Two human BC cell lines MCF-7 (CVCL_0031) (ER-positive) and MDA-MB-231 (CVCL_0062) (ER-negative) were purchased from Cell Bank of Chinese Academy of Sciences. Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (HyClone) supplemented with 10% (v/v) fetal bovine serum (FBS) (Gibco) and maintained at 37°C under 5% CO₂ in a humidified incubator.

For this study, a total of 70 BC patients from the Fifth Medical Center of PLA General Hospital were enrolled in this prospective open non-randomized study and their blood samples were collected from September 2017 to June 2018. Also, 67 female controls who volunteered to perform early screening for cancers were enrolled, and their blood samples were collected from January 2019 to July 2019 for CTC analysis. The study protocol was approved by the Ethics Committees of the Fifth Medical Center of PLA General Hospital. During the treatment period, some patients received different chemotherapies of metastatic settings, including anthracyclines, taxanes, capecitabine, and vinorelbine. Meanwhile, the others underwent endocrine treatment administrating drugs such as tamoxifen, aromatase inhibitors, and fulvestrant (Supplementary Table S1). The patients were divided into two groups (ER-positive patients and ER-negative patients) based on ER typing of primary tumor foci assayed by clinical immunohistochemistry (IHC) protocol. All patients and healthy volunteers had given written informed consent.

Eligible patients were required to have measurable or evaluable disease, with an Eastern Cooperative Oncology Group (ECOG) performance status score of 0 to 3, their ER and PR status, and the pathology report depicting their histological type and nodal status. PVB was drawn during a therapy cycle and delivered to the laboratory for CTC analysis. Only 2.0 ml of blood was used for CTC enumeration via Pep@MNPs method as described previously (28). CTCs were further analyzed for ER expression, and compared with primary tumor and metastases for ER status. The ER assessment of biopsies was determined by IHC in the Department of Pathology and ER is considered positive when at least 1% positive tumor nuclei exist in the sample as described (31). ER level was classified by an evaluation of the percentage of stained tumor cells and staining intensity. All treatment decisions for the patients were made according to the National Comprehensive Cancer Network (NCCN) clinical practice guidelines. The endocrine therapy selection was based on ER status in any one of primary tumors, metastases, or CTCs. Assessment of disease status was made according to Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria. After follow-up, the relationship between CTC ER expression and clinical outcome was evaluated by statistical analysis.

To establish the fluorescent quantification system assessing ER expression by FACS and immunofluorescence, 1×10⁶ MCF-7 and MDA-MB-231 BC cells were digested with 0.25% trypsin, fixed in 4% paraformaldehyde for 15 min, and blocked with 5% BSA in PBST for 30 min. Subsequently, the cells were incubated with AlexaFluor 647-anti-ER antibody (Abcam, ab196159, 1:300) for 1 h at room temperature. After washing three times with PBS, the cells were incubated with secondary antibody for 1 h at room temperature. Nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI) (Abcam, ab285390, 1:2000). ER expression assessment was performed on the BD flow cytometer (BD Accuri™ C6), and the data were analyzed using Flowjo software (Tree Star). The staining results were obtained using a fluorescence microscope (OLYMPUS IX73). The immunofluorescent images were processed and measured by ImageJ software for quantification analysis.

CTCs were isolated and counted by the Pep@MNPs method as previously described (29, 32). In brief, the biotin-conjugated recognition peptide targeting epithelial cell adhesion molecule (EpCAM) with considerably high binding affinity was linked to streptavidin-conjugated iron oxide magnetic nanoparticles. For each detection, 10 μl pre-vortexed Pep@MNPs was added into 2.0 ml of the PVB sample, and incubated with gentle shaking at 37°C for 1 h. Subsequently, the captured CTCs were isolated under a magnetic field for 30 min and stained by multiple monoclonal antibodies (Abcam), including DAPI for cell nuclei, cytokeratin mix (Abcam, ab264485, 1:200) for positive selection, and CD45-phycoerythrin (Abcam, ab223183, 1:200) for negative selection (leukocytes). CTCs were identified as cells with the molecular feature of CKmix⁺/DAPI⁺/CD45^- under OLYMPUS IX73 fluorescence microscope. The expression of ER in CTCs was assessed with a Cy5-labeled anti-ER antibody (Abcam). Cell images were obtained and analyzed under OLYMPUS IX73 fluorescence microscope for final characterization of the magnetically captured cells.

