The risk of tumors development is currently increasing due to drug resistance and high consumption of alcohol and tobacco worldwide [1] . Breast cancer (BC) is the most pressing issue in modern oncology due to its high morbidity rate [2] [3] . Although surgery is still considered the main method in BC treatment, organ-sparing strategies have recently become leading in medicine, and surgeon-oncologists have been trying to perform operations to reduce the axillary lymph dissection volume [4] - [8] , because in most cases, metastases are not identified during surgically removed axillary lymph nodes (ALN) histological examination. Currently in practice, signal lymph node (LN) biopsy is considered an alternative to surgery for preventing postoperative complications; however, the literature lacks sufficient information on the impact of molecular subtypes on tumor lymphogenic differentiation. In this regard, the LN clinical and morphological parameters study is of great practical importance. Guo Q. (2022) and co-authors showed that BC directly affects ALN, causing morphological changes in the ultrasonography (USG) pattern due to changes in sonography sensitivity. It is believed that in ALN metastatic BC, the application of additional USG tests may increase the diagnostic value of the examination [9] . Bedi R.G. et al. (2008) detected USG examination values for malignant and benign LNs according to their morphological classification [10] ; however, the data on this issue in the literature are contradictory [11] [12] . Therefore, we studied the relationship between changes in the morphological parameters of the ALN along with the extent of breast cancer dissemination and described the lymph nodes that can and should be removed during the elimination of primary lesions without the need for a rather expensive procedure, namely, biopsy.

We confirmed that clarification of the tumor histological subtype exerts clinical significance in predicting metastases to ALN. Some studies have been conducted to determine BC molecular subtypes based on the characteristics of metastatic LNs; however, the results obtained in this field remain ambiguous [13] [14] . Initially, depending on receptor expression, luminal A, luminal B, HER2+, and triple-negative subtypes of BC were distinguished. Further, the St. Gallen International Commission experts identified BC subtypes as luminal A (estrogen receptor positive (ER+) with progesterone receptor positive or negative (PR+/−), HER2− (human epidermal growth factor receptor 2 negative), Ki-67 < 14%; luminal B/HER2−; luminal B/HER2+; HER2+; and triple negative subtype (TNBC) [15] - [18] . Further investigations of the relationship between LN indicators ultrasonography (USG) and cancer histological characteristics serve to minimize costs for detecting metastatic foci before biopsy, and therefore, reduce expenses for tumor treatment.

Object of study: A total of 184 female patients with BC aged 24 - 75 (53.9 ± 0.8) were examined and treated at the Azerbaijan Medical University Oncology Clinic from 2023 to 2024.

Inclusion criteria: patients with BC

Exclusion criteria: patients with BC aggravated by non-cancerous diseases, such as diabetes mellitus, liver inflammation, pneumonia, congestive heart failure, cardiovascular disease, and previous history of cancer localized in foci other than the breast.

All patients were staged according to the TNM system of the American Joint Committee on Cancer 8th edition [19] , and when determining cancer stages, we considered surgical and histopathological results. The identification of cancer molecular subtypes was based on tumor immunohistological examination using the tru-cut biopsy method. Ki-67 expression was determined by immunohistochemical nuclear staining. LN size, shape, structure, and presence of conglomerates were assessed and recorded according to the USG results (LOGIQ C5-Premium, 2012). We found the lymph node length as the longest axis and width perpendicular to the length, less than 12 mm across were accepted as normal nodes. During the ultrasound, we paid attention to such parameters as: loss of ultrasound texture, nodular cortex thickening, round shape. The shape was identified by Tensorflow graph in C++. Lymph node cortex thickening was considered if cortical layer exeeded 3 mm. Areas of lymph node cortex with varying thickening like 3, 2, 5 mm, etc. thickness from different sides was described as thickened unevenly. The device resolution allowed to determine the depth of the lymph node LI, LII, and LIII.

Blood CA-15-3 levels was analyzed on COBAS-e 411 automated analyzer and interpreted vs. control reference 16.4 (4 - 23) U/mL.

Statistical analysis of the results was carried out using the SPSS-26 software package and based on the t-Student-Bonferroni and H-Kruskal-Wallis criteria.

