CXCL12 loaded-dermal filler captures CXCR4 expressing melanoma circulating tumor cells

Development of distant metastasis relies on interactions between cancer and stromal cells. CXCL12, also known as stromal-derived factor 1α (SDF-1α), is a major chemokine constitutively secreted in bone marrow, lymph nodes, liver and lung, playing a critical role in the migration and seeding of neoplastic cells. CXCL12 activates the CXCR4 receptor that is overexpressed in several human cancer cells. Recent evidence reveals that tumors induce pre-metastatic niches in target organ producing tumor-derived factors. Pre-metastatic niches represent a tumor growth-favoring microenvironment in absence of cancer cells. A commercially available dermal filler, hyaluronic acid (HA) -based gel, loaded with CXCL12 (CLG) reproduced a “fake” pre-metastatic niche. In vitro, B16-hCXCR4-GFP, human cxcr4 expressing murine melanoma cells efficiently migrated toward CLG. In vivo, CLGs and empty gels (EGs) were subcutaneously injected into C57BL/6 mice and 5 days later B16-hCXCR4-GFP cells were intravenously inoculated. CLGs were able to recruit a significantly higher number of B16-hCXCR4-GFP cells as compared to EGs, with reduced lung metastasis in mice carrying CLG. CLG were infiltrated by higher number of CD45-positive leukocytes, mainly neutrophils CD11b+Ly6G+ cells, myeloid CD11b+Ly6G- and macrophages F4/80. CLG recovered cells recapitulated the features of B16-hCXCR4-GFP (epithelial, melanin rich, MELAN A/ S100/ c-Kit/CXCR4 pos; α-SMA neg). Thus a HA-based dermal filler loaded with CXCL12 can attract and trap CXCR4+tumor cells. The CLG trapped cells can be recovered and biologically characterized. As a corollary, a reduction in CXCR4 dependent lung metastasis was detected.


Introduction
Cancer metastases contribute to over 90% of cancer deaths. The metastatic process is a complex process that involves primary tumors, tumor microenvironment and distant organs 1,2 . Recent studies suggest that stromal cells control the effective organ colonization and growth of circulating tumor cells (CTCs) 1 . CXCL12, also known as stromal-derived factor 1α (SDF-1α), is a major chemokine, constitutively secreted in bone marrow, lymph nodes, liver and lung 3 ; it plays a key role in physiologic processes, such as lymphocyte homing, chemotactic migration, 4,5 and metastatic development 6 . CXCL12 binds the chemokine receptor, CXCR4, overexpressed in at least 23 different human cancers [7][8][9] and CXCR7, a recently identified CXCL12 receptor also overexpressed in tumors [10][11][12] . Fibroblasts remodel extracellular matrix (ECM) at the pre-metastatic niche, a tumor growthfavoring microenvironment developed in the absence of neoplastic cells [13][14][15] , secreting inflammatory cytokines and growth factors such as CXCL12, TGF-β, S100A4, and expressing fibronectin and matrix metalloproteinases (MMPs) 16 . Tumor-secreted factors regulate the expression of molecules such as S100A8, S100A9, LOX, fibronectin, MMP-9, MMP-2 17 that promote the recruitment of specific bone marrow-derived cells (VEGFR1+, CD11b+, CD34+), myeloid cells (CD11b+) as well as differentiated innate and adaptive immune cells 17,18 . In the ECM, CXCL12 interacts with heparin sulfates 19 and co-localizes with hyaluronic acid (HA) [20][21][22] ; hematopoietic progenitor cells that migrate on HA toward a gradient of CXCL12 acquire spread and polarization expressing CD44 receptor mainly at the leading edge 22,23 . Hyaluronic acid (HA, or hyaluronan) is a glycosaminoglycan which consists of regular repeating non-sulfated disaccharide units of glucuronic acid and N-acetyl glucosamine that exhibits no tissue or species specificity 24 . HA is a principal component of the extracellular matrix, interacts with specific cell surface receptors (CD44, RHAMM and ICAM-1) and contributes to cell-cell adhesion, migration, proliferation and differentiation 25,26 . Injectable hyaluronic acid (HA) gels such as Intense Belotero® dermal filler are used in aesthetic medicine and thought to be applicable as injectable drug delivery systems 26 . CXCL12, cationic chemokine, binds to glycosaminoglycans (GAGs) through ionic interactions between basic amino acid residues and acidic groups along the disaccharide backbone 27 . Since CXCL12 is highly unstable and rapidly inactivated by CD26, several drug delivery systems have been developed to improve its in vivo residence time 28 . Polymeric hydrogels can hide and protect chemokines from enzymatic cleavage and glycosaminoglycans (GAGs) immobilize and enhance chemokines local concentrations, promoting oligomerization and improving of presentation to the receptors 29,30 . HA, abundant in the bone marrow where CXCL12 is expressed, serves as an anchoring molecule for breast metastatic cells homing through the CD44 receptor 22,31 Taking advantage of CXCR4 expression on metastatic, aggressive cells we speculate that an artificial tool loaded with CXCL12 might establish a condition similar to pre-metastatic niche that with recruitment of neutrophils and bone marrow derived cells will attract B16-hCXCR4 cells diverting them from secondary metastatic sites. Herein a commercially available dermal filler, HA-based gel, loaded with CXCL12 reproduced a "fake" pre-metastatic niche. The efficacy of this tool was evaluated in (i) recruiting circulating cells to be biologically characterized; (ii) trapping tumor cells and consequently reducing metastases.