All statistical computations were performed based on R version 3.6.3. The correlation between the number of CTCs that were detected in a sample and the percentage of ER-positive CTCs were analyzed using Spearman’s rank correlation coefficient. McNemar’s test was used to determine whether a statistically significant difference existed regarding variations in ER status between CTCs and histological results. P values <0.05 were considered statistically significant. P values <0.01 were considered extremely significant. Receiver operating characteristic (ROC) analyses were performed to evaluate the accordance of ER expression between CTCs and IHC results and implemented in python3.7.6. According to the proper threshold, samples were divided into two categories to compare median using the Wilcox test.

To assess the quantitative system for ER analysis, we first evaluated ER expressions in two extensively-used BC cell lines: MCF-7 and MDA-MB-231. Previous studies have shown that MCF-7 BC cells are ER-positive and low metastatic, whereas MDA-MB-231 BC cells are ER negative and highly metastatic (33). Our FACS results also showed that the fluorescence intensity of ER was 6.7×10³ in MCF-7 BC cells, 12.88 times higher than the unstained cells (Figure 1A). The fluorescence intensity was 4.8×10² in MDA-MB-231 BC cells (Figure 1B), being similar with the unstained cells. Therefore, both BC cell lines were suitable for ER quantitative analysis.

Next, MCF-7 and MDA-MB-231 BC cells were isolated using the Pep@MNPs method and stained with Cy5-labeled antibodies against ER. In line with the FACS results, MCF-7 BC cells were positive for ER, while MDA-MB-231 BC cells were negative for ER under a fluorescence microscope (Figure 1C). The mean fluorescent intensity (MFI) of ER in each cell was calculated as described previously (32). Based on the MFI of isolated and stained cells, the threshold was set and defined an ER quantitative system: negative (ER^-, MFI<150) and positive (ER⁺, MFI≥150).

Considering the importance of ER expression, clinical and histological characteristics of BC patients have been routinely assessed. After analysis of primary tumor samples from BC patients, we confirmed that one patient was positive for ER while another patient was negative for ER based on IHC assays (Figure 1D).

Using the previously established Pep@MNPs isolated system, CTCs characterized as CKmix⁺/DAPI⁺/CD45^- were captured from blood samples of BC patients and analyzed for ER expression. Specifically, the MFI of CKmix is equal to or greater than 120 and the MFI of CD45 is less than 80. As expected, in some patients (No. 10250893 as an example), CTCs being positive for ER (CTC^ER+) were observed, while in other patients (No. 10198263 as an example) the results were negative (CTC^ER-) (Figure 1E). Notably, among five CTCs isolated from the patient (No. 10250893), three CTCs were positive for ER, whereas two CTCs were negative for ER (Figure 1E), which presented a heterogeneity of ER expression in CTCs from an individual with BC.

A summary of patient and tumor characteristics was listed in Table 1. A total of 70 BC patients enrolled whose blood samples were collected between September 2017 and June 2018 fulfilled the study criteria. The mean age of the patients was 51 years (range: 30 to 81 years). Among the 70 patients, 29 patients (41%) were in the early phase with a mean age of 52.5, while 41 patients (59%) were in the metastatic phase, with a mean age of 50 years. The available biopsies of a primary tumor and/or metastases in all patients were collected and analyzed. The biopsy analysis by IHC assays showed that a large proportion of patients were positive for ER. The percentage of patients with ER-positive primary tumor was 80% (56/70), while the percentage of patients with available metastases being ER-positive was 69% (20/29). 2.0 ml of the PVB samples was collected from each patient during the treatment to detect CTCs. We enrolled 67 females who volunteered to perform early screening for cancers, and their blood samples were collected for CTC analysis as controls to BC patients. The mean age of the control group was 49 years (range: 30 to 81 years), relatively following the characteristics of BC patients.

