In Early Breast Cancer, the Ratios of Neutrophils, Platelets and Monocytes to Lymphocytes Significantly Correlate with the Presence of Subsets of Circulating Tumor Cells but Not with Disseminated Tumor Cells

Simple Summary Circulating tumor cells (CTCs) are potential precursors of metastasis and while travelling through the peripheral blood, they crosstalk with different blood cells before a few of them manage to settle down as disseminated tumor cells (DTCs). Little is known about the correlation of blood cells with CTCs/DTCs in early breast cancer (BC). We retrospectively recorded clinical data, results for CTCs, DTCs and blood cell counts from 171 early staged diagnosed BC patients and demonstrated that the presence of epithelial CTCs was related to reduced lymphocyte and monocyte counts, to elevated neutrophil to lymphocyte and platelet to lymphocyte ratios while CTCs in epithelial mesenchymal transition associated with a reduced monocyte to lymphocyte ratio. No significant correlations were found for DTCs, however, DTC-positive patients, harboring a lower platelet to lymphocyte ratio, had a significant shorter overall survival. We confirm that pro-inflammatory markers in blood are closely related to the presence of CTC subtypes, the precursors of metastasis. Abstract Circulating tumor cells (CTCs) crosstalk with different blood cells before a few of them settle down as disseminated tumor cells (DTCs). We evaluated the correlation between CTC subtypes, DTCs and the neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR) and monocyte to lymphocyte ratio (MLR) for better prognostication of 171 early staged diagnosed breast cancer (BC) patients. —Clinical data and blood values before treatment were retrospectively recorded, representing the 75% percentile, resulting in 3.13 for NLR, 222.3 for PLR and 0.39 for MLR, respectively. DTCs were analyzed by immunocytochemistry using the pan-cytokeratin antibodyA45-B/B3. CTCs were determined applying the AdnaTests BreastCancerDetect and EMT (Epithelial Mesenchymal Transition) Detect. —Reduced lymphocyte (p = 0.007) and monocyte counts (p = 0.012), an elevated NLR (p = 0.003) and PLR (p = 0.001) significantly correlated with the presence of epithelial CTCs while a reduced MLR was related to EMT-CTCs (p = 0.045). PLR (p = 0.029) and MLR (p = 0.041) significantly related to lymph node involvement and monocyte counts significantly correlated with OS (p = 0.034). No correlations were found for NLR, PLR and MLR with DTCs, however, DTC-positive patients, harboring a lower PLR, had a significant shorter OS (p = 0.043). —Pro-inflammatory markers are closely related to different CTC subsets. This knowledge might improve risk prognostication of these patients.


Introduction
Breast cancer (BC) is the leading cause of cancer associated mortality in women worldwide. Over the years, the prognosis of these patients has steadily improved and

Clinical Characteristics of Patients
The clinical characteristics of all chemo-naive, early staged diagnosed, non-metastatic BC patients (n = 171) at the time of primary diagnosis are shown in Table 1. The median age of the patients was 61 years (range 31-83 years). The majority of the patients were post-menopausal (71.9%), had T1 (59.6%) and T2 (37.4%) tumors, 59.1% were node-negative (38.6% = N1) and most of the patients presented with a poor or moderately (GII and III) differentiated tumor (93.6%). Expression of the estrogen (ER) and progesterone receptor (PR) was observed in 78.4% and 69.6% of the tumors and human epidermal growth factor receptor (HER)2 was overexpressed in 15.8% of cases, respectively. When patients were stratified according to their histological subtypes, 69% were ER-and or PR-positive and HER2-negative, 15.2% were triple-negative, 11.1% triple-positive and 4.7% showed HER2 overexpression (ER and PR-negative), respectively. 58/170 (34.1%) of the patients were DTC-positive and 44/155 (28%) of the patients were CTC-positive, respectively.

Establishment of Cut-Off Values
NLR, PLR and MLR were calculated from peripheral blood cell counts (Table 2) and their cut-off levels were determined by the 75% percentile resulting in the values 3.13 for NLR, 222.3 for PLR and 0.39 for MLR, respectively ( Figure 1). Table 2. Cut-off values for peripheral blood cell counts and their ratios. Cut-off levels for blood values of all patients were determined by the 75% percentile (gray column). MLR: monocyte-to-lymphocyte ratio; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio.

