aimed to construct symptom networks and visualize the interrelationships among arm symptoms in breast cancer survivors with and without lymphedema.
Implications for Practice.
Our study revealed a high prevalence of arm symptoms in BCS with or without LE and modeled symptom networks, providing a new perspective of understanding arm symptoms. The findings emphasize the importance for clinicians working with BCS to prioritize the management of arm symptoms. Special attention should be paid to survivors receiving ALND, RT, and a longer post-surgery duration. Based on the symptom clusters identified, a rapid assessment can be conducted to determine the primary symptom concerns in patients, facilitating targeted care. Additionally, the core symptoms should be prioritized when developing symptom management strategies, to increase the efficacy of interventions.
Introduction
As one of the most commonly diagnosed cancers worldwide, advancements in cancer screening and treatment have led to a notably high survival rate and an increasing number of breast cancer survivors (BCS).1 Yet, the combination of breast cancer surgery along with radiotherapy and systemic therapy leaves the risk of local late toxicity following breast cancer treatment.2 Among these, arm symptoms such as lymphedema (LE), limited arm mobility, pain, heaviness, etc., specifically triggered by axillary lymph node dissection (ALND) and regional nodal irradiation (RNI), were frequently reported and would persist into long-term survivorship.3-5 A recent study disclosed that approximately one-third of 882 BCS reported arm symptoms at a mean time of 10.5 years since surgery.6 The presence of arm symptoms was associated with poorer upper-body function, lower physical activity levels, diminished work productivity, and decreased quality of life.3,5,7
Breast cancer-related lymphedema (BCRL), a chronic lymphatic condition characterized by interstitial accumulation of protein-rich fluid, leading to inflammation, adipose tissue hypertrophy, and fibrosis,8,9 is a progressing and uncured complication, with an pooled incidence of 21.9% (95% CI, 19.8%-24.0%).10 Once established, it required costly and time-consuming management, which might explain the mounting research endeavored to study the risk prediction and management of BCRL.9 Additionally, the onset of BCRL is associated with a range of arm symptoms, such as swelling, stiffness, tightness, fatigue, heaviness, etc., which might in turn exacerbate the arm symptoms caused by breast cancer treatments.11 This has been confirmed by a previous study showing that BCRL survivors reported a significantly higher number of symptoms than those without a diagnosis of LE.5
A previous study highlighted that arm symptoms, rather than LE, significantly influenced the quality of life of BCS.5 This emphasizes the critical importance of addressing and managing arm symptoms in this population. The implication is that interventions focused on alleviating arm symptoms may offer greater benefits in enhancing the overall well-being of BCS. However, clinicians have primarily focused their attention on detecting and treating LE, overlooking the assessment and management of arm symptoms among BCS, both with and without BCRL. Arm symptoms manifest as phenotypic features of LE and may precede its diagnosis. Armer et al12 published a landmark study demonstrating that self-reported arm symptoms including “swelling now” and “heaviness in the past year,” were significant predictors of BCRL. This study underscored the importance of arm symptom management for patients with breast cancer. Attending to these symptoms is crucial not just for early diagnosis but, more significantly, early symptom management can decelerate edema progression and enhance survivors’ quality of life. Presently, there are no established clinical practice guidelines or instructions that clinicians or survivors can use, leaving the burden of arm symptoms unaddressed. This lack of evidence-based practice underscored the need to better understand arm symptoms among BCS, to lay foundation for further targeted interventions.
The symptoms rarely appear individually but typically occur in clusters. Nevertheless, existing literature predominantly revolves around detailing the incidence and severity of isolated arm symptoms, the clustering pattern and interrelationship among multiple arm symptoms remains unclear. Symptom network represents an innovative paradigm that allows for the visualization and exploration of internal symptom network structures within a specific population.13,14 This approach aids researchers in identifying associations between symptom severities and exploring mechanistic indicators of symptoms from a holistic perspective.13 Building on this approach, the present study aimed to (1) generate symptom networks of arm symptoms among BCS; (2) explore the complex relationship and core symptoms among arm symptoms; (3) assess differences in symptom networks among different demographic and clinical covariates, and different LE status.
Patients and methods
Study design and settings
We conducted a secondary analysis based on 3 cross-sectional studies.15-17 A total of 1129 BCS were involved in parent studies, with 341 (30.2%) being diagnosed with BCRL. The incidence of BCRL was within the average range reported by prior evidence.18 The parent studies were approved by the Biomedical Ethics Committee of Peking University (approval numbers: IRB00001052-11051, IRB00001052-15073, and IRB00001052-21123), and were conducted in accordance with the Helsinki Declaration. As this study involved a secondary analysis of existing anonymous data, informed consent is not required.
Participants
Participants involved in the parent studies met the following criteria: (1) pathologically diagnosed with primary breast cancer; (2) aged 18 years or older; (3) more than 1 month post-breast cancer surgery; (4) being able to read and communicate in Chinese; and (6) provided informed consent. Patients with other concurrent malignant tumors, with history of arm surgery or arm dysfunction, with history of lymphatic diseases or primary LE, with presence cardiogenic/nephrogenic/hepatic edema, were excluded. After excluding cases with missing data on arm circumference measurement and self-reported arm symptoms, 1116 cases were included for analysis.
Variables and measures
Data on participants’ sociodemographic and clinical characteristics, including age, education, employment, marital status, medical insurance, tumor location, types of breast cancer surgery, types of axilla surgery, treatments received (eg, chemotherapy, radiotherapy, and endocrine therapy), arm circumference, and 24 self-reported symptoms measured by The Part I of the Breast Cancer and Lymphedema Symptom Experience Index (BCLE-SEI), were extracted from the parent datasets.17 Arm circumferences were measured by a retractable non-stretch soft tape from the ulna styloid to the axilla at 4-cm or 10-cm intervals. The Maximum interarm circumference difference ≥2 cm in the ipsilateral arm compared to the contralateral arm, was used to diagnose BCRL.19,20 The sociodemographic and clinical characteristics, and arm symptoms were collected by self-reported questionnaires. Symptoms were scored from 0 to 4, and higher scores indicated severe symptoms.17
Data analysis
Statistical analyses were performed with IBM SPSS 26.0 for descriptive analyses, difference test, regression analysis, and exploratory factor analysis (EFA), R version 4.2.3 with the qgraph package for network visualization, bootnet for stability analyses, mgm for predictability test, and NetworkComparisonTest for network comparison tests (NCT). There was no missing data to be handled in this study.
