Circulating tumour cells at baseline and late phase of treatment provide prognostic value in breast cancer

To determine the prognostic value of the timing of circulating breast tumour cell measurement during treatment, peripheral blood from 164 patients with breast disease was collected. Circulating tumour cells (CTCs) were enriched by using immunomagnetic nanospheres (IMNs) and were identified by using immunofluorescent staining. The CTC shows nuclear-positive, EpCAM-positive, CK19-positive, and CD45-negative. Patients with CTC positivity (> 19/7.5 mL blood) had shorter progression-free survival (PFS) and overall survival (OS) than those with negative results (≤ 19/7.5 mL blood) at baseline. Surgery caused an increase in the number and prevalence of CTCs, and the effect disappeared on day 14 after surgery. During adjuvant chemotherapy, CTCs detected before therapy was only correlated with PFS; however, CTCs at the end of adjuvant chemotherapy were correlated with both PFS and OS. The PFS and OS of the CTC-positive group were significantly shorter than those of the CTC-negative group at the end-point follow-up visit. The prognostic value of CTCs at different measurement time points was demonstrated during breast cancer treatment. Surgery and chemotherapy affected the prevalence of CTCs, leading to different prognostic relevance of CTCs at different treatment stages. CTCs detected at baseline or in the late phase of treatment are preferable for prognosis.

www.nature.com/scientificreports/ breast cancer. Although some studies have compared the prognostic relevance of CTCs in breast cancer before and after treatment 8,19,20 , few studies have included different treatment stages.
On the other hand, the capture efficiency of the CTC enrichment method also affects its prognostic accuracy. Although epitope-based CTC enrichment methods (such as CellSearch system) may miss some CTCs that do not express the corresponding antigens, they are still the most commonly used CTC enrichment methods because of the simple and convenient operation. Therefore, when using epitope-based CTC enrichment method to study the prognostic value of CTC, it is necessary to choose a CTC enrichment method with high capture efficiency, so as to avoid further loss of CTC. The capture efficiency of epitope-based CTC enrichment method is mainly affected by the surface to volume ratio of the capture medium. Magnetic nanoparticles have a relatively high surface to volume ratio, inducing the high binding ability and CTC capture efficiency. However, the magnetic response of magnetic nanoparticles is not fast enough, and they suffer higher loss rate. Immunomagnetic nanospheres (IMNs) with quick magnetic response overcome this defect of magnetic nanoparticles 21 . IMNs are made of Pst-AAm-COOH as the core, and coated with 5 layers of nano-γ-Fe 2 O 3 to get a large magnetic saturation value 21 , so that the IMNs could be captured within 1 min with commercial magnetic medium 21 . Therefore, the whole progress of CTC enrichment could be completed within 5 min with ~ 94% recovery rate 21 .
Therefore, by using IMNs to capture the CTC, the prognostic relevance of CTCs at the time point of presurgery, pre-adjuvant chemotherapy, final adjuvant chemotherapy, and perioperative period was demonstrated to clarify which time points the CTC has prognostic value.