Gel preparation
The commercially available HA based gels (Belotero Intense®) were purchased by Merz Pharma. HA fillers were commercially packaged in sterile ready-to-go syringes and appear as a clear, colorless, low-viscosity gels. The recombinant human/ CXCL12/SDF-1 alpha (R&D System) (300 ng/ml) in PBS/ 0.5% BSA was dropped onto the sterile gel, gently mixed by using a micropipette without introducing bubbles and immediately used. For in vivo experiments, CXCL12 (5 µg/ml) in 100 µl was dropped onto the sterile gel (1 ml) and immediately injected in the subcutaneous flank of C57BL/6 mice.
In vivo metastases assay CXCL12 (5 µg/ml)-loaded gels (CLGs) and empty gels (EGs) were injected in the subcutaneous flank of 18 (9/ group) 6-8-week-old female C57BL/6 mice (Harlan). Five days later 5 x 10 5 B16hCXCR4-GFP cells were inoculated into the tail vein. The mice were divided into three groups: (I) control mice (no gels) (6/group), (II) mice inoculated with empty gels (EG) (9/group), (III) mice inoculated with CXCL12-loaded gels (CLG) (9/group). Mice were sacrificed 22 days post-cell injection; lungs and gels were analyzed. Gels were digested with hyaluronidase (HAase, Bioindustria) (100 Units/ml) in 4 Units/ml in PBS. Lungs were fixed in 10% buffered formalin, paraffinembedded and hematoxylin/eosin stained. Microscopic analysis was conducted with a computer-assisted image measurement program by a microscope (BX51 microscope and DP-1 microscope digital camera; Olympus Japan). On days 10 and 22 post cells injection mice were anesthetized with pentobarbital (70 mg/kg) and subjected to retro-orbital blood sampling. Blood was collected using 1 ml syringes with 10 μl fresh EDTA (Spinreact) and processed for detection of the circulating tumor cells (CTCs). The Istituto Nazionale Tumori, IRCCS, Fondazione Giovanni Pascale Independent Ethical Committee approved the study and experiments were performed in accordance with the Institutional Animal Care and Use Committee guidelines.

Flow cytometer analyses
Tumoral cells recovered from gels were stained with a PE-hCXCR4 and APC anti-mouse CD45 Ab (BD Biosciences) and analyzed with a FacsDiva software 8.18 (BD Bioscience). The cells were stained with a viability dye 7-AAD to identify living and dead cells. The percentage of CXCR4 positive/ GFP positive cells was evaluated on live cells, 7-AAD negative / CD45 negative cells gate. Single stained samples were performed for compensation controls, and isotype control was used to determine the level of non-specific binding. Red blood cells in 100 μl murine blood cells were lysed and remaining cells were stained with human CXCR4 antibody (clone 11G5), PE-hCXCR4, and APC anti-mouse CD45 Ab (clone 30-F11, BD Biosciences). Also, EG/CLG hydrogels were recovered and digested with hyaluronidase. Cells were stained using 7-AAD (Thermo-Fisher), APC anti-mouse CD45 Ab (BD Biosciences), PE-Cy7 anti-mouse Ly-6G (clone1A8 BD Biosciences) and AlexaFluor 488 anti-mouse CD11b (clone M1/70, Elabscience). Lymphocytes were characterized as CD45 bright/ SSC low on base of cell scattering characteristics and CD45 expression.

