This cross-sectional study included patients who underwent routine surgical resection and were pathologically confirmed as having breast cancer at the First Affiliated Hospital of Henan University of Traditional Chinese Medicine between January 2018 and December 2020. This study was approved by the ethics committee of the First Affiliated Hospital of Henan University of Traditional Chinese Medicine. The requirement for informed consent was waived due to the retrospective nature of the study. The inclusion criteria were: 1) available clinical data, 2) diagnosis of breast cancer and available tissue specimens, and 3) available peripheral blood samples. The exclusion criteria were: 1) incomplete clinical data or 2) recipient of neoadjuvant treatments.

The demographic and clinical data of the patients were collected, including age, sex, lymph node metastasis, carcinogenic molecular detection, pathological type, tumor size, and Ki-67 expression. Molecular subtypes included Luminal A, Luminal B, human epidermal growth factor receptor-2 (HER-2) amplification, and basal-like. Luminal A type showed estrogen-receptor (ER) positive or progesterone-receptor (PR) positive and PR high expression (≥20%), HER2 negative, or Ki-67 low expression (<14%). Luminal (luminal or hormone receptor-positive) type B was divided into two subtypes, Luminal B (HER-2 negative) and Luminal B (HER-2 positive): Luminal B (HER-2 negative): ER and/or PR positive, HER-2 negative, and high expression of Ki-67 (greater than or equal to 14%). Luminal B (HER-2 positive): ER and/or PR positive, HER- 2 overexpression or proliferation, and any level of Ki-67. HER-2 amplification type, also known as HER2 positive (non-Luminal) type, showed ER and PR deletion, HER-2 overexpression or proliferation, and any level of Ki-67. Basal-like breast cancers have low expressions of ER, PR, and HER2 and high expressions of markers of breast basal or myoepithelial cells, such as caveolin-1, p63.

Peripheral venous blood (5 mL) was collected from an antecubital vein using a vacuum EDTA anticoagulant tube and transferred to a sample storage tube. The tube was inverted >10 times and incubated at 15°C–30°C for 30 min to lyse the cells. The blood was centrifuged at 1850 rpm for 5 min. The supernatant was discarded, and 4 mL of PBS and 1 mL of RI fixative solution were added, vortexed, and let to stand for 8 min at room temperature. The liquid was poured into the filter connected to a vacuum pump with a negative pressure of −0.06 MPa. The fixative (1 mL) was added for 1 h. The membrane was dehydrated with gradient alcohol to capture the cells. Labels for EP-CAM, CK8, and CK8/18 (as epithelial markers) and Vimentin (as mesenchymal markers) produced by CanPatrol Biotechnology Co., Ltd. were used for mRNA fluorescence in situ hybridization. Fluorescence from any of the epithelial/mesenchymal makers are indicative of epithelial/mesenchymal-type circulating tumor cells. The membrane was scanned, and the count of each type of cell was analyzed using Isis5 and Metafer3 fluorescence analysis software produced by German Midas. Red fluorescence represented epithelial-type CTCs (E-CTC), green fluorescence represented mesenchymal-type CTCs (M-CTC), and both red and green represented mixed-type CTCs (E/M-CTC).

The postoperative specimens were fixed in 10% formalin, dehydrated, paraffin-embedded, and serially cut into 4-μm sections. After dewaxing and rehydrating, the sections were put into ethylenediaminetetraacetic acid (EDTA) solution (pH 8.0) produced by Beijing Soleibao Technology Co., Ltd. at 100°C for 20 min. The endogenous peroxidase activity was blocked with 3% H₂O₂ for 10 min. After rinsing with phosphate buffer solution (PBS), the sections were incubated with an anti-rabbit primary antibody for Claudin-4 (Shanghai Jiehao Biotechnology Technology Co., Ltd., CRM-1251) at 37°C for 1 h. After washing with PBS, the sections were incubated with the secondary antibody (Roche Shanghai Co., Ltd., J19565) and rinsed. The sections were revealed with diaminobenzidine (DAB), stained with hematoxylin, dehydrated, and mounted. The cells were counted in 10 randomly selected visual fields at ×400 magnification. For number scoring, <5% of positive cells were scored zero, 5%–25% was one, 26%–50% was two, and >50% was three. For intensity scoring, no staining was zero, light yellow was one, brown was two, and tan was three. The number and intensity scores were multiplied; zero points were negative, one to two represented 1+, three to four represented 2+, and five to six represented 3+. Claudin-4 negative and Claudin-4 1+ were considered low expressions. Claudin-4 2+ and 3+ were considered high expressions. The scores were determined by two pathologists with 10 years of experience, and the average value was taken as the final result.

