Stained region lymph node biopsy for axillary restaging after neoadjuvant systemic therapy in node-positive breast cancer: a cohort study Authors: Rui Chen, Hailing Zha, Qiannan Zhu, Xinpei Liu, Lifeng Huang, Cuiying Li, Jue Wang, Xiaoming Zha Published: 2026 ABSTRACT AbstractBackgroundAccurate assessment of axillary lymph node status after neoadjuvant systemic therapy (NST) is essential for de-escalating surgery in node-positive breast cancer. Although various axillary surgeries based on sentinel lymph node biopsy (SLNB) are widely reported, they still have limitations in this setting. Since metastatic lymph nodes inherently encompass sentinel nodes, we hypothesized that directly marking and retrieving these nodes could enable reliable restaging without SLNB.Materials and methodsIn this single-arm, prospective cohort study, patients with pathologically confirmed axillary metastasis underwent ultrasound-guided injection of carbon nanoparticle suspension into the most suspicious lymph node(s) before NST. After completing NST, stained region lymph node biopsy (SrLNB) was performed to retrieve the marked nodes, followed by completion axillary lymph node dissection (ALND), without performing SLNB. The detection rate and false-negative rate (FNR) of SrLNB were evaluated to determine its feasibility.ResultsBetween September 2020 and December 2022, 159 patients were included. The axillary pathological complete response (pCR) rate was 40.9% (65/159). SrLNB achieved a detection rate of 100% (159/159) and an overall FNR of 5.3% (5/94). The FNR remained within the 10% safety threshold across all patient subgroups.ConclusionSrLNB is a feasible, economical, and straightforward technique for assessing axillary response after NST. With its high detection rate and acceptable FNR, SrLNB represents an alternative SLNB-independent strategy for axillary restaging in patients with initially node-positive breast cancer.Keywords: Sentinel lymph node biopsy, Stained region lymph node biopsy, Neoadjuvant systemic therapy, Breast cancer Introduction Neoadjuvant systemic therapy (NST) significantly reduces the tumor burden of primary breast cancer, enabling surgery for previously ineligible patients and facilitating breast conservation in those who would otherwise require mastectomy. However, in terms of axillary management, most patients with initially clinically or pathologically positive lymph nodes (cN + or pN +) still undergo axillary lymph node dissection (ALND) in real-world practice, even when no residual tumor is found in axillary lymph nodes after NST (ypN0). The pathological complete response (pCR) rate in the axilla is approximately 40%, which is higher than that in the breast. Therefore, de-escalating axillary surgery is more critical than de-escalating breast surgery. The key challenge lies in identifying reliable methods to accurately assess axillary lymph node status after NST. Sentinel lymph node biopsy (SLNB) is the most readily considered approach; however, for patients with cN + /pN + disease, the detection rate of SLNB after NST is relatively low (80.1%), and its false-negative rate (FNR) is unacceptably high (14.2%) (Kuehn et al. 2013; Nijnatten et al. 2015; Hage et al. 2016). To reduce the FNR, SLNB must meet the following criteria according to the literature and NCCN guidelines: (1) marking metastatic lymph nodes before NST, (2) using dual tracers such as radiocolloids and dyes, and (3) retrieving at least two lymph nodes (Pilewskie and Morrow 2017; Burstein et al. 2019; Cardoso et al. 2019; Barrio et al. 2021). Adherence to these criteria can significantly reduce the FNR and enhance oncologic safety (Cabioglu et al. 2026). Nevertheless, these stringent requirements limit the broader application and accessibility of the procedure, reducing the number of eligible patients. Moreover, the use of radiocolloids is subject to legal restrictions in many countries/regions and cannot be routinely incorporated into SLNB. Given that in cN + /pN + breast cancer, sentinel lymph nodes are part of the metastatic burden and thus image-detectable, we therefore hypothesized that precise pre-NST marking and post-NST retrieval of these nodes might allow axillary restaging without routine SLNB. Accordingly, we developed and evaluated a novel “stained-region” lymph node biopsy (SrLNB) method using carbon nanoparticle tattooing, by assessing its detection rate and false-negative rate (FNR). Methods This was a single-arm, prospective study conducted at a single institution. This study aimed to recruit breast cancer candidates