Abstract
Purpose
Supportive care resources are limited, requiring a strategic, tailored approach, especially with the growing population of breast cancer patients. We examined whether routinely collected patient-reported outcome measures (PROMs) can identify meaningful health-related quality of life (HRQoL) trajectories and help guide resource allocation of supportive care.
Methods
As part of routine care at the Netherlands Cancer Institute, 2181 early-stage breast cancer patients completed the EORTC QLQ-C30 and BR23 before treatment and six months later. Through Latent Class and Latent Transition Analyses, we identified HRQoL subgroups and transitions between subgroups over time. Multinomial regression examined sociodemographic and clinical correlates. Network analysis detected key domains and inter-domain connections within subgroups.
Results
Three HRQoL subgroups were identified at baseline: Excellent HRQoL (53%), Good HRQoL with Psychosocial concerns (33%), and Poor HRQoL with severe functional limitations (13%). Multiple comorbidities and history of depression were independently and strongly associated with membership to less favorable subgroups. Subgroup transitions were rare (1–5%). Network analysis showed subgroup-specific key domains: emotional functioning in the Psychosocial concerns-subgroup, and menopausal symptoms, social, and role functioning in the Poor HRQoL-subgroup; the latter subgroup showing the highest interconnectivity between HRQoL-domains.
Conclusion
Baseline HRQoL and sociodemographic factors, independent of treatment, is associated with HRQoL trajectories and can guide efficient re-allocation of resources to those patients with complex needs. Most patients maintain excellent HRQoL and may benefit from low-intensity or self-management support, whereas targeted, multidisciplinary interventions should be prioritized for those with complex needs. Different key domains and connectivity suggest different types of supportive care are needed for different subgroups.
Supplementary Information
The online version contains supplementary material available at 10.1186/s13058-026-02228-5.
Keywords: Breast cancer, Health-related quality of life, Supportive care, Latent class analysis, Latent transition analysis, Network analysis, Value-based health care
Introduction
The assessment and management of health-related quality of life (HRQoL) is an essential component of breast cancer care [1]. HRQoL, and consequently the need for supportive care, varies widely among breast cancer patients [2]. While most patients maintain a relatively good HRQoL throughout treatment [3, 4], a subset experiences persistently low HRQoL, including impairments in physical, role, and social functioning [5, 6]. Importantly, the variation in HRQoL is already apparent at the time of diagnosis. Several studies have shown that lower baseline HRQoL is associated with experiencing more toxicities [7], early termination of chemotherapy and endocrine therapy [7, 8], poorer surgical outcomes [9], and lower survival rates [10, 11].
Timely provision of supportive care interventions, such as psychosocial support and guided exercise, can mitigate those adverse outcomes [12–16]. Meanwhile, the rising number of breast cancer survivors [17], and persistent healthcare workforce shortages [18], underscore the need to restructure supportive care to safeguard access for those in need. It is therefore equally important to identify patients likely to sustain adequate HRQoL and navigate their treatment journey with minimal intervention, such as self-management tools [19, 20]. Current discourse in healthcare highlights a shift from equality (offering identical care to all), to emphasizing equity (tailoring support to individual needs) [21, 22]. This approach prioritizes intensive care for patients with low HRQoL while offering minimal intervention to those maintaining good HRQoL. Reflecting this, the American Society of Clinical Oncology (ASCO) recommends supportive care tailored to individual needs [23, 24]. Yet, structured strategies for implementing tailored care remain limited in practice [2, 25–27], partly because clinicians struggle to determine which patients require specific types of support [14, 28, 29].
Implementing Patient-Reported Outcome Measures (PROMs) into routine practice could structure tailored supportive care for patients with diverse HRQoL trajectories. PROMs have shown to effectively identify HRQoL issues [30], yet are often reported as aggregated scores, and fail to capture interpatient heterogeneity and the interplay of multiple co-occurring issues. This limits their clinical use in reorganizing supportive care pathways. Advanced statistical approaches can potentially offer better clinically meaningful insights. Classifying patients into subgroups with similar HRQoL trajectories could optimize resource allocation for those with complex needs [31]. Data-driven methods such as latent class analysis (LCA) and latent transition analysis (LTA) can identify such unobserved subgroups with shared HRQoL trajectories [32]. Prior studies used LCA and LTA to identify breast cancer patient subgroups characterized by low, intermediate, and high overall HRQoL and with specific patterns in sleep, cognitive, physical, and emotional functioning from 5 months after the diagnosis [32–37]. Recent insights from the French CANTO (CANcer TOxicities) cohort in breast cancer patients receiving chemotherapy, showed those trajectories might already be present at baseline, before treatment has started [36]. Also, to further guide efficient targeted interventions network analysis may help to uncover central symptoms that drive others [38–41]. A recent study by Cai et al. [42] suggests that symptom relationships may vary across HRQoL-subgroups, and thus different interventions may be needed for each subgroup.
The present study builds on previous research by integrating latent class and transition analyses with network analysis of PROMs in real-world data from early-stage breast cancer patients at diagnosis and after six months. By doing so, our findings aim to aid the development of tailored referral pathways to supportive care, helping us re-organize supportive care for a growing group of breast cancer patients. We hypothesize that distinct HRQoL trajectory subgroups already emerge early in the care pathway for all early-stage breast cancer patients. Additionally, we hypothesize that there are unique central symptoms that drive other symptoms in each subgroup.
Methods
Population
At the Netherlands Cancer Institute-Antoni van Leeuwenhoek hospital, all new patients with early stage breast cancer are invited to completed PROMs as part of routine care since 2022 [43]. Patients are invited by email and asked to complete the PROMS through the patient portal of the electronic health record (Hix 6.3, Chipsoft). Eligibility criteria are: access to the digital patient portal and sufficient proficiency in Dutch (as our institute only offers PROMs in Dutch). All patients provide written informed consent for the use of their data for research purposes [44]; the use of their data for this study was approved by Internal Review Board (IRBd21-262). For the present study, we included patients who completed their baseline PROMs (after diagnosis and before the start of therapy) between February 2022 and September 2024 (70% of the eligible patients).
Sociodemographic and clinical characteristics
Sociodemographic (i.e. age, sex, postal code, menopausal status, BMI, alcohol use and smoking status) and treatment-related data were extracted from the electronic health record; data on treatment plans was used for baseline analysis. Comorbidities were assessed using the self-administered comorbidity questionnaire (SCQ) [45] at baseline. Postal codes were linked to Statistics Netherlands household income data as a proxy for socioeconomic status (SES)[46].
PROMs
Participants completed the EORTC QLQ-C30 [47] and the breast cancer-specific QLQ-BR23 module [48] at baseline (T0; after diagnosis, but before initiation of treatment) and six months thereafter (T1). The QLQ-C30 consists of 14 domains (physical-, role-, emotional-, cognitive-, and social functioning, fatigue, nausea/vomiting, pain, dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties). The QLQ-BR23 comprises nine domains: systemic therapy side effects, arm symptoms, breast symptoms, body image, future perspective, upset by hair loss, sexual functioning, and sexual enjoyment. As baseline questionnaires were conducted prior to treatment initiation, the systemic therapy side effects domain (hot flushes, dry mouth, irritated eyes, headaches, feeling unwell) was interpreted as reflecting hormonal symptoms, given the overlap with menopausal symptomatology. All items are scored on a 1–4 Likert scale and linearly transformed to 0–100. For interpretability, scores were recoded so that higher scores reflect better HRQoL or lower symptom burden.
