Clinical benefit of palbociclib retreatment after abemaciclib exposure in hormone receptor positive, HER2 negative metastatic breast cancer Authors: Muhan Yu, Michiko Yamashita, Mamoru Takada, Daisuke Takabatake, Mina Takahashi, Mariko Kochi, Maya Uno, Midori Kondo, Takeshi Nagashima, Hiroshi Fujimoto, Junta Sakakibara, Ryotaro Teranaka, Kenjiro Aogi, Masayuki Otsuka Published: 2026 ABSTRACT Cyclin-dependent kinase 4/6 inhibitors (CDK4/6is) combined with endocrine therapy are standard first-line treatment for HR + /HER2– metastatic breast cancer (MBC). However, disease progression and treatment-limiting adverse events (AEs) remain major challenges, and the clinical benefit of CDK4/6i retreatment after prior exposure is uncertain. We retrospectively analyzed 78 patients with HR + /HER2 − MBC treated at two institutions (2018–2025) who received at least two CDK4/6i-containing therapies. Patients were classified according to treatment sequence: abemaciclib → abemaciclib (cohort 1), abemaciclib → palbociclib (cohort 2), or palbociclib → abemaciclib (cohort 3). Median time to treatment failure (TTF) of CDK4/6i retrial was 6, 10, and 6 months across the three cohorts, respectively. Cohort 2 showed the most favorable retrial outcomes, with 64.5% of patients achieving a longer TTF on retrial than on initial CDK4/6i therapy. Discontinuation due to AEs was associated with a trend toward longer TTF compared with progression, whereas visceral metastasis was associated with shorter TTF in cohort 2 (hazard ratio 12.03, p = 0.01). CDK4/6i retrial may provide clinically meaningful benefit for selected patients, particularly with palbociclib after abemaciclib exposure or discontinuing initial therapy due to AEs, whole visceral metastasis predicts poorer outcomes with retreatment. These findings support individualized patient selection and warrant prospective validation to optimize sequencing strategies for CDK4/6is.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-58952-7. Introduction There is a significant degree of molecular heterogeneity in breast cancer, and treatment options and prognosis have typically been guided by immunohistochemistry (IHC) markers such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth receptor 2 (HER2). Approximately 60–65% of all breast cancers are ER-positive HER2-negative breast cancers, making them the most prevalent subtype of breast cancer1. This subtype is characterized by a risk of both local and distant recurrence, with about 20–30% of patients developing metastatic breast cancer (MBC)2. Hormone receptor-positive, human epidermal growth receptor 2-negative metastatic breast cancer (HR + /HER2– MBC) is currently incurable, with only 24% of patients surviving more than 5 years3. Cyclin-dependent kinase 4/6 inhibitors (CDK4/6is) combined with endocrine therapy (ET) have been authorized by the US Food and Drug Administration (FDA) for use as first-line and second-line treatments for HR + /HER2- MBC since 20154, this combination has been shown to significantly improve both progression-free survival (PFS) and overall survival (OS) in contrast to ET single treatment5–7. Despite these advancements, nearly all patients eventually acquire resistance to ET and CDK4/6is, and the prognosis remained limited, with 5-year breast cancer-specific survival around 35% in recent years post-20158–10. Following progression on first-line CDK4/6i treatment, several treatment options are available. Guidelines by the National Comprehensive Cancer Network (NCCN) and the European Society for Medical Oncology (ESMO) recommend various options for the treatment in the second-line setting. These include: fulvestrant monotherapy; mammalian target of rapamycin inhibitors (mTORi; e.g., everolimus) ± ET; PI3K inhibitor (PI3Ki; e.g., alpelisib) + fulvestrant (if phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha mutation positive [PIK3CA-m +]); AKT inhibitor (AKTi; e.g., capivasertib) + fulvestrant; poly-ADP ribose polymerase inhibitors (PARPi) including olaparib or talazoparib (if breast cancer gene/partner and localizer of BRCA2 positive [BRCA/PALB2m +]); and chemotherapy (in cases when a life-threatening disease is present)11,12. According to the guidelines by the Japanese Breast Cancer Society (JBCS), unused endocrine therapies (including combination therapy with mTORi) may be considered. Data are also being reported on the use of oral selective estrogen receptor degraders (SERDs) and CDK4/6is13. However, their efficacy and tolerability tend to decrease with each subsequent line of therapy, and there is no established treatment sequencing after disease progression with CDK4/6i14. Rechallenging CDK4/6i also offers additional avenues for molecularly targeted care, while a key question is whether patients might benefit from continuing ET + CDK4/6i after initial progression. This question is particularly relevant as CDK4/6is are increasingly used in the adjuvant setting15. Several phase II clinical trials were conducted to explore this issue. The PACE and PALMIRA trials found no PFS improvement when comparing palbociclib + ET to ET alone in patients with HR + /HER2- MBC who had previously progressed on an aromatase inhibitor (AI) + a CDK4/6i (any CDK4/6i in PACE and palbociclib specifically in PALMIRA)16,17. The MAINTAIN trial evaluated ribociclib + ET versus ET alone in patients previously treated with any CDK4/6i (mostly with palbociclib) and found significant PFS improvement with ribociclib + ET 18. In postMONARCH, a phase 3 clinical trial evaluating the efficacy and safety of