Abstract
Background
Predicting long-term benefit from CDK4/6 inhibitors in HR+/HER2− advanced breast cancer remains challenging in routine clinical practice. We evaluated whether a simple composite index integrating early neutropenia and mean corpuscular volume change, termed the MCV–Neutropenia Risk Score (MNRS), could improve early prognostic stratification.
Methods
In this multicenter retrospective study, 432 patients were screened, and 273 patients who were alive and evaluable at the predefined 3-month landmark were included in the final analysis. Grade 2–3 neutropenia during the first three months and ΔMCV ≥10 fL at the landmark assessment were assigned one point each to construct the MNRS (range, 0–2). Landmark overall survival (OS), measured from the 3-month landmark, was evaluated using Kaplan–Meier estimates and Cox proportional hazards regression. Internal bootstrap validation with 1,000 resamples was performed.
Results
After a median landmark follow-up of 44.9 months, 77 deaths were recorded. Grade 2–3 neutropenia occurred in 205 patients (75.1%), ΔMCV ≥10 fL in 93 patients (34.1%), and 66 patients (24.2%) had an MNRS of 2. Higher MNRS categories were associated with progressively more favorable landmark OS (log-rank p = 0.009). In the final center-stratified multivariable model, each one-point increase in MNRS was independently associated with a lower risk of death (HR 0.555, 95% CI 0.381–0.808; p = 0.002). Internal bootstrap validation supported the stability of MNRS in the multivariable model.
Conclusion
Among patients who were alive and evaluable at the 3-month landmark, higher MNRS was independently associated with more favorable subsequent landmark OS. MNRS may represent an exploratory and complementary approach to early risk stratification using routinely available hematologic parameters. Because both neutropenia and MCV changes may be influenced by treatment exposure and other clinical factors, prospective external validation is required before clinical application.
Keywords: advanced breast cancer, CDK4/6 inhibitors, treatment-emergent neutropenia, mean corpuscular volume, early on-treatment biomarker, overall survival
Introduction
CDK4/6 inhibitors have substantially improved progression-free and overall survival in patients with hormone receptor–positive, HER2-negative advanced breast cancer across multiple Phase III trials.1–3 Hematologic adverse events—particularly neutropenia—are among the most common toxicities, with grade 3–4 neutropenia reported in approximately half of treated patients depending on the specific agent and population studied.4,5 Unlike chemotherapy-induced myelosuppression, CDK4/6 inhibitor–associated neutropenia is generally reversible and reflects transient cell-cycle arrest of hematopoietic progenitors rather than direct cytotoxic marrow injury.6–8 This biological distinction has raised interest in whether treatment-emergent hematologic changes might represent pharmacodynamic markers of effective target inhibition rather than toxicity alone.9
Despite the survival benefits demonstrated in randomized trials, real-world outcomes remain heterogeneous.10–13 Observational cohorts have confirmed the effectiveness of CDK4/6 inhibitors while highlighting variability in treatment exposure, dose modifications, and adverse event management.12,14,15 These factors complicate interpretation of on-treatment laboratory changes and may confound simple associations between toxicity and survival.
Several retrospective analyses and post hoc evaluations have suggested that the development of neutropenia during CDK4/6 inhibitor therapy may be associated with improved clinical outcomes, potentially reflecting adequate drug exposure.9,16 However, findings have been inconsistent across real-world datasets, and interpretation is limited by time-dependent bias, treatment duration, dose adjustments, and patient selection.16–18 Consequently, neutropenia alone appears insufficient to fully capture the prognostic information embedded in treatment-related hematologic changes.
In parallel, increases in mean corpuscular volume (MCV) are frequently observed during CDK4/6 inhibitor therapy.19 Although macrocytosis is commonly encountered in oncology practice, its biological basis in this context is thought to involve altered erythroid maturation and cell-cycle regulation,7,19 and its prognostic relevance remains uncertain.20,21 Because MCV is a multifactorial parameter and prior studies have largely evaluated it in isolation, we hypothesized that integrating longitudinal MCV change with treatment-emergent neutropenia might provide complementary prognostic information.21 To our knowledge, such a composite hematologic index has not been systematically evaluated.
In this multicenter retrospective cohort study, we evaluated the prognostic significance of treatment-related hematologic changes in patients with advanced breast cancer receiving CDK4/6 inhibitors in routine clinical practice. Specifically, we assessed clinically significant neutropenia, longitudinal changes in mean corpuscular volume (ΔMCV), and an exploratory composite index—the MCV–Neutropenia Risk Score (MNRS)—in relation to landmark overall survival, aiming to explore whether a simple composite hematologic index could improve early prognostic stratification in this setting.
Methods
We conducted a multicenter retrospective cohort study across three tertiary oncology centers with established experience in CDK4/6 inhibitor use. Patients with hormone receptor-positive, human epidermal growth factor receptor 2-negative (HR+/HER2−) advanced breast cancer were retrospectively identified from institutional electronic medical records and oncology clinic treatment databases covering the period from January 2020 through June 2025. Among patients included in the final analytic cohort, first-line CDK4/6 inhibitor therapy in combination with endocrine treatment was initiated between April 2020 and November 2023. Survival status and follow-up information were updated through the administrative data cutoff in June 2025.
Eligible patients had histologically confirmed HR+/HER2− advanced breast cancer, initiated first-line CDK4/6 inhibitor-based therapy, and had complete baseline clinical and laboratory data. Patients were required to be alive and evaluable at the predefined 3-month landmark, with an on-treatment complete blood count available within the landmark window. Patients treated beyond the first-line setting or lacking required baseline, landmark, or follow-up data were excluded. No imputation was performed.
The primary outcome was landmark overall survival (OS), defined as the time from the predefined 3-month landmark to death from any cause; patients alive at last follow-up were censored at that date. Accordingly, the estimand represented survival conditional on being alive and evaluable at the 3-month landmark rather than survival from treatment initiation.
Demographic, clinical, and treatment-related variables included age, menopausal status, ECOG performance status, disease presentation, treating center, CDK4/6 inhibitor, treatment initiation date, and metastatic sites. Metastatic sites were not mutually exclusive; bone-only disease was defined as bone metastasis without recorded liver, lung, brain, or soft-tissue involvement. Menopausal status was based on clinical documentation, with age ≥60 years used to infer postmenopausal status when unavailable.22 All patients received first-line CDK4/6 inhibition with endocrine therapy, and premenopausal patients received ovarian function suppression. The endocrine therapy partner was not analyzed because patient-level data could not be reliably harmonized across centers.