Statistical analysis
Patients were included if they completed PROMs at both T0 and T1. Descriptive statistics summarized sociodemographic and clinical characteristics. Analyses were performed in R (v4.1.2) [48] and LatentGold (v5.2) [49]. Missing sociodemographic and clinical data were addressed using multilevel multiple imputation with chained equations (MICE) (MICE, v3.15.0) [50], accounting for correlation between repeated measures. Network analysis employed the estimateNetwork function from the bootnet-package (v1.4.3) [51], visualized with qgraph (v1.9.8) [51]. Networks of different subgroups were compared using the NetworkComparisonTest-package (v2.2.2) [38].
Latent class and transition analysis
We applied LCA to identify subgroups with distinct HRQoL patterns in QLQ-C30 and QLQ-BR23 scores, entered as ordinal variables. Model selection used fit indices (BIC, AIC, VLMR, BLRT, entropy) and expert clinical judgement on interpretability, following the method described by Sinha et al. [52] and were reported accordingly. QLQ-C30 scores were compared with Giesinger et al.’s thresholds for clinical relevance [53], indicating scores warranting clinical attention. As BR23 thresholds are still under development [54]; differences were interpreted descriptively. Subgroups were labelled to reflect distinctive HRQoL profiles. To assess change over 6 months (T1), subgroup transition probabilities were estimated via latent transition analysis (LTA) [55]. Associations between sociodemographic/clinical characteristics and subgroup membership or transitions were examined using multinomial logistic regression, using the bias-adjusted three-step approach [56].
Network analysis
Within each subgroup, network analysis followed Epskamp and Borsboom [57] and reporting standards of Burger et al. [58], PROM domains were represented as nodes in a Fruchterman–Reingold layout [59], with edges denoting partial correlations adjusted for all other nodes; edge thickness indicated association strength. Global connectivity, defined as the sum of absolute edge weights, quantified overall symptom interrelatedness. Standardized centrality indices (z-scores) assessed domain importance: strength (total connectivity), closeness (proximity to other nodes), and betweenness (role in connecting nodes). Differences in network structure, edge strength, and global connectivity between subgroups and over time were tested using the Network Comparison Test (NCT) [38], which applies permutation-based testing.
Results
A total of 2181 patients (mean age = 58.3 years, SD = 10.8) completed the EORTC QLQ-C30 and BR23 module at both T0 and T1 (Table 1). Data were complete on age and EORTC QLQ-C30 and BR23 at both time points. Missing data occurred across covariates (level of education and comorbidities (n = 829, 38.0%), menopausal status (n = 546, 25.0%), WHO performance status (n = 511, 23.4%), smoking and alcohol use (n = 459, 21.0%), tumor and treatment characteristics (n = 197, 9.0%), body mass index (n = 55, 2.5%), and socioeconomic status (n = 22, 1.0%). Under the assumption that data were missing at random, we employed multilevel multiple imputation by chained equations to generate 5 imputed datasets. All subsequent analysis were conducted on the pooled imputed datasets.
Table 1.
Sociodemographic and clinical variables at baseline by subgroup
| Characteristics | Overall | Excellent HRQOL | Good HRQoL with psychosocial concerns | Poor HRQoL with severe limitations in functioning |
|---|---|---|---|---|
| N = 2,1811 | N = 1,1731 | N = 7231 | N = 2851 | |
| Sociodemographic | ||||
| Age Median ((Q1, Q3)) | 58 (49, 67) | 59 (52, 68) | 56 (47, 67) | 54 (46, 66) |
| WHO status (n, %) | ||||
| 0 | 1920 (88%) | 1030 (88%) | 638 (88%) | 252 (88%) |
| 1 | 215 (9.9%) | 120 (10%) | 71 (9.8%) | 24 (8.4%) |
| 2 | 29 (1.3%) | 14 (1.2%) | 8 (1.1%) | 7 (2.5%) |
| 4 | 17 (0.8%) | 9 (0.8%) | 6 (0.8%) | 2 (0.7%) |
| Body weight (n, %) | ||||
| Underweight (BMI < 18.5) | 32 (1.5%) | 20 (1.7%) | 8 (1.1%) | 4 (1.4%) |
| Normal weight (BMI > 18.5 < 25) | 1061 (49%) | 569 (49%) | 356 (50%) | 136 (48%) |
| Overweight (BMI > 25 < 29.9) | 633 (29%) | 340 (29%) | 211 (29%) | 82 (29%) |
| Obese (BMI > 30) | 441 (20%) | 237 (20%) | 142 (20%) | 62 (22%) |
| Marital status | ||||
| Married or living together | 1572 (72%) | 858 (73%) | 510 (71%) | 204 (72%) |
| In a relationship and living apart | 148 (6.8%) | 77 (6.6%) | 53 (7.3%) | 18 (6.3%) |
| Divorced | 234 (11%) | 109 (9.3%) | 91 (13%) | 34 (12%) |
| Widowed | 105 (4.8%) | 61 (5.2%) | 34 (4.7%) | 10 (3.5%) |
| Never been married or lived together with a partner | 122 (5.6%) | 68 (5.8%) | 35 (4.8%) | 19 (6.7%) |
| Highest completed level of education | ||||
| Only primary school | 98 (4.5%) | 49 (4.2%) | 29 (4.0%) | 20 (7.0%) |
| Secondary education | 207 (9.5%) | 104 (8.9%) | 79 (11%) | 24 (8.4%) |