Descriptive statistics were calculated for sociodemographic and clinical variables. To ensure the stability of statistical analysis, we included 17 symptoms with a prevalence above 15% for symptom cluster and symptom network analysis. Linear regression analysis was performed to identify covariates for overall 17-symptom severity. Significant factors with P value < .001 were included as covariates to examine the relationships among symptoms. We chose a stringent P value threshold of <.001 to enhance the reliability of our findings, given our large sample size.21 We performed EFA using principal components and oblique rotation to examine symptom clusters, defining by factor loadings ≥0.4, Cronbach’s α ≥ 0.70, and cumulative variance contribution >50%.22
Based on the identified symptom clusters, regularized partial correlation networks analyses were estimated using the least absolute shrinkage and selection operator algorithm and extended Bayesian information criterion model selection, for the total sample (including and excluding covariates) and for subgroups of LE diagnosis (LE or Non-LE), post-surgery years (0-1 year, 1-3 years, 3-5 years, and 5+ years), and RT (yes or no), axillary surgery types (ALND or sentinel lymph node biopsy [SLNB]), separately. Each node denotes a distinct symptom, and each edge within the network represents the conditional independent relationships between 2 nodes. Edge thickness reflected the intensity of the association between interconnected nodes and green edge indicated positive relationships. We conducted a centrality analysis with 3 indices: strength, betweenness, and closeness. Core symptoms were identified based on the highest centrality coefficients.23 ∑s (the absolute value of all Spearman coefficients between 2 nodes) was calculated to indicate network density.
Nonparametric Bootstrapping (nBoots = 1000) was used to assess the accuracy and stability of the network. Edge weights with 95% CIs were bootstrapped to measure the edge’s accuracy. Then, a case‐dropping subset bootstrap was used to determine the centrality stability of the coefficient (CS-coefficient). The CS-coefficient is recommended to be above 0.50 and at least >0.25 for the centrality indices to be trustworthy.23 We computed predictability for each node to quantify the mean explained variance of the estimated network. Predictability assesses how connected nodes can predict the value of each node, offering a practical metric for targeting interventions. To formally test for between-group network differences (ie, ALND vs SLNB, RT vs Non-RT), NCTs were performed on subsamples with 1000 permutations to assess global network strengths (absolute sums of all edge weights) and network structures (distributions of edge weights) between the 2 networks. The strength of each edge between the 2 networks was assessed using Holm-Bonferroni correlations for multiple comparisons. As the NCT can lose power when sample sizes are not equal, we chose to include the covariates including types of axillary surgery, RT, and post-surgery duration, to ensure balanced sample sizes between subgroups (n > 200).
Results
Sample characteristics
In total, we included 1116 cases of BCS, with 29.84% (333/1116) with LE. The mean age was 55.25 (SD = 10.95, range: 24-85) years. The average post-surgery duration was 39.42 (Mean = 26.5, IQR: 7-52) months. Most participants were coupled (92.7%) and with medical insurance (97.0%). 76.8% of them were unemployed. In terms of clinical characteristics, 80.4% underwent mastectomy, 76.3% received ALND, and 84.7% were treated with chemotherapy. Nearly half of the participants received radiotherapy (54.3%) and endocrine therapy (52.0%). There were significant differences between groups with and without LE on post-surgery duration, type of breast cancer surgery, type of axillary surgery, chemotherapy, and radiotherapy (P < .05). Detailed information is presented in Table 1.
Table 1.
Characteristics of the participants (n = 1116).
| Characteristics | Overall (n = 1116) N (%)/mean ± SD (IQR) |
Non-LE (n = 783) N (%)/mean ± SD (IQR) |
LE (n = 333) N (%)/mean ± SD (IQR) |
t/Z/χ2 | P |
|---|---|---|---|---|---|
| Age (years) | 55.25 ± 10.95, 55 (47,63) |
54.94 ± 11.17, 55 (47, 62) |
55.97 ± 10.39, 56 (59, 63) |
−1.439a | 0.150 |
| Post-surgery months | 39.42 ± 49.14, 26.5 (7, 52) |
28.76 ± 29.52, 21 (6, 47) |
64.48 ± 71.83, 44 (21, 69) |
−11.776a | <0.001 |
| Education | |||||
| Primary school or below | 97 (8.7) | 66 (8.4) | 31 (9.3) | 8.837c | 0.116 |
| Middle school | 287 (25.7) | 207 (26.4) | 80 (24.0) | ||
| High school | 298 (26.7) | 204 (26.1) | 94 (28.2) | ||
| Junior college | 176 (15.8) | 122 (15.6) | 54 (16.2) | ||
| University or above | 243 (21.8) | 178 (22.7) | 65 (19.5) | ||
| Unclear | 15 (1.3) | 6 (0.8) | 9 (2.7) | ||
| Marital status | |||||
| Single | 81 (7.3) | 55 (7.0) | 26 (7.8) | 0.213c | 0.644 |
| Coupled | 1035 (92.7) | 728 (93.0) | 307 (92.2) | ||
| Employment | |||||
| Unemployed | 857 (76.8) | 604 (77.1) | 253 (76.0) | 0.177c | 0.674 |
| Employed | 259 (23.2) | 179 (22.9) | 80 (24.0) | ||
| Medical insurance | |||||
| Without insurance | 34 (3.0) | 21 (2.7) | 13 (3.9) | 1.181c | 0.277 |
| With insurance | 1082 (97.0) | 762 (97.3) | 320 (96.1) | ||
| Tumor location | |||||
| Left breast | 536 (48.0) | 364 (46.5) | 172 (51.7) | 2.583d | 0.256 |
| Right breast | 570 (51.1) | 412 (52.6) | 158 (47.4) | ||
| Bilateral breast | 10 (0.9) | 7 (0.9) | 3 (0.9) | ||
| Type of breast surgery | |||||
| Mastectomy | 897 (80.4) | 600 (76.6) | 297 (89.2) | 23.371c | <0.001 |
| Lumpectomy | 219 (19.6) | 183 (23.4) | 36 (10.8) | ||
| Type of axillary surgery | |||||
| SLNB | 264 (23.7) | 228 (29.1) | 36 (10.8) | 104.442c | <0.001 |
| ALND | 852 (76.3) | 555 (70.9) | 297 (79.2) | ||
| Adjuvant chemotherapy | |||||
| No | 171 (15.3) | 147 (18.8) | 24 (7.2) | 24.092c | <0.001 |
| Yes | 945 (84.7) | 636 (81.2) | 309 (92.8) | ||
| Radiotherapy | |||||
| No | 510 (45.7) | 387 (49.4) | 123 (36.9) | 14.684c | <0.001 |
| Yes | 606 (54.3) | 396 (50.6) | 210 (63.1) | ||
| Endocrine therapy | |||||
| No | 536 (48.0) | 367 (46.9) | 169 (50.8) | 1.409c | 0.235 |
| Yes | 580 (52.0) | 416 (53.1) | 164 (49.2) | ||
| Post-surgery duration (years) | |||||
| <1 | 378 (33.9) | 323 (41.3) | 24 (7.2) | 96.813c | <0.001 |
| 1-3 | 287 (25.7) | 204 (26.1) | 83 (24.9) | ||
| 3-5 | 248 (22.2) | 160 (20.4) | 88 (26.4) | ||
| >5 | 203 (18.2) | 96 (12.3) | 107 (32.1) | ||
aIndependent t test.