aged 18–70 years diagnosed with cN + and scheduled for NST. Before the NST, the most abnormal lymph nodes were selected for fine-needle aspiration (FNA) or core needle biopsy (CNB) (Fig. 1). Subsequently, a titanium clip (Ultraclip 863,017; Bard Peripheral Vascular Inc. USA) was placed under ultrasound guidance within or adjacent to the biopsied lymph node. Finally, a carbon nanoparticle suspension injection (0.5 mL:25 mg, H20073246, Chongqing Lummy Pharmaceutical Co., Ltd, China) was diluted (concentration: 0.05 mL per 0.15 mL saline) and injected into the cortex of the biopsied lymph nodes under ultrasound guidance to ensure that the biopsied lymph nodes were accurately stained and marked (Fig. 1). Not all abnormal nodes were injected unless multiple, widely distributed abnormal nodes were present; in such cases, carbon nanoparticles were also injected into additional nodes in more distant regions. Procedure of carbon nanoparticle suspension injection and SrLNB a FNA of the abnormal axillary lymph node under ultrasound guidance. b Carbon nanoparticle suspension injection was diluted and injected under ultrasound guidance. c Ultrasound images show that the needle has penetrated the cortex of the lymph node. d For patients who received a modified radical mastectomy, the incision of the breast was used for SrLNB. e For breast-conserving surgery, a separate curved incision was made at the lower margin of the axillary hair area for SrLNB. f Pathological examination was used to identify metastatic cancer cells beside carbon nanoparticles. SrLNB, stained region lymph node biopsy; FNA, fine needle aspiration The NST regimen (eight cycles every 21 days) was based on anthracycline, cyclophosphamide, and taxans. For patients with HER2 + , trastuzumab and pertuzumab were added, while carboplatin was optional for those with HR-/HER2-. Axillary lymph node status was assessed every two cycles of NST. Surgery was performed approximately 3–4 weeks after NST completion. During surgery, SrLNB was conducted according to the carbon nanoparticle and clip-marked regions, followed by ALND. For patients undergoing modified radical mastectomy, the breast incision was utilized for SrLNB. For those undergoing breast-conserving surgery, a separate curved incision was made at the lower margin of the axillary hair line for SrLNB (Fig. 1). After SrLNB, all stained lymph nodes were pathologically examined, and ALND was performed on all patients (Fig. 1). SrLNB was conducted by four surgeons from the same surgical team. The procedure focused on retrieving all lymph nodes within the stained region and any palpably abnormal nodes nearby, without emphasis on identifying the titanium clip and without performing SLNB. The pathological results of SrLNB and other axillary lymph nodes were recorded. Immunohistochemistry or cytological imprinting was not used for the pathological assessment (Fig. 1). Many studies have demonstrated that ultrasound findings correlate with the extent and number of nodal metastases found during surgery. (Chung et al. 2021) Therefore, a slight modification was applied to the clinical nodal (cN) staging definition in this study. Based on ultrasound, cN1, cN2, and cN3 were defined as the presence of 1–3, 4–9, and ≥ 10 abnormal lymph nodes, respectively. Lymph node fusion was recorded as two or more nodes only if clearly distinguishable on ultrasound; otherwise, it was counted as one node. Patients with negative FNA/CNB pathology were excluded from the study. The primary outcome was the feasibility of SrLNB for assessing axillary status after NST, measured by the detection rate and FNR. The detection rate was defined as the successful retrieval of at least one carbon nanoparticle-stained lymph node. A false-negative event was defined as negative pathology on SrLNB but positive pathology on final axillary dissection (ypN +). The study was conducted and reported in accordance with the STROCSS criteria (Mathew et al. 2021). Statistical analysis Data were analyzed using SPSS version 25 for Windows (SPSS Inc., Chicago, IL, USA). Categorical and ordinal variables are presented as numbers and percentages. Descriptive variables were presented as mean ± standard deviation. Pearson’s χ2 or Fisher’s exact test was used for comparing categorical variables. A P value < 0.05 was considered statistically significant. Results Study participants Between September 24, 2020, and December 2, 2022, 159 patients were enrolled. Fifteen patients were excluded: four did not complete the full NST cycles, one did not undergo surgery