A total of 57 confirmed breast cancer cases treated at Shenzhen People’s Hospital (China) from September 2022 to May 2024 were selected for CTCs detection and MRP2 protein quantification. Ethical approval for the peripheral blood study was obtained from the Ethics Committee of the Shenzhen People’s Hospital (Shenzhen, China; ethics approval number: LL-KY-2024078-01). The clinical characters of 52 patients (five patients were excluded because of incomplete clinical information) were collected, including age, gender, tumor size, ER, PR, HER-2, and other clinicopathological features. The exclusion criteria were defined as follows: 1) incomplete clinical information (e.g., missing pathological subtype, tumor stage, or biomarker status [ER/PR/HER2]); 2) concurrent diagnosis of other malignant tumors (to avoid CTC interference from non-breast cancer sources); 3) severe hepatic or renal dysfunction (estimated glomerular filtration rate <30 mL/min/1.73m² or Child-Pugh Class C), as organ dysfunction may affect CTC survival in circulation; 4) history of hematological diseases (e.g., leukemia, lymphoma) that could confound leukocyte depletion during CTC isolation; 5) inability to cooperate with peripheral blood collection (e.g., severe coagulation disorders). The data on breast cancer type and stratification by therapy received of these patients are shown in Supplementary Table 1 .

In order to prevent contamination of cells resulting from puncturing the skin veins, the initial 2 ml of peripheral blood was discarded. Subsequently, 5 ml of blood was collected into an ethylenediaminetetraacetic acid (EDTA) tube (Becton Dickinson, Shanghai, China). The Canpatrol System (SurExam Biotech, Guangzhou, China) was used for analysis within 4h after blood collection.

The previously reported Canpatrol platform was used to enrich and identify CTCs from blood (13). First, collect 5 ml of peripheral blood sample from the patient using an EDTA tube, invert, and mix well. Next, 15 ml of red blood cell lysis buffer (154 mM NH4Cl, 10 mM KHCO3, and 0.1 mM EDTA) was added, mixed again, and allowed to stand at room temperature for 30 min to lyse red blood cells. Subsequently, the sample at 500 g for 5 min, the supernatant was removed, and the cell pellet was resuspended in PBS. Next, the remaining cell pellet was fixed with 4% formaldehyde (final concentration) for 8 min. After fixation, the cells were transferred to a filter tube containing a filter membrane (SurExam Biotech, Guangzhou, China) with a pore size of 8 µm, and the cells were filtered onto the membrane using a vacuum filtration pump (Auto Science, Tianjin, China). Finally, the filtered cell membrane sample was fixed with 4% formaldehyde at room temperature for 1 h.

The fixed membrane samples were washed three times with PBS and placed in a 24-well plate. Proteinase K (0.1 mg/mL; Qiagen, Hilden, Germany) was added for treatment, and the samples were left to stand at room temperature for 1 h to increase cell membrane permeability. The samples were then washed three times with PBS, followed by the addition of specific capture probes for hybridization, including epithelial biomarkers (EpCAM, CK8/18/19), mesenchymal biomarkers (vimentin and twist), and the leukocyte marker CD45. The probes were synthesized by Shanghai Sangon Bioengineering Company (Shanghai, China), and the probe sequences are listed in Table 1 . The hybridization reaction was performed at 40 °C for 3 h. Unbound probes were washed three times with 1000 μl of eluent (formulation: 0.1×SSC (Sigma, St. Louis, USA)). Subsequently, a volume of 100 μl of the pre-amplification solution was introduced, comprising 30% horse serum (Sigma, St. Louis, USA), 1.5% sodium dodecyl sulfate (Sigma, St. Louis, USA), 3 mM Tris-HCl (pH 8.0), and 0.5 fmol of pre-amplification probes as detailed in Table 2 . The samples were then incubated at 40 °C for 30 min to facilitate the reaction involving the signal amplification probes. This process involved the conjugation of capture probes to branched DNA (b-DNA) signal amplification probes, resulting in the formation of a branched structure. After cooling the membrane, it was washed three times with 1000 μl of 0.1×SSC eluent, then incubated with 100 μl of amplification solution (containing 30% horse serum, 1.5% sodium dodecyl sulfate, 3 mM Tris-HCl (pH 8.0), and 1 fmol pre-amplification probes ( Table 2 )) at 40 °C for 30 min.

Subsequently, three fluorescent protein-labeled probes (Shanghai Sangon Bioengineering Company, Shanghai, China) ( Table 2 ) were added: Alexa Fluor 594 (for labeling epithelial biomarkers EpCAM, CK8/18/19), Alexa Fluor 488 (for mesenchymal biomarkers vimentin and twist), Alexa Fluor 750 (for leukocyte marker CD45), and Alexa Fluor 647 (for MRP2 mRNA), followed by incubation at 40 °C for 30 min. Finally, the samples were eluted with 0.1×SSC and stained with 4’, 6-diamidino-2-phenylindole (DAPI) (Louis, USA) for nuclear staining for 5 min and observed under a 100× oil immersion lens using an automated fluorescence scanning microscope (ZEISS, Germany). Red and green fluorescent signal dots represent the expression of epithelial and mesenchymal genes in CTCs, respectively, while white signal dots represent the gene expression of the leukocyte marker CD45. The purple signal dots represent MRP2 expression (a cut-off point was set at two signal dots: “low expression” and “high expression”). The CTCs classification criteria are shown in Table 3 . The study flowchart is shown in Figure 1 .

Data were analyzed using SPSS software (version 21.0; SPSS Inc.). Continuous variables were documented as medians accompanied by their respective ranges, while categorical variables were expressed in terms of frequency and percentage. Patients were divided into two groups according to high and low CTC/cluster numbers and high and low gene expression levels. The chi-square test was employed to assess the associations between clinicopathological characteristics and the expression of CTCs or MRP2 expression in CTCs. Statistical significance was set at P < 0.05.

