Abstract
Purpose: Recently several reverse transcription-PCR techniques have been proven to be useful for the detection of circulating micrometastases. However, this way intact cell clusters that were found in animal experiments of prognostic value could not be detected. In this study, evaluation and modification of a commercial, cytokeratin-based, immunomagnetic cell separation method was performed for the detection of intact cell clusters in colorectal carcinoma patients.
Experimental Design: Thirty-two colon cancer patients (6 were in Dukes stage B, 13 in stage C, and 13 in stage D) and 20 healthy donor samples were evaluated. Immunomagnetic cell separation was performed from the buffy coat of peripheral blood samples (20 ml) using the Carcinoma Cell Enrichment Kit (Miltenyi Biotec, Bergisch Gladbach, Germany), avoiding any filtering steps. The enriched cell fraction was cytocentrifuged and immunocytochemically labeled using a pancytokeratin antibody (MNF116; Dako).
Results: Of 20 healthy samples, 2 contained one cytokeratin-positive cell. Of 32 single samples from malignant cases, 24 showed cytokeratin-positive cells. Tumor cell clusters, mixed-cell doublets (one cytokeratin-positive and -negative cell), and mixed-cell clusters were detected in 22 of 24 patients. In six cases, cytokeratin-positive dendritic-like cells were detected. Follow-up data indicate that chemotherapy cannot destroy all of the circulating tumor cell clusters.
Conclusions: Using the methods presented, we could detect circulating colon cancer cells and cell clusters in colon carcinoma patients. Similar cellular structures were described previously only in rats. Present data prove that such structures are present in human colorectal cancer, too.
INTRODUCTION
Detection of circulating tumor cells by immunocytochemistry in the bone marrow and in the sentinel lymph nodes is an independent prognostic marker for colorectal cancer, proved in several studies (1, 2, 3) . The detection of circulating tumor cells in the peripheral blood would offer several advantages over the bone marrow analysis. For example, the investigation could be performed more frequently.
With the RT-PCR2 method, several organ- or tissue- specific markers have been evaluated for micrometastases detection. The most frequent markers and tumor types are as follows: prostate-specific antigen for prostate cancer (4) , CK-19, CK-20, K-ras, MUC, and guanylyl cyclase for the colon (5 , 6) and tyrosinase for the melanoma (7) . The proven detection rate of these methods are about one tumor cell/ml blood. The most severe disadvantage of the RT-PCR methods is the loss of the morphological information and the false-positive results found in healthy blood donors (8) .
For morphologically intact circulating tumor cell detection, the microscopic evaluation of cytochemically stained peripheral blood smears was applied first (9) ; later, the microscopic evaluation of immunocytochemically labeled cells was preferred (3) . Another method of automated recognition of circulating tumor cells is the application of flow cytometric analysis after fluorescent immunocytochemical labeling (10 , 11) .
Recently a new immunomagnetic cell-separation technique was developed and used for the isolation of viable hematoprogenitor cells and specific immune cells (12 , 13) . There are two major types of the immunomagnetic sorting systems. The ones using 2–5-μm, large magnetic beads have several hundreds or thousands of antibodies on the bead surface (14) . The techniques using nanobeads use supermagnetic nanoparticles bound to antibodies (15 , 16) .
The first results with immunomagnetic cell enrichment and consecutive fluorescent immunocytochemical labeling and flow cytometrical detection show much promise. The detection limit of these methods is lower than that of the RT-PCR techniques (17 , 18) , one tumor cell/5 ml blood, but the characterization of cells for the presence of other markers is also possible (19) . However, the application of flow cytometry for detection of labeled cells does not support the morphological analysis of cells, either, and cell clusters and clumps could not be detected for technical reasons.
However, the presence of circulating tumor cell clusters and tumor-lymphocyte mixed clusters was found to be a more important prognostic factors in the metastatic process as compared with single circulating cells in animal experiments in several studies (20 , 21) . Data concerning the presence, type, and amount of circulating tumor cell clusters in human colorectal patients are not available.
