Low‐Dose Antiangiogenic Therapy With Different Chemotherapy Backbones Among Patients With Heavily Pretreated Metastatic Breast Cancer Authors: Yuanyi Wang, Yalong Qi, Cheng Zeng, Ruixia Song, Yuhan Wei, Haili Qian, Fei Ma Published: 2026 ABSTRACT ABSTRACTBackgroundThe prognosis for metastatic breast cancer (MBC) remains poor, and treatment options are limited for patients with heavily pretreated disease. Although antiangiogenic agents may provide clinical benefit in this setting, real‐world evidence regarding their efficacy, safety, and optimal combination strategies remains limited. This study evaluated antiangiogenic therapy combined with chemotherapy in heavily pretreated MBC, with exploratory analyses across clinically relevant subgroups.MethodsThis multicenter retrospective study enrolled 348 patients with heavily pretreated MBC who received antiangiogenic therapy plus chemotherapy or chemotherapy alone. A 1:1 propensity score‐matched cohort of 174 pairs was generated for the primary comparative analysis. Chemotherapy backbones were categorized as microfilament–microtubule inhibitor (MMI)‐based or non‐MMI‐based regimens. The primary endpoint was progression‐free survival (PFS), and secondary endpoints included objective response rate (ORR) and safety. Exploratory subgroup analyses were conducted according to antiangiogenic agent, chemotherapy backbone, prior lines of therapy, and dosing strategy.ResultsIn the propensity score‐matched cohort, antiangiogenic therapy combined with chemotherapy was associated with significantly longer progression‐free survival than chemotherapy alone (mPFS, 122 days [95% CI, 104–151] vs. 90 days [95% CI, 83–113]; HR, 0.62; 95% CI, 0.49–0.78; p < 0.001). The incidence of treatment‐related adverse events was similar between the antiangiogenic therapy plus chemotherapy and chemotherapy‐alone groups (72.4% vs. 76.4%), with most events being grade 1–2. Exploratory analyses suggested that PFS outcomes varied according to antiangiogenic agent and chemotherapy backbone, with more favorable estimates observed for bevacizumab‐ or apatinib‐based combinations and for MMI‐based chemotherapy backbones. In exploratory descriptive dose‐stratified analyses, PFS estimates varied across dose groups, and these findings should not be interpreted as evidence of comparable efficacy between dose levels.ConclusionsAntiangiogenic therapy combined with chemotherapy was associated with longer progression‐free survival and a manageable safety profile in patients with heavily pretreated metastatic breast cancer. Exploratory analyses suggested that treatment outcomes may vary according to antiangiogenic agent, chemotherapy backbone, and prior treatment exposure. Dose–stratified analyses were descriptive and should not be interpreted as evidence of equivalence or non‐inferiority between dose levels. Given the retrospective observational design, these findings should be considered hypothesis‐generating and warrant prospective validation. Abbreviations adverse events alanine aminotransferase aspartate aminotransferase body mass index confidence interval common terminology criteria for adverse events Eastern Cooperative Oncology Group performance status fibroblast growth factor receptor hazard ratio hazard ratios invasive ductal carcinoma invasive lobular carcinoma metastatic breast cancer microfilament–microtubule inhibitors median PFS objective response rate overall survival platelet‐derived growth factor receptor progression‐free survival propensity score matching randomized controlled trial response evaluation criteria in solid tumors tyrosine kinase inhibitor vascular endothelial growth factor vascular endothelial growth factor receptor Introduction Breast cancer is the most common malignancy among women, and metastatic breast cancer (MBC) remains the major cause of breast cancer‐related mortality. Approximately 4%–6% of patients with breast cancer are diagnosed with advanced disease at initial presentation, and 30%–40% of patients with early‐stage breast cancer eventually develop metastatic disease despite active treatment [1, 2]. The prognosis for MBC is poor, characterized by a median survival of only 2–3 years and a 5‐year survival rate as low as 20% [2, 3]. MBC is not curable, and there is no standard treatment regimen for MBC patients who have received multiple lines of treatment. To address the unmet clinical needs, further research is warranted to investigate novel therapeutic agents and refine evidence‐based treatment approaches, aiming to improve survival outcomes and quality of life in patients with MBC. Tumor blood vessels continuously supply oxygen and nutrients to cancerous tissues, a process that is closely associated with tumor initiation and progression [4, 5]. The VEGF signaling pathway, mediated by vascular endothelial growth factor (VEGF) and its receptor (VEGFR), plays a crucial role in tumor angiogenesis [6]. Extensive research on tumor angiogenesis has established antiangiogenic therapy as a pivotal strategy in cancer treatment. Common antiangiogenic agents, such as monoclonal antibodies and small‐molecule VEGFR tyrosine kinase inhibitors, have demonstrated clinical efficacy in a variety of cancers. In breast cancer, the overexpression of VEGF and VEGFR promotes abnormal angiogenesis, making this signaling pathway a prime target for molecular targeted therapy [7]. Given the prominent role of VEGF signaling in breast cancer pathogenesis, antiangiogenic therapy has been extensively explored in this disease. Apatinib is a small‐molecule antiangiogenic tyrosine kinase inhibitor (TKI) that functions as a highly selective VEGFR‐2 inhibitor. By blocking the VEGF–VEGFR‐2 signaling pathway, apatinib suppresses tumor angiogenesis and disrupts tumor vascularization. Due to its demonstrated clinical efficacy, apatinib has been approved by the China Food and Drug Administration for the treatment of advanced gastric adenocarcinoma or gastroesophageal junction adenocarcinoma refractory to chemotherapy, as well as for advanced hepatocellular carcinoma [8, 9, 10]. Additionally, apatinib has been investigated and has shown promising antitumor activity in melanoma [11], non‐small cell lung cancer [12], colorectal cancer [13], and cervical cancer [14]. In breast cancer, apatinib has been explored both as monotherapy [15, 16] and in combination with chemotherapy, with studies reporting encouraging disease control and progression‐free survival outcomes in selected patients with advanced disease [17]. Anlotinib, another small‐molecule TKI that targets angiogenesis, primarily acts on VEGFR1–3, fibroblast growth factor receptor (FGFR) 1–4, platelet‐derived growth factor receptor (PDGFR)‐α, PDGFR‐β, and stem cell factor receptor c‐Kit. By inhibiting these pathways, anlotinib suppresses tumor angiogenesis and growth while promoting autophagy and apoptosis [17]. Anlotinib has demonstrated significant clinical efficacy in medullary thyroid cancer, non‐small cell lung cancer, small cell lung cancer, and soft tissue sarcoma, leading to its approval by the China Food and Drug Administration for these indications [18, 19, 20]. In breast cancer, anlotinib‐based regimens have also been evaluated in real‐world settings, with reported antitumor activity and manageable safety profiles in patients with metastatic breast cancer [21]. Bevacizumab, a recombinant humanized monoclonal antibody, was the first antiangiogenic agent approved for clinical use. Its mechanism of action, which involves binding to VEGF and preventing its interaction with VEGFR, differs fundamentally from that of small‐molecule TKIs. Clinical trials have consistently demonstrated that adding bevacizumab to chemotherapy regimens significantly improves progression‐free survival (PFS) in