abemaciclib plus fulvestrant versus placebo plus fulvestrant in patients with HR + /HER2- advanced breast cancer who have progressed on a prior CDK4/6i and ET, abemaciclib plus fulvestrant significantly improved PFS compared to placebo plus fulvestrant19. These trials highlight the ongoing uncertainty regarding the optimal approach to retreatment with CDK4/6i in HR + /HER2- MBC. Based on the results of these trials, the question of whether patients can benefit from CDK4/6i retreatment after the progression of initial CDK4/6i exposure remains unresolved in real-world clinical settings. Thus, it is essential to determine the efficacy of CDK4/6i retrial, including identifying the patients who can benefit most from CDK4/6i retrial, to help guide clinical decision-making in the real world. This in turn has the potential to assist in making treatment decisions for patients who relapse after using CDK4/6is in an adjuvant setting15. To further evaluate the association between the duration of CDK4/6i retrial and patient characteristics or tumor features, we conducted a retrospective clinical analysis. The analysis focused on patients treated at Chiba University Hospital or National Hospital Organization Shikoku Cancer Center who had received at least two lines of therapy containing CDK4/6is. Our study cohorts included patients who were retreated with the same CDK4/6i and those who were treated with a different CDK4/6i. We report the real-world efficacy of CDK4/6i retrial and examine the relationship between the TTF of CDK4/6i retrial and other clinical variables in our patient population. Our study aims to identify which patients are most likely to benefit from CDK4/6i retreatment after progression on first-line ET + CDK4/6i in HR + /HER2- MBC in a real-world clinical setting. Methods Patients Eligible patients had biopsy-confirmed unresectable stage III or stage IV HR + /HER2- breast cancer, were treated at Chiba University Hospital or National Hospital Organization Shikoku Cancer Center, and received two or more lines of treatment for advanced disease, with at least two prior lines containing a CDK4/6i. Furthermore, to analyze the treatment sequencing of patients who ever received CDK4/6i therapy, patients who received only one line containing a CDK4/6i were also reviewed. Patients with initial HR + /HER2 + breast cancer were excluded unless their disease reverted to a HER2-negative state by the time of CDK4/6i exposure. The Institutional Review Board of Chiba University Hospital approved this study on February 24, 2023 (Approval No. HK202302-04). As Shikoku Cancer Center participated under the centralized ethical approval obtained by Chiba University Hospital, the same approval number was applied at both institutions. The study was not associated with a clinical trial in Japan. This study complied with all relevant ethical regulations regarding patient data and was conducted in accordance with the ethical principles of the Declaration of Helsinki. Study design We performed a retrospective analysis of patients treated from April 2018 to June 2025 with at least two separate treatment lines containing a CDK4/6i for advanced HR + /HER2- breast cancer. Efficacy outcomes were evaluated in three different patient cohorts. Cohort 1 (AA) patients (n = 5) underwent initial abemaciclib treatment and abemaciclib retrial. Cohort 2 (AP) patients (n = 31) received initial abemaciclib treatment and changed to palbociclib retrial. Meanwhile, cohort 3 (PA) patients (n = 42) underwent initial palbociclib treatment and abemaciclib retrial. Ribociclib was not included in this analysis, as it was not approved for clinical use in Japan during the study period. Palbociclib was approved approximately one year earlier than abemaciclib and was the only available CDK4/6 inhibitor during the initial approval phase. Consequently, patients treated earlier in the study period predominantly received palbociclib as the initial agent. No participating institution had formulary restrictions or administrative policies influencing CDK4/6 inhibitor selection; treatment decisions were made at the discretion of the treating physician based on drug availability and contemporaneous clinical evidence. Moreover, to analyze the post-treatment after CDK4/6i failure, patients treated from April 2018 to March 2025 who received separate treatment lines containing a CDK4/6i for advanced HR + /HER2- breast cancer in Chiba University Hospital (n = 42) and patients who ever received CDK4/6i therapy for advanced HR + /HER2- breast cancer in Chiba University Hospital (n = 102) were also analyzed. Outcomes The primary objective of this study was to evaluate OS and TTF on CDK4/6i retreatment in the 3 different pre-specified cohorts. OS was defined as the time in months from when a patient started first CDK4/6i exposure to death. TTF was selected as the primary endpoint to reflect real-world clinical practice. Unlike PFS, TTF captures treatment discontinuation for any reason, including disease progression, adverse events, patient preference, or physician decision-making. TTF of initial CDK4/6i treatment was defined as the time in months from when a patient began first CDK4/6i exposure to discontinuation of initial CDK4/6i for any reason. TTF of CDK4/6i retrial was defined as the time in months from when a patient began CDK4/6i retreatment to discontinuation of CDK4/6i retreatment for any reason. To better understand potential associations between certain clinical variables and response to CDK4/6i retrial, we included the following variables in our