bMann-Whitney U test.
cFisher’s test.
dChi-square test.
Prevalence and severity of arm symptoms
As shown in Supplementary Table S1, the average number of self-reported arm symptoms was 4 (IQR: 2, 9) for the total sample. Participants in LE group reported significantly more symptoms than those in non-LE group (median = 10, IQR: 5, 14 vs median = 3, IQR:1, 6, Z = 15.952, P < .001). The LE group reported more prevalent and severe symptoms that the non-LE group (Table 2). The most frequently reported symptoms were swelling (49.2%), fatigue (48.7%), and heaviness (47.0%) in total sample. For LE group, swelling (84.7%), heaviness (75.7%), and fatigue (69.1%) exhibited the highest prevalence. For non-LE group, the prevalence of fatigue (40.0%), limited-shoulder-movement (38.7%), and numbness (35.8%) ranked the top. The average severity for all 24 symptoms was 0.36, and 0.46 for the selected 17 symptoms. The most severe symptoms in non-LE group were limited-shoulder-movement and fatigue, and arm-swelling and heaviness in LE group. The prevalence and severity of all symptoms were significantly higher in LE group than those in non-LE group (all P < .001), except the prevalence of breast-swelling and numbness. The results of linear regression model analysis indicated that ALND (vs SLNB), longer post-surgery duration, and radiotherapy (vs Non-RT) had significantly impact on the severity of 17 symptoms (P < .001, Table 3). These factors were further included in the network analysis as covariates.
Table 2.
Prevalence and severity of arm symptoms among breast cancer survivors with or without lymphedema.
| Symptoms | Overall (n = 1116) | Non-LE (n = 783) | LE (n = 333) | Comparison of prevalence | Comparison of severity | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| Frequency (%) | Median (IQR), Average | Frequency (%) | Median (IQR), Average | Frequency (%) | Median (IQR), Average | χ2 | P | t | P | |
| Limited-shoulder-movement | 504 (45.2) | 0 (0, 1), 0.66 | 303 (38.7) | 0 (0, 1), 0.51 | 201 (60.4) | 1 (0, 2), 1.00 | 44.272 | <.001 | −8.048 | <.001 |
| Limited-elbow-movement | 251 (22.5) | 0 (0, 0), 0.34 | 75 (9.6) | 0 (0, 0), 0.14 | 176 (52.9) | 1 (0, 2), 0.84 | 250.981 | <.001 | −12.946 | <.001 |
| Limited-wrist-movement | 186 (16.7) | 0 (0, 0), 0.25 | 59 (7.5) | 0 (0, 0), 0.09 | 127 (38.1) | 0 (0, 1), 0.61 | 157.544 | <.001 | −10.100 | <.001 |
| Limited-fingers-movement | 228 (20.4) | 0 (0, 0), 0.29 | 117 (14.9) | 0 (0, 0), 0.18 | 111 (33.3) | 0 (0, 1), 0.55 | 48.61 | <.001 | −6.985 | <.001 |
| Limited-arm-movement | 335 (30.0) | 0 (0, 1), 0.48 | 139 (17.8) | 0 (0, 0), 0.25 | 196 (58.9) | 1 (0, 2), 1.03 | 187.931 | <.001 | −12.569 | <.001 |
| Arm-swelling | 549 (49.2) | 0 (0, 1), 0.84 | 267 (34.1) | 0 (0, 1), 0.41 | 282 (84.7) | 2 (0, 3), 1.84 | 239.199 | <.001 | −20.567 | <.001 |
| Stiffness | 330 (29.6) | 0 (0, 1), 0.43 | 143 (18.3) | 0 (0, 0), 0.22 | 187 (56.2) | 1 (0, 2), 0.92 | 161.084 | <.001 | −11.886 | <.001 |
| Tightness | 416 (37.3) | 0 (0, 1), 0.56 | 183 (23.4) | 0 (0, 0), 0.28 | 233 (70.0) | 1 (0, 2), 1.19 | 216.98 | <.001 | −14.880 | <.001 |
| Heaviness | 524 (47.0) | 0 (0, 1), 0.76 | 272 (34.7) | 0 (0, 1), 0.44 | 252 (75.7) | 1 (0, 2), 1.50 | 157.203 | <.001 | −15.345 | <.001 |
| Fibrosis (toughness or thickness of skin) | 252 (22.6) | 0 (0, 0), 0.35 | 86 (11.0) | 0 (0, 0), 0.15 | 166 (49.8) | 0 (0, 1), 0.82 | 201.886 | <.001 | −11.460 | <.001 |
| Firmness | 397 (35.6) | 0 (0, 1), 0.49 | 201 (25.7) | 0 (0, 1), 0.31 | 196 (58.9) | 1 (0, 1), 0.93 | 112.285 | <.001 | −10.685 | <.001 |
| Tenderness | 204 (18.3) | 0 (0, 0), 0.25 | 109 (13.9) | 0 (0, 0), 0.18 | 95 (28.5) | 0 (0, 1), 0.40 | 33.374 | <.001 | −4.803 | <.001 |
| Pain/aching/soreness | 358 (32.1) | 0 (0, 1), 0.40 | 208 (26.6) | 0 (0, 1), 0.31 | 150 (45.0) | 0 (0, 1), 0.59 | 36.622 | <.001 | −5.774 | <.001 |
| Numbness | 419 (37.5) | 0 (0, 1), 0.46 | 280 (35.8) | 0 (0, 1), 0.42 | 139 (41.7) | 0 (0, 1), 0.56 | 3.565 | .059 | −2.816 | .005 |
| Tingling (pins and needles) | 172 (15.4) | 0 (0, 0), 0.19 | 94 (12.0) | 0 (0, 0), 0.14 | 78 (23.4) | 0 (0, 0), 0.31 | 23.365 | <.001 | −4.395 | <.001 |
| Fatigue | 543 (48.7) | 0 (0, 1), 0.70 | 313 (40.0) | 0 (0, 1), 0.48 | 230 (69.1) | 1 (0, 2), 1.20 | 79.166 | <.001 | −11.336 | <.001 |
| Weakness | 373 (33.4) | 0 (0, 1), 0.44 | 218 (27.8) | 0 (0, 1), 0.34 | 155 (46.5) | 0 (0, 1), 0.66 | 36.735 | <.001 | −5.967 | <.001 |
Table 3.
