A relevant immunomagnetic assay to detect and ... - Wiley Online Library

693

A Relevant Immunomagnetic Assay to Detect and Characterize Epithelial Cell Adhesion MoleculePositive Cells in Bone Marrow from Patients with Breast Carcinoma Immunomagnetic Purification of Micrometastases

Vale´rie Choesmel, A.I.1 Philippe Anract, M.D.2 Hanne Høifødt, A.I.3 Jean-Paul Thiery, Ph.D.1 Nathalie Blin, Ph.D.1 1

UMR144 CNRS, Research Division, Institut Curie, Paris, France.

2

Orthopedic Surgery Department, Hoˆpital Cochin, Paris, France.

3

Tumor Biology Department, The Norwegian Radium Hospital, Oslo, Norway.

Supported by the Programme Incitatif et Coope´ratif on Micrometastases at the Institut Curie. The authors thank Dr. Jacqueline Jouanneau for constructive comments in writing the article. Special thanks to patients for giving their consent to participate to this study. Address for reprints: Nathalie Blin, Ph.D., Institut Curie, CNRS UMR144, 26 rue d’Ulm, 75248 Paris Cedex 05, France; Fax: (011) 33 142346349; E-mail: [email protected] Received February 29, 2004; revision received April 26, 2004; accepted April 29, 2004.

BACKGROUND. The efficient detection and characterization of micrometastatic cells in the bone marrow of patients with breast carcinoma are of prognostic and therapeutic importance. The technique used must overcome the challenges that result from the small number of target cells (1 per 1 million hematopoietic cells) and the heterogeneous expression of micrometastatic cell markers. In this study, the authors assessed and improved the current methods for purifying and characterizing rare disseminated carcinoma cells. METHODS. The authors developed a single-step assay that does not require densitygradient separation. This assay can be performed directly on crude human bone marrow aspirates and is based on the use of immunomagnetic beads coated with an antibody that recognizes an epithelial cell-surface epitope, the epithelial cell adhesion molecule (EpCAM). To determine the specificity of the assay, the authors evaluated bone marrow specimens from 46 control patients. RESULTS. The novel method was highly reproducible and was capable of detecting as few as 10 carcinoma cells among 50 million hematopoietic cells. The yield was nearly 100%, with only 0.01% nonspecific cell draining. The authors found that 68 ⫾ 51 cells were trapped per 50 million cells in control crude aspirates and that density-gradient separation increased this number by 2-fold to 29-fold. These trapped cells expressed EpCAM, represented 1.4 ⫻ 10⫺4 % of the sample, and were characterized as of hematopoietic cell origin (CD45 positive) or progenitor cell origin (CD34 positive). CONCLUSIONS. The authors developed a highly efficient and reproducible, singlestep immunomagnetic assay that may be performed directly on crude human bone marrow aspirates. The authors believe the current study is the first to demonstrate that some rare bone marrow cells (CD45-positive or CD34-positive cells) may express EpCAM and, to some extent, may contaminate the purified micrometastatic cell fraction. Thus, a universal marker for micrometastatic cells remains to be discovered. Cancer 2004;101:693–703. © 2004 American Cancer Society. KEYWORDS: breast carcinoma, bone marrow, micrometastases, immunomagnetic selection, epithelial cell adhesion molecule.

B

reast carcinoma remains one of the leading causes of tumorrelated deaths in developed countries and affects one in every eight women. Breast carcinoma can recur due to the primary tumor spreading even before the initial diagnosis. In recent years, the lymphatic and hematogenous routes of tumor dissemination have been

© 2004 American Cancer Society DOI 10.1002/cncr.20391 Published online 21 June 2004 in Wiley InterScience (www.interscience.wiley.com).

694

CANCER August 15, 2004 / Volume 101 / Number 4

explored. Analysis of lymph nodes and blood appears to be an appealing approach for the detection of occult metastatic cells. However, reports in the field are controversial, and the clinical relevance of such analyses remains unclear. More recently, the presence of micrometastatic cells in the bone marrow (BM) from patients with breast carcinoma has been reported as an independent indicator of the risk of recurrence.1,2 Several more recent studies have assessed the clinical relevance of BM micrometastases by using different techniques. Immunocytochemistry directed against epithelial cytokeratins (CK) expressed in disseminated carcinoma cells has been studied extensively.3–5 However, the frequency of CK-positive cells in BM cell preparations from patients with malignant disease is approximately 10⫺5 to 10⫺6. Hence, this method requires screening numerous slides and involves large cohorts of patients. Reverse transcriptasepolymerase chain reaction targeting CK or mucin-like glycoproteins is not suitable for the detection of rare cells disseminated in BM, mainly because of illegitimate transcripts.6 – 8 Immunomagnetic (IM) techniques have been developed to purge tumor cells for BM transplantation and to improve the detection of tumor cells disseminated in blood samples.9 –13 IM beads coated with antibodies directed against the epithelial cell adhesion molecule (EpCAM) have been used to enrich, detect, and characterize disseminated carcinoma cells in BM samples from patients with breast, colorectal, and ovarian carcinomas.14 –17 These studies found no correlation between the amount of micrometastatic cells in BM and the stage of the disease. Improved methods for the detection of disseminated tumor cells in patients in the early stages of the disease at the time of diagnosis may allow a more accurate assessment of prognosis and aid in selecting candidates for adjuvant systemic therapy. Purifying BM micrometastases from patients with malignancies should help the characterization of these cells and therefore improve our understanding of the metastatic process, allowing us to develop novel immunotherapies. The objective of the current study was to improve the IM method for the quantitative and qualitative detection/purification of micrometastatic cells in BM samples from patients with breast carcinoma. Using a model system, we first developed a single-step assay that does not involve density-gradient centrifugation, thus improving the cell capture yield. Next, the specificity of the assay was evaluated using BM samples from control donors. The related trapped cells were characterized further.

