Identification of Circulating Melanoma Cells in Uveal Melanoma Patients by Dual-Marker Immunoenrichment

Abstract

Purpose.: Despite successful local tumor control, uveal melanoma (UM) patients may develop lethal metastases. To reliably identify circulating melanoma cells (CMC) in UM patients, we set out to test a new immunomagnetic enrichment assay and screened UM patients for the presence of CMC. We also determined whether we could find CMC in culture; for example, for future drug testing.

Methods.: A dual-immunomagnetic enrichment assay using antibodies against two melanoma markers (NKI/C3 and NKI/beteb) was used to determine the presence of UM cells in blood. The sensitivity of the assay was determined by spiking normal blood with 92.1 cells (concentration range, 1–10⁴ cells/mL). Isolated cells were characterized by immunocytochemistry directly after immunoenrichment and after a 2-week culture. The presence of CMC was determined in the peripheral blood of 31 patients with UM, and results were compared to clinical prognostic factors at the time of presentation.

Results.: The CMC were detected in 93.5% (n = 29 of 31) of the patients with primary nonmetastatic UM at a median density of 3.5 cells/10 mL blood (range, 0–10.2 cells), as well as in blood cultures. No significant association was observed between the presence or number of CMC and any clinical prognostic factors.

Conclusions.: The improved dual-immunoenrichment assay enabled the detection of intact and viable CMC in the majority of UM patients. We also were able to identify CMC after short-term culturing. Molecular characterization of the CMC rather than the prevalence of these cells is expected to provide relevant information on the individual risk of metastasis.

Introduction

Uveal melanoma (UM) is the most frequent intraocular malignancy in adults. Despite advances in diagnosis and local tumor control, up to 50% of the patients may have metastases within 5 to 10 years after treatment of the primary tumor. ^1,2 Metastases usually are detected at a stage when no effective treatment is available, resulting in an approximate survival time of 12 to 14 months following diagnosis. ^3,4 The latency in the development of metastases suggests that the UM cells already disseminated at the time of initial diagnosis and remained dormant in the new microenvironment until the conditions became favorable for their growth. ⁴ Since metastases of UM occur mainly through the hematogenous route, ^5,6 dissemination of the primary tumor cells into peripheral blood is a prerequisite. ⁷ Accordingly, the analysis of circulating melanoma cells (CMC) is expected to provide valuable information on the course of disease and isolation of such cells may provide information on the molecular mechanisms, which would be indispensable for the development of adjuvant therapies. 

Various approaches have been developed in the past years to detect CMC and analyze their clinical significance in UM patients. ⁷ Using RT-PCR, melanoma-associated transcripts were detected in the blood cell lysates of UM patients who exhibited no clinical evidence of metastasis. ⁸ However, the detection frequency varied greatly and ranged from 0% to 97%. ^9–12 Our group previously showed that immunomagnetic enrichment allowed the isolation of intact CMC (median, 2.5 cells/50 mL blood) in 19% (n = 10 of 52) of UM patients without clinical metastases. ¹³ The percentage of patients with metastatic disease that were positive for CMC ranged from 30% to 63%, as determined by the analysis of intact CMC using the automated CellSearch system and by RT-PCR, respectively. ^14,15 Proper identification of CMC clearly is a difficult issue. ⁷ Moreover, when intact CMC have been obtained by the currently available methods, they are not suitable for culturing, impeding the development of early therapies against metastases. 

The discrepancy in the detection frequency of CMC might have arisen from the presumably discontinuous nature of the shedding of these cells ¹² and the molecular changes that the CMC are likely to undergo in the circulation. ¹⁶ Additionally, the approaches currently used for the detection of CMC have some limitations. Amplification of pseudogenes and illegitimate transcripts may lead to the overestimation of the CMC ratio by RT-PCR. The cytometric method we have reported previously involved the use of a single monoclonal antibody (clone 9.2.27) against the melanoma-associated chondroitin sulfate proteoglycan (MCSP) for capturing the UM cells. ¹³ However, the use of this monoclonal antibody alone may not be sufficient for the immunomagnetic enrichment of all types of UM cells, as demonstrated by the inability to isolate the UM cell line SP6.5 from the mononuclear cell fraction when these tumor cells were spiked at a concentration below 1 × 10⁶ cells/mL human blood. ¹⁷ The CellSearch system employs antibodies against the melanoma-associated glycoprotein (CD146/NKI/beteb) for the capture as well as MCSP-antibodies for immunostaining of the CMC. ¹⁴ However, the MCSP antigen does not exhibit a uniform expression pattern in all the currently established UM cell lines ¹⁷ despite the detection of this protein in 95% of the UM samples. ¹⁸ Accordingly, only the combination of two monoclonal antibodies (NKI/beteb and NKI/C3) against the melanoma-associated glycoprotein allowed immunomagnetic isolation of as few as 10 cells/mL of a UM cell line spiked into normal blood. ¹⁷ The NKI/beteb and NKI/C3 antibodies could reveal a moderate to high staining intensity in surgically removed UM lesions, serving as appropriate markers for UM. ^19–21 However, this promising novel approach involving the combination of these two antibodies for immunoenrichment was, to our knowledge, not yet used for the screening of CMC in UM patients. 