analysis: age, menopausal status, presence of bone-only disease, presence of visceral disease, de novo metastatic disease, and time to progression on CDK4/6i treatment. Statistical analysis OS and TTF were estimated using the Kaplan–Meier method. Survival distributions were compared using the log-rank test., and corresponding median values with 95% confidence intervals were calculated. Swimmer plots were generated to visualize individual treatment durations. Associations between clinical variables and TTF following CDK4/6i rechallenge were evaluated using Cox proportional hazards regression models. All statistical analyses were performed using R software (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria) and RStudio (version 2026.01.0 + 392; Posit Software, PBC, Boston, MA, USA). A two-sided P value < 0.05 was considered statistically significant. Results Characteristics of patients A total of 78 HR + /HER2- patients with MBC treated at Chiba University Hospital (n = 42) or National Hospital Organization Shikoku Cancer Center (n = 36) with at least 2 separate treatment regimens containing a CDK4/6i from April 2018 to June 2025 were identified. The median age for all patients identified was 66. Patients were divided into 3 cohorts. Cohort 1 patients (n = 5) underwent initial abemaciclib treatment and abemaciclib retrial. Cohort 2 (n = 31) patients received initial abemaciclib treatment and were changed to palbociclib retrial. Meanwhile, cohort 3 patients (n = 42) underwent initial palbociclib treatment and abemaciclib retrial. The clinical characteristics of all patients in all cohorts are summarized in Table 1. All 78 patients were treated for metastatic disease by the time of first CDK4/6i exposure, and no patients were treated for adjuvant disease.Table 1Clinical characteristics of all patients in all cohorts.Cohort 1 (n = 5)Cohort 2 (n = 31)Cohort 3 (n = 42)First CDK4/6i exposure (Abemaciclib)CDK4/6i retrial (Abemaciclib)First CDK4/6i exposure (Abemaciclib)CDK4/6i retrial (Palbociclib)First CDK4/6i exposure (Palbociclib)CDK4/6i retrial (Abemaciclib)Median age58 (49–80)67 (37–85)62.5 (34–83)Bone-only metastases07 (22.6%)13 (31.0%)Visceral disease3 (60.0%)16 (51.6%)18 (42.9%)De novo metastatic disease1 (20.0%)8 (25.8%)17 (40.5%)Endocrine partnerAromatase inhibitor1 (20.0%)2 (40.0%)12 (38.7%)14 (45.2%)18 (42.9%)13 (31.0%)SERD (fulvestrant)3 (60.0%)2 (40.0%)19 (61.3%)17 (54.8%)23 (54.8%)30 (71.4%)Tamoxifen1 (20.0%)1 (20.0%)002 (4.8%)0Median lines of therapy for metastatic disease1 (1–15)3 (2–17)1 (1–6)3 (2–10)1 (1–10)3 (2–11)First line4 (80.0%)017 (54.8%)022 (52.4%)0Second line02 (40.0%)6 (19.4%)14 (45.2%)10 (23.8%)19 (45.2%)Third Line01 (20.0%)4 (12.9%)9 (29.0%)5 (11.9%)10 (23.8%)Fourth and beyond1 (20.0%)2 (40.0%)4 (12.9%)8 (25.8%)5 (11.9%)13 (31.0%)Clinical characteristics are shown by cohort. Data are presented as median (range) or number (%), as appropriate. Endocrine partner refers to the endocrine therapy administered in combination with CDK4/6is. Lines of therapy indicate the number of prior systemic treatments for metastatic disease at the time of CDK4/6i initiation. Table 1Clinical characteristics of all patients in all cohorts.Cohort 1 (n = 5)Cohort 2 (n = 31)Cohort 3 (n = 42)First CDK4/6i exposure (Abemaciclib)CDK4/6i retrial (Abemaciclib)First CDK4/6i exposure (Abemaciclib)CDK4/6i retrial (Palbociclib)First CDK4/6i exposure (Palbociclib)CDK4/6i retrial (Abemaciclib)Median age58 (49–80)67 (37–85)62.5 (34–83)Bone-only metastases07 (22.6%)13 (31.0%)Visceral disease3 (60.0%)16 (51.6%)18 (42.9%)De novo metastatic disease1 (20.0%)8 (25.8%)17 (40.5%)Endocrine partnerAromatase inhibitor1 (20.0%)2 (40.0%)12 (38.7%)14 (45.2%)18 (42.9%)13 (31.0%)SERD (fulvestrant)3 (60.0%)2 (40.0%)19 (61.3%)17 (54.8%)23 (54.8%)30 (71.4%)Tamoxifen1 (20.0%)1 (20.0%)002 (4.8%)0Median lines of therapy for metastatic disease1 (1–15)3 (2–17)1 (1–6)3 (2–10)1 (1–10)3 (2–11)First line4 (80.0%)017 (54.8%)022 (52.4%)0Second line02 (40.0%)6 (19.4%)14 (45.2%)10 (23.8%)19 (45.2%)Third Line01 (20.0%)4 (12.9%)9 (29.0%)5 (11.9%)10 (23.8%)Fourth and beyond1 (20.0%)2 (40.0%)4 (12.9%)8 (25.8%)5 (11.9%)13 (31.0%)Clinical characteristics are shown by cohort. Data are presented as median (range) or number (%), as appropriate. Endocrine partner refers to the endocrine therapy administered in combination with CDK4/6is. Lines of therapy indicate the number of prior systemic treatments for metastatic disease at the time of CDK4/6i initiation. Clinical characteristics of all patients in all cohorts. Clinical characteristics are shown by cohort. Data are presented as median (range) or number (%), as appropriate. Endocrine partner refers to the endocrine therapy administered in combination with CDK4/6is. Lines of therapy indicate the number of prior systemic treatments for metastatic disease at the time of CDK4/6i initiation. In cohort 1, none of the patients had bone-only diseases, whereas 60.0% had visceral diseases. Across all patients in the cohort, the median number of therapy lines for metastatic disease of CDK4/6i retrial was 3, and 40.0% of patients in this cohort received CDK4/6is as the immediately preceding therapy before retrial. In cohort 2, 22.6% of patients had bone-only diseases, whereas 51.6% had visceral diseases. The median number of treatment lines of CDK4/6i retrial was 3. In cohort 3, 31.0% of patients had bone-only diseases and 42.9% had visceral diseases. The median number of treatment lines of retrial was 3. SERDs (fulvestrant) were the most common endocrine partner at re-treatment over all