at our center, three were excluded due to excessive carbon nanoparticle injection causing overstaining early in the study, and seven had negative FNA/CNB pathology. Consequently, 159 patients (mean age 48.5 ± 10.3 years) were included in the final analysis (Fig. 2). The stained region was visually identifiable during surgery in all cases, resulting in a 100% (159/159) detection rate for SrLNB (Fig. 1). Flowchart of this study cN + , clinically assessed positive lymph nodes; pN + , pathologically positive lymph nodes; NST, neoadjuvant systemic therapy; SrLNB, stained region lymph node biopsy; FNA, fine needle aspiration; CNB, core needle biopsy; ALND, axillary lymph node dissection Clinical outcome Patient characteristics are summarized in Table 1. Briefly, the cohort was nearly equally distributed between age < 50 and ≥ 50 years (51.6% vs. 48.4%) and between premenopausal and postmenopausal status (54.7 vs. 45.3%). Most patients presented with cT2 (76.1%) and cN1 or cN2 disease (42.8 and 40.3%, respectively). The most common molecular subtype was HR + /HER2- (39.0%). Axillary staging was primarily performed by FNA (84.3%). After NST, the breast pCR (ypT0/is) rate was 27.7%, and 56.0% of patients achieved a significant breast partial response (pPR). Only one patient exhibited disease progression in the breast. The axillary pCR (ypN0) rate was 40.9% (65/159), which was higher than the breast pCR rate. Regarding SrLNB, the number of retrieved lymph nodes per patient was as follows: 1 node in 7.5% (12/159), 2 nodes in 21.4% (34/159), 3 nodes in 11.3% (18/159), and ≥ 4 nodes in 59.7% (95/159) (Table 1). Table 1General characteristics of enrolled patientsCharacteristics (n = 159)Number of patients%Age (48.5 ± 10.3 years) < 508251.6% ≥ 507748.4%Menstrual status at diagnosis Premenopausal8754.7% Postmenopausal7245.3%Clinically assessed breast tumor size at diagnosis (cT) cT185.0% cT212176.1% cT32616.4% cT442.5%Clinically assessed metastatic axillary lymph nodes by ultrasound (cN) cN1 (1–3)6842.8% cN2 (4–9)6440.3% cN3 (≥ 10)2717.0%Molecular subtypes of breast tumor HR + /HER2-6239.0% HR + /HER2 + 3823.9% HR-/HER2 + 3220.1% HR-/HER2-2717.0%Types of axillary lymph node biopsy Fine needle aspiration13484.3% Core needle biopsy2515.7%Pathological response evaluation of breast tumor after NST Breast pCR (ypT0/is)4427.7% Breast pPR8956.0% Breast pSD2515.7% Breast pPD10.6%Pathological staging of axillary lymph nodes after NST (ypN) ypN06540.9% ypN14830.2% yPN23119.5% ypN3159.4%Number of retrieved SrLN 000.0% 1127.5% 23421.4% 31811.3% ≥ 49559.7%HR, hormone receptor; HER2, human epidermal growth factor receptor-2; pCR, pathologically complete response; ypT0/is, no invasive tumor residuals in the breast; pPR, pathologically partial response; pSD, pathologically stable disease; pPD, pathologically progressive disease; SrLN, axillary lymph nodes retrieved in stained regions General characteristics of enrolled patients HR, hormone receptor; HER2, human epidermal growth factor receptor-2; pCR, pathologically complete response; ypT0/is, no invasive tumor residuals in the breast; pPR, pathologically partial response; pSD, pathologically stable disease; pPD, pathologically progressive disease; SrLN, axillary lymph nodes retrieved in stained regions Statistical analysis revealed that age, menopausal status, and cT or cN stage did not significantly influence the axillary pCR rate (Table 2). However, patients who achieved breast pCR had a significantly higher axillary pCR rate (86.4%, 38/44) compared to those without breast pCR (23.5%, 27/115; P < 0.001). Analysis by molecular subtype showed the highest axillary pCR rate in the HR-/HER2 + subgroup (78.1%, 25/32), followed by HR + /HER2 + (50.0%, 19/38) and HR-/HER2- (48.1%, 13/27). In contrast, the HR + /HER2- subgroup had the lowest axillary pCR rate (12.9%, 8/62; P < 0.001) (Table 2). Table 2Analysis of factors affecting the axillary pCR ratesCharacteristics (n = 159)ypN0, n = 65 (%)ypN + , n = 94 (%)P valueAge (years)0.425 < 5031 (37.8)51 (62.2) ≥ 5034 (45.4)43 (54.6)Menstrual status at diagnosis0.873 Premenopausal35 (40.2)52 (59.8) Postmenopausal30 (41.7)42 (58.3)Clinically assessed tumor size at diagnosis (cT)0.364 cT14 (50.0)4 (50.0) cT251 (42.1)70 (57.9) cT310 (38.5)16 (61.5) cT40 (0)4 (100.0)Clinically assessed metastatic axillary lymph nodes by ultrasound (cN)0.881 cN1 (1–3)29 (42.6)39 (57.4) cN2 (4–9)26 (40.6)38 (59.4) cN3 (≥ 10)10 (37.0)17 (63.0)Pathological response evaluation of breast tumor after NST < 0.001 Breast pCR (ypT0/is)38 (86.4)6 (13.6) Without breast