ALN metastases were found in 116 (63.0%) out of 184 BC patients examined at Oncology Clinic in 2023-2024. We found 116 cases of ALN metastases in Luminal A, Luminal B/HER2−, Luminal B/HER2+, TNBC, and HER2 + tumors ( Table 1 ). Regarding cortex structure of total LN, 29,9 % (n = 55) of total ALN were thickened somewhat evenly (<3.0 mm), while 17,9% (n = 33) thickened unevenly (>3.0 mm)), and in 52,2% (n = 96) of ALNs, a poorly visualized cortex with complete disruption of structure (P_H < 0.001) was detected. Amid metastatic ALN, 1.7% (n = 2) had slightly uniformly thickened cortex (<3.0 mm), 16.4% (n = 19) an unevenly thickened (>3.0 mm), and 81, 9% (n = 95) had a complete disruption of cortex structure, so that the cortex layer was poorly visualized (P_H < 0.001) ( Table 2 ). These data indicate that metastatic ALNs can be easily distinguished based on changes in their cortical structure before biopsy. Cortical structure USG simplifies the diagnosis of tumor metastasis to the ALN and should be considered during surgery. According to our results, vast majority—10 of 11 HER2+ subtype BC patients had metastatic ALN (90.9%), indicating that this subtype will most probably end with ALN metastasis; therefore, it is better to remove ALN in this type of cancer as soon as it is diagnosed. For comparison, even in TNBC subtype, ALN metastasis has been detected only in 16 out of 36 (57.1%) cases.

We also observed that subclavian lymph node metastasis was more common in HER2+ subtype, whereas Luminal B/HER2+ subtype metastasized mainly to supraclavicular LN (n = 13; 43.3%; P_H = 0.066).

Note: LN: lymph node, ALN: axillary lymph node, mts: metastasis, ER: estrogen receptor, PR: progesteron receptor, HER2: receptor for human epidermal growth factor 2, TNBC: triple negative breast cancer, G: tumor grade indicator.

Note: N: number of cases, N%: percentage of cases, ALN: axillary lymph node, LN: lymph node, mts: metastasis, TGR: tumor growth rate, N: cancer spread to LN, M: metastasis type, G: tumor maldifferentiation grade, ER: estrogen receptor, PR: progesteron receptor, Ki-67: cancer prognostic factor antibody protein, HER2: receptor for human epidermal growth factor 2.

In general, luminal A subtype covered 19.8%, luminal B/HER2− subtype 40.5%, luminal B/HER2+ subtype – 17.2%, HER2+ subtype – 8.6% of all subtypes with ALN, while TNBC subtype was spread in 13.8% of our BC patients with ALN metastasis ( Table 2 ). Some researchers have also studied the relationship between ALN metastasis and various molecular subtypes of BC [13] [14] [16] [17] and discovered that BC histological subtypes are noteworthy predictors of LN metastasis. According to these studies, the HER−/HER2− subtype is least likely to cause LN metastasis [20] . Results obtained by Ahmed R.H. (2016) showed that patients with the True HER2+ subtype exhibited the highest risk of ALN metastatic lesions vs other subtypes. We also observed multiple ALN metastases in HER2+ cases, but interestingly, the incidence of ALN metastases was highest in both luminal B/HER2+ and luminal B/HER2− subtypes, whereas luminal A and TNBC subtypes were characterized by the lowest risk of ALN metastasis. According to our results, in Luminal A and TNBC subtypes, ALN metastasis can be expected with an accuracy 57%, in Luminal B/HER2− subtype, in 62.7% of patients, while in the case of Luminal B/HER2+ with an accuracy of 66.7%, and in True HER2+ with probability of 90.9%.

In 82 (44.6%) out of 184 patients, neoplasms were localized in the right mammary gland, while in 102 (55.4%) in the left one, and 116 (63.0%) of these patients had metastases to ALN ( Table 2 ). Seven (3.8%) of total BC patients exhibited metastases to liver, seven (3.8%) to the lungs, and 28 (15.2%) to the bones. Depending on the number of metastatic LN number, groups with few (1 - 3) and multiple (> 4) metastases were distinguished. In 69 (37.5%) of them, only few LN metastases were observed, while in 47 (25.5%) cases, multiple metastases have been detected. 39 (21.3%) ALN metastatic patients had metastases to the subclavian and 42 (31.9%) to the supraclavicular LN. In 25 (13.6%) patients few, while in 14 (7.6%) multiple metastases to the subclavian LN were registered. In BC with metastases to supraclavicular LN, 37 (31.9%) patients were diagnosed with few, and 5 (2.7%) with multiple metastases.

According to our study results, in BC patients with metastasis to the ALN, the primary tumor was more likely localized in the left mammary gland (n = 68; 58.6%; P_H = 0.257). Tumor foci number was also found to be a risk factor for ALN metastasis: availability of multiple foci in primary tumor more likely to end with metastasis (n = 43; 38.1% vs. n = 10; 14.7%, P_H < 0.001). Examinations showed that BC patients with tumor radiation type (n = 93; 80.2%; P_H = 0.035) had significantly more ALN metastases.