Correlation of NLR, PLR and MLR with Clinical Characteristics
When patients characteristics as well as DTCs and CTCs were correlated with NLR, MLR and PLR, only the lymph node status was significantly related to PLR (p = 0.029) and MLR (p = 0.041) while CTCs significantly related to NLR (p = 0.025) ( Table 3). In more detail, patients with a high NLR (≥3.13) level had significantly more often a positive CTC-status. Patients with a high PLR (≥222.3) or high MLR (≥0.39) level, had significantly more often a pN2-status.

Prognostic Role of Blood Cell Counts, Ratios, CTCs and DTCs
Using Spearman-Rho Test (Table 4), no significant correlations with regard to PFS and OS were found for neutrophils, lymphocytes, platelets, NLR, PLR and MLR, respectively. Nevertheless, high monocyte counts significantly related to a reduced OS (p = 0.034). Whereas the presence of DTCs did not correlate with PFS (p = 0.254) or OS (p = 0.185), DTCpositive patients, harboring a lower PLR, had a significant shorter OS (p = 0.043) ( Figure 3). The presence of CTCs was significantly correlated with a shorter PFS (p = 0.038) and OS (p = 0.018) (Figure 4), however, neither eCTCs nor EMT-CTCs alone were of prognostic significance. In patients still alive, a reduced PFS/OS significantly correlated with enhanced lymphocyte (p = 0.025/0.011) and monocyte counts (p = 0.039/p = 0.037) as well as a low PLR (p = 0.032/0.023) whereas an enhanced MLR showed a high correlation with a shorter PFS (p = 0.007) and OS (p = 0.021) in deceased patients (data not shown).    Figure S1).

Univariate and Multivariate Analysis
As apparent from Table 5, in univariate analysis, good prognostic factors with regard to PFS and OS were age >60 years (PFS: p = 0.046; OS: p = 0.027) and no lymph node involvement (PFS: p = 0.047; OS: p = 0.020). For multivariate analysis, patients who presented with a N0 status, OS was close to significance (p = 0.050). For all other variables including NLR, PLR, MLR, CTCs and DTCs, no significant relationships were found ( Table 5).