In this study, we initially evaluated the sensitivity of a slightly modified version of this dual-immunoenrichment method by retrieving cultured UM cells that were spiked into normal blood or the mononuclear blood cell fraction at different densities. We then used the modified dual-immunoenrichment assay to isolate intact CMC from UM patients with no clinical evidence of metastasis at the time of testing. The presence and number of CMC were compared to established clinical prognostic factors, as we hypothesized that patients at high risk of metastases would shed more UM cells into the bloodstream, and at a higher frequency. A subpopulation of the isolated cells was cultivated for 2 weeks before being analyzed for purity as an initial attempt to establish CMC cultures. 

Materials and Methods

Patient Selection

Between December 2010 and December 2012, 31 consecutive patients with localized UM, presenting at the Department of Ophthalmology, University of Lübeck, Germany, were enrolled in the study. The condition of UM was diagnosed with clinical and ultrasound examination performed by a specialized ophthalmologist. Standardized A- and B-scans (I3 eye cubed System-ABD; Ellex, Inc., Sacramento, CA, USA) and ultrasound biomicroscopy (VuMax II; Sonomed, Inc., NY, USA) were obtained to evaluate the size of the intraocular tumor, the exact localization, and to determine ciliary body involvement. In cases of enucleation, resection, or biopsies, the diagnosis was verified histologically by the Institute of Pathology at the University of Lübeck. The study was authorized by the local ethics committee and conformed to the guidelines of the Declaration of Helsinki as revised in Tokyo and Venice. All patients gave their informed consent before inclusion in the study. 

Patients underwent liver function tests (alkaline phosphatase, aspartate aminotransferase, alanine aminotransferase, bilirubin); ultrasound of the abdomen; computer tomography of the chest and abdomen; and magnetic resonance tomography of the head. The data were coded and analyzed according to the seventh edition of tumor–node–metastasis (TNM) classification. ²²  

Controls

Nine control samples (n = 4 female, n = 5 male; median age, 58 years; range, 26–73 years) were obtained from seven “nonmelanoma” patients treated at the Department of Ophthalmology, University of Lübeck and two healthy volunteers. 

Culture of the Primary Tumor Cells

Primary tumor samples were obtained from UM patients undergoing endoresection or enucleation. Tumor samples were washed two times in 0.01 M PBS (GIBCO, Invitrogen, Darmstadt, Germany) with penicillin/streptomycin and dissociated into single cells using a neural tissue dissociation kit with papain (Miltenyi Biotec, Bergisch Gladbach, Germany). The cells were seeded into 25-cm² culture flasks and grown in normal medium (RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, all purchased from GIBCO) at 37°C under 5% CO₂, with the replacement of medium by one-half every 2 to 3 days. 

Isolation of the Peripheral Blood Mononuclear Cells

Approximately 50 mL of peripheral blood was collected into Li-Heparin tubes (Sarstedt, Nümbrecht, Germany) and processed within 3 hours. Peripheral blood was diluted 1:1 (vol/vol) with PBS prewarmed to room temperature and centrifuged at 200g for 15 minutes. The sedimented cells were resuspended in PBS at a total volume of 35 mL/tube, layered carefully onto the 15 mL aliquots of Histopaque-1077 (Sigma-Aldrich, Steinheim, Germany) in 50-ml conical tubes, which were prewarmed to room temperature, and centrifuged at 400g for 30 minutes without acceleration/brake. The upper layer was discarded and the intermediate opaque layer containing the mononuclear blood cells was collected into new 50-ml tubes. The cells were washed three times with Hank's balanced salt solution (GIBCO) at 250g for 10 minutes as instructed and resuspended in PBS with 0.1% BSA and 2 mM EDTA (Sigma-Aldrich) at a final concentration of 0.7 to 1 × 10⁷cells/ml. 

