Limited genomic heterogeneity of circulating melanoma cells in advanced stage patients

The melanoma cell lines WM278, 1617 and 1789 were purchased from the Wistar Institute Collection at the Coriell Institute for Medical Research, Camden, NJ. They were maintained in melanoma growth medium (Tu2%), consisting of four parts of MCDB153 (Sigma-Aldrich, Saint Louis, MO) and one part of L-15 (Invitrogen, Carlsbad, CA), supplemented with 1.68 mmol L⁻¹ calcium chloride, 5 μg ml⁻¹ bovine insulin and 2% fetal bovine serum (Invitrogen). The culture medium was changed every 2 days. To determine expression of markers, melanoma cells were spiked into PBMCs at 1:100 and plated as a monolayer onto custom made cell-adhesion glass slides. To evaluate sensitivity while mimicking patient samples, 0, 10, 50, 100 and 500 melanoma cells were spiked into three million PBMCs each. The experiment was repeated three times using each cell line to validate reproducibility.

The CSPG4-specific mAb 149.53, 763.74, TP61.52, VF1-TP41.2, VF4-TP108, VF4-TP109.2, and VT80.12 were developed and characterized as described [23, 24, 27]. mAb were purified from ascitic fluid by sequential precipitation with ammonium sulfate and caprylic acid [24]. The purity and activity of mAb preparations were assessed by SDS-PAGE analysis, and by testing with CSPG4-bearing melanoma cells in a binding assay. HMB-45-specific mAb clone gp-100 was purchased from Dako, Carpinteria, CA.

The optimal CSPG4 staining was performed as follows: after the slides (containing melanoma cell lines spiked into PBMCs or patient blood samples) were thawed and fixed with 2% neutral buffered formalin solution (VWR, San Dimas, CA), non-specific binding sites were blocked with 10% goat serum (Millipore, Billerica, MA). Slides were subsequently incubated with CSPG4-specific mAbs (5 μg ml⁻¹ total concentration) and Alexa Fluor^® 647 pre-conjugated anti-CD45 antibody (MCA87A647, AbD serotec, Raleigh, NC) for 40 min at 37 °C. Slides were then incubated with Alexa Fluor^® 555 goat anti-mouse IgG1 antibody (A21127, invitrogen) for 20 min at 37 °C. Cells were counterstained with Hoechst 33 258 and mounted with an aqueous mounting media. The CSPG4/HMB-45 staining protocol includes minor modifications. After incubation with Alexa Fluor^® 555 goat anti-mouse IgG1 antibody (A21121, invitrogen), cells were permeabilized using cold methanol for 5 min at RT. Then, cells were incubated with HMB-45-specific mAb (1 μg ml⁻¹) for 40 min at 37 °C and subsequently with Alexa Fluor^® 488 goat anti-mouse IgG1 antibody.

After CSPG4 staining, all nucleated cells in the specimen were imaged as previously reported [28]. Identified candidate cells were evaluated by direct visualization to eliminate possible false positives such as dye aggregates and classified as CMCs based on CSPG4 expression and morphometric parameters. Other cells related to CMCs but lacking essential features were also tracked and classified. The concentration of CMCs ml⁻¹ is calculated by counting the total number of nucleated cells on the glass slide used and comparing it to the PBMC count in the patient's blood specimen. For this reason, fractional values of CMCs ml⁻¹ are possible.

Peripheral blood (8 ml) was collected from 40 metastatic melanoma patients and 10 normal blood donor (NBD) into anti-coagulated Cell-Free DNA BCT^® tubes (Streck, Omaha, NE). Samples from non-local sites (Comprehensive Cancer Centers of Nevada, Las Vegas, NV) were shipped overnight in temperature stabilized containers and processed within 24 h of the blood draw. Samples from local sites (Pacific Oncology and Hematology, Encinitas, CA) were held at RT for 24 h to mimic samples coming from non-local sites. NBD specimens were collected at The Scripps Research Institute's Normal Blood Donor Service. The analysis of the samples was conducted with no previous knowledge of patient's disease status.

Whole blood specimens were prepared as previously described [28]. Briefly, blood samples were subjected to erythrocyte lysis in isotonic NH₄Cl (ammonium chloride) solution. Nucleated cells were re-suspended in PBS (4 × 10⁶ cells ml⁻¹) and a volume of 750 μl was plated as a monolayer on custom made cell-adhesion glass slides (Marienfeld, Lauda-Königshofen, Germany) up to 16 slides. Slides were incubated at 37 °C for 40 min and then stored at −80 °C for future evaluation. For CMC detection in melanoma patients, four slides holding approximately twelve million cells were used as a test.