pCR27 (23.5)88 (76.5)Molecular subtypes of breast tumor < 0.001 HR + /HER2-8 (12.9)54 (87.1) HR + /HER2 + 19 (50.0)19 (50.0) HR-/HER2 + 25 (78.1)7 (21.9) HR-/HER2-13 (48.1)14 (51.9)HR, hormone receptor; HER2, human epidermal growth factor receptor-2; pCR, pathologically complete response; ypT0/is, no invasive tumor residuals in the breast Analysis of factors affecting the axillary pCR rates HR, hormone receptor; HER2, human epidermal growth factor receptor-2; pCR, pathologically complete response; ypT0/is, no invasive tumor residuals in the breast Among the 94 patients with a negative SrLNB (SrLNB-), 5 were found to have residual disease in non-stained nodes (ypN +), resulting in an overall FNR of 5.3% (5/94). Subgroup analyses are detailed in Table 3 and Fig. 3. The FNR was 0% in subgroups where 1, 2, or 3 lymph nodes were retrieved via SrLNB. However, in the subgroup where ≥ 4 lymph nodes were retrieved, the FNR was 8.9% (5/56). When analyzed by clinical nodal stage, the FNR was 7.7% (3/39) for cN1, 2.6% (1/38) for cN2, and 5.9% (1/17) for cN3 patients (Table 3, Fig. 3). Table 3Analysis of the FNR of SrLNBGroupsSrLNBypN + ypN0FNR(%)All patients + 8905.3−565 Number of retrieved SrLNs 1 + 900−03 2 + 1800−016 3 + 1100−07 ≥ 4 + 5108.9−539 cN stage cN1 (1–3) + 3607.7−329 cN2 (4–9) + 3702.6−126 cN3 (≥ 10) + 1605.9−110FNR, false-negative rate; SrLNB, stained region lymph node biopsy; SrLN, axillary lymph nodes retrieved in stained region; ypN + , tumor residue in the axillary lymph node after NST; ypN0, no tumor residue in the axillary lymph node after NST; cN, clinically assessed metastatic axillary lymph nodes by ultrasonography Analysis of the FNR of SrLNB FNR, false-negative rate; SrLNB, stained region lymph node biopsy; SrLN, axillary lymph nodes retrieved in stained region; ypN + , tumor residue in the axillary lymph node after NST; ypN0, no tumor residue in the axillary lymph node after NST; cN, clinically assessed metastatic axillary lymph nodes by ultrasonography FNRs of SrLNB were within the safe threshold of 10% in overall patients and subgroups FNR, false-negative rate; SrLNB, stained region lymph node biopsy; SrLN, axillary lymph nodes retrieved in stained region; cN, clinically assessed metastatic axillary lymph nodes by ultrasound A comparative analysis of the median number of retrieved lymph nodes and corresponding FNR is shown in Table 4. The median number retrieved overall was 4 (range 1–12). By cN subgroup, the median was 4 (range 1–10) for cN1, 5 (range 1–12) for cN2, and 4 (range 1–10) for cN3 patients (Table 4). Table 4Analysis for the FNR and No. of retrieved lymph nodescN stageNo. of retrieved axillary lymph nodesFNR (%)MedianRangeAll patients41–125.3cN141–107.7cN251–122.6cN341–105.9FNR, false-negative rate; cN, clinically assessed metastatic axillary lymph nodes by ultrasonography Analysis for the FNR and No. of retrieved lymph nodes FNR, false-negative rate; cN, clinically assessed metastatic axillary lymph nodes by ultrasonography Discussion The axillary pCR rate in this study was 40.9%, consistent with rates reported in the literature (Simons et al. 2019; Li et al. 2024). Notably, this was achieved in a cohort with a relatively high axillary tumor burden, as over half (57.2%) of the patients had cN2–3 disease, supporting the rationale and feasibility of axillary surgery de-escalation. Furthermore, for patients who achieved breast pCR (ypT0/is), the likelihood of axillary pCR reached 86.4%, significantly higher than in those without breast pCR. Consistent with prior report (Kümmel et al. 2014), axillary pCR rates varied significantly by molecular subtype. Rates were higher in the HR + /HER2 + , HR-/HER2 + , and HR-/HER2- subgroups (50.0, 78.1, and 48.1%, respectively). In contrast, the HR + /HER2- subgroup had a significantly lower axillary pCR rate (12.9%; P < 0.001). However, given that HR + /HER2- is the most prevalent subtype in clinical practice, the absolute number of patients achieving axillary pCR in this group remains substantial. Thus, NST offers an opportunity for axillary downstaging across all molecular subtypes in patients initially diagnosed with node-positive disease. SLNB is the most common de-escalation approach, offering a favorable safety profile and fewer upper extremity complications, making it the standard for cN0 patients (Noordaa et al. 2021; Park et al. 2025). It is therefore a natural consideration for cN + patients post-NST. However, studies such as SENTINA have reported suboptimal performance for SLNB in this setting, with a detection rate of 