Immunomagnetic Enrichment of CMC

The CMC were isolated using a combination of the NKI/C3 and NKI/beteb antibodies as described ¹⁷ with slight modifications. Briefly, isolated mononuclear blood cells were incubated with a mixture of the mouse antibodies against the melanoma markers NKI/beteb and NKI/C3 (at a final concentration of 1 μg/mL each; Abcam, Cambridge, UK) for 30 minutes under gentle rotation. The cells were washed with 2 mL of 0.1% BSA-PBS per 10⁷ cells and centrifuged at 300g for 10 minutes. The cell pellet was resuspended in 0.1% BSA-PBS at a concentration of 10⁷ cells/mL and incubated with prewashed magnetic immunobeads conjugated with anti-mouse IgGs (CELLection Pan Mouse IgG Kit; DYNAL, Invitrogen) under gentle rotation for 45 minutes. The bead-bound cells were washed three times with 0.1% BSA-PBS and released from the immunobeads by DNase treatment in releasing medium (RPMI-1640 supplemented with 1% fetal bovine serum, 1 mM CaCl₂, and 4 mM MgCl₂) using a magnet as instructed. The pooled suspension of the released cells was centrifuged at 300g for 8 minutes and resuspended in normal medium at a maximal volume of 2 ml. Approximately 90% of the cell volume was processed directly for making cytospins as described below, whereas the rest was partitioned equally into two wells of an 8-well slide (Sarstedt) and grown in normal medium for 2 weeks at 37°C under 5% CO₂ with the replacement of the medium by one-half every 2 to 3 days. 

Immunomagnetic Enrichment of 92.1 Cells Spiked Into Whole Blood

The 92.1 uveal melanoma cell line was grown as described previously. ²³ Confluent cells were detached from the culture flasks by washing with warm PBS and incubation with 0.025% Trypsin-1 mM EDTA (GIBCO) for 2 to 3 minutes at room temperature. After centrifuging at 300g for 8 minutes, the cells were resuspended in PBS. Peripheral blood (60 mL) was collected from a healthy donor (female, 33 years) and divided into 6 equal aliquots, five of which were mixed with 1 mL of the 92.1 cell suspension at concentrations of 1 to 10⁴ cells/mL blood, whereas the remaining aliquot was mixed with 1 mL PBS only as the negative control. After immunoenrichment, the isolated cells from each subgroup were seeded into a 6-well plate and grown in normal medium for 2 weeks. For immunocytochemistry, cells were trypsinized and processed for the preparation of cytospins as described below. 

Immunomagnetic Enrichment of 92.1 Cells Spiked Into Mononuclear Blood Cells

Mononuclear blood cells from a healthy donor (female, 36 years) were isolated from the peripheral blood as described above. 92.1 cells were mixed 1:1 (vol/vol) with the mononuclear blood cells at a final volume of 1 mL and densities corresponding to 2, 10, 100, and 1000 tumor cells/10 mL whole blood. The concentration of the mononuclear blood cells during immunoenrichment was 5 to 7.5 × 10⁶ cells/mL assay buffer. Dual-immunoenrichment was performed immediately after spiking and the isolated cells from each subgroup were directly processed for the preparation of cytospins as described without culturing. 

Cytospins

Before the preparation of the cytospins of freshly isolated CMC or the spiked 92.1 cells, four cytospin chambers were prewet by filling with 50 μL of releasing medium and centrifuging at 135g for 4 minutes in a cytocentrifuge (Tharma, Waldsolms, Germany). The suspension of isolated CMC or 92.1 cells was equally partitioned into the precoated cytospin chambers and centrifuged on to Superfrost Plus slides (Fisher Scientific, Houston, TX, USA) at 135g for 8 minutes. 

The cytospins of 92.1 cells used for analyzing the specificity of markers were prepared by filling the chambers directly with 500 μL of cell suspension (10⁵ cells/cytospin) without prewetting and centrifuging on to Superfrost Plus slides at 135g for 8 minutes. Cytospins of the cultured primary tumor cells (2000–3500 cells/spin) and the aliquots of mononuclear blood cell fraction from control patients (1000–2000 cells/spin) also were prepared without prewetting the cytospin chambers as described above. All cytospins were dried overnight at room temperature and stored at −20°C. 

Immunofluorescence Staining

Frozen cytospins were air-dried at room temperature for 30 minutes. Cytospins and the cultured cells in 8-well slides were fixed for 10 minutes in 2% paraformaldehyde-PBS, washed three times in PBS, and incubated in the sterile-filtered blocking buffer (3% BSA in 10 mM Tris-HCl, pH 7.5, 120 mM KCl, 20 mM NaCl, 5 mM EDTA, 0.1% Triton X-100) for 30 minutes, followed by the primary mouse antibodies against NKI/beteb (1/10 dilution in the blocking buffer; Abcam), NKI/C3 (1/20 dilution; Abcam), MCSP (clone 9.2.27, 1/10 dilution; Abcam), or rabbit antibodies against CD45 (1/50 dilution; Abcam) overnight at 4°C. Cells were washed for three times in PBS and incubated with Alexa 488-conjugated anti-mouse (1:100 in blocking buffer; Molecular Probes, Invitrogen, Eugene, OR, USA) or Cy3-conjugated anti-rabbit antibodies (1:400 dilution; Jackson Immunoresearch, West Grove, PA, USA) for 1 hour at room temperature. Nuclei were counterstained with DAPI (0.5 μg/mL in PBS; Molecular Probes) for 10 minutes. Samples were mounted in Mowiol (Roth, Karlsruhe, Germany) and stored at 4°C. Double-immunostainings were performed following the same procedure by incubating the samples with a mixture of the primary antibodies against CD45 and MCSP or CD45 and NKI/C3 followed by the mixture of the secondary anti-rabbit and anti-mouse antibodies at the concentrations indicated. Samples were visualized under a fluorescence microscope (Leica, Wetzlar, Germany) with the appropriate filter sets (A4, Excitation [Ex] 360/40, Emission [Em] 470/40 nm; L5, Ex 460/40, Em 527/30 nm; Cy3, Ex 545/30, Em 610/75 nm). Images were acquired using a monochrome digital camera (DFC 350 FX; Leica) attached to the microscope and the Leica Application Software (Advanced Fluorescence 2.3.0, build 5131). 

