Anlotinib combined with neoadjuvant chemotherapy for HR+/HER2− breast cancer (ACNTBC): a prospective, single-arm, single-center phase II clinical study with real-world validation

Abstract

Despite being an option for high-risk or inoperable cases, existing neoadjuvant therapies for hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2−) breast cancer yield suboptimal pathological responses, necessitating novel strategies. This prospective, phase II ACNTBC trial (NCT05558722) investigated the efficacy and safety of combining anlotinib with chemotherapy as a neoadjuvant treatment for HR+/HER2− breast cancer. Enrolled patients received 5 cycles of anlotinib (12 mg qd, d1‒14; q3w) plus 6 cycles of nab-paclitaxel (200 mg/m², q3w), pirarubicin (50 mg/m², q3w), and cyclophosphamide (500 mg/m², q3w). The primary endpoint was the total pathological complete response (tpCR) rate, with the major secondary endpoints including residual cancer burden (RCB) 0/I rate, objective response rate (ORR), and safety. The study enrolled 31 patients, and the ORR was 93.5% (29/31, 95% CI, 78.6–99.2). In the per-protocol set of 28 patients completing treatment and surgery, the tpCR rate was 14.3% (4/28, 95% CI, 4.0–32.7), with RCB 0/I rate of 25.0% (7/28, 95% CI, 10.7–44.9). Grade 3/4 adverse events included hematologic toxicities, hypertension, proteinuria, and headache, all manageable. Post-hoc biomarker analysis showed that high baseline VEGFR2 expression and microvessel density (MVD) were significantly associated with superior efficacy, and the treatment significantly reduced Ki-67 index, VEGFR2 expression, and MVD. Supporting evidence from an exploratory inverse probability of treatment weighting (IPTW) analysis using a real-world cohort receiving chemotherapy alone indicated a potential benefit of the combination over chemotherapy alone. In conclusion, neoadjuvant anlotinib plus chemotherapy demonstrates promising efficacy and manageable safety in HR+/HER2− breast cancer with a high Ki-67 index.

Introduction

As the most common molecular subtype, hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2−) breast cancer accounts for approximately 70% of all invasive breast cancer cases.¹ For breast cancer patients with high-risk features, neoadjuvant therapy (NAT) serves a critical component of systemic treatment. Its value lies not only in tumor downstaging and increasing rates of breast and axillary conservation, but also in serving as a valuable in vivo platform for assessing treatment sensitivity and exploring new therapeutic strategies.^2,3 While pathological complete response (pCR) has been acknowledged as a well-established indicator of favorable long-term prognosis in triple-negative, HER2-positive, and HR+/HER2− breast cancer with a high Ki-67 index, its low incidence in HR+/HER2− disease specifically presents a distinct clinical challenge.^4–6 Identifying more effective strategies to enhance the pCR rate in HR+/HER2− breast cancer patients undergoing NAT remains a critical challenge.

Several strategies have been explored to improve the pCR rate in HR+/HER2− breast cancer. A key approach to enhance the efficacy of NAT for HR+/HER2− breast cancer is neoadjuvant endocrine therapy (NET). Nevertheless, available clinical trial data have not shown that NETs, with or without CDK4/6 inhibitors, significantly increase the response over standard neoadjuvant chemotherapy (NACT).^7–10 Although the combination of targeted agents, such as histone deacetylase or poly ADP-ribose polymerase inhibitors with chemotherapy is another strategy pursued to enhance pCR rates, their efficacy in the neoadjuvant setting remains relatively limited for this population.^11–13 Immune checkpoint inhibitors (ICIs) have demonstrated some activity in increasing pCR rates, yet their effect is modest compared to the significant improvements observed in triple-negative breast cancer (TNBC) and may depend on biomarker expression, such as PD-L1.^14,15 Thus, identifying novel, effective, and tolerable combination regimens remains an urgent clinical challenge.

Given the critical role of angiogenesis in tumorigenesis, targeting this pathway has long been a compelling therapeutic strategy.¹⁶ Bevacizumab, the first widely used anti-angiogenic monoclonal antibody against vascular endothelial growth factor (VEGF),¹⁷ initially showed the potential to increase pCR rates^18,19 but failed to confer a consistent overall survival (OS) benefit and was associated with increased toxicity.^20,21 This discrepancy resulted in the U.S. Food and Drug Administration withdrawing its approval for bevacizumab in breast cancer.²² This experience highlights the demand for anti-angiogenic agents with broader mechanisms of action and improved therapeutic indices. Anlotinib is a novel oral multi-targeted receptor tyrosine kinase inhibitor (TKI) that specifically targets VEGFR, PDGFR, FGFR, and c-kit.²³ Preclinically, anlotinib has been confirmed to promote vessel normalization and reprogram the immunosuppressive tumor microenvironment, enhancing cytotoxic drug delivery, as evidenced by the modulation of the extracellular matrix.^24,25 Although anlotinib has shown antitumor activity combined with chemotherapy in other solid malignancies,^26–28 its role in the NAT of HR+/HER2− breast cancer remains unexplored.

To address this gap, we conducted a prospective, exploratory single-arm, phase II trial of anlotinib combined with neoadjuvant chemotherapy for HR+/HER2− breast cancer (ACNTBC; trial number, NCT05558722) to assess the efficacy and safety of anlotinib. To our knowledge, this is the first prospective trial designed specifically to investigate anlotinib in the neoadjuvant setting for this patient population. In addition to primary and secondary endpoints, we performed post-hoc exploratory biomarker analyses, including demonstrating VEGFR2 expression and microvessel density (MVD), to identify potential predictors of response and elucidate mechanisms of action. Furthermore, to better contextualize our findings, an inverse probability of treatment weighting (IPTW) analysis using real-world data from a contemporaneous cohort receiving standard NACT in our center was also performed. This study aimed to provide both clinical evidence and mechanistic insight into a novel therapeutic strategy that may improve pathological responses in HR+/HER2− breast cancer with a high Ki-67 index.

