Sage Journals: Discover world-class research

Abstract

Introduction

This study aimed to assess the predictive value of integrating ultrasonographic features, pathological characteristics, and inflammatory markers for axillary lymph node metastasis (ALNM) in early-stage breast cancer (BC), and to construct a corresponding nomogram.

Methods

A retrospective review was conducted on clinical data from 287 early-stage BC patients who underwent surgery at Shenzhen Luohu People’s Hospital between January 2020 and March 2024. Based on histopathological evaluation, patients were categorized into ALNM-positive (ALNM⁺) and ALNM-negative (ALNM⁻) groups. Independent predictors of ALNM were identified using univariate and multivariate logistic regression analyses. These variables were used to develop a predictive nomogram. Model performance was evaluated by concordance index (C-index), receiver operating characteristic (ROC) curve, calibration plot, and decision curve analysis (DCA), assessing its accuracy, discrimination, calibration, and clinical utility.

Results

Multivariate analysis identified vascular invasion, neutrophil-to-lymphocyte ratio (NLR), lymphocyte count, tumor size, lymph node echogenicity, and margin characteristics as independent predictors of ALNM. The nomogram showed excellent discriminative ability (AUC = 0.944, 95% CI: 0.906-0.981; C-index = 0.944, 95% CI: 0.906-0.982) and good calibration (Brier score = 0.063). DCA indicated meaningful clinical benefit across relevant threshold probabilities.

Conclusion

The nomogram developed in this study demonstrates strong predictive performance and clinical value for preoperative ALNM assessment in early-stage BC. It may serve as a practical tool to guide individualized surgical and therapeutic decision-making.

Keywords

breast cancer (BC)axillary lymph node metastasis (ALNM)inflammatory markers nomogram risk

Introduction

Breast cancer (BC) is the most prevalent malignancy among women worldwide and is marked by significant heterogeneity and variable aggressiveness.¹ Despite advances in standardized treatments, axillary lymph node metastasis (ALNM) remains a major indicator of poor prognosis. Axillary lymph node (ALN) status is essential not only for TNM staging but also for guiding surgical and adjuvant treatment decisions.^2,3 Sentinel lymph node biopsy (SLNB) has replaced axillary lymph node dissection (ALND) as the preferred method for evaluating ALN status in early-stage BC.⁴ However, SLNB lacks perfect diagnostic accuracy, with reported false-negative rates in high-quality studies ranging from approximately 4.6% to 16.7%, and it is associated with complications such as lymphedema, hematoma, nerve injury, and upper limb dysfunction, all of which can compromise postoperative quality of life.^5,6 With improved imaging techniques, most patients are now diagnosed at earlier stages, with smaller tumor burdens and lower ALNM rates,⁷ reducing the clinical benefit of both SLNB and ALND. These challenges highlight the urgent need for accurate, non-invasive methods to assess ALN status preoperatively, aiming to improve diagnostic precision while minimizing surgical morbidity.

Conventional imaging techniques—including mammography, ultrasonography, computed tomography (CT), and magnetic resonance imaging (MRI)—are commonly used to evaluate ALNM. However, their sensitivity in detecting micrometastases remains limited.⁸ Diagnostic accuracy varies across modalities and is often influenced by operator experience, introducing potential bias in preoperative assessment.^9,10 Current clinical prediction models mainly rely on imaging findings, clinicopathological features, and conventional tumor biomarkers,^8,11-13 but largely neglect the critical role of the tumor-associated inflammatory microenvironment in metastasis. Inflammation contributes to tumor progression by modulating immune responses and establishing a pro-metastatic niche. Chronic inflammation, in particular, promotes oncogenesis through sustained tissue damage and transformation, with approximately 20% of cancers linked to long-standing inflammatory conditions.^14,15 Recent studies highlight inflammation as a key driver of breast cancer metastasis.¹⁶ The inflammatory microenvironment facilitates lymphatic spread through two main mechanisms. First, tumor-associated neutrophils (TANs) enhance lymphangiogenesis by releasing matrix metalloproteinase-9 (MMP-9) and vascular endothelial growth factor (VEGF), which remodel the extracellular matrix and create pathways for tumor invasion.¹⁷ Second, systemic immune-inflammatory markers—such as the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII)—indicate a shift toward pro-inflammatory, immunosuppressive states that support tumor dissemination.¹⁸ Unlike static anatomical imaging, blood-based inflammatory biomarkers offer dynamic, cost-effective, and readily available insight into tumor-host interactions. Their integration into diagnostic workflows may improve the accuracy of ALNM prediction and support more personalized therapeutic approaches in breast cancer.

This study presents a novel nomogram that combines inflammatory markers, ultrasound features, and pathological parameters to predict preoperative ALNM risk. Integrating these complementary indicators enhances predictive accuracy and supports more informed clinical decision-making.