Quantification of marker expression on the leukocytes and primary tumor cells was performed on at least 106 cells per group. Quantification of CMC was performed on cytospins, immunostained for NKI/C3 (n = 31 patients) and MCSP (n = 28 patients). The mean CMC number per patient was calculated from the number of NKI/C3 and MCSP-positive cells for n = 28 of the 31 patients. For the remaining 3 patients for whom the expression of MCSP could not be analyzed, the number of NKI/C3-positive cells in 10 mL blood was used for the statistical analysis. 

Statistical Analysis

The data set was analyzed using the NCSS statistical software (Version 8.0.13; NCSS, LLC, Kaysville, UT, USA) under Windows 7. Association of the number of native and cultured CMC with established clinical prognostic factors was evaluated by a logistic regression analysis, taking the absence of a given condition as the reference (baseline) value for the categorical variables and providing odds ratios (ORs) with the corresponding Wald confidence intervals (CI). Comparison of the number of native CMC cells in the cytospins with the success rate of CMC cultures was performed using a 2-sided t-test, assuming equal variance among the independent samples. All P values were explorative and the P values less than 0.05 were considered significant. 

Results

Expression of Melanoma Markers on Cultured UM Cells and Leukocytes

Primary cultures of UM cells established from three tumors showed expression of all three melanoma markers (NKI/C3, NKI/beteb, and MCSP) on at least 92% of the cells. Immunocytochemical staining of UM cell line 92.1 demonstrated a strong expression of NKI/C3 and NKI/beteb, with MCSP expression being restricted to a smaller fraction of cells and at a considerably lower level, which was consistent with previous findings. ¹⁷ Although the immunostainings performed on leukocytes from control patients (n = 4) demonstrated positive staining of some granulocytes and monocytes with anti-NKI/C3, these cells could be visually classified as leukocytes based on their segmented or indented nuclear morphology, respectively. Over 93% of the leukocytes were negative for the melanoma markers NKI/beteb and MCSP (Figs. 1A–C). Common leukocyte antigen CD45 was detected strongly on granulocytes, lymphocytes and monocytes, but not on UM cells (Fig. 1D). Based on these findings, we considered the three tested markers as being suitable for distinguishing CMC from the surrounding leukocytes. 

Figure 1

 

Specificity of the melanoma markers NKI/C3, NKI/beteb, and MCSP used in this study. Cell line 92.1 showed a high expression of the first two markers, whereas MCSP was restricted to a smaller fraction of cells at considerably lower levels, while three cultures of primary UM expressed all three. The majority of the leukocytes exhibited a negligible immunoreactivity to the melanoma markers NKI/beteb and MCSP (A–C). Granulocytes, lymphocytes (L) and monocytes, but not UM cells, expressed the common leukocyte antigen CD45 (D). For the primary tumor cells and normal leukocytes, representative images obtained from the cells of a UM patient (female, 78 years) who underwent enucleation in our clinic and a control patient (female, 71 years) are presented, respectively. Scale bar: 10 μm. Images were acquired at an original magnification of ×400.

Immunomagnetic Enrichment of 92.1 Cells Spiked Into Blood

To validate the sensitivity of the modified immunomagnetic enrichment assay, 92.1 cells were spiked into aliquots of purified mononuclear blood cells at concentrations corresponding to 2, 10, 100, and 1000 tumor cells/10 mL blood. An additional aliquot of mononuclear blood cells was mixed with the vehicle alone as negative control. Following dual-immunoenrichment, the isolated cells were processed immediately for the preparation of cytospins and immunocytochemistry. This procedure yielded highly accurate and reproducible results in the first two groups, with a minimum recovery rate of 80% (corresponding to 2.4 ± 0.8 tumor cells/10 mL blood in the first group and 9.0 ± 0.9 cells in the second group, mean ± SD of n = 5 independent experiments, Fig. 2A). The recovery rate was reduced to 51.2 ± 20.6% and 38.2 ± 20.5% in the groups spiked at concentrations corresponding to 100 and 1000 cells/10 mL blood, respectively (mean ± SD of n = 5 experiments). No tumor cells were detected in the negative controls. Isolated cells exhibited a well-preserved morphology and a strong expression of the NKI/C3 marker, which enabled their distinction from the surrounding CD45-positive leukocytes (Fig. 2B). 