Results

Patient characteristics

From December 2022 to February 2024, a total of 36 patients with biopsy-confirmed HR+/HER2− breast cancer who provided informed consent to participate were screened for eligibility. Five patients were excluded for failing to meet the eligibility criteria, resulting in 31 patients being finally enrolled and receiving treatment, and were included in the full analysis set (FAS) and safety analysis set (SAS). Three patients discontinued the study: one due to HER2 amplification detected upon re-examination after surgery, one due to adverse events (AEs) (grade 4 neutropenia and leukopenia), and one due to disease progression without surgery. Ultimately, 28 (90.3%) patients underwent surgery and were consequently included in the per-protocol set (PPS). For exploratory sensitivity analysis, a modified full analysis set (mFAS) with 30 patients was defined as the FAS excluding the patient with a major protocol violation (Fig. 1a, b).

Fig. 1.

Trial design and efficacy analysis of ACNTBC trial. a Study design and treatment schedule for ACNTBC trial; b Flow diagram of patient enrollment, treatment, and follow-up; c Different pCR in primary tumors and axillary lymph nodes (ypT0/is ypN0), primary tumors (ypT0/is) and axillary lymph nodes (ypN0) after neoadjuvant treatment in the PPS; d Pie plot of the patient distribution of RCB classes after neoadjuvant therapy in the PPS; e Pie plot of the patient distribution of treatment responses according to RECIST v1.1 criteria in the FAS; f Waterfall plot of the best percent change in tumor burden from baseline based on RECIST v1.1 criteria in the PPS (N = 28), bars are color-coded by RCB class, the x-axis indicates individual patients ordered by the magnitude of tumor reduction and the y-axis shows the best percent changes from baseline, the horizontal dashed lines denote the thresholds for SD, PR, and CR; g Magnetic resonance imaging of representative clinical responses in CR (Patient No. 17, 26th from left in panel f), PR (Patient No. 12, 13th from left in panel f), and SD (Patient No. 11, 1st from left in panel f) patients, with patient numbers assigned in order of enrollment. nab-P-EC nab-paclitaxel, pirarubicin, and cyclophosphamide, FAS full analysis set, PPS per-protocol set, NAT neoadjuvant therapy, RECIST response evaluation criteria in solid tumors, CR complete response, PR partial response, SD stable disease, PD progressive disease, pCR pathological complete response, tpCR total pCR, bpCR breast pCR, apCR, axillary pCR, RCB residual cancer burden. Data cutoff: Dec. 31, 2025

The patient characteristics are reported in Supplementary Table 1. The median age of PPS was 45 years. The majority (60.7%) of patients presented with clinical stage III (according to American Joint Committee on Cancer, 8th edition), and 23 patients (82.1%) had lymph node involvement.

Efficacy and survival outcomes

Of the 28 patients in the PPS, 4 patients [14.3%, 95% confidence interval (CI), 4.0–32.7] achieved total pCR (tpCR, defined as ypT0/Tis, ypN0), further efficacy assessments focusing specifically on primary tumors and axillary lymph nodes showed that 25.0% (7/28, 95% CI, 10.7–44.9) of patients attained breast pCR (bpCR, ypT0/Tis), and an identical proportion achieved axillary pCR (apCR, ypN0) (Fig. 1c). Additionally, the distribution of residual cancer burden (RCB) outcomes was shown in Fig. 1d. Seven patients with RCB 0 or I corresponded to a combined RCB 0/I rate of 25.0% (7/28, 95% CI, 10.7–44.9). In the FAS, the tpCR rate was 12.9% (4/31, 95% CI, 3.6–29.8), and the RCB 0/I rate was 25.8% (8/31, 95% CI, 11.9–44.6). As an exploratory analysis with regard to the mFAS, the tpCR rate was 13.3% (4/30, 95% CI, 3.8–30.7), and the RCB 0/I rate was 23.3% (7/30, 95% CI, 9.9–42.3) (Supplementary Table 2).

Moreover, in the FAS, according to Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1), 5 patients exhibited a clinically complete response (CR), 24 showed a partial response (PR), and 1 had stable disease (SD) (Fig. 1e). This corresponded to an objective response rate (ORR) of 93.5% (29/31, 95% CI, 78.6–99.2) and a disease control rate (DCR) of 96.8% (30/31, 95% CI, 83.3–99.9) (Supplementary Table 2). The best percent change in tumor burden from baseline after neoadjuvant treatment for all evaluable patients in the PPS is depicted in Fig. 1f. Representative imaging cases before and after NAT are presented in Fig. 1g.

In a post‑hoc analysis, Spearman’s rank correlation analysis revealed no significant association between radiological response and pathological response in the PPS [Spearman’s ρ = 0.112 (95% CI, −0.28 to 0.48), p = 0.57] (Supplementary Fig. 1), suggesting that radiological response does not reliably reflect pathological outcomes in this cohort, highlighting the importance of surgical resection for accurate pathological assessment.

Regarding the survival outcome, by the data cutoff (Dec. 31, 2025), the median follow-up time was 27.9 months (range: 2.1–36.9 months), and the estimated 2-year event-free survival (EFS) was 93.5% (95% CI, 85.3–100) for patients in the FAS (Supplementary Fig. 2). No events were observed in the PPS. Owing to the limited follow-up duration and the small number of events, definitive conclusions regarding long‑term survival benefit cannot be drawn, and extended follow-up is ongoing.

Surgical outcomes for the PPS are summarized in Supplementary Table 3. Breast-conserving surgery was performed in 4 patients (14.3%), mastectomy with immediate reconstruction in 3 patients (10.7%), and mastectomy only in the remaining 21 patients (75.0%). As for regional nodal management, 5 patients (17.9%) received sentinel lymph node biopsy, 23 patients (82.1%) underwent axillary lymph node dissection, 7 patients (25.0%) underwent infraclavicular lymph node dissection, 3 patients (10.7%) received supraclavicular lymph node dissection, and 1 patient (3.6%) received internal mammary lymph node dissection. Adjuvant radiotherapy was administered to 23 patients (82.1%).