Materials and Methods

Study Population

In this retrospective study, clinical data were collected from patients who were pathologically diagnosed with breast cancer and underwent surgical treatment at the Department of Breast and Thyroid Surgery, Luohu People’s Hospital of Shenzhen, between January 2020 and March 2024. Based on postoperative pathology, patients were categorized into axillary lymph node metastasis-positive (ALNM⁺) and ALNM-negative (ALNM^-) groups. The inclusion criteria were as follows: (1) Primary early-stage breast cancer confirmed by histopathological examination; (2) Complete clinical records available; (3) Preoperative breast ultrasonography and routine blood testing performed; (4) Axillary lymph node status determined via SLNB or ALND; (5) All surgeries performed by the same surgical team. The exclusion criteria included: (1) Presence of severe preoperative infection; (2) Diagnosis of multiple primary malignant tumors; (3) Receipt of neoadjuvant therapy or evidence of distant metastasis before surgery; (4) Age over 80 years. A total of 287 patients were included, with 233 assigned to the ALNM⁺ group and 54 to the ALNM⁻ group. The patient selection process is outlined in Figure 1. The study was conducted in accordance with the TRIPOD guidelines.¹⁹ Ethical approval was granted by the Ethics Committee of Shenzhen Luohu People’s Hospital (Approval No. 2024-LHQRMYY-KYLL-044; Approval Date: July 21, 2024). A waiver of informed consent was approved due to the retrospective nature of the study.

Figure 1.

Flow Chart. Illustration of Patient Inclusion

Ultrasound Examination

Preoperative breast ultrasonography was performed on all patients by two experienced sonographers, each with more than 5 years of specialization in breast imaging. All examinations were conducted independently, and the sonographers were blinded to the histopathological results. The following breast sonographic features were recorded: tumor size, shape, echogenicity (hypoechoic or hyperechoic), posterior acoustic features, lymph node echotexture, margin characteristics, presence of a hyperechoic halo, calcifications, and color Doppler vascularity (classified as absent, minimal, moderate, or marked). All imaging parameters were classified according to the fifth edition of the American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) Ultrasound Lexicon.²⁰

Clinical Data Collection

Clinical and pathological data were extracted from the hospital’s electronic medical record system. Collected variables included patient age, tumor location, histological grade (I-III), estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), Ki-67, CK5/6, P63, vascular invasion, preoperative hematologic parameters (neutrophil, lymphocyte, monocyte, and platelet counts), and serum tumor markers (CEA, CA125, and CA153). Inflammatory indices were calculated as follows: NLR = neutrophil count/lymphocyte count; PLR = platelet count/lymphocyte count; LMR = lymphocyte count/monocyte count; and SII = (platelet count × neutrophil count)/lymphocyte count. Hormone receptor and HER2 statuses were assessed in accordance with established clinical guidelines.²¹ ER and PR positivity were defined as nuclear staining in ≥1% of tumor cells by immunohistochemistry (IHC). HER2 expression was considered negative with IHC scores of 0 or (+), positive with (+++), and equivocal with (++). Equivocal cases were further evaluated by fluorescence in situ hybridization (FISH). HER2 amplification confirmed by FISH was classified as positive; non-amplified cases were deemed HER2-negative.

Cutoff Determination and Variable Stratification

Optimal cutoff values for tumor markers (CEA, CA125, CA153), hematologic indices (neutrophil, lymphocyte, monocyte counts), inflammatory markers (NLR, PLR, LMR, SII), and tumor size were determined using the Youden index derived from receiver operating characteristic (ROC) curve analysis. All continuous variables were subsequently dichotomized into high and low levels based on these thresholds.

Statistical Analysis

All statistical analyses were performed using IBM SPSS Statistics (version 23.0) and R software (version 4.3.2). Continuous variables following normal distribution were expressed as mean ± standard deviation (SD) and compared using independent t-tests. Non-normally distributed variables were presented as median (Q1, Q3) and analyzed using the Mann–Whitney U test. Categorical variables were summarized as counts and percentages [n (%)] and compared using the Chi-square test. Univariate logistic regression was conducted to examine associations between clinical variables and ALN status. Variables with P < 0.05 were ranked by importance using the random forest algorithm and included in multivariate logistic regression to identify independent predictors of ALN metastasis. Significant predictors were incorporated into a nomogram model. Model performance was evaluated by assessing discrimination (concordance index [C-index] and area under the curve [AUC]), calibration (calibration plots), and clinical utility (decision curve analysis [DCA]). A two-sided P-value <0.05 was considered statistically significant.

Results

Clinicopathological Features of Patients

A total of 287 patients with early-stage breast cancer were included based on predefined eligibility criteria. Patients were stratified into axillary lymph node metastasis-positive (ALNM⁺, n = 233) and metastasis-negative (ALNM⁻, n = 54) groups based on histopathological findings. The mean age was 51 years (range: 24-80 years), with no significant difference between groups (P > 0.05).

The incidence of vascular invasion was significantly higher in the ALNM⁺ group (48.2%) than in the ALNM^- group (15.1%) (P < 0.001), as shown in Figure 2A. Histological grade II-III tumors were more prevalent in the ALNM⁺ group, reflecting a higher malignancy grade (P < 0.01).

Figure 2.