Results
Optimization of CTC enrichment method and threshold demarcation. Quick-response IMNs can capture tumour cells in blood with an efficiency of more than 94% after only 5 min of incubation 21 . Therefore, it is a rapid and sensitive CTC detection method. However, the previously described method recommends using 1.5 mL of blood to detect CTCs, which is less than the sample volume (7.5 mL) in the CellSearch system. Because a larger volume of blood could increase the sensitivity of CTC detection, we increased the blood sample volume to 7.5 mL for the IMN-based CTC detection.
To accurately and efficiently enrich CTCs in a 7.5 mL blood sample, we optimised the capture conditions by performing experiments with different concentrations of IMNs and incubation times, and then compared the capture recovery rate of each condition. The recovery rate was calculated by dividing the number of captured MCF-7 cells by the number of input MCF-7 cells. As shown in Fig. S1a, 0.13 mg/mL IMNs gave a 97.1% recovery rate, which was higher than that of 0.07 mg/mL IMNs (85.7% recovery rate) and 0.20 mg/mL IMNs (91.4% recovery rate). Incubation times of 5 min and 10 min showed recovery rates of over 95% (Fig. S1b), but 1 min and 15 min incubation times showed lower recovery rates. Therefore, 0.13 mg/mL IMNs and 5-min incubation were used in subsequent experiments.
To investigate the capture specificity of IMNs, three kinds of epithelial tumour cells (HCT116, A549, and MCF-7) and leukocytes were added to PBS (7.5 mL) and captured by IMNs. The recovery rates of all epithelial tumour cells were over 92%, but the recovery rate of leukocytes was 16.6% (Fig. S1c). The non-specific capture of leukocytes may be caused by the non-specific absorption of IMNs. Therefore, we further increased the washing time, and non-specific adsorption was reduced by washing multiple times (Fig. S1d). In addition, leukocytes could be distinguished by CD45 staining, thus, it did not affect CTC counts. Leukocyte residues are inevitable. But it is no significant difference in the number of leukocytes between the breast cancer and healthy donors (Fig. S1e).
Anti-EpCAM antibody also captures epithelial non-tumour cells in the blood. Thus cells displaying nuclearpositive, CK19-positive, and CD45-negative by immunocytochemistry, were also observed in the blood samples from healthy donors (Fig. S1f). Blood samples from 19 healthy donors were collected and tested for CTC-positive threshold demarcation, and the results showed that the mean epithelial cell count was 8.89, and the standard deviation (SD) was 3.72. Therefore, the threshold of the CTC-positive sample was 19, calculated by adding the mean epithelial cell counts and three times the SD.
CTCs were detected in the peripheral blood of the patients. CTCs were wrapped in magnetic beads and were nuclear-positive, EpCAM-positive, CK19-positive, and CD45-negative (Figs. 1 and S3). Interestingly, we found that CTCs were also detected in 42.3% of the patients who were diagnosed with benign breast diseases (Figure S1f). This might be related to the presence of epithelial cells in the blood of different patients due to personal condition or medication, indicating that CTCs are not a good biomarker for diagnosing benign breast diseases.  [22][23][24][25] . There were no significant differences in intrinsic subtype, stage, risk categories, tumour size, lymph node status, oestrogen receptor (ER) status, progesterone receptor (PR) status, human epidermal growth factor receptor 2 (HER2) status, and Ki67 status between the CTC-positive and CTC-negative groups of patients (P > 0.05).
CTCs in the perioperative period and their prognostic relevance. Surgery can affect the number of CTCs in a patient's blood 17 . We investigated the CTCs during the perioperative period. The number of CTCs significantly increased on the first day after surgery (P = 0.0001) (Fig. S4a), and the CTC prevalence increased from 44.6% to 65.8% (P = 0.0073). The prevalence increase persisted until 7 d after surgery (P = 0.0147), and the number of CTCs levelled (P = 0.0634) (Fig. S4b). There was no significant difference between the number of CTCs 14 d after surgery and the number of CTCs before surgery (P = 0.3718), and the CTC prevalence decreased to 54.2% at 14 d after surgery (P = 0.2899) (Fig. S4c). Therefore, CTC count varied in the perioperative period. It is necessary to further investigate the prognostic value of CTCs at different measurement time points during the perioperative period.