All patients were prospectively enrolled according to IRB approved protocols. Clinical data were retrospectively collected and are shown in table 1. All patients had metastatic disease, as determined by radiological and clinical criteria at the time of blood draw. Thirty patients were stage IV and ten patients were stage IIIC. BRAF mutational status was assessed in 25 patients and 11 had the V600E mutation. Twenty-one patients with stage IV and one with stage IIIC had died by the time of CMC analysis, 3 were in complete remission, 15 were alive with disease and one was lost to follow-up. Although melanoma has a poor prognosis with a median survival of 6–9 months [29], the follow-up period for this study ranged from 0.08 to 22 months. Patients received different therapies including chemotherapy (cisplatin, carboplatin), immunotherapy (anti-CTLA4 mAb) and targeted therapy (BRAF-MEK inhibitors).

CMCs were relocated on the glass slide and reimaged at 40× for detailed cytomorphometric analysis as previously described [28]. CMC were then captured by micromanipulation and whole genome amplification and CNV analysis were performed as previously described [22].

Relative CSPG4 intensity measured in cell lines was compared by one-way ANOVA with a Mann–Whitney correction test using GraphPad Prism version 5.04 (GraphPad Software, San Diego, CA). A receiver operating characteristic (ROC) curve was generated using clinical outcome for distinguishing high risk from low risk patients. These are reported in the results as the CMC ml⁻¹ value with the highest sensitivity and specificity with 95% confidence intervals, and was calculated using GraphPad Prism 5.04.

The WM1617, WM278 and WM1789 melanoma cell lines which represent the radial growth phase, the vertical growth phase and metastasis of melanoma progression, respectively, were used to optimize the HDSCA methodology. To determine CSPG4 expression, melanoma cell lines were spiked into PBMCs at 1:100 and plated as a monolayer onto custom made cell-adhesion glass slides. CSPG4 protein expression was detected on the three melanoma cell line utilizing the mAbs 149.53, 763.74, TP61.5, VF1-TP41.2, VF4-TP108, VF4-TP109.2, and VT80.12, that recognize distinct and spatially distant CSPG4 epitopes (figure 1). Relative CSPG4 signal expression using mAb VT80.12 and TP61.5 was on average between 1.6 and 7 times higher than that obtained using mAb VF4-TP108 and 149.53. The staining yielded by the combination of all seven CSPG4-specific mAbs had a higher intensity than that obtained with each individual mAb at the same concentration (figure 1). PBMCs from a NBD were used to normalize signal intensities. Based on immunofluorescent parameters, an intact CMC was defined as: CSPG4 positive, CD45 negative, with an intact, non-apoptotic nucleus detected by Hoechst 33 258 imaging. Positivity for CSPG4 was defined as the fluorescent signal being at least 2-fold higher than the background signal of surrounding PBMCs, which was determined by measuring the CSPG4 signal in 1.7 × 10⁶ PBMCs. Negativity of CD45 was defined as having signal below visual detection under the condition that >99% of all surrounding PBMCs were detectable, as described for the HDSCA methodology.

Zoom In Zoom Out Reset image size

To evaluate CMC assay linearity and reproducibility using the CSPG4 antibody cocktail while mimicking patient samples, 0, 10, 50, 100 and 500 melanoma cells were spiked into approximately 3 × 10⁶ PBMCs from a NBD. As displayed in supplementary figure S1 (stacks.iop.org/PB/12/016008/mmedia), the number of WM1617, WM278 and WM1789 cells detected is plotted against expected cells. A percentage of detection of 97.0, 98.3 and 97.3 with a linear detection coefficient of 0.99, 0.99 and 0.97 was obtained in WM1617, WM278 and WM1789 cells, respectively.

To assess the specificity of the assay using the combination of all 7 CSPG4-specific mAbs, 10 blood samples from normal donors and 40 from melanoma patients were evaluated. In addition to the immunofluorescent criteria described above, a morphometric analysis of candidate cells was performed to determine quantitative differences that may functionally contribute to improve the criteria for a cell to be defined as a CMC in patient samples. Relative nuclear size (RNS) was found to be a crucial informative parameter (figure 2). Approximately, 70% of candidate cells (517) detected in melanoma patients had a RNS smaller than 2.5 with an average of 1.5 (quadrant a), and had a relative CSPG4 intensity range of 2 to 289-fold brighter (top left cells) than surrounding PBMCs (mean 47.2 ± 55.1). The remaining 30% (224 cells) had a RNS up to 13-fold larger than the surrounding PBMCs with an average close to 5 (quadrant b, gray dots), and a relative CSPG4 intensity range of 2 to 8-fold brighter than PBMCs (top right cell). The results obtained from NBD samples (quadrant b, blue dots) showed that 100% of candidate cells had a RNS of at least 2.5-fold larger than surrounding PBMCs, with a relative CSPG4 signal lower than 8 (bottom right cell). Based on this evaluation, and assuming that normal blood cells may exist among candidate cells in melanoma patient samples, we defined an exclusion criterion for those cells near the lower limit of CSPG4 intensity. Thus, cells from melanoma patients that were CD45 negative, had a relative CSPG4 intensity below 8, an intact nucleus and a RNS of 2.5 or larger (figure 2, top right cell) were excluded.

