Abstract
Background
In the clinical management of patients with invasive breast cancer (IBC), the precise identification of axillary lymph node metastasis (ALNM) is of paramount importance for guiding axillary surgery and formulating corresponding treatment strategies. Currently, the clinical need to accurately predict the risk of ALNM in patients with IBC, so as to determine whether axillary surgery (including sentinel lymph node biopsy) can be safely omitted, has not been fully met.
Methods
The clinical data of 454 patients with IBC were retrospectively analyzed, and the patients were randomly divided into a training group and a validation group at a ratio of 3:1. Independent predictors related to ALNM were identified through univariate and multivariate Logistic regression analysis. Accordingly, a nomogram model integrating clinicopathological, ultrasound and serological indicators along with additional clinical parameters was constructed and validated by receiver operating characteristic (ROC) curve analysis, calibration plots, and decision curve analysis (DCA). The primary outcome measure was the incidence of ALNM.
Results
The ALNM rates were 31.4% and 27.4% in the training and verification groups, respectively. Multivariate regression analysis indicated that tumor size, circumscribe margin, ultrasonic lymph node status, WBC, ER, and BI-RADS were independent risk factors for ALNM in patients with IBC. In both the training group and the validation group, the nomogram exhibited modest predictive performance (training group AUC = 0.741; validation group AUC = 0.705). The Ultrasonic lymph node status (normal vs. abnormal) alone yielded an AUC of 0.620 in the training cohort and 0.627 in the validation cohort for predicting ALNM. In comparison, our nomogram achieved significantly higher AUCs (training: 0.741, P = 0.0017; validation: 0.705, P = 0.1772; DeLong test). Additionally, the nomogram demonstrated satisfactory calibration and clinical utility as evidenced by the calibration curve and DCA.
Conclusion
By integrating clinicopathological, ultrasonic and serological indicators, this nomogram can enhance the accuracy of predicting ALNM in patients with IBC. However, given its moderate discriminative ability, the model can support preoperative risk stratification but cannot replace standard axillary staging procedures (including sentinel lymph node biopsy). Our findings provide a basis for decision-making regarding individualized axillary lymph node surgery.
Keywords: ALNM, clinicopathological indicators, invasive breast cancer, nomogram, serological indicators, ultrasound indicators
Introduction
Breast cancer is the most common malignant tumor in women worldwide, and in recent years, with the advancement of treatment, the survival rate of patients has improved significantly (1). However, recurrence and distant metastasis remain the leading causes of death in patients. In patients with IBC, ALNs are the main route for breast cancer cell metastasis, and approximately 70% of breast lymph fluid passes through the axillary area. Therefore, the status of ALNs is a key factor in assessing cancer staging and developing treatment plans, and early identification of ALNM is essential for treatment (2, 3).
Currently, sentinel lymph node biopsy (SLNB) has become the standard of care for axillary staging in clinically node-negative breast cancer patients, as recommended by major guidelines including the National Comprehensive Cancer Network (NCCN) and European Society for Medical Oncology (ESMO) (4, 5). Compared with axillary lymph node dissection (ALND), SLNB offers comparable survival outcomes while significantly reducing surgical morbidity (6). A recent meta-analysis of seven randomized controlled trials demonstrated that SLNB is associated with a 65% lower risk of lymphedema compared with ALND, with no significant differences in 5-, 8-, or 10-year overall survival, disease-free survival, or recurrence rates (6). However, SLNB is not without limitations. The procedure carries a false-negative rate of approximately 5-10% in clinically node-negative patients, which can be reduced to below 10% through dual-tracer mapping techniques and removal of three or more sentinel nodes (7–9). Potential complications, though substantially less frequent than with ALND, include temporary sensory disturbances (reported in approximately 10% of patients), transient shoulder mobility restriction (approximately 7%), and mild lymphedema (approximately 7-8%) (7). These considerations underscore the clinical need for accurate preoperative prediction of axillary lymph node metastasis to optimize patient selection for SLNB and potentially avoid unnecessary axillary surgery altogether.
With the continuous advancement of medical technology, it is particularly urgent to develop non-invasive methods to predict ALNM in IBC patients. As a routine screening method for breast cancer, breast ultrasound not only performs well in evaluating the size and location of tumors and delineating the internal structure of lymph nodes, but also shows great potential in diagnosing early metastatic lesions (10). Compared with MRI or other radiomics techniques, breast ultrasound is a non-invasive diagnostic method with low cost, convenient operation, strong repeatability, and no radiation (11). Some studies have also confirmed that the clinicopathological characteristics, including age, tumor size, location, histological grade, and receptor status, are key factors in predicting axillary lymph node metastasis (12–14). In recent years, indicators such as serum markers and blood cell counts have also been found to be closely related to the metastasis, prognosis and treatment effect of malignant tumors, and the acquisition of these indicators is relatively easy and less invasive (15). However, the relationship between serological indicators and ALNM is still limited, and the exact mechanism needs to be further clarified.
Given the high heterogeneity of IBC, different patients may have different disease progression and prognosis. Although several predictive models for ALNM have been developed, most are based on Western populations with limited applicability to Chinese patients, rely on a single category of predictors (e.g., ultrasound alone or clinicopathological features alone), or require complex radiomics or genetic testing that hinders clinical adoption. More importantly, in the current era of axillary surgical de−escalation, where landmark trials such as SOUND and INSEMA have demonstrated that sentinel lymph node biopsy can be safely omitted in selected low−risk patients, the clinical need has shifted from simply predicting ALNM to accurately identifying patients with a sufficiently low risk of metastasis to safely forgo any axillary surgery. To address these gaps, we developed a practical nomogram that integrates widely accessible clinicopathological, ultrasound, and serological indicators (including white blood cell count, which has rarely been incorporated into such models) to preoperatively predict ALNM risk in a Chinese cohort. Our model is designed to support individualized axillary de−escalation decisions in routine clinical practice.
Materials and methods
Patients
This study included 454 patients with IBC diagnosed in the Second Clinical College of Shanxi Medical University from January 2020 to July 2024, which constituted the main study cohort. Exclusion criteria included: patients who had received preoperative treatment such as neoadjuvant chemotherapy; Patients with a history of other malignancies, immune system or blood system disorders; And patients with multifocal, bilateral, recurrent, metastatic, or other pathological types of breast cancer. Patients were randomly allocated to a training set and a validation set in a 3:1 ratio using a fixed random seed (set.seed(123) in R version 4.1.3) to ensure reproducibility. This retrospective study follows the ethical guidelines of the Declaration of Helsinki and has been approved by the Ethics Committee of the Second Clinical Hospital of Shanxi Medical University. Considering that the study will not adversely affect the rights and health of the participants and that the level of risk does not exceed the minimum risk, the procedure of informed consent is exempted. In addition, patient privacy and personally identifiable information are properly protected.