MATERIALS AND METHODS Patients Control BM samples were collected from 46 patients who underwent hip surgery in the Orthopedic Department of Hoˆpital Cochin (Paris, France). The mean age (⫾ standard deviation [SD]) of the control group was 62 ⫾ 14 years. The medical record for each patient was carefully checked for any previous carcinoma, and patients with doubtful diagnoses were excluded. Additional control BM samples from healthy organ donors, patients with Ewing (or synovial) sarcoma, and patients with melanoma were kindly provided by Prof. M. Cavazzana-Calvo (Hoˆpital Necker, Paris, France), Dr. O. Delattre (Institut Curie, Paris, France) and Prof. O. Fodstad (The Norwegian Radium Hospital, Oslo, Norway), respectively. BM samples from patients with breast carcinoma were collected in the Surgical and Medical Oncology Departments of Institut Curie, thanks to Drs. C. Nos and J. Y. Pierga. BM aspirates were obtained from the upper iliac crest during surgery or under local anesthesia before chemotherapy was intiated. Written informed consent was obtained from all patients.

BM Processing BM samples were collected in sodium citrate Vacutainer tubes (Becton Dickinson, Franklin Lakes, NJ). Sample volumes ranged from 4 mL to 22 mL (mean ⫾ SD, 14 ⫾ 4 mL), yielding 656 ⫾ 447 million nucleated cells. This yielded a mean concentration of 47 million nucleated cells per mL, which was used as a reference value in all of our experiments. BM aspirates were processed immediately after collection or were stored at 4 °C for no longer than overnight. After washing in a 10-fold volume of Hank balanced salt solution (HBSS), BM samples were resuspended in HBSS and layered over a Ficoll solution (HistoPaque1077; Sigma Diagnostic, St Louis, MO) in 10-mL LeucoSep tubes (VWR International, Darmstadt, Germany) before they were centrifuged at ⫻ 400 g for 20 minutes. The resulting density-gradient interface was collected and washed in 0.1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS). The Lymphoprep (Axis-Shield, Oslo, Norway) and the OncoQuick (Greiner Bio-One, Frickenhausen, Germany) density-gradient centrifugation systems were also tested on BM samples from several control patients and were compared with the HistoPaque results. For experiments that did not include a Ficoll processing step, clots and clumps were first removed from the crude sample by pipetting up and down. An aliquot containing 50 million nucleated BM cells was washed in a 10-fold volume of HBSS. After centrifugation at

IM Assay for Detecting EpCAM in Bone Marrow/Choesmel et al.

⫻ 570 g for 15 minutes, the pellet was resuspended in 2 mL of 1% BSA PBS at 4 °C and was kept on ice until it was processed.

695

reaction of their MOC31 monoclonal antibody coat with the secondary antimouse antibody.

Immunomagnetic Enrichment Cell Lines MCF7 and T47D cell lines originate from human breast carcinoma pleural effusions (American TypeCulture Collection, Manassas, VA), and BC-H1 is a BM micrometastatic cell line (from a patient who had breast carcinoma) that was engineered and kindly provided by Prof. K. Pantel (Institute of Tumor Biology, University Hospital of Hamburg, Eppendorf, Germany). All cell lines were cultured under appropriate conditions and were mycoplasma-free. For spiking assays, cell lines were first labeled for 45 minutes at 37 °C with a vital dye probe (5-chloromethylfluorescein diacetate [CMFDA]; Molecular Probes, Eugene, OR) before they were added to BM samples. This brightgreen, long-lasting, fluorescent dye enables the detection, in an experimental model, of as few as 10 live cells among 50 million nucleated BM cells.

Antibodies and Immunofluorescence Assays The MOC31 monoclonal antibody (kindly provided by Prof. O. Fodstad, The Norwegian Radium Hospital, Oslo, Norway) and the VU1D9 monoclonal antibody (Biogenesis, Poole, U.K.) are directed against the same extracellular epitope of EpCAM, and were both used at a final concentration of 1 ␮g/mL. The gp94 antibody (from Prof. O. Fodstad’s laboratory) targets a 94-kilodalton (kD) glycoprotein that is present in human melanoma and carcinoma cells and was used at a final concentration of 1 ␮g/mL. A45-B/B3 (Chromavision, San Juan Capistrano, CA) recognizes a common epitope of the epithelial CKs: CK8, CK18, and CK19. Immunolabeling with A45-B/B3 (1 ␮g/mL final concentration) is possible after cells are fixed in 4% formaldehyde and 0.2% sucrose in PBS. The antihuman leukocyte common antigen CD45 (4 ␮g/mL final concentration), antihuman leukocyte-differentiation antigen CD34 (1 ␮g/mL final concentration), and murine immunoglobulin G1 (IgG1) negative control (2 ␮g/mL final concentration) were purchased from Dako (Glostrup, Denmark). For immunofluorescence experiments, Alexa-Fluor 488 (green) and 594 (red) goat antimouse secondary antibodies (Molecular Probes) were used at a final concentration of 2 ␮g/mL, and control experiments were performed in the absence of primary antibody. Cells on slides were incubated simultaneously with primary and secondary antibodies for 1 hour at room temperature. After mounting with Dako fluorescent mounting medium, the slides were observed using fluorescence microscopy. Magnetic beads appeared fluorescent due to the