| Medium vocational training | 607 (28%) | 313 (27%) | 207 (29%) | 87 (31%) |
| High vocational training | 1269 (58%) | 707 (60%) | 408 (56%) | 154 (54%) |
| Social economic status (n, %) | ||||
| Low | 78 (3.6%) | 48 (4.1%) | 19 (2.6%) | 11 (3.9%) |
| Below average | 298 (14%) | 165 (14%) | 98 (14%) | 35 (12%) |
| Above average | 21 (1.0%) | 13 (1.1%) | 6 (0.8%) | 2 (0.7%) |
| High | 1784 (82%) | 947 (81%) | 600 (83%) | 237 (83%) |
| Comorbidities | ||||
| Cardiac history | 136 (6.2%) | 69 (5.9%) | 44 (6.1%) | 23 (8.1%) |
| Stroke | 12 (0.6%) | 4 (0.3%) | 3 (0.4%) | 5 (1.8%) |
| Hypertension | 337 (15%) | 174 (15%) | 114 (16%) | 49 (17%) |
| Respiratory diseases | 239 (11%) | 114 (9.7%) | 81 (11%) | 44 (15%) |
| Diabetes | 94 (4.3%) | 43 (3.7%) | 38 (5.3%) | 13 (4.6%) |
| Stomach ulcer | 10 (0.5%) | 3 (0.3%) | 2 (0.3%) | 5 (1.8%) |
| Kidney diseases | 41 (1.9%) | 18 (1.5%) | 11 (1.5%) | 12 (4.2%) |
| Liver diseases | 38 (1.7%) | 19 (1.6%) | 13 (1.8%) | 6 (2.1%) |
| Blood diseases | 127 (5.8%) | 53 (4.5%) | 44 (6.1%) | 30 (11%) |
| Thyroid diseases | 162 (7.4%) | 80 (6.8%) | 54 (7.5%) | 28 (9.8%) |
| Depression | 136 (6.2%) | 39 (3.3%) | 63 (8.7%) | 34 (12%) |
| Arthritis | 420 (19%) | 235 (20%) | 131 (18%) | 54 (19%) |
| Back pain | 624 (29%) | 265 (23%) | 256 (35%) | 103 (36%) |
| Rheumatic arthritis | 103 (4.7%) | 40 (3.4%) | 44 (6.1%) | 19 (6.7%) |
| Amount of comorbidities (Median (Q1, Q3)) | 6.0 (5.0, 7.0) | 3.0 (5.0, 7.0) | 4.0 (5.0, 7.0) | 7.0 (5.0, 8.0) |
| Tumor & treatment characteristics | ||||
| Hormone receptor status | ||||
| Negative | 146 (6.7%) | 70 (6.0%) | 49 (6.8%) | 27 (9.5%) |
| Mixed | 1047 (48%) | 608 (52%) | 332 (46%) | 107 (38%) |
| Positive | 988 (45%) | 495 (42%) | 342 (47%) | 151 (53%) |
| Stage | ||||
| 1 | 1037 (48%) | 599 (51%) | 317 (44%) | 121 (42%) |
| 2 | 1076 (49%) | 542 (46%) | 380 (53%) | 154 (54%) |
| 3 | 68 (3.1%) | 32 (2.7%) | 26 (3.6%) | 10 (3.5%) |
| Surgery | ||||
| Mastectomy | 612 (33%) | 297 (30%) | 219 (37%) | 96 (41%) |
| Lumpectomy | 1078 (59%) | 630 (63%) | 326 (55%) | 122 (52%) |
| Axillary dissection | 110 (6.0%) | 55 (5.5%) | 41 (6.9%) | 14 (6.0%) |
| Chemotherapy neo-adjuvant | 1838 (84%) | 1005 (86%) | 598 (82%) | 235 (100%) |
| Chemotherapy adjuvant | 326 (15%) | 148 (13%) | 122 (17%) | 56 (20%) |
| Immunotherapy neo-adjuvant | 182 (8%) | 93 (8%) | 62 (8%) | 17 (8%) |
| Endocrine therapy | 908 (42%) | 495 (42%) | 307 (42%) | 106 (37%) |
| Radiotherapy adjuvant | 1329 (61%) | 721 (61%) | 444 (61%) | 164 (58%) |
| Other | ||||
| Smoking status (n, %) | ||||
| Never smoked | 1218 (56%) | 670 (57%) | 395 (55%) | 153 (54%) |
| Current smoker | 244 (11%) | 137 (12%) | 74 (10%) | 33 (12%) |
| Stopped smoking | 719 (33%) | 366 (31%) | 254 (35%) | 99 (35%) |
| Alcohol use (n, %) | ||||
| Never consumed alcohol | 760 (35%) | 410 (35%) | 247 (34%) | 103 (36%) |
| Currently consuming alcohol | 1342 (62%) | 720 (61%) | 452 (63%) | 170 (60%) |
| Stopped consuming alcohol | 79 (3.6%) | 43 (3.7%) | 24 (3.3%) | 12 (4.2%) |
| Menopausal status (n, %) | ||||
| Pre-menopausal | 645 (30%) | 285 (24%) | 236 (33%) | 124 (44%) |
| Peri-menopausal | 210 (9.6%) | 114 (9.7%) | 73 (10%) | 23 (8.1%) |
| Post-menopausal | 1,326 (61%) | 774 (66%) | 414 (57%) | 138 (48%) |
1n (%)
Description of groups at baseline
Based on fit statistics, subgroup sizes (see Data Supplement), and clinical interpretability, a three-subgroup solution was selected. We labelled the largest subgroup Excellent HRQoL (n = 1173; 53%) as scores in all domains exceeded the thresholds for clinical importance, with scores approximately 5–10 points above Dutch normative values. The second subgroup was labelled Good HRQoL with psychosocial concerns, hereafter Psychosocial concerns (n = 723; 33%), with scores above thresholds for most domains, except emotional functioning (mean 63.54, SD 19.46) and insomnia (mean 58.92, SD 27.38). Pain (mean 77.34, SD 18.15) and fatigue (mean 64.08, SD 15.95) scores were relatively low but remained above thresholds. Future perspective scored notably low (mean 45.55, SD 27.51). We labelled the smallest subgroup Poor HRQoL with severe limitations in functioning domains, hereafter Poor HRQoL (n = 285; 13%), as almost all domain scores were below thresholds of clinical importance, particularly in physical, role, emotional, and social functioning. Unlike other subgroups, financial impact (mean 78.60, SD 29.04) was also below threshold.
Sexual functioning and sexual enjoyment scores were low across all subgroups, with means of 23.57 (SD 21.27) and 46.21 (SD 28.45) in the Excellent HRQoL-subgroup, 18.42 (SD 19.18) and 39.51 (SD 28.70) in the Psychosocial concerns-subgroup, and 15.09 (SD 19.03) and 33.33 (SD 26.39) in the Poor HRQoL-subgroup, respectively. Full descriptive statistics are presented in Fig. 1 and the Data Supplement.
Fig. 1.