Definitions
For the diagnosis of lymph node metastasis, the following criteria were employed: Positive axillary lymph node (ALN): Positive outcomes are affirmed through axillary lymph node dissection and needle biopsy; Axillary lymph node (ALN) negative: A negative finding is verified by axillary lymph node dissection and sentinel lymph node biopsy. In patients with negative sentinel lymph node biopsy (SLND) results, axillary lymph nodes are regarded as free of metastasis. Based on these criteria, patients were classified into those with axillary lymph node metastasis and those without.
Data collection
In this study, we collected the following clinical and laboratory indicators from patients:
(1) Demographic characteristics: age and menstrual status. (2) Pathological indicators: androgen receptor (AR), estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), P53, Ki67. (3) Ultrasound indicators: tumor location, size, margin, morphology, color Doppler flow imaging (CDFI) characteristics, axillary lymph node status, and breast Imaging Reporting and Data System (BI-RADS) class. (4) Serological indicators: alanine aminotransferase, aspartate aminotransferase, total bilirubin, urea, creatinine, D dimer, white blood cell count, red blood cell count, platelet count, platelet volume distribution width, neutrophil absolute value, lymphocyte absolute value, monocyte absolute value, eosinophils absolute value, basophils absolute value.
For ultrasound evaluation, the following definitions were used: Tumor size was measured as the maximum diameter (in cm) of the primary tumor on gray−scale ultrasound. Circumscribe margin was classified as “clear” (well−defined, sharp border between tumor and surrounding tissue) or “vague” (indistinct, irregular, or spiculated border). Tumor morphology was categorized as “regular” (oval or round shape) or “irregular” (lobular, spiculated, or ill−defined shape). CDFI (color Doppler flow imaging) was graded as “no abnormality” (absence of blood flow signal within the tumor) or “positive” (presence of any detectable blood flow signal). Axillary lymph node status was considered abnormal if any of the following criteria were met: (1) round or oval lymph node shape; (2) absent or displaced echogenic hilum; (3) minimal nodal diameter >5 mm; (4) irregular cortical thickening >2 mm. Otherwise, the axillary lymph node was classified as normal. BI-RADS category was assigned according to the American College of Radiology Breast Imaging Reporting and Data System (5th edition) (16).
Statistical analysis
The final cohort of patients was randomly allocated into a training group and a validation group in a 3:1 ratio using a fixed random seed (set.seed(123) in R version 4.1.3) to ensure reproducibility. Continuous variables adhering to a normal distribution were expressed as mean ± standard deviation, whereas non-normally distributed continuous variables were presented as median (interquartile range). Categorical variables were reported as percentages. Prior to multivariable logistic regression, we assessed multicollinearity among the independent variables using the variance inflation factor (VIF). A VIF > 5 was considered indicative of significant multicollinearity. Continuous variables, including tumor size and WBC count were entered into the model as continuous measurements without categorization. Logistic regression analysis was employed to identify potential risk factors for LNM, and a nomogram was subsequently constructed using multivariate logistic regression analysis. The Receiver Operating Characteristic (ROC) curve was utilized to validate the nomogram model and evaluate its accuracy in predicting LNM. Calibration curves were employed to assess the concordance between observed outcomes and predicted probabilities. Additionally, Decision Curve Analysis (DCA) was conducted to determine the clinical net benefit of the predictive model.
Statistical analyzes were performed using R version 4.1.3, with a significance threshold set at P < 0.05.
Results
Basic characteristics
The patient’s flow chart is shown in Figure 1. The study included 454 patients with invasive breast cancer, with a median age of 54.5 years. The clinical characteristics of these eligible patients are shown in Table 1. Baseline clinical data were comparable between the training and validation groups, and among all participants, 138 patients with invasive breast cases developed axillary lymph node metastases.
Figure 1.
The flowchart of patient selection from the Shanxi Hospital.
Table 1.
Baseline characteristics of participants.
| Variable | Total (n =454) | Training (n=341) | Validation (n=113) | P value |
|---|---|---|---|---|
| LNM, n (%) | 0.502 | |||
| No | 318 (70.0) | 236 (68.6) | 82 (72.6) | |
| Yes | 138 (30.4) | 107 (31.4) | 31 (27.4) | |
| Age (years) | 54 (46,63) | 55 (46,63) | 52 (46,62) | 0.361 |
| Menopause state, n (%) | 0.434 | |||
| Premenopausal | 340 (74.8%) | 259 (76%) | 81 (71.7%) | |