M450 magnetic beads (4.5 ␮m) coated with sheep antimouse IgG (Dynal, Oslo, Norway) were conjugated to the MOC31 (anti-EpCAM) monoclonal antibody. After washing in the presence of a magnet, beads were coated with 2 ␮g antibody per mg of beads for at least 2 hour at 4 °C in 500 ␮L of 1% BSA PBS (bead buffer). After washing to remove excess antibody, the bead suspension (4 ⫻ 108 beads/mL) was ready to use and was stored in bead buffer at 4 °C for up to 3 months. After numerous calibration experiments in which control BM samples were spiked with CMFDA-labeled cells, the commonly described IM protocol was modified as follows: for each experiment, 50 million nucleated BM cells per tube were subjected to IM enrichment. The reaction was performed with 10 ␮L of bead suspension (4 ⫻ 106 beads) in bead buffer at 4 °C for 30 minutes under rotation in a 2-mL total volume. After washing twice in 5 mL of ice-cold bead buffer, 50 ␮L of RosetteSep antibody cocktail (StemCell Technologies, Vancouver, British Columbia, Canada) were added and incubated for an additional 20 minutes at 4 °C. Thus, CD45-positive cells (lymphoid cells) and CD66b-positive cells (granulocytes) were depleted during a third washing step. Trapped cells were then collected in a 50-␮L total volume, and the whole fraction was analyzed on glass slides using light microscopy. Cells with visible nuclei or membrane, measuring more than 12 ␮m, and rosetted with at least 5 beads were considered positive. For negative selection or depletion IM experiments, commercially available CD45-coated beads (Dynal) were compared with M450 Dynabeads conjugated to the Dako CD45 antibody. The CELLection Pan Mouse IgG kit (Dynal) was used to detach magnetic beads from trapped cells, according to the manufacturer’s instructions, with slight modifications. The cell recovery yield was increased by including 2 consecutive detachment cycles with 0.3 mg/mL DNase I (Roche Diagnostic, Mannheim, Germany) for 15 minutes at 37 °C in a 500-␮L volume. With this method, only approximately 50% of the IM trapped fraction is recovered, and cells are prone to damage, which means that only immunofluorescence techniques can be performed.

RESULTS Loss of Micrometastatic Cells in the Ficoll Enrichment Step and New Method for BM Processing A density-gradient centrifugation (Ficoll) isolation step is used routinely to enrich mononuclear cells

696

CANCER August 15, 2004 / Volume 101 / Number 4

(mnc) from blood and BM before detecting micrometastases by immunocytochemistry or IM sorting. Because there are only 1–10 micrometastases per 1 million BM mnc in patients with carcinoma, all micrometastases need to be recovered. To evaluate the proportion of micrometastases lost during the steps preceding their detection, we set up a model experiment in which control BM samples were spiked with labeled MCF7, T47D and BC-H1 cell lines. Due to the large heterogeneity of expression of intracellular CKs in cell lines (data not shown), we chose the CMFDA fluorescent dye. This dye is trapped in living cells and labels them irreversibly and homogeneously. We showed that 34 ⫾ 2% of tumor cells were recovered after Ficoll enrichment (n ⫽ 4 experiments), indicating that only approximately one-third of the carcinoma cells exhibit a buoyancy close enough to the mnc fraction to be collected after density-gradient separation. Therefore, we tested several erythrocytelysing reagents. However, this resulted in cell damage and consecutive loss during subsequent centrifugation steps. For all of these reasons, we developed a new method of BM processing. First, the freshly harvested BM sample was used directly to recover micrometastatic cells. Second, two additional steps were introduced into the standard IM protocol to lessen the impact of working with crude BM aspirates. We included an initial washing step to remove excess lipids and serum and a CD66b/CD45-positive cell depletion step to trap and to remove granulocytes and other hematopoietic cells before recovering the fraction of interest.

A Simple, Fast, Efficient, Sensitive and Reproducible IM Purification Assay The optimization steps described earlier were introduced into the standard IM enrichment assay using 4.5 ␮m Dynabeads coated with EpCAM-specific antibodies. We chose MOC31 rather than BerEp4, the two most commonly used antibodies for carcinoma cell immunopurification because of its higher affinity as determined by fluorescent-activated cell sorter analysis with MCF7, T47D and BC-H1 metastatic cell lines (data not shown). IM calibration experiments, which involved the spiking of crude BM samples with CMFDA-labeled cells, led us to lower the usual amount of MOC31-coated Dynabeads to 10 ␮L. This amount allowed us to recover all spiked cells and to ensure that excess free beads did not form unspecific aggregates. It also allowed us to analyze the whole IM-trapped cell fraction on a single slide. This inexpensive assay is simple to set up and can be performed in less than 1 hour.

FIGURE 1.

Spiking experiments involving the addition of MCF7 cells to control bone marrow (BM) samples. (A) Approximately 100 MCF7 cells were mixed with 50 million nucleated BM cells, immunopurified using M450 Dynabeads conjugated to the MOC31 antibody, and analyzed by light microscopy. (B) After immunomagnetic (IM) purification, cytokeratins were detected by indirect immunofluorescent labeling with AF488-coupled antimouse antibody (green). (C) Spiked cells were labeled with the 5-chloromethylfluorescein diacetate fluorescent (CMFDA) dye (green), IM purified, and labeled for cytokeratins using AF594-coupled antimouse antibody (red). In Panels B and C, the TIFF images of the visible and fluorescence records were merged using Adobe姞 Photoshop姞 software (version 7.0, Adobe Systems Inc., San Jose, CA). Magnetic beads appear fluorescent due to the reactivity of the MOC31 monoclonal antibody with the secondary antimouse antibody. Original magnification ⫻ 100(A); ⫻ 400 (B,C).

A qualitative approach showed that the MCF7, T47D and BC-H1 cell lines could be immunopurified by using this optimized IM protocol. Cells are surrounded by a crown or a husk of beads, depending on their EpCAM expression level (Fig. 1A). Some cells were not immunoreactive to the anti-CK antibody (Fig. 1B), whereas some were immunoreactive (Fig. 1C), confirming the heterogeneous expression pattern of this intracellular antigen18 and emphasizing the benefit of using EpCAM IM capture. We next performed quantitative IM spiking assays: we mixed 10 –100 CMFDA-labeled cells with 50 million control BM cells, thus simulating the physiopathologic dissemination of micrometastatic cells in the BM of patients with breast carcinoma (Table 1). Despite some inherent variability because of the dif-


697

TABLE 1 Yield and Sensitivity of the Immunomagnetic Assay CMFDA-labeled cells recovered by IM sorting (MOC31-positive) Cell lines Control BM