The patients were followed up every 2 months by telephone contact after enrollment. Patients’ vital signs and disease progression were recorded. If necessary, the patients were admitted to the hospital to evaluate their progress. The expected follow-up time was 24 months. The follow-up endpoint was all-cause death. The follow-up deadline was 30 October 2020.

All data were analyzed using SPSS 20.0 (IBM, Armonk, NY, United States). Continuous data were expressed as means ± standard deviation and analyzed using Student’s t-test. Categorical data were presented as n (%) and analyzed using the chi-square test. The relationship between CTCs and Claudin-4 expression was analyzed using Spearman rank correlation and the Kaplan-Meier method was used to draw the survival curve. Log-rank test was used to compare the differences. A Multivariate COX risk regression model was established to analyze the independent adverse factors of prognosis. The statistical analysis of the data from two groups was performed using a t-test. The comparisons of multiple groups were performed by one-way ANOVA and then an LSD-t test. p < 0.05 was considered to be significant. Two-sided p-values < 0.05 were considered statistically significant.

Sixty-five breast cancer patients, 39 to 61 (median, 49) years of age, were included. All patients were women, including 51 with invasive carcinoma and 14 cases of ductal carcinoma in situ. There were 48 patients with a tumor <5 cm and 17 with a tumor >5 cm. There were 21 patients with lymph node metastasis and 44 without. There were 12 patients with Ki- 67 < 14% and 53 with Ki-67 ≥ 14% (Table 1).

Among the 65 patients, there were 11 Claudin-4-negative tumors, 27 with Claudin-4 +, 22 with Claudin-4 2+, and five with Claudin-4 3+ (Figure 1). Most patients with high Claudin-4 expression had lymph node metastasis (55.6% vs. 15.8%, p < 0.001 vs. without low Claudin-4 expression). Most patients with high Claudin-4 expression had a tumor >5 cm (51.9% vs. 7.9%, p < 0.001 vs. low Claudin-4 expression) (Table 2).

CanPatrol CTC enrichment and in situ hybridization were used to monitor the number and classification of CTC in the blood of 65 patients. Red fluorescence was used to represent epithelial CTC, and green fluorescence was used to represent interstitial CTC. CTC was divided into epithelial type (E-CTC), mixed type (E/M-CTC), and interstitial type (M-CTC). The number of CTCs in patients aged ≥50 was significantly high (11.8 vs. 9.1, p < 0.001 vs. Age < 50). The number of CTCs in patients with lymph node metastasis was high (11.2 vs. 6.5, p < 0.001 vs. without lymph node metastasis). There are significant differences in the number of CTCs in different molecular types of breast cancer (p < 0.001) (Table 3).

The number of CTCs was higher in patients with high Claudin-4 expression (11.7 vs. 7.4, p < 0.001). A more detailed analysis revealed that a high Claudin-4 expression was associated with a lower count of E-CTCs (3.4 vs. 5.0, p = 0.033) but high counts of E/M-CTCs (5.0 vs. 3.5, p = 0.021) and M-CTCs (4.4 vs. 1.1, p < 0.001) (Table 4 and Figure 2). Spearman correlation analysis showed that the intensity of Claudin-4 was positively correlated with CTC (r_s = 0.43, p = 0.001) (Figure 3).

E-CTC in breast cancer patients with lymph node metastasis was significantly higher than that in the lymph node metastasis group (3.4 vs. 1.4, p < 0.001). E/M-CTC was significantly lower than that in the lymph node metastasis group (1.5 vs. 4.3, p = 0.002), and the M-CTC count was lower than that in the lymph node metastasis group (1.6 vs. 6.4, p < 0.001) (Table 5).

By the last follow-up, the average survival time of 65 patients with high expressions of Claudin-4 (n = 27) was 18.22 months, and the average survival time of the low-expression group (n = 38) was 21.81 months. The survival rate of the high-expression group was significantly lower than that of the low-expression group (Log Rank Χ² = 4.71, p = 0.030) (Figure 4).

The results showed that CTC counts (HR = 1.3, p < 0.001), Claudin-4 (HR = 4.6, p = 0.008), and lymph node metastasis (HR = 12.9, p = 0.001) were independent prognostic factors for poor prognosis (Table 6).

Our analysis showed that E/M-CTC (HR = 1.9, p = 0.001) and M-CTC (HR = 1.5, p = 0.001) were independent influencing factors of poor prognosis in breast cancer patients (Table 7).