| Postmenopausal | 114 (25.1%) | 82 (24%) | 32 (28.3%) | |
| AR (%) | 90 (60,90) | 90 (65,90) | 90 (30,90) | 0.387 |
| ER (%) | 90 (60,90) | 90 (5,90) | 90 (0,90) | 0.463 |
| PR (%) | 30 (0,80) | 30 (0,80) | 40 (0,90) | 0.747 |
| HER2 state, n (%) | 0.372 | |||
| negative | 267 (58.8%) | 196 (57.5%) | 71 (62.8%) | |
| positive | 187 (41.1%) | 145 (42.5%) | 42 (37.2%) | |
| P53 (%) | 15 (15,30) | 15 (15,30) | 15 (15,30) | 0.295 |
| Ki67 (%) | 30 (15,50) | 30 (15,50) | 30 (19,60) | 0.303 |
| Tumor size (cm²) | 2.23 (1.03,3.64) | 2.28 (1.01,3.65) | 2.11 (1.04,3.75) | 0.962 |
| Circumscribe margin, n (%) | 0.990 | |||
| Vague | 279 (61.4%) | 209 (61.3%) | 70 (61.9%) | |
| Clear | 175 (38.5%) | 132 (38.7%) | 43 (38.1%) | |
| Tumor morphology, n (%) | 0.875 | |||
| regular | 48 (10.5%) | 37 (10.9%) | 11 (9.7%) | |
| irregular | 406 (89.4%) | 304 (89.1%) | 102 (90.3%) | |
| CDF1, n (%) | 0.088 | |||
| No abnormality | 153 (33.7%) | 107 (31.4%) | 46 (40.7%) | |
| Blood flow signal positive | 301 (66.2%) | 234 (68.6%) | 67 (59.3%) | |
| Ultrasonic lymph node status, n (%) | 0.104 | |||
| Normal | 213 (46.9%) | 152 (44.6%) | 61 (54%) | |
| Abnormal | 241 (53%) | 189 (55.4%) | 52 (46%) | |
| Tumor location, n (%) | 0.543 | |||
| Outer upper quadrant | 191 (42%) | 144 (42.2%) | 47 (41.6%) | |
| Outer lower quadrant | 35 (7.7%) | 27 (7.9%) | 8 (7.1%) | |
| Inner lower quadrant | 72 (15.8%) | 58 (17%) | 14 (12.4%) | |
| Inner upper quadrant | 156 (34.4%) | 112 (32.8%) | 44 (38.9%) | |
| BI-RADS, n (%) | 0.975 | |||
| Level 2 | 1 (0.2%) | 1 (0.3%) | 0 (0%) | |
| Level 3 | 19 (4.1%) | 14 (4.1%) | 5 (4.4%) | |
| Level 4 | 349 (76.8%) | 263 (77.1%) | 86 (76.1%) | |
| Level 5 | 70 (15.4%) | 52 (15.2%) | 18 (15.9%) | |
| Level 6 | 15 (3.3%) | 11 (3.2%) | 4 (3.5%) | |
| ALT (U/L) | 16.5 (12.5,23.4) | 16.3 (12.7,23.6) | 17.9 (11.8,23.2) | 0.898 |
| AST (U/L) | 20.7 (17.6,24.8) | 20.7 (17.6,25.0) | 20.5 (17.4,24.0) | 0.459 |
| TBIL (umol/L) | 11.9 (9.5,15.5) | 12.0 (9.5,15.7) | 11.7 (9.6,15.1) | 0.863 |
| UREA (mmol/L) | 4.8 (4.0,5.6) | 4.8 (4.1,5.7) | 4.7 (3.9,5.5) | 0.322 |
| CREA (umol/L) | 56 (51,62) | 56 (51,62) | 55 (50,63) | 0.629 |
| GLU (mmol/L) | 5.21 (4.69,5.93) | 5.23 (4.67,5.97) | 5.14 (4.73,5.90) | 0.742 |
| D-dimer (ng/mL) | 87.5 (57.8,143.0) | 91 (57,139) | 83 (58.5,148) | 0.883 |
| WBC (/L) | 5.82 (4.88,6.85) | 5.87 (4.90,6.88) | 5.76 (4.82,6.73) | 0.646 |
| RBC (/L) | 4.48 (4.20,4.73) | 4.48 (4.19,4.71) | 4.47 (4.23,4.81) | 0.358 |
| PLT (/L) | 245 (197,287) | 244 (198,287) | 247 (196,291) | 0.841 |
| PDW (fL) | 14.8 (11.7,16.0) | 15.0 (11.7,16.0) | 14.2 (11.5,16.0) | 0.731 |
| NEU (/L) | 3.54 (2.83,4.37) | 3.54 (2.81,4.37) | 3.53 (2.93,4.41) | 0.841 |
| LYM (/L) | 1.70 (1.38,2.13) | 1.72 (1.40,2.14) | 1.68 (1.30,2.13) | 0.355 |
| MON (/L) | 0.36 (0.28,0.45) | 0.37 ( (0.28,0.46) | 0.36 (0.28,0.44) | 0.325 |
| EOS (/L) | 0.07 (0.04,0.12) | 0.07 (0.04,0.12) | 0.07 (0.04,0.12) | 0.818 |
| BASO (/L) | 0.03 (0.02,0.04) | 0.03 (0.02,0.04) | 0.03 (0.02,0.04) | 0.351 |
LYM, Lymph Node Metastasis; AR, Androgen Receptor; ER, Estrogen Receptor; PR, Progesterone Receptor; HER2, Human Epidermal Growth Factor Receptor 2; P53, Tumor Protein P53; Ki-67, A proliferation marker in cancer; CDFI, Color Doppler Flow Imaging; BI-RADS, Breast Imaging Reporting and Data System; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; TBIL, Total Bilirubin; UREA, Urea; CREA, Creatinine; GLU, Glucose; D-dimer, A protein fragment found in the blood; WBC: White Blood Cell count; RBC, Red Blood Cell count; PLT, Platelet count; PDW, Platelet Distribution Width; NEU, Neutrophil count; LYM, Lymphocyte count; MON, Monocyte count; EOS, Eosinophil count; BASO, Basophil count.
Development of nomogram
Univariate logistic regression analysis identified 8 variables that were significantly associated with LNM, including ER, CDFI, BI-RADS, WBC, BE, Ultrasonic lymph node status, Tumor size, Circumscribe margin, Tumor morphology (Table 2). Multicollinearity diagnostics revealed no significant collinearity among the six independent variables included in the final multivariable model (all VIF values < 2.5). A multivariate regression analysis of the training set, utilizing the significantly different variables, indicated that ER, BI-RADS, WBC, Tumor size, circumscribe margin and Ultrasonic lymph node status were identified as independent risk factors for the development of LNM in patients with IBC (Table 3). Based on these variables, a nomogram was constructed to predict LNM in invasive breast cancer patients (Figure 2). In the nomogram, the following variable assignments were used: circumscribe margin (0 = vague, 1 = clear), ultrasonic lymph node status (0 = normal, 1 = abnormal), and BI-RADS class (2 to 6, treated as continuous). ER, tumor size, and WBC were entered as continuous measurements.
Table 2.
Univariate analysis of predictive variables of LNM in the training cohort.