cohorts. Clinical outcomes for patients receiving CDK4/6i retreatment after progression on a CDK4/6i At the time of data cutoff (June. 30, 2025), most patients had experienced failure on CDK4/6i retreatment (n = 55; 70.5%), a minority of patients (n = 17; 21.8%) discontinued CDK4/6i retrial because of toxicity without disease progression included, 1 patient discontinued CDK4/6i retrial because of shrinkage of tumor that could be surgically removed, and a subset remained on treatment (n = 22; 28.2%). The median OS on CDK4/6i treatment was 62 months (Fig. 1a). The median OS for cohort 1 patients was 37 months (Supplementary Fig. S1). The median OS for cohort 2 and 3 patients was 35 and 62 months, respectively (Fig. 1b, c). The Kaplan–Meier curves for OS demonstrated no statistically significant difference between cohort 2 and 3 patients (p = 0.66). The estimated survival probabilities remained similar over time, suggesting that treatment sequence was not associated with overall survival.Fig. 1Overall survival (OS) for patients receiving cyclin-dependent kinase 4/6 inhibitors (CDK4/6is) after first CDK4/6i exposure. a OS for all patients receiving CDK4/6i after first CDK4/6i exposure. b OS for patients in cohort 2. c OS for patients in cohort 3. d OS for patients in cohort 2 (dark blue) and cohort 3 (light blue). No significant difference in OS was observed between the two groups (log-rank p = 0.66). Overall survival (OS) for patients receiving cyclin-dependent kinase 4/6 inhibitors (CDK4/6is) after first CDK4/6i exposure. a OS for all patients receiving CDK4/6i after first CDK4/6i exposure. b OS for patients in cohort 2. c OS for patients in cohort 3. d OS for patients in cohort 2 (dark blue) and cohort 3 (light blue). No significant difference in OS was observed between the two groups (log-rank p = 0.66). Time to treatment failure with CDK4/6i retrial Kaplan–Meier curves summarizing median TTF of both initial CDK4/6i treatment and CDK4/6i retrial are organized per cohort in Fig. 2 and Supplementary Fig. S2. Median TTF for CDK4/6i retrial compared to initial CDK4/6i exposure in cohort 1 was 6 months vs 13 months, in cohort 2 was 10 months vs 3 months, and in cohort 3 was 6 months vs 12 months.Fig. 2Kaplan–Meier curves of time to treatment failure (TTF) for CDK4/6i retrial the table above shows the Median TTF for the first CDK4/6i exposure and CDK4/6i retrial. Patients were divided into those who discontinued the initial CDK4/6i due to adverse events (AEs) and those who discontinued due to progressive disease (PD). a TTF for abemaciclib retrial in cohort 2, b TTF for palbociclib retrial in cohort 3, c Patients who discontinued the initial CDK4/6i due to AEs vs. PD in cohort 2, d Patients who discontinued the initial CDK4/6i due to AEs vs. PD in cohort 3, e Patients who discontinued the initial CDK4/6i due to AEs in cohort 2 vs. cohort 3, f Patients who discontinued the initial CDK4/6i due to PD in cohort 2 vs. cohort 3. Kaplan–Meier curves of time to treatment failure (TTF) for CDK4/6i retrial the table above shows the Median TTF for the first CDK4/6i exposure and CDK4/6i retrial. Patients were divided into those who discontinued the initial CDK4/6i due to adverse events (AEs) and those who discontinued due to progressive disease (PD). a TTF for abemaciclib retrial in cohort 2, b TTF for palbociclib retrial in cohort 3, c Patients who discontinued the initial CDK4/6i due to AEs vs. PD in cohort 2, d Patients who discontinued the initial CDK4/6i due to AEs vs. PD in cohort 3, e Patients who discontinued the initial CDK4/6i due to AEs in cohort 2 vs. cohort 3, f Patients who discontinued the initial CDK4/6i due to PD in cohort 2 vs. cohort 3. The TTF of CDK4/6i retrial was better in cohort 2 than in cohort 3 (Fig. 2a, b). We hypothesized that this difference might be attributed to the higher proportion of patients in cohort 2 who had discontinued the initial CDK4/6i treatment due to adverse events (AEs). To explore this hypothesis, we analyzed the TTF of CDK4/6i retrial after categorizing patients in both cohorts into two groups based on the reason for discontinuation of the initial treatment: the AE group and the progressive disease (PD) group (Fig. 2c, d). In both cohorts, patients who had discontinued the initial CDK4/6i treatment due to AEs demonstrated numerically longer TTF upon CDK4/6i retrial compared with those who had discontinued due to PD. Although these differences did not reach statistical significance (cohort 2: p = 0.098; cohort 3: p = 0.10), a consistent numerical trend was observed across both cohorts, suggesting that patients who discontinued initial CDK4/6i for reasons other than disease progression may retain sensitivity to CDK4/6 inhibition upon rechallenge. This observation should be interpreted cautiously and requires validation in larger, independent cohorts to determine whether the reason for initial discontinuation independently influences retreatment outcomes. Therefore, we further compared the TTF of CDK4/6i retrial between cohort 2 and cohort 3 after stratifying patients by the reason for discontinuation of the initial CDK4/6i treatment. Among patients who discontinued due to AEs (Fig. 2e), those in cohort 2 demonstrated a numerically longer TTF on CDK4/6i retrial than those in cohort 3. No statistically significant difference was observed (p = 0.50), although the Kaplan–Meier curves showed separation favoring cohort 2. Similarly, among patients who discontinued the initial CDK4/6i treatment due to PD (Fig. 2f), no significant difference in TTF was observed