MCV and ANC were obtained from routine complete blood counts at baseline and during treatment. To reduce immortal-time and other time-dependent biases, a predefined 3-month landmark was applied.18 Only patients alive and evaluable at the landmark were included; exposures were defined by the maximum neutropenia grade during the first 3 months and the MCV measurement closest to the landmark. All survival times were measured from the 3-month landmark.
Neutropenia was graded according to CTCAE version 5.0.23 For MNRS construction, the maximum grade 2–3 neutropenia recorded during the first 3 months was classified as the favorable neutropenia feature; grade 0–1/no neutropenia and grade 4 neutropenia were classified as absence of this feature.
ΔMCV was calculated as the MCV measurement closest to the 3-month landmark minus baseline MCV. A prespecified cutoff of ≥10 fL, selected a priori based on previous reports of CDK4/6 inhibitor–associated macrocytosis, was used for categorization.
MCV–Neutropenia Risk Score (MNRS)
MNRS was explored as a composite on-treatment hematologic index integrating two distinct but complementary treatment-related variables: (i) favorable ΔMCV and (ii) the occurrence of grade 2–3 neutropenia within the first three months of therapy. Each favorable component was assigned one point, resulting in a total score ranging from 0 to 2.
Patients were classified into three groups:
MNRS = 0: neither favorable ΔMCV nor grade 2–3 neutropenia
MNRS = 1: presence of either favorable ΔMCV or grade 2–3 neutropenia
MNRS = 2: presence of both favorable ΔMCV and grade 2–3 neutropenia
MNRS was evaluated as an exploratory composite marker in relation to landmark overall survival.
Statistical Analysis
Baseline characteristics were summarized for the overall cohort and by MNRS category. Continuous variables were reported as mean ± SD or median (IQR), as appropriate, and categorical variables as n (%). No inferential between-group comparisons were performed for Table 1.
Table 1.
Demographic, Clinical, Treatment-Related, and Hematologic Characteristics of the Study Population According to MNRS Category
| Variable | Overall (N=273) | MNRS = 0 (n=41) | MNRS = 1 (n=166) | MNRS = 2 (n=66) |
|---|---|---|---|---|
| Demographic characteristics | ||||
| Total patients, n | 273 | 41 | 166 | 66 |
| Age, mean ± SD, years | 48.7 ± 15.0 | 49.3 ± 14.6 | 48.2 ± 14.7 | 49.6 ± 16.3 |
| Menopausal status | ||||
| Premenopausal, n (%) | 116 (42.5%) | 18 (43.9%) | 70 (42.2%) | 28 (42.4%) |
| Postmenopausal, n (%) | 157 (57.5%) | 23 (56.1%) | 96 (57.8%) | 38 (57.6%) |
| Clinical and disease characteristics | ||||
| ECOG performance status | ||||
| ECOG PS 0–1, n (%) | 236 (86.4%) | 34 (82.9%) | 149 (89.8%) | 53 (80.3%) |
| ECOG PS 2–3, n (%) | 37 (13.6%) | 7 (17.1%) | 17 (10.2%) | 13 (19.7%) |
| Disease presentation | ||||
| De novo metastatic disease, n (%) | 192 (70.3%) | 30 (73.2%) | 125 (75.3%) | 37 (56.1%) |
| Recurrent metastatic disease, n (%) | 81 (29.7%) | 11 (26.8%) | 41 (24.7%) | 29 (43.9%) |
| Sites of metastatic involvement | ||||
| Liver, n (%) | 40 (14.7%) | 10 (24.4%) | 18 (10.8%) | 12 (18.2%) |
| Lung, n (%) | 78 (28.6%) | 11 (26.8%) | 47 (28.3%) | 20 (30.3%) |
| Bone, n (%) | 142 (52.0%) | 17 (41.5%) | 93 (56.0%) | 32 (48.5%) |
| Brain, n (%) | 14 (5.1%) | 1 (2.4%) | 9 (5.4%) | 4 (6.1%) |
| Soft tissue, n (%) | 35 (12.8%) | 6 (14.6%) | 24 (14.5%) | 5 (7.6%) |
| Bone-only metastatic disease, n (%) | 72 (26.4%) | 9 (22.0%) | 47 (28.3%) | 16 (24.2%) |
| Treatment characteristics | ||||
| Participating center | ||||
| Center 1, n (%) | 118 (43.2%) | 12 (29.3%) | 69 (41.6%) | 37 (56.1%) |
| Center 2, n (%) | 101 (37.0%) | 21 (51.2%) | 59 (35.5%) | 21 (31.8%) |
| Center 3, n (%) | 54 (19.8%) | 8 (19.5%) | 38 (22.9%) | 8 (12.1%) |
| CDK4/6 inhibitor | ||||
| Palbociclib, n (%) | 78 (28.6%) | 11 (26.8%) | 47 (28.3%) | 20 (30.3%) |
| Ribociclib, n (%) | 195 (71.4%) | 30 (73.2%) | 119 (71.7%) | 46 (69.7%) |
| Treatment initiation period | ||||
| 2020–2022, n (%) | 227 (83.2%) | 30 (73.2%) | 142 (85.5%) | 55 (83.3%) |
| 2023, n (%) | 46 (16.8%) | 11 (26.8%) | 24 (14.5%) | 11 (16.7%) |
| Hematologic characteristics | ||||
| Baseline MCV, median (IQR), fL | 88.0 (81.0–96.0) | 89.0 (84.0–98.0) | 91.0 (83.0–98.0) | 82.5 (74.0–87.0) |
| 3-month MCV, median (IQR), fL | 90.0 (82.0–98.0) | 85.0 (81.0–93.0) | 88.0 (80.3–96.0) | 98.0 (95.0–103.0) |
| ΔMCV, median (IQR), fL | 4.0 (−8.0–12.0) | −1.0 (−9.0–4.0) | −2.0 (−11.0–6.0) | 14.5 (12.0–20.0) |
| ΔMCV ≥10 fL, n (%) | 93 (34.1%) | 0 (0.0%) | 27 (16.3%) | 66 (100.0%) |
| Maximum neutropenia grade within 3 months | ||||
| Grade 0, n (%) | 12 (4.4%) | 9 (22.0%) | 3 (1.8%) | 0 (0.0%) |
| Grade 1, n (%) | 26 (9.5%) | 15 (36.6%) | 11 (6.6%) | 0 (0.0%) |
| Grade 2, n (%) | 65 (23.8%) | 0 (0.0%) | 38 (22.9%) | 27 (40.9%) |
| Grade 3, n (%) | 140 (51.3%) | 0 (0.0%) | 101 (60.8%) | 39 (59.1%) |
| Grade 4, n (%) | 30 (11.0%) | 17 (41.5%) | 13 (7.8%) | 0 (0.0%) |
| Grade 2–3 neutropenia, n (%) | 205 (75.1%) | 0 (0.0%) | 139 (83.7%) | 66 (100.0%) |