80.1% and an FNR of 14.2% using a single tracer (Kuehn et al. 2013). While subsequent trials using dual tracers and retrieval of ≥ 2 nodes improved outcomes (FNR = 9.1%), the FNR often remained near or above the 10% safety threshold unless specific criteria were met (Boughey et al. 2013). The targeted axillary dissection (TAD) procedure (Caudle et al. 2016; Wild et al. 2024), which combines retrieval of a clip-marked node with SLNB, has demonstrated excellent accuracy (FNR 1.4%) and oncological safety (Kuemmel et al. 2023; Wu et al. 2023). These data confirm that SLNB-based procedures are feasible but require adherence to specific, often stringent, technical requirements. According to Occam’s razor, the need for multiple additional requirements suggests that conventional SLNB may not be the optimal approach for cN + patients. The fundamental premise of SLNB is the identification of the first nodal drainage site. In cN0 patients, the tracer-defined sentinel node and the cancer cell-defined sentinel node are typically congruent, validating the clinical use of tracer-based SLNB (Classe et al. 2019; Kühn et al. 2018). However, in cN + disease, lymphatic drainage can be disrupted by tumor burden. A prior study reported that 23% of clip-marked metastatic nodes were not identified by SLNB (Caudle et al. 2017), indicating that the tracer may not reach the true metastatic sentinel node, leading to false-negative results. This necessitates additional technical steps to reduce the FNR, which in turn limits the procedure’s applicability and accessibility, compounded by regulatory restrictions on radiocolloids in many regions. Since the sentinel nodes are contained within the initially metastatic nodal basin, the therapeutic response of these specific nodes to NST is of direct relevance. We hypothesized that directly marking and retrieving these initially involved nodes (SrLNB) might be more reliable than tracer-based SLNB for assessing response in cN + /pN + patients. Our results support this: SrLNB achieved a 100% detection rate and an FNR of 5.3%, outperforming average reported values for SLNB-based techniques (detection rate ~ 89%, FNR ~ 17%) (Simons et al. 2019). This suggests that SLNB may be omitted in this context. While other studies have also suggested the sufficiency of retrieving marked nodes alone, they typically combined it with SLNB, making it difficult to isolate the contribution of the marked node retrieval (Kuemmel et al. 2022; Khallaf et al. 2020; Montagna et al. 2024). Our study design specifically eliminated SLNB, providing clearer evidence for the standalone utility of the stained-region retrieval concept. Regarding the subgroup analysis of SrLNB, the FNRs for the cN1, cN2, and cN3 subgroups were 7.7, 2.6, and 5.9%, respectively, all within the 10% safety threshold. These findings tentatively suggest that SrLNB is suitable for patients across different nodal disease burdens. Notably, the FNR was 0% in subgroups where only one or two lymph nodes were retrieved, differing from SLNB-based studies which typically require retrieval of at least two nodes to achieve an acceptable FNR. We also observed a safe but elevated FNR (8.9%) in the subgroup where four or more lymph nodes were retrieved. This raises the question of whether retrieving fewer nodes in patients with a heavier tumor burden might increase the FNR, given the known heterogeneous response of metastatic lymph nodes to NST (Glaeser et al. 2019). To address this, we examined the cN3 subgroup, where the median number of retrieved lymph nodes was 4 (range 1–10), fewer than the number of abnormal nodes suggested by ultrasound (cN3 implies ≥ 10 abnormal nodes). The fact that the FNR remained low (5.9%) may be attributed to our strategy of staining the most abnormal node(s) at baseline and, in cases of widely distributed disease, injecting carbon nanoparticles into additional nodes in more distant regions. Currently, there is no consensus on the optimal number of nodes to mark. Some studies advocate marking only the most suspicious node (Caudle et al. 2016), while others propose marking all abnormal nodes (Natsiopoulos et al. 2019). A small-scale study marking all abnormal nodes reported an FNR of 7.1% for the first clipped node alone, which dropped to 0% when a second clipped node was considered (Lim et al. 2020). Collectively, these data suggest that accurate retrieval of the key metastatic node(s) may be more critical than the sheer number of nodes retrieved. Fewer but precisely targeted biopsies align with the goals of precision surgery