Radar plots of the three HRQoL-subgroups. Green, excellent HRQOL; Yellow, good HRQoL with psychosocial concerns; Red, poor HRQoL with severe limitations in functioning domains. Grey areas, scores below thresholds of clinical importance for all EORTC-QLQ-C30 items (QL, PF, RF, EF, CF, SF, FA, NV, PA, DY, SL, AP, CO, DI, FI.) For the EORTC QLQ-BR23 (ST, HL, AS, BS, BI, FU, SEF, SEE) items thresholds were not yet established, therefore not represented in the figure. Interpretation: Scale scores range from 0 to 100, with higher scores indicating a better HRQOL or lower symptom burden. QL, quality of life; PF, physical functioning; RF, role functioning; EF, emotional functioning; CF, cognitive functioning; SF, social functioning; FA, fatigue; NV, nausea/vomiting; PA, pain; DY, dyspnoea; SL, sleep disturbances/insomnia; AP, appetite loss; CO, constipation; DI, diarrhoea; FI, financial impact; ST, systemic therapy side effects; HL, hair loss; AS, arm symptoms; BS, breast symptoms; BI, body image; FU, future perspective; SEF, sexual functioning; SEE, sexual enjoyment
Factors associated with membership of subgroup
Multinomial logistic regression, with the Excellent HRQoL-subgroup as reference category, identified several factors independently associated with membership in less favorable HRQoL-subgroups. For the Psychosocial concerns-subgroup, having ≥ 3 comorbidities (OR 3.80, 95% CI 2.85–6.85), a history of depression (OR 2.58, 95% CI 0.37–2.91), scheduled chemotherapy (neo-adjuvant OR 1.69, 95% CI 1.56–1.98; adjuvant OR 1.72, 95% CI 1.54–1.96), and lower education level (OR 1.17, 95% CI 1.05–1.30) increased odds of group membership. For the Poor HRQoL-subgroup, significant factors included ≥ 3 comorbidities (OR 5.61, 95% CI 3.10–8.23), single marital status (OR 3.98, 95% CI 2.81–5.87), obesity (OR 2.68, 95% CI 2.66–2.99), peri-menopausal status (OR 2.77, 95% CI 2.65–2.91), history of depression (OR 2.59, 95% CI 1.09–3.32), lower socioeconomic status (OR 1.98, 95% CI 1.03–9.87), current smoking (OR 1.11, 95% CI 1.00–1.23), and PR-negative status (OR 1.47, 95% CI 1.09–1.99). Age and other tumor characteristics were not associated with subgroup membership. See Fig. 2 and Data Supplement for detailed results.
Fig. 2.
Forest plot displaying odds ratios (ORs) with 95% confidence intervals (CIs) for covariates associated with latent subgroup membership. The x-axis represents the ORs on a logarithmic scale, and the y-axis lists all covariates included in the multinomial logistic regression model. Two subgroups are shown: the Good HRQoL with Psychosocial Concerns subgroup (yellow, left panel) and the Poor HRQoL with Severe Functional Limitations subgroup (red, right panel). Error bars indicate 95% CIs. Statistically significant associations (p < 0.05) are marked with an asterisk (*)
Transitions between subgroups from baseline (T0) to 6 months (T1)
Transition probabilities between subgroups were low (1–5%), indicating that most patients remained stable from baseline to 6 months, See Fig. 3. Several factors were independently associated with higher odds of transitioning to less favorable subgroups. Patients with a history of depression (OR = 4.95, 95 CI% 2.52–9.72), a history of back pain (OR = 2.00, 95 CI% 1.31–3.06) or lower education (OR 0.77, 95 CI% 0.62–0.88) are more likely to transition from Excellent HRQoL to Psychosocial concerns-group. Patients with a history of depression (OR = 4.62, 95% CI 2.03–10.51), peri-menopausal status (OR = 0.64, 95 CI% 0.46), lower education (OR = 0.77, 95 CI% 0.62–0.88), and negative ER-status (OR 2.82 95 CI% 1.61–4.95) are more likely to transition from the Psychosocial concerns-subgroup to the Low HRQoL-subgroup. Data Supplement for a complete overview.
Fig. 3.
a Sankey diagram illustrating transitions in HRQoL-subgroup membership between baseline (T0) and six months post-diagnosis (T1). The width of each flow is proportional to the number of participants transitioning between subgroups. Group 1 (green) represents Excellent HRQoL, Group 2 (yellow) indicates Good HRQoL with Psychosocial Concerns, and Group 3 (red) corresponds to Poor HRQoL with Severe Functional Limitations. b Transition matrix showing the number of patients classified in each HRQoL subgroup at T1, stratified by their subgroup membership at T0, representing transition probabilities between time points.Network analysis
Connectivity and network structure
The network structures of the Psychosocial Concerns- and Poor HRQoL-subgroups were compared to the Excellent HRQoL-subgroup. The Psychosocial Concerns-subgroup showed moderately higher domain connectivity (average edge strength: 2.76 vs. 1.17; T = 0.66, p = 0.016), with no significant difference in overall network structure (T = 0.21, p = 0.09) or connectivity over time (T = 4.83, p = 0.67). Conversely, the Poor HRQoL-subgroup exhibited markedly greater connectivity (5.15 vs. 1.17; T = 4.03, p < 0.001) and a significantly different network structure (T = 0.28, p = 0.017), characterized by closer domain proximity and direct links between Social- and Role Functioning. Financial Impact was centrally integrated, connecting with Future Perspective, Emotional, Role, and Social Functioning. This subgroup also demonstrated a significant decline in connectivity over time (Δ = − 1.20; T = 2.84, p < 0.001) (Fig. 4).
Fig. 4.
Networks constructed via graphical LASSO visualizing the regularized partial correlations between items in different subgroups. Legend and interpretation: Green edges represent positive partial correlations and red edges negative ones. Thicker and more saturated edges represent stronger partial correlations. Edges with absolute weight above 0.05 are displayed. The distance between two nodes reflects the absolute edge weight between them (Fruchterman–Reingold layout). Highlighted symptoms represent the symptoms with the highest strength. QL, quality of life total; PF, physical functioning; RF, role functioning; EF, emotional functioning; CF, cognitive functioning; SF, social functioning; FA, fatigue; NV, nausea/vomiting; PA, pain; DY, dyspnoea; SL, sleep disturbances/insomnia; AP, appetite loss; CO, constipation; DI, diarrhoea; FI, financial impact; ST, systemic therapy side effects/hormonal symptoms; HL, hair loss; AS, arm symptoms; BS, breast symptoms; BI, body image; FU, future perspective; SEF, sexual functioning; SEE, sexual enjoyment
Centrality
Across the total sample, Fatigue (1.93), Systemic Treatment Side Effects (1.43), and Emotional Functioning (0.91) were most central. In the Excellent HRQoL-subgroup, central domains included Fatigue (1.71), Physical Functioning (1.53), Emotional Functioning (1.45), and Pain (1.56), with Sexual Functioning and Enjoyment (0.51 vs. − 0.81; T = 1.34, p < 0.001) and Future Perspective (0.73 vs. − 0.40; T = 0.93, p = 0.002) significantly more central than the total sample. In the Psychosocial Concerns-subgroup, Emotional Functioning (2.53 vs. 0.91; T = 0.80, p = 0.02), Pain (1.61 vs. 1.01; T = 1.21, p = 0.03), Physical Functioning (1.91 vs. 1.15; T = 1.52, p = 0.01), and Future Perspective (0.91 vs. − 0.40; T = 0.78, p < 0.001) showed higher centrality relative to the total sample. In the Poor HRQoL-subgroup, hormonal symptoms (2.61 vs. − 0.62; T = 0.78, p < 0.001), Role Functioning (1.32 vs. 0.42; T = 0.68, p = 0.003), and social functioning (1.31 vs. − 0.82; T = 1.03, p < 0.001) were most central, whereas domains central in other subgroups (sexual functioning and enjoyment, emotional functioning, pain, physical functioning, future perspective) were significantly less central here (Fig. 5). Full centrality metrics are provided in the Data Supplement.
Fig. 5.