Notes: Data are presented as n (%) unless otherwise indicated; percentages are column percentages and may not total 100% because of rounding. Metastatic sites were not mutually exclusive. Bone-only metastatic disease was defined as bone metastasis in the absence of recorded liver, lung, brain, and soft-tissue metastases. ΔMCV was calculated as 3-month MCV minus baseline MCV. Grade 2–3 neutropenia denotes a maximum grade 2 or 3 event during the first 3 months; grade 4 neutropenia was not counted as the favorable neutropenia component. MNRS was calculated by assigning 1 point each for ΔMCV ≥10 fL and grade 2–3 neutropenia within the landmark window. Center 1, Center 2, and Center 3 denote the three participating institutions. Treatment initiation period was grouped as 2020–2022 versus 2023 because no patient initiated CDK4/6 inhibitor therapy after 2023 in the analytic cohort. Patient-level endocrine therapy partner was not presented because this variable could not be reliably harmonized across centers; all patients received concomitant endocrine therapy, and premenopausal patients received ovarian function suppression. No inferential between-group comparisons were performed for this descriptive table.
Abbreviations: CDK, cyclin-dependent kinase; ECOG PS, Eastern Cooperative Oncology Group performance status; IQR, interquartile range; MCV, mean corpuscular volume; MNRS, MCV–Neutropenia Risk Score; SD, standard deviation.
Landmark OS was estimated using the Kaplan–Meier method and compared with the Log rank test. Median follow-up from the landmark was calculated using the reverse Kaplan–Meier method, and survival estimates were reported with 95% CIs where estimable. Center-specific landmark OS was also evaluated using Kaplan–Meier analysis and the Log rank test.
Patients were also classified into four mutually exclusive component-defined groups: neither component, ΔMCV ≥10 fL alone, grade 2–3 neutropenia alone, and both components. Landmark OS was compared across groups using Kaplan–Meier analysis and the global Log rank test. Center-stratified Cox models adjusted for menopausal status were used to estimate group-specific hazard ratios, with neither component as the reference, and to test the formal ΔMCV-by-neutropenia multiplicative interaction.
Univariable Cox regression was performed for age, menopausal status, ECOG performance status, liver and lung metastases, recurrent versus de novo disease, grade 2–3 neutropenia, ΔMCV category, and MNRS. Variables with univariable p < 0.10 were entered into a backward likelihood-ratio multivariable model; menopausal status, grade 2–3 neutropenia, and MNRS met this criterion, whereas age and ΔMCV category did not. The primary multivariable model was stratified by center, with a sensitivity analysis modeling center as a categorical fixed effect. All Cox models used time from the 3-month landmark. Proportional hazards were assessed using scaled Schoenfeld residuals, including covariate-specific and global tests. HRs with 95% CIs were reported. Internal validation was performed in R using 1,000 bootstrap resamples to assess variable-selection stability.
Statistical analyses were performed using IBM SPSS Statistics (version 26.0) and R software (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two-sided, and a p value < 0.05 was considered statistically significant.
This study was approved by the Ethics Committee of Göztepe Prof. Dr. Süleyman Yalçın City Hospital (approval number: 2025/0120; approval date: 28 August 2025) and was conducted in accordance with the ethical principles of the Declaration of Helsinki. Owing to the retrospective nature of the study, the requirement for informed consent was waived by the Ethics Committee. All patient data were anonymized prior to analysis.
Results
A total of 432 patients were screened, of whom 273 were included in the final landmark cohort (Figure 1). The mean age was 48.7 ± 15.0 years; 116 patients (42.5%) were premenopausal, 236 (86.4%) had ECOG performance status 0–1, and 192 (70.3%) presented with de novo metastatic disease. Bone was the most common metastatic site (142 patients, 52.0%). Ribociclib was administered to 195 patients (71.4%) and palbociclib to 78 (28.6%). Centers 1, 2, and 3 contributed 118 (43.2%), 101 (37.0%), and 54 (19.8%) patients, respectively. Detailed demographic, clinical, disease-related, and treatment characteristics, including their distribution across MNRS categories, are presented in Table 1.
Figure 1.
Flow diagram illustrating patient screening, eligibility assessment, exclusion criteria, and inclusion in the final landmark survival analysis cohort.
The median baseline MCV was 88.0 fL (IQR, 81.0–96.0), increasing to 90.0 fL (IQR, 82.0–98.0) at the 3-month assessment. The median ΔMCV was 4.0 fL (IQR, −8.0 to 12.0); 93 patients (34.1%) had ΔMCV ≥10 fL, and 205 (75.1%) developed grade 2–3 neutropenia. According to the MNRS, 41 patients (15.0%) were classified as MNRS = 0, 166 (60.8%) as MNRS = 1, and 66 (24.2%) as MNRS = 2. Detailed hematologic characteristics are presented in Table 1, and the cross-tabulation of the two MNRS components is provided in Supplementary Table 1.
Among the 273 patients who were alive and evaluable at the predefined 3-month landmark, 77 deaths (28.2%) were recorded during subsequent follow-up. The median follow-up from the landmark, estimated using the reverse Kaplan–Meier method, was 44.9 months (95% CI, 43.4–47.2). Landmark Kaplan–Meier analyses were performed according to ΔMCV category, the occurrence of grade 2–3 neutropenia, and MNRS category. Among patients with ΔMCV <10 fL, 56 deaths occurred in 180 patients, whereas 21 deaths occurred among 93 patients with ΔMCV ≥10 fL. Patients with ΔMCV ≥10 fL had numerically higher landmark survival probabilities than those with ΔMCV <10 fL; however, the difference was not statistically significant (log-rank χ2 = 2.53; p = 0.112). The estimated 2- and 3-year landmark OS rates were 91.4% (95% CI, 83.5–95.6) and 83.1% (95% CI, 73.4–89.5), respectively, in the ΔMCV ≥10 fL group, compared with 83.3% (95% CI, 77.0–88.0) and 73.5% (95% CI, 66.2–79.4) in the ΔMCV <10 fL group. Median landmark OS was not reached in either group. These findings are illustrated in Figure 2.