T47D BM1

BM2

BC-H1 BM2

BM3

BM4 Mean ⴞ SD

No. of cells spiked 10 50 100

42 117 188

20 107 220

18 69 121

4 ND 25

6 28 66

18 ⫾ 7 80 ⫾ 18 124 ⫾ 36

CMFDA: 5-chloromethylfluorescein diacetate; IM: immunomagnetic; BM: bone marrow; SD: standard deviation; ND: not determined.

ficulty in accurately picking up 10 cells, 50 cells or 100 cells, the mean cell recovery yield (⫾ SD) was 123 ⫾ 41% (n ⫽ 10 experiments). We estimated the nonspecific cell rate to only 0.01%. It was mostly due to erythrocytes, which are easily recognizable under light microscopy (n ⫽ 5 experiments). This assay was very sensitive, with the ability to detect 10 cells among 50 million nucleated BM cells. The assays were also highly reproducible, both for the metastatic T47D cell line and for the micrometastatic BC-H1 cell line, as they were using Dynabeads coated with the VU1D9 antibody directed against the same EpCAM epitope as MOC31 (data not shown).

IM Purification using BM Aspirates from Patients with Breast Carcinoma and from Control Patients We evaluated the ability of our IM assay, which was developed on model systems, to purify micrometastatic cells in BM aspirates from patients with breast carcinoma. The mean number of cells (⫾ SD) that were trapped from 50 million nucleated BM cells was 84 ⫾ 74 cells for patients with “localized disease” (n ⫽ 10 experiments) and 192 ⫾ 100 cells for patients with “advanced disease” (n ⫽ 22 experiments). In some specimens of “advanced disease,” an impressive number of tumor cells could be immunopurified (Fig. 2). It is interesting to note that the T47D metastatic cell line exhibits morphologic and adhesive properties similar to those of the trapped BM micrometastases, leading to cell clustering. To assess the specificity of this powerful IM assay, we analyzed 46 BM samples from control patients who underwent hip surgery. Surprisingly, light microscopy revealed several MOC31-positive cells in most samples (Fig. 3). These cells were of different sizes and morphologies. We detected a mean (⫾ SD) of 68 ⫾ 51 MOC31-positive cells per 50 million nucleated BM

FIGURE 2. Immunomagnetic (IM) purification of micrometastases in bone marrow (BM) samples from patients with “advanced disease” breast carcinoma. BM cells were incubated with MOC31-conjugated Dynabeads in the presence of CMFDA-labeled (green) T47D cells (depicted in the insert). After IM purification, cells were recovered and analyzed by light/fluorescence microscopy. Note the clustering of T47D metastatic cells with BM micrometastases. Original magnification ⫻ 200 (A); ⫻ 400 (B).

cells, which is equivalent to a frequency of 1.36 ⫻ 10⫺6. This small cell fraction, however, is sufficient to contaminate the IM detection of rare events, such as micrometastases disseminated in patient’s BM. We estimated that this contamination rate is 0.74 –7.4% for patients with “advanced disease”, but it may reach 7.4-74% for patients with “localized disease.”

Significance of IM Trapping in Control Patient BM Aspirates The significance of trapping cells in control BM samples by IM sorting was addressed first relative to patient characteristics. The control patients displayed no obvious present or past carcinoma at the time of hip surgery and were not receiving any po-

698

CANCER August 15, 2004 / Volume 101 / Number 4 TABLE 2 The Impact of Ficoll Processing on the Trapped Cell Fraction in Control Bone Marrow Samples Control BM/processing BM5 Crude Ficoll (Histopaque) Ficoll (Histopaque) BM6 Crude Ficoll (Histopaque) Ficoll (OncoQuick) BM7 Crude Ficoll (Histopaque) Ficoll (OncoQuick) BM8 Crude Ficoll (Histopaque) Ficoll (LymphoPrep) BM6 Crude Ficoll (Histopaque) Ficoll (LymphoPrep)

MOC31-positive cells (for 50 ⴛ 106 BM cells)

19 184 190 1 235 143 150 360 250 36 147 262 56 178 238

BM: bone marrow.

FIGURE 3. Immunomagnetic (IM) trapping of cells in bone marrow (BM) aspirates from control patients. Note the differences in the size and morphology of cells trapped from crude BM samples using MOC31conjugated Dynabeads (original magnification, ⫻ 400). Similar discrepancies in cell size and morphology were observed using BM samples that were processed by density-gradient centrifugation (Ficoll). tentially interfering pharmacologic treatment. The mean age (⫾ SD) in the control population (61 ⫾ 13 years; n ⫽ 46 patients) was similar to the mean age in the group of patients with overt clinical carcinoma (50 ⫾ 9 years; n ⫽ 32 patients), and no clear relation was found between the number of cells trapped and patient age. Further evidence for the

presence of these cells was provided by screening control BM samples from various origins. BM samples from 2 healthy organ donors, 2 patients with Ewing sarcoma, 1 patient with synovial sarcoma, and 1 patient with melanoma all were positive, with 5–284 MOC31-positive cells per 50 million nucleated BM cells. Passive adsorption of beads also may contribute to the trapping of normal BM cells. We assayed uncoated epoxy-Dynabeads, as well as antimouse antibody-coated M450 Dynabeads, with control BM samples and never trapped more than four cells. We also excluded the possibility of a purely mechanistic effect, because the number of cells trapped in control BM was proportional to the number of nucleated BM cells added. We hypothesized that clearance of the BM aspirate by Ficoll separation should reduce this phenomenon. Several density gradients (HistoPaque, LymphoPrep, and OncoQuick) were assessed in parallel (Table 2). Unexpectedly, Ficoll processing increased cell capture in the control BM samples by 2–29-fold (n ⫽ 7 experiments). This suggests that Ficoll, de novo activated cryptic epitopes for some hematopoietic cells. Another potential pitfall is that receptors at the surface of monocytes and macrophages specifically bind the IgG Fc-fragments, leading to improper trapping of IgG-coated Dynabeads. We therefore used the IgG1 control isotype of the MOC31 anti-