Analysis of factors associated with efficacy

Our exploratory post-hoc analysis (Fig. 2) did not identify any clinical or pathological subgroup that consistently benefited substantially from anlotinib-based NAT in the tpCR or RCB 0/I rate, despite a numerical trend favoring node-negative (N0) patients. This is likely attributable to the limited overall sample size and the very small N0 subgroup (n = 5).

Fig. 2.

Exploratory subgroup analyses stratified by baseline characteristics. The tpCR rates (a) and RCB 0/I rates (b) in various subgroups based on different baseline patient characteristics. Data are presented as rates (%) with 95% CIs. The dotted line indicates the rates in all patients. tpCR total pathological complete response, RCB residual cancer burden, CI confidence interval, AJCC American Joint Committee on Cancer, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2

Next, we performed a post-hoc analysis on the relationship between several neovascularization-related biomarkers and treatment response. Although we initially sought to assess periodic acid–schiff (PAS)⁺/CD31⁻ vasculogenic mimicry density (VMD), its baseline expression was uniformly low, consistent with the known biology of luminal breast cancer (Supplementary Fig. 3). Consequently, VMD was excluded from further comparative analysis, and we focused on VEGFR2 expression, CD31⁺ MVD, and CD31⁺/CD34⁺ MVD. After applying the Holm‒Bonferroni correction, all associations remained statistically significant (all adjusted p < 0.05). Specifically, patients achieving a favorable pathological response (tpCR or RCB 0/I) had significantly higher baseline levels of VEGFR2 expression, CD31⁺ MVD, and CD31⁺/CD34⁺ MVD than non-responders (Fig. 3a–f). Furthermore, we analyzed the alterations in the Ki-67 index, VEGFR2 expression, and CD31⁺ MVD before and after treatment. Significant decreases in all these biomarkers (all adjusted p < 0.05) could be observed following anlotinib-based NAT, regardless of whether bpCR patients were included (Fig. 3g–i and Supplementary Fig. 4). For the cohort including bpCR patients, the mean Ki-67 index decreased from 42.68% (95% CI, 36.03–49.32) at baseline to 6.79% (95% CI, 3.91–9.67) in surgical samples (Supplementary Fig. 5).

Fig. 3.

The dynamic changes of VEGFR2 and MVD in response to anlotinib and their association with pathological response. VEGFR2 expression levels in baseline biopsies comparing patients with tpCR versus non-tpCR (a) and patients with RCB 0/I versus RCB II/III (b), representative images (left; scale bar, 50 μm) and quantitative analysis of IHC score (right; mean ± SD) are shown; CD31⁺ MVD in baseline biopsies comparing patients with tpCR versus non-tpCR (c) and patients with RCB 0/I versus RCB II/III (d), representative images (left; scale bars, 100 μm) and quantitative analysis of CD31⁺ MVD (right; mean ± SD) were shown; CD31⁺/CD34⁺ MVD in baseline biopsies comparing patients with tpCR versus non-tpCR (e) and patients with RCB 0/I versus RCB II/III (f), representative images by IF staining (left; scale bars, 50 μm and 20 μm) and quantitative analysis of CD31⁺/CD34⁺ MVD (right; mean ± SD) are shown. g The alterations in the Ki-67 index between biopsy and post-operative tumor specimens; h The changes in VEGFR2 expression levels between biopsy and post-operative tumor specimens of patients. Representative images (Patient No.30) are shown left (scale bar, 50 μm), and the graph of IHC score is shown right; i The changes in CD31⁺ MVD between biopsy and post-operative tumor specimens of patients, representative images (Patient No.19) are shown left (scale bar, 100 μm), and the graph of density is shown right. IHC immunohistochemistry, tpCR total pathological complete response, RCB residual cancer burden, MVD microvessel density. p values were adjusted for multiple testing using the Holm‒Bonferroni step-down procedure within each analysis family. Paired pre‑ and post‑treatment samples were available for VEGFR2 and MVD analysis in 27 of the 28 patients in the PPS. Ki‑67 analysis was performed on paired samples from all 28 patients

Safety and feasibility

For patients in the SAS, other frequent treatment-related adverse events (TRAEs) (incidence ≥ 10%) and anlotinib-associated AEs are summarized in Table 1. Alopecia (71.0%), neutropenia (64.5%), leukopenia (58.1%), and nausea/vomiting (51.6%) were the top four most frequent TRAEs of any grade, occurring in more than half of the patients. The most common grade 3/4 TRAEs were hematologic, including neutropenia (16.1%), leukopenia (12.9%), and anemia (6.5%). All patients with grade 3 AEs received appropriate management, and no treatment delays or dose adjustments were needed. One patient experienced grade 4 neutropenia and leukopenia after the third cycle of treatment despite receiving pegylated recombinant human granulocyte colony-stimulating factor (PEG-rhG-CSF) support. The patient was admitted for supportive care, including antibiotics and therapeutic rhG‑CSF. Following physician assessment and in accordance with the patient’s preference, the patient was withdrawn from the trial and underwent surgery ahead of schedule. The most frequent anlotinib-associated AEs were hypertension (32.3%), hand-foot syndrome (29.0%), proteinuria (16.1%), and headache (16.1%). One patient experienced grade 3 hypertension in the second cycle, and another suffered grade 3 proteinuria and headache in the fourth cycle. Anlotinib administration was interrupted until the symptoms resolved to ≤ grade 2, after which the dose was reduced to 10 mg/day and continued until the end of treatment. Three patients experienced treatment delays attributable to the COVID-19 pandemic, all of whom successfully completed all planned cycles thereafter.

Table 1.