| Variables | OR | 95% CI | P value |
|---|---|---|---|
| Age (years) | 0.997 | 0.978-1.016 | 0.740 |
| Menopause state | 0.947 | 0.553-1.622 | 0.842 |
| AR | 1.002 | 0.995-1.009 | 0.529 |
| ER | 1.008 | 1.002-1.014 | 0.013 |
| PR | 1.004 | 0.998-1.010 | 0.250 |
| HER2 state | 0.906 | 0.612-1.545 | 0.906 |
| P53 | 0.991 | 0.975-1.007 | 0.289 |
| Ki67 | 1.001 | 0.992-1.010 | 0.850 |
| Tumor size(cm²) | 1.121 | 1.035-1.213 | 0.005 |
| Circumscribe margin | 0.443 | 0.268-0.732 | 0.001 |
| Tumor morphology | 3.232 | 1.222-8.543 | 0.018 |
| CDF1 | 2.204 | 1.285-3.779 | 0.004 |
| Ultrasonic lymph node status | 2.796 | 1.707-4.581 | <0.001 |
| Tumor location | 1.179 | 0.949,1.464 | 0.137 |
| BI-RADS | 2.400 | 1.480-3.892 | <0.001 |
| ALT(U/L) | 0.998 | 0.972-1.017 | 0.870 |
| AST(U/L) | 1.005 | 0.980-1.030 | 0.715 |
| TBIL (umol/L) | 0.978 | 0.936-1.022 | 0.317 |
| UREA (mmol/L) | 0.905 | 0.759-1.080 | 0.270 |
| CREA (umol/L) | 1.001 | 0.977-1.026 | 0.921 |
| GLU (mmol/L) | 1.111 | 0.974-1.268 | 0.117 |
| D-dimer(ng/mL) | 1.002 | 1.000-1.004 | 0.100 |
| WBC(/L) | 0.867 | 0.753-0.997 | 0.046 |
| RBC(/L) | 1.018 | 0.976-1.061 | 0.413 |
| PLT(/L) | 0.998 | 0.995-1.002 | 0.321 |
| PDW (fL) | 0.938 | 0.857-1.026 | 0.161 |
| NEU(/L) | 0.198 | 0.754-1.060 | 0.198 |
| LYM(/L) | 1.003 | 0.910-1.104 | 0.957 |
| MON(/L) | 1.094 | 0.836-1.430 | 0.513 |
| EOS(/L) | 0.952 | 0.346-2.614 | 0.923 |
| BASO(/L) | 178.957 | 0.056-570501.761 | 0.208 |
AR, Androgen Receptor; ER, Estrogen Receptor; PR, Progesterone Receptor; HER2, Human Epidermal Growth Factor Receptor 2; P53, Tumor Protein P53; Ki-67, A proliferation marker in cancer; CDFI, Color Doppler Flow Imaging; BI-RADS, Breast Imaging Reporting and Data System; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; TBIL, Total Bilirubin; UREA, Urea; CREA, Creatinine; GLU, Glucose; D-dimer, A protein fragment found in the blood; WBC: White Blood Cell count; RBC, Red Blood Cell count; PLT, Platelet count; PDW, Platelet Distribution Width; NEU, Neutrophil count; LYM, Lymphocyte count; MON, Monocyte count; EOS, Eosinophil count; BASO, Basophil count.
Table 3.
Multivariate logistic regression analysis of independent predictors of LNM in the training cohort.
| Variables | β | SE | Wald | P-value | OR (95% Cl) |
|---|---|---|---|---|---|
| ER | 0.009 | 0.003 | 6.763 | 0.009 | 1.009 (1.002-1.016) |
| Tumor size | 0.070 | 0.033 | 4.362 | 0.037 | 1.072 (1.004-1.144) |
| Circumscribe margin | -0.677 | 0.274 | 6.095 | 0.014 | 0.508 (0.297-0.870) |
| Ultrasonic lymph node status | 0.853 | 0.271 | 9.878 | 0.002 | 2.346 (1.379-3.994) |
| BI-RADS class | 0.578 | 0.230 | 6.320 | 0.012 | 1.782 (1.136-2.797) |
| WBC | -0.221 | 0.083 | 7.061 | 0.008 | 0.802 (0.682-0.944) |
| Constant | -2.902 | 1.062 | 7.459 | <0.05 | —— |
Continuous variables (tumor size, WBC) were entered as continuous measurements. VIF values for all variables were < 2.5, indicating no significant multicollinearity. For categorical variables: circumscribe margin (0 = vague, 1 = clear); ultrasonic lymph node status (0 = normal, 1 = abnormal).
BI-RADS, Breast Imaging Reporting and Data System; WBC, White Blood Cell count; ER, Estrogen Receptor.
Figure 2.
Nomogram to estimate the risk of LNM in patients with breast cancer. Variable assignments: ER (estrogen receptor, % positive, continuous); Tumor size (cm, continuous); Circumscribe margin (0 = vague, 1 = clear); Ultrasonic lymph node status (0 = normal, 1 = abnormal); BI-RADS class (continuous scale from 2 to 6); WBC (white blood cell count, ×109/L, continuous). BI-RADS, Breast Imaging Reporting and Data System; WBC, White Blood Cell count; ER, Estrogen Receptor.
Verification of nomogram
The ROC curve analysis showed that the model could differentiate between LNM and non-LNM patients with an AUC of 0.741 in the training cohort and 0.705 in the validation cohort, indicating acceptable discriminative performance (Figures 3A, B). According to common classification schemes for discriminative ability (e.g., Hosmer–Lemeshow: AUC 0.7–0.8 indicates acceptable discrimination), our model’s performance falls within the acceptable range. To further validate the clinical utility of our nomogram, we compared its predictive performance with that of the preoperative axillary ultrasound assessment performed by experienced breast radiologists. The Ultrasonic lymph node status (normal vs. abnormal) alone yielded an AUC of 0.620 in the training cohort and 0.627 in the validation cohort for predicting ALNM (Figures 3C, D). In comparison, our nomogram achieved significantly higher AUCs (training: 0.741, P = 0.0017; validation: 0.705, P = 0.1772; DeLong test). These results suggest that the nomogram provides improved discriminative ability over routine ultrasound assessment alone.
Figure 3.
The ROC curve of the nomogram for predicting LNM in patient with breast cancer. The AUC of the nomogram for the prediction of LNM in the training set (A) and the validation set (B). The AUC of the Ultrasonic lymph node status for the prediction of LNM in the training set (C) and the validation set (D). Ultrasonic lymph node status (0 = normal, 1 = abnormal).
Additionally, the prediction probabilities generated by the model showed consistency with the observed outcomes in both the training set (Figure 4A) and the validation set (Figure 4B), indicating good calibration. The examination of the clinical impact curve (Figures 5A, B) demonstrates that the nomogram exhibits a substantial overall net benefit across a broad spectrum of threshold probabilities, suggesting a significant potential for clinical application.
Figure 4.
Calibration curves and DCA of the nomogram for predicting LNM. Calibration curves of the predicted nomogram in training set (A) and validation set (B).
Figure 5.
DCA of the nomogram in the training set (A) and the validation set (B). DCA, decision curve analysis.