Cell viability analysis
B16-CXCR4-GFP cells were allowed to migrate onto gels at 3, 7, 14, 21 and 28 days. Gels were digested with hyaluronidase as above described, cells recovered and dissociated with 0.05% trypsin/EDTA. Cell suspension was incubated with 10 µL of trypan blue solution and live cells counted on hemocytometer.

Real time PCR
Total RNA from murine blood (100 ul) was extracted using QIAamp RNA Blood Mini Kit (Qiagen), according to the manufacturer's instructions. DNase-treated RNA (100 ng) was reverse transcribed by Superscript II RNase H-reverse transcriptase according to the manufacturer's instructions (Invitrogen-Life Technologies, Carlsbad, CA, USA).

CXCL12/SDF-1α plasma concentration
Blood samples were obtained from retroorbital sinus of anesthetized mice in EDTA and stored at −70°C until assayed. Mouse SDF-1α was evaluated with CXCL12α Sandwich-ELISA kit (e-Bioscience).

Statistical analysis
Student's t-tests were used to compare the means of two groups when the assumptions of normality assessed with the Shapiro-Wilks test are met. For independent (unpaired) groups which are non-normally distributed, the non-parametric Mann-Whitney U test was used. The non-parametric Kruskal-Wallis test, used in the in vivo experiments, evaluated the significance of the differences of the mean ranks, owing to a lack of compatibility to the normal distribution. Per-comparison two-sided p values less than 0.05 were considered statistically significant. The values given are means ± standard deviation (SPSS statistics).

CXCL12 released from CLG induces long term chemotaxis
To investigate the efficacy of CLG-trapped CXCL12 to induce chemotaxis, the B16-hCXCR4-GFP migration toward CLG or EG was evaluated at longer time points. As shown in Fig. 1C B16-hCXCR4-GFP cells specifically migrated toward CLG as compared to EG for 14 days. At 21 and 28 days there were no detectable differences in migration toward CLG or EG suggesting CXCL12 exhaustion. Thus CXCL12 embedded into the gel is able to attract CXCR4 expressing cells. These data were confirmed by the slow in vitro CXCL12 release from CLG [5.43% of the total loaded cytokine released after 7 day incubation (up to 100 h) (data not shown)].

B16-hCXCR4-GFP cells trapped in EG or CLG gels progressively die
To evaluate the viability and the growth capability of B16-hCXCR4-GFP trapped cells, the gels were digested and cells recovered. In Fig. 2, trypan-blue exclusion test demonstrated that recovered cells were viable at three days. In CLG, on day 14 approximately 30% of live cells and 70% of dead cells were detected while in EG the percentage of live cells was 14% and dead cells was 86%. After four weeks the majority of cells were dead in all gels.

CLG in vivo captured Circulating Tumor Cells (CTCs)
To investigate the efficacy of CLG in attracting CXCR4 positive circulating tumor cells, the syngeneic model of melanoma lung metastasis was employed. CLG [CXCL12 (5 µg/ml) in 100 µl] was subcutaneously injected into the right flank of C57BL/6 female mice. Five days later B16-hCXCR4-GFP cells (5 x 10 5 ) 39,40 were intravenously injected into the mice tail vein. Twenty-two days later mice were euthanized. The gels were recovered, digested and cancer cells characterized as CD45 neg /hCXCR4 plus/ GFP plus cells. The number of CD45neg /hCXCR4plus /GFPplus cells recovered from CLGs was significantly higher than the number of cells isolated from EG (120 ± 36.7 in CLG vs16 ± 6.3 in EG) (Fig. 3, upper  panel). In Fig. 3