Linear regression model of overall symptom severity of 17 symptoms (n = 1116).
| Characteristics | β | P |
|---|---|---|
| Age | 0.005 | .889 |
| High education level | −0.030 | .370 |
| Coupled (compared to single) | −0.024 | .403 |
| Employed (compared to unemployed) | −0.028 | .441 |
| With medical insurance (compared to No) | −0.077 | .008 |
| Tumor location | 0.001 | .964 |
| Lumpectomy (compared to mastectomy) | −0.085 | .007 |
| ALND (compared to SLNB) | 0.114 | <.001 |
| Post-surgery years | 0.126 | <.001 |
| Adjuvant chemotherapy (compared to No) | 0.065 | .031 |
| Radiotherapy (compared to No) | 0.117 | <.001 |
| Endocrine therapy (compared to No) | −0.049 | .111 |
| High level LK | 0.020 | .511 |
R 2 = 0.086, Adjusted R2 = 0.075, F = 7.955, P < .001.
Symptom clusters identified through PCA
Three symptom classes were identified with EFA (χ2 = 11501.35, P < .001), explaining 64.58% of the total variance (Table 4). Each factor was named based on the constellation of symptoms within the factor: “Lymphatic Stasis” symptom cluster (swelling, stiffness, tightness, heaviness, fibrosis, firmness, and fatigue), “Nerve Injury” symptom cluster (tenderness, pain/aching/soreness, numbness, tingling), and “Movement Limitation” symptom cluster (limited-shoulder-movement, limited-elbow-movement, limited-wrist-movement, limited-fingers-movement, and limited-arm-movement). The Cronbach’s α value of each symptom cluster were 0.919, 0.778, and 0.843, respectively.
Table 4.
Factor loading promax rotation for symptom occurrence.
| Symptoms | Components | ||
|---|---|---|---|
| Factor 1 lymphatic stasis symptom cluster | Factor 2 nerve injury symptom cluster | Factor 3 movement limitation symptom cluster | |
| Limited-shoulder-movement | 0.415 | 0.369 | 0.270 |
| Limited-elbow-movement | 0.508 | 0.223 | 0.598 |
| Limited-wrist-movement | 0.276 | 0.181 | 0.858 |
| Limited-fingers-movement | 0.111 | 0.190 | 0.864 |
| Limited-arm-movement | 0.485 | 0.245 | 0.616 |
| Swelling | 0.789 | 0.001 | 0.256 |
| Stiffness | 0.711 | 0.340 | 0.284 |
| Tightness | 0.777 | 0.294 | 0.261 |
| Heaviness | 0.841 | 0.172 | 0.156 |
| Fibrosis (toughness or thickness of skin) | 0.697 | 0.229 | 0.304 |
| Firmness | 0.697 | 0.385 | 0.185 |
| Tenderness | 0.181 | 0.716 | 0.260 |
| Pain/aching/soreness | 0.300 | 0.696 | 0.093 |
| Numbness | 0.113 | 0.652 | 0.164 |
| Tingling (pins and needles) | 0.111 | 0.741 | 0.140 |
| Fatigue | 0.760 | 0.280 | 0.060 |
| Weakness | 0.388 | 0.593 | 0.061 |
| Cronbach’s α | 0.919 | 0.778 | 0.843 |
| % of variance | 29.66% | 18.53% | 16.40% |
| % of total variance | 29.66% | 48.19% | 64.58% |
Bold values indicated that the symptoms were categorized as the corresponding factors.
Symptom network estimation of all cases
Figure 1A shows the networks of self-reported arm symptoms among BCS. The density of the total sample network without covariates is 99.63. Limited-wrist-movement had a strong connection with limited-fingers-movement (r = .61); stiffness had a moderate connection with tightness (r = .44); heaviness had a moderate connection with fatigue (r = .37). The node predictability values ranged from 28.1% to 74.4%. Stiffness, tightness, and heaviness had the highest predictability, showing that 74.4%, 69.9%, and 68.4% of their variance can be explained by their neighboring symptoms.
Figure 1.
Symptom networks of arm symptoms for the total sample with and without covariates, and symptom networks of LE and Non-LE group: (A) all sample network without covariates; (B) all sample network with covariates; (C) network of non-LE group; (D) network of LE group.
After introducing clinical covariates into the network (Figure 1B), the weight of each connection changed, but the connections between symptoms were almost identical. Some new connections appeared, such as the connection between firmness and stiffness (r = .10), pain and numbness (r = .11), etc. Post-surgery years positively correlated with arm-swelling (r = .18). The prevalence (<1 year: 33.07%, 1-3 years:55.05%, 3-5 years:60.07%, and >5 years:67.49%) and severity (<1 year: mean = 0.45, SD = 0.73; 1-3 years: mean = 0.87, SD = 1.01; 3-5 years: mean = 0.94, SD = 1.12; and >5 years: mean = 1.39, SD = 1.30) of arm-swelling increased progressively with the number of post-surgery years. Type of axillary surgery, and radiotherapy showed weak connections with all symptoms. More details about the weight of each connection in the network with and without clinical covariates are presented in Supplementary Tables S2 and S3. Moreover, the results of centrality analyses indicated that tightness (rs = 1.404, rb = 1.320, rc = 1.465) was the most central symptom (Figure 2A). Limited-arm-movement (rbs = 5.506), limited-elbow-movement (rbs = 5.447) had the largest values of bridge strength (Figure 2B).