IM Assay for Detecting EpCAM in Bone Marrow/Choesmel et al. TABLE 3 Evaluation, using IM Depletion, of the Contribution of Fc-Receptors to the Trapped Cell Fraction in Control Bone Marrow Samples Control BM/IM selection BM10 (crude) First cycle MOC31 Second cycle IgG1 Third cycle IgG1 Fourth cycle MOC31 BM11 (crude) First cycle MOC31 Second cycle IgG1 Third cycle IgG1 Fourth cycle MOC31 BM11 (Ficoll) First cycle MOC31 Second cycle IgG1 Third cycle IgG1 Fourth cycle MOC31

TABLE 4 Evaluation, using Blocking Reagent, of the Contribution of FcReceptors to the Trapped Cell Fraction in Control Bone Marrow Samples


96 6 3 101 39 0 1 47 520 64 12 414

BM: bone marrow; IM: immunomagnetic; IgG1: immunoglobulin G1.

body to determine whether two incubation periods with IgG1 beads decreased the number of MOC31positive cells detected in control BM samples (Table 3). We observed no significant Fc-receptor-specific binding that could explain IM trapping in crude or in Ficoll-processed control BM samples. Further evidence was provided by Ficoll-processed BM samples that were treated or not treated with an Fcreceptor-blocking reagent (Table 4). Finally, we performed immunofluorescence and IM experiments using the gp94 antibody directed against a surface protein unrelated to EpCAM, which is expressed in human melanoma and carcinoma cells.19 T47D and BC-H1 cells were indeed immunoreactive to the gp94 antibody, but four out of six BM cytospots from independent control patients were also labeled. Moreover, gp94-coated Dynabeads trapped 1.3–5.7-fold more cells from 3 different control BM samples compared with MOC31-coated Dynabeads (data not shown).

Characterization of Cells IM-Trapped in BM Samples from Control Patients To characterize this cell fraction better, we performed immunofluorescence experiments with IM-trapped cells or direct cytospots of crude and Ficoll-processed control BM aspirates, respectively. Because the beads hindered visual observation, we attempted to remove them by using an assay in which the conjugated Dynabeads included a DNase-sensitive linker. This demonstrated that some BM cells of hematopoietic origin (CD45-negative or CD34-positive cells) indeed did ex-

699


Control BM

Without FcR blocking reagent

With FcR blocking reagent

BM12 (Ficoll) BM13 (Ficoll) BM14 (Ficoll)

100 144 214

82 130 196

BM: bone marrow; FcR: Fc-receptors.

press EpCAM (MOC31-positive) and, to a lesser extent, CKs (A45-B/B3-positive) (Fig. 4). Control breast carcinoma cell lines, as expected for tumor cells, were negative for CD45 and CD34. The IM-trapped cells in control BM samples, in fact, were rare hematopoietic cells, which do express EpCAM at their surface and may contaminate the micrometastatic cell fraction sorted out from BM samples from patients with breast carcinoma. Therefore, we depleted the CD45-reactive hematopoietic population prior to IM sorting to determine the effect of eliminating leukocytes. However, IM depletion using CD45-coated beads before MOC31 IMpositive selection did not appear to decrease the number of cells trapped in control BM aspirates (Table 5). This may be explained by an incomplete depletion procedure due to the large numbers of CD45-positive cells, and because of large variations in the expression of CD45 at the cell surface.

DISCUSSION Immunocytochemical (IC) detection of B7 micrometastases has been studied extensively in patients with breast carcinoma and other malignancies.2,20 –25 In recent years, many IM-enrichment techniques have been developed for micrometastasis detection.10 –12,15–17 However, to our knowledge, none of those studies found a clear correlation between the IM detection of micrometastatic cells and the stage of disease or prognosis. Moreover, the use of different types of beads (Dynal or Miltenyi), different magnetic selection procedures (positive or negative), different markers (EpCAM or MUC1), and different antibodies (BerP4 or MOC31) make it difficult to assess the data and the clinical impact of the technique. Because we are convinced that enrichment is the critical step when trying to detect rare cells, we believe that it is essential to evaluate the performance of the IM technique rigorously before undertaking clinical studies.

700

CANCER August 15, 2004 / Volume 101 / Number 4 TABLE 5 CD45 Immunodepletion in Control Bone Marrow Samples to Assess the Contribution of Hematopoietic Cells to the Immunoselected Cell Fraction Control BM

IM depletion/selection cycles


BM15

MOC31 Two cycles of M450-CD45 ⫹ MOC31 MOC31 Two cycles of M450-CD45 ⫹ MOC31 One cycle of DYNAL-CD45 ⫹ MOC31 Two cycles of DYNAL-CD45 ⫹ MOC31 MOC31 Two cycles of M450-CD45 ⫹ MOC31 Two cycles of DYNAL-CD45 ⫹ MOC31

59 53 20 19 31 28 80 79 75

BM16

BM17

BM: bone marrow; IM: immunomagnetic.

FIGURE 4. Characterization of the cell fraction trapped in bone marrow (BM) samples from control patients. Control BM samples were immunopurified using MOC31-conjugated Cellection Dynabeads. After bead removal, immunofluorescence experiments were performed using anticytokeratin/AF488 (green) and anti-CD45/AF594 (red) antibodies (A) or anti-CD34/AF488 (green) antibodies (B) and were analyzed by light/fluorescence microscopy. CD34 labeling was performed on three independent BM specimens. Original magnification ⫻ 200 (B); ⫻ 400 (A). For this purpose, we have established model systems in which breast carcinoma cell lines are labeled before they are mixed with control BM specimens. Experiments were performed with different metastatic and micrometastatic cell lines to mimic the physiologic model of tumor cell dissemination as closely as possible, and to avoid bias due to approximate morphologic or densitometric parameters. Because human breast carcinoma are highly heterogeneous, with notable differences in morphology and antigenicity,26 and because additional heterogeneity may arise from the differential expression of markers between the primary site and the BM metastatic compartment during disease progression,6,27 we avoided antibody labeling and set up a rigorous calibration assay using a fluorescent dye, which is trapped in living cells and labels them irreversibly and homogeneously. We used this model system to assess different methods of IM selection on the basis of tumor cell recovery rate, enrichment factor, and ease of use. The magnetic column-based Miltenyi/MACS system works well with blood, although tumor cell recovery yields are low.12 We obtained low tumor cell