In vitro characterization of CLG/EG recovered cells
Cells recovered from EG and CLG are shown in Fig. 6A at 3-7 days of cell culture. Cells isolated from CLG displayed spherical shape (Fig. 6A, Panel B), stained positive for c-Kit 43 , negative for Alpha-Smooth Muscle Actin (α-SMA) (Fig. 6B), conversely EG recovered cells shaped as fibroblasts (Fig. 6A, Panel A) and stained negative for c-Kit, positive for α-SMA (Fig. 6B). In addition, c-kit and α-SMA expression was evaluated on CLG-and EG-isolated cells through qRT-PCR confirming high expression of α-SMA and low c-kit in EG, and high expression of c-kit and low α-SMA in CLG (Supplemental Fig. 4A). CLG isolated cells appeared epithelial, melanin rich (Fig. 6A, Panel B), express the Melan A 44 and S100 45 suggesting a melanoma cell line (Fig. 6C) while EG isolated cells appeared negative for Melan A and S100 (Fig. 6C). In Fig. 5D the CLG recovered cells expressed human CXCR4 as compared to the injected cells B16-hCXCR4 cells and to PES43, human melanoma cells. As shown in Supplemental Fig. 4B, CLG recovered cells expressed GFP RNA level as compared with B16-hCXCR4-GFP cells. The CLG isolated cells actively migrated toward CXCL12 and migration is inhibited by AMD3100 (Supplemental Fig.  4C). The ability to isolate tumor cells while subtracting them to conventional sites of metastasis allows isolation and characterization of circulating tumoral cells.

CTCs in mice carrying CLG/EG
The possible impact of CLG and EG on the number of circulating tumor cells (CTCs) was evaluated at 10 and at 22 days post cell injection. Briefly 100 µl of retrorbital blood was obtained and CTCs identified as CD45 neg/ hCXCR4 pos/GFP pos. The number of CTCs isolated in mice CLG carrying was significantly higher compared to CTCs retrieved in mice EG carrying. As shown in Fig. 7 (upper-left panel) higher number of CTCs was detected in the blood at 10 days from mice CLG (0.09 ± 0.02%) versus EG (0.03 ± 0.02%). This difference, although lower, is still significant at 22 days (CLG 0.04 ± 0.01% versus EG 0.02 ± 0.01%) (Fig. 7, lower-left panel). In concordance, GFP mRNA was higher in CTCs isolated from CLG compared to EG carrying mice at 10 and 22 days (respectively, CLG 0.02 ± 0.009% versus EG 0.01 ± 0.003% and CLG 9.7E-04 ± 0.002% versus EG 2.7E-05 ± 2.3E-05%) (Fig. 7, right  panel).

CXCL12 increased in mice carrying CLG/EG
To investigate on possible mechanisms responsible for increased numbers of CTCs in CLG, circulating CXCL12 was evaluated in peripheral blood collected from CLG/EG bearing C57Bl/6 mice. As shown in Fig. 8, both CLG and EG carrying mice displayed higher murine CXCL12 plasma level at 24 h after gel inoculation consistent with inflammatory reaction induced by gel implantation. Circulating CXCL12 at 24 h was 11.3 ± 0.2 ng/ml in mice carryng EG and 10.2 ± 0.4 ng/ml in mice carryng CLG compared to 1.1 ± 0.3 in gel-not injected mice (p < 0.01). CXCL12 blood concentration was not statistically different between mice EG or CLG-inoculated at 24 h post gel implantation. In both groups, reduction of blood CXCL12 was observed at 72 h and 21 days post B16-hCXCR4-GFP cells inoculation (at 72 h CXCL12 was 10.6 ± 0.3 ng/ml in EG -mice and 7.7 ± 1.0 ng/ml in CLG-mice while at 21days CXCL12 was 7.1 ± 1.8 ng/ml and 4.1 ± 2.0 ng/ml respectively, in EG and CLG-mice) suggesting that plasma CXCL12 was not influencing the propensity of tumor cell homing to the lungs/CLG.