between cohort 2 and cohort 3 (p = 0.81). The survival curves of the PD groups overlapped throughout the observation period, indicating comparable clinical outcomes upon rechallenge. Taken together, these findings raise the possibility that the longer TTF observed in cohort 2 might be influenced, at least in part, by the higher proportion of patients who discontinued their initial CDK4/6i for reasons other than disease progression. Accordingly, these findings should be considered exploratory and hypothesis-generating. In addition, the cases of initial CDK4/6i discontinued due to AEs in cohort 2 (n = 27) and cohort 3 (n = 7) are shown in Tables 2 and 3, respectively. In cohort 2, the most common AE were gastrointestinal (n = 8, 29.6%) and hepatic (n = 7, 25.9%) events, followed by dermatologic (n = 5, 18.5%), respiratory (cough, n = 4, 14.8%; interstitial lung disease [ILD], n = 4, 14.8%), hematologic (n = 2, 7.4%), and renal (n = 1, 3.7%) events. One patient (3.7%) discontinued due to ILD associated with COVID-19. Median TTF varied by AE type, ranging from 3 months (hematologic) to 17 months (respiratory ILD). In cohort 3, hematologic AEs were predominant (n = 6, 85.7%), while dermatologic AEs accounted for 14.3% (n = 1). Median TTF was 11 months for hematologic events (range, 1–69 months) and less than 1 month for dermatologic events.Table 2Cases discontinued the initial CDK4/6i due to AEs in cohort 2 (n = 27).Number of cases (%)Median TTF of CDK4/6i retrial (months)Gastrointestinal8 (29.6)11 (1–33)Dermatologic5 (18.5)16 (1–25)Hematologic2 (7.4)3 (1–4)Hepatic7 (25.9)7 (1–32)Renal1 (3.7)12Respiratory (Cough)4 (14.8)15 (3–29)Respiratory (Interstitial lung disease)4 (14.8)17 (5–26)ILD by Covid 191 (3.7)6This table summarizes AEs leading to discontinuation of the initial CDK4/6i in cohort 2 (n = 27) and the corresponding TTF following CDK4/6i retrial. Data are presented as number (%) and median TTF in months (range).AE categories were not mutually exclusive. Patients with overlapping toxicities were included in all relevant categories; therefore, totals may exceed 100%Table 3Cases discontinued the initial CDK4/6i due to AEs in cohort 3 (n = 7).Number of casesMedian TTF of CDK4/6i retrial (months)Dermatologic1 (14.3%)1Hematologic6 (85.7%)11 (1–69)This table summarizes AEs leading to discontinuation of the initial CDK4/6i in cohort 3 (n = 7) and the corresponding TTF following CDK4/6i retrial. Data are presented as number (%) and median TTF in months (range).AE categories were not mutually exclusive. Patients with overlapping toxicities were included in all relevant categories; therefore, totals may exceed 100% Table 2Cases discontinued the initial CDK4/6i due to AEs in cohort 2 (n = 27).Number of cases (%)Median TTF of CDK4/6i retrial (months)Gastrointestinal8 (29.6)11 (1–33)Dermatologic5 (18.5)16 (1–25)Hematologic2 (7.4)3 (1–4)Hepatic7 (25.9)7 (1–32)Renal1 (3.7)12Respiratory (Cough)4 (14.8)15 (3–29)Respiratory (Interstitial lung disease)4 (14.8)17 (5–26)ILD by Covid 191 (3.7)6This table summarizes AEs leading to discontinuation of the initial CDK4/6i in cohort 2 (n = 27) and the corresponding TTF following CDK4/6i retrial. Data are presented as number (%) and median TTF in months (range).AE categories were not mutually exclusive. Patients with overlapping toxicities were included in all relevant categories; therefore, totals may exceed 100% Cases discontinued the initial CDK4/6i due to AEs in cohort 2 (n = 27). This table summarizes AEs leading to discontinuation of the initial CDK4/6i in cohort 2 (n = 27) and the corresponding TTF following CDK4/6i retrial. Data are presented as number (%) and median TTF in months (range). AE categories were not mutually exclusive. Patients with overlapping toxicities were included in all relevant categories; therefore, totals may exceed 100% Table 3Cases discontinued the initial CDK4/6i due to AEs in cohort 3 (n = 7).Number of casesMedian TTF of CDK4/6i retrial (months)Dermatologic1 (14.3%)1Hematologic6 (85.7%)11 (1–69)This table summarizes AEs leading to discontinuation of the initial CDK4/6i in cohort 3 (n = 7) and the corresponding TTF following CDK4/6i retrial. Data are presented as number (%) and median TTF in months (range).AE categories were not mutually exclusive. Patients with overlapping toxicities were included in all relevant categories; therefore, totals may exceed 100% Cases discontinued the initial CDK4/6i due to AEs in cohort 3 (n = 7). This table summarizes AEs leading to discontinuation of the initial CDK4/6i in cohort 3 (n = 7) and the corresponding TTF following CDK4/6i retrial. Data are presented as number (%) and median TTF in months (range). AE categories were not mutually exclusive. Patients with overlapping toxicities were included in all relevant categories; therefore, totals may exceed 100% Swimmer plots comparing individual TTF for both initial CDK4/6i treatment and CDK4/6i retrial side-by-side per patient are illustrated in Fig. 3 and Supplementary Fig. S3. In the swimmer plot, navy bars on the left indicate patients who discontinued the initial CDK4/6 inhibitor due to AEs, whereas light blue bars indicate those who discontinued due to PD.Fig. 3TTF at first CDK4/6i exposure vs. retrial. The figure shows a two-headed swimmer plot for patients receiving abemaciclib after prior progression on initial abemaciclib exposure. For each patient, both the TTF for initial CDK4/6i exposure (pointing leftward) and for CDK4/6i retrial (pointing rightward) are shown side-by-side. The TTFs for retrial color-coded depend upon the treatment