and help ensure patients derive the full benefit of less invasive axillary procedures. In terms of techniques for marking metastatic lymph nodes, several methods have been explored, including titanium clips (Caudle et al. 2016), magnetic seeds (Kühn et al. 2025), radioactive iodine-125 seeds (Donker et al. 2015; Straver et al. 2010; Simons et al. 2022; Hemert et al. 2026) and carbon nanoparticle suspension (Li et al. 2025). Titanium clips are easy to place but, due to their small size, often require a separate preoperative wire localization for intraoperative identification. To avoid this additional invasive step, alternative materials like magnetic or radioactive markers have been introduced. However, magnetic materials may interfere with MRI-based treatment monitoring, and radioactive seeds are not approved for clinical use in many regions. Furthermore, these markers typically require specialized intraoperative detection equipment, increasing costs and limiting widespread adoption. Considering accessibility and practicality, we employed a carbon nanoparticle suspension in this study. Its key advantages are prolonged retention at the injection site, allowing direct visual identification of the stained region during surgery without the need for additional localization procedures or specialized detection devices. Limitations This study has several limitations. First, it was conducted at a single center. Second, because ALND was performed after SrLNB in all cases, we could not assess the specific impact of SrLNB alone on upper-extremity complications such as lymphedema or mobility impairment. Third, although titanium clips were placed, we did not systematically collect data on the concordance between clip location and the stained nodes, as the clip was often difficult to visualize amidst the dark carbon staining. The 100% detection rate based on the stained region alone suggests the clip may be non-essential for this technique. Fourth, other potential influencing factors, such as precise axillary tumor burden and patient body mass index (BMI), were not evaluated. Finally, while our protocol of injecting the most abnormal node proved simple and preliminarily safe in this cohort, the sample size remains moderate, and the long-term oncological safety of omitting SLNB in favor of SrLNB requires validation in larger, prospective trials. Conclusion Stained region lymph node biopsy (SrLNB) demonstrates a high detection rate and a safe false-negative rate for assessing axillary response after neoadjuvant systemic therapy. As an economical and straightforward technique that requires neither radioactive tracers nor specialized detection equipment, SrLNB is independent of SLNB. Thus, it represents a feasible and self-contained strategy for restaging the axilla in initially node-positive patients. Acknowledgements We would like to thank Dr. Jin Liu (the First Affiliated Hospital of Nanjing Medical University) for statistical analysis. We would also like to thank Editage (www.editage.com) for English language editing. Author contributions C.L. and X.Z.designed this study and conducted quality control on it. R.C. and J.W. wrote the main manuscript text. H.Z. and X.L. prepared Figs. 1–3. Q.Z. and L.H. participated in the analysis of data. All authors reviewed the manuscript. Funding This research was partially supported by the Chinese Society of Clinical Oncology Foundation (Y-HR2020MS-0421, Y-SY201901-0168), Bethune Research Foundation (2024-YJ-156-J-004), Wu Jieping Medical Foundation Special Fund for Clinical Research (320.6750.2023-11-29), and National Natural Science Foundation of China Project (82503790). Data availability No datasets were generated or analysed during the current study. Declarations Conflict of interest The authors declare no competing interests. Ethics approval and consent to participate The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (Approval No. 2020-SR-311) and was conducted in accordance with the Declaration of Helsinki. This study was registered at ClinicalTrials.gov (NCT04482803). All the procedures were followed in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and those of the Helsinki Declaration, and informed consent was obtained from all the participants. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rui Chen, Hailing Zha, and Qiannan Zhu have contributed equally. Contributor Information Cuiying Li, Email: lynx_ko@163.com. Jue Wang, Email: wangjue200011@njmu.edu.cn. Xiaoming Zha, Email: njzhaxm@njmu.edu.cn. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.