Two Groups of Patients With Axillary Lymph Node Metastasis (A) Proportion of Vascular invasion; (B) Proportion of Tumor Size

Tumors in the ALNM⁺ group were most commonly located in the upper outer quadrant, followed by the lower outer quadrant; inner quadrant tumors were less frequent. Tumor size was significantly larger in the ALNM⁺ group (mean: 26.5 mm, range: 8-65 mm) compared to the ALNM^- group (mean: 12.3 mm, range: 3-50 mm) (P < 0.001), as shown in Figure 2B. Abnormal lymph node echogenicity was detected in 86 patients via preoperative ultrasound, 40 (46.5%) of whom were confirmed to have ALNM. Irregular tumor margins were observed more frequently in the ALNM⁺ group (79.6%) than in the ALNM^- group (46.4%) (P < 0.01). Univariate analysis revealed significant group differences in several laboratory parameters, including LYMPH, NLR, SII, CA125, and CA153 (P < 0.05). Comprehensive clinicopathological and ultrasound imaging data are summarized in Tables 1 and 2.

Table 1.

Univariate Analysis of Preoperative Clinicopathological Characteristics

Characteristic	Overall	ALNM-negative (n = 233)	ALNM-positive (n = 54)	P-value	OR (95%CI)
Age	287
<50	136	110 (47.2%)	26 (48.1%)		1.00 (Reference)
≥50	151	123 (52.8%)	28 (51.9%)	0.856	0.95 (0.52 ∼ 1.71)
Primary site
Upper-outer	153	115 (49.4%)	38 (70.4%)		1.00 (Reference)
Upper-inner	36	28 (12%)	8 (14.8%)	0.742	0.86 (0.36 ∼ 2.06)
Lower-outer	55	49 (21%)	6 (11.1%)	0.035	0.37 (0.15 ∼ 0.93)
Lower-inner	43	41 (17.6%)	2 (3.7%)	0.011	0.15 (0.03 ∼ 0.64)
Vascular invasion
Negative	226	198 (84.9%)	28 (51.8%)		1.00 (Reference)
Positive	61	35 (15.1%)	26 (48.2%)	<0.001	5.25 (2.76 ∼ 10.00)
ER
Negative	72	55 (23.6%)	17 (31.5%)		1.00 (Reference)
Positive	215	178 (76.4%)	37 (68.5%)	0.231	0.67 (0.35 ∼ 1.29)
PR
Negative	79	57 (24.6%)	22 (40.7%)		1.00 (Reference)
Positive	208	176 (75.4%)	32 (59.3%)	0.017	0.47 (0.25 ∼ 0.88)
HER2
Negative	225	187 (80.3%)	38 (70.4%)		1.00 (Reference)
Positive	62	46 (19.7%)	16 (29.6%)	0.114	1.71 (0.88 ∼ 3.34)
Ki67
<40%	203	171 (73.4%)	32 (59.3%)		1.00 (Reference)
≥40%	84	62 (26.6%)	22 (40.7%)	0.042	1.90 (1.02 ∼ 3.51)
P63
Negative	196	157 (67.4%)	39 (72.2%)		1.00 (Reference)
Positive	45	43 (18.5%)	2 (3.7%)	0.025	0.19 (0.04 ∼ 0.81)
Not record	46	33 (14.1%)	13 (24.1%)	0.217	1.59 (0.76 ∼ 3.30)
Tumor-histologic grade
Grade I	43	42 (18%)	1 (1.9%)		1.00 (Reference)
Grade Ⅱ	176	146 (62.7%)	30 (55.5%)	0.037	8.63 (1.14 ∼ 65.17)
Grade Ⅲ	68	45 (19.3%)	23 (42.6%)	0.003	21.47(2.78∼ 166.05)
CEA
<0.965	47	39 (16.7%)	8 (14.8%)		1.00 (Reference)
≥0.965	240	194 (83.3%)	46 (85.2%)	0.731	1.16 (0.51 ∼ 2.64)
CA125
<25.01	251	212 (91.0%)	38 (70.4%)		1.00 (Reference)
≥25.01	36	21 (9%)	16 (29.6%)	<.001	3.10 (1.69 ∼ 5.68)
CA153
<12.65	190	166 (71.2%)	24 (44.4%)		1.00 (Reference)
≥12.65	97	67 (28.8%)	30 (55.6%)	<.001	3.88 (1.84 ∼ 8.18)
NEUT
<3.505	121	103 (44.2%)	18 (33.3%)		1.00 (Reference)
≥3.505	166	130 (55.8%)	36 (66.7%)	0.147	1.58 (0.85 ∼ 2.95)
LYMPH
<3.125	281	231 (99.1%)	50 (92.6%)		1.00 (Reference)
≥3.125	6	2 (0.9%)	4 (7.4%)	0.012	9.24 (1.65 ∼ 51.84)
MONO
<0.425	243	202 (86.7%)	41 (75.9%)		1.00 (Reference)
≥0.425	44	31 (13.3%)	13 (24.1%)	0.051	2.07 (1.00 ∼ 4.29)
PLR
<145	126	108 (46.4%)	18 (33.3%)	0.085	1.00 (Reference)
≥145	161	125 (53.6%)	36 (66.7%)	0.085	1.73 (0.93 ∼ 3.22)
NLR
<3.43	42	28 (12.0%)	14 (25.9%)	0.011	1.00 (Reference)
≥3.43	245	205 (88.0%)	40 (74.1%)	0.011	2.56 (1.24 ∼ 5.29)
LMR
<10.2	273	223 (95.7%)	50 (92.6%)	0.344	1.00 (Reference)
≥10.2	14	10 (4.3%)	4 (7.4%)	0.344	1.78 (0.54 ∼ 5.92)
SII
<776	213	180 (77.3%)	33 (61.1%)	0.016	1.00 (Reference)
≥776	73	53 (22.7%)	21 (38.9%)	0.016	2.16 (1.15 ∼ 4.05)

Table 2.