Treatment-related adverse events in the safety analysis set population (N = 31)

TRAEs	All grades N (%)	Grade ½ N (%)	Grade 3 N (%)	Grade 4 N (%)
Other frequent TRAEs (≥ 10%)
Alopecia	22 (71.0%)	22 (71.0%)	0 (0.0%)	0 (0.0%)
Neutropenia	20 (64.5%)	15 (48.4%)	4 (12.9%)	1 (3.2%)
Leukopenia	18 (58.1%)	14 (45.2%)	3 (9.7%)	1 (3.2%)
Nausea/vomiting	16 (51.6%)	16 (51.6%)	0 (0.0%)	0 (0.0%)
Oral Mucositis	9 (29.0%)	9 (29.0%)	0 (0.0%)	0 (0.0%)
Fatigue	9 (29.0%)	9 (29.0%)	0 (0.0%)	0 (0.0%)
AST/ALT Increased	8 (25.8%)	8 (25.8%)	0 (0.0%)	0 (0.0%)
Paresthesia	6 (19.4%)	6 (19.4%)	0 (0.0%)	0 (0.0%)
Anemia	6 (19.4%)	4 (12.9%)	2 (6.5%)	0 (0.0%)
Thrombocytopenia	4 (12.9%)	4 (12.9%)	0 (0.0%)	0 (0.0%)
Blood Bilirubin Increased	4 (12.9%)	4 (12.9%)	0 (0.0%)	0 (0.0%)
Diarrhea	4 (12.9%)	4 (12.9%)	0 (0.0%)	0 (0.0%)
Anlotinib-associated adverse events
Hypertension	10 (32.3%)	9 (29.0%)	1 (3.2%)	0 (0.0%)
Hand-foot Syndrome	9 (29.0%)	9 (29.0%)	0 (0.0%)	0 (0.0%)
Proteinuria	5 (16.1%)	4 (12.9%)	1 (3.2%)	0 (0.0%)
Headache	5 (16.1%)	4 (12.9%)	1 (3.2%)	0 (0.0%)
Oral Mucosal Hemorrhage	4 (12.9%)	4 (12.9%)	0 (0.0%)	0 (0.0%)
Rhinorrhagia	3 (9.7%)	3 (9.7%)	0 (0.0%)	0 (0.0%)
Hypothyroidism	3 (9.7%)	3 (9.7%)	0 (0.0%)	0 (0.0%)

TRAEs treatment-related adverse events, AST aspartate aminotransferase, ALT alanine aminotransferase

The selection of the chemotherapy group from the real world and the IPTW analysis

During the same period of our trial (from December 2022 to February 2024), there were 312 patients diagnosed with stage II‒III HR+/HER2− breast cancer receiving NACT in our center. After excluding 35 patients without complete information (including 7 patients without surgery), 111 patients did not meet the inclusion criteria of the trial, and 41 patients did not receive a regimen similar to nab-paclitaxel, pirarubicin, and cyclophosphamide (nab-P-EC). This final exclusion comprised patients with incomplete administration of taxane-anthracycline-cyclophosphamide-based combination chemotherapy (n = 6), those who received alternative chemotherapy agents (n = 5), those with incomplete treatment cycles (n = 19), and those who received concurrent immunotherapy (n = 11). This strict exclusion process was implemented to ensure that the real-world cohort was fully comparable to the PPS cohort in terms of baseline characteristics, treatment regimen, and follow-up completeness, thereby minimizing bias. Consequently, a total of 125 patients were included as the control chemotherapy group (Fig. 4a). The clinicopathological characteristics, chemotherapy regimen, and evaluation of therapeutic efficacy of the 125 patients are shown in Supplementary Table 4. The decrease in Ki-67 indices between the two groups was statistically significant (p < 0.05, Fig. 4b).

Fig. 4.

Treatment efficacy and anti-angiogenic effects of anlotinib plus chemotherapy versus chemotherapy alone. a Flowchart of patient selection for the chemotherapy group from December, 2022 to February, 2024; b Decrease in the Ki-67 index between the chemotherapy group and anlotinib plus chemotherapy group; c Baseline clinicopathological characteristics of the 27 matched patients randomly selected for IHC staining from the chemotherapy group compared with the patients who received anlotinib plus chemotherapy. d Variations in VEGFR2 expression levels before and after treatment between the chemotherapy group (N = 27) and anlotinib plus chemotherapy group (N = 27), representative images (above; scale bar, 50 μm) and quantitative analysis of IHC score (below; mean ± SD) are shown; e Changes in CD31⁺ MVD before and after treatment between the chemotherapy group (N = 27) and anlotinib plus chemotherapy group (N = 27), representative images (above; scale bar, 100 μm) and quantitative analysis of density (below; mean ± SD) are shown. HR hormone receptor, HER2 human epidermal growth factor receptor 2, BC breast cancer, NACT neoadjuvant chemotherapy, TAC taxanes, anthracycline, and cyclophosphamide, RCB residual cancer burden, IHC immunohistochemistry, MVD microvessel density, SMD standardized mean difference. ns not significant; * p < 0.05

There were several differences in clinicopathological characteristics between the anlotinib and chemotherapy groups before IPTW, while after adjusting via IPTW analysis, no significant differences were observed in baseline characteristics between the two groups [all standardized mean differences (SMDs) < 0.1] (Supplementary Table 5). After balancing baseline characteristics, we found that patients in the anlotinib group had a significantly higher likelihood of achieving tpCR, RCB 0/I and bpCR than those receiving chemotherapy only, with risk ratios for benefit of 3.33 (95% CI, 1.26–8.79), 2.31 (95% CI, 1.15–4.64) and 2.83 (95% CI, 1.32–6.06), respectively (Table 2). The median follow-up time for patients in the PPS of anlotinib group was 28.1 months (range: 22.2–36.9 months) and for patients in the control group was 29.3 months (range: 7.1–37.0 months). The corresponding estimated 2‑year EFS rates were 100.0% (95% CI, 87.7–100.0) and 96.8% (95% CI, 93.7–99.9), respectively (Supplementary Fig. 6).

Table 2.