Figure 2

 

Sensitivity of the dual-immunoenrichment assay as validated by the recovery of spiked 92.1 cells from the mononuclear blood cell fraction (A, B) or whole blood (C, D). (A) Spiking of 92.1 cells into the purified fraction of mononuclear blood cells. Following dual-enrichment, cytospins were made and double-stained for NKI/C3 and CD45. No tumor cells were detected in the subgroups of blood cells spiked with the vehicle alone. Data represent the mean ± SD of five independent experiments. (B) Representative double-immunostainings on the cytospins of the isolated cells (recovered from the subgroup spiked at a density corresponding to 2 tumor cells/10 mL blood), demonstrating an NKI/C3-positive tumor cell (green) adjacent to two CD45-positive monocytes (red). The nuclei were counterstained in blue with DAPI. Scale bar: 10 μm; original magnification: ×400. (C) Light microscopy images of the spiked cells cultured for 2 weeks after isolation from blood by dual-immunoenrichment. Tumor cells isolated from the blood samples (arrows) spiked at the densities of 1 to 100 cells/mL were detected with a high degree of accuracy. No cells could be detected in the blood aliquot that was spiked with the vehicle (PBS) alone. Scale bar: 100 μm; original magnification: ×100. (D) The identity of the recovered cells was confirmed by the NKI/C3-immunostainings on the cytospins of the cultured cells. Scale bar: 10 μm, original magnification: ×400.

To further evaluate the sensitivity of the modified immunomagnetic enrichment assay, and to determine whether we could obtain CMC for culturing, we tested this method on 92.1 cells spiked into the aliquots of peripheral blood from a healthy donor and allowed the cells to recover in culture as recommended ¹⁷ for 2 weeks. We created a serial concentration range of 1 to 10⁴ cells/mL blood and all aliquots were processed in parallel. In the aliquots spiked with 10³ and 10⁴ tumor cells/mL blood, a high number of visible aggregates formed while processing the samples. After density gradient centrifugation, these aggregates sedimented below the interphase layer consisting of the mononuclear blood cells (data not shown) and, therefore, were not collected within the mononuclear cell fraction. 

Our assay did not detect any cells in the blood aliquot that was spiked with the vehicle (PBS) alone, and enabled the recovery of the 92.1 cells spiked at all the densities tested (Fig. 2C). The isolated 92.1 cells were clearly detectable by light microscopy in all the subgroups after culturing for 2 weeks. The detected cell number matched the expected cell number exactly for the group spiked with 1 tumor cell/mL blood, and exceeded the expected cell number by 6.7% in the group spiked with 10 tumor cells/mL blood. Conversely, the detected cell number was 6.7% less than the expected cell number in the group spiked with 100 tumor cells/mL blood. In the subgroups spiked with 1000 and 10,000 cells/mL blood, the isolated cells exhibited a better morphology (Fig. 2C). However, the detected cell number was considerably less than the expected cell number. All the isolated cells were positive for NKI/C3 as demonstrated by the immunostaining performed on cytospins of cultured cells (Fig. 2D). 

Immunomagnetic Enrichment of CMC From UM Patients

As the modified immunomagnetic enrichment assay was developed to determine the presence of CMC in UM patients, we analyzed the blood of 31 UM patients. Intact CMC were isolated from the blood samples of UM patients without clinically-detectable metastases (Table 1; see Supplementary Table S1 for the individual results of each patient). Approximately 90% of the isolated cells from each patient were processed immediately for the preparation of cytospins without further cultivation, whereas the remaining 10% was maintained in culture for 2 weeks. To ascertain the specificity of the staining method, we initially performed double-immunostainings for the leukocyte antigen CD45 and the melanoma markers NKI/C3 or MCSP on the cytospins of the cells isolated from the blood samples from 10 UM patients. The leukocytes could not be completely eliminated after immunoenrichment and constituted the majority of the isolated cell population, as demonstrated by the expression of CD45 on 97.1 ± 2.1% of the cells (mean ± SD of n = 10 patients with a minimum of 110 cells analyzed in each patient). We also clearly identified CD45-negative cells with a strong and uniform expression pattern of NKI/C3 or MCSP, representing the CMC (Figs. 3A, 3B, respectively). Unlike normal leukocytes, these cells possessed a round or oval nucleus together with a larger cytoplasm similar to the dissociated primary UM cells that we had started with. Therefore, we concluded that MCSP is equally as good as NKI/C3 as an individual marker for the identification of CMC after immunomagnetic enrichment. 