Discussion
Although early staged diagnosed BC is considered potentially curable and therapeutic options have substantially progressed over the last years, it is estimated that 20-30% of all BC cases will become metastatic, even after 10-20 years. In addition, BC prognosis depends on a variety of patient and tumor factors, however, markedly different clinical outcomes are seen among patients with similar prognostic factors [1]. Thus, research in early staged diagnosed BC currently aims to discover new biomarkers to better estimate patients' clinical outcome, identify patients who might have a higher risk for relapse and to get insights to adjust treatment.
In this regard, the presence and persistence of CTCs as well as DTCs were shown to be of prognostic significance [4,21] which is also confirmed in our study for CTCs with regard to PFS and OS while DTCs showed no prognostic significance. The latter finding goes in line with our previously published studies which explain these results with the intake of bisphosphonates in case of DTC-positivity which was shown to markedly improve the outcome of our patients [22][23][24][25].
However, while travelling from the primary site into secondary organs, tumor cells leave their original surrounding, loose contact to epithelial cells and have to adapt to cells in the blood stream where they are directly exposed to the immune system which should, in principle, try to eliminate them [26]. Although inflammatory cells and mediators in the tumor microenvironment are thought to play an important role in cancer progression, the interplay with CTCs and DTCs in early staged diagnosed BC has rarely been addressed [27].
The prognostic and predictive value of pro-inflammatory markers including ratios to lymphocytes has often been studied in BC patients receiving neo-adjuvant chemotherapy with mainly neutrophils as well as NLR in their focus [12,28,29]. Pre-treatment values of NLR and MLR in these patients were shown to be suitable biomarker for predicting treatment efficacy, pathological complete remission and survival [30][31][32][33]. Up to now, only a very few studies have focused on early staged diagnosed BC. In our study, we could not demonstrate any significances for neutrophils, lymphocytes, platelets, NLR, PLR and MLR with regard to PFS and OS, respectively. This is in contrast to the other published articles, however, it is difficult to compare these studies because different cut-off values, in general lower values, for all ratios were used [34]. In this context, a cut-off above 1.8 for NLR was shown to relate to a postoperative recurrence [35], above 1.97 to a significant worse prognosis [36] and above 2.65 to a shorter OS, repectively [37]. Referring to meta analyses including early staged diagnosed and locally advanced BC patients, the most frequently used cut-off for NLR was 3.0 (as median) which is similar to the value we used (3.13 as 75% percentile) [34]. Applying exactly the cut-off of 3.0, only NLR among several other immune-based scores tested, was significantly related to a reduced PFS in a cohort of 284 early staged diagnosed BC patients [20].
The prognostic potential of PLR and MLR was identified less frequently in early staged diagnosed BC. Published PLR scores of >185 and >190.9 (thus, below our applied score of 222.3 with no significant results obtained for PFS and OS), were related to a shorter OS in univariate analyses [37,38]. In contrast, high monocyte counts in our patient cohort significantly correlated with a reduced OS which goes in line with a large cohort study, which demonstrated that elevated pre-operative circulating absolute monocyte counts were related to a shorter OS [39]. For MLR, using a cut-off of 0.39, no significant relationships were found for PFS and OS which was confirmed in a comprehensive study in early staged diagnosed BC applying a cut-off of 0.34 [20].
For the interplay of immune cells with tumor cells, our retrospective analysis confirms that pro-inflammatory markers in blood, namely NLR, PLR and MLR, are closely related to the presence of different CTC subtypes. While eCTCs were closely related to reduced lymphocyte and monocyte counts and an elevated NLR and PLR, the presence of EMT-CTCs significantly correlated with a reduced MLR. In addition, DTC-positive patients with a documented lower PLR had a significant shorter OS. With regard to CTCs, only a few studies have addressed this topic in solid tumors including BC. Zheng et al., correlated pre-operative inflammation-based indexes, including a systemic immune-inflammation index (SII), NLR, PLR as well as a prognostic nutrition index (PNI), with CTC counts in 60 patients with gastric cancer. Selecting epithelial tumor cells by size using ISET (Isolation by SizE of Tumor Cells) with subsequent staining of the cells with ostom Y and methylene blue for Wright staining, they demonstrated that NLR and PLR were clinically meaningful to predict eCTC counts and eCTC detection ratios and the correlation suggested that improved scores of NLR and PLR were consistent with increased eCTC counts [40]. These data are, somehow, comparable with our finding and emphasize that the analysis of different CTC subtypes might help to improve the prognostication of patients. In a retrospective analysis of 516 metastatic BC patients, eCTC counts obtained with the CellSearch System significantly correlated with the number of monocytes and neutrophils and in a multivariate analysis, in combination with a MLR ≥0.34, eCTCs were predictors of a decreased OS [19]. Comparable with our study, Miklikova et al., evaluated NLR, MLR, PLR and SIL (systemic immune-inflammation index), referred to as complete blood count (CBC)-derived inflammation-based scores, in 284 early BC patients and correlated these findings with the EMT-CTC subtype. Using CD45 depletion (RossetteSep TM ), followed by reverse transcription of enriched mRNA to detect the EMT-related transcription factors TWIST1, SNAIL1, SLUG, and ZEB1 gene transcripts, a strong relationship between adverse outcome and elevated NLR and MLR in EMT-CTC-positive patients was observed. They concluded that their findings support the hypothesis that immune cells in the bloodstream can expand the metastatic potential of CTCs [20]. Although different methods were used to detect CTC subtypes in these studies, it seems as if NLR, PLR and MLR are closely related to the presence of CTC subtypes, even in early BC patients where CTCs, in general, are rare [21].
The interplay between CTCs/DTCs and the tumor microenvironment is complex and poorly understood. A variety of important infiltrating and circulating immune cells hinder or favor the dissemination of CTCs, including natural killer cells, T-cells, neutrophils, monocytes and macrophages as well as platelets [27]. In this regard, a variety of mechanisms have been hypothesized through which CTCs escape and survive. In this context, some of the CTCs are known to express CD47, a 'don t eat me signal' and PD-L1 has been detected on BC CTCs which prevents T-cell mediated destruction [41][42][43]. Furthermore, CTCs are frequently present inside circulating tumor micro-emboli or white blood cell clusters (WBCs) which protect them from being recognized [44]. Among all immune cells discussed to support CTCs, neutrophils seem to play a major role. Besides NETs that help CTCs to spread and adhere to distant sites [17], CTCs were shown to co-localize with neutrophils in the pre-metastatic vascular network, suggesting that neutrophils can retain cancer cells and facilitate their extravasation. A recently published study showed that WBCs-CTC clusters of BC patients, although quite rare, were related to neutrophils [16]. Interestingly, the transcriptomic profiles of CTCs associated with neutrophils were different from those of single CTCs with regard to cell cycle progression, thus, probably leading to metastasis. These data also support our study which documents a significant relationship between NLR and eCTCs. In addition, CTC clusters detected in advanced BC exhibited mesenchymal features and showed attached CD61-positive platelets that are known to induce EMT-like features in CTCs [45]. We did not find significant relationships between platelets and CTC subtypes, however, an enhanced PLR was closely related to the presence of eCTCs but not to EMT-CTCs which in turn were correlated with a reduced MLR. We can only speculate about the cause of the latter finding. In general, circulating monocytes can extravasate and differentiate into macrophages with pro-tumor and pro-metastatic functions. In this context, Zhang et al., showed that a subgroup of macrophages was able to phagocyte CTCs, incorporate their content and thus, express malignant features [46]. Furthermore, tumor associated macrophages were shown to facilitate cancer cell migration by activating EMT [47] which might partly explain our findings.
To the best of our knowledge, no data have been published with regard to the relationship between pro-inflammatory markers in blood and their relationship to DTCs that settle down in the BM. Our current study did not identify significant correlations between blood cell counts, NLR, PLR, MLR and DTCs, however, DTC-positive patients with a documented lower PLR had a significant shorter OS. For homing of DTCs in secondary organs, other immune cells or factors than those we evaluated, might be relevant. In this context, the chemokine receptor CXCR4 has been favored to help tumor cells to settle down in the BM [48] and we recently demonstrated that CXCR4/JUNB/CK-expressing DTCs were frequently detected in the BM of early staged diagnosed BC patients and seem to identify a subgroup of patients at higher risk for relapse [10].