Discussion
Among patients with IBC, predicting ALNM is of paramount importance for clinical treatment decision-making. This study conducted an in-depth analysis of the clinicopathological characteristics, serological indicators, and ultrasound indices of the patients and discovered that there was a significant correlation between these factors and ALNM. Based on these findings, we selected key variables such as ER, Tumor size, circumscribe margins, Ultrasound Lymph node status, BI-RADS class and WBC and constructed a prediction model through multivariate regression analysis. Our results showed that in the training set and validation set, the proportion of IBC patients with axillary lymph node metastasis was 31.4% and 27.4%, respectively, which is consistent with reports in the existing literature (12). In addition, compared with MSKCC and SOC models, while the absolute discriminative performance (AUC 0.741 and 0.705) is considered acceptable for clinical decision support (17). Our nomogram demonstrated superior discriminative performance compared with ultrasound specialist assessment based on Ultrasonic lymph node status alone (AUC 0.741 vs. 0.620 in the training cohort). This improvement is likely attributable to the integration of additional predictors (tumor size, margin, ER, WBC, and BI-RADS) that capture complementary aspects of tumor biology. To our knowledge, this is one of the few nomograms that simultaneously incorporates ultrasound features, serological markers (including WBC), and pathological indicators in a Chinese population, while being explicitly designed to align with current axillary de−escalation strategies.
The paradigm of axillary surgery has been progressively narrowed by landmark de−escalation trials. ACOSOG Z0011 and AMAROS established that completion ALND can be safely omitted or replaced by axillary radiotherapy in selected patients with limited sentinel node involvement (18, 19). More recently, SOUND, INSEMA, and SENOMAC demonstrated that even SLNB can be safely omitted in cT1−2N0 patients with negative preoperative ultrasound, and that completion ALND is unnecessary for sentinel node macrometastases (6, 20, 21); a meta−analysis (CADONOT) confirmed a 65% lower lymphedema risk with SLNB alone versus ALND (6). The impetus for axillary de−escalation stems largely from the substantial morbidity associated with more extensive axillary surgery. ALND carries a significant risk of postoperative complications, including seroma formation (reported in 15–30% of patients), wound infections (5–10%), hematoma (3–5%), and prolonged drainage, which can delay adjuvant therapy and impair quality of life (18, 19, 22). In contrast, sentinel lymph node biopsy (SLNB) substantially reduces these risks: seroma rates drop to 5–10%, wound complications to 2–5%, and lymphedema to approximately 5–8% compared with 20–30% after ALND (7). Lymphedema, in particular, is a feared long−term sequela that can lead to chronic swelling, functional impairment, and recurrent infections. Landmark de−escalation trials have consistently demonstrated that omitting axillary surgery altogether in carefully selected patients (e.g., the SOUND and INSEMA trials) results in even lower rates of these complications without compromising oncologic outcomes (20, 21). Therefore, accurate preoperative identification of patients with a very low risk of ALNM is crucial to safely extend the benefits of avoiding any axillary surgery. Our nomogram is designed to support this precise risk stratification. Our nomogram directly addresses this clinical need by integrating clinicopathological, ultrasound, and serological indicators to predict ALNM risk preoperatively, thereby providing supportive information for preoperative risk stratification. However, due to its moderate discriminative ability, it cannot replace standard axillary staging or directly determine the omission of SLNB.
Beyond axillary staging, accurate preoperative assessment of primary breast tumors is equally critical for tailored surgical planning, particularly for non-palpable breast lesions which account for a significant proportion of screen-detected breast cancers. Precise localization of these lesions is essential to achieve negative surgical margins while minimizing the volume of healthy breast tissue resected, thereby improving cosmetic outcomes and patient quality of life. Traditional wire-guided localization, while widely used, has several limitations including wire migration, patient discomfort, and the need for preoperative radiological placement immediately before surgery. In recent years, advanced image-guided localization techniques have emerged as promising alternatives, offering enhanced precision and flexibility. Parisi et al. (23) compared radiofrequency (RF) localization alone with a combined RF and intraoperative ultrasound (IOUS) technique for non-palpable breast cancer, demonstrating that the combined approach resulted in significantly lower positive margin rates (11.1% vs. 27.8%) and reduced reoperation rates. Similarly, the same group reported their initial experience with the combined LOCalizer™ radiofrequency identification system and IOUS, which achieved 100% successful localization with no positive margins in their cohort of 25 patients (24). These studies highlight the value of integrating multiple imaging modalities and advanced localization technologies to optimize breast-conserving surgery. Our nomogram, which provides preoperative risk stratification for ALNM, complements these localization techniques by enabling comprehensive individualized surgical planning.
This study covers preoperative ultrasound imaging features commonly used in the clinic. Ultrasound technology is non-invasive, reproducible, and can be observed in real time, and has shown unique value in detecting small calcifications inside tumors and in imaging blood flow. In recent years, several clinicians have constructed models to predict axillary lymph node metastasis based on ultrasound features. Our study further optimizes and validates these existing models. Consistent with previous models, our results show that Tumor size, circumscribe margin features, BI-RADS classification, and Ultrasound lymph node status are independent predictors of ALNM in patients with IBC. Tumor size is the strongest predictor of ALN metastasis, a fact that has been demonstrated in multiple studies. A larger tumor volume means an increase in the number of tumor cells, which can lead to an increase in the complexity of the tumor microenvironment, thereby enhancing its metastasis potential (25). Circumscribe margins are key areas where tumors interact with surrounding normal tissue, and cells here tend to exhibit higher genomic instability, epithelial-mesenchymal transition (EMT) phenomena, and greater aggressiveness. By changing their morphology and movement patterns, these cells may play a key role in tumor invasion and metastasis, a finding that is consistent with previous observations by researchers (26–28).
The BI-RADS classification takes into account the morphological characteristics of the tumor, which are strongly associated with the risk of tumor metastasis (29). In addition, we integrated the positive indicators of ultrasound lymph node status assessment from previous studies, and patients meeting any of the following conditions were judged to have axillary ultrasound lymph node status abnormalities (30): (1) Round or oval lymph nodes were detected; (2) Hypoechoic nucleus or lymph node central portal disappeared; (3) The minimum diameter of lymph nodes is more than 5mm; 4. Irregular cortex or cortex thickness greater than 2mm. These changes are associated with modification of the lymph node microenvironment by tumor cells, including angiogenesis and recruitment of immune cells. In addition, Tumor morphology and CDFI were independent predictors of ALNM. Recent studies have shown that tumor morphology may be related to the remodeling of tumor extracellular mechanisms, and the high expression of mechanism metalloproteinases may lead to the degradation of extracellular matrix, thus promoting tumor invasion and morphological irregularity (31). The increased blood flow signal may be related to the inflammatory response in the tumor microenvironment, and the infiltration of inflammatory cells can release various cytokines, such as tumor necrosis factor α (TNF-α) and interleukin, which can promote blood vessel dilation and the formation of new blood vessels, making tumors more likely to metastasize (32).These ultrasound features incorporated into our model are readily available in all breast cancer patients, which guarantees the applicability and utility of our model.