Discussion
With the primary intent to collect circulating tumor cells expressing CXCR4, a new device composed of a commercially available dermal filler, hyaluronic acid based gel (Belotero Intense®), loaded with CXCL12 was realized. The CXCL12-loaded gel (CLG) attracted circulating CXCR4 positive melanoma cells diverting them from secondary sites. The cells isolated from CLG mimicked the original injected cells and could be expanded in vitro providing a valid source of tumor cells for further characterization. Moreover, the CLG, capturing CTCs and diverting them from conventional metastatic sites, induced a net decrease in lung metastases. Interestingly, an increased number of circulating tumor cells (CTCs) were observed in CLG bearing mice. Mechanical devices impairing metastases were previously described 39,42,46-48 . To mimic a functional and reproducible bone, silk scaffolds coupled with bone morphogenetic protein-2 (BMP-2) seeded with bone marrow stromal cells (BMSC) were developed 46 . Similarly, a chimeric bone construct was realized with biodegradable tubular composite scaffolds seeded with human mesenchymal progenitor cells and loaded with osteogenic protein-1 47 . Although these examples provided "proof of principle" for a bioengineered humanized model of bone metastasis, the clinical translation is complex. A sort of pre-metastatic niche was developed with a 3D-scaffold embedded with exosomes (M-Trap) capturing human ovarian cancer cells disseminated in the peritoneal cavity 39 . This model demonstrated that an added matrix could segregate metastatic tumor cells in a defined environment such as peritoneal cavity. Micro particles (MPs) containing carcinoma-associated fibroblasts (CAFs), which continuously deposit a pro-adhesive matrix at the surface, selectively bound ovarian neoplastic cells within the peritoneal compartment of the MPs 48 . In a xenograft model of breast cancer an implanted biomaterial (PLG) scaffold recruited circulating metastatic breast cancer cells (MDA-MB-231BR) 42,49 and human breast cancer brain metastasis derived three-dimensional CAFs aggregates expressed significantly higher levels of CXCL12 and CXCL16 than CAFs aggregates generated from primary breast tumors or normal breast, suggesting that chemokine modulation is crucial in regulating metastasis 1 .
The herein described tool CLG represents an innovative and simple solution to trap circulating tumor cells and, secondarily, to reduce lung metastasis. The main hypothesis herein was that interfering in the process that drives tumor cells from tumor to secondary organs, we may reduce the number of effective cells able to develop metastasis. Metastasis development is a very inefficient process with 0.01% or less of circulating tumor cells able to develop secondary tumors 50 . The CXCL12 gel will subtract CXCR4 overexpressing cells and trap them in a pseudo-niche, where the cells will eventually die within 2 weeks. In doing so it might reduce the critical number of metastasis initiating cells. We may also hypothesize that 'sensing' of CXCL12 inside the gel increases the "signal noise" attracting CXCR4 positive cells and affecting seeding in secondary organs as demonstrated by the largest number of circulating cells detected in mice carrying CLG. Since the hyaluronic acid of the hydrogel exerts itself an effect of pseudo-niche, we expect that the cells will be trapped also there. This is consistent with the reduction of metastases in EG compared to the number of metastasis developed in mice in absence of gel (5.9 versus 8.2 metastatic lesions per section). Interestingly, recent evidence demonstrated that local delivery of lauroylgemcitabine lipid nanocapsule based hydrogel (GemC12-LNC) in the tumor resection cavity of glioblastoma could prevent local recurrence 51 . The CXCL12 loaded hydrogels Fig. 6 In vitro characterization of CLG/EG recovered cells. A Cells recovered from EG and CLG gel were seeded on plates immediately after collection. Microphotographs from EG (panel A) and CLG (panel B) at 3 and 7days (Axiovert10 Microscopy-Zeiss). Scale bar is 20 µm. CLG cells express B c-Kit, C Melan A, S100 and D CXCR4 while EG cells express B α-SMA but do not express c-Kit, Melan A and S100. Cells (1 x 10 4 ) were seeded on glass coverslips, fixed with 4% paraformaldehyde (15 min, 4°C) and stained with rabbit anti-c-Kit primary antibody, mouse anti-αSMA primary antibody, mouse anti-S100 primary antibody mouse, mouse anti-Melan A primary antibody and mouse anti-human CXCR4 primary antibody (clone 12G5 R&D). DAPI was used to stain the cell nucleus (Carlo Zeiss, Axio Scope.A1). CXCR4 expression was compared to PES43, human melanoma cell line-CXCR4 expressing, FB-1, human anaplastic thyroid cancer cell line-low CXCR4 expressing and murine B16-hCXCR4-GFP cells evaluated in the in vitro/in vivo assay. DAPI was used to stain the cell nucleae. Magnification 400X. Scale bar is 