Node strength of each HRQoL domain displayed as standardized Z-scores. Green, excellent HRQOL; Yellow, good HRQoL with psychosocial concerns; Red, poor HRQoL with severe limitations in functioning domains. QL, quality of life total; PF, physical functioning; RF, role functioning; EF, emotional functioning; CF, cognitive functioning; SF, social functioning; FA, fatigue; NV, nausea/vomiting; PA, pain; DY, dyspnoea; SL, sleep disturbances/insomnia; AP, appetite loss; CO, constipation; DI, diarrhoea; FI, financial impact; ST, systemic therapy side effects/hormonal symptoms; HL, hair loss; AS, arm symptoms; BS, breast symptoms; BI, body image; FU, future perspective; SEF, sexual functioning; SEE, sexual enjoyment
Discussion
In a large sample of real world HRQoL data of 2181 early stage breast cancer patients, we identified three distinct subgroups of breast cancer patients based on baseline HRQoL using latent class analyses. The largest subgroup (52%) reported excellent HRQoL; a second subgroup (34%) reported good HRQoL but clinically relevant psychosocial concerns, and a third smaller subgroup (13%) reported poor HRQoL with severe limitations in functioning domains. Transition probabilities between subgroups at six months were low (1–5%), indicating that most patients remained in their initial subgroup. Network analysis revealed distinct patterns and different key symptoms per subgroup.
Notably, the Excellent HRQoL-subgroup reported clinically meaningful higher QLQ-C30 scores than Dutch normative values for women aged 50–59 [60, 61]. Although general population references for the BR23 are unavailable, the Excellent HRQoL-subgroup consistently exceeded scores from a cohort of women with benign breast conditions at the Charité Breast Clinic, except for sexual functioning and enjoyment [62]. These findings align with prior HRQoL trajectory studies in early-stage breast cancer. For instance, the Carolina Breast Study identified a comparable “excellent HRQoL” subgroup (11%) maintaining superior scores in comparison to the general U.S. population at 5 and 25 months post-diagnosis [32]. Similarly, a “good HRQoL with psychosocial concerns” subgroup appeared in both our cohort and the Carolina Breast Study (28%) and French CANTO cohort (32%) [36] though the latter reported lower fatigue and pain levels. Lastly, a smaller subgroup with poor HRQoL across multiple domains was consistently observed across studies.
Interestingly, in our cohort, distinct HRQoL subgroups were identifiable at baseline, prior to treatment initiation, and remained largely stable over six months, with low transition probabilities. This aligns with findings from the CANTO cohort, which reported stable HRQoL trajectories persisting up to four years post-treatment in chemotherapy patients, except for a small subset (15%) exhibiting higher transition rates linked to reduced physical activity 1 to 2 years post-diagnosis [63]. Our results suggest that stable HRQoL subgrouping may be a generalizable phenomenon in early-stage breast cancer populations, independent of treatment modality. This is supported by stronger associations of sociodemographic and psychosocial factors with subgroup membership than tumor or treatment-related variables. Consistent with previous literature, multiple comorbidities, obesity, smoking, lower socioeconomic status, single marital status [32, 33, 64], and menopausal status [42, 65] were associated with poorer HRQoL subgroups. Notably, a history of depression was also strongly linked to membership in less favorable groups. Although prior clustering studies have not included history of depression, it has been associated with lower HRQoL [66, 67] and greater symptom burden at baseline [63] in patients with breast cancer. It further correlates with sociodemographic factors such as lower income, obesity, and smoking, all more prevalent in our Poor HRQoL subgroup [67]
Network analysis revealed distinct patterns of domain connectivity and centrality across subgroups. The Poor HRQoL-subgroup exhibited the highest overall connectivity, indicating a more tightly interlinked symptom structure. This aligns with prior studies reporting denser networks in cancer patients with lower HRQoL [42, 68, 69]. Whether high severity leads to denser networks or vice versa remains unresolved [70], and the underlying mechanisms are unclear. One hypothesis suggests that stronger inter-domain associations may facilitate a cascade. In those patients, highly connected domains can amplify minor disturbances, triggering widespread deterioration [71]. Furthermore, denser networks may reinforce persistence through mutual activation [72]. Accordingly, the optimal supportive care approach for this group is still unknown. Interventions could either target multiple domains simultaneously, or focus on the most central symptoms to induce more rapid downstream improvements across other domains [73, 74].
Subgroup-specific analyses also revealed variation in central domains. While previous studies in breast cancer populations have identified fatigue and emotional functioning as central domains [68, 75, 76] this pattern was only evident at the total sample level in our data. Subgroup analyses showed that emotional functioning was most central in the Psychosocial Concerns-subgroup, whereas hormonal symptoms, role functioning, and social functioning were central in the Poor HRQoL-subgroup. Over time, network connectivity remained stable in the Excellent HRQoL and Psychosocial Concerns subgroups, consistent with previous research suggesting that psychosocial symptom clusters are relatively stable throughout treatment [68, 77, 78]. In contrast, the Poor HRQoL-subgroup exhibited substantial changes in network structure, particularly in role and social functioning. This reinforces the notion that those domains may be more dynamic during active treatment [68, 77, 78], and therefore warrant closer monitoring to detect early deterioration.
Strengths and limitations
We had access to a large longitudinal database of women with early stage breast cancer completing PROMS in routine clinical care, with a relatively high response rate [79]. The dataset included HRQoL-data prior to treatment, alongside detailed clinical and sociodemographic information, enabling robust clustering and network analyses. By integrating those techniques we were able to thoroughly understand HRQoL subgroup trajectories before and during the initial phase of treatment.
While the high proportion of patients with excellent HRQoL scores and limited transition between subgroups may reflect clinical practice, this pattern could also be influenced by sample characteristics. Because the study was conducted at the Netherlands’ only comprehensive cancer center, the sample may under-represent individuals with lower SES. Patients with lower SES are more sensitive to travel burden [80], less inclined to seek second opinions [81], and less frequently receive treatment because they are often judged medically unfit [82], and this may reduce the chance of referral to a comprehensive cancer center. Additionally, PROMs were administered via a Dutch-language digital portal, potentially excluding patients with limited (digital) literacy or non-native speakers, groups consistently linked to lower HRQoL [83]. This potential selection bias is further illustrated by findings from the Carolina Breast Cancer Study (who purposefully sampled for greater ethnic and socioeconomic diversity) [32], where ‘only’ 11% of women was classified as having “Excellent” health, while 25% transitioned to a less favorable subgroup, associated with Black race, smoking, and lack of insurance. Replication of this study in a more diverse population is needed to improve representativeness and generalizability, which requires PROMs (research) to be more inclusive [84] to lower the barriers for patients with lower SES, limited (digital) literacy, and non-native speakers to participate.
Finally, as our cohort included only early-stage breast cancer patients, it remains unclear whether similarly stable HRQoL trajectories are found in patients with metastatic disease, who typically experience more pronounced HRQoL deterioration [64], underlining that future studies this group are needed.