Figure 2.
Centrality index, bridge centrality index, accuracy and stability of the total sample symptom network, and results of difference tests. (A) Centrality index; (B) bridge centrality index; (C) bootstrap analysis results of the edge weights; (D) correlation stability coefficient for strength, expected influence, and closeness; (E) bootstrapped difference test for edges; (F) bootstrapped difference test for nodes.
Supplementary Figure S1A shows the bootstrap analysis results of edge weights for the total sample network. The bootstrapped CIs were small, indicating good accuracy of the network. For the bootstrap subset (Supplementary Figure S1B), the correlation stability coefficient was 0.75 for expected influence and 0.59 for strength, suggesting that the network remained stable. Supplementary Figure S2A shows the results of the bootstrapped edge difference test. The bootstrapped difference test for edge weights showed that the 2 strongest edge weights, “limited-wrist-movement and limited-fingers-movement” and “stiffness and tightness,” were significantly different from approximately 95% of other edge weights (Supplementary Figure S2A). Supplementary Figure S2B shows the results of the bootstrapped node difference test. Numbness and tightness significantly differed from other nodes (DTs = 1.20).
Comparison of networks between LE and non-LE groups
Figure 3A, 3B presents the networks of non-LE and LE groups, showing that tightness (rs = 1.547, rb = 0.400, rc = 0.664) and stiffness (rs = 1.500, rb = -0.845, rc = 0.621) were the most central symptoms, respectively (Supplementary Figures S3, S4). Tightness and stiffness showed the highest predictability (tightness: R2 = 0.755; stiffness: R2 = 0.604) in the LE and non-LE network, respectively. Bootstrap analysis results of edge weights and the subset bootstrap results showed the networks of LE and non-LE group were accurate and stable (Supplementary Figures S3, S4). For NCT analysis, no significant differences were found between LE and non-LE networks based on the network invariance test (P = .347) and global strength invariance test (network strength for non-LE group: 7.30 vs LE group: 8.07, P = .119). However, regarding the edge invariance test, the network of LE group compared with the network of non-LE group showed stronger connections between symptoms (Figure 3A, 3B), but connection between numbness and tingling significantly weakened (P < .05).
Figure 3.
Symptom networks of LE and Non-LE group: (A) network of non-LE group; (B) network of LE group.
Network comparisons by covariates
We compared symptom networks between different subgroups of post-surgery duration, type of axillary surgery, and RT (Supplementary Figures S5-S7). However, there were no significant differences between RT groups (network strength for non-RT: 7.47 vs RT: 7.99, P = .287; network invariance test: P = .911), and post-surgery duration groups (network strength for 0-1 year: 7.81 vs 1-3 years: 7.67 vs 3-5 years: 8.03 vs 5+ years: 7.80, all P < .05; network invariance test: P ≥ .05). For networks between different axillary surgery groups, there was significant difference in network structure (network invariance test: P = .010), but not in network strength (network strength for SLNB: 7.56 vs ALND: 8.23, P = .307). The edge invariance test showed that 13 out of 288 edges were significantly different between networks of ALND and SLNB (P < .05). The symptom clustering characteristics of ALND network were more pronounced than the SLNB network (Supplementary Figure S7).
Discussion
This study is the first to examine the patterns and interrelationships among arm symptoms in BCS using network analysis. Our results identified the most frequently reported and most severe arm symptoms in LE and non-LE group, and 3 symptom clusters including “Lymphatic Stasis,” “Nerve Injury,” and “Movement Limitation.” Tightness was the core symptom for the total sample network, while firmness and tightness were identified as core symptoms for networks of non-LE and LE group, respectively. Types of axillary surgery, RT, and post-surgery duration were identified as covariates for overall symptom severity but did not significantly influence the network strength and structure. The network structure of LE group was significantly different from that of non-LE group. Exploring the intricate networks of arm symptoms in BCS has deepened our comprehension of symptomatology.
The study of arm symptom clusters in BCS has been relatively understudied in the field. Only one previous study was found, reporting 3 symptom clusters: impaired limb mobility, fluid accumulation and discomfort, in BCS at 4-8 weeks and 12 months post-surgery.24 The symptom clusters were similar to our findings, although they were named differently and encompassed different specific symptoms. The most notable differences were observed between the “discomfort” and “Nerve Injury” symptom clusters. The “discomfort” cluster, observed at 4-8 weeks post-surgery, comprised finger limitation and tenderness, and included tenderness, blistering, and pain at 12 months. In contrast, the “Nerve Injury” cluster in our study encompassed tenderness, pain, numbness, tingling, and weakness. The difference might be explained by the sample size and short post-surgery periods, which had been acknowledged as limitations by the authors.24 We can assume that the symptom clusters remain stable among different patient groups; however, further validation in diverse populations is still needed. It had been reported that BCS reported up to 35 upper limb symptoms or signs, which would be challenging for health care providers to consider so many symptoms.25 Symptom clusters streamline the understanding and assessment of arm symptoms by reducing and categorizing dimensions of interconnected symptoms.13
Additionally, we found that the most frequently reported arm symptoms identified in this study were consistent with the reports of a systematic review, where the most frequently reported symptoms were swelling (80.9%) and heaviness (66.7%) in LE group, and tenderness (37%) and numbness (27%) in non-LE group.25 These results could be pathophysiologically explained, since swelling and heaviness were typical symptoms of lymph fluid accumulation, while tenderness and numbness were more likely to be caused by nerve injury.12 In addition to symptom prevalence, our study revealed that swelling and heaviness were the most prevalent and severe symptoms for LE group. Our findings partially verified the study by Armer et al12 which reported that “swelling” and “heaviness” were predictive of BCRL (a maximal difference of at least 2 cm), while numbness was more common in non-BCRL patients with a maximal difference less than 2 cm. Thus, these self-reported symptoms, such as swelling, heaviness, numbness and tenderness, should be particularly attended to and managed for breast cancer survivors by clinicians to prevent and early detect BCRL, along with objective assessments. Additionally, focusing on the alleviation of the most prevalent and severe symptoms may help reduce distress for survivors. However, the effectiveness of managing symptoms in isolation is likely to be limited.