patients with MBC [22, 23]. However, the magnitude of benefit and optimal patient selection remain areas of ongoing investigation, particularly in heavily pretreated populations. These findings suggest that antiangiogenic therapy may provide clinical benefit for patients with MBC, particularly in the setting of limited treatment options after multiple lines of therapy. However, evidence from real‐world practice remains limited. Several clinically relevant issues remain to be clarified in real‐world settings, including the overall effectiveness and safety of antiangiogenic therapy combined with chemotherapy in heavily pretreated patients, as well as the heterogeneity of treatment outcomes across different therapeutic settings. In this multicenter retrospective study, we reviewed the clinical characteristics of patients with metastatic breast cancer who received antiangiogenic therapy in combination with conventional chemotherapy. Propensity score matching (PSM) was applied to identify a comparator cohort treated with chemotherapy alone. By comparing efficacy and safety outcomes between these two groups, we primarily aimed to evaluate the overall clinical benefit of antiangiogenic therapy combined with chemotherapy. Exploratory analyses were further conducted to describe potential differences across antiangiogenic agents, chemotherapy backbones, and prior treatment exposure. Patients and Methods Study Design This multicenter retrospective study was conducted at the Cancer Hospital, Chinese Academy of Medical Sciences, and Huanxing Cancer Hospital. Clinical data of patients with MBC treated between January 2016 and April 2024 were collected and analyzed. The study aimed to evaluate the real‐world efficacy and safety of antiangiogenic therapy combined with chemotherapy in patients with heavily pretreated MBC. In this real‐world setting, treatment selection was not guided by a predefined study protocol. All treatment decisions were made by treating physicians according to routine clinical practice, taking into account prior treatment history, Eastern Cooperative Oncology Group performance status (ECOG PS), disease burden, and treatment tolerability. This study was approved by the institutional review board of the National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College and was conducted in accordance with the Declaration of Helsinki, International Council for Harmonisation Good Clinical Practice guidelines, and applicable local ethical and legal requirements. Participants Patients with metastatic breast cancer treated at the participating centers during the study period were screened for eligibility. Eligible patients were required to have experienced disease progression after at least one prior line of systemic therapy and to have measurable disease according to Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1). The inclusion criteria were as follows: (1) confirmed diagnosis of metastatic breast cancer; (2) measurable disease according to RECIST 1.1; (3) disease progression following at least one prior line of systemic therapy; and (4) receipt of chemotherapy with or without an antiangiogenic agent. Patients with other active malignancies or concurrent uncontrolled infections, fever, or hematological disorders were excluded. Treatment Exposure and Patient Grouping Based on the treatment strategy administered at initiation of the analyzed chemotherapy regimen, patients were categorized into two groups: those treated with antiangiogenic therapy in combination with chemotherapy (Antiangio‐Chemo group) and those treated with chemotherapy alone (Chemo group). Patients in the Chemo group were retrospectively identified from the same participating centers and during the same study period as those receiving antiangiogenic therapy plus chemotherapy, and were required to meet the same eligibility criteria. The index date was defined consistently for both groups as the initiation date of the corresponding chemotherapy regimen. To ensure comparability of outcome assessment, all patients were required to have at least one radiologic evaluation after treatment initiation. Patients were excluded from the comparative analysis if antiangiogenic agents were initiated after chemotherapy start or if treatment crossover occurred before the first post‐baseline tumor assessment, to minimize treatment misclassification and immortal time bias. In the Antiangio‐Chemo group, antiangiogenic agents included bevacizumab, apatinib, and anlotinib, which were administered in combination with chemotherapy in routine clinical practice. Patients were further classified into bevacizumab‐based (Bev‐Chemo), apatinib‐based (Apa‐Chemo), and anlotinib‐based (Anlo‐Chemo) subgroups according to the specific antiangiogenic agent used. Chemotherapy backbones were also categorized based on their mechanism of action as microfilament–microtubule inhibitor (MMI) or non‐MMI regimens. Chemotherapy regimens were categorized according to the dominant cytotoxic agent administered at the index treatment line. MMI‐based regimens included taxanes (paclitaxel, docetaxel), eribulin, Vinca alkaloids, and utidelone. Non‐MMI regimens included capecitabine, gemcitabine, platinum‐based agents, and anthracyclines when administered without an MMI. For combination chemotherapy regimens, classification was based on the primary cytotoxic agent; for example, utidelone plus cisplatin was classified as an MMI‐based regimen. These subgroup classifications were applied for exploratory analyses. Low‐dose antiangiogenic therapy was defined as administration at doses lower than the commonly recommended or label‐based doses used for the corresponding antiangiogenic agents. For the descriptive dose‐stratified analyses, low‐dose bevacizumab was defined as < 10 mg/kg per administration, whereas higher dose bevacizumab was defined as ≥ 10 mg/kg per administration. Apatinib dose was categorized as 250 mg once daily versus 425 mg once daily, both of which were lower than the standard approved doses used in gastric adenocarcinoma or hepatocellular carcinoma. Low‐dose anlotinib was defined as 8 mg or 10 mg once daily; because nearly all patients receiving anlotinib were treated at these doses, no formal dose comparison was performed for anlotinib. These dose‐stratified analyses were descriptive and were not designed to assess equivalence or non‐inferiority between dose levels. Efficacy and Safety Evaluation Tumor response was assessed according to RECIST 1.1. Radiologic evaluations using computed tomography or magnetic resonance imaging were performed every two treatment cycles or whenever clinically indicated. The primary endpoint of the study was PFS, defined as the time from treatment initiation to documented disease progression or death from any cause. Given the heavily pretreated population, heterogeneous subsequent treatments, and immature overall survival (OS) data at the time of analysis, PFS was selected as the primary endpoint because it directly reflects disease control during the analyzed treatment line and is less affected by post‐progression therapies. Patients who remained progression‐free at the time of last follow‐up were censored. Secondary endpoints included objective response rate (ORR) and safety. ORR was defined as the proportion of patients achieving a complete or partial response as their best overall response according to RECIST 1.1. Adverse events (AEs) were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), version 5.0. Statistical Analysis Given the retrospective and non‐randomized nature of this study, adjustment for potential confounding factors was performed using a combination of PSM and multivariable regression analyses. Baseline characteristics were compared using Pearson's χ 2 test or Fisher's exact test, as appropriate. For paired binary outcomes such as objective response rate in the propensity score‐matched cohort, between‐group comparisons were performed using McNemar's test. PFS was estimated using the Kaplan–Meier method, and survival curves were compared using the log‐rank test. Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were calculated using Cox proportional hazards regression models. Patients treated with antiangiogenic therapy plus chemotherapy and those treated with chemotherapy alone were first identified from the overall study population. The overall study flow is shown in Supporting Information S2: Figure S1. PSM was applied as the primary method to balance baseline characteristics between the two treatment groups. Propensity scores were estimated using a logistic regression model incorporating baseline hormone receptor status, HER2 status, treatment regimen, prior lines of therapy, and ECOG PS. One‐to‐one nearest‐neighbor matching without replacement was performed with a caliper width of 0.2 of the standard deviation of the logit of the propensity score. Covariate balance before and after matching was assessed using standardized mean differences (SMDs) and visualized using a Love plot (Supporting Information S2: Figure S2). An absolute SMD < 0.1 was considered indicative of adequate balance. Treatment effects were estimated in the matched cohort using Cox proportional hazards models with robust sandwich standard errors clustered on matched pairs. To further address potential residual confounding, an adjusted Cox model was additionally fitted including clinically relevant covariates that showed minor residual imbalance after matching (absolute SMD between 0.1 and 0.2), specifically histopathological type, lung metastasis, and brain metastasis. Subgroup analyses were performed in an exploratory manner. As PSM was conducted only for the primary comparison between antiangiogenic therapy plus chemotherapy and chemotherapy alone, subgroup comparisons were additionally adjusted using multivariable regression models. For subgroup comparisons between individual antiangiogenic agent–based regimens and chemotherapy alone, analyses were performed in the full unmatched cohort using multivariable Cox models with adjustment for prespecified baseline covariates. For subgroup comparisons within the antiangiogenic‐treated population, analyses were restricted to the antiangiogenic cohort and adjusted for the same set of baseline covariates. To account for multiple exploratory comparisons, the false discovery rate (FDR) was controlled using the Benjamini‐Hochberg (BH) procedure. BH‐adjusted q values were calculated separately within each prespecified set of exploratory comparisons (Sections A–D), corresponding to analyses by antiangiogenic agent, chemotherapy backbone, prior lines of therapy, and dose strata. Nominal p values and BH–FDR‐adjusted q values are reported. Exploratory findings were interpreted descriptively and considered hypothesis‐generating. All statistical analyses were performed using SPSS version 26.0 (IBM Corp., USA) and R statistical software version 4.3. All statistical tests were two‐sided. A p value < 0.05 was considered statistically significant for primary analyses, whereas for exploratory analyses, BH–FDR‐adjusted q values < 0.10 were considered noteworthy. Results Patients and Baseline Characteristics A total of 348 patients with metastatic breast cancer were included in this study. Using the reverse Kaplan–Meier method, the median follow‐up duration was 113 days. Patients were treated with either antiangiogenic therapy combined with chemotherapy (Antiangio‐Chemo group) or chemotherapy alone (Chemo group). Patient screening, exclusion, treatment allocation, and subsequent propensity score matching are summarized in the patient flow diagram (Supporting Information S2: Figure S1). After 1:1 propensity score matching, a matched cohort was generated for the primary comparative analyses. Baseline characteristics of patients in the propensity score–matched cohort are presented in Table 1. After matching, baseline variables were well balanced between the Antiangio‐Chemo and Chemo groups. Most covariates achieved good balance, with SMDs below 0.1, while a small number of variables showed minor residual imbalance (SMD between 0.1 and 0.2). These residual imbalances were considered acceptable and were further adjusted for in subsequent multivariable analyses. Detailed balance diagnostics are illustrated in the Love plot (Supporting Information S2: Figure S2). Table 1Baseline characteristics of patients after propensity score matching.VariableAfter propensity score matching p valueSMDAntiangio‐ChemoChemo N = 174, n (%) N = 174, n (%)Age (years)0.4420.094< 5064 (36.8)72 (41.4)≥ 50110 (63.2)102 (58.6)ECOG PS10.025≤ 1164 (94.3)165 (94.8)> 110 (5.7)9 (5.2)BMI (kg/m2)0.6670.058< 2497 (55.8)92 (52.9)≥ 2477 (44.2)82 (47.1)Histopathological types0.1940.194IDC140 (80.5)150 (86.2)ILC10 (5.8)4 (2.3)Others24 (13.7)20 (11.5)Hormone‐receptor status0.3340.115Positive84 (48.3)94 (54.0)Negative90 (51.7)80 (46.0)HER2 status0.7340.055HER2 negative156 (89.7)153 (87.9)HER2 positive18 (10.3)21 (12.1)Molecular subtype0.4650.133TNBC66 (37.9)77 (44.3)Luminal A/B94 (54.0)83 (47.7)HER2 positive14 (8.0)14 (8.0)Liver metastasis0.5710.073Yes62 (35.6)56 (32.2)No112 (64.4)118 (67.8)Lung metastasis0.2740.129Yes75 (43.1)64 (36.8)No99 (56.9)110 (63.2)Bone metastasis1< 0.001Yes73 (42.0)73 (42.0)No101 (58.0)101 (58.0)Brain metastasis0.3450.127Yes12 (6.9)7 (4.0)No162 (93.1)167 (96.0)Number of metastatic sites0.9130.023< 3102 (58.6)104 (59.8)≥ 372 (41.4)70 (40.2)Prior lines of therapy0.5860.070< 468 (39.1)74 (42.5)≥ 4106 (60.9)100 (57.5)Abbreviations: BMI, body mass index; ECOG PS, Eastern Cooperative Oncology Group performance status; HER2, human epidermal growth factor receptor 2‐positive breast cancer; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; SMD, standardized mean difference; TNBC, triple‐negative breast cancer.John Wiley & Sons, Ltd. Baseline characteristics of patients after propensity score matching. Abbreviations: BMI, body mass index; ECOG PS, Eastern Cooperative Oncology Group performance status; HER2, human epidermal growth factor receptor 2‐positive breast cancer; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; SMD, standardized mean difference; TNBC, triple‐negative breast cancer. Detailed baseline characteristics and treatment composition of patients treated with antiangiogenic in combination with chemotherapy are summarized in Table 2. In the Antiangio‐Chemo group, the median age at diagnosis was 51.0 years (range: 28–74 years). Visceral metastases were present in 146 patients, including liver metastases in 62 patients, lung metastases in 75 patients, bone metastases in 73 patients, and brain metastases in 12 patients. Overall, 156 patients (89.7%) were HER2‐negative and 84 patients (48.3%) were hormone receptor–positive. A total of 106 patients (60.9%) had received four or more prior lines of systemic therapy. Table 2Characteristics of patients treated with antiangiogenic therapy plus chemotherapy.VariableTotal No., N (%)Bevacizumab, n (%)Apatinib, n (%)Anlotinib, n (%) p 174 (100.0)97 (55.8)51 (29.3)26 (14.9)Age (years)0.740< 5064 (36.8)34 (35.0)21 (41.2)9 (34.6)≥ 50110 (63.2)63 (65.0)30 (58.8)17 (65.4)ECOG PS0.226≤ 1164 (94.3)89 (91.8)50 (98.0)25 (96.2)> 110 (5.7)8 (8.2)1 (2.0)1 (3.8)BMI (kg/m2)0.968< 2497 (55.7)54 (55.7)29 (56.9)14 (53.8)≥ 2477 (44.3)43 (44.3)22 (43.1)12 (46.2)Histopathological types0.598IDC140 (80.5)79 (81.4)43 (84.3)18 (69.2)ILC10 (5.7)5 (5.2)2 (3.9)3 (11.5)Others24 (13.8)13 (13.4)6 (11.8)5 (19.3)Hormone‐receptor status0.381Positive84 (48.3)51 (52.6)23 (45.1)10 (38.5)Negative90 (51.7)46 (47.4)28 (54.9)16 (61.5)HER2 status0.852HER2 negative156 (89.7)87 (89.7)45 (88.2)24 (92.3)HER2 positive18 (10.3)10 (10.3)6 (11.8)2 (7.7)Liver metastasis0.569Yes62 (35.6)37 (38.1)18 (35.3)7 (26.9)No112 (64.4)60 (61.9)33 (64.7)19 (73.1)Lung metastasis0.014Yes75 (43.1)37 (38.1)20 (39.2)18 (69.2)No99 (56.9)60 (61.9)31 (60.8)8 (30.8)Bone metastasis0.234Yes73 (42.0)44 (45.4)22 (43.1)7 (26.9)No101 (58.0)53 (54.3)29 (56.9)19 (73.1)Brain metastasis0.178Yes12 (6.9)9 (9.3)1 (2.0)2 (7.7)No162 (93.1)88 (90.7)50 (98.0)24 (92.3)Number of metastatic sites0.119< 3102 (58.6)52 (53.6)36 (70.6)14 (53.8)≥ 372 (41.4)45 (46.4)15 (29.4)12 (46.2)Prior lines of therapy0.039< 468 (39.1)34 (35.1)18 (35.3)16 (61.5)≥ 4106 (60.9)63 (64.9)33 (64.7)10 (38.5)Abbreviations: BMI, body mass index; ECOG PS, Eastern Cooperative Oncology Group performance status; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma.John Wiley & Sons, Ltd. Characteristics of patients treated with antiangiogenic therapy plus chemotherapy. Abbreviations: BMI, body mass index; ECOG PS, Eastern Cooperative Oncology Group performance status; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma. Within the Antiangio‐Chemo group, bevacizumab‐containing regimens were administered in 97 patients (55.8%), apatinib‐containing regimens in 51 patients (29.3%), and anlotinib‐containing regimens in 26 patients (14.9%). Chemotherapy backbones combined with antiangiogenic therapy were primarily MMI, with 112 patients (64.4%) receiving MMI‐based regimens and 62 patients (35.6%) receiving non‐MMI regimens (Supporting Information S1: Table S1). Efficacy Progression‐Free Survival and Response Outcomes in the Antiangio‐Chemo and Chemo Groups In the propensity score‐matched cohort, patients treated with antiangiogenic therapy combined with chemotherapy exhibited a longer PFS than those treated with chemotherapy alone. The median PFS was 122 days (95% CI, 104–151 days) in the Antiangio‐Chemo group and 90 days (95% CI, 83–113 days) in the Chemo group (HR, 0.62; 95% CI, 0.49–0.78; p < 0.001) (Figure 1a). Overall efficacy of antiangiogenic therapy combined with chemotherapy. (a) PFS in patients treated with antiangiogenic therapy combined with chemotherapy versus chemotherapy alone in the propensity score–matched cohort. (b) ORR in patients treated with antiangiogenic therapy combined with chemotherapy and chemotherapy alone. ORR, objective response rate; PFS, progression‐free survival. The ORR was also higher in the Antiangio‐Chemo group than in the Chemo group in the propensity score–matched cohort (25.9% vs. 14.4%; McNemar's test, p = 0.007) (Figure 1b). Exploratory Analysis by Antiangiogenic Agent Exploratory subgroup analyses were performed according to the specific antiangiogenic agent used in combination with chemotherapy, including bevacizumab, apatinib, and anlotinib. Nominal p values and BH–FDR‐adjusted q values are reported in Supporting Information S1: Table S2. In the bevacizumab‐based subgroup, patients treated with bevacizumab plus chemotherapy had a median PFS of 129 days (95% CI, 109–167 days), compared with 90 days (95% CI, 83–113 days) in the Chemo group (HR, 0.57; 95% CI, 0.43–0.75; p < 0.001) (Figure 2a). In the apatinib‐based subgroup, patients treated with apatinib plus chemotherapy had a median PFS of 109.5 days (95% CI, 87–153 days), compared with 90 days (95% CI, 83–113 days) in the Chemo group (HR, 0.70; 95% CI, 0.49–0.97; p = 0.032) (Figure 2b). In the anlotinib‐based subgroup, patients treated with anlotinib plus chemotherapy had a median PFS of 100 days (95% CI, 90–167 days), compared with 90 days (95% CI, 83–113 days) in the Chemo group. The association did not reach statistical significance (HR, 0.68; 95% CI, 0.43–1.06; nominal p = 0.086) (Supporting Information S2: Figure S3), and this finding should be interpreted with caution given the limited sample size. Exploratory subgroup analyses of PFS according to the antiangiogenic agent used in combination with chemotherapy. PFS in patients treated with bevacizumab (a) or apatinib (b) combined with chemotherapy compared with chemotherapy alone. PFS, progression‐free survival. Exploratory Analysis by Chemotherapy Backbone PFS was further examined in exploratory analyses according to the chemotherapy backbone used in combination with antiangiogenic therapy. Patients were categorized into an antiangiogenic therapy plus MMI group (Antiangio‐MMI) and an antiangiogenic therapy plus non‐MMI group (Antiangio‐non‐MMI). The median PFS was 151 days (95% CI, 123–168 days) in the Antiangio‐MMI group and 90 days (95% CI, 76–108 days) in the Antiangio‐non‐MMI group (HR, 0.58; 95% CI, 0.42–0.81; p = 0.001) (Figure 3a). In exploratory analyses by antiangiogenic agent, patients treated with bevacizumab combined with MMI (Bev‐MMI group) had a median PFS of 152 days (95% CI, 123–180 days), compared with 72 days (95% CI, 58–86 days) in those treated with bevacizumab combined with non‐MMI regimens (HR, 0.41; 95% CI, 0.23–0.72; p = 0.002) (Figure 3b). Similarly, patients treated with apatinib combined with MMI (Apa‐MMI group) had a median PFS of 151 days (95% CI, 122–334 days), compared with 91 days (95% CI, 76–140 days) in those treated with apatinib plus non‐MMI regimens (HR, 0.51; 95% CI, 0.27–0.96; p = 0.036) (Figure 3c). PFS according to chemotherapy backbone among patients receiving antiangiogenic therapy‐containing regimens. (a) Antiangio‐MMI versus Antiangio‐non‐MMI. (b) Bev‐MMI versus Bev‐non‐MMI. (c) Apa‐MMI versus Apa‐non‐MMI. MMI, microfilament–microtubule inhibitors; PFS, progression‐free survival. For patients treated with anlotinib, no statistically significant differences in PFS were observed between MMI and non‐MMI chemotherapy backbones (Supporting Information S2: Figure S4). Exploratory Analysis by Prior Lines of Therapy PFS was further evaluated in exploratory analyses according to the number of prior lines of systemic therapy (< 4 vs. ≥ 4). Among patients treated with bevacizumab plus chemotherapy, those who had received ≥ 4 prior lines of therapy had a median PFS of 165 days (95% CI, 123–180 days), compared with 94 days (95% CI, 77–152 days) in those who had received fewer than four prior lines of therapy (p = 0.016) (Figure 4a). Among patients treated with apatinib plus chemotherapy, those who had received fewer than four prior lines of therapy had a median PFS of 157.5 days (95% CI, 131–334 days), whereas those who had received ≥ 4 prior lines had a median PFS of 89 days (95% CI, 76–123 days) (p = 0.011) (Figure 4b). Exploratory analyses of PFS according to prior lines of therapy in patients receiving antiangiogenic therapy‐containing regimens. (a) PFS in patients receiving bevacizumab‐containing regimens stratified by prior lines of therapy. (b) PFS in patients receiving apatinib‐containing regimens stratified by prior lines of therapy. PFS, progression‐free survival. Among patients treated with anlotinib plus chemotherapy, no statistically significant difference in PFS was observed according to the number of prior treatment lines (Supporting Information S2: Figure S5). Exploratory Descriptive Analysis According to Antiangiogenic Drug Dose PFS outcomes were further assessed in exploratory descriptive analyses according to the dose of antiangiogenic agents administered in combination with chemotherapy. Among patients treated with bevacizumab plus chemotherapy, 85 patients (87.6%) received a low‐dose bevacizumab, defined as a dose below 10 mg/kg per administration. The median PFS was 139 days (95% CI, 104–166 days) in the low‐dose group and 128 days (95% CI, 35–221 days) in the higher dose group (HR, 0.70; 95% CI, 0.37–1.33; p = 0.30) (Figure 5a). Exploratory descriptive analyses of PFS according to antiangiogenic drug dose. (a) PFS