Clinical implications and future directions
Our findings underscore that a one-size-fits-all approach to supportive care in breast cancer is suboptimal and that it might be possible to strategically re-allocate supportive care. This allows us to concentrate resources on a small subgroup experiencing poor HRQoL and high symptom interconnectivity. Notably, different central symptoms per subgroup suggests that both the intensity and the type of supportive care should be tailored accordingly.
A stratified care pathway should move away from standardized referrals based solely on treatment trajectories and anticipated side effects (as currently practiced in our institute) towards a personalized model guided by baseline HRQoL and sociodemographic factors. Baseline HRQoL may be sufficient to stratify patients into different supportive care pathways, given the generally low transition probabilities. Nonetheless, some patients might still needed monitoring: patients with history of depression, as this was associated with transitions to poorer HRQoL subgroups, and patients with lower activity levels, as the CANTO cohort linked reduced physical activity to transition to poorer subgroups [36]. Patients in the Good HRQoL-subgroup may benefit sufficiently from low-intensity interventions, including peer support, e-health tools promoting healthy behaviors, and patient-led access to healthcare professionals. Evidence from patient-led follow-up in breast cancer indicates that this approach can reduce consultations while preserving patient satisfaction and HRQoL [85–87]. Qualitative data further suggest that patients often feel empowered and less burdened, although a small subgroup reports unmet supportive care needs and may prefer more intensive support [88]. For patients with psychosocial burdens, short-term cognitive behavioral therapy targeting emotional functioning, pain, and future perspective may be particularly beneficial [89]. In contrast, patients in the Poor HRQoL and broad functional impairments-subgroup may require a multidisciplinary approach due to the involvement of multiple HRQoL-domains and the association with several actionable factors, including obesity, menopausal status, and a history of depression. Given the centrality of social and role functioning, as well as menopausal symptoms, interventions involving social work and proactive management of menopausal symptoms are crucial, as they may produce downstream benefits across other HRQoL-domains. As many in this group are single or have a lower socioeconomic status, interventions should extend beyond healthcare into social and policy domains. It should be noted that, in the Dutch healthcare system, not all supportive care services are reimbursed by basic health insurance, posing a significant barrier for patients with limited financial means [90, 91].
Conclusion
This study provides insights into different trajectories of HRQoL of patients with early stage breast cancer before the start of treatment until six months after treatment initiation. It suggests that prior functional status, baseline HRQoL, and sociodemographic factors, are indicative of the HRQoL trajectory in the first 6 months of treatment, and therefore should be used to strategically (re)allocate supportive care. Most patients reported excellent, stable HRQoL trajectories, suggesting that a low-intensity, self-management oriented support may suffice. Consequently, more intensive supportive care resources can be strategically directed towards those patients with the most complex needs.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
Not applicable.
Author contributions
All authors contributed to the study conception and design. Data collection and analysis were performed by Eva Boomstra.The first draft of the manuscript was written by Eva Boomstra and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding
The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
Data availability
The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request. Data are located in controlled access data storage at the Netherlands Cancer Institute.
Declarations
Ethics approval and consent to participate
All patients provide written informed consent for the use of their data for research purposes [44]; the use of their data for this study was approved by Internal Review Board (IRBd21-262).
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests to declare.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Elliott MJ, et al. Enhancing early-stage breast cancer survivorship: evidence-based strategies, surveillance testing, and imaging guidelines. Am Soc Clin Oncol Educ Book. 2024;44(3):e432564. [DOI] [PubMed] [Google Scholar]
- 2.Khajoei R, et al. Breast cancer survivors-supportive care needs: systematic review. BMJ Support Palliat Care. 2023;13(2):143–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Javan Biparva A, et al. Global quality of life in breast cancer: systematic review and meta-analysis. BMJ Support Palliat Care. 2023;13(e3):e528–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gonzalez L, et al. Health-related quality of life in patients with breast cancer in Latin America and the Caribbean: a systematic review and meta-analysis. Oncologist. 2021;26(5):e794–806. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.de Ligt KM, et al. The impact of health symptoms on health‑related quality of life in early‑stage breast cancer survivors. Breast Cancer Res Treat. 2019;178:703–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.de Ligt KM, et al. Patient-reported health problems and healthcare use after treatment for early-stage breast cancer. Breast. 2019;46:4–11. [DOI] [PubMed] [Google Scholar]
- 7.Licaj I, et al. Baseline quality of life and chemotherapy toxicities in patients with early breast cancer. Cancer. 2023;129(7):1085–95. [DOI] [PubMed] [Google Scholar]
- 8.Yanez B, et al. Association of modifiable risk factors with early discontinuation of adjuvant endocrine therapy: a post hoc analysis of a randomized clinical trial. JAMA Oncol. 2021;7(8):1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Thronicke A, et al. Self-reported baseline quality of life mirrors treatment-specific characteristics of cancer patients. Cancers. 2023;15(24):5763. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Husson O, et al. The EORTC QLQ‐C30 summary score as prognostic factor for survival of patients with cancer in the “real‐world”: results from the population‐based PROFILES registry. Oncologist. 2019;25(4):e722–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Efficace F, et al. Baseline health-related quality-of-life data as prognostic factors in a phase III multicentre study of women with metastatic breast cancer. Eur J Cancer. 2004;40(7):1021–30. [DOI] [PubMed] [Google Scholar]
- 12.Pergolotti M, et al. A health services research agenda to fully integrate cancer rehabilitation into oncology care. Cancer. 2019;125(22):3908–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Silver JK, Baima J, Mayer RS. Impairment-driven cancer rehabilitation: an essential component of quality care and survivorship. CA Cancer J Clin. 2013;63(5):295–317. [DOI] [PubMed] [Google Scholar]