In the clinical pathology parameters, our research findings indicate that ER is independent predictors of ALNM in IBC patients. As one of the key indicators in the diagnosis of breast cancer, ER has been widely confirmed to be closely related to the clinical outcome and prognosis of patients. ER status is not only an important parameter for evaluating tumor mutation load in breast cancer patients, but also its expression level is closely related to the expression of Ki67, PR and HER2, which has a significant impact on lymph node metastasis, tumor grade and lymphatic vessel invasion of invasive breast cancer (33). Recent studies have shown that ER promotes the growth and survival of invasive breast cancer cells by activating multiple signaling pathways, such as the PI3K/AKT/mTOR and MAPK/ERK pathways (34). Activation of these pathways enhances the aggressiveness and metastasis potential of tumor cells. In addition, estrogen can enhance the migration and invasion ability of breast cancer cells by regulating the expression of genes associated with tumor metastasis, such as promoting the expression of genes associated with epithelial stromal transformation (EMT) (35).
Existing research indicates that WBC counts are closely correlated with prognosis across multiple cancers. For instance, in colorectal cancer patients, elevated white blood cell counts correlate with poorer outcomes and an immunosuppressive tumor microenvironment (36). Additionally, white blood cell count is recognized as a key prognostic biomarker in non-smoking lung adenocarcinoma patients (37). In breast cancer, white blood cell count and its derived inflammatory markers, such as neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), have been extensively studied to evaluate their role in cancer prognosis (38). Interestingly, our multivariate analysis revealed that a higher WBC count was associated with a reduced risk of ALNM (OR = 0.802, 95% CI: 0.682–0.944). However, this finding should be interpreted with caution, as it is exploratory and derived from a retrospective single−center cohort. This protective effect may reflect competent systemic immune surveillance, where adequate circulating leukocytes facilitate the elimination of metastatic tumor cells in axillary lymph nodes. Alternatively, lower WBC counts in some patients might indicate poor nutritional status, bone marrow suppression, or advanced disease burden, all of which could correlate with higher metastatic potential. Nevertheless, we cannot exclude the possibility of residual confounding. Unmeasured factors – such as subclinical infections, concomitant medications (e.g., corticosteroids or anti−inflammatory drugs), systemic inflammatory states, or differences in disease stage – may influence both WBC count and ALNM risk. Reverse causality is also possible, whereby more advanced tumors might suppress bone marrow function or alter leukocyte trafficking. While some previous studies have reported that elevated WBC counts are associated with increased breast cancer risk, our findings suggest that the relationship between WBC and ALNM is complex and may depend on disease stage, patient population, and the balance between pro-tumor and anti-tumor immune responses (39). Given the exploratory nature of this finding, prospective studies with comprehensive covariate adjustment are needed to validate the independent protective association of WBC and to clarify the underlying mechanisms.
Limitations
This study used a single-center retrospective design, and although we applied strict inclusion and exclusion criteria to ensure the accuracy of the data, selection bias is still a limitation that cannot be completely avoided. Furthermore, breast ultrasound is inherently operator dependent. A more specific limitation is that breast ultrasound is inherently operator dependent. In this study, all preoperative ultrasound evaluations were performed by a team of three experienced breast radiologists (each with more than 5 years of specialized experience) following a standardized scanning protocol. However, we did not formally assess inter-observer variability, which may affect the reproducibility of ultrasound-derived variables such as tumor margin, morphology, and axillary lymph node status. This represents an additional limitation of our study. However, we did not formally assess interobserver variability (e.g., using kappa statistics). This is a major limitation, as ultrasound variables such as tumor margin (circumscribe vs. vague), tumor morphology (regular vs. irregular), and axillary lymph node status (normal vs. abnormal) are subjective and may vary between operators. The lack of interobserver reproducibility assessment means that the performance of our nomogram when applied by different radiologists or in different institutions remains uncertain. Moreover, our model was developed and internally validated only in a single-center retrospective Chinese cohort. External validation in independent, preferably multicenter and multinational, cohorts is essential to assess its reproducibility across different populations (e.g., Western or other Asian populations), ultrasound operators, and healthcare settings. The current model’s performance in other surgical environments remains uncertain. In order to enhance the universality and reliability of the study results, future studies should adopt a multicenter, randomized controlled, large sample size prospective study design to further validate the findings of this study. Prospective studies should also include rigorous training and quality control measures for ultrasound operators, and report inter-observer agreement (e.g., kappa statistics) to ensure the generalizability of predictive models.
Conclusion
In the era of axillary surgical de−escalation, accurately identifying patients with a sufficiently low risk of axillary lymph node metastasis is of paramount clinical importance. Our nomogram offers several practical advantages: it integrates three readily available data categories (clinicopathological, ultrasound, and serological indicators), uses only routine preoperative tests without requiring advanced imaging or genetic profiling, and is specifically developed and validated in a Chinese cohort. With acceptable discriminative performance (AUC = 0.741 in the training cohort, 0.705 in the validation cohort). However, given its moderate predictive accuracy, this model cannot directly determine whether sentinel lymph node biopsy (SLNB) may be omitted, nor can it replace standard axillary staging. At present, the nomogram should be viewed only as auxiliary support for preoperative risk stratification — for example, to identify patients with a higher probability of metastasis who may warrant more rigorous preoperative evaluation. Decisions to omit axillary surgery must continue to follow established trial criteria (e.g., SOUND, INSEMA) and current guideline recommendations. Prospective, multicenter external validation is necessary before this nomogram can be considered for routine clinical implementation.
Acknowledgments
We acknowledged the contributions of the Second Clinical College of Shanxi Medical University for providing clinical data.
Funding Statement
The author(s) declared that financial support was not received for this work and/or its publication.
Footnotes
Edited by: Cristian Scatena, University of Pisa, Italy
Reviewed by: Claudio Gambardella, University of Campania Luigi Vanvitelli, Italy
Remzi Can Çakır, Hatay Training and Research Hospital, Türkiye
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Ethics statement
This retrospective study follows the ethical guidelines of the Declaration of Helsinki and has been approved by the Ethics Committee of the Second Clinical Hospital of Shanxi Medical University (2024 YX NO.360). The studies were conducted in accordance with the local legislation and institutional requirements. The human samples used in this study were acquired from a by- product of routine care or industry. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.