Figure 2.
Kaplan–Meier curves for landmark overall survival according to ΔMCV category Landmark overall survival was compared between patients with ΔMCV ≥10 fL and those with ΔMCV <10 fL. Survival time was measured from the predefined 3-month landmark among patients who were alive and evaluable at that timepoint. Tick marks indicate censored observations, and numbers at risk are shown below the x-axis.
Among patients who did not develop grade 2–3 neutropenia, 28 deaths occurred in 68 patients, whereas 49 deaths occurred among 205 patients who developed grade 2–3 neutropenia. Landmark OS differed significantly according to the occurrence of grade 2–3 neutropenia within the first three months of treatment (log-rank χ2 = 7.01; p = 0.008). The estimated 2- and 3-year landmark OS rates were 86.8% (95% CI, 81.4–90.8) and 80.2% (95% CI, 73.9–85.2), respectively, among patients with grade 2–3 neutropenia, compared with 83.8% (95% CI, 72.7–90.7) and 66.0% (95% CI, 53.0–76.2) among patients without grade 2–3 neutropenia. Median landmark OS was not reached in either group. These findings are illustrated in Figure 3.
Figure 3.
Kaplan–Meier curves for landmark overall survival according to grade 2–3 neutropenia Landmark overall survival was compared between patients who developed grade 2–3 neutropenia within the first three months of treatment and those who did not. Survival time was measured from the predefined 3-month landmark among patients who were alive and evaluable at that timepoint. Median landmark overall survival was not reached in either group. Tick marks indicate censored observations, and numbers at risk are shown below the x-axis.
Landmark OS differed across MNRS categories. Seventeen deaths occurred among 41 patients with MNRS = 0, 50 deaths among 166 patients with MNRS = 1, and 10 deaths among 66 patients with MNRS = 2. Kaplan–Meier analysis demonstrated graded separation of landmark survival curves across MNRS categories (global log-rank χ2 = 9.32; p = 0.009). The estimated 2- and 3-year landmark OS rates were 92.4% (95% CI, 82.8–96.8) and 89.0% (95% CI, 78.2–94.6), respectively, in patients with MNRS = 2; 84.9% (95% CI, 78.5–89.6) and 74.8% (95% CI, 67.2–80.9) in patients with MNRS = 1; and 80.5% (95% CI, 64.8–89.7) and 64.7% (95% CI, 47.6–77.4) in patients with MNRS = 0. Median landmark OS was not reached in any MNRS category. These findings are illustrated in Figure 4.
Figure 4.
Kaplan–Meier curves for landmark overall survival according to the MCV–Neutropenia Risk Score Landmark overall survival was compared across MNRS categories 0, 1, and 2. MNRS was calculated by assigning one point each for ΔMCV ≥10 fL and grade 2–3 neutropenia occurring within the first three months of treatment. Survival time was measured from the predefined 3-month landmark among patients who were alive and evaluable at that timepoint. Median landmark overall survival was not reached in any MNRS category. Tick marks indicate censored observations, and numbers at risk are shown below the x-axis.
Abbreviations: ΔMCV, change in mean corpuscular volume from baseline to 3 months; MCV, mean corpuscular volume; MNRS, MCV–Neutropenia Risk Score; OS, overall survival.
To examine the individual and joint associations of the two MNRS components, patients were additionally classified into four mutually exclusive groups: neither component (n = 41; 17 deaths), ΔMCV ≥10 fL alone (n = 27; 11 deaths), grade 2–3 neutropenia alone (n = 139; 39 deaths), and both components (n = 66; 10 deaths). Landmark OS differed significantly across the four groups (global log-rank χ2 = 10.56, 3 degrees of freedom; p = 0.014). In the center-stratified Cox model adjusted for menopausal status, ΔMCV ≥10 fL alone was not significantly associated with landmark OS compared with neither component (HR 0.823, 95% CI 0.379–1.788; p = 0.623). Grade 2–3 neutropenia alone showed a numerically favorable but statistically nonsignificant association (HR 0.600, 95% CI 0.335–1.076; p = 0.087), whereas the presence of both components was associated with a significantly lower risk of death (HR 0.292, 95% CI 0.131–0.653; p = 0.003). The formal ΔMCV-by-neutropenia interaction term was not statistically significant (HR 0.591, 95% CI 0.209–1.671; p = 0.321), indicating no statistical evidence that the joint association departed from multiplicativity. Detailed Kaplan–Meier estimates and adjusted Cox regression results are presented in Supplementary Table 4.
Center-specific landmark survival analyses were also performed. Centers 1, 2, and 3 included 118, 101, and 54 patients, with 32, 26, and 19 deaths, respectively. The estimated 2- and 3-year landmark OS rates were 87.3% and 78.3% at Center 1, 85.1% and 79.5% at Center 2, and 85.2% and 67.9% at Center 3, respectively. Landmark OS did not differ significantly across the three participating centers (global log-rank χ2 = 2.07; p = 0.355).
Univariable Cox proportional hazards regression analyses were performed using landmark OS as the outcome (Table 2). Grade 2–3 neutropenia was associated with a lower risk of death in the univariable analysis (HR 0.539, 95% CI 0.339–0.858; p = 0.009), whereas ΔMCV ≥10 fL showed a numerically favorable but statistically nonsignificant association (HR 0.667, 95% CI 0.404–1.102; p = 0.114). Each one-point increase in MNRS was associated with a lower risk of death (HR 0.572, 95% CI 0.399–0.820; p = 0.002). Postmenopausal status was associated with a lower estimated hazard of death, although this did not reach statistical significance (HR 0.640, 95% CI 0.409–1.001; p = 0.051), whereas age, ECOG performance status, liver and lung metastases, and recurrent versus de novo metastatic disease were not significantly associated with landmark OS (Table 2).