- 14.Toohey K, et al. A systematic review of multimodal prehabilitation in breast cancer. Breast Cancer Res Treat. 2023;197(1):1–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hui D, Heung Y, Bruera E. Timely palliative care: personalizing the process of referral. Cancers. 2022;14(4):1047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Arving C, et al. Early rehabilitation of cancer patients—an individual randomized stepped-care stress-management intervention. Psychooncology. 2019;28(2):301–8. [DOI] [PubMed] [Google Scholar]
- 17.Sung H, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–49. [DOI] [PubMed] [Google Scholar]
- 18.American Society of Clinical Oncology. The state of cancer care in America, 2016: a report by the American Society of Clinical Oncology. J Oncol Pract. 2016;12(4):339–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Li J, et al. Effect of telehealth interventions on quality of life in cancer survivors: A systematic review and meta-analysis of randomized controlled trials. Int J Nurs Stud. 2021;122:103970. [DOI] [PubMed] [Google Scholar]
- 20.Spahrkäs SS, et al. Beating cancer-related fatigue with the Untire mobile app: results from a waiting-list randomized controlled trial. Psychooncology. 2020;29(11):1823–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Barcellini A, et al. Please mind the gap-about equity and access to care in oncology. ESMO Open. 2021;6(6):100335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Chan RJ, et al. Supportive care 2030 movement: towards unifying ambitions for global excellence in supportive cancer care—an international Delphi study. eClinicalMedicine. 2024;76:102825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Andersen BL, et al. Management of anxiety and depression in adult survivors of cancer: ASCO guideline update. J Clin Oncol. 2023;41(18):3426–53. [DOI] [PubMed] [Google Scholar]
- 24.Ligibel JA, et al. Exercise, diet, and weight management during cancer treatment: ASCO guideline. J Clin Oncol. 2022;40(22):2491–507. [DOI] [PubMed] [Google Scholar]
- 25.Stubblefield MD. The underutilization of rehabilitation to treat physical impairments in breast cancer survivors. PM R. 2017;9(9, Supplement 2):S317–23. [DOI] [PubMed] [Google Scholar]
- 26.Harrison JD, et al. What are the unmet supportive care needs of people with cancer? A systematic review. Support Care Cancer. 2009;17(8):1117–28. [DOI] [PubMed] [Google Scholar]
- 27.Chou Y-H, et al. Unmet supportive care needs of survival patients with breast cancer in different cancer stages and treatment phases. Taiwan J Obstet Gynecol. 2020;59(2):231–6. [DOI] [PubMed] [Google Scholar]
- 28.Rosario-Concepción RA, et al. Oncologists’ attitude and knowledge about cancer rehabilitation. PM&R. 2021;13(12):1357–61. [DOI] [PubMed] [Google Scholar]
- 29.Capozzi LC, et al. Feasibility and implementation of an oncology rehabilitation triage clinic: assessing rehabilitation, exercise need, and triage pathways within the Alberta Cancer Exercise–Neuro-Oncology Study. Curr Oncol. 2023;30(7):6220–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Naehrig DN, et al. Patient-reported outcome measures and supportive care need assessment in patients attending an Australian comprehensive care centre: a multi-method study. Support Care Cancer. 2021;29(9):5037–46. [DOI] [PubMed] [Google Scholar]
- 31.Cox CE, et al. Palliative care phenotypes among critically ill patients and family members: intensive care unit prospective cohort study. BMJ Support Palliat Care. 2022. 10.1136/spcare-2022-003622. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Pinheiro LC, et al. Examining health-related quality of life patterns in women with breast cancer. Qual Life Res. 2017;26(7):1733–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.De Ligt KM, et al. Varying severities of symptoms underline the relevance of personalized follow-up care in breast cancer survivors: latent class cluster analyses in a cross-sectional cohort. Support Care Cancer. 2022. 10.1007/s00520-022-07229-6. [DOI] [PubMed] [Google Scholar]
- 34.Chang G, et al. Symptom cluster study undergoing chemotherapy in breast cancer patients: latent class analysis and contemporaneous network analysis. Asia-Pac J Oncol Nurs. 2024;11(6):100499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Cai T, et al. Heterogeneity of symptoms and functions among women receiving chemotherapy for breast cancer in China: a multicentre, cross-sectional study. Front in Publ Health. 2022;10:952710. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Di Meglio A, et al. Dynamics of long-term patient-reported quality of life and health behaviors after adjuvant breast cancer chemotherapy. J Clin Oncol. 2022;40(27):3190–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Park J-H, et al. Trajectories of quality of life in breast cancer survivors during the first year after treatment: a longitudinal study. BMC Womens Health. 2023;23(1):12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.van Borkulo CD, et al. Comparing network structures on three aspects: a permutation test. Psychol Methods. 2023;28(6):1273–85. [DOI] [PubMed] [Google Scholar]
- 39.Khamboon T, Pakanta I. Intervention for symptom cluster management of fatigue, loss of appetite, and anxiety among patients with lung cancer undergoing chemotherapy. Asia Pac J Oncol Nurs. 2021;8(3):267–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Miaskowski C, et al. Advancing symptom science through symptom cluster research: expert panel proceedings and recommendations. JNCI J Natl Cancer Inst. 2017;109(4):djw253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.So WKW, et al. The effect of nonpharmacological interventions on managing symptom clusters among cancer patients: a systematic review. Cancer Nurs. 2020;43(6):E304–27. [DOI] [PubMed] [Google Scholar]
- 42.Cai T, et al. Cancer–related symptoms among young and middle–aged women undergoing chemotherapy for breast cancer: Application of latent class analysis and network analysis. Eur J Oncol Nurs. 2023;63:102287. [DOI] [PubMed] [Google Scholar]
- 43.Boomstra E, et al. Moving beyond barriers: a mixed-method study to develop evidence-based strategies to improve implementation of PROMs in clinical oncology care. Qual Life Res. 2024;34:1–16. [DOI] [PubMed] [Google Scholar]
- 44.Rebers S, et al. Consent for the secondary use of human residual tissue: the patient is best served by an ‘opting-out’ procedure. Ned Tijdschr Geneeskd. 2012;156(21):A4485. [PubMed] [Google Scholar]
- 45.Sangha O, et al. The self-administered comorbidity questionnaire: a new method to assess comorbidity for clinical and health services research. Arthritis Rheum. 2003;49(2):156–63. [DOI] [PubMed] [Google Scholar]
- 46.Van Duijn C. Sociaal-economische status indicator op postcode niveau. Maandsta Bevolk. 2002;50:32–5. [Google Scholar]
- 47.Aaronson NK, et al. The European organisation for research and treatment of cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst. 1993;85:365–76. [DOI] [PubMed] [Google Scholar]
- 48.R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2021.