Comparison of pathological response status between anlotinib and chemotherapy groups after adjustment by IPTW

Pathological evaluation	Anlotinib + Chemotherapy	Chemotherapy	Risk ratioa (95% CI)	p value
Weighted number	28.6	124.8
tpCR Status, N (%)			3.33 (1.26–8.79)	0.035*
tpCR	6.1 (21.3%)	8.0 (6.4%)
non-tpCR	22.5 (78.7%)	116.8 (93.6%)
RCB status, N (%)			2.31 (1.15–4.64)	0.044*
RCB 0/I	9.0 (31.5%)	17.0 (13.6%)
RCB II/III	19.6 (68.5%)	107.8 (86.4%)
bpCR status, N (%)			2.83 (1.32–6.06)	0.018*
bpCR	8.5 (29.7%)	13.1 (10.5%)
non-bpCR	20.1 (70.3%)	111.7 (89.5%)
apCR status, N (%)			1.23 (0.68–2.21)	0.562
apCR	9.7 (33.9%)	34.3 (27.5%)
non-apCR	18.9 (66.1%)	90.5 (72.5%)

tpCR total pathological complete response, RCB residual cancer burden, bpCR breast pathological complete response, apCR axillary pathological complete response, CI confidence interval

^*p < 0.05

^aThe risk ratio represents the ratio of the probability of achieving a favorable pathological response in the anlotinib group to that in the chemotherapy group

Finally, for each patient in the anlotinib group, a matched control was randomly selected from a pool of eligible candidates with similar baseline characteristics. In total, 27 matched control patients were included. After matching, all measured baseline characteristics were well balanced between the two groups (all SMDs <  0.1, Fig. 4c). We then examined VEGFR2 expression level and CD31⁺ MVD in tissue samples before and after treatment. As shown in Fig. 4d, e, before treatment, there were no significant differences in VEGFR2 expression level or CD31⁺ MVD between the two groups. However, after treatment, both biomarkers were significantly lower in the anlotinib group compared with the control group, which suggested that anlotinib significantly inhibited VEGFR2 expression and CD31⁺ MVD.

Discussion

ACNTBC trial demonstrates that the combination of anlotinib with chemotherapy as NAT yields promising efficacy and a manageable safety profile in high-Ki-67 HR+/HER2− breast cancer patients, making this combination regimen a potential option. The significance of NAT for breast cancer is evolving, not only for tumor downstaging but also for providing early information on the response to treatment, which can serve as a prognostic factor for long-term outcomes.^29,30 Although the long-term benefit of pCR is less pronounced in HR+/HER2− breast cancer, it confers a favorable prognosis, especially in patients with intermediate to high risk.^4,6 And pCR is still recommended as the validated surrogate endpoint for neoadjuvant clinical trials by NeoSTEEP.³¹ However, the historically low pCR rate in HR+/HER2− breast cancer is a persistent and critical challenge in breast oncology. Generally, the pCR rate for HR+/HER2− breast cancer patients is 7.5–16.2%, which is lower than that in other subtypes.⁴ While multiple strategies, including NETs (with or without CDK4/6 inhibitors),^7–10 combinations with other targeted agents,^13,32 and more recently, ICIs^14,15 have been explored, the improvement in pCR rates for HR+/HER2− breast cancer remains modest. Previous studies have indicated that the anti-angiogenic agent bevacizumab could increase the pCR rate,^18,19,33 but its application has been limited due to a failure to translate into survival benefits and additional toxicity concerns.^20,21 Thus, a key question arises, is there an anti-angiogenic agent with a broader mechanism of action capable of overcoming the resistant tumor microenvironment in HR+/HER2− breast cancer?

Our study provided new insight and evidence addressing this question. For the first time in a prospective clinical design, we have demonstrated that combining anlotinib, a novel multi-targeted TKI with chemotherapy might represent a promising NAT option. The observed tpCR rate in our PPS was 14.3%; while being numerically modest, it is important to note that our trial enrolled more patients with stage III breast cancer (60.7% in the PPS) than other clinical trials, including I-SPY2 (33.2% stage III)³² and KEYNOTE-756 (37.2% stage III in the pembrolizumab group).¹⁴ This heavier baseline tumor burden may also explain the relatively lower rates of breast-conserving surgery and axilla-conserving procedures observed in our study, which were influenced by both clinical feasibility and patient preference. Beyond the tpCR rate, the favorable RCB 0/I rate³⁴ of 25% and the profound reduction in the Ki-67 index³⁵ in 96.4% (27/28) of patients also constitute effective surrogate metrics that correlate with treatment response and prognosis.

More importantly, in exploratory hypothesis-generating analyses, we observed that a high baseline VEGFR2 expression level and high MVD were associated with a higher likelihood of better pathological responses, providing potential predictive biomarkers for identifying a “vessel-dependent” subgroup most likely to benefit. Meanwhile, the significant post-treatment decreases in the Ki-67 index, VEGFR2 expression, and MVD constitute direct pharmacodynamic evidence of anlotinib’s dual anti-proliferative and anti-angiogenic effects in vivo. Furthermore, we used IPTW to generate a balanced comparable real-world control cohort. Compared with patients who only received chemotherapy, those receiving anlotinib achieved superior pathological responses and more pronounced reductions in Ki-67, VEGFR2, and MVD. While this IPTW‑adjusted comparison cannot establish definitive causality, it provides meaningful contextual support and strengthens the validity of our findings.

The unique therapeutic value of anlotinib stems from its mechanism as a multi-targeted TKI. Unlike single-targeted agents such as bevacizumab, anlotinib potently and simultaneously inhibits multiple targets, including VEGFR1-3, PDGFR-α/β, FGFR1-3, and c-Kit. By blocking the crosstalk among VEGF/PDGF/FGF signaling pathways, it more comprehensively inhibits tumor angiogenesis and reduces the risk of resistance due to bypass activation.²³ Additionally, preclinical studies have confirmed that anlotinib promotes vascular structural normalization, improves blood perfusion in the tumor microenvironment, and reduces interstitial fluid pressure, thereby enhancing the delivery and penetration of chemotherapeutic drugs within tumor tissue.²⁵ Our observed universal decrease in the Ki-67 index and the correlation between vascular marker reduction and response clinically validate this preclinical rationale. Multiple clinical studies have also validated the anti-tumor activity of anlotinib across various tumor types, including colorectal cancer,²⁶ soft tissue sarcoma,³⁶ and head and neck squamous cell carcinoma.³⁷ As for TNBC, the neoadjuvant regimen of anlotinib plus chemotherapy was associated with a superior pCR rate of 51.52%.³⁸ Additionally, the safety profile observed was manageable and consistent with the known toxicities of the individual agents, with no new safety signals identified. This favorable risk-benefit profile is crucial for a treatment intended for the curative setting.