Figure 3

 

Distinction of the isolated CMC from the surrounding leukocytes. (A) CMC were isolated from the blood sample of a UM patient (male, 42 years). Cytospins of the isolated cells were processed for the double-immunostaining for the melanoma marker MCSP (green) and the common leukocyte antigen CD45 (red). All leukocyte types, including the activated (*) lymphocytes, exhibited a strong expression of CD45 and a negligible immunoreactivity for MCSP. In contrast, we could detect a few CD45− cells with a strong and uniform expression of MCSP, representing CMC. Arrows indicate the immunobeads. (B) NKI/C3+/CD45− CMC isolated from a UM patient (male, 53 years) compared to a neighboring granulocyte. Despite the moderate to strong immunoreactivity to NKI/C3 (green) on some granulocytes and monocytes, these cells could be identified clearly as leukocytes based on the prevalent CD45 expression and their nuclear morphology. Scale bar: 10 μm. All images were acquired at an original magnification of ×400.

Table 1

 

Association of Established Clinical Prognostic Factors With the Number of CMC (Median, 3.5 cells/10 mL Blood)

Variable	#CMC ≥ Median, n = 16 Patients	#CMC < Median, n = 15 Patients	All Patients, n = 31		OR (95% CI)
Age, y
Median	66.5	60	62	0.16	1.0 (0.9, 1.1)
Min, max	45, 80	25, 79	25, 80
Sex
Female	9	8	17	0.85	1.1 (−0.3, 4.7)
Male	7	7	14
Eye
Right	7	8	15	0.59	0.7 (0.2, 2.8)
Left	9	7	16
LBD, mm
Median	9.1	9.6	9.6	0.96	1.0 (0.9, 1.2)
Min, max	2, 16.9	3.0, 16.5	2, 16.9
>14	5	4	9	0.78	1.3 (0.3, 5.9)
<14	11	11	22
Tumor height, mm
Median	4.5	3.5	3.9	0.14	1.2 (0.9, 1.6)
Min, max	1.4, 12.5	1.0, 11.4	1.0, 12.5
Ciliary body infiltration
Yes	3	3	6	0.93	0.9 (0.2, 5.5)
No	13	12	25
Optic nerve infiltration
Yes	0	1	1	0.98	0.0 (0.0, 10,000+)
No	16	14	30
TNM stage
TNM 1	6	7	13	0.61	0.7 (0.2, 2.9)
TNM 2	3	4	7	0.6	0.6 (0.1, 3.5)
TNM 3	7	4	11	0.32	2.1 (0.5, 9.7)

 Association of CMC number with the clinical prognostic factors was evaluated by logistic regression analysis, with #CMC < median taken as the reference (baseline) value for the categorical dependent variable. The age, LBD, and tumor height were evaluated as numeric independent variables, whereas the remaining parameters were analyzed as categorical independent variables, with the absence of a given condition as the reference value.

Analysis of CMC number was performed on cytospins prepared immediately after the isolation of NKI/C3 and NKI/beteb positive cells. Immunocytochemistry for NKI/C3 or MCSP on the cytospins revealed the presence of CMC expressing at least one of these markers in n = 29 of the 31 melanoma patients examined (93.6%, with 95% CI of 78.6%–99.2%, Fig. 3, Table 1, Supplementary Table S1). The absolute number of CMC detected in these patients varied between 0 and 10.2 cells (median, 3.5 cells) per 10 mL of whole blood. No cells positive for NKI/C3, MCSP, or NKI/beteb were detected in any of the control samples (n = 9). 

For analyzing the viability of the isolated CMC, approximately 10% of the cells isolated from all patients were put into culture and allowed to grow for 2 weeks. Double-immunostainings for NKI/C3 and CD45 were performed to identify the nature of the cultured cells. The CMC were detected in the cultures derived from 15 of the 29 patients (51.7%) whose direct blood samples had been found to be positive (Fig. 4, Supplementary Table S1). Successful culture apparently was dependent on the number of seeded cells, as suggested by the higher median CMC number of these 15 patients (4.1 cells/10 mL blood; range, 0.8–8.4 cells in these positive cases versus 1.5 cells/10 mL blood in the 14 negative cases, P = 0.17). The 15 positive CMC cultures had a purity of 78.8 ± 26.7% (mean ± SD of the NKI/C3+/CD45− cell ratio; median, 87.5%; Supplementary Table S1). The majority of the cultured CMC exhibited a healthy morphology with a well-established adhesion as opposed to the NKI/C3−/CD45+ cells, which mostly appeared contracted (Fig. 4). No CMC were detected in the cultured samples of the remaining two patients whose cytospins had been negative for these cells and in any of the age-matched controls (n = 9). 