Univariate Analysis of Preoperative US Characteristics

Characteristic	Overall	ALNM-negative (n = 233)	ALNM-positive (n = 54)	P-value	OR (95%CI)
Tumor size
<14 mm	173	167 (71.7%)	6 (11.1%)		1.00 (Reference)
≥14 mm	114	66 (28.3%)	48 (88.9%)	<.001	20.24 (8.2∼-49.55)
Lymph-node echo
Normal	167	156 (67.0%)	11 (20.4%)		1.00 (Reference)
Abnormality	86	46 (19.7%)	40 (74.0%)	<.001	12.33(5.86∼25.95)
Not record	34	31 (13.3%)	3 (5.6%)	0.642	1.37 (0.36 ∼ 5.21)
Echogenicity
Hypoechoic	219	179 (76.8%)	40 (74.0%)		1.00 (Reference)
Hyperechoic	33	22 (9.4%)	11 (20.4%)	0.049	2.24 (1.01 ∼ 4.98)
Not record	35	32 (13.8%)	3 (5.6%)	0.167	0.42 (0.12 ∼ 1.44)
Margin
Clear	100	92 (39.5%)	8 (14.8%)		1.00 (Reference)
Unclear	151	108 (46.4%)	43 (79.6%)	<.001	4.58(2.05 ∼ 10.23)
Not record	36	33 (14.1%)	3 (5.6%)	0.950	1.05 (0.26 ∼ 4.18)
Posterior acoustic decrease
No	139	107 (45.9%)	32 (59.3%)		1.00 (Reference)
Yes	90	72 (30.9%)	18 (33.3%)	0.589	0.84 (0.44 ∼ 1.60)
Not record	58	54 (23.2%)	4 (7.4%)	0.012	0.25 (0.08 ∼ 0.74)
Hyperechoic halo
No	283	231 (99.1%)	52 (96.3%)		1.00 (Reference)
Yes	4	2 (0.9%)	2 (3.7%)	0.141	4.44(0.61 ∼ 32.27)
Calcification
No	210	172(73.8%)	38(70.4%)		1.00 (Reference)
Yes	77	61(26.2%)	16(29.6%)	<.001	6.59(2.97 ∼ 14.60)
Color Doppler flow
No flow, minimal	165	143 (61.4%)	22 (40.7%)		1.00 (Reference)
Moderate, abundant	90	61 (26.2%)	29 (53.7%)	<.001	3.09 (1.65 ∼ 5.80)
Not recorded	32	29 (12.4%)	3 (5.6%)	0.540	0.67 (0.19 ∼ 2.40)
BI-RADS
3-4	86	74 (31.8%)	12 (22.2%)		1.00 (Reference)
5-6	159	121 (51.9%)	38 (70.4%)	0.068	1.94 (0.95 ∼ 3.94)
Not recorded	42	38 (16.3%)	4 (7.4%)	0.479	0.65 (0.20 ∼ 2.15)

Predictors of ALNM in Early Breast Cancer

Figure 3A illustrates the correlations among 28 clinical and imaging variables. Univariate logistic regression identified 18 factors significantly associated with ALNM, including tumor location, vascular invasion, PR status, Ki67, P63, histological grade, CA125, CA153, lymphocyte count, NLR, SII, tumor size, lymph node echotexture and echogenicity, calcification, posterior acoustic features, margin characteristics, and blood flow (P < 0.05; Tables 1 and 2). A random forest algorithm ranked these variables by predictive importance, highlighting tumor size and lymph node echotexture as the most influential (Figure 3B). Multivariate logistic regression identified six independent predictors of ALNM: vascular invasion (OR = 6.24, P < 0.001), lymphocyte count (OR = 161.04, P < 0.001), NLR (OR = 3.58, P < 0.05), tumor size (OR = 17.63, P < 0.001), lymph node echotexture (OR = 16.70, P < 0.001), and margin characteristics (OR = 5.44, P < 0.001) (Table 3).

Figure 3.

(A) Correlations Between Variables; (B) the Importance Ranking of Significant Variables in Univariate Analysis

Table 3.