Identifying core symptoms and underlying mechanisms will guide efficient symptom management. Through network analysis, for the first time, we were able to examine core symptoms of arm symptoms in BCS, discovering tightness as core symptom for the LE group. Generally, we consider tightness, the sensation of increased skin tension due to lymphatic fluid accumulation in the interstitial spaces, as a signal symptom for LEs.11 Hence, the results that tightness was the core symptom in LE network align with our expectation. To our surprise, tightness was also the most central symptom in the total sample network. We noticed that 23.4% of patients without LE reported tightness in our study. Supposing that this could be explained by the fact that patients with subclinical LE might have experienced the symptom of tightness even before exhibiting an increase in limb circumference.26 However, the current study could not distinguish patients with subclinical LE due to limited data from the parent studies, highlighting the need for further investigation. Stiffness categorized to the “Lymphatic Stasis” symptom cluster emerged as a core symptom with the highest predictability in non-LE symptom network. Previous evidence supported self-reported symptoms like tightness and stiffness as predictors for LE development, underscoring the importance of early detection and management.12,25,27 Core symptoms refer to the central symptoms that have the strongest connections with other symptoms in symptom networks. By understanding and targeting core symptoms, researchers and health care professionals can develop more effective interventions for managing and alleviating the overall arm symptomatology in BCS.
Our findings showed that ALND, radiotherapy, and longer post-surgery duration were correlated with more severe arm symptoms, supporting previous studies.4,28 However, we did not detect any significant differences in network structure and global network strength between subgroups of these covariates. One possible explanation is that symptom network is constructed by modeling the interactions among the constituent factors, rather than symptom incidence or severity.29 Another reason could be that we were not able to control all covariates in subgroup network analyses.30 The results should be further validated in primary studies considering more possible covariates. Additionally, we noticed an obvious symptom clustering phenomenon of “lymph-stasis symptom” and “nerve injury symptom” in ALND network compared to SLNB network. As reported, ALND is an independent risk factor of developing LE, comparing to SLNB (Pooled OR = 3.098, 95% CI, 2.482-3.866).10 In this study, a higher proportion of participants receiving ALND (34.96%) developed LE than those undergone SLNB (13.64%). This could explain the network structure change toward lymph-stasis and nerve injury-related symptoms. However, the rate of ALND (76.3%) in our study was high than the reported rate in modern cohort, which could be explained by the selection bias of purposive sampling and convenience sampling of the primary studies, the high proportion of participants receiving breast cancer surgery over 10 years ago (almost half of the cases), and the high proportion of late-stage patients due to poor breast cancer screening rate. As SLNB was first officially recommended by the Chinese Anti-Cancer Association in 2011,31 its adoption in clinical practice took some time especially for the first few years and for nondeveloped regions. Despite of insufficient representativeness of our sample, the arm symptom network patterns and relationships in this study still provide valuable insights for symptom management in modern patients. Special attention should be given to the assessment, prevention, and management of arm symptoms in patients with ALND.
Radiotherapy was associated with severe arm symptoms, consistent with previous reports.32 Nevertheless, no significant differences were detected in network differences between groups (radiation vs nonradiation). RNI has been identified as a contributing factor for lymphedema.10 However, this study could not offer additional insights into the relationship between RNI and arm symptoms due to the limited data available. Future research should investigate the association of RNI with arm symptom networks. In addition, post-surgery duration was associated with the severity of arm symptoms, consistent with the findings of Nielsen et al.4 This result could be explained by the fact that BCRL is an uncured and progressive complication that most commonly occurs between 3 and 5 years post-surgery. Without timely treatment, the severity of symptoms increases and the number of lymphedema cases accumulates.
Post-surgery years showed a weak but positive correlation with arm-swelling in the symptom network (r = 0.18). Despite the weak correlation, this finding underscores the importance of long-term follow-up, continuous monitoring, and support for patients with breast cancer following surgery to prevent and manage arm-swelling and other potential arm symptoms. Clinicians should be aware that arm-swelling is an early sign and primary symptom of BCRL, which requires early detection and management. However, the weak correlation constrained the confidence in the findings, underscoring the need for future validation studies, particularly those with longitudinal designs.
Our results emphasize the importance for clinicians working with BCS to prioritize the long-term follow-up and continuous assessment and management of arm symptoms. It highlights the need for attention to arm symptoms even in survivors who may not fully meet the diagnostic criteria for LE.6,33 Special attention should be paid to survivors receiving ALND, RT, and a longer post-surgery duration who are more likely to suffer from arm symptoms. Based on the symptom clusters identified, a rapid assessment can be conducted to determine the primary symptom concerns in patients, facilitating targeted care. Additionally, the core symptoms should be prioritized when developing symptom management strategies, to increase the efficacy of interventions. Patient education on self-monitoring and reporting any changes in arm symptoms might help with early detection and management.
Limitations
The findings of this study should be considered in light of several limitations. First, due to limitations in the original dataset, we could not convert the interlimb circumference differences into limb volume. Currently, using interarm differences as a diagnostic criterion for lymphedema is not recommended and could potentially affect the accuracy of lymphedema diagnosis, leading to a high probability of false positives. Second, the secondary data analysis design constrained our capacity to thoroughly investigate additional potential covariates related to arm symptoms in BCS, such as the fields of radiation. Third, the utilization of cross-sectional data from the parent studies restricted our ability to observe the sequence of symptoms occurrence and discern causal relationships. Fourth, the exclusion of 7 symptoms due to their low incidence and clinical considerations may introduce bias in estimating the overall prevalence of arm symptoms. Fifth, the representativeness of our sample was limited by a high rate of ALND compared to modern breast cancer patient cohorts. Hence, the prevalence of arm symptoms should be interpreted with caution. Last, the weak correlation between post-surgery years and arm-swelling limited the confidence for clinical implications. These limitations underscore the need for further multicenter research, particularly through longitudinal studies, to validate and build upon our findings.
Conclusion
In conclusion, our study revealed a high prevalence of arm symptoms in BCS, both with and without LE. Three distinct symptom clusters, namely “Lymph Stasis,” “Nerve Injury,” and “Movement Limitation,” were detected. Tightness emerged as a core symptom in LE networks, while firmness was central in non-LE networks. Survivors who underwent ALND, received radiotherapy, and had longer post-surgery durations reported more severe arm symptoms. However, no significant differences in global strength and network structure were detected between various subgroups. Survivors with LE reported more prevalent and severe arm symptoms, and showed stronger connections between symptoms in the network compared to the non-LE network. The findings of this study provide insights into arm symptoms interrelationships of BCS, implying that clinicians and researchers to prioritize the assessment of arm symptoms in this population. Further longitudinal study is warranted to validate our findings and to understand dynamic arm symptom relationships over time.