Figure 4

 

Culturing of the isolated CMC for two weeks. Cultured CMC strongly expressing NKI/C3 and exhibiting no immunoreactivity for the leukocyte marker CD45 were detected in the samples of approximately 52% (n = 15 of 29) of the patients, whose cytospin samples of freshly isolated cells were positive for CMC. Most of the cultured CMC exhibited a healthy morphology with a well-defined adhesion. Scale bar: 25 μm. In contrast, the majority of the leukocytes appeared significantly contracted. Images of a leukocyte detected in the cultured sample of a 78-year-old UM patient (female) were included as controls for the NKI/C3 and CD45 immunostainings. Scale bar: 10 μm. All images were acquired at an original magnification of ×400.

Analysis of the Presence and Number of CMC With Established Markers for Tumor Progression

No significant association was detected between the presence or the number of uncultured CMC and the established prognostic factors analyzed; including the age, sex, affected eye, largest basal diameter (LBD), tumor height, ciliary body infiltration, optic nerve infiltration, and TNM Stage (Table 1, Supplementary Table S1). Optic nerve infiltration was detected in one of the two patients negative for CMC. The median tumor height tended to be higher and exceeded 4 mm in the patients carrying a higher number of CMC (≥ median value of n = 3.5 CMC/10 mL blood, n = 16 of 31 patients, P = 0.14, Table 1). When comparing clinical characteristics between patients with positive versus negative culture results, we did not observe any significant association between the success rate of the cultures and the clinical prognostic factors analyzed (Table 2). 

Table 2

 

Association of Established Clinical Prognostic Factors With the Success of CMC Cultures

Variable	CMC Detected in Culture, n = 15 Patients	No CMC in Culture, n = 14 Patients	All Patients With Native CMC, n = 29		OR (95% CI)
Age, y
Median	62	62	62	0.63	1.0 (0.9, 1.0)
Min, max	25, 80	28, 78	25, 80
Sex
Female	9	7	16	0.36	2.0 (0.5, 8.8)
Male	6	7	13
Eye
Right	7	7	14	0.86	0.9 (0.2, 3.8)
Left	8	7	15
LBD, mm					1.0 (0.9, 1.2)
Median	9.6	9.1	9.6	0.88
Min, max	3.5, 16.9	2.0, 16.5	2, 16.9		0.7 (0.1, 3.2)
>14	4	5	9	0.59
<14	11	9	20
Tumor height, mm
Median	3.7	4.1	3.9	0.78	1.0 (0.8, 1.3)
Min, max	1.4, 12.0	1.0, 12.5	1.0, 12.5
Ciliary body infiltration
Yes	4	2	6	0.42	2.2 (0.3, 14.4)
No	11	12	23
TNM stage
TNM 1	7	5	12	0.55	1.6 (0.4, 7.0)
TNM 2	4	3	7	0.74	1.3 (0.2, 7.4)
TNM 3	4	6	10	0.36	0.5 (0.1, 2.3)

 NOTE: Association of cultured CMC with the clinical prognostic factors was evaluated by logistic regression analysis, with the absence of CMC in culture taken as the reference (baseline) value for the categorical dependent variable. The age, LBD, and tumor height were evaluated as numeric independent variables, whereas the remaining parameters were analyzed as categorical independent variables, with the absence of a given condition as the reference value. Optic nerve infiltration was not detected in any of the patients.

Discussion

In the present study, we used a novel dual-immunomagnetic enrichment assay using the combination of two antibodies (NKI/C3, NKI/beteb) to isolate intact CMC from patients with primary UM. This method initially was described by Cools-Lartigue et al. ¹⁷ with a detection sensitivity of 10 spiked UM cells per mL blood. In our study, we slightly modified this assay by incubating the mononuclear blood cell fraction initially with a cocktail of the NKI/C3 and NKI/beteb antibodies followed by the immunobeads rather than directly using prelabeled immunobeads. This indirect labeling approach was recommended by the manufacturer to increase the likelihood of cell capture when the target cells are expected to be present at a very low density. To determine the sensitivity of this modified assay, we spiked mononuclear blood cells with 92.1 cells and processed the isolated cells for immunostainings on cytospin. This approach allowed a highly accurate and reproducible recovery of the 92.1 cells spiked at very low concentrations corresponding to 2 and 10 tumor cells/10 mL blood. However, the recovery rate underwent a gradual reduction as the concentration of the spiked cells was increased to 100 and 1000 cells/10 mL blood. These findings suggested that the concentration of the antibodies and/or immunobeads may need to be increased for the isolation of target cells at such high densities. 