Author contributions
QX: Data curation, Conceptualization, Investigation, Validation, Software, Project administration, Writing – review & editing, Supervision, Writing – original draft, Methodology. ZY: Writing – review & editing, Software.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
- 1. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, et al. Erratum: Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. Ca: A Cancer J For Clin. (2020) 70(4):313. doi: 10.3322/caac.21609. PMID: [DOI] [PubMed] [Google Scholar]
- 2. Ning M, Ping Y, Qin L, Qinglin W, Meijie L, Heng M, et al. Radiomics nomogram of contrast-enhanced spectral mammography for prediction of axillary lymph node metastasis in breast cancer: a multicenter study. Eur Radio. (2020) 30(12):6732–9. doi: 10.2139/ssrn.3469725 [DOI] [PubMed] [Google Scholar]
- 3. Chenao Z, Yongbin H, Xinrong W, Huan L, Liming X, Tao A. Prediction of axillary lymph node metastasis in breast cancer using intra-peritumoral textural transition analysis based on dynamic contrast-enhanced magnetic resonance imaging. Acad Radiol. (2022) 29(Suppl 1):S107–5. doi: 10.1016/j.acra.2021.02.008. PMID: [DOI] [PubMed] [Google Scholar]
- 4. Loibl S, André F, Bachelot T, Barrios CH, Bergh J, Burstein HJ, et al. Early breast cancer: ESMO clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol. (2024) 35:159–82. doi: 10.1016/j.annonc.2023.11.016. PMID: [DOI] [PubMed] [Google Scholar]
- 5. Gradishar WJ, Moran MS, Abraham J, Abramson V, Aft R, Agnese D, et al. NCCN Guidelines® Insights: Breast cancer, version 5.2025. J Natl Compr Canc Netw. (2025) 23:426–36. doi: 10.6004/jnccn.2025.0053. PMID: [DOI] [PubMed] [Google Scholar]
- 6. Mattar A, Antonini M, Cavalcante FP, Zerwes F, Millen EC, Brenelli FP, et al. CADONOT: Comparing axillary dissection or not in breast cancer surgery. Breast. (2025) 81:104453. doi: 10.1016/j.breast.2025.104453. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Lorek A, Stojčev Z, Zarębski W, Kowalczyk M, Szyluk K. Analysis of postoperative complications after 303 sentinel lymph node identification procedures using the SentiMag® method in breast cancer patients. Med Sci Monitor: Int Med J Exp Clin Res. (2019) 25:3154–60. doi: 10.12659/msm.912758. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Banys-Paluchowski M, Gasparri ML, de Boniface J, Gentilini O, Stickeler E, Hartmann S, et al. Surgical management of the axilla in clinically node-positive breast cancer patients converting to clinical node negativity through neoadjuvant chemotherapy: Current status, knowledge gaps, and rationale for the EUBREAST-03 AXSANA study. Cancers (Basel). (2021) 13(7):1565. doi: 10.3390/cancers13071565. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wang Z, Chen D. Letter to the editor "Performance of sentinel lymph node biopsy after neoadjuvant chemotherapy in clinically node-positive breast cancer patients systematic review and meta-analysis. Int J Surg (London England). (2025) 111:5713–4. doi: 10.1097/js9.0000000000002585. PMID: [DOI] [PubMed] [Google Scholar]
- 10. Rongrong G, Guolan L, Binjie Q, Baowei F. Ultrasound imaging technologies for breast cancer detection and management: A review. Ultrasound Med Biol. (2018) 23(Suppl 6):vi23–9. doi: 10.1016/j.ultrasmedbio.2017.09.012. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Sigrist R, Joy L, Ahmed El K, Maria Cristina C, Jüergen KW. Ultrasound elastography: Review of techniques and clinical applications. Theranostics. (2017) 7(5):1303–29. doi: 10.7150/thno.18650. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Castelo BB, Brito LGO, Torresan RZ, Filho CC, Duarte GM. Omission of axillary surgery in early breast cancer with negative lymph nodes: a systematic review and meta-analysis of randomized clinical trials. Breast Cancer Res Treat. (2026) 216(3):30. doi: 10.1007/s10549-026-07910-y. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Park WK, Nam SJ, Kim SW, Yu J, Chae BJ, Lee SK, et al. Axillary surgery in breast cancer: Evidence-based de-escalation across upfront and post-neoadjuvant settings. J Breast Cancer. (2026) 29:17–32. doi: 10.4048/jbc.2025.0290. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Jian Z, Xiao L, Rong H, Wei F, Yunhong K, Feng X, et al. A nomogram to predict the probability of axillary lymph node metastasis in female patients with breast cancer in China: A nationwide, multicenter, 10-year epidemiological study. Oncotarget. (2016) 8(21):35311–25. doi: 10.18632/oncotarget.13330. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ayca Nazli M, Mahima G, Aminata L, Debapriya D, Betsy B, Maria Grosse P, et al. Abstract 6286: Metabolic abnormalities are early predictors of liver metastasis in patients with ER+ breast tumors. Cancer Res. (2024) 84(6_Supplement):6286. doi: 10.1158/1538-7445.am2024-6286. PMID: 36230740 [DOI] [Google Scholar]
- 16. Asada T, Yamada T, Kanemaki Y, Fujiwara K, Okamoto S, Nakajima Y. Grading system to categorize breast MRI using BI-RADS 5th edition: a statistical study of non-mass enhancement descriptors in terms of probability of Malignancy. Jpn J Radiol. (2018) 36:200–8. doi: 10.1007/s11604-017-0717-9. PMID: [DOI] [PubMed] [Google Scholar]
- 17. Jin Y, Liu X, Zhang X, Wang Y, Cheng X, Cao S, et al. Developing and evaluating a nomogram model predicting axillary lymph node metastasis of triple-negative breast cancer based on multimodal imaging characteristics. Acad Radiol. (2025) 32:4382–94. doi: 10.1016/j.acra.2025.04.031. PMID: [DOI] [PubMed] [Google Scholar]
- 18. Giuliano AE, Ballman KV, McCall L, Beitsch PD, Brennan MB, Kelemen PR, et al. Effect of axillary dissection vs no axillary dissection on 10-year overall survival among women with invasive breast cancer and sentinel node metastasis: The ACOSOG Z0011 (Alliance) randomized clinical trial. Jama. (2017) 318:918–26. doi: 10.1001/jama.2017.11470. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bartels SAL, Donker M, Poncet C, Sauvé N, Straver ME, van de Velde CJH, et al. Radiotherapy or surgery of the axilla after a positive sentinel node in breast cancer: 10-year results of the randomized controlled EORTC 10981–22023 AMAROS trial. J Clin Oncol Off J Am Soc Clin Oncol. (2023) 41:2159–65. doi: 10.1200/jco.22.01565. PMID: [DOI] [PubMed] [Google Scholar]