Multivariate Analysis of Preoperative Clinicopathological Characteristics Associated With LNM

Characteristic	Category	OR	95% CI	P-value
Vascular tumor embolus	Negative	Reference
Vascular tumor embolus	Positive	6.24	2.04 ∼ 19.08	0.001
LYMPH	<3.125	Reference
LYMPH	≥3.125	161.04	8.45∼ 3068.15	0.001
NLR	<3.43	Reference
NLR	≥3.43	3.58	1.01 ∼ 12.61	0.047
Tumor size	<14 mm	Reference
Tumor size	≥14 mm	17.63	5.23 ∼ 59.46	<0.001
Lymph-node echogenicity	Normal	Reference
	Abnormality	16.70	5.32 ∼ 52.46	<0.001
	Not record	0.94	0.13 ∼ 6.99	0.955
Margin	Clear	Reference
	Unclear	5.44	1.66 ∼ 17.80	0.005
	Not record	2.08	0.26 ∼ 16.50	0.490

Nomogram Construction and Validation

A nomogram was constructed using six independent predictors to estimate the preoperative risk of ALNM in early-stage breast cancer (Figure 4). Each predictor was assigned a score based on its relative contribution, enabling calculation of a total risk score and corresponding probability of metastasis. The model demonstrated excellent discriminative performance, with an AUC of 0.944 (95% CI: 0.906-0.981) (Figure 5A). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 89.3%, 90.7%, 97.6%, and 66.2%, respectively. The C-index was 0.944 and remained robust at 0.929 after 1000 bootstrap iterations, indicating strong accuracy and internal validity. Calibration analysis showed close alignment between predicted and observed outcomes (Figure 5B), with a low Brier score of 0.063, reflecting high calibration precision. DCA further confirmed the model’s clinical utility across a wide range of threshold probabilities (Figure 6), supporting its potential to inform individualized surgical decision-making.

Figure 4.

A Nomogram for Predicting the Risk of Lymph Node Metastasis in Early Breast Cancer

Figure 5.

(A) The Receiver Operating Characteristic Curve (ROC) of the Nomogram; (B) the Calibration Curve

Figure 6.

DCA Analysis of ALNM Prediction Model for Breast Cancer

Discussion

Breast cancer remains one of the most prevalent malignancies among women globally.¹ Despite advances in multimodal treatments—such as surgery, chemotherapy, radiotherapy, targeted therapy, and immunotherapy—accurate preoperative staging of axillary lymph nodes is crucial for guiding personalized treatment planning and optimizing clinical outcomes.²¹ However, conventional assessment methods often lack sufficient specificity, sensitivity, or objectivity, and their invasiveness limits feasibility for widespread screening of ALNM. To overcome these challenges, this study integrated ultrasonographic features, pathological factors, and inflammatory biomarkers to identify independent predictors of ALNM. Based on these variables, a predictive nomogram was developed as a practical tool to enhance individualized clinical decision-making.

Unlike conventional predictive models, our approach incorporates immune-inflammatory biomarkers, which have gained increasing attention in oncology for their clinical relevance.^12,22 These markers offer several advantages: they are non-invasive, low-cost, objective, and easily obtainable, making them practical tools for clinical application. Furthermore, tumor development and progression are closely associated with chronic inflammation and immune dysregulation within the tumor microenvironment.^23,24 Inflammatory indices such as the NLR, PLR, and LMR reflect the dynamic interplay between tumor-promoting inflammation and anti-tumor immunity. Guo et al.²³ reported that elevated preoperative NLR and PLR correlate with poor prognosis in breast cancer. A high NLR typically results from increased neutrophils and decreased lymphocytes, indicating heightened proinflammatory activity and suppressed immune surveillance. This imbalance promotes tumor proliferation, angiogenesis, and metastasis, supporting the utility of NLR as a predictive marker for ALNM. In our multivariate analysis, NLR was identified as an independent risk factor for ALNM in early-stage breast cancer (OR = 6.04, P < 0.05). In contrast, PLR and LMR were not independently associated with ALNM, possibly due to sample variability, selection bias, or inconsistent threshold definitions.

The SII—derived from neutrophil, lymphocyte, and platelet counts—has been proposed as a comprehensive marker of systemic inflammation and immune function.²⁵ Although prior studies linked elevated SII with increased ALNM risk,²⁶ our multivariate analysis did not confirm SII as an independent predictor (P > 0.05). This discrepancy may stem from variations in study populations, cutoff criteria, or limited sample size. Further large-scale, multicenter research is warranted to validate the prognostic value of SII. Lymphocyte profiling has also emerged as a key factor in ALNM risk stratification. Previous studies²⁷ have reported a significant association between elevated lymphocyte levels and both axillary metastasis and locoregional recurrence. Notably, patients with three or more metastatic lymph nodes exhibited higher lymphocyte counts than those with one or two, suggesting that lymphocyte-mediated immune responses may influence the extent of nodal involvement.

Vascular invasion is a key step in the invasion–metastasis cascade of cancer progression. During tumor development, cancer cells secrete growth and angiogenic factors that stimulate fibrous tissue proliferation and upregulate VEGF, promoting neovascularization and lymphangiogenesis. These microenvironmental changes facilitate tumor expansion, invasion, and metastatic spread.²⁸ Our findings reinforce vascular invasion as a significant predictor of ALNM in early-stage breast cancer (OR = 6.24, P < 0.05), aligning with established models of metastatic progression.^29,30 This highlights the importance of microvascular dynamics in predicting nodal involvement and justifies its inclusion in our model.