We also added the 92.1 cells to blood at higher densities and after isolation, cultured the cells for two weeks. In contrast to the direct analysis of the isolated cells, we were able to detect the 92.1 cells spiked at a concentration range of 10 to 100 cells/mL blood with a very high accuracy after culturing the isolated cells for two weeks. A possible reason underlying this observation might be the regeneration of cell surface proteins during the culture period, which would promote better cell adhesion and prevent cell loss compared to the direct analysis after immunoenrichment. However, we could not exclude the possibility of cell proliferation during the culture period particularly in the groups with a higher number of seeded cells, where the paracrine signals from the neighboring cells also might have promoted cell growth. Cells spiked at the higher densities of 10³ and 10⁴ cells/mL blood also could be recovered efficiently and were readily discernible already after one day in culture. However, the rate of recovery was considerably lower than the expected yield, suggesting further that the concentration of the antibodies and/or immunobeads may not be sufficient to isolate the target cells present at such high numbers. Interestingly, we also observed the formation of large aggregates after spiking normal blood with 92.1 cells at these high densities, which also might have resulted in the loss of a significant proportion of the spiked tumor cells during the isolation of the mononuclear blood cell fraction. Nevertheless, these results demonstrated the sensitivity of the assay for the isolation of tumor cells at physiologically relevant concentrations and the ability to characterize the isolated cells without culturing. Therefore, we considered the assay conditions as being optimal for the screening of CMC in patients. 

Using this assay, we were able to achieve a higher detection rate of intact CMC compared to our former study, amounting to 93.6% compared to less than 20%. ¹³ The recovery yield also was considerably higher in this study, with a median CMC number of 3.5 cells per 10 mL blood, possibly due to the increased likelihood of cell capture by the simultaneous targeting of two marker epitopes under continuous rotation. In contrast to our former study, in which we demonstrated a significant association between the presence of intact CMC and largest tumor diameter, ¹³ this was not the case in this study. We could not detect a significant correlation between the presence or number of CMC and the other established clinical prognostic factors, either, possibly due to the detection of CMC in the majority of the patients. 

Our results also were similar to the findings of a previous RT-PCR–based work, in which CMC transcripts were detected in 96.7% (n = 29 of 30) of the nonmetastatic UM patients. The detection rate in this earlier RT-PCR based study indicates the presence of a minimum of one or two intact CMC in 20 mL blood. ¹² However, we identified a higher number of intact CMC in this study, which was similar to the CMC density detected by the CellSearch system (median, 3 CMC/7.5 mL blood; range, 1–20 CMC) in 30% (n = 12 of 40) of the metastatic UM patients analyzed. ¹⁴ In the CellSearch study, 67.5% (n = 27 of 40) of the patients were at an age of 60 years or above, whereas in our study, this ratio was 58.1% (n = 18 of 31 patients). All the patients in the CellSearch study had liver metastasis, with 20% (n = 8 of 40) of the patients exhibiting other metastatic sites as well. ¹⁴ In contrast, our study included patients without clinical evidence of metastases. A possible reason underlying the difference in CMC number detected in our study versus the RT-PCR–based study might be the markers selected for the identification of CMC. In our study, CMC were identified based on the expression of melanoma markers NKI/C3 or MCSP at the protein instead of the mRNA level. The mRNA profile of UM cells disseminated into blood might, indeed, undergo some alterations as the cells try to survive in the circulation. This is suggested by an in vivo study in which an immune-suppressed rabbit was xenografted with UM cells in the eye. The CMC detected in this animal exhibited differences in the mRNA expression pattern, including a significant reduction in the levels of the Melan-A transcripts, compared to the primary tumor cells in the eye. ¹⁶  

As lack of cells was no longer the limiting factor, the molecular characteristics of CMC rather than the presence of these cells alone should be analyzed and this will provide more information on the risk of metastasis and the tumor cells sensitivity to, for example, drugs. ⁴ The improved immunomagnetic enrichment assay, therefore, is a valuable tool for obtaining intact and viable CMC, which can be analyzed further for the presence of established prognostic factors, such as monosomy of chromosome 3, or other mutations that can influence the prognosis. ¹ All the patients enrolled in this study within a time frame of two years were diagnosed with primary UM with no evidence of metastasis. It would be of particular interest to conduct long-term follow-up studies to evaluate the course of disease along with the prevalence and characteristics of CMC in these patients. 

In summary, we demonstrated here the sensitivity of an improved immunomagnetic cell enrichment approach, which allowed the isolation of viable CMC from the majority of UM patients. This major step in the acquisition of viable CMC would, in turn, enable further molecular characterizations, which would be indispensable to unravel the mechanisms that provide some of these intermediary cells with a higher metastatic potential and to develop early therapies against metastatic disease. This cost-effective method, which does not necessitate specialized equipment, also is expected to serve as a valuable tool for monitoring the course of disease and evaluating the efficacy of treatments. 

Supplementary Materials

docx S1 

Acknowledgments

The authors thank Angelika Tyzak and Regina Kupfer for the collection of blood samples. 

Supported by the internal research fund of the University of Lübeck (FC 20222). The authors alone are responsible for the content and writing of the paper. 

Disclosure: A. Tura, None; J. Lüke, None; H. Merz, None; M. Reinsberg, None; M. Lüke, None; M.J. Jager, None; S. Grisanti, None 

Footnotes

 AT and JL contributed equally to the work presented here and should therefore be considered as equivalent authors.