In total, 70 blood samples from BC patients were collected and analyzed, and CTCs were detected in 67 cases. The detection rate for CTCs was 95.71%, which was consistent with our previous studies (32). The median number of CTCs in these patients was 8 (range: 1 to 24). By contrast, among 67 female controls, CTCs were detected in 27 volunteers, with the median number of CTCs being 2 (range: 1 to 3). The difference of CTC numbers within 2 ml of blood between BC patients and controls was extremely significant (P<0.0001), with area under the curve (AUC) being 95.81% (Figures 2A, B). The optimal threshold based on precision-recall curve suggested that ≤2 CTCs within 2 ml of blood predict healthy females, whereas more than 2 CTCs predict patients (Figure 2C). Therefore, the cut-off of >2 CTCs per 2 ml blood was applied for further analysis to evaluate that patients were considered CTC-positive.

To evaluate the correlation of ER expression between CTCs and the corresponding primary tumor, samples were analyzed from 38 patients whose primary tumor IHC results and CTCs were both available but metastatic information was not available. The number of CTCs, the number of ER-positive and ER-negative CTCs, the status of ER in primary tumor, and the status of ER in CTCs (CTC^ER+≥1 designated as positive) were all evaluated in these patients. As illustrated in Figure 3A, 32 out of 38 (84.21%) patients were positive for ER in primary tumor, 25 out of 38 (65.79%) patients were positive for ER in CTCs, and 24 out of 38 (63.16%) patients were positive for ER both in primary tumor and CTCs. On the contrary, 5 out 38 (13.16%) patients were negative for ER both in primary tumors and CTCs. These data suggested that the concordance in ER status between CTCs and the corresponding primary tumor be 76.32%.

Then we further investigated the results from 26 MBC patients whose primary tumor IHC results, metastatic IHC results and CTCs were all available for assessment. As shown in Figure 3B, if these two subgroups with available results of primary tumor IHC and CTCs were both included, the concordance in ER status between CTCs and the corresponding primary tumor was 76.56% (49/64 cases). Among 26 MBC patients, metastases and CTCs displayed a concordant ER status in 69.23% (18/26) and a discordant ER status in 30.77% (8/26) of cases. Notably, discordance of ER expression between the primary tumor and metastases was found in four patients.

The ER detection results of 64 patients mentioned above were presented in Figure 3C, and the McNemar’s test demonstrated that the difference of ER detection rate between IHC results and CTCs in all cases was not significant (P=0.08). Assuming that the accuracy of ER detection by IHC was 100%, the AUC was calculated to be 70.57%, presenting a concordance to a certain degree (Figure 3D). Based on the precision-recall curve, the optimal threshold suggested that at least one ER-positive CTC within 2 ml of blood be defined as CTC^ER+ (Figure 3E).

Taken together, these results indicated that there existed a relatively high concordance for ER status between CTCs and the corresponding primary tumor or metastatic IHC results, suggesting that ER expression in CTCs may be conducive to supplement the IHC results.

To further compare the prognostic value of ER expression in CTCs and corresponding IHC results in response to endocrine therapy, we analyzed the data from patients whose primary tumor or metastatic IHC results, CTCs, and the first-line endocrine therapy efficacy when received CTC detection were available. The ER status and endocrine therapy efficacy in the subgroup of patients with available primary tumor IHC and CTC results (Table 2) and the subgroup of patients with available metastatic IHC and CTC results (Table 3) were listed. The highest disease control rate (DCR) was observed in patients whose IHC results and CTCs were positive for ER in both subgroups (79.2% and 86.7%, respectively). The second-highest DCR was shown in patients whose CTCs were positive for ER but IHC results were negative for ER in both subgroups (66.7% and 80.0%, respectively).