line for metastatic disease corresponding to CDK4/6i retrial. TTF2 (2–3) = 2nd or 3rd line; TTF2 (4–5) = 4th or 5th line; TTF2 (> 5) = 6th line and beyond. Navy bars: Patients who discontinued the initial CDK4/6i due to AEs. Light blue bars: Patients who discontinued due to PD. a Swimmer plot for cohort 2, b Swimmer plot for cohort 3. TTF at first CDK4/6i exposure vs. retrial. The figure shows a two-headed swimmer plot for patients receiving abemaciclib after prior progression on initial abemaciclib exposure. For each patient, both the TTF for initial CDK4/6i exposure (pointing leftward) and for CDK4/6i retrial (pointing rightward) are shown side-by-side. The TTFs for retrial color-coded depend upon the treatment line for metastatic disease corresponding to CDK4/6i retrial. TTF2 (2–3) = 2nd or 3rd line; TTF2 (4–5) = 4th or 5th line; TTF2 (> 5) = 6th line and beyond. Navy bars: Patients who discontinued the initial CDK4/6i due to AEs. Light blue bars: Patients who discontinued due to PD. a Swimmer plot for cohort 2, b Swimmer plot for cohort 3. To compare the duration of CDK4/6i retrial to that of initial therapy, we calculated TTF2/TTF1 ratio by dividing TTF of retrial by TTF of initial CDK4/6i. In cohort 1, the median TTF2/TTF1 ratio was 0.2, with only 20.0% of patients having a longer TTF on retrial compared to initial treatment (Supplementary Fig. S3). Cohort 2 had a median TTF2/TTF1 ratio of 2.8, and 64.5% of patients had a longer TTF2 with CDK4/6i retrial compared to initial treatment (Fig. 3a). In this cohort, at the time of data censoring, 15 patients (48.4%) remained on treatment without further progression. One patient discontinued CDK4/6i retrial because of tumor shrinkage of the tumor, which could be removed surgically. Cohort 3 had a median TTF2/TTF1 ratio of 0.4, with 23.8% of patients having a longer CDK4/6i retrial duration than initial treatment (Fig. 3b). At the time of data censoring, 6 patients in cohort 3 (14.3%) remained on treatment without further progression. Association between the duration of CDK4/6i retrial and other clinical variables The correlation between the duration of CDK4/6i retrial and other factors were analyzed. A Cox proportional hazards regression model was conducted to evaluate prognostic factors, including age, menopausal status, bone-only metastasis, presence of visceral metastasis, duration of previous CDK4/6i treatment, or de novo metastases, associated with TTF of CDK4/6i retrial in cohort 2 and cohort 3 (Fig. 4). Accordingly, analyses were confined to univariate models.Fig. 4Correlation between TTF of CDK4/6i retrial and other factors. a Cox proportional hazards regression model of the correlation between prognostic factors and TTF of CDK4/6i retrial in cohort 2, b Kaplan–Meier analysis of TTF of CDK4/6i retrial in patients with visceral metastasis compared to those without in cohort 2, c Cox proportional hazards regression model of the correlation between prognostic factors and TTF of CDK4/6i retrial in cohort 3, d Kaplan–Meier analysis of TTF of CDK4/6i retrial in patients with de novo disease compared to those without in cohort 3. Correlation between TTF of CDK4/6i retrial and other factors. a Cox proportional hazards regression model of the correlation between prognostic factors and TTF of CDK4/6i retrial in cohort 2, b Kaplan–Meier analysis of TTF of CDK4/6i retrial in patients with visceral metastasis compared to those without in cohort 2, c Cox proportional hazards regression model of the correlation between prognostic factors and TTF of CDK4/6i retrial in cohort 3, d Kaplan–Meier analysis of TTF of CDK4/6i retrial in patients with de novo disease compared to those without in cohort 3. In cohort 2, in the univariate Cox proportional hazards analysis, visceral metastasis was significantly associated with shorter TTF (hazard ratio = 12.03, 95% CI 1.82–79.8, p = 0.01), indicating a potential adverse prognostic association (Fig. 4a). The width of the confidence interval further underscores the uncertainty surrounding this estimate. Age (hazard ratio = 1.13, 95% CI 0.999–1.30, p = 0.052) and bone-only metastasis (hazard ratio = 15.47, 95% CI 0.93–256.2, p = 0.056) showed borderline associations with shorter TTF, whereas other covariates, including menopausal status, duration of previous CDK4/6i treatment, and de novo metastasis, were not statistically significant (Fig. 4a). Kaplan–Meier analysis further demonstrated a significantly shorter TTF in patients with visceral metastasis compared to those without (p = 0.026) (Fig. 4b). Taken together, these findings indicate a potential association between visceral metastasis and poorer retreatment outcomes in cohort 2. In light of the limited sample size and exploratory analyses, this observation requires validation in larger, independent cohorts. In cohort 3, none of the examined variables showed a statistically significant association with TTF (Fig. 4c). Although patients with de novo metastatic breast cancer demonstrated a numerically shorter TTF (hazard ratio 1.78, 95% CI 0.87–3.61, p = 0.113), this result was not statistically significant (Fig. 4c-d). Treatment sequencing after initial CDK4/6i The treatment sequencing of patients treated with CDK4/6i in Chiba University Hospital is shown in Supplementary Fig. S4a. The innermost to outermost rings show initial CDK4/6i treatment (light green ring for CDK4/6i, green ring for abemaciclib, and blue ring for palbociclib) and treatment patterns after initial CDK4/6i treatment (subsequent rings). In patients who did not undergo CDK4/6i retrial (n = 102), once they started chemotherapy after CDK4/6is failed, their treatment was chemotherapy-centered and they were (or are) unlikely to return to hormone therapy (Supplementary Fig. S4a). However, in patients who underwent CDK4/6i retrial (n = 42), some still could return to hormone therapy after chemotherapy (Supplementary Fig. S4b). Chemotherapy becomes the primary treatment after CDK4/6i failure, limiting the potential for returning to hormone therapy. Discussion The analysis of TTF for both initial CDK4/6i treatment and CDK4/6i retrial across the three cohorts reveals significant variability in patient responses, demonstrating the complexity of retrial efficacy. TTF was selected as the primary endpoint given the retrospective, real-world design of the study. In routine clinical practice, treatment discontinuation may occur due to factors beyond radiologic progression, including toxicity and clinical judgment. Accordingly, TTF was considered an appropriate measure of treatment durability in this setting. Moreover, our results suggested that TTF of CDK4/6i retrial correlates with specific patient characteristics, providing insights into patient subgroups that derive clinical benefit more or less from retreatment strategies. In addition, due to the limited number of cases in the palbociclib retrial followed by the initial palbociclib treatment, no cases were selected for analysis in our study. In cohort 1, the median TTF for CDK4/6i retrial was substantially shorter than that of the initial treatment (6 months vs. 13 months), and only 20.0% of patients in this cohort experienced a longer TTF during retrial compared to their initial treatment. These findings are consistent with acquired resistance contributing to reduced efficacy of abemaciclib retrial in this cohort. However, the results of phase Ⅱ AGAIN Study (WJOG14220B) suggest that continuing abemaciclib after progression on an initial abemaciclib treatment may provide clinical benefit19,20. This inconsistency may be attributable to the limited sample size. In contrast, in cohort 2, a more favorable response to CDK4/6i retrial was demonstrated, with a median TTF of 10 months compared to 3 months for initial treatment, 64.5% of patients in this cohort had a longer TTF during retrial (Fig. 2). This suggests that palbociclib retreatment may be effective after abemaciclib failure, and a subset of patients in cohort 2 may derive substantial benefit from CDK4/6i retrial. This finding is noteworthy because abemaciclib is generally considered the more potent CDK4/6 inhibitor and has demonstrated substantial efficacy 21–24. Therefore, the favorable outcomes observed with palbociclib retreatment in cohort 2 are unlikely to be explained simply by differences in intrinsic drug activity. One possible explanation is that a considerable proportion of patients in cohort 2 discontinued initial abemaciclib because of AEs rather than PD. Patients who discontinue treatment due to toxicity may retain sensitivity to CDK4/6 inhibition, whereas those who discontinue because of PD may have already acquired resistance mechanisms. Consistent with this hypothesis, patients who discontinued initial CDK4/6i therapy due to AEs tended to experience longer retreatment TTF. Although this observation requires validation in larger cohorts, it suggests that toxicity-related discontinuation may identify patients who remain candidates for subsequent CDK4/6i rechallenge. Conversely, cohort 3 demonstrated a median TTF of 6 months during retrial compared with 12 months during initial treatment, with only 23.8% of patients experiencing a longer TTF upon retrial (Fig. 2), suggesting that abemaciclib retreatment may provide limited clinical benefit after palbociclib exposure. The limited benefit observed in this cohort may be explained by the relatively small proportion of patients who discontinued initial palbociclib because of AEs. As most patients underwent treatment switching following PD, acquired resistance to CDK4/6 inhibition may have already developed before retrial. Taken together, the differing efficacy of CDK4/6i retreatment observed between cohorts 2 and 3 may reflect differences in the reasons for discontinuation of the initial CDK4/6 inhibitor rather than the activity of the retreatment regimen itself. While postMONARCH trials primarily evaluated CDK4/6i retrial after PD19, real-world practice frequently involves retrial due to AEs. Thus, although most clinical trials provide important insights into treatment selection after disease progression, very few offer evidence on therapeutic decision-making following AEs. Consequently, optimal post-AE treatment strategies remain uncertain. Our findings suggest that sequential CDK4/6i administration after discontinuation due to AEs may be a clinically relevant option in selected patients. Prospective studies are needed to confirm its clinical utility and to refine patient selection. We additionally evaluated the total TTF from initiation of the first CDK4/6 inhibitor to failure of CDK4/6 inhibitor retreatment. Similar to the OS analysis, no statistically significant differences were observed among the treatment-sequence cohorts (Fig. 1, Supplementary Fig. S5a-b). However, interpretation of these findings requires caution because abemaciclib was approved approximately one year later than palbociclib in Japan, resulting in a shorter follow-up period in cohort 2 than in cohort 3. Therefore, potential differences in long-term outcomes may not yet be fully captured, and further follow-up is warranted before