We also evaluated the utility of ultrasonographic features in predicting ALNM. As the primary imaging modality for assessing axillary lymph nodes and guiding interventions, ultrasound plays a central role in clinical management. Multivariate analysis identified significant associations between ALNM and tumor size (P < 0.001), lymph node echogenicity (P < 0.001), and tumor margin characteristics (P < 0.001). Tumors exceeding 14 mm in diameter were strongly associated with increased ALNM risk (OR = 17.63, 95% CI: 5.23-59.46), consistent with the progressive rise in metastatic potential as tumor volume increases. This may be due to the tumor’s enhanced ability to invade the basement membrane and infiltrate surrounding glandular tissues, increasing access to periductal lymphatics and facilitating lymphatic dissemination. Approximately 75% of breast lymph drains via the axillary pathway. Tumors in the outer quadrants, especially the upper outer quadrant, are more likely to metastasize to axillary nodes, resulting in higher ALNM rates than tumors in other locations. In contrast, inner quadrant tumors preferentially spread through the internal mammary chain. Accordingly, surgical resection and thorough assessment remain essential to achieve accurate staging and guide optimal therapeutic strategies. These anatomical differences in lymphatic drainage likely explain the variation in ALNM incidence by tumor location.^30-32 In our study, most ALNM cases involved tumors in the upper outer quadrant (70.4%), followed by the lower outer quadrant (14.8%), a pattern consistent with previous anatomical and clinical reports.

The tumor margin, where cancer cells often accumulate, is a common site of invasive behavior.³³ Prior research has identified irregular margins as a risk factor for ALNM^11,34; our results confirmed irregular margins as an independent predictor. This may reflect increased collagen deposition at the invasive front, where fibrosis correlates with aggressiveness. VEGF overexpression, stimulated by tumor-secreted growth factors, promotes angiogenesis and lymphangiogenesis, while MMP-9 contributes to basement membrane degradation, enabling local invasion.³⁵ Guo et al.³⁶ proposed that color Doppler flow imaging (CDFI) grading could serve as a predictive marker for ALNM. While our univariate analysis supported this, CDFI grade did not remain significant in multivariate analysis. This may reflect limitations of our retrospective, single-center study and the inherent subjectivity of CDFI interpretation, which could introduce bias.

Of the sonographic features analyzed, only abnormal lymph node echogenicity emerged as an independent predictor of ALNM (OR = 15.44, P < 0.001). Other primary tumor characteristics—including vascularity, echogenicity, hyperechoic halo, posterior acoustic shadowing, and calcification—were not independently associated with ALNM. This contrasts with prior studies linking some of these features, particularly hyperechoic halo and calcification, to metastatic potential.^28,37 These inconsistencies may be due to variations in imaging techniques or population differences. Standardized, multicenter studies are needed to clarify the prognostic value of these sonographic features in ALNM.

Various nomogram models have been developed to predict ALNM in breast cancer, with the goal of guiding individualized treatment decisions.^8,38,39 Among them, the Memorial Sloan Kettering Cancer Center (MSKCC) model is a well-established example, incorporating nine clinicopathological variables and achieving an AUC of 0.754. However, it does not include imaging features, particularly those derived from ultrasonography. As interest in noninvasive imaging increased, researchers began integrating preoperative imaging data into predictive models. Xiong et al.¹¹ developed a nomogram combining conventional ultrasound features with clinicopathological factors, yielding an AUC of 0.705. More recently, advances in artificial intelligence (AI) have enabled the creation of more accurate and personalized models. For example, Li et al.⁴⁰ applied deep learning to ultrasound images to predict ALNM, achieving an AUC of 0.72, with 72.6% accuracy, 65.5% sensitivity, and 78.9% specificity—highlighting the potential of AI-enhanced imaging in preoperative evaluation. Our model incorporates ultrasonographic characteristics, histopathological findings, and preoperative inflammatory markers—all routinely assessed in standard clinical practice—without adding cost or procedural complexity. It demonstrated excellent predictive performance, with an AUC of 0.944 (95% CI: 0.906-0.981) and a C-index of 0.944. Calibration curves indicated strong agreement between predicted and actual outcomes, supported by a low Brier score of 0.068. Moreover, DCA confirmed substantial clinical benefit across a broad range of threshold probabilities. Compared to existing models, our nomogram shows superior discriminative power by integrating multimodal clinical data. It offers a practical, accurate, and visually accessible tool for preoperative ALNM risk stratification in early-stage breast cancer, enhancing clinical decision-making and improving communication between physicians and patients.

This study has several limitations. First, its single-center, retrospective design and relatively small sample size may introduce selection bias. Validation through large-scale, multicenter prospective studies is warranted. Second, the analysis did not include other ALNM-related variables such as preoperative medical history, molecular subtypes, and additional inflammatory markers (eg, C-reactive protein-to-albumin ratio [CAR]), which may affect the model’s generalizability. Third, future research should investigate the molecular pathways linking tumor-associated inflammation and ALNM to support the development of more precise therapeutic strategies.

Conclusion

This study presents a predictive model incorporating vascular invasion, neutrophil-to-lymphocyte ratio, lymphocyte count, tumor size, lymph node echogenicity, and margin irregularity, which demonstrates high predictive accuracy and clinical utility in identifying patients at elevated risk for ALNM preoperatively, thereby guiding precision treatment and helping avoid overtreatment.