The aim of the present study was to evaluate whether the immunomagnetic cell sorting with subsequent cytocentrifugation and immunocytochemical labeling produces clinically valuable data for detecting the presence of circulating tumor cell clusters in colorectal cancer patients.
MATERIALS AND METHODS
Cell Lines and Spiking Experiments.
HT29 colorectal cells were used as a positive control in immunocytochemical labeling and spiking experiments. The HT29 cells were maintained in DMEM containing 10% FCS. Cells for magnetic separation were suspended in PBS containing 5 mm EDTA and counted using a Buerker hemocytometer.
Collection of Blood Samples.
Colon cancer patient blood evaluations were done after informed consent. From each patient, 20 ml of blood were taken from an antecubital vein without using a tourniquet and evaluated immediately after sampling. Patient age, sex, date of diagnosis, therapeutic interventions, clinical status, and biopsy report were retrieved from the patients’ charts. The institutional review board approved the protocol. The blood samples were collected in patients in Dukes B and C stages before starting the chemotherapy and in some patients who were undergoing chemotherapy (patients 1, 6, and 12). In Dukes stage D, the patients were undergoing chemotherapy based on the Mayo protocol. Healthy blood samples from 20 individuals were used as negative controls.
Seeding of HT29 Cells in Blood/Recovery Experiments.
Dilution series were prepared for the determination of the sensitivity of the detection system. HT29 tumor cells were serially diluted 10-fold into 20 ml of normal blood taken from healthy volunteers. The final concentration of tumor cells was 106, 105, 104, 103, 102, 101, and 1 HT29 cell/1 ml blood, respectively. Samples under 102, 101, and 1 HT29 cell/ml cell concentration were prepared using a microscopic micromanipulator.
Magnetic Labeling.
Twenty ml of anticoagulated blood (1 ml of heparin) were centrifuged with 400 × g for 35 min. Approximately 5 ml of buffy coat containing ∼1.2 × 108 leukocytes were collected into 50-ml conical tubes. For magnetic labeling, the cells were first permeabilized using CellPerm Solution (40 ml) from the Carcinoma Cell Enrichment Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) and incubated for 5 min at room temperature. Fixation was done using 5 ml of CellFix Solution for 30 min at room temperature. The cells were then washed twice in dilution buffer. Two hundred μl of FcR Blocking Reagent (Miltenyi Biotec) was used for the prevention of unspecific binding. Two hundred μl of anti-cytokeratin (7/8) Micro Bead was used for magnetic labeling. The magnetic labeling was performed at room temperature for 45 min. Finally, the cells were washed once in cell stain solution and resuspended in 1 ml of dilution buffer for magnetic cell separation.
Magnetic Separation of Cytokeratin 7/8-Positive Tumor Cells.
For magnetic separation, the labeled cells were passed through MS+ (Miltenyi Biotec) separation columns that had been equilibrated with dilution buffer. The negative cells were washed off the column with 3 ml of dilution buffer. The retained cells were eluted from the column outside the magnetic field by pipetting 1 ml of dilution buffer onto the column. At this step, we changed the application protocol of the manufacturer (22) so that the filtering of the cellular suspension through a mesh with 30-μm diameter holes was omitted before the separation on the magnetic columns.
Immuncytochemical Labeling and Microscopic Analysis.
Cells were magnetically separated as described above, and the enriched cell fraction was spun down on slides using a cytocentrifuge. Slides were air-dried overnight at room temperature. The preparation was fixed with acetone (−18C°) for 5 min. The inhibition of endogenous peroxidase activity was performed with H2O2-methanol solution for 30 min (37C°). The slides were washed three times in distilled water. The blocking of nonspecific binding was done using PBS containing 1% BSA for 20 min at room temperature. The cytospins were stained with the Dako MNF 116 pan-anticytokeratin antibody at 37C° for 90 min in a humid chamber. After washing five times with PBS, the secondary, biotin-labeled, antimouse antibody (Dako Kit) was applied at room temperature for 30 min. After washing five times, streptavidin-bound peroxidase was added for 25 min. The amino-ethyl-carbazol chromogen was supplied after the washing steps. A red product is formed by the reaction of peroxidase with the amino-ethyl-carbazol.