definitive conclusions can be drawn regarding whether abemaciclib should be used upfront or reserved for later-line treatment. The optimal choice of first-line CDK4/6 inhibitor remains a matter of ongoing debate. This interpretation may also be supported by evidence from major clinical trials of CDK4/6 inhibitors. Although neither MONARCH 3 nor PALOMA-2 was powered to definitively detect an OS difference, MONARCH 3 showed a clinically meaningful trend toward improved OS despite its relatively limited sample size, whereas PALOMA-2 did not demonstrate a similarly convincing OS signal24,25. A broadly consistent pattern has been observed in other settings: MONARCH 2 demonstrated a significant OS benefit, whereas palbociclib-based studies have yielded less consistent long-term survival results, and in the adjuvant setting, abemaciclib-based therapy in monarchE showed sustained benefit in survival outcomes, while palbociclib-based strategies in PALLAS and PENELOPE-B failed to show clear long-term benefit23,26,27. In addition, approximately 20% of patients in the abemaciclib arm of MONARCH 3 remained progression-free beyond 7 years, supporting the possibility of durable disease control in a substantial subset of patients24. Comparative data from PALMIRA also support this view, as abemaciclib was associated with better PFS than palbociclib (aHR 0.71) despite being used in patients with less favorable baseline characteristics, including more frequent liver metastasis, fewer de novo metastatic cases, and lower PgR expression. These findings may provide additional support for considering abemaciclib in treatment selection.28. Although these comparisons should be interpreted cautiously, the overall consistency of the available evidence suggests that abemaciclib may have greater long-term treatment potential than palbociclib, and this possibility may merit consideration when selecting first-line CDK4/6 inhibitor therapy and subsequent treatment sequencing in clinical practice. Regarding treatment options other than CDK4/6i rechallenge, several effective therapeutic strategies have emerged following progression on first-line CDK4/6i therapy, including PI3K/AKT/mTOR pathway inhibitors, selective estrogen receptor degraders, chemotherapy, and antibody–drug conjugates. Reported median PFS for these strategies generally ranges from approximately 4 to 8 months29,30, depending on patient population and molecular characteristics. The median TTF in cohort 2 supports a potential role for CDK4/6i retrial in selected patients. These findings underscore the need to consider retrial strategies within the broader therapeutic landscape. Furthermore, identifying the predictors of response to CDK4/6i retrial after prior CDK4/6i treatment failure is required. In cohort 2, univariate analyses demonstrated that visceral metastasis was significantly associated with shorter TTF of CDK4/6i retrial. Both univariate Cox regression and Kaplan–Meier analyses showed shorter TTF among patients with visceral involvement. However, given that only univariate analyses were performed and the confidence intervals were wide, these findings should be interpreted cautiously. Age and bone-only metastasis showed borderline associations but require validation in larger cohorts. In contrast, none of the examined variables in cohort 3 were significantly associated with TTF. Although a trend toward shorter TTF was observed in patients with de novo metastatic breast cancer, the association did not reach statistical significance. This may be because of the limitations of sample size. Collectively, these findings underscore the importance of considering metastatic site, particularly visceral disease, when evaluating the potential benefit of CDK4/6i retreatment. They also highlight the need for further studies to identify predictive factors for response, which could help optimize patient selection and improve clinical outcomes in both real-world and trial settings. Resistance to CDK4/6is is multifactorial and has been associated with alterations in cell-cycle regulators and activation of alternative signaling pathways31–34. The differential efficacy observed among cohorts in this study may therefore reflect distinct resistance mechanisms depending on prior CDK4/6i exposure. These biological considerations support further investigation into optimal sequencing strategies and predictive biomarkers. Collectively, our findings indicate that the efficacy of CDK4/6i retrial is heterogeneous and influenced by prior treatment exposure and clinical characteristics. Notably, palbociclib rechallenge after abemaciclib exposure was associated with clinically meaningful benefit in selected patients, including those who discontinued prior therapy due to toxicity. These results support further prospective evaluation to refine patient selection and optimize CDK4/6i retreatment strategies. Conclusions Our findings suggest that the efficacy of CDK4/6i retrial varies according to prior treatment exposure and clinical context, including the reason for initial discontinuation. In particular, selected patients—such as those discontinuing therapy due to toxicity—may derive benefit from sequential CDK4/6 inhibition. These results should be interpreted in light of the retrospective design and the use of TTF as a pragmatic endpoint. Prospective studies incorporating standardized progression assessment and molecular profiling are needed to validate these observations and refine patient selection for CDK4/6i retreatment. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1