Footnotes

ORCID iDs

Xinhua Zhang

Jian Zhang

Xiuming Zhang

Ethical Approval

Ethical approval was granted by the Ethics Committee of Shenzhen Luohu People’s Hospital (Approval No. 2024-LHQRMYY-KYLL-044; Approval Date: July 21, 2024). A waiver of informed consent was approved due to the retrospective nature of the study.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Shenzhen Basic Research Project (No. JCYJ20190812171215641); Shenzhen Key Medical Discipline (SZXK054).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Original data are available upon reasonable request to the corresponding author.*

References

Bray

Laversanne

Sung

, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-263. doi:10.3322/caac.21834

Bartels

SAL

Donker

Poncet

, 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. 2023;41(12):2159-2165. doi:10.1200/JCO.22.01565

Islami

Goding Sauer

Miller

, et al. Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. CA Cancer J Clin. 2018;68(1):31-54. doi:10.3322/caac.21440

Marino

Avendano

Zapata

Riedl

Pinker

. Lymph node imaging in patients with primary breast cancer: concurrent diagnostic tools. Oncologist. 2020;25(2):e231-e242. doi:10.1634/theoncologist.2019-0427

Lee

, et al. Risk factors for a false-negative result of sentinel node biopsy in patients with clinically node-negative breast cancer. Cancer Res Treat. 2018;50(3):625-633. doi:10.4143/crt.2017.089

Pesapane

Mariano

Magnoni

, et al. Future directions in the assessment of axillary lymph nodes in patients with breast cancer. Medicina (Kaunas). 2023;59(9):1544. doi:10.3390/medicina59091544

Geng

Zhang

. Predictive nomogram based on serum tumor markers and clinicopathological features for stratifying lymph node metastasis in breast cancer. BMC Cancer. 2022;22(1):1328. doi:10.1186/s12885-022-10436-3

Dihge

Bendahl

Rydén

. Nomograms for preoperative prediction of axillary nodal status in breast cancer. Br J Surg. 2017;104(11):1494-1505. doi:10.1002/bjs.10583

Chen

Wang

, et al. Feasibility of using negative ultrasonography results of axillary lymph nodes to predict sentinel lymph node metastasis in breast cancer patients. Cancer Med. 2018;7(7):3066-3072. doi:10.1002/cam4.1606

10.

Zhu

Zhou

Jia

Huang

Zhan

. Preoperative axillary ultrasound in the selection of patients with a heavy axillary tumor burden in early-stage breast cancer: what leads to false-positive results? J Ultrasound Med. 2018;37(6):1357-1365. doi:10.1002/jum.14545

11.

Xiong

Zuo

, 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

12.

Liu

Wang

Zheng

Huang

Zhou

. Elaboration and validation of a nomogram based on axillary ultrasound and tumor clinicopathological features to predict axillary lymph node metastasis in patients with breast cancer. Front Oncol. 2022;12:845334. doi:10.3389/fonc.2022.845334

13.

Han

Shen

Duan

. Construction of a comprehensive predictive model for axillary lymph node metastasis in breast cancer: a retrospective study. BMC Cancer. 2023;23(1):1028. doi:10.1186/s12885-023-11498-7

14.

Coussens

Werb

. Inflammation and cancer. Nature. 2002;420(6917):860-867. doi:10.1038/nature01322

15.

Siegel

Miller

Jemal

. Cancer statistics, 2019. CA Cancer J Clin. 2019;69(1):7-34. doi:10.3322/caac.21551

16.

Fridman

Zitvogel

Sautès-Fridman

Kroemer

. The immune contexture in cancer prognosis and treatment. Nat Rev Clin Oncol. 2017;14(12):717-734. doi:10.1038/nrclinonc.2017.101

17.

Coffelt

Wellenstein

de Visser

. Neutrophils in cancer: neutral no more. Nat Rev Cancer. 2016;16(7):431-446. doi:10.1038/nrc.2016.52

18.

Mantovani

Allavena

Sica

Balkwill

. Cancer-related inflammation. Nature. 2008;454(7203):436-444. doi:10.1038/nature07205

19.

Collins

Reitsma

Altman

Moons

KGM

. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. 2015;350:g7594. doi:10.1136/bmj.g7594

20.

Shi

Dong

Tan

, et al. Accuracy of conventional ultrasound, contrast-enhanced ultrasound and dynamic contrast-enhanced magnetic resonance imaging in assessing the size of breast cancer. Clin Hemorheol Microcirc. 2022;82(2):157-168. doi:10.3233/CH-221456

21.

Gradishar

Anderson

Balassanian

, et al. NCCN Guidelines insights: breast cancer, version 1.2017. J Natl Compr Cancer Netw. 2017;15(4):433-451. doi:10.6004/jnccn.2017.0044

22.

Pang

Wang

Jia

Nie

. Predictive value for axillary lymph node metastases in early breast cancer: based on contrast-enhanced ultrasound characteristics of the primary lesion and sentinel lymph node. Clin Hemorheol Microcirc. 2024;86(3):357-367. doi:10.3233/CH-231973

23.