recovery rates with the MACS technology and BM specimens (range, 48 –59%). Similar to other authors,28 we experienced additional problems related to the thickness of BM samples, leading to blockage of the magnetic columns. Moreover, a high level of nonspecific trapping occurred using these nanobeads (data not shown). This most likely was due to their propensity for hydrophobic interactions. Conversely, the Dynal IM selection system using superparamagnetic microbeads was highly feasible in the micrometastases detection field but could not predict clinical outcomes.10,14,16,17,29 We have chosen to evaluate Dynal IM enrichment rather than IM depletion assays, which are unsatisfactory with BM samples because of poor efficiency and of the loss of tumor cells.22 Due to the heterogeneous nature of metastatic marker expression, we selected cells on the basis of their epithelial cell surface antigenicity. MUC1 is a large, heavily glycosylated mucin that is expressed on the apical surfaces of mammary gland secretory epithelia. Although it was believed originally that MUC1 was an epithelial-specific protein, it is now known that MUC1 is also expressed on a variety of hematopoietic cells,30 –32 meaning that it cannot be used in IM enrichment assays. EpCAM, which is also named 17-1A, ESA, EGP40, EGP2, or GA733.2, is an epithelial transmembrane glycoprotein and a homophilic cell-cell adhesion molecule, the expression of which correlates with cell proliferation and dedifferentiation along with a progression in tumorigenicity.33,34 EpCAM is overexpressed in most malignancies, especially colorectal carcinoma16,35 and breast carcinoma.36 –38 Several antibodies have been raised against the external domain of the molecule, and the most commonly used are MOC31 and BerEP4. We chose MOC31 as a reference after comparing the affinity of MOC31 and BerEP4


toward several tumor cell lines using flow cytometry and IM techniques (data not shown). One of the main steps in micrometastatic cell detection, that to our knowledge has never been explored carefully, is the density-gradient centrifugation separation (Ficoll) of mnc from BM samples. We have shown here that only 34% of the tumor cells are recovered after Ficoll. We describe, for what to our knowledge is the first time, an IM enrichment protocol in which the whole crude BM aspirate is treated at once and is analyzed on a single slide. This simple, fast and reproducible IM purification assay exhibits recovery yields close to 100%, only 0.01% nonspecific cell draining (mainly erythrocytes) and a sensitivity of 10 cells per 50 million nucleated BM cells. The specificity of this powerful IM assay was assessed further by analyzing a large number of BM samples from control patients originating from different sources. Previous studies reported the absence of cell IM-trapping when testing control patients.15–17 In the current study, we demonstrate that few contaminating cells (frequency ⫽ 1.4 ⫻ 10⫺6) are immunopurified in most cases, and may contaminate the equally rare carcinoma cells isolated in BM aspirates from patients with breast carcinoma. It is unclear how these contaminants influence the results. However, we have demonstrated elsewhere18 the clinical relevance of this optimized IM method for the detection of medullar micrometastatic cells in patients with breast carcinoma.18 In the current study, we demonstrated that Ficoll processing of control BM samples increased the number of trapped cells by 2–29-fold in a Ficoll compositionindependent manner. This phenomenon, described here for the first time in a study on micrometastases detection, may arise from the activation of lymphocytes by Ficoll gradients39,40 or by endotoxins associated with Ficoll preparations.41 Ficoll may also affect cells by exposing them sequentially to dextran, gradient centrifugation and hypotonic conditions.42 These findings emphasize the importance of avoiding density-gradient centrifugation separation when processing BM specimens. Our objective was to characterize these control BM EpCAM-positive cells, that to our knowledge had never been described before with IM or IC techniques. Cellection Dynabeads, which include a cleavable DNA linker, have been used successfully to confirm tumor cell morphology in cytologic fluid specimens.43,44 We optimized this technique for BM processing. According to one report,43 the mean cell recovery rate is less than 60%, and cells that are subjected to DNase become sensitive to shear stress. Consequently, morphologic staining, IC and fluorescent in situ hybridization techniques were too drastic for the analysis of the few

701

remaining EpCAM-positive cells. Therefore, we determined whether the cells expressed antigens typical of hematopoietic cells using the gentle immunofluorescence technique following IM selection. We detected hematopoietic lineage antigens, such as CD45 and CD34, on EpCAM-positive cells. CD45, or the leukocyte common antigen, is a heavily glycosylated cell surface protein-tyrosine phosphatase that is highly expressed on hematopoietic cells. It is believed to be a critical marker of lymphocyte activation, differentiation and maturation, and is expressed on all hematopoietic cells except erythrocytes.45 The CD34 antigen is a heavily glycosylated type I transmembrane protein that is expressed on hematopoietic progenitor cells, and is involved in cell-cell adhesion and maintenance of a phenotypically plastic state in undifferentiated cells.46 According to our results, a subpopulation of EpCAM-positive cells was found in BM using an antibody directed against CD34-positive cells.47 In that study, the EpCAM-positive/CD34-positive cells were characterized as immature erythroid cells, and peripheral blood CD34-positive cells, in vitro induced to differentiate into the erythroid lineage, were found to strongly express EpCAM. In addition, CD34-positive progenitor cells purified from the peripheral blood of healthy volunteers also expressed EpCAM.48 It is noteworthy that the MUC1 carcinoma-associated mucin antigen is expressed on normal BM cells, and particularly on CD34-positive cells of the erythroid lineage.30 The sensitive reverse transcriptase-polymerase chain reaction technique has detected illegitimate transcripts for EpCAM and MUC1 genes in blood cells.49,50 Ectopic transcription in CD34-positive cells may reflect incomplete inactivation of thousands of leaky genes in pluripotent cells. CD34-positive cells, therefore, may contaminate the purified fraction when these markers are used to detect rare micrometastatic cells. We have developed a simple, fast, efficient, sensitive, and reproducible IM purification assay for the detection and purification of BM micrometastatic cells from patients with breast carcinoma. However, the specificity of the method is hampered by the fact that subpopulations of hematopoietic pluripotent cells have an expression pattern that involves nearly all human genes, and by the fact that a universal marker for micrometastatic cells remains to be discovered.