in patients treated with low‐ or higher‐dose bevacizumab combined with chemotherapy. (b) PFS in patients treated with 250 or 425 mg apatinib combined with chemotherapy. PFS, Progression‐free survival. Among patients treated with apatinib plus chemotherapy, all patients received apatinib at doses lower than the standard approved doses used for gastric adenocarcinoma or hepatocellular carcinoma, with 41 patients (80.4%) receiving 250 mg once daily and 10 patients (19.6%) receiving 425 mg once daily. The median PFS was 119 days (95% CI, 0–248 days) in the 250 mg group and 100 days (95% CI, 86–153 days) in the 425 mg group (HR, 0.78; 95% CI, 0.38–1.62; p = 0.50) (Figure 5b). Among patients treated with anlotinib plus chemotherapy, 25 patients (96.2%) received a daily dose of 8 mg or 10 mg. PFS outcomes according to anlotinib dose are shown in Supporting Information S2: Figure S6. Given the small number of patients in the higher dose strata and the absence of a formal non‐inferiority design, these dose‐stratified findings should be interpreted descriptively and should not be considered evidence of comparable efficacy between dose levels. Safety Treatment‐related AEs are summarized in Table 3 and Supporting Information S1: Table S3. Overall, AEs occurred in 126/174 patients (72.4%) in the antiangiogenic therapy plus chemotherapy group and in 133/174 patients (76.4%) in the chemotherapy‐alone group. Most AEs were grade 1–2. No treatment‐related deaths were documented. Treatment‐related dose reductions were not systematically captured, although no such reductions were identified in the available medical records. Table 3Treatment‐related adverse events in patients treated with antiangiogenic therapy plus chemotherapy or chemotherapy alone.Adverse eventsAntiangio‐chemo (N = 174)Chemotherapy (N = 174)Grade, 1/2 n (%)Grade 3/4, n (%)All grade, n (%)Grade, 1/2 n (%)Grade, 3/4 n (%)All grade, n (%)All Events92 (52.9)34 (19.5)126 (72.4)88 (50.6)45 (25.9)133 (76.4)Leukopenia72 (41.4)17 (9.8)89 (51.1)48 (27.6)27 (15.5)75 (43.1)Neutropenia63 (36.2)23 (13.2)86 (49.4)47 (27.0)41 (23.6)88 (50.6)Nausea/vomiting44 (25.3)1 (0.6)45 (25.9)37 (21.3)1 (0.6)38 (21.8)AST increased46 (26.4)2 (1.1)48 (27.6)34 (19.5)0 (0)34 (19.5)Anemia34 (19.5)0 (0)34 (19.5)35 (20.1)1 (0.6)36 (20.7)Alopecia12 (6.9)a 21 (12.1)b 33 (19.0)9 (5.2)a 17 (9.8)b 26 (15.0)Peripheral neuropathy24 (13.8)2 (1.1)26 (14.9)21 (12.1)2 (1.1)23 (13.2)Hypertension19 (10.9)3 (1.7)22 (12.6)0 (0)0 (0)0 (0)ALT increased19 (10.9)0 (0)19 (10.9)26 (14.9)0 (0)26 (14.9)Thrombocytopenia7 (4.0)4 (2.3)11 (6.3)6 (3.4)2 (1.1)8 (4.6)Fatigue11 (6.3)0 (0)11 (6.3)11 (6.3)1 (0.6)12 (6.9)Elevated bilirubin8 (4.6)0 (0)8 (4.6)3 (1.7)0 (0)3 (1.7)Diarrhea7 (4.0)0 (0)7 (4.0)12 (6.9)2 (1.1)14 (8.0)Hand‐foot skin reaction4 (2.3)2 (1.1)6 (3.4)6 (3.4)2 (1.1)8 (4.6)Proteinuria6 (3.4)0 (0)6 (3.4)0 (0)0 (0)0 (0)Elevated creatinine5 (2.9)0 (0)5 (2.9)1 (0.6)1 (0.6)2 (1.1)Mucositis4 (2.3)0 (0)4 (2.3)3 (1.7)0 (0)3 (1.7)Rash1 (0.6)0 (0)1 (0.6)2 (1.1)0 (0)2 (1.1)Bleeding4 (2.3)0 (0)4 (2.3)3 (1.7)0 (0)3 (1.7)Thromboembolism (VTE/ATE)5 (2.9)0 (0)5 (2.9)3 (1.7)0 (0)3 (1.7)Heart failure/LV dysfunction2 (1.1)0 (0)2 (1.1)3 (1.7)0 (0)3 (1.7) Note: Data are n (%), based on the denominator of each treatment group. For recurrent AEs, the highest CTCAE grade was counted.Abbreviations: AEs, Adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATE, arterial thromboembolism; LV, left ventricular; VTE, venous thromboembolism. a Grade 1. b Grade 2.John Wiley & Sons, Ltd. Treatment‐related adverse events in patients treated with antiangiogenic therapy plus chemotherapy or chemotherapy alone. Note: Data are n (%), based on the denominator of each treatment group. For recurrent AEs, the highest CTCAE grade was counted. Abbreviations: AEs, Adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATE, arterial thromboembolism; LV, left ventricular; VTE, venous thromboembolism. Grade 1. Grade 2. The most common AEs in both groups included hematologic toxicity, alopecia, peripheral neuropathy, and elevated aminotransferase levels. Hypertension occurred in 22/174 patients (12.6%) in the antiangiogenic therapy plus chemotherapy group, whereas no cases were reported in the chemotherapy‐alone group. In the antiangiogenic therapy plus chemotherapy group, the most frequent grade 3–4 AEs were neutropenia in 23/174 patients (13.2%), leukopenia in 17/174 patients (9.8%), and thrombocytopenia in 4/174 patients (2.3%). Antiangiogenic therapy–associated toxicities of particular interest, including proteinuria, bleeding events, thromboembolic events, and cardiac dysfunction, were uncommon in this cohort, as detailed in Table 3 and Supporting Information S1: Table S3. However, because of the retrospective design and limited sample size, underreporting of mild, asymptomatic, or subclinical events cannot be excluded. Treatment‐related AEs stratified by antiangiogenic agent are summarized in Supporting Information S1: Table S3. Among patients receiving bevacizumab‐containing regimens, the most commonly reported AEs were leukopenia in 46/97 patients (47.4%), neutropenia in 45/97 patients (46.4%), increased aspartate aminotransferase (AST) levels in 33/97 patients (34.0%), nausea/vomiting in 21/97 patients (21.6%), and anemia in 17/97 patients (17.5%). Among patients receiving apatinib‐containing regimens, the most frequently reported AEs were leukopenia in 27/51 patients (52.9%), neutropenia in 22/51 patients (43.1%), nausea/vomiting in 15/51 patients (29.4%), hypertension in 12/51 patients (23.5%), increased AST levels in 11/51 patients (21.6%), and anemia in 11/51 patients (21.6%). Among patients receiving anlotinib‐containing regimens, the most commonly reported AEs were leukopenia in 16/26 patients (61.5%), neutropenia in 19/26 patients (73.1%), nausea/vomiting in 9/26 patients (34.6%), alopecia in 8/26 patients (30.8%), and anemia in 6/26 patients (23.1%). Hypertension was numerically most frequent in the apatinib subgroup, whereas neutropenia was numerically more common in the anlotinib subgroup. Safety profiles stratified by antiangiogenic drug dose are summarized descriptively in Tables 4 and 5. Among patients treated with bevacizumab, overall AEs occurred in 10/12 patients (83.3%) in the higher dose group and 60/85 patients (70.6%) in the low‐dose group. Among patients treated with apatinib, overall AEs occurred in 8/10 patients (80.0%) in the higher dose group and 28/41 patients (68.3%) in the low‐dose group. The distributions of specific AEs by dose level are provided in Tables 4 and 5. Among patients treated with anlotinib, 25/26 patients (96.2%) received a daily dose of 8 mg or 10 mg, and only 1/26 patient (3.8%) received a dose of 12 mg; therefore, no formal comparison of AEs by dose was performed for anlotinib‐treated patients. Table 4Adverse events for patients treated with regimen containing bevacizumab.Adverse eventsLow‐dose (n = 85)Higher‐dose (n = 12)Grade 1/2, n (%)Grade 3/4, n (%)All grade, n (%)Grade, 1/2 n (%)Grade 3/4, n (%)All grade, n (%)Leukopenia34 (40.0)4 (4.7)38 (44.7)5 (41.7)3 (25)8 (66.7)Neutropenia29 (34.1)6 (7.1)35 (41.2)4 (33.3)6 (50)10 (83.3)Nausea/vomiting16 (18.8)0 (0)16 (18.8)5 (41.7)0 (0)5 (41.7)AST increased27 (31.8)1 (1.2)28 (32.9)4 (33.3)1 (8.3)5 (41.7)Anemia13 (15.3)0 (0)13 (15.3)4 (33.3)0 (0)4 (33.3)Alopecia4 (4.7)a 7 (8.3)b 11 (12.9)1 (8.3)a 3 (25)b 4 (33.3)Peripheral neuropathy9 (10.6)1 (1.2)10 (11.8)4 (33.3)1 (8.3)5 (41.7)Hypertension7 (8.2)0 (0)7 (8.2)5 (41.7)1 (8.3)6 (50)ALT increased8 (9.4)0 (0)8 (9.4)4 (33.3)0 (0)4 (33.3)Thrombocytopenia2 (2.4)1 (1.2)3 (3.5)1 (8.3)1 (8.3)2 (16.7)Fatigue2 (2.4)0 (0)2 (2.4)2 (16.7)0 (0)2 (16.7)Elevated bilirubin3 (3.5)0 (0)3 (3.5)1 (8.3)0 (0)1 (8.3)Diarrhea0 (0)0 (0)0 (0)2 (16.7)0 (0)2 (16.7)Hand‐foot