Table 2.
Univariable, Initial Multivariable, and Final Center-Stratified Cox Proportional Hazards Regression Analyses for Landmark Overall Survival
| Variable | Univariable Analysis | Initial Center-Stratified Multivariable Model | Final Center-Stratified Backward Model | |||
|---|---|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | |
| Age, per year | 0.991 (0.976–1.006) | 0.220 | — | — | — | — |
| Menopausal status (postmenopausal vs premenopausal) | 0.640 (0.409–1.001) | 0.051 | 0.605 (0.385–0.950) | 0.029 | 0.607 (0.386–0.953) | 0.030 |
| ECOG PS 2–3 vs 0–1 | 1.444 (0.795–2.623) | 0.227 | — | — | — | — |
| Liver metastasis | 0.552 (0.254–1.201) | 0.134 | — | — | — | — |
| Lung metastasis | 1.160 (0.716–1.879) | 0.546 | — | — | — | — |
| Recurrent vs de novo metastatic disease | 0.747 (0.445–1.255) | 0.271 | — | — | — | — |
| Grade 2–3 neutropenia | 0.539 (0.339–0.858) | 0.009 | 0.844 (0.436–1.634) | 0.614 | — | — |
| ΔMCV ≥10 fL vs <10 fL | 0.667 (0.404–1.102) | 0.114 | — | — | — | — |
| MNRS, per 1-point increase |
0.572 (0.399–0.820) | 0.002 | 0.608 (0.364–1.014) | 0.057 | 0.555 (0.381–0.808) | 0.002 |
Notes: All Cox models used time from the predefined 3-month landmark as the survival time scale. MNRS was modeled as an ordinal variable (range, 0–2), and its HR represents the relative change in the hazard of death per one-point increase. Variables with a univariable p value <0.10 were entered into the initial center-stratified multivariable Cox model. The initial model included menopausal status, grade 2–3 neutropenia, and MNRS. Grade 2–3 neutropenia was removed during backward likelihood-ratio selection; postmenopausal status and MNRS remained in the final model. The Cox models were stratified by participating center, allowing the baseline hazard function to vary across centers. A dash indicates that the variable was not included in the respective model. All p values are two-sided.
Abbreviations: CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; LR, likelihood ratio; MCV, mean corpuscular volume; MNRS, MCV–Neutropenia Risk Score; OS, overall survival; ΔMCV, change in mean corpuscular volume from baseline to the 3-month landmark.
Based on the prespecified univariable p-value threshold of <0.10, menopausal status, grade 2–3 neutropenia, and MNRS were entered into the initial center-stratified multivariable Cox model. In this initial model, postmenopausal status remained significantly associated with a lower risk of death (HR 0.605, 95% CI 0.385–0.950; p = 0.029), whereas grade 2–3 neutropenia was no longer independently associated with landmark OS (HR 0.844, 95% CI 0.436–1.634; p = 0.614). The association between MNRS and landmark OS did not reach statistical significance in the initial model (HR 0.608 per one-point increase, 95% CI 0.364–1.014; p = 0.057). Following backward likelihood-ratio elimination, grade 2–3 neutropenia was removed from the model. In the final center-stratified model, postmenopausal status (HR 0.607, 95% CI 0.386–0.953; p = 0.030) and MNRS (HR 0.555 per one-point increase, 95% CI 0.381–0.808; p = 0.002) remained independently associated with landmark OS. Detailed univariable, initial multivariable, and final multivariable Cox regression results are presented in Table 2.
In the fixed-effect sensitivity model, the associations of postmenopausal status (HR 0.594, 95% CI 0.378–0.934; p = 0.024) and MNRS (HR 0.554 per one-point increase, 95% CI 0.381–0.806; p = 0.002) with landmark OS were materially unchanged, and neither center covariate was independently associated with outcome (Supplementary Table 2).
The proportional hazards assumption was assessed for covariates retained in the final center-stratified Cox model. No evidence of nonproportional hazards was detected for postmenopausal status (χ2 = 3.35; p = 0.067), MNRS (χ2 = 0.28; p = 0.594), or globally (χ2 = 3.63, 2 degrees of freedom; p = 0.163) (Supplementary Table 3).
To assess the robustness of the variable selection procedure, internal bootstrap validation with 1,000 resamples was performed using the same center-stratified backward likelihood-ratio approach. MNRS was retained in 80.1% of bootstrap samples, postmenopausal status in 70.1%, and grade 2–3 neutropenia in 20.2%. These findings supported the relative stability of MNRS and menopausal status, whereas the low selection frequency of grade 2–3 neutropenia was consistent with its exclusion from the final multivariable model. Detailed bootstrap results are presented in Table 3.
Table 3.
Internal Bootstrap Validation and Variable-Selection Stability of the Final Center-Stratified Cox Model for Landmark Overall Survival
| Variable | Original Final Center-Stratified Model HR (95% CI) | p | Bootstrap Selection Frequency (%) |
|---|---|---|---|
| Postmenopausal vs premenopausal | 0.607 (0.386–0.953) | 0.030 | 70.1 |
| Grade 2–3 neutropenia | — | — | 20.2 |
| MNRS, per 1-point increase | 0.555 (0.381–0.808) | 0.002 | 80.1 |
Notes: Internal bootstrap validation was performed using 1,000 resamples and the same center-stratified backward likelihood-ratio selection procedure used in the primary analysis. The candidate variables were menopausal status, grade 2–3 neutropenia, and MNRS. In each bootstrap resample, the Cox model was stratified by participating center, and selection frequency was calculated as the percentage of resamples in which each variable was retained after backward elimination. Original HRs and p values are shown only for variables retained in the final primary model; a dash indicates that the variable was not retained. Bootstrap selection frequencies assess variable-selection stability and do not constitute external validation or replace formal assessment of model discrimination or calibration. All p values are two-sided.
Abbreviations: CI, confidence interval; HR, hazard ratio; LR, likelihood ratio; MNRS, MCV–Neutropenia Risk Score; OS, overall survival.