Study Population
This retrospective study was conducted in the Department of Gynecology and Obstetrics at the University Hospital of Essen and included 171 early staged diagnosed, non-metastatic BC patients who presented with first diagnosis of BC between July 2006 and December 2012. The eligibility criteria were as follows: histologically proven BC, no severe uncontrolled co-morbidities or medical conditions, no further malignancies at present or in history, completion of adjuvant treatment according to guidelines including adjuvant chemotherapy (anthracyclines, 5-fluorouracil, taxanes, cyclophosphamide), anti-hormonal therapy in case of hormone responsive tumors (tamoxifen or an aromatase inhibitor), trastuzumab in case of HER2-positivity and radiotherapy. All DTC-positive patients were recommended an additional therapy with oral clodronate (2 × 520 mg per day for at least two years). Patients treated with neo-adjuvant chemotherapy were excluded.
All specimens were obtained after written informed consent from all subjects prior to inclusion in the study and collected using protocols approved by the clinical Ethic committee of the University Hospital Essen (05/2856). All methods were carried out in accordance with the approved guidelines.
From a total number of 623 early staged diagnosed BC patients, patients' selection for this study was based on the availability of differential blood cell counts at primary diagnosis, before any treatment. 171/623 BC patients fulfilled these criteria and for these patients, DTC as well as CTC analysis was performed at the same time point, before the start of any therapy.
After a median follow-up time of 96.3 months (range: 1.8 to 152 months and 95% CI was 93.0 to 99.5 month). The OS rate was 91.2% (156/171 patients) and relapses occurred in 5.84% (10/171 patients) of cases.