skin reaction1 (1.2)0 (0)1 (1.2)1 (8.3)0 (0)1 (8.3)Proteinuria3 (3.5)0 (0)3 (3.5)1 (8.3)0 (0)1 (8.3)Elevated creatinine1 (1.2)0 (0)1 (1.2)2 (16.7)0 (0)2 (16.7)Mucositis1 (1.2)0 (0)1 (1.2)1 (8.3)0 (0)1 (8.3)Rash0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)Bleeding1 (1.2)0 (0)1 (1.2)1 (8.3)0 (0)1 (8.3)Thromboembolism (VTE/ATE)2 (2.4)0 (0)2 (2.4)1 (8.3)0 (0)1 (8.3)Heart failure/LV dysfunction0 (0)0 (0)0 (0)1 (8.3)0 (0)1 (8.3) Note: Data are n (%), based on the denominator of each treatment group. For recurrent AEs, the highest CTCAE grade was counted.Abbreviations: AEs, Adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATE, arterial thromboembolism; LV, left ventricular; VTE, venous thromboembolism. a Grade 1. b Grade 2.John Wiley & Sons, Ltd. Adverse events for patients treated with regimen containing bevacizumab. Note: Data are n (%), based on the denominator of each treatment group. For recurrent AEs, the highest CTCAE grade was counted. Abbreviations: AEs, Adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATE, arterial thromboembolism; LV, left ventricular; VTE, venous thromboembolism. Grade 1. Grade 2. Table 5Adverse events for patients treated with regimen containing apatinib.Adverse eventsLow‐dose (n = 41)Higher‐dose (n = 10)Grade 1/2, n (%)Grade 3/4, n (%)All grade, n (%)Grade 1/2, n (%)Grade 3/4, n (%)All grade, n (%)Leukopenia18 (43.9)2 (4.9)20 (48.8)5 (50.0)2 (20.0)7 (70.0)Neutropenia11 (26.8)3 (7.3)14 (34.1)5 (50.0)3 (30.0)8 (80.0)Nausea/vomiting10 (24.4)0 (0)10 (24.4)4 (40.0)1 (10.0)5 (50.0)AST increased7 (17.1)0 (0)7 (17.1)4 (40.0)0(0)4 (40.0)Anemia6 (14.6)0 (0)6 (14.6)5 (50.0)0 (0)5 (50.0)Alopecia3 (7.3)a 2 (4.9)b 5 (12.2)1 (10.0)a 4 (40.0)b 5 (50.0)Peripheral neuropathy4 (9.8)0 (0)4 (9.8)3 (30.0)0 (0)3 (30.0)Hypertension6 (14.6)0 (0)6 (14.6)3 (30.0)3 (30.0)6 (60.0)ALT increased1 (2.4)0 (0)1 (2.4)4 (40.0)0 (0)4 (40.0)Thrombocytopenia2 (4.9)0 (0)2 (4.9)1 (10.0)2 (20.0)3 (30.0)Fatigue2 (4.9)0 (0)2 (4.9)2 (20.0)0 (0)2 (20.0)Elevated bilirubin1 (2.4)0 (0)1 (2.4)3 (30.0)0 (0)3 (30.0)Diarrhea0 (0)0 (0)0 (0)2 (20.0)0 (0)2 (20.0)Hand‐foot skin reaction1 (2.4)0 (0)1 (2.4)0 (0)1 (10.0)1 (10.0)Proteinuria0 (0)0 (0)0 (0)2 (20.0)0 (0)2 (20.0)Elevated creatinine0 (0)0 (0)0 (0)1 (10.0)0 (0)1 (10.0)Mucositis0 (0)0 (0)0 (0)1 (10.0)0 (0)1 (10.0)Rash0 (0)0 (0)0 (0)1 (10.0)0 (0)1 (10.0)Bleeding0 (0)0 (0)0 (0)1 (10.0)0 (0)1 (10.0)Thromboembolism (VTE/ATE)0 (0)0 (0)0 (0)1 (10.0)0 (0)1 (10.0)Heart failure/LV dysfunction0 (0)0 (0)0 (0)0 (0)0 (0)0 (0) Note: Data are n (%), based on the denominator of each treatment group. For recurrent AEs, the highest CTCAE grade was counted.Abbreviations: AEs, Adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATE, arterial thromboembolism; LV, left ventricular; VTE, venous thromboembolism. a Grade 1. b Grade 2.John Wiley & Sons, Ltd. Adverse events for patients treated with regimen containing apatinib. Note: Data are n (%), based on the denominator of each treatment group. For recurrent AEs, the highest CTCAE grade was counted. Abbreviations: AEs, Adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ATE, arterial thromboembolism; LV, left ventricular; VTE, venous thromboembolism. Grade 1. Grade 2. Discussion Patients with heavily pretreated MBC have limited therapeutic options and poor prognosis, highlighting an urgent need for effective and tolerable treatment strategies in later‐line settings. Tumor angiogenesis plays a central role in cancer progression, and antiangiogenic therapies have demonstrated clinical benefit in several solid tumors. However, evidence supporting their use in heavily pretreated MBC remains limited, particularly in real‐world practice. In this multicenter retrospective study, antiangiogenic therapy combined with chemotherapy was associated with longer PFS and a higher ORR than chemotherapy alone after adjustment for baseline characteristics, suggesting that this approach may warrant further investigation as a later‐line treatment strategy for selected patients with advanced MBC. These findings should be interpreted as hypothesis‐generating because of the retrospective, non‐randomized design and the possibility of residual confounding. The study population represented a particularly challenging clinical cohort. More than 60% of patients had received four or more prior lines of systemic therapy, and over 40% had metastases at three or more sites, reflecting a high tumor burden and limited remaining treatment options. In this setting, conventional palliative chemotherapy alone often provides only modest clinical benefit. The association between antiangiogenic combination therapy and longer PFS in this heavily pretreated population may therefore be clinically relevant, although the absolute survival outcomes should be interpreted in the context of the advanced disease burden and extensive prior treatment exposure in this cohort. Across antiangiogenic agents, PFS estimates generally favored combination therapy over chemotherapy alone, although the magnitude and statistical certainty of these associations varied across subgroups. Apatinib primarily exerts antiangiogenic effects by targeting VEGFR‐2. In addition to its effects on tumor angiogenesis, apatinib has been reported to induce autophagy and apoptosis in tumor cells [24]. Apatinib may also reverse multidrug resistance by decreasing the expression of P‐glycoprotein (ABCB1) and breast cancer resistance protein (ABCG2) in vitro, thereby potentially enhancing the drug susceptibility of drug‐resistant cell lines. Preclinical studies have further suggested that apatinib may enhance the antitumor activity of several chemotherapeutic agents, including doxorubicin, taxanes, Vinca alkaloids, and chidamide [25, 26, 27, 28]. These findings provide a biological rationale for further evaluation of apatinib in combination with chemotherapy, rather than definitive evidence of clinical benefit. Several recent studies have evaluated the clinical activity of apatinib, either as monotherapy or in combination with chemotherapy, in pretreated MBC. In these studies, median PFS ranged from 3.3 to 4.6 months with apatinib monotherapy and from 4.4 to 6.9 months with apatinib plus chemotherapy [29]. Although these results appear numerically favorable for apatinib plus chemotherapy, cross‐study comparisons should be interpreted cautiously because of differences in study design and patient populations. Moreover, most prior studies were single‐arm and did not include a chemotherapy‐alone control group, limiting direct assessment of the added value of apatinib. By establishing a matched chemotherapy‐alone control cohort, our study adds real‐world comparative data suggesting that apatinib plus chemotherapy may be associated with longer PFS than chemotherapy alone in selected patients. Anlotinib differs from apatinib, in that it is a multitarget TKI that blocks VEGFR, PDGFR, FGFR, and c‐Kit, thereby exerting broad antiangiogenic activity [30]. In vivo studies have demonstrated the antitumor activity of anlotinib in several human tumor xenograft models, including renal, colon, and lung cancer [31, 32, 33]. By contributing to tumor vascular normalization, anlotinib may enhance the intratumoral delivery of chemotherapy, providing a potential rationale for combination strategies. Anlotinib‐based combination therapy has also been evaluated in several cancers, including esophageal squamous cell carcinoma [34], advanced gastric adenocarcinoma and esophageal squamous cell carcinoma [35], hepatocellular carcinoma [36], and pancreatic cancer [37]. In advanced breast cancer, one clinical study involving patients with locally recurrent or metastatic disease previously treated with anthracycline‐ or taxane‐based chemotherapy reported a median PFS of 3.5 months