Discussion
Despite the established survival benefits of CDK4/6 inhibitors, outcomes remain heterogeneous in real-world practice,1–3,10–13 prompting interest in early treatment-emergent markers of prognosis. In unadjusted analyses, grade 2–3 neutropenia was associated with more favorable subsequent landmark OS, whereas ΔMCV ≥10 fL showed a numerically favorable but statistically nonsignificant survival pattern among patients who were alive and evaluable at the 3-month landmark. When evaluated individually, grade 2–3 neutropenia did not remain independently associated with landmark OS after multivariable adjustment, whereas ΔMCV ≥10 fL did not meet the prespecified threshold for entry into the multivariable model. Their combined evaluation as the MCV–Neutropenia Risk Score (MNRS) demonstrated graded stratification of landmark overall survival. Importantly, MNRS remained independently associated with landmark OS after multivariable adjustment, supporting further evaluation of this composite as an exploratory prognostic marker.
Neutropenia has been proposed as a pharmacodynamic marker of CDK4/6 inhibitor exposure, with early studies linking treatment-emergent neutropenia to improved response.9 However, real-world analyses indicate that neutropenia is closely related to treatment duration, dose intensity, dose modifications, and time-dependent bias, including immortal-time bias.16–18 In our cohort, grade 2–3 neutropenia was associated with improved survival in univariable analysis but not after multivariable adjustment, suggesting that it may reflect treatment exposure rather than serve as an independent surrogate of long-term benefit.
In parallel, ΔMCV and treatment-emergent macrocytosis have attracted increasing attention as potential biomarkers during CDK4/6 inhibitor therapy. Biologically, macrocytosis is plausible, as CDK4/6 inhibition may alter erythroid maturation through cell-cycle modulation.7,19 Early observations described palbociclib-associated macrocytosis and suggested that greater increases in MCV were associated with prolonged treatment exposure, while also emphasizing the cumulative and time-dependent nature of this phenomenon.19 Subsequent real-world studies have yielded inconsistent results. For example, some cohorts have reported improved survival among patients exhibiting macrocytosis,20,24 whereas other analyses have not demonstrated an independent association with progression-free or overall survival after multivariable adjustment.21 These discrepant findings are not unexpected, as ΔMCV is influenced by multiple non–treatment-related factors and is particularly susceptible to time-related bias when interpreted in isolation. In addition, because nutritional status, vitamin deficiencies, and other non-treatment-related clinical factors were not systematically assessed in the present retrospective cohort, residual confounding related to MCV cannot be excluded. In our study, higher ΔMCV values were associated with numerically improved survival and early curve separation; however, ΔMCV ≥10 fL did not meet the prespecified threshold for entry into the multivariable model, supporting a cautious interpretation of this marker when assessed alone.
Taken together, the limitations observed for neutropenia and ΔMCV when evaluated separately provided the rationale for exploring a composite approach such as MNRS. Neutropenia may reflect host myeloid sensitivity and drug exposure during CDK4/6 inhibition, whereas ΔMCV may capture erythroid effects of sustained cell-cycle modulation; however, both parameters are also influenced by treatment duration, dose modifications, supportive care, and competing clinical factors. Therefore, MNRS should be interpreted as an exploratory host- and treatment-related hematologic signal rather than a direct measure of tumor biology or molecular resistance. In the landmark cohort, MNRS demonstrated graded separation of survival curves and remained independently associated with landmark OS after multivariable adjustment, supporting its further evaluation as a composite exploratory prognostic marker. Although patients with both components had the most favorable adjusted survival estimate in the four-group analysis, the formal ΔMCV-by-neutropenia interaction was not statistically significant; therefore, these findings should not be interpreted as evidence of a synergistic interaction between the two components.
From a clinical perspective, MNRS may have potential as an early risk-stratification framework rather than a tool for direct treatment modification. Because it is derived from routinely collected hematologic parameters, it could be evaluated without requiring additional biomarker testing. Lower MNRS values were associated with poorer survival outcomes in this cohort, whereas higher MNRS values were associated with more favorable survival estimates. However, the potential clinical utility of these observations requires prospective validation before any practice implications can be considered. Importantly, MNRS should not be interpreted as a tool to induce toxicity or justify dose escalation; rather, it provides an observational framework to contextualize naturally occurring on-treatment hematologic changes and to complement established clinical and radiologic assessments. Accordingly, the present findings should be considered hypothesis-generating and should not be used to guide treatment decisions without prospective validation. While survival outcomes remain a major priority, treatment-related toxicity, quality of life, and patient preferences also influence therapeutic decisions in advanced breast cancer. Therefore, if validated prospectively, MNRS may serve as a complementary tool to facilitate individualized patient counseling alongside established clinical, radiologic, and molecular assessments.25
interpretation of MNRS should also be placed within the broader biological context of HR+/HER2− advanced breast cancer. Outcomes with CDK4/6 inhibitors are influenced by multiple tumor- and treatment-related factors, including disease burden, endocrine sensitivity, metastatic pattern, ESR1 and PIK3CA pathway alterations, RB pathway integrity, CCNE1 amplification, HER2-low biology, germline BRCA/PALB2 status, and the availability of subsequent targeted therapies. In this evolving treatment landscape, early molecular monitoring strategies, such as ctDNA-guided detection of emerging ESR1 mutations in studies including SERENA-6, represent more direct approaches to assessing endocrine resistance biology.26 By contrast, MNRS is not a genomic or tumor-intrinsic biomarker, and our study cannot determine whether hematologic changes are linked to molecular resistance mechanisms. Rather, MNRS may reflect a composite host-treatment interaction during CDK4/6 inhibitor exposure and should be considered complementary to, not a substitute for, tumor genomic profiling and molecular resistance monitoring. Importantly, toxicity should not be universally interpreted as a marker of efficacy across anticancer drug classes; hematologic or systemic toxicity from agents such as PARP inhibitors or antibody–drug conjugates may instead lead to dose interruption, treatment discontinuation, or worse outcomes in some clinical contexts. Therefore, the significant unadjusted association observed with early grade 2–3 neutropenia and the numerically favorable but nonsignificant pattern observed with ΔMCV should be interpreted as context-specific and hypothesis-generating. Grade 2–3 neutropenia did not retain statistical significance after multivariable adjustment, whereas ΔMCV did not meet the prespecified threshold for entry into the multivariable model.