Englander K. et al. found a correlation between tumor size and the risk of ALN metastasis [21] . According to our study, the invasive ductal (infiltrative) tumors (n = 83; 71.6%; P_H = 0.596) showed a tendency for LN metastasis, and the risk of metastasis to ALN increased with tumor size ( Table 2 ). The results of our study show that the tumor size (37.5 ± 1.6 mm; 10.0 - 106.0 mm) in BC with ALN metastasis is significantly greater vs BC without metastasis (25.6 ± 1.3 mm; 10.0 - 56.0 mm; P_H < 0.001). This result was confirmed by the Receiver Operating Characteristic (ROC) analysis. Thus, according to ROC statistical analysis (AUC = 0.737, 95EI: 0.664 - 0.811; P_H < 0.001), tumor size is important in terms of specificity and sensitivity for predicting ALN metastasis ( Graph 1 ), since the maximum AUC was 0.737.

Graph 2. ROC analysis of hormonal receptors and Ki-67 impact on ALN metastasis in BC patients. Arrows show the risk of metastasis to the ALN: red—direct dependence on the Ki-67 protein, blue—inverse dependence on the estrogen receptors availability.

Our data allow us to conclude that the absence or low PR is characterized by higher expression of the growth factor receptors HER2 (ρ = −0.221; p = 0.003) and tumor spread to the bone (ρ = −0.201; p = 0.006). Ki-67 high expression and levels above the reference line were also found to be negatively correlated with ER availability on the tumor surface (ρ = −0.350; P_H < 0.001), indicating that the less ER, the more Ki-67, and the more aggressive the tumor is. We claim negative correlation between Ki-67 expression and PR (ρ = −0.395; P_H < 0.001) as well. Besides, in patients with ER+ and PR+ tumors, Ki-67 expression protein was below the reference line, namely 14 (ρ = −0.259; P_H < 0.001). However, ER+ patients tended to develop multiple tumors (ρ = 0.192; P_H = 0.009) and bone metastases (ρ = 0.194; P_H = 0.008). In these patients, predominantly G2 differentiation rate (ρ = −0.204; p = 0.006) and high PR expression (ρ = 0.822; P_H < 0.001) were detected. PR+ tumors also exerted a higher risk of metastases to bone tissue (ρ = 0.204; P_H = 0.005) and weakened expression of Ki-67 (ρ = −0.324; P_H < 0.001) and HER2+ (ρ = -0.264; P_H < 0.001).

For patients with Ki-67 > 14 tumors, the ALN shape in this group was found to be mostly spherical (ρ = 0.209; p = 0.004), with multiple (ρ = 0.149; P_H = 0.043), particularly supraclavicular (ρ = 0.162; P_H = 0.028) LN metastases. This group was characterized by decreased expression of ER (ρ = −0.259; P_H < 0.001) and PR (ρ = −0.325; P_H < 0.001), and increased expression of HER2 (ρ = 0.164; P_H = 0.026).

In patients with HER2+ cancer subtype, the LN shape changed (ρ = 0.182; P_H = 0.013), and the risk of multiple metastases (ρ = 0.174; P_H = 0.018), in particular to the supraclavicular LN (ρ = 0.172; P_H = 0.020) and lungs (ρ = 0.167; P_H = 0.018) = 0.024) increased, while PR expression decreased (ρ = −0.233; P_H = 0.001). We note again that when metastasis occurs, the ALN shape changes to sphericity (whereas the non-metastatic LN shape is elliptical) with diffused and/or completely damaged cortex structure ( Table 1 ).

We also noticed that with ALN metastases, the tumor was transferred both to the supraclavicular (n = 42; 36.2%) and the subclavian (n = 39; 43.8%; with P_H < 0.001) regions, but in relation to malignancy and multiple lesions, the subclavian (n = 14; 12.1%) lesions significantly predominate over supraclavicular ones.

When True HER2+ developed, we did not find a single case of ALN with a relatively normal cortical structure. Moreover, in 91% of cases node cortical layer was found unevenly thickened, indicating malignancy. Further, the algorithm of ALN metastasis according to BC subtypes is presented ( Figure 1 ).

According to our data, HER2+ is the type of lymph node that should be removed without any doubt during surgical removal of localized in mammary gland tumor.

1) We identified BC True HER2+ subtype as the most aggressive in terms of metastasis to ALN (91%) with 100% damaged lymph node cortex, which is why in this BC subtype, ALN is better to be removed along with primary tumor foci.

2) According to our results, one of the main breast cancer indicators, CA 15-3 expression is also found to be highest in the Luminal HER2(+) subtype.

3) In cases observed, an increase in Ki-67 was associated with a decrease in ER/PR and increased expression of HER2, and vice versa.

The study was approved by Azerbaijan Medical University Ethics Committee, Protocol No. 26, 2023.