This study provides several promising implications for translational development and future research. First, the preliminary identification of baseline VEGFR2 expression and MVD as potential predictive biomarkers warrants prospective validation. A biomarker-guided therapeutic strategy demonstrates potential for optimizing patient selection and maximizing clinical benefit while sparing potential non-responders from unnecessary toxicity. Moreover, the inhibition of anlotinib on the PDGFR/FGFR pathways can modulate the activity of cancer-associated fibroblasts and reduce the infiltration of myeloid-derived suppressor cells, thereby reversing the immunosuppressive microenvironment and enhancing anti-tumor immune responses.²⁴ This mechanism provides a theoretical basis for combining anlotinib with ICIs, making the “chemotherapy + anti-angiogenesis + immunotherapy” triple regimen a possibility. This triple regimen approach has been explored in TNBC and significantly improves the pCR rate even in PD-L1-negative patients.³⁹ Our study has preliminarily demonstrated the feasibility and efficacy of the combination of anlotinib and chemotherapy as NAT for HR+/HER2− breast cancer patients. Based on the synergistic regulatory potential of anlotinib on the tumor vascular and immune microenvironment, further exploration of the triple therapy will be a highly promising research direction in the future, with the potential to overcome the limitations of ICIs in this subtype. According to the current findings from our study and other relevant evidence, future research could not only focus on conducting a randomized phase III trial to directly validate the efficacy of anlotinib on HR+/HER2− breast cancer, with multi-omics analyses on serial biopsy samples to systematically elucidate mechanisms of response and resistance, but also explore the feasibility of treatment strategies based on anlotinib and other similar multi-targeted TKIs in combination with chemotherapy, immunotherapy, and other therapeutic approaches.

We acknowledge the limitations of this study. First, although the strict criteria for the real-world control cohort ensured baseline comparability with the PPS cohort of the trial, and IPTW was applied to balance baseline covariates, residual confounding might still exist due to variations in specific chemotherapy protocols and other unmeasured factors. More fundamentally, the inherent single-arm, non-randomized design and the limited sample size affect the reliability of efficacy. The contribution of anlotinib cannot be fully determined. Subsequent validation in larger randomized controlled trials is warranted. Second, the 3-weekly nab-P-EC regimen for 6 cycles was selected to synchronize with anlotinib’s dosing schedule, differing from the more widely adopted dose-dense sequential regimens. This difference may compromise efficacy and increase the incidence of AEs. Meanwhile, the chemotherapeutic agent pirarubicin, which had lower cardiotoxicity than doxorubicin, as well as the targeted agent anlotinib, were not approved worldwide, which may limit the immediate generalizability of our findings to other regions. Third, although we reported encouraging EFS, the current 2-year follow-up period is insufficient to assess long-term efficacy in HR+/HER2− breast cancer patients, and the absence of events in the PPS limits the statistical power for comparative analyses. Consequently, despite the numerical difference, the 2-year EFS rates did not reach statistical significance compared to the IPTW-weighted control group. The impact of this anlotinib-based NAT on EFS and OS remains to be determined with longer follow-up. Fourth, limitations in tissue availability constrained our biomarker analyses. Tissue samples before and after NAT were not available for all patients, and resources permitted assessment in only a subset of the real-world cohorts. Furthermore, our exploration of biomarkers was restricted to VEGFR2 expression and MVD and was conducted post hoc. Consequently, all biomarker correlations must be considered hypothesis-generating and need prospective validation. Finally, the absence of high-throughput sequencing may preclude a deeper investigation into the molecular effects of anlotinib and the identification of a more precisely defined beneficiary population.

In conclusion, anlotinib combined with nab-P-EC as NAT shows promise in improving the pCR rate with a manageable safety profile in stage II‒III HR+/HER2− breast cancer. Further large-scale validation of these findings is necessary to establish its role in clinical practice. The findings also suggest VEGFR2 expression and MVD as potential predictive biomarkers for treatment efficacy. Moreover, our study provides a robust rationale for continued exploration of chemotherapy combined with multi-targeted anti-angiogenic TKIs as a neoadjuvant strategy for HR+/HER2− breast cancer patients.

Materials and methods

Ethics approval and consent to participate

This study was conducted in accordance with the World Medical Association Declaration of Helsinki. The trial protocol and all subsequent amendments were approved by Review Approval Document of the Medical Ethics Committee of the First Affiliated Hospital of the Air Force Medical University (Number: KY-20202075-F-2). Written informed consent was obtained from all patients. All information collected from study participants was kept confidential by assigning a random number to each patient, following institutional guidelines and policies. Patients received no financial or other compensation other than the treatment for participating in the study. For patients in the chemotherapy-only group, data collection was approved by the institution’s ethics committee (Review Approval Document of the Medical Ethics Committee of the First Affiliated Hospital of the Air Force Medical University, Number: KY-20232266-C-1).

Study design and population

This trial is an investigator-driven, prospective, single-arm, single-center, phase II study of anlotinib combined with nab-P-EC as neoadjuvant chemotherapy with long-term follow-up in HR+/HER2− breast cancer patients. Sample size calculation was performed using NCSS PASS 2021 statistical software (LLC, Kaysville, Utah, USA, ncss.com/software/pass). The exact (Clopper‒Pearson) confidence interval method was employed, with a tpCR rate of 13%, according to the upper end of the published historical range,⁴ aiming to limit the total width of the 95% CI to ≤ 30%. The calculation required 25 evaluable patients to achieve a 95% CI with a total width of 29.5% (3.0–32.5%). Accounting for a 20% dropout rate, 31 patients with HR+/HER2− breast cancer were planned for enrollment.