with eribulin and 5.1 months with eribulin plus anlotinib [38]. Another study involving 153 MBC patients reported a median PFS of 6.0 months with anlotinib‐based treatment [21]. In our study, anlotinib combined with chemotherapy was associated with a numerical trend toward longer median PFS compared with chemotherapy alone. However, this observation should be interpreted cautiously because the anlotinib subgroup was small and the comparison did not reach statistical significance. The median PFS reported in prior studies appears longer than that observed in our cohort; however, these outcomes are not directly comparable because of differences in study design, treatment setting, prior treatment exposure, and disease burden. In our cohort, more than 60% of patients had received four or more prior lines of systemic therapy and over 40% had three or more metastatic sites, indicating a heavily pretreated population with high disease burden. This clinical context may partly explain the shorter absolute PFS observed in our study compared with earlier‐line clinical trials. Bevacizumab was the first approved antiangiogenic monoclonal antibody. By neutralizing VEGF, bevacizumab inhibits angiogenesis, a critical process in tumor growth and metastasis. Previous studies, including E2100 and GIM11‐BERGI, have reported PFS benefit or clinically meaningful activity with bevacizumab‐containing chemotherapy regimens in patients with MBC, mainly in earlier‐line settings [39, 40]. Bevacizumab has also been evaluated in the neoadjuvant setting [41]. For patients with MBC who have progressed after multiple lines of therapy, no standard treatment regimen has been established. This highlights the need for additional treatment strategies in this challenging patient population. In this real‐world cohort, bevacizumab combined with chemotherapy was associated with longer PFS than chemotherapy alone after adjustment for baseline characteristics. These findings suggest that bevacizumab‐based combination therapy warrants further evaluation in selected patients with metastatic breast cancer in later‐line settings. Although multiple chemotherapy agents have been combined with antiangiogenic therapy, limited evidence is available to guide the selection of chemotherapy backbones in this setting. Previous studies of apatinib‐based combinations have evaluated capecitabine [42], vinorelbine [43], and etoposide [44] as chemotherapy partners, with reported median PFS ranging from 3.9 to 6.9 months. In our previous phase II trial involving 40 patients with HER2‐negative advanced breast cancer refractory to anthracycline‐ and taxane‐based chemotherapy, apatinib plus oral vinorelbine yielded a median PFS of 5.2 months [17]. However, heterogeneity in patient populations and study designs limits cross‐trial comparability, and most prior studies were not designed to compare chemotherapy backbones head‐to‐head. Similarly, anlotinib‐ and bevacizumab‐containing regimens have been evaluated with various chemotherapy agents in MBC, but direct comparative data across chemotherapy backbones remain limited. To explore this question, we categorized chemotherapy backbones by mechanism of action into MMI‐based and non‐MMI‐based regimens for exploratory analyses. In our heavily pretreated cohort, exploratory analyses suggested that antiangiogenic therapy combined with MMI‐based regimens, such as taxanes or eribulin, was associated with longer PFS than combinations using non‐MMI backbones, such as capecitabine or gemcitabine. This observed association between MMI‐based backbones and longer PFS may have several biologically plausible explanations. Antiangiogenic agents can transiently normalize abnormal tumor vasculature, thereby improving perfusion, oxygenation, and intratumoral drug delivery. MMI agents exert their cytotoxic effects predominantly during mitosis and may therefore be particularly dependent on adequate drug exposure and tumor proliferative activity. Improved vascular function during the window of vascular normalization may therefore enhance the activity of MMI‐based regimens. In addition, certain MMI agents, such as eribulin, have been reported to induce vascular remodeling and reverse epithelial–mesenchymal transition, which could further support a biological rationale for combining MMI agents with antiangiogenic therapy. In contrast, the interaction between antiangiogenic therapy and non‐MMI agents may differ according to drug‐specific pharmacologic properties and dependence on intratumoral delivery. These biological considerations may partly explain the observed association between MMI‐based backbones and longer PFS; however, this finding remains exploratory and hypothesis‐generating and requires prospective validation. The standard approved doses of bevacizumab commonly used in oncology include 10 or 15 mg/kg, while apatinib is approved for gastric adenocarcinoma and hepatocellular carcinoma at doses of 850 mg once daily and 750 mg once daily, respectively. However, standard‐dose antiangiogenic therapy can be associated with clinically relevant toxicities, which may lead to dose reduction or treatment discontinuation in clinical practice [39, 45]. In this real‐world cohort, many patients had received multiple prior lines of therapy, and antiangiogenic drug doses were selected by treating physicians according to patient condition, prior treatment exposure, and anticipated tolerability. Specifically, 87.6% of patients received bevacizumab at a dose of less than 10 mg/kg, while apatinib was administered at 250 mg (80.4%) or 425 mg (19.6%) once daily. Similarly, 25/26 patients (96.2%) receiving anlotinib were treated with a daily dose of 8 mg or 10 mg, and only one patient received 12 mg daily. Given the retrospective design and the absence of a formal non‐inferiority framework, the dose‐stratified efficacy analyses should be interpreted descriptively. In these descriptive analyses, PFS estimates and AE frequencies varied across dose groups, with numerically lower AE rates observed in the low‐dose groups; however, these comparisons were limited by small subgroup sizes and potential treatment‐selection bias. Several recent studies have explored low‐dose bevacizumab‐ or apatinib‐based regimens and reported encouraging activity with acceptable tolerability in selected patient populations [46, 47, 48]. From a mechanistic perspective, lower‐intensity antiangiogenic therapy may prolong the window of vascular normalization, improve tumor perfusion, and reduce hypoxia‐associated immunosuppression, potentially contributing to a more favorable tumor microenvironment. Such effects may also improve treatment delivery or activity when antiangiogenic therapy is combined with chemotherapy or radiotherapy [49]. Our findings generate the hypothesis that low‐dose antiangiogenic therapy combined with chemotherapy may be feasible in selected heavily pretreated patients, but they do not establish equivalence or non‐inferiority to standard‐dose regimens. Prospective randomized studies are needed to validate these observations and to clarify the biological and clinical implications of dose optimization. The safety profile observed in the antiangiogenic therapy plus chemotherapy group was broadly characterized by hematologic toxicities, gastrointestinal symptoms, aminotransferase elevations, alopecia, peripheral neuropathy, and hypertension. Most events were grade 1/2, suggesting an overall manageable toxicity profile in this real‐world cohort. Previous studies have identified hypertension as a characteristic toxicity of antiangiogenic agents, including apatinib, anlotinib, and bevacizumab, which may require dose modification or treatment interruption in clinical practice [8, 10, 19, 50, 51, 52]. Consistent