25 µm in adjuvant setting may trap the undetectable, occult and harboring micrometastases cells reducing future recurrences. This is based on the capability of the matrix, dermal filler, plus the properties of CXCL12 to develop a suitable pre-metastatic niche with attractive capability for CXCR4 expressing circulating tumor cells. VEGFR1 + hematopoietic precursor cells, along with fibronectin and associated stromal cells modify the local microenvironment and regulate the homing and retention of hematopoietic precursor cells as well as tumor cells 17,52 . Liu et al. developed a CXCL12 biomimetic tumoral niche with a thin and soft polyelectrolyte film. CXCL12 presentation was spatially controlled at the ventral side of breast cancer cells inducing lamellipodia and filopodia mediated by CXCR4. CD44 acted in concert to drive a specific matrixbound CXCL12-induced cell response associated with ERK signaling 53 . Herein CLG recruited higher number of cells compared to EG; although at lesser extent, EG was also able to recruit circulating cancer and mesenchymal cells suggesting that the hyaluronate matrix exerts attraction 21,22,54,55 . Ko Cy et al. developed in vivo model to investigate inflammation-mediated cancer metastasis through biomaterial microspheres. Interestingly, metastatic cancer cells B16F10 injected into the peritoneal cavity migrated into subcutaneous microsphere 56 . The immune cells recruitment to implanted biomaterials may be different in tumor bearing animals versus healthy animals 42 . In our manuscript the CXCL12 loaded gel was able to recruit and trap B16-hCXCR4-GFP cells as previously shown 57 . Quantitative and qualitative differences were detected within the CLGs and EGs inflammatory infiltrate. CLG showed higher number of CD45-positive leukocytes mainly neutrophils CD11b+Ly6G+cells, myeloid CD11b+Ly6G-cells and macrophages F4/80 as compared to EG 58 . CXCL12 insists also on CXCR7, which is phylogenetically closely related to chemokine receptors, binds CXCL12 with a higher affinity than CXCR4, but fails to couple with G proteins to induce typical chemokine receptor-mediated cellular responses 59 . Although in our gels CXCR7 antagonism is not able to interfere with migration of CXCR4 expressing cells toward CLG/EGs, it was recently demonstrated that CXCR7 is expressed on CD14 + CD16 + mature monocytes and a small molecule CXCR7 antagonist (CCX771) can prevent CD14 + CD16 + monocyte transmigration into the central nervous system from uninfected and HIV infected individuals 60 . Intriguingly, the number of CTCs isolated in mice carrying CLG was significantly higher compared to CTCs retrieved in mice carrying EG. To better define this phenomenon blood CXCL12 was evaluated but CXCL12 clearly increased in mice gels-inoculated, either empty or CXCL12-loaded consistent with inflammation gels induced. In both groups, EG/CLG-inoculated, reduction of blood CXCL12 was observed at 72 h and 21 days post B16-hCXCR4-GFP cells inoculation suggesting that plasma CXCL12 was not influencing the propensity of tumor cell homing to the lungs/CLG. Bertolini et al. defined lung cancer initiating cells (CICs) as CD133 +CXCR4+cells with metastatic potential inhibited by CXCR4 antagonism 61 . In fact CTC enumeration and CXCR4 expression are promising prognostic biomarkers for CTCs in Extensive-Disease Small Cell Lung Cancer (ED-SCLC) at baseline and post-treatment 62 . For patients at risk of recurrence, CLGs would allow characterization of CTCs and praecox recognition of metastasis disease. CTCs isolation is still a high unmet need due to the limited number of cells generally retrieved (1 in 10 9 red blood cells) 63 . Injecting 1x10 6 of mouse breast cancer GFP-Luc-4T1 in tail vein of BALB/c syngeneic mice resulted in detection of 16 Luc-CTCs comparing three CTC detection methods 64 . The patient subcutaneous devices implantation is currently evaluated in clinical trials. Agarose devices containing murine renal cancer cells (RENCA cells) were safely inoculated in patients abdominal cavity. RENCA cells, encapsulated in the agarose gel, selected a stem cell-like subpopulation which drove colony formation in the macrobeads and produced diffusible substances that markedly inhibited in vitro and in vivo proliferation of epithelial-derived tumor cells outside the macrobeads 65 (NCT02046174). Phase II studies in patients with colorectal, pancreatic or prostate cancers are in progress (NCT01174368, NCT01053013).
In conclusion, our study shows that a commercially available dermal filler loaded with CXCL12 is able to capture and divert neoplastic CXCR4 expressing cells. Behind the suggestion of a diminution of the metastatic efficiency, it is possible to speculate that the device will allow identification and characterization of potential metastatic cells. The praecox identification and definition of metastatic cells may enable efficient interventional strategies.