- 20. Gentilini OD, Botteri E, Sangalli C, Galimberti V, Porpiglia M, Agresti R, et al. Sentinel lymph node biopsy vs no axillary surgery in patients with small breast cancer and negative results on ultrasonography of axillary lymph nodes: The SOUND randomized clinical trial. JAMA Oncol. (2023) 9:1557–64. doi: 10.1001/jamaoncol.2023.3759. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Reimer T, Stachs A, Veselinovic K, Kühn T, Heil J, Polata S, et al. Axillary surgery in breast cancer - primary results of the INSEMA trial. N Engl J Med. (2025) 392:1051–64. doi: 10.1056/nejmoa2412063. PMID: [DOI] [PubMed] [Google Scholar]
- 22. Gambardella C, Clarizia G, Patrone R, Offi C, Mauriello C, Romano R, et al. Advanced hemostasis in axillary lymph node dissection for locally advanced breast cancer: New technology devices compared in the prevention of seroma formation. BMC Surg. (2019) 18:125. doi: 10.1186/s12893-018-0454-8. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Parisi S, Gambardella C, Conzo G, Ruggiero R, Tolone S, Lucido FS, et al. Advanced localization technique for non-palpable breast cancer: Radiofrequency alone vs combined technique with ultrasound. J Clin Med. (2023) 12(15):5076. doi: 10.3390/jcm12155076. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Parisi S, Ruggiero R, Gualtieri G, Volpe ML, Rinaldi S, Nesta G, et al. Combined LOCalizer™ and intraoperative ultrasound localization: First experience in localization of non-palpable breast cancer. In Vivo. (2021) 35:1669–76. doi: 10.21873/invivo.12426. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Litao R, Yunyue W, Ronghua S, Shangan L, Yan S, Huan W, et al. Ultrasound and clinicopathological characteristics of breast cancer for predicting axillary lymph node metastasis. Clin Hemorheol Microcirc. (2023) 85(2):147–62. doi: 10.3233/ch-231777. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Jiajia X, Wei Z, Yaxin W, Xiuhua W, Wenqu L, Qiaodan W, et al. Ultrasonography and clinicopathological features of breast cancer in predicting axillary lymph node metastases. BMC Cancer. (2022) 22(1):1155. doi: 10.1186/s12885-022-10240-z. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Xie H, Jingyi X, Shujia P, Ping Y, Zhenyu Y, Lin Y, et al. Preoperative nomogram for predicting sentinel lymph node metastasis risk in breast cancer: A potential application on omitting sentinel lymph node biopsy. Front Oncol. (2021) 11:665240. doi: 10.3389/fonc.2021.665240 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Jiao L, Weimei M, Xinhua J, Chunyan C, Hongli W, Jiewen C, et al. Development and validation of nomograms predictive of axillary nodal status to guide surgical decision-making in early-stage breast cancer. J Cancer. (2019) 10(5):1263–74. doi: 10.7150/jca.32386. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Ren T, Gao Z, Yang L, Cheng W, Luo X. Development of a nomogram for predicting Malignancy in BI-RADS 4 breast lesions using contrast-enhanced ultrasound and shear wave elastography parameters. Sci Rep. (2025) 15:1356. doi: 10.1038/s41598-025-85862-x. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Bowen Z, Qingshuang C. Novel model based on ultrasound predict axillary lymph node metastasis in breast cancer. BMC Med Imaging. (2023) 23(1):135. doi: 10.1186/s12880-023-01090-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Fang F, Wu T, Wang M, Li W, You Z, Chen M, et al. Matrix metalloproteinase-9 is associated with tumor microenvironment remodeling of bladder cancer. Biol Direct. (2025) 20:8. doi: 10.1186/s13062-025-00599-x. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Wang Y, Wang E, Anany M, Füllsack S, Huo YH, Dutta S, et al. The crosstalk between neuropilin-1 and tumor necrosis factor-α in endothelial cells. Front Cell Dev Biol. (2024) 12:1210944. doi: 10.3389/fcell.2024.1210944. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Huang L, Xu C, Song Y, Sun F, Sun X, Yao H, et al. Construction of a human epidermal growth factor receptor 2-related gene risk model for predicting breast cancer prognosis. Oncol Lett. (2026) 31:61. doi: 10.3892/ol.2025.15414. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Noyan S, Gür Dedeoğlu B. Upregulation of miR-99b-5p modulates ESR1 expression as an adaptive mechanism to circumvent drug response via facilitating ER/HER2 crosstalk. Balkan Med J. (2025) 42:150–6. doi: 10.4274/balkanmedj.galenos.2025.2024-12-47. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Sonia N, Saikia S, Limaye AM. Estrogenic control of matrix metalloproteinases: a perspective on breast tumor invasion and metastasis. Mol Biol Rep. (2025) 52:453. doi: 10.1007/s11033-025-10555-7. PMID: [DOI] [PubMed] [Google Scholar]
- 36. Wang B, Ling D, Li L, Zhang J, Xu J. Impact of preoperative white blood cell count on outcomes in different stage colorectal cancer patients undergoing surgical resection: a single-institution retrospective cohort study. BMC Cancer. (2024) 24:242. doi: 10.1186/s12885-024-11983-7. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Xiang R, Hu P, Xiao X, Li W, Liao X, Li J, et al. Development of a prognostic prediction model for non-smoking lung adenocarcinoma based on pathological information and laboratory hematologic indicators: a multicenter study. Front Immunol. (2025) 16:1566195. doi: 10.3389/fimmu.2025.1566195. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Cheng YH, Dong J, Wang Z, Zhao H, Chen M, Ma T. The value of artificial intelligence in ultrasound imaging for predicting molecular subtypes of breast cancer: a meta-analysis. Front Oncol. (2026) 16:1748473. doi: 10.3389/fonc.2026.1748473. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Rehana Q, Manuel P-R, Maiko S, Derek Van B, Vanessa Nunes de P, Anna B-R, et al. Estrone, the major postmenopausal estrogen, binds ERa to induce SNAI2, epithelial-to-mesenchymal transition, and ER+ breast cancer metastasis. Cell Rep. (2022) 41(7):111672. doi: 10.1016/j.celrep.2022.111672. PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.