with this known toxicity profile, hypertension was more frequently reported in the antiangiogenic therapy plus chemotherapy group than in the chemotherapy‐alone group and was numerically most frequent among patients receiving apatinib‐containing regimens. Although the observed incidence of hypertension appeared lower than that reported in some studies using standard‐dose apatinib, such cross‐study comparisons should be interpreted cautiously because of differences in dose intensity, patient populations, AE ascertainment, and follow‐up duration. Proteinuria and hand‐foot syndrome are also commonly reported toxicities of VEGFR‐TKIs such as apatinib and anlotinib. In the present cohort, proteinuria and hand–foot syndrome were uncommon; however, their frequency should be interpreted in light of the retrospective design and the possibility of under‐ascertainment, particularly for mild or asymptomatic events. Bevacizumab‐associated toxicities, including bleeding events and gastrointestinal perforation, were uncommon in this cohort [53, 54]. Hematologic toxicity and peripheral neuropathy remained common, likely reflecting the chemotherapy backbones used in the combination regimens, particularly MMI‐based agents. Overall, the safety profile of antiangiogenic therapy plus chemotherapy appeared manageable in this cohort, but definitive conclusions regarding comparative tolerability are limited by the retrospective nature of AE assessment. Similarly, dose‐stratified safety findings should be considered descriptive and hypothesis‐generating rather than evidence that lower doses reduce toxicity. This study has several limitations that warrant consideration. First, given the retrospective, non‐randomized design, selection bias and residual confounding cannot be fully excluded. Although propensity score matching was used to balance key baseline characteristics between groups, treatment decisions in routine practice were made by treating physicians based on clinical judgment, prior treatment exposure, disease burden, and anticipated tolerability. Accordingly, unmeasured or incompletely captured confounders may have influenced both treatment selection and outcomes, even after additional adjustment for minor residual imbalances in the matched cohort. Therefore, the observed associations should not be interpreted as evidence of causality. Second, the evaluated regimens are not routinely used as standard treatment options in this setting, which limited their uptake and resulted in a relatively small sample size, particularly for the anlotinib subgroup. The limited statistical power may have contributed to the absence of statistical significance in certain exploratory subgroup analyses, and the dose‐stratified analyses were not powered to assess equivalence, non‐inferiority, or dose‐response relationships. In addition, adverse event assessment was based on retrospective medical record review, which may have led to under‐ascertainment of mild, asymptomatic, or subclinical events. Third, multiple chemotherapy backbones were included, each represented by a small number of patients, precluding reliable head‐to‐head comparisons among individual agents when combined with antiangiogenic therapy. To mitigate this heterogeneity, chemotherapy regimens were grouped according to their primary mechanism of action; however, this strategy may have masked regimen‐specific effects. Finally, median OS had not been reached at the time of analysis, precluding reliable assessment of OS. In this heavily pretreated real‐world cohort, PFS was considered an appropriate primary endpoint because it directly captures disease control during the evaluated treatment line and is less affected than OS by heterogeneous post‐progression therapies. Nevertheless, whether the observed association with longer PFS will translate into improved OS remains uncertain, and longer follow‐up is needed to better characterize long‐term survival outcomes. Given these limitations, the findings should be interpreted cautiously as hypothesis‐generating, and prospective randomized studies are warranted to validate these observations. Conclusion In this real‐world analysis, antiangiogenic therapy combined with chemotherapy was associated with longer PFS and a higher ORR than chemotherapy alone, with an overall manageable safety profile in patients with heavily pretreated metastatic breast cancer. Exploratory analyses suggested that MMI‐based chemotherapy backbones were associated with longer PFS when combined with antiangiogenic agents. Exploratory subgroup analyses also suggested that PFS patterns may vary according to prior lines of therapy, with bevacizumab‐based regimens appearing more favorable in patients who had received four or more prior lines and apatinib‐based regimens appearing more favorable in less heavily pretreated patients. Dose‐stratified analyses of bevacizumab and apatinib were descriptive and suggested numerical differences in AE frequencies across dose groups, but these findings do not establish improved tolerability, equivalence, or non‐inferiority of low‐dose regimens. Overall, these findings should be considered hypothesis‐generating and support further prospective evaluation of antiangiogenic therapy combined with chemotherapy in heavily pretreated metastatic breast cancer. Author Contributions Yuanyi Wang: methodology, formal analysis, visualization, writing – original draft, writing – review and editing. Yalong Qi: conceptualization, data curation, investigation, writing – original draft, writing – review and editing. Cheng Zeng: methodology, formal analysis, visualization, writing – review and editing. Ruixia Song: data curation, investigation. Yuhan Wei: data curation, investigation. Haili Qian: investigation, resources. Fei Ma: supervision, project administration, funding acquisition, writing – review and editing. Ethics Statement This study was approved by the National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College review board (approval number: 25/237‐5183). Huanxing Cancer Hospital recognized and relied upon the ethics review opinion issued by the Ethics Committee of Cancer Hospital, Chinese Academy of Medical Sciences for relevant clinical studies. Consent As this was a retrospective study using de‐identified clinical data, the requirement for written informed consent was waived by the ethics committee of the National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College. Conflicts of Interest Professor Fei Ma and Haili Qian are members of Cancer Innovation Editorial Board. To minimize bias, they were excluded from all editorial decision‐making related to the acceptance of this article for publication. The remaining authors declare no conflicts of interest. Supporting information Figure S1: Flow diagram of patient screening, eligibility assessment, treatment group assignment, and 1:1 propensity score matching. Figure S2: Love plot displaying absolute standardized mean differences of baseline covariates before and after PSM, used to assess covariate balance between treatment groups. Figure S3: PFS for patients treated with anlotinib combined with chemotherapy versus chemotherapy alone, respectively. Figure S4: PFS for patients treated with anlotinib combined with different chemotherapy backbone, including MMI and non‐MMI. Figure S5: PFS for patients who had received ≥4 or <4 lines of treatment when treated with anlotinib combination therapy. Figure S6: PFS for patients treated with low‐/higher‐dose antiangiogenic drugs combined with chemotherapy, respectively. Table S1: Chemotherapy regimens administered in the study population. Table S2: Exploratory comparisons with BH–FDR adjustment. Table S3: AEs for patients treated with regimens containing antiangiogenic drugs.