Supplementary Material
Supplementary material is available at The Oncologist online.
Acknowledgments
We extend our gratitude to the researchers involved in the primary studies and the participants who contributed. Special thanks to the support provided by the China Scholarship Council and the Center for Global Initiatives at Johns Hopkins University School of Nursing.
Contributor Information
Aomei Shen, Tianjin Medical University Cancer Institute & Hospital, National Clinical Research Center for Cancer, Tianjin’s Clinical Research Center for Cancer, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Medical University Ministry of Education, Tianjin, 300060, People’s Republic of China; Peking University School of Nursing, Beijing, 100191, People’s Republic of China.
Zhongning Zhang, Tianjin Medical University Cancer Institute & Hospital, National Clinical Research Center for Cancer, Tianjin’s Clinical Research Center for Cancer, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Medical University Ministry of Education, Tianjin, 300060, People’s Republic of China; Tianjin Medical University School of Nursing, Tianjin, 300070, People’s Republic of China.
Jingming Ye, Department of Thyroid and Breast Surgery, Peking University First Hospital, Beijing, 100034, People’s Republic of China.
Yue Wang, Department of Thyroid and Breast Surgery, Peking University First Hospital, Beijing, 100034, People’s Republic of China.
Hongmeng Zhao, Tianjin Medical University Cancer Institute & Hospital, National Clinical Research Center for Cancer, Tianjin’s Clinical Research Center for Cancer, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Medical University Ministry of Education, Tianjin, 300060, People’s Republic of China.
Xin Li, Tianjin Medical University Cancer Institute & Hospital, National Clinical Research Center for Cancer, Tianjin’s Clinical Research Center for Cancer, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Medical University Ministry of Education, Tianjin, 300060, People’s Republic of China.
Peipei Wu, Tianjin Medical University Cancer Institute & Hospital, National Clinical Research Center for Cancer, Tianjin’s Clinical Research Center for Cancer, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Medical University Ministry of Education, Tianjin, 300060, People’s Republic of China.
Wanmin Qiang, Tianjin Medical University Cancer Institute & Hospital, National Clinical Research Center for Cancer, Tianjin’s Clinical Research Center for Cancer, Key Laboratory of Breast Cancer Prevention and Therapy, Tianjin Medical University Ministry of Education, Tianjin, 300060, People’s Republic of China.
Qian Lu, Peking University School of Nursing, Beijing, 100191, People’s Republic of China.
Author Contributions
Conception/design: A.M.S., Q.L.; Provision of study material or patients: J.M.Y., Y.W., H.M.Z., P.P.W.; Collection and assembly of data: J.M.Y., Y.W., H.M.Z., P.P.W., Z.N.Z., X.L., Y.W., A.M.S.; Data analysis and interpretation: A.M.S., Q.L.; Manuscript writing: A.M.S., Z.N.Z.,
Final approval of manuscript: All authors.
Funding
This study was supported by grants from the National Natural Science Foundation of China (72174011), Nursing Innovation Talent Fund of Tianjin Medical University Cancer Institute and Hospital (HL2021-27), Tianjin Key Medical Discipline (Specialty) Construction Project (TJYXZDXK 011A). The authors declare that no other funds, grants, or other support were received during the preparation of this manuscript. None of these funding impacted the study design, process, data analysis, and paper writing.
Conflict of Interest
None declared.
Data Availability
The data underlying this article will be shared on reasonable request to the corresponding author.
References
- 1. Giaquinto AN, Sung H, Miller KD, et al. Breast cancer statistics, 2022. CA Cancer J Clin. 2022;72(6):524-541. https://doi.org/ 10.3322/caac.21754 [DOI] [PubMed] [Google Scholar]
- 2. Batenburg MCT, Mink van der Molen DR, van der Leij F, et al. ; UMBRELLA study team. Patient-reported symptoms of late toxicity in patients with breast cancer treated with hypofractionated radiation therapy and the association with quality of life. Int J Radiat Oncol Biol Phys. 2023;115(5):1181-1191. https://doi.org/ 10.1016/j.ijrobp.2022.11.008 [DOI] [PubMed] [Google Scholar]
- 3. Hayes SC, Dunn M, Plinsinga ML, et al. Do patient-reported upper-body symptoms predict breast cancer-related lymphoedema: results from a population-based, longitudinal breast cancer cohort study. Cancers (Basel) 2022;14(23):5998. https://doi.org/ 10.3390/cancers14235998 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Nielsen AWM, Lundorff M, Nielsen HM, et al. Symptom trajectories in breast cancer survivors: growth mixture analysis of patient-reported pain, fatigue, insomnia, breast and arm symptoms. Acta Oncol. 2021;60(12):1659-1667. https://doi.org/ 10.1080/0284186X.2021.1981550 [DOI] [PubMed] [Google Scholar]
- 5. Hormes JM, Bryan C, Lytle LA, et al. Impact of lymphedema and arm symptoms on quality of life in breast cancer survivors. Lymphology. 2010;43(1):1-13. [PubMed] [Google Scholar]
- 6. Laws A, Lagendijk M, Grossmith S, et al. Long-term patient-reported arm symptoms in breast cancer survivors. Ann Surg Oncol. 2024;31(3):1623-1633. https://doi.org/ 10.1245/s10434-023-14711-w [DOI] [PubMed] [Google Scholar]
- 7. Zomkowski K, Cruz de Souza B, Pinheiro da Silva F, et al. Physical symptoms and working performance in female breast cancer survivors: a systematic review. Disabil Rehabil. 2018;40(13):1485-1493. https://doi.org/ 10.1080/09638288.2017.1300950 [DOI] [PubMed] [Google Scholar]