Next we focused on 10 patients who had a discordant ER status between the IHC results and CTCs (Table 4). Particularly, among 5 patients with ER-positive IHC results and ER-negative CTCs, 4 cases suggested poor results in response to endocrine therapy, while only 1 case was associated with favorable efficacy in response to endocrine therapy. For these patients, the accuracy of employing ER status in IHC results to instruct clinical treatment was only 20%. For 2 patients with ER-negative (primary tumor) IHC results and ER-positive CTCs, their responses to endocrine therapy were good, revealing that the accuracy of utilizing ER status in CTCs to predict clinical treatment be 100% for these patients. Among 5 patients with ER-negative (metastases) IHC results and ER-positive CTCs, 4 cases resulted in appreciative responses to endocrine therapy, demonstrating that the accuracy of applying ER status in CTCs to show clinical treatment was 80% for these patients. Therefore, the expression level of ER on CTC might better predict the efficacy of endocrine therapy when compared with IHC results.

We also compared the PFS between patients with ER-positive CTCs and those with ER-negative CTCs. The results showed that a significant difference existed between these two groups (P<0.05), and the efficacies of endocrine therapy in patients with ER-positive CTCs was better than those of patients with ER-negative CTCs (Figure 4).

The advantage of ER detection in CTCs using our Pep@MNPs method was that we could conduct a multifaceted test that covered not only the determination of ER-positivity but also the acquisition of other metrics, including the numbers of total CTCs and ER-positive CTCs within 2 ml of PVB. In this study, we also found that ER-positive CTCs and ER-negative CTCs could coexist in the same patient, suggesting that the ratio of ER-positive CTCs to total CTCs (CTC^ER+/CTC) may be relevant to the efficacies of endocrine therapy. When CTC^ER+/CTC=0.25 was set as a critical value and all patients were divided into two groups accordingly, we found an obvious difference in PFS between the group of CTC^ER+/CTC<0.25 and the group of CTC^ER+/CTC≥0.25 (P=0.013) (Supplementary Figure S1A). Thus, we defined the patients of CTC^ER+/CTC<0.25 as high-risk patients and the patients of CTC^ER+/CTC≥0.25 as low-risk patients. Notably, in high-risk patients, the outcome of endocrine therapy was not effective, but in low-risk patients, the endocrine therapy had obvious curative effects. The PFS of high-risk patients was shorter than that of low-risk patients.

To sum up, our data showed that in diversified subgroups of patients with ER-positive CTCs (CTC ^ER+) had longer PFS and better efficacy in response to endocrine therapy than those without ER-positive CTCs (CTC ^ER-), demonstrating that ER status in CTCs might exert a predictive and prognostic significance whereas further evaluation was needed in larger prospective trials.

To further evaluate the potential roles of ER status in CTCs in predicting OS, we assessed the association between factors of interest and OS in BC patients. Interestingly, among patients whose CTCs were negative for ER, there was an inverse correlation between the number of CTC^ER- and OS (Figure 5A). The absence of ER in CTCs might compromise the efficacy of therapies, while the increasing number of ER-negative CTCs might facilitate metastasis and predict shorter OS.

Next, we compared the OS in patients with discordant status of ER in CTCs or IHC results. Remarkably, the probability of OS in patients with CTC^ER+ was substantially different from that in CTC^ER- group (P<0.01, CTC^ER+ median OS=143.9 months vs CTC^ER- median OS=71.5 months) (Figure 5B). Nevertheless, if the patients were grouped into ER-positive and ER-negative based on the ER status of primary tumor or metastatic IHC results, the OS showed no significant difference between these two groups (P=0.52) (Figure 5C and Supplementary Figure S2). These results suggested that the presence of CTC^ER+ can predict patients with better OS outcomes. It is worth noting that between the group of CTC^ER+/CTC<0.25 defined as high-risk and the group of CTC^ER+/CTC≥0.25 defined as low-risk, the analytical results of OS were in accord with those of PFS (P<0.05) (Supplementary Figure S1B), indicating that the value of CTC^ER+/CTC=0.25 was also applicable to predict OS.