The clinical characteristics and CTCs detection data of the 52 patients with breast cancer are shown in Table 4 . Most patients were female (98.1%), with a median age of 47 years (range: 34–72 years; 11.5% < 40 years). The median tumor size was 1.5 (range: 0.2-6), and 79% had tumors ≤ 2. The pathological stage had a median of 1 (range: 0-4), with 76.9% of patients at stage ≤ II.

CTCs were detected (≥ 1/5 ml) in 94.2% (49/52) of the breast cancer patients ( Table 4 ). The median number of CTCs was 15 (range: 0-108) in 5 ml peripheral blood samples from all 52 patients. Using the Canpatrol™ CTC detection platform, all the separated CTCs were classified into three distinct EMT categories: E, H, and M types, utilizing various labeled mRNA probes. As illustrated in Figure 2 , E-type cells correspond to the epithelial type, H-type cells denote the hybrid epithelial/mesenchymal type, and M-type cells are indicative of the mesenchymal type. E-type CTCs showed a median of 3.5 (range: 0-63), with 76.9% having ≥ 1/5 ml. H-type CTCs had a median of 7.5 (range: 0-43), and 88.5% had ≥ 1/5 ml. M-type CTCs had a median of 0 (range: 0-5), with 88.5% having < 1/5 ml. MRP2 (+) CTCs (MRP2 signal dots ≥ 1) had a median of 13 (range: 0-84), with 48.9% having ≤ 13/5 ml.

The association between CTCs and clinical variables in 52 patients with breast cancer is presented in Table 5 . For gender (1 male, 51 females), all CTC-related types (total, E, H, M, H + M, MRP2 (+)) showed no significant differences (P > 0.05, e.g., total CTCs: median 23 vs. 15 (0–108), P = 0.654). In terms of age (< 40 years: 6 patients; ≥ 40 years: 46 patients), no significant differences were observed for all CTC-related types (P > 0.05, e.g., total CTCs: median 26.5 (0-108) vs 14 (0-39), P = 0.332). For tumor size (≤ 2:39 patients; > 2:13 patients), significant differences were found in H-type CTCs (median 6 (0-30) vs 15 (2-43), P = 0.044) and MRP2 (+) CTCs (median 12 (0-35) vs 17 (3-84), P = 0.039), suggesting that CTCs may play an important role in the progression of breast cancer. Other CTC-related parameters (total, E, M, and H + M) showed no significant differences (P > 0.05). Regarding pathological stage (≤ II: 40 patients; > II: 12 patients), no significant differences were detected for all CTC-related types (P > 0.05, e.g., total CTCs: median 14 (0-108) vs 19 (6-37), P = 0.192). Overall, only H-type and MRP2 (+) CTCs exhibited significant differences between the different tumor sizes. Most CTC-related types showed no significant associations with gender, age, or pathological stage.

The mRNA expression of MRP2 was detected using the MRIA assay ( Table 6 ). The results showed that the MRP2 gene was expressed in 94.2% (49/52) of the CTC-positive patients. The median number of MRP2-positive CTCs was 2 in 5 ml of blood samples from patients. Further study showed that the expression rates of MRP2 in different types of CTCs were different: 76.3% (271/355) in the E-type CTCs and 85.9% (505/588) in the (H+M)-type CTCs. The difference in MRP2 expression between the two groups was statistically significant (P < 0.001; Table 6 ).

An analysis was conducted to investigate the correlation between the expression levels of MRP2 in circulating tumor cells (CTCs) and the clinical characteristics of the patients, as presented in Table 7 . The median count of MRP2-positive CTCs was determined to be 13 per 5 ml of blood samples, leading to the stratification of patients into two distinct groups: those with >13 MRP2-positive CTCs and those with ≤13 MRP2-positive CTCs per 5 ml of blood. A significant association was observed between the presence of >13 MRP2-positive CTCs and tumor size, in contrast to those with ≤13 MRP2-positive CTCs (P < 0.001). However, no significant differences were noted concerning other clinicopathological characteristics (P > 0.05).

EMT levels were classified based on the ratio of (mixed-type + mesenchymal cell phenotypes) to total cells, divided into four grades: grade 1 (0-24%), grade 2 (25%-49%), grade 3 (50%-74%), and grade 4 (75-100%). According to the expression level of the MRP2 protein, the patients were divided into four groups. When the number of MRP2 expressions in CTCs is zero, it is negative; when it is greater than zero, it is positive. In CTCs, if the number of MRP2 signal points is ≤ 2, it is low expression; if it is > 2, it is high expression. The expression level of MRP2 was divided into four grades: 0-24% was grade 1, 25%-49% was grade 2, 50%-74% was grade 3, and 75-100% was grade 4. Spearman’s rank correlation test showed a significant correlation between EMT levels and MRP2 (+) CTCs grade (R = 0.341, P = 0.013) and high MRP2 expression (R = 0.283, P = 0.042) ( Table 8 ), but no significant correlation between EMT levels and MRP2 low expression (R = -0.01, P = 0.945). In addition, these correlation scatter plots show how different CTC subtypes (Epithelial, Hybrid, Mesenchymal, (H+M)) relate to their MRP2 expression. All have strong positive correlations: Epithelial (R = 0.98, P < 2.2e−16; Figure 3A ), Hybrid (R = 0.97, P < 2.2e−16; Figure 3B ), Mesenchymal (R = 0.89, P < 2.2e−16; Figure 3C ), and (H+M) (R = 0.97, P < 2.2e−16; Figure 3D ). Therefore, MRP2 expression is closely related to CTC subtype counts, providing a new perspective for further investigation of the mechanisms of drug resistance in breast cancer.