- 8. Fu MR. Breast cancer-related lymphedema: symptoms, diagnosis, risk reduction, and management. World J Clin Oncol. 2014;5(3):241-247. https://doi.org/ 10.5306/wjco.v5.i3.241 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Donahue PMC, MacKenzie A, Filipovic A, Koelmeyer L.. Advances in the prevention and treatment of breast cancer-related lymphedema. Breast Cancer Res Treat. 2023;200(1):1-14. https://doi.org/ 10.1007/s10549-023-06947-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Shen A, Lu Q, Fu X, et al. Risk factors of unilateral breast cancer-related lymphedema: an updated systematic review and meta-analysis of 84 cohort studies. Support Care Cancer. 2022;31(1):18. https://doi.org/ 10.1007/s00520-022-07508-2 [DOI] [PubMed] [Google Scholar]
- 11. De Groef A, Devoogdt N, Gursen C, et al. Sensory signs and symptoms in women with self-reported breast cancer-related lymphedema: a case-control study close up. J Cancer Surviv. 2022;16(5):1096-1106. https://doi.org/ 10.1007/s11764-021-01100-z [DOI] [PubMed] [Google Scholar]
- 12. Armer JM, Radina ME, Porock D, Culbertson SD.. Predicting breast cancer-related lymphedema using self-reported symptoms. Nurs Res. 2003;52(6):370-379. https://doi.org/ 10.1097/00006199-200311000-00004 [DOI] [PubMed] [Google Scholar]
- 13. Zhu Z, Xing W, Hu Y, Wu B, So WKW.. Paradigm shift: moving from symptom clusters to symptom networks. Asia Pac J Oncol Nurs. 2022;9(1):5-6. https://doi.org/ 10.1016/j.apjon.2021.12.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Hevey D. Network analysis: a brief overview and tutorial. Health Psychol Behav Med. 2018;6(1):301-328. https://doi.org/ 10.1080/21642850.2018.1521283 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Li F, Lu Q, Jin S, et al. A scoring system for predicting the risk of breast cancer-related lymphedema. Int J Nurs Sci. 2020;7(1):21-28. https://doi.org/ 10.1016/j.ijnss.2019.12.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Shen A, Wu P, Qiang W, et al. Factors associated with lymphedema self-management behaviours among breast cancer survivors: a cross-sectional study. J Clin Nurs. 2023;32(19-20):7330-7345. https://doi.org/ 10.1111/jocn.16833 [DOI] [PubMed] [Google Scholar]
- 17. Shi S, Lu Q, Fu MR, et al. Psychometric properties of the breast cancer and lymphedema symptom experience index: the Chinese version. Eur J Oncol Nurs. 2016;20:10-16. https://doi.org/ 10.1016/j.ejon.2015.05.002 [DOI] [PubMed] [Google Scholar]
- 18. Shen A, Wei X, Zhu F, et al. Risk prediction models for breast cancer-related lymphedema: a systematic review and meta-analysis. Eur J Oncol Nurs. 2023;64:102326. https://doi.org/ 10.1016/j.ejon.2023.102326 [DOI] [PubMed] [Google Scholar]
- 19. Brunelle CL, Taghian AG.. Breast cancer–related lymphedema: the prospective surveillance model, early intervention strategies, and role of complete decongestive therapy. Curr Breast Cancer Rep. 2023;15(1):12-23. https://doi.org/ 10.1007/s12609-023-00471-9 [DOI] [Google Scholar]
- 20. Dylke E. Measurement of breast cancer-related lymphoedema. J Physiother. 2022;68(4):238-243. https://doi.org/ 10.1016/j.jphys.2022.09.008 [DOI] [PubMed] [Google Scholar]
- 21. Button KS, Ioannidis JP, Mokrysz C, et al. Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci. 2013;14(5):365-376. https://doi.org/ 10.1038/nrn3475 [DOI] [PubMed] [Google Scholar]
- 22. Watkins MW. Exploratory factor analysis: a guide to best practice. J Black Psychol. 2018;44(3):219-246. https://doi.org/ 10.1177/0095798418771807 [DOI] [Google Scholar]
- 23. Freichel R. Symptom network analysis tools for applied researchers with cross-sectional and panel data - a brief overview and multiverse analysis. Psychol Rep. 2023;332941231213649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Fu MR, Conley YP, Axelrod D, et al. Precision assessment of heterogeneity of lymphedema phenotype, genotypes and risk prediction. Breast. 2016;29:231-240. https://doi.org/ 10.1016/j.breast.2016.06.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Gursen C, Dylke ES, Moloney N, et al. Self‐reported signs and symptoms of secondary upper limb lymphoedema related to breast cancer treatment: systematic review. Eur J Cancer Care (Engl). 2021;30(5):1-22. [DOI] [PubMed] [Google Scholar]
- 26. Svensson BJ, Dylke ES, Ward LC, Black DA, Kilbreath SL.. Screening for breast cancer-related lymphoedema: self-assessment of symptoms and signs. Support Care Cancer. 2020;28(7):3073-3080. https://doi.org/ 10.1007/s00520-019-05083-7 [DOI] [PubMed] [Google Scholar]
- 27. Wei X, Lu Q, Jin S, et al. Developing and validating a prediction model for lymphedema detection in breast cancer survivors. Eur J Oncol Nurs. 2021;54:102023. https://doi.org/ 10.1016/j.ejon.2021.102023 [DOI] [PubMed] [Google Scholar]
- 28. Ebaugh D, Spinelli B, Schmitz KH.. Shoulder impairments and their association with symptomatic rotator cuff disease in breast cancer survivors. Med Hypotheses. 2011;77(4):481-487. https://doi.org/ 10.1016/j.mehy.2011.06.015 [DOI] [PubMed] [Google Scholar]
- 29. van Borkulo CD, van Bork R, Boschloo L, et al. Comparing network structures on three aspects: a permutation test. Psychol Methods. 2023;28(6):1273-1285. https://doi.org/ 10.1037/met0000476 [DOI] [PubMed] [Google Scholar]
- 30. Jing F, Zhu Z, Qiu J, et al. Contemporaneous symptom networks and correlates during endocrine therapy among breast cancer patients: a network analysis. Front Oncol. 2023;13:1081786. https://doi.org/ 10.3389/fonc.2023.1081786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Chinese Anti-Cancer Association Breast Cancer Committee. Guidelines and norms for breast cancer diagnosis and treatment by the Chinese anti-cancer association. China Oncol. 2011;21(5):367-417. [Google Scholar]
- 32. Togawa K, Ma H, Smith AW, et al. Self-reported symptoms of arm lymphedema and health-related quality of life among female breast cancer survivors. Sci Rep. 2021;11(1):10701. https://doi.org/ 10.1038/s41598-021-89055-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Wennman-Larsen A, Petersson LM, Saboonchi F, Alexanderson K, Vaez M.. Consistency of breast and arm symptoms during the first two years after breast cancer surgery. Oncol Nurs Forum. 2015;42(2):145-155. https://doi.org/ 10.1188/15.ONF.145-155 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article will be shared on reasonable request to the corresponding author.