REFERENCES 1.

Pantel K, Cote RJ, Fodstad O. Detection and clinical importance of micrometastatic disease. J Natl Cancer Inst. 1999; 91:1113–1124.

702 2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

CANCER August 15, 2004 / Volume 101 / Number 4 Braun S, Pantel K, Muller P, et al. Cytokeratin-positive cells in the bone marrow and survival of patients with Stage I, II, or III breast cancer [see comments]. N Engl J Med. 2000;342: 525–533. Janni W, Gastroph S, Hepp F, et al. Prognostic significance of an increased number of micrometastatic tumor cells in the bone marrow of patients with first recurrence of breast carcinoma. Cancer. 2000;88:2252–2259. Braun S, Schindlbeck C, Hepp F, et al. Occult tumor cells in bone marrow of patients with locoregionally restricted ovarian cancer predict early distant metastatic relapse. J Clin Oncol. 2001;19:368 –375. Pierga JY, Bonneton C, Vincent-Salomon A, et al. Clinical significance of immunocytochemical detection of tumor cells using digital microscopy in peripheral blood and bone marrow of breast cancer patients. Clin Cancer Res. 2004;10: 1392–1400. Zippelius A, Kufer P, Honold G, et al. Limitations of reversetranscriptase polymerase chain reaction analyses for detection of micrometastatic epithelial cancer cells in bone marrow [see comments]. J Clin Oncol. 1997;15:2701–2708. Aerts J, Wynendaele W, Paridaens R, et al. A real-time quantitative reverse transcriptase polymerase chain reaction (RTPCR) to detect breast carcinoma cells in peripheral blood. Ann Oncol. 2001;12:39 – 46. Jung YS, Lee KJ, Kim HJ, et al. Clinical significance of bone marrow micrometastasis detected by nested rt-PCR for keratin-19 in breast cancer patients. Jpn J Clin Oncol. 2003;33: 167–172. Rye PD, Hoifodt HK, Overli GE, et al. Immunobead filtration: a novel approach for the isolation and propagation of tumor cells. Am J Pathol. 1997;150:99 –106. Naume B, Borgen E, Nesland JM, et al. Increased sensitivity for detection of micrometastases in bone-marrow/peripheral-blood stem-cell products from breast-cancer patients by negative immunomagnetic separation. Int J Cancer. 1998; 78:556 –560. Zhong XY, Kaul S, Lin YS, et al. Sensitive detection of micrometastases in bone marrow from patients with breast cancer using immunomagnetic isolation of tumor cells in combination with reverse transcriptase/polymerase chain reaction for cytokeratin-19. J Cancer Res Clin Oncol. 2000; 126:212–218. Kruger W, Datta C, Badbaran A, et al. Immunomagnetic tumor cell selection—implications for the detection of disseminated cancer cells. Transfusion. 2000;40:1489 –1493. Witzig TE, Bossy B, Kimlinger T, et al. Detection of circulating cytokeratin-positive cells in the blood of breast cancer patients using immunomagnetic enrichment and digital microscopy. Clin Cancer Res. 2002;8:1085–1091. Forus A, Hoifodt HK, Overli GE, et al. Sensitive fluorescent in situ hybridisation method for the characterisation of breast cancer cells in bone marrow aspirates. Mol Pathol. 1999;52: 68 –74. Ree AH, Engerbraaten O, Hovig E, et al. Differential display analysis of breast carcinoma cells enriched by immunomagnetic target cell selection: gene expression profiles in bone marrow target cells. Int J Cancer. 2002;97:28 –33. Flatmark K, Bjornland K, Johannessen HO, et al. Immunomagnetic detection of micrometastatic cells in bone marrow of colorectal cancer patients. Clin Cancer Res. 2002;8:444– 449. Marth C, Kisic J, Kaern J, et al. Circulating tumor cells in the peripheral blood and bone marrow of patients with ovarian carcinoma do not predict prognosis. Cancer. 2002;94:707–712.