Patients were recruited from the First Affiliated Hospital of the Air Force Medical University from December 2022 to February 2024. Eligible patients were female aged 18–70 years with newly diagnosed, previously untreated, histologically confirmed stage II‒III HR+/HER2− invasive breast cancer and were considered candidates for NAT based on local multidisciplinary evaluation. Other key inclusion criteria were as follows: (1) stage II‒III (tumor size ≥ 2.0 cm regardless of lymph node status or positive lymph node status regardless of tumor size); (2) Ki-67 index of the primary breast tumor > 20%; (3) Eastern Cooperative Oncology Group (ECOG) performance status of 0–1 and normal bone marrow, liver and renal function. The key exclusion criteria included stage IV breast cancer and contraindications to anlotinib or chemotherapy.

Treatment

All enrolled patients received 5 cycles of anlotinib (12 mg qd, d1–14; q3w) plus 6 cycles of nab-paclitaxel, pirarubicin, and cyclophosphamide (nab-P-EC) (200 mg/m² nab-paclitaxel, 50 mg/m² pirarubicin, and 500 mg/m² cyclophosphamide all administered every 3 weeks). Upon completion of the NAT, patients received surgery within 2–6 weeks. The choice of surgical procedure was determined by the surgeon’s clinical judgment and patient preference. Options for the affected breast included breast-conserving surgery and mastectomy with or without immediate reconstruction. Regional nodal management included sentinel lymph node biopsy, axillary lymph node dissection, and extended lymph node dissection (infraclavicular, supraclavicular, or internal mammary) when clinically indicated. NAT and surgery were performed at the First Affiliated Hospital of the Air Force Medical University, while anlotinib was administered orally by the patient at home. Regular reminder messages were sent to patients to enhance adherence. Adjuvant therapy (including radiotherapy and endocrine therapy) was administered at the discretion of the physician according to the standard clinical guidelines.

For the management of AEs during the trial, all grades of AEs were monitored and treated per protocol. AEs graded ≥ 2 were managed according to the American Society of Clinical Oncology and Chinese Society of Clinical Oncology guidelines. Dose adjustments, including interruptions and reductions, were allowed for the management of TRAEs. If grade 3/4 anlotinib-related AEs occurred, the dose could be lowered to 10 mg/day or 8 mg/day. PEG-rhG-CSF was routinely administered after each cycle of chemotherapy according to institutional guidelines for high‑risk regimens.

Assessments

Pretreatment biopsy specimens were histopathologically evaluated to confirm the diagnosis and to compare baseline tumor morphology with any residual tumor present after treatment. Pathologic response was assessed in the post-treatment surgical specimens according to standard pathologic evaluation recommendations and reviewed by a dedicated pathologist to ensure reporting consistency.

At baseline, patients underwent radiological tumor assessment, including ultrasound imaging, magnetic resonance imaging (MRI), with mammography as an optional component when clinically indicated. In addition, physical examination, routine work-up staging, assessment of vital signs, laboratory tests, and 12-lead electrocardiography were also carried out. Laboratory tests, vital sign assessments, and physical examinations were performed before and after every cycle or before surgery. Tumor response was assessed every 2 cycles using ultrasonography and before surgery using MRI. The objective response for all assessment modalities was defined per RECIST v1.1, with confirmation by both the principal investigator and radiology staff. Long-term follow-up schedule for disease status and survival entailed evaluations every 3 months in the first 2 years after surgery, every 6 months for the subsequent three years, and then annually thereafter until the 10th year.

Endpoints

The primary endpoint of the study was tpCR, defined as the absence of invasive cancer cells in the breast and axillary lymph node (ypT0/Tis, ypN0). The key secondary endpoints included ORR, the rates of RCB 0/I, bpCR (ypT0/Tis), and apCR (ypN0), and safety. The ORR was defined as the proportion of patients achieving the best overall CR or PR according to RECIST v1.1. All CR and PR responses were confirmed by three independent radiologists. Additionally, this study also regarded EFS as the secondary endpoint, which was defined as the time from the initiation of treatment until the first occurrence of any of the following events: disease progression precluding surgical intervention, local or distant recurrence, and death from any cause. For safety evaluation, the severity grade (according to National Cancer Institute Common Terminology Criteria for Adverse Events, v 5.0) and the relationship to study treatment of AEs were assessed by physical examination and laboratory tests before and after every cycle, during follow-up visits, and upon indication by symptoms. Other frequent AEs were defined as AEs mostly related to chemotherapy occurring in ≥ 10% of patients, while anlotinib-associated AEs were defined as AEs potentially related to anlotinib.⁴⁰

Analysis sets

The SAS included all patients who received at least one dose of treatment, which in our study was also identical to the FAS. The PPS was defined as all patients who completed the predefined neoadjuvant anlotinib plus nab-P-EC, underwent radical surgery, and had evaluable pathological specimens. Efficacy was analyzed in the PPS and in the FAS for sensitivity analysis. In the FAS, the endpoints were analyzed using a modified intention-to-treat (ITT) principle, whereby for patients who did not undergo surgery, their pCR status was conservatively regarded as “non-pCR,” and their RCB status was conservatively regarded as “RCB III”.

Furthermore, we defined an mFAS, which excluded one patient with a major protocol deviation (postoperative confirmation of HER2 positivity) from the FAS, applying the same conservative principle to handle patients without surgery. This set was used for exploratory sensitivity analysis.