Several limitations and interpretative considerations should be acknowledged. First, survival estimates in the present study were calculated conditionally from the predefined 3-month landmark and therefore should not be directly compared with overall survival estimates measured from treatment initiation in pivotal phase III trials of CDK4/6 inhibitors.1–3 CDK4/6 inhibitors were incorporated into national reimbursement policies in Türkiye later than in many Western healthcare systems; therefore, our cohort reflects a more recent therapeutic era. At the time of analysis, the median follow-up from the 3-month landmark was 44.9 months (95% CI, 43.4–47.2), and 77 deaths (28.2%) had been recorded. Median landmark OS was not reached in any MNRS category, indicating that longer observation and additional event accrual are required to characterize the later portions of the survival curves. Additionally, advances in subsequent-line treatment strategies in more contemporary practice settings may have favorably influenced post-progression outcomes compared with earlier registration trials. These factors should be considered when interpreting cross-study comparisons of long-term survival outcomes.
Second, the retrospective design introduces potential selection bias and residual confounding despite multivariable adjustment. By design, the landmark analysis estimated survival conditional on patients being alive and evaluable at the 3-month landmark; therefore, the findings cannot be generalized to patients who died, discontinued treatment, or lacked evaluable hematologic data before the landmark. Although the landmark approach reduced immortal-time and other time-dependent biases, residual time-related confounding cannot be fully excluded. In particular, detailed longitudinal data regarding dose reductions, dose interruptions, relative dose intensity, and cumulative treatment exposure were not available for formal time-dependent modeling. Because treatment-emergent neutropenia itself may influence subsequent dose modifications, the relationship between hematologic changes and survival outcomes should be interpreted cautiously. Laboratory assessments were performed as part of routine clinical care and were not strictly standardized across centers. In addition, detailed patient-level information regarding the endocrine therapy partner could not be reliably harmonized across centers and was therefore not included in the stratified descriptive analyses. Accordingly, the observed association between higher MNRS and improved survival should not be interpreted as evidence that MNRS directly reflects pharmacodynamic efficacy, as treatment exposure itself may represent an important unmeasured confounder. Prospective studies incorporating longitudinal dose intensity, cumulative exposure, and time-dependent treatment modifications will be required to clarify whether MNRS reflects treatment pharmacodynamics, treatment exposure, or a combination of both.
Third, dichotomization of hematologic variables, including ΔMCV, may have resulted in some loss of information compared with continuous modeling approaches. Although the number of observed events allowed construction of a parsimonious multivariable Cox regression model, regression models derived using backward variable selection may be less stable in datasets with a moderate number of events. To address this potential limitation, we performed an internal bootstrap validation using 1,000 resamples, which demonstrated consistent retention of the final prognostic variables and supported the robustness of the final multivariable model. Nevertheless, bootstrap validation cannot substitute for independent external validation. Furthermore, discrimination metrics such as the C-index and calibration analyses were not formally evaluated. The absence of patients treated with abemaciclib in our cohort, largely reflecting national reimbursement policies during the study period, may also limit the generalizability of our findings across all CDK4/6 inhibitors. Finally, only overall survival was evaluated. Given the potential influence of subsequent-line therapies in HR+/HER2− advanced breast cancer, future prospective studies incorporating progression-free survival, external validation, and formal model performance assessment are warranted.
study also has several important strengths. The multicenter design enhances generalizability across routine clinical settings, while center stratification and sensitivity analyses allowed potential between-center variation to be explicitly considered. The sample size and number of observed events provide a clinically relevant real-world dataset for evaluating on-treatment biomarkers in advanced breast cancer. Importantly, unlike prior studies that examined treatment-emergent neutropenia or macrocytosis in isolation, our analysis integrates complementary hematologic parameters within a predefined landmark framework, offering a structured exploratory framework for evaluating early treatment-emergent hematologic changes.
Despite these limitations, our findings suggest that the exploratory MCV–Neutropenia Risk Score (MNRS) may provide complementary prognostic information based on routinely available hematologic parameters obtained early during CDK4/6 inhibitor therapy. However, MNRS should not be regarded as a surrogate marker of treatment efficacy or pharmacodynamic activity, nor should it replace established clinical, radiologic, or molecular assessment. Rather, it should be considered a hypothesis-generating exploratory tool requiring prospective external validation, standardized assessment timing, and comprehensive adjustment for treatment exposure, dose intensity, and other potential confounders before any clinical application can be recommended.
Funding Statement
This research received no external funding.
Generative AI Statement
Generative AI tools (ChatGPT, OpenAI; GPT-5.3) were used only for language refinement, grammar editing, and improvement of manuscript readability. All scientific content, data interpretation, statistical analyses, and final editorial decisions were performed and verified by the authors.
Data Sharing Statement
The datasets generated and/or analyzed during the current study are not publicly available because they contain potentially identifiable patient information but are available from the corresponding author on reasonable request.
Author Contributions
All authors made a significant contribution to the work reported, whether in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising, or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.
Disclosure
The authors declare that they have no competing interests.