Discussion
CTCs have been proven associated with the prognosis of breast cancer. However, therapy can affect CTCs, especially surgery 15 . It has been reported that the number of CTCs increases 3-4 days after surgery 17 . Our data showed that the number and prevalence of CTCs increased at 1 d and 7 d after surgery. No significant differences in PFS and OS were observed between days 1 and 7 after surgery for CTC-positive and CTC-negative patients, but CTCs measured on day 14 after surgery were related to PFS. Therefore, the measurement time point of CTCs affects their prognostic value.  www.nature.com/scientificreports/ CTCs at baseline showed prognostic relevance in many reports [26][27][28] as well as in the present study, because most CTCs at baseline were measured before any new line of treatment thus the effects of therapy were limited.  www.nature.com/scientificreports/ However, some patients missed the CTC measurement time point at baseline, therefore if these patients want to benefit from the prognosis of CTCs, another CTC-measurement time point, at which the number of CTCs has prognostic value, should be determined. Therefore, we investigated the prognostic relevance of CTCs before and after adjuvant chemotherapy in addition to the perioperative period. The prevalence of CTCs before adjuvant chemotherapy (51.7%) was higher   www.nature.com/scientificreports/ than that after adjuvant chemotherapy (46.7%), indicating that chemotherapy had an impact on CTCs. However, CTCs after chemotherapy showed more associations with PFS and OS than those before adjuvant chemotherapy, revealing that a CTC-measurement time point closer to the later phase of the treatment had a more obvious prognostic value. To validate this hypothesis, the prognostic relevance of CTCs at the end-point follow-up visit was investigated. Although the end-point follow-up visit did not occur after a fixed period, CTCs at the end-point follow-up visit showed a prognostic value. Therefore, in addition to the baseline, CTCs at the late phase of the disease or at the end of treatment also have prognostic value.
The research of the present study has some limitations: (1) It was a single-centre prospective study and the number of patients enrolled was limited. Patient stratification analysis did not proceed because of the limited number of patients. The mobility of patients resulted in a smaller sample size at follow-up visit and the treatment monitoring; (2) Although the performance of IMNs for CTC collection was validated, this method was not compared to other platforms such as CellSearch system; (3) More treatment stages, such as neoadjuvant chemotherapy, should be considered for inclusion in a study to assess the prognostic value of CTCs measured at different time points.

Conclusion
In this prospective trial, the prognostic value of the timing of CTC measurement for breast cancer in the Chinese population was demonstrated. CTCs change dynamically during treatment, which can affect prognosis. Surgery led to an increase in the number and prevalence of CTCs on the first day after surgery and did not return to the preoperative level until 14 d after surgery. The CTC prevalence at the baseline and end-point follow-up visits was related to PFS and OS, while the CTCs detected before chemotherapy were only related to PFS. The CTCs detected at the last adjuvant chemotherapy were more correlated with prognosis than those before adjuvant chemotherapy in the analysis of paired samples. Therefore, CTCs at baseline or in the late phase of treatment are preferable for prognosis.

Materials and methods
Patients and study design. Between 2015 and 2016, 164 patients with breast disease who were diagnosed or treated at Jinling Hospital were included in our study. Peripheral blood was collected before surgery and recorded as a baseline then when available before each treatment until the end of the adjuvant chemotherapy. A discard tube was used to collect the first 1 mL of bloods to avoid epithelial skin cells contamination. Progression or non-progression was evaluated according to clinical practice based on clinical and radiological evaluations. Peripheral blood specimens from 19 healthy donors were used as negative controls.
Detection of CTCs in breast disease patients and healthy donors. Eight millilitres of blood was drawn into EDTAVacutainer tubes (BD, NJ, USA). The blood samples were stored at 4 °C and processed within two h after collection. Seven point five millilitres of blood was incubated with IMNs for 5 min. After enrichment, the subsequent process is the same as the process described in "Capture of spiked tumour cells in mock samples" section.
Statistical analysis. Statistical analyses were performed using IBM SPSS Statistics (version 24.0, IBM, Armonk, NY, USA). Pearson's chi-square test was used to compare the categorical variables. Pearson's chisquared test for trend was used to compare the ordinal data. The Mann-Whitney U test was used to compare continuous variables. Kaplan-Meier estimator and log-rank tests were used to compare survival rates between groups. Cox proportional hazards regression was used to determine the univariate and multivariate hazard ratios (HRs) of the potential predictors of PFS and OS.