Counterstaining was done using hematoxylin for 30 s. Slides were coverslipped using glycerin + PBS (1:1).
The light microscopic analysis was done using a conventional light microscope with ×10 and ×40 magnification without knowledge of the patients’ clinical details. Only the cells with circular-form, uniformly cytoplasmically distributed staining were scored as cytokeratin-positive cells (Fig. 1)⇓ . Two independent investigators did the evaluation. The mean value of data given by the two observers was taken as the number of positive cells.
Circulating tumor cell doublet and circulating tumor cell mixed cluster. Cytokeratin labeling black; ×20.
Statistical Analysis.
The statistical analysis was performed using the Statistica program package (Statsoft, Inc., Tulsa, OK; V.4.3). Significance of difference was determined by the post hoc LSD test.
RESULTS
In the spiking experiments, the lowest sensitivity of the assay was one tumor cell/1 ml. In the microscopic analysis, we found five different types of cells or cell clumps, as follows: (a) cytokeratin-positive single-standing tumor cells; (b) cytokeratin-positive cell clumps containing at least two cells, consisting only of cytokeratin-positive cells; (c) cell doublets consisting of one cytokeratin-positive and one cytokeratin-negative cell (Fig. 1)⇓ ; (d) mixed cell clusters containing more than three cells, with at least one of them cytokeratin-positive (Fig. 1)⇓ ; and (e) dendritic-like cytokeratin-positive cells (Fig. 2)⇓ .
Cytokeratin-positive tumor cells, cytokeratin-negative cells, and cytokeratin-positive dendritic-like cells in interaction (×200). Cytokeratin labeling black; dendritic-like cells are shown in high magnification (×400).
From the negative controls, in two cases, we found one circular cytokeratin-positive cell; the other controls were all negative.
In 22 of 32 colon carcinoma cases, we found more than one cytokeratin-positive cell. However, the cells were not usually simple; rather, they were doublets or clusters. The cytokeratin-positive cells were attached to round-shaped cytokeratin-negative cells. They appeared either in doublets or in mixed clusters (Tables 1⇓ and 2⇓ ).
Circulating tumor cells and tumor cell clusters in the peripheral blood of colorectal cancer patientsa
Detailed data about the circulating tumor cells and tumor cell clusters in the peripheral blood of colorectal cancer patients
No significant difference was found in the mean values of the observed cell types between the groups of Dukes stages B, C, and D patients (Table 1)⇓ .
For some patients (5) , follow-up data are now available. These show that the circulating tumor cells are in the blood stream for quite a long time. There was one patient who was negative at repeated investigations, as well. In six patients, we found a very specific type of cell (Fig. 2)⇓ : cytokeratin-positive cells showing dendrite-like alterations on their surfaces. This cell type could be detected again during the repeated investigation of the patients mentioned above (Table 3)⇓ .
Follow-up examination for the presence of peripheral circulating tumor cells and cell clusters
In two patients, blood was analyzed before and after chemotherapy (5-fluorouracil and leucovorin). A decrease of cytokeratin-positive cells was found in one case, and all of the circulating tumor cells disappeared in the other case.
In one case, blood samples were obtained and evaluated before and after operation. Although there were no cytokeratin cells before the operation, we found cytokeratin-positive cells in the blood hours and months after the surgery (Table 4)⇓ .
Changes in circulating tumor cell number after surgical operation
DISCUSSION
The detection of circulating tumor cells is very important from both a theoretical and a practical point of view. With the on-going detection and evaluation of these cells, we can get insight into the metastatic process in vivo. Additional information can be expected about the relation between the micrometastatic cells and the host immune system. Furthermore, we can use the data obtained from the characterization of these cells for the development of new therapeutic regimens.