Guo

Liu

, et al. Prognostic value of neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio for breast cancer patients: an updated meta-analysis of 17079 individuals. Cancer Med. 2019;8(9):4135-4148. doi:10.1002/cam4.2281

24.

Wang

Xiao

, et al. Prognostic nutritional index and systemic immune-inflammation index predict the prognosis of patients with HCC. J Gastrointest Surg. 2021;25(2):421-427. doi:10.1007/s11605-019-04492-7

25.

Zhang

, et al. Prognostic significance of prognostic nutritional index and systemic immune-inflammation index in patients after curative breast cancer resection: a retrospective cohort study. BMC Cancer. 2022;22(1):1128. doi:10.1186/s12885-022-10218-x

26.

Wang

. Prognostic prediction of systemic immune-inflammation index for patients with gynecological and breast cancers: a meta-analysis. World J Surg Oncol. 2020;18(1):197. doi:10.1186/s12957-020-01974-w

27.

Rotstein

Blomgren

Petrini

Wasserman

Nilsson

Baral

. Blood lymphocyte counts with subset analysis in operable breast cancer. Relation to the extent of tumor disease and prognosis. Cancer. 1985;56(6):1413-1419. doi:10.1002/1097-0142

28.

Kariri

Aleskandarany

Joseph

, et al. Molecular complexity of lymphovascular invasion: the role of cell migration in breast cancer as a prototype. Pathobiology. 2020;87(4):218-231.

29.

Lale

Yur

Özgül

, et al. Predictors of non-sentinel lymph node metastasis in clinical early stage (cT1-2N0) breast cancer patients with 1-2 metastatic sentinel lymph nodes. Asian J Surg. 2020;43(4):538-549. doi:10.1016/j.asjsur.2019.07.019

30.

Bevilacqua

Kattan

Fey

Cody

3rd Borgen

Van Zee

. Doctor, what are my chances of having a positive sentinel node? A validated nomogram for risk estimation. J Clin Oncol. 2007;25(24):3670-3679. doi:10.1200/JCO.2006.08.8013

31.

Desai

Hoskin

Day

Habermann

Boughey

. Effect of primary breast tumor location on axillary nodal positivity. Ann Surg Oncol. 2018;25(10):3011-3018. doi:10.1245/s10434-018-6590-7

32.

Estourgie

Nieweg

Olmos

Rutgers

EJT

Kroon

BBR

. Lymphatic drainage patterns from the breast. Ann Surg. 2004;239(2):232-237. doi:10.1097/01.sla.0000109156.26378.90

33.

Uematsu

Kasami

Nicholson

. Rim-enhancing breast masses with smooth or spiculated margins on magnetic resonance imaging: histopathology and clinical significance. Jpn J Radiol. 2011;29(9):609-614. doi:10.1007/s11604-011-0612-8

34.

Zong

Deng

Chen

. Establishment of simple nomograms for predicting axillary lymph node involvement in early breast cancer. Cancer Manag Res. 2020;12:2025-2035. doi:10.2147/CMAR.S241641

35.

Kang

Chengrong

Wen

, et al. The relationship between collagen fiber hyperplasia at the site of ultrasound burr sign and prognostic factors in breast cancer. Chin J Med Imaging. 2018;34(12):1820-1824. doi:10.13929/J.1003-3289.201803168

36.

Guo

Dong

Zhang

, et al. Ultrasound features of breast cancer for predicting axillary lymph node metastasis. J Ultrasound Med. 2018;37(6):1353-1354. doi:10.1002/jum.14469

37.

Chen

Liu

Jacobs

. Development of nomograms to predict axillary lymph node status in breast cancer patients. BMC Cancer. 2017;17(1):561. doi:10.1186/s12885-017-3535-7

38.

van den Hoven

van Klaveren

Verheuvel

, et al. Predicting the extent of nodal involvement for node positive breast cancer patients: development and validation of a novel tool. J Surg Oncol. 2019;120(4):578-586. doi:10.1002/jso.25644

39.

Akissue

CLF

Shimizu

Filassi

Maesaka

de Barros

. Axillary lymph node sonographic features and breast tumor characteristics as predictors of malignancy: a nomogram to predict risk. Ultrasound Med Biol. 2017;43(9):1837-1845. doi:10.1016/j.ultrasmedbio.2017.05.003

40.

Lee

Huang

Shih

Chang

. Axillary lymph node metastasis status prediction of early-stage breast cancer using convolutional neural networks. Comput Biol Med. 2021;130:104206. doi:10.1016/j.compbiomed.2020.104206

Establishment of Prediction Model of Axillary Lymph Node Metastasis Before Operation for Early-Stage Breast Cancer

Abstract

Introduction

Methods

Results

Conclusion

Keywords

Introduction

Materials and Methods

Study Population

Ultrasound Examination

Clinical Data Collection

Cutoff Determination and Variable Stratification

Statistical Analysis

Results

Clinicopathological Features of Patients

Predictors of ALNM in Early Breast Cancer

Nomogram Construction and Validation

Discussion

Conclusion

Footnotes

ORCID iDs

Ethical Approval

Funding

Declaration of Conflicting Interests

Data Availability Statement

References