- 49.Vermunt JK, Magidson J. Latent GOLD 4.0 user’s guide. Belmont: Statistical Innovations Inc. UK; 2005. [Google Scholar]
- 50.van Buuren S, Groothuis-Oudshoorn K. mice: multivariate imputation by chained equations in R. J Stat Softw. 2011;45(3):1–67. [Google Scholar]
- 51.Epskamp S, et al. qgraph: network visualizations of relationships in psychometric data. J Stat Softw. 2012;48:1–18. [Google Scholar]
- 52.Sinha P, Calfee CS, Delucchi KL. Practitioner’s guide to latent class analysis: methodological considerations and common pitfalls. Crit Care Med. 2021;49(1):e63–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Giesinger JM, et al. Thresholds for clinical importance were established to improve interpretation of the EORTC QLQ-C30 in clinical practice and research. J Clin Epidemiol. 2020;118:1–8. [DOI] [PubMed] [Google Scholar]
- 54.Pilz MJ, et al. PCR98 establishing thresholds for clinical importance for EORTC questionnaire modules for eight specific groups of cancer patients. Value Health. 2024;27(12):S524. [Google Scholar]
- 55.Abarda A, et al. Latent transition analysis (LTA): a method for identifying differences in longitudinal change among unobserved groups. Procedia Comput Sci. 2020;170:1116–21. [Google Scholar]
- 56.Vermunt JK. Latent class modeling with covariates: two improved three-step approaches. Polit Anal. 2017;18(4):450–69. [Google Scholar]
- 57.Epskamp S, Borsboom D, Fried EI. Estimating psychological networks and their accuracy: a tutorial paper. Behav Res Methods. 2018;50(1):195–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Burger J, et al. Reporting standards for psychological network analyses in cross-sectional data. Psychol Methods. 2023;28(4):806–24. [DOI] [PubMed] [Google Scholar]
- 59.Zhu Z, et al. Assessing psychological symptom networks related to HIV-positive duration among people living with HIV: a network analysis. AIDS Care. 2022;34(6):725–33. [DOI] [PubMed] [Google Scholar]
- 60.Musoro JZ, et al. Minimally important differences for interpreting EORTC QLQ-C30 change scores over time: a synthesis across 21 clinical trials involving nine different cancer types. Eur J Cancer. 2023;188:171–82. [DOI] [PubMed] [Google Scholar]
- 61.de Ligt KM, et al. Updated normative data for the EORTC QLQ-C30 in the general Dutch population by age and sex: a cross-sectional panel research study. Qual Life Res. 2023;32(9):2477–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Karsten MM, et al. Real-world reference scores for EORTC QLQ-C30 and EORTC QLQ-BR23 in early breast cancer patients. Eur J Cancer. 2022;163:128–39. [DOI] [PubMed] [Google Scholar]
- 63.Pagliuca M, et al. Long-term behavioral symptom clusters among survivors of early-stage breast cancer: Development and validation of a predictive model. J Natl Cancer Inst. 2025;117(1):89–102. [DOI] [PubMed] [Google Scholar]
- 64.Pat-Horenczyk R, et al. Trajectories of quality of life among an international sample of women during the first year after the diagnosis of early breast cancer: a latent growth curve analysis. Cancers. 2023;15(7):1961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Yeo W, et al. Menopausal symptoms in relationship to breast cancer-specific quality of life after adjuvant cytotoxic treatment in young breast cancer survivors. Health Qual Life Outcomes. 2020;18(1):24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Jim HSL, et al. History of major depressive disorder prospectively predicts worse quality of life in women with breast cancer. Ann Behav Med. 2011;43(3):402–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Ward EP, Sood D. Evaluating the impact of depression on self-reported quality of life measurements for patients with a history of cancer: a 2022 behavioral risk factor surveillance system questionnaire study. J Clin Oncol. 2024;42(16_suppl):e23131–e23131. [Google Scholar]
- 68.Doppenberg-Smit GE, et al. Network analysis used to investigate the interplay among somatic and psychological symptoms in patients with cancer and cancer survivors: a scoping review. J Cancer Surviv. 2024. 10.1007/s11764-024-01543-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 69.Bergsneider BH, et al. Discovery of clinical and demographic determinants of symptom burden in primary brain tumor patients using network analysis and unsupervised clustering. Neuro-Oncol Adv. 2023;5(1):vdac188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Neijenhuijs KI, et al. Symptom clusters among cancer survivors: what can machine learning techniques tell us? BMC Med Res Methodol. 2021;21(1):166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Robinaugh DJ, et al. The network approach to psychopathology: a review of the literature 2008–2018 and an agenda for future research. Psychol Med. 2020;50(3):353–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.Jiménez S, de Arango Montis I, Garza-Villarreal EA. Modeling vulnerability and intervention targets in the Borderline Personality Disorder system: a network analysis of in silico and in vivo interventions. PLoS ONE. 2023;18(7):e0289101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Fried EI, Cramer AOJ. Moving forward: challenges and directions for psychopathological network theory and methodology. Perspect Psychol Sci. 2017;12(6):999–1020. [DOI] [PubMed] [Google Scholar]
- 74.Schellekens MPJ, et al. Exploring the interconnectedness of fatigue, depression, anxiety and potential risk and protective factors in cancer patients: a network approach. J Behav Med. 2020;43(4):553–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75.Poikonen-Saksela P, et al. A graphical LASSO analysis of global quality of life, sub scales of the EORTC QLQ-C30 instrument and depression in early breast cancer. Sci Rep. 2022;12(1):2112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 76.Kim Y, et al. Network analysis of quality of life among young and middle-aged Korean cancer survivors. Asia Pac J Oncol Nurs. 2025. 10.1016/j.apjon.2025.100684. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 77.Airaksinen J, et al. Connectivity of depression symptoms before and after diagnosis of a chronic disease: a network analysis in the U.S. health and retirement study. J Affect Disord. 2020;266:230–4. [DOI] [PubMed] [Google Scholar]
- 78.So WKW, et al. Symptom clusters experienced by breast cancer patients at various treatment stages: a systematic review. Cancer Med. 2021;10(8):2531–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 79.Al-Antary N, et al. Interventions influencing patient-reported outcomes (PROs) response rates in cancer: a scoping review. J Cancer Surviv. 2025. 10.1007/s11764-025-01801-9. [DOI] [PubMed] [Google Scholar]
- 80.Versteeg SE, et al. Centralisation of cancer surgery and the impact on patients’ travel burden. Health Policy. 2018;122(9):1028–34. [DOI] [PubMed] [Google Scholar]
- 81.Greenfield G, et al. Patient-initiated second medical consultations-patient characteristics and motivating factors, impact on care and satisfaction: a systematic review. BMJ Open. 2021;11(9):e044033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 82.Filipe MD, et al. The association of socioeconomic status on treatment strategy in patients with stage I and II breast cancer in the Netherlands. Breast Cancer Res Treat. 2021;189(2):541–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 83.Holden CE, et al. The role of health literacy in cancer care: a mixed studies systematic review. PLoS ONE. 2021;16(11):e0259815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 84.Calvert MJ, et al. Patient reported outcome assessment must be inclusive and equitable. Nat Med. 2022;28(6):1120–4. [DOI] [PubMed] [Google Scholar]
- 85.Chapman D, et al. Patient-led breast cancer follow up. Breast. 2009;18(2):100–2. [DOI] [PubMed] [Google Scholar]
- 86.Riis CL, et al. Satisfaction with care and adherence to treatment when using patient reported outcomes to individualize follow-up care for women with early breast cancer—a pilot randomized controlled trial. Acta Oncol. 2020;59(4):444–52. [DOI] [PubMed] [Google Scholar]
- 87.Fjell M, et al. Reduced symptom burden with the support of an interactive app during neoadjuvant chemotherapy for breast cancer—a randomized controlled trial. Breast. 2020;51:85–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 88.Moore L, et al. Optimising patient-initiated follow-up care - a qualitative analysis of women with breast cancer in the UK. Eur J Oncol Nurs. 2022;60:102183. [DOI] [PubMed] [Google Scholar]
- 89.Anghel T, et al. Review of psychological interventions in oncology: current trends and future directions. Medicina (Kaunas). 2025;61(2):279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 90.Cheville A, et al. Policy and reimbursement considerations for exercise programming in cancer. In: Schmitz KH, editor., et al., Exercise oncology: prescribing physical activity before and after a cancer diagnosis. Cham: Springer International Publishing; 2020. p. 405–27. [Google Scholar]
- 91.Navntoft S, et al. Barriers and facilitators to cancer rehabilitation for patients with head and neck or lung cancer—a scoping review mapping structural and healthcare professionals’ perspectives. Disabil Rehabil. 2024;46(20):4617–29. [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 that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request. Data are located in controlled access data storage at the Netherlands Cancer Institute.