The applied immunomagnetic cell separation with immunocytochemical labeling includes several technical steps, which is rather time and manpower consuming. It takes >2 days from the sampling time of the blood until the light microscopic evaluation.
Presently, there is no method for the automation of the process. The number of reactions that can be performed in parallel by one person is limited to four to five samples/day. The light microscopic analysis can be more time consuming than the flow cytometric one; however, important new observations were achieved this way.
Our results are comparable with that of the recently published studies using immunomagnetic separation with flow cytometric evaluation (17 , 18) . Martin et al. (17) found circulating tumor cells in 50–70% of malignant cancer cases. In this study, no cytokeratin-positive cells were found in the blood of healthy donors. Racila et al. (18) showed positive results in 29 of 30 cases. However, the mean value of the cytokeratin-positive cells of healthy donors was 1.5 ± 1.8. Our results show that 2 of 20 healthy control samples had one cytokeratin-positive cell. As data are accumulating, a threshold value of cytokeratin-positive cells in healthy donors seems to be evident, but their role is not yet clarified.
We found not only circulating tumor cells in our microscopic analysis, but tumor cell complexes consisting of cytokeratin-positive or cytokeratin-positive and -negative cells as well. This is the first such observation in colorectal cancer patients. Liotta et al. (19) and Glaves (20) found similar results in animal experiments. In these studies, using rat models, mixed cellular complexes were found shortly after injections of tumor cells into the rats’ veins. The first generation of granulocyte-tumor cell complexes was soon replaced by tumor-lymphocyte cell complexes. In our study, we have not performed the characterization of the cytokeratin-negative cells, which are presumed to be lymphocytes, based on their morphology. However, in a later experiment, Glaves (20) showed that these lymphocytes could destroy the attached cytokeratin-positive cells.
We had another new reproducible finding of dendritic-like cells in the blood of patients with long-lasting metastatic-free periods. These cells have to be characterized further in fluorescent microscopic studies.
The observed results concerning the shedding of tumor cells in the blood by surgery are comparable with those obtained by other publications (23 , 24) . Our data show, however, that the number of these cells decreases after operation; but they remain in the blood for a longer period.
We also had a longer observation period for selected patients. The appearance of circulating tumor cells and tumor cell clusters in the blood seem to be a dynamic but reproducible process. This was influenced by tumor-cell release, immunoresponse, and chemotherapy.
The results obtained from the patients before and after adjuvant chemotherapy demonstrated that the number of cells is decreased, but not eliminated, by the therapy. The explanation of this observation, also shown by others (25) , could be that the majority of the circulating tumor cells are dormant cells.
The present data also show the necessity of developing automated microscopic techniques for the automation of the cytospins microscopic image analysis. Similar efforts already can be observed in the bone marrow evaluation for micrometastases (26) .
The developed technique can be enhanced further using fluorescent labels. This way, the cytokeratin-positive tumor cells can be characterized further for the presence of proliferation or differentiation but also for other antigens such as epithelial growth factor receptor and CD44 homing receptor (17) . The origin and type of the attached cytokeratin-positive cells could also be clarified.
The applied technique uses intracellular cytokeratin for the immunomagnetic separation and labeling. Other methods use surface markers such as the epithelial cell adhesion molecule. This antigen, called EPCAM, BER-EP 4, or 17-A1, is also being used as the target of a recently developed immunotherapy (27 , 28) .
The integration of the presented method into routine immuno- or chemotherapy could be suggested for the research of the metastatic process and the determination of an eventual correlation between circulating tumor cell clusters and metastases in humans.
Footnotes
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 To whom requests for reprints should be addressed, at II. Department of Medicine, Semmelweis University, Szentkirályi St. 46, 1088 Budapest, Hungary. Phone: 36-1-2660926-5596; Fax: 36-1-2660816; E-mail: mb{at}bel2.sote.hu
-
↵2 The abbreviation used is: RT-PCR, reverse transcription-PCR.
- Received January 18, 2001.
- Revision received September 7, 2001.
- Accepted September 21, 2001.
References
Volume 7, Issue 12