18. Choesmel V, Pierga JY, Noo C, et al. Enrichment methods to detect bone marrow micrometastases in breast carcinoma patients: clinical relevance. Breast Cancer Res. In press. 19. Imai K, Wilson BS, Bigotti A, et al. A 94,000-dalton glycoprotein expressed by human melanoma and carcinoma cells. J Natl Cancer Inst. 1982;68:761–769. 20. Diel IJ, Kaufmann M, Costa SD, et al. Micrometastatic breast cancer cells in bone marrow at primary surgery: prognostic value in comparison with nodal status [see comments]. J Natl Cancer Inst. 1996;88:1652–1658. 21. Mansi JL, Gogas H, Bliss JM, et al. Outcome of primarybreast-cancer patients with micrometastases: a long-term follow-up study. Lancet. 1999;354:197–202. 22. Naume B, Borgen E, Kvalheim G, et al. Detection of isolated tumor cells in bone marrow in early-stage breast carcinoma patients: comparison with preoperative clinical parameters and primary tumor characteristics. Clin Cancer Res. 2001;7: 4122– 4129. 23. Gebauer G, Fehm T, Merkle E, et al. Epithelial cells in bone marrow of breast cancer patients at time of primary surgery: clinical outcome during long-term follow-up. J Clin Oncol. 2001;19:3669 –3674. 24. Gerber B, Krause A, Muller H, et al. Simultaneous immunohistochemical detection of tumor cells in lymph nodes and bone marrow aspirates in breast cancer and its correlation with other prognostic factors. J Clin Oncol. 2001;19:960 –971. 25. Wiedswang G, Borgen E, Karesen R, et al. Detection of isolated tumor cells in bone marrow is an independent prognostic factor in breast cancer. J Clin Oncol. 2003;21: 3469 –3478. 26. Denk H. Immunohistologic heterogeneity of malignant tumors. Pathol Res Pract. 1988;183:693– 697. 27. Cardoso F, Di Leo A, Larsimont D, et al. Evaluation of HER2, p53, bcl-2, topoisomerase II-alpha, heat shock proteins 27 and 70 in primary breast cancer and metastatic ipsilateral axillary lymph nodes. Ann Oncol. 2001;12:615– 620. 28. Thurm H, Ebel S, Kentenich C, et al. Rare expression of epithelial cell adhesion molecule on residual micrometastatic breast cancer cells after adjuvant chemotherapy. Clin Cancer Res. 2003;9:2598 –2604. 29. Nakamura T, Yasumura T, Hayashi K, et al. Immunocytochemical detection of circulating esophageal carcinoma cells by immunomagnetic separation. Anticancer Res. 2000; 20:4739 – 4744. 30. Brugger W, Buhring HJ, Grunebach F, et al. Expression of MUC-1 epitopes on normal bone marrow: implications for the detection of micrometastatic tumor cells. J Clin Oncol. 1999;17:1535–1544. 31. Berois N, Varangot M, Aizen B, et al. Molecular detection of cancer cells in bone marrow and peripheral blood of patients with operable breast cancer. Comparison of CK19, MUC1 and CEA using RT-PCR. Eur J Cancer. 2000;36:717–723. 32. Gendler SJ. MUC1, the renaissance molecule. J Mammary Gland Biol Neoplasia. 2001;6:339 –353. 33. Balzar M, Winter MJ, de Boer CJ, et al. The biology of the 17-1A antigen (Ep-CAM). J Mol Med. 1999;77:699 –712. 34. de Boer CJ, van Krieken JH, Janssen-van Rhijn CM, et al. Expression of Ep-CAM in normal, regenerating, metaplastic, and neoplastic liver. J Pathol. 1999;188:201–206. 35. Gottlinger HG, Funke I, Johnson JP, et al. The epithelial cell surface antigen 17-1A, a target for antibody-mediated tumor therapy: its biochemical nature, tissue distribution and recognition by different monoclonal antibodies. Int J Cancer. 1986;38:47–53.

IM Assay for Detecting EpCAM in Bone Marrow/Choesmel et al. 36. Braun S, Hepp F, Kentenich CR, et al. Monoclonal antibody therapy with edrecolomab in breast cancer patients: monitoring of elimination of disseminated cytokeratin-positive tumor cells in bone marrow. Clin Cancer Res. 1999;5:3999 – 4004. 37. Gastl G, Spizzo G, Obrist P, et al. Ep-CAM overexpression in breast cancer as a predictor of survival. Lancet. 2000;356: 1981–1982. 38. Spizzo G, Obrist P, Ensinger C, et al. Prognostic significance of Ep-CAM and Her-2/neu overexpression in invasive breast cancer. Int J Cancer. 2002;98:883– 888. 39. Kol R, Friedlander M, Riklis E, et al. Separation of human lymphocytes on Ficoll-Paque gradients: stimulation of cells and depletion of a concanavalin-A responsive radioresistant subpopulation. Radiat Res. 1983;95:108 –115. 40. Jahr H, Hering B, Federlin K, et al. Activation of human complement by collagenase and Ficoll. Exp Clin Endocrinol Diabetes. 1995;103(Suppl 2):27–29. 41. Jahr H, Pfeiffer G, Hering BJ, et al. Endotoxin-mediated activation of cytokine production in human PBMCs by collagenase and Ficoll. J Mol Med. 1999;77:118 –120. 42. Berkow RL, Tzeng DY, Williams LV, et al. The comparative responses of human polymorphonuclear leukocytes obtained by counterflow centrifugal elutriation and FicollHypaque density centrifugation. I. Resting volume, stimulus-induced superoxide production, and primary and specific granule release. J Lab Clin Med. 1983;102:732–742. 43. Werther K, Normark M, Hansen BF, et al. The use of the

44.

45. 46.

47.

48.

49.

50.

703

CELLection kit in the isolation of carcinoma cells from mononuclear cell suspensions. J Immunol Methods. 2000; 238:133–141. Kielhorn E, Schofield K, Rimm DL. Use of magnetic enrichment for detection of carcinoma cells in fluid specimens. Cancer. 2002;94:205–211. Thomas ML. The leukocyte common antigen family. Annu Rev Immunol. 1989;7:339 –369. Andrews RG, Singer JW, Bernstein ID. Monoclonal antibody 12-8 recognizes a 115-kd molecule present on both unipotent and multipotent hematopoietic colony-forming cells and their precursors. Blood. 1986;67:842– 845. Lammers R, Giesert C, Grunebach F, et al. Monoclonal antibody 9C4 recognizes epithelial cellular adhesion molecule, a cell surface antigen expressed in early steps of erythropoiesis. Exp Hematol. 2002;30:537–545. Fava TA, Desnoyers R, Schulz S, et al. Ectopic expression of guanyl cyclase C in CD34⫹ progenitor cells in peripheral blood. J Clin Oncol. 2001;19:3951–3959. De Graaf H, Maelandsmo GM, Ruud P, et al. Ectopic expression of target genes may represent an inherent limitation of RT-PCR assays used for micrometastasis detection: studies on the epithelial glycoprotein gene EGP-2. Int J Cancer. 1997;72:191–196. Bostick PJ, Chatterjee S, Chi DD, et al. Limitations of specific reverse-transcriptase polymerase chain reaction markers in the detection of metastases in the lymph nodes and blood of breast cancer patients. J Clin Oncol. 1998;16:2632–2640.