Selection and Detection of CTCs
CTCs were isolated from 4 × 5 mL blood drawn before the application of therapeutic substances and before surgery with an S-Monovette (Sarstedt AG & Co., Nümbrecht, Germany) and stored at 4 • C until further examination. The samples were processed immediately or latest four hours after blood withdrawal. Establishment and validation of this assay has been described in detail elsewhere [24]. Briefly, CTCs were selected by positive immune-magnetic selection targeting EpCAM and MUC1 using the AdnaTest BreastCancerSelect (QIAGEN, Hilden, Germany), followed by mRNA isolation from lysed, enriched cells and subsequent reverse transcription resulted in cDNA, which was the template for tumor cell detection and characterization by multiplex PCR. eCTCs were determined applying the AdnaTest BreastCancerDetect (QIAGEN, Hilden, Germany) for gene expression analysis of EpCAM (GA733-2), MUC-1 and HER2 (ERBB2). The test is considered positive if an amplicon of at least one tumor-associated transcript was detected over the threshold defined in the handbook. EMT-CTCs were detected using the AdnaTest EMT Detect (QIAGEN, Hilden, Germany) for the expression of PI3K, AKT and TWIST. The test is considered positive if an amplicon of at least one tumor-associated transcript was detected over the threshold defined in the handbook. We defined a patient CTC positive if one of the two tests were positive. Visualization of the PCR fragments was carried out with a 2100 Bio-Analyzer using the DNA 1000 LabChips (Agilent Technologies) and the Expert Software Package (version B.02.03.SI307) both Santa Clara, CA, USA.

Selection and Detection of DTCs
Between 10 and 20 mL BM were aspirated from the anterior iliac crests of all patients at the beginning of surgery of the primary tumor, before start of any therapy and processed within 24 h. BM tumor cell isolation and detection have been described elsewhere [22][23][24][25]. Briefly, BM cells were isolated from heparinized BM (5000 U/mL BM) by Ficoll-Hypaque density gradient centrifugation (density 1.077 g/mol; Pharmacia, Freiburg, Germany) at 400× g for 30 min. Slides were analyzed for DTCs by immunocytochemistry using the pan-cytokeratin antibody A45-B/B3. Microscopic evaluation of the slides was carried out using the ARIOL system (Applied Imaging). A patient was defined as DTC-positive in case one cytokeratin-positive cell was detected.

Statistical Analysis
Statistical analysis included descriptive reporting, U-test for group differences, Rho correlations for dependencies and Fisher's chi-square test for distributional differences. Rho correlations were used to assess the relationship between NLR, MLR, PLR and tumor cells as well as PFS and OS. We utilized cox regression analysis to generate a hazard ratio with confidence intervals for OS and PFS and a variety of markers and clinical parameters.

Limitations of the Study and Conclusions
The relationship between tumor cells and their environment is very complex, however, might be promising to develop new treatment strategies in the future. The strength of our study is reflected by significant relations between CTC subtypes and blood cell ratios which encourages to get deeper insights into the origin of circulating neutrophils, platelets, monocytes as well as T-and B-cells in the future. However, the study also has some limitations. Non-consecutive patients, undergoing adjuvant treatment were enrolled until about 10 years ago, and the patient population largely consisted of hormone receptor-positive, HER2-negative patients. Consequently, the results are applicable almost exclusively to this BC subgroup. Upcoming studies in this direction should include other BC subgroups, mostly receiving neoadjuvant chemotherapy nowadays as well as more sampling time points to draw more precise conclusions.
Among all markers evaluated up to now, neutrophils and NLR seem to be the most promising markers and could serve as a predictor of patients' survival to help with treatment decisions. Although therapeutic options are rarely available right now, interesting experimental studies demonstrated that blocking the interaction of tumor cells and platelets [49,50] or antibodies against NET-remodeled laminin, that prevent awakening of dormant cells, which might be a promising approach in the future [18].