References
- 1.Goetz MP, Toi M, Huober J, et al. Abemaciclib plus a nonsteroidal aromatase inhibitor as initial therapy for HR+, HER2− advanced breast cancer: final overall survival results of MONARCH 3. Ann Oncol. 2024;35(8):718–15. doi: 10.1016/j.annonc.2024.04.013 [DOI] [PubMed] [Google Scholar]
- 2.Slamon DJ, Neven P, Chia S, et al. Ribociclib plus fulvestrant for postmenopausal women with hormone receptor-positive, human epidermal growth factor receptor 2-negative advanced breast cancer in the phase III randomized MONALEESA-3 trial: updated overall survival. Ann Oncol. 2021;32(8):1015–1024. doi: 10.1016/j.annonc.2021.05.353 [DOI] [PubMed] [Google Scholar]
- 3.Tripathy D, Im SA, Colleoni M, et al. Ribociclib plus endocrine therapy for premenopausal women with hormone-receptor-positive, advanced breast cancer (MONALEESA-7): a randomised Phase 3 trial. Lancet Oncol. 2018;19(7):904–915. doi: 10.1016/S1470-2045(18)30292-4 [DOI] [PubMed] [Google Scholar]
- 4.Finn RS, Rugo HS, Gelmon KA, et al. Long-term pooled safety analysis of palbociclib in combination with endocrine therapy for hormone receptor-positive/human epidermal growth factor receptor 2-negative advanced breast cancer: updated analysis with up to 5 years of follow-up. Oncologist. 2021;26(5):e749–e755. doi: 10.1002/onco.13684 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Diéras V, Rugo HS, Schnell P, et al. Long-term pooled safety analysis of palbociclib in combination with endocrine therapy for HR+/HER2− advanced breast cancer. J Natl Cancer Inst. 2019;111(4):419–430. doi: 10.1093/jnci/djy109 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Finn RS, Martin M, Rugo HS, et al. Palbociclib and letrozole in advanced breast cancer. N Engl J Med. 2016;375(20):1925–1936. doi: 10.1056/NEJMoa1607303 [DOI] [PubMed] [Google Scholar]
- 7.Maurer B, Brandstoetter T, Kollmann S, Sexl V, Prchal-Murphy M. Inducible deletion of CDK4 and CDK6 – deciphering CDK4/6 inhibitor effects in the hematopoietic system. Haematologica. 2021;106(10):2624–2632. doi: 10.3324/haematol.2020.256313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.He S, Roberts PJ, Sorrentino JA, et al. Transient CDK4/6 inhibition protects hematopoietic stem cells from chemotherapy-induced exhaustion. Sci Transl Med. 2017;9(387):eaal3986. doi: 10.1126/scitranslmed.aal3986 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.McAndrew NP, Dickson MA, Clark AS, et al. Early treatment-related neutropenia predicts response to palbociclib. Br J Cancer. 2020;123(6):912–918. doi: 10.1038/s41416-020-0967-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Low JL, Lim E, Bharwani L, et al. Real-world outcomes from use of CDK4/6 inhibitors in the management of advanced/metastatic breast cancer in Asia. Ther Adv Med Oncol. 2022;14:17588359221139678. doi: 10.1177/17588359221139678 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gullick G, Owen CN, Watkins WJ, et al. UK multicentre real-world data of the use of cyclin-dependent kinase 4/6 inhibitors in metastatic breast cancer. ESMO Real World Data Digit Oncol. 2024;5:100064. doi: 10.1016/j.esmorw.2024.100064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rugo HS, Layman RM, Lynce F, et al. Comparative overall survival of CDK4/6 inhibitors plus an aromatase inhibitor in HR+/HER2− metastatic breast cancer in the US real-world setting. ESMO Open. 2025;10(1):104103. doi: 10.1016/j.esmoop.2024.104103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Patel R, Mathews J, Hamm C, et al. Real-world experience with CDK4/6 inhibitors in the first-line palliative setting for HR+/HER2− advanced breast cancer. Curr Oncol. 2025;32(1):52. doi: 10.3390/curroncol32010052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lin W, Zeng Y, Weng L, Yang J, Zhuang W. Comparative analysis of adverse events associated with CDK4/6 inhibitors based on FDA’s adverse event reporting system: a case control pharmacovigilance study. BMC Pharmacol Toxicol. 2024;25(1):47. doi: 10.1186/s40360-024-00770-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Li X, Song Y, Lv M, et al. A real-world study of palbociclib plus endocrine therapy with or without a short course chemotherapy in the first-line treatment of HR-positive HER2-negative metastatic breast cancer. Front Oncol. 2025;15:1512496. doi: 10.3389/fonc.2025.1512496 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Buijs SM, Mohmaed Ali MI, Oomen-de Hoop E, et al. Palbociclib exposure in relation to efficacy and toxicity in patients with advanced breast cancer. ESMO Open. 2025;10(3):104290. doi: 10.1016/j.esmoop.2025.104290 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kahraman S, Erul E, Gumusay O, et al. Real-world treatment efficacy of ribociclib or palbociclib plus fulvestrant in hormone receptor-positive/HER2-negative metastatic breast cancer: Turkish Oncology Group (TOG) study. Clin Breast Cancer. 2025;25(5):e635–e644. doi: 10.1016/j.clbc.2025.03.003 [DOI] [PubMed] [Google Scholar]
- 18.Bjerrum A, Henriksen AF, Knop AS, et al. 391P Detrimental effect on overall survival of CDK4/6 inhibitor dose reduction if immortal time bias is considered. Ann Oncol. 2023;34:S344–S345. doi: 10.1016/j.annonc.2023.09.568 [DOI] [Google Scholar]
- 19.Anampa J, Haque T, Murakhovskaya I, et al. Macrocytosis and dysplastic anemia is associated with the cyclin-dependent kinase 4/6 inhibitor palbociclib in metastatic breast cancer. Haematologica. 2018;103(3):e98–e102. doi: 10.3324/haematol.2017.181941 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Inci B K, Kubilay Tolunay P, Öztekin Ş, et al. Monitoring the response of cyclin-dependent kinase 4/6 inhibitors with mean corpuscular volume. Curr Oncol. 2024;31(10):5722–5729. doi: 10.3390/curroncol31100424 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Barroso T, Costa L, Gonçalves L, et al. CDK4/6 inhibitors-induced macrocytosis is not associated with hemolysis and does not impact hemoglobin homeostasis. Cancers. 2025;17(9):1567. doi: 10.3390/cancers17091567 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Goetz MP, Toi M, Campone M, et al. MONARCH 3: abemaciclib as initial therapy for advanced breast cancer. J Clin Oncol. 2017;35(32):3638–3646. doi: 10.1200/JCO.2017.75.6155 [DOI] [PubMed] [Google Scholar]
- 23.National Cancer Institute. Common Terminology Criteria for Adverse Events (CTCAE) Version 5.0. Bethesda, MD: National Cancer Institute; 2017. [Google Scholar]
- 24.Gültürk I, Colak R, Kapar C, Guliyev M, Yıldırım C, Yilmaz M. Could macrocytosis predict survival in advanced breast cancer patients that were treated with CDK 4–6 inhibitors? Breast. 2024;78:103820. doi: 10.1016/j.breast.2024.103820 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chua PX, Soh SH, Ng QX, et al. What matters most? Understanding patient preferences in breast cancer treatment: a systematic review of discrete choice experiments. Am J Surg. 2025;250:116642. doi: 10.1016/j.amjsurg.2025.116642 [DOI] [PubMed] [Google Scholar]
- 26.Bidard F-C, Mayer EL, Park YH, et al. First-line camizestrant for emerging ESR1-mutated advanced breast cancer. N Engl J Med. 2025;393(6):569–580. doi: 10.1056/NEJMoa2502929 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and/or analyzed during the current study are not publicly available because they contain potentially identifiable patient information but are available from the corresponding author on reasonable request.