Immunohistochemistry (IHC), PAS staining, and immunofluorescence (IF)

For IHC, the tumor tissues were fixed in formalin and processed for paraffin embedding. The sections were first dewaxed and rehydrated in xylene and graded alcoholic solutions and then treated with high-temperature antigen retrieval in citrate buffer (pH 6.0), followed by blocking with ready-to-use normal animal serum (Boster Biological Technology, AR0009). The primary antibody for VEGFR2 (Santa Cruz, sc-6251, 1:50) was incubated at 4 °C overnight, and secondary antibodies were incubated at room temperature for 30 min. After washing, the sections were stained with DAB peroxidase substrate kit (Maixin Biotech, DAB-1031) until the desired intensity was achieved. The IHC score was calculated using a semi-quantitative method. Specifically, five random fields of view were selected for assessment. In each field, two parameters were evaluated, staining intensity (graded as follows: no staining, 0 points; pale yellow, 1 point; brownish yellow, 2 points; dark brown, 3 points) and percentage of positive cells (0%, 0 points; 1–25%, 1 point; 26–50%, 2 points; 51–75%, 3 points; 76–100%: 4 points). A composite score for each field was obtained by multiplying the intensity score by the proportion score. The final IHC score was determined by calculating the average of the scores from the five fields.

For tumor vessels, PAS staining (Beyotime, C0142M) was performed after CD31 (Santa Cruz, sc-13537, 1:50) IHC staining according to the manufacturer’s protocol. Tumor vessel density (including MVD and VMD) was calculated by the average number of vessels in 5 random fields.

For IF staining, after dewaxed and rehydrated, the sections were subjected to antigen retrieval in Tris-EDTA buffer (pH 9.0) and blocked. Then the sections were incubated with a mixture of directly conjugated primary antibodies CD31 (Proteintech, CL488-66065, 1:200) and CD34 (Proteintech, CL594-60180, 1:50) for 1.5 h at room temperature in the dark. After washing, the cell nuclei were counterstained with DAPI, and sections were mounted with an anti-fade mounting medium containing DAPI (Beyotime, P0131). Images were captured and visualized by a slide scanner (Olympus VS200). CD31⁺/CD34⁺ MVD was calculated by the average number of vessels in 5 random fields.

Of the 28 patients in the PPS, paired pre‑ and post‑treatment tissue samples were available for VEGFR2 and MVD analysis in 27, with one patient excluded due to insufficient tissue quality. For patients who achieved bpCR, the tumor bed was identified by an experienced pathologist based on characteristic histological features (for example, fibrosis, inflammatory cell infiltration, foam cell aggregation, and calcification). IHC scoring for VEGFR2 and quantification of CD31⁺ MVD were performed on the tumor bed area using the same protocol applied to non-bpCR cases. In such patients, no residual invasive tumor cells were detected, with the Ki-67 index of 0. Meanwhile, VEGFR2 expression levels were extremely low, and CD31⁺ MVD was negligible, resulting in both the IHC score and CD31⁺ MVD being recorded as 0 post-treatment.

Real-world data collection and IPTW analysis

We identified 312 patients with primary HR+/HER2− stage II–III breast cancer from the database at the First Affiliated Hospital of the Air Force Medical University between December 2022 and February 2024. To construct a real-world comparator group that would mirror the stringent eligibility and treatment completion standards of the PPS of the anlotinib trial, we applied the following sequential exclusion criteria to the initial cohort: (1) the relevant inclusion and exclusion criteria identical to those of the trial; (2) the necessity to have undergone surgical resection; (3) the requirement to have completed all the planned neoadjuvant chemotherapy cycles and the regimen consisting exclusively of taxanes, anthracyclines, and cyclophosphamide. Following this screening process, a control group of patients largely comparable in clinical characteristics to those in the anlotinib treatment group was obtained. The balance in covariates was assessed using the SMD approach.^41,42 Baseline characteristics were considered imbalanced between the two groups if SMD > 0.1. Baseline differences between the two groups were adjusted using the IPTW method via R library ipw (version 1.0–11).^43,44 To obtain the propensity score, we included age, clinical T stage, clinical N stage, ER status, PR status, HER2 status, and Ki-67 index in the models. Treatment efficacy was comparable when there were no differences in all covariates. EFS was assessed in the weighted dataset using the weighted Kaplan‒Meier method, and differences between groups were evaluated with the weighted log-rank test.

Statistical analysis

R software (version 4.5.0) was used for statistical analyses. Statistical significance was set as p < 0.05 unless otherwise specified. Student’s t-test or Mann‒Whitney U test was appropriately used to compare biomarkers between different clinical outcome groups. The 95% CIs for response rates and other proportions were calculated using the Clopper‒Pearson exact method. Paired t-test was used for comparing pre- and post-treatment parameters. A post‑hoc exploratory analysis was performed using Spearman’s rank correlation coefficient to evaluate the association between radiological response (assessed according to RECIST v1.1) and pathological response (graded according to RCB class).

Two families of biomarker analyses were conducted post-hoc and considered exploratory. The first involved 6 group comparisons, including assessments of baseline VEGFR2 expression, CD31⁺ MVD, as well as CD31⁺/CD34⁺ MVD between the response and non-response groups for tpCR and RCB 0/I endpoints. The second involved 6 groups, comparing changes in Ki-67, VEGFR2, and CD31⁺ MVD following treatment with or without bpCR patients. To control the family-wise error rate for multiple testing, p-values within each distinct family of comparisons were adjusted separately using the Holm‒Bonferroni step-down procedure. Adjusted p-values < 0.05 were considered statistically significant. All findings from these analyses should be interpreted as hypothesis-generating.

Author contributions

J.F. and T.W. conceived and designed the clinical trial; H.M., Y.W., M.L., and J.K. recruited patients. J.D. and H.Z. collected the data. C.F. and J.K. verified the data. J.D., H.Z., and L.S. contributed to the analysis and interpretation of data. H.Z., H.C., and S.W. collected tissue samples and performed the staining of sections. L.Z. provided expert guidance in the pathological evaluation of tissue sections. J.D. drafted the manuscript, and all authors have read and approved the article. T.W. contributed to the administrative support and was responsible for the decision to submit the manuscript. All authors have read and approved the article.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to ethical restrictions, patient privacy regulations, and institutional data governance policies. However, de-identified participant data may be made available by the corresponding author upon reasonable request, subject to review and approval of a formal research proposal by the institutional ethics committee.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41392-026-02788-0.