Zoom In Zoom Out Reset image size

A comparison of the immunocytochemical features between melanoma cell lines (quadrant a, orange dots) and CMCs from melanoma patients revealed significant differences, including a relative CSPG4 intensity mean 12 times greater in cell lines than in CMCs. Moreover, these cell lines displayed a wider RNS range (up to 10 times larger than PBMCs), with an average of 2.4.

Between 1 and 250 CMCs in 22 of 40 (55%) metastatic melanoma patients were detected (table 2). The number of CMCs ranged between 0.5 and 371.5 ml⁻¹ (mean 14.9 CMCs ml⁻¹), and no CMCs were detected in NBDs (figure 3(a)). Nineteen (47.5%) of the positive patients had ≥1 CMCs ml⁻¹ and 2 (5%) ≥ 100 CMC ml⁻¹. The CSPG4 signal intensity within the CMC population varied within and across patients (figure 3(b)). Four patients had only CSPG4 bright (from 8 to 389-fold brighter than the surrounding PBMCs) CMCs, 9 only CSPG4 dim (from 2 to 8-fold brighter than the surrounding PBMCs) CMCs, and 9 both CSPG4 bright and CSPG4 dim CMCs. Figure 3(c) shows the cytomorphology and immunophenotype of four representative CMCs from two melanoma patients (#30 and #37). In order to evaluate the morphometric heterogeneity of CMCs, we analyzed the cells found in the two patients (#30 and #37) with more than 100 CMC ml⁻¹ (figures 3(d) and (e)). CMC shapes varied within and between these two patients. For example, most of the cells from patient 30 were highly pleomorphic in shape and presented polygonal nuclei. Cell roundness mean was 1.01 ± 0.01 on average for patient PBMCs, and 1.11 ± 0.12 and 1.05 ± 0.02 for CMCs in patient 30 and 37, respectively, indicating that nuclei from both the PBMCs and the HD-CMS were essentially round with the CMCs tending very slightly toward oblong. Diameter plots indicate that CMCs in melanoma patient 30 and 37 were significantly larger than their corresponding PBMCs (diameter: 12.6 ± 2.3 versus 8.9 ± 0.5 μm, P < 0.0001 and 11.8 ± 1.2 versus 8.9 ± 0.6 μm, P < 0.0001, respectively). In most patients, only single cells were detected, except for seven patients (#5, 13, 17, 18, 30, 35 and 37) containing two-cell to six-cell clusters.

Zoom In Zoom Out Reset image size

Figure 4(a) shows the distribution of 124 CMCs found in 40 melanoma patients that were either CSPG4⁺/HMB-45⁻ (left) or CSPG4⁺/ HMB-45⁺ (right). Sixty-one (49%) of 124 cells were CSPG4⁺/HMB-45⁻, while the remaining 68 cells (51%) were CSPG4⁺/HMB-45⁺. To evaluate HMB-45 signal heterogeneity of CMCs within and across patients, we analyzed CMCs found in the two melanoma patients (#30 and #37) with the highest number of cells available for assay. In patient #30, 58 cells were reviewed. In general, CMCs from this patient had a high relative CSPG4 intensity (mean 62.9). Thirty-three (57%) of the CMCs were positive for both CSPG4 and HMB-45 (relative HMB-45 intensity mean was 20.7). In patient #37, 37 cells were evaluated. In this case, 24 CMCs (68%) were positive for both CSPG4 (relative CSPG4 intensity mean 16.3) and HMB-45 (relative HMB-45 intensity mean 34.6) markers. Despite the low sensitivity of the HMB-45 marker observed in our assay, its high specificity supported the inclusion of those CMCs with low CSPG4 expression, especially in patient #37 in whom 26% of CMCs had a relative CSPG4 intensity signal close to the lower limit cutoff (figure 4(b)).

Zoom In Zoom Out Reset image size

DNA CNV analyses were assessed in single CMCs, WBCs and 'excluded candidate cells' isolated from melanoma patient #30 (40 cells) and #37 (23 cells) (figure 5). Chromosomal alterations were found in 100% of the CMCs analyzed. A unique clonal population (38 CMCs) in patient #30 and two clonal populations (18 CMCs) in patient #37 were observed. Chromosomal gains and deletions of chr5, 7, 9, 10, 12, 17, and 19 were detected in both patients. Candidate genes encoding components of commonly altered pathways in melanoma were located at these amplified/deleted areas. The amplification of mixed-lineage leukemia 3 (MLL3) in chromosome 7, an important histone regulator gene, and the loss of cyclin-dependent kinase inhibitor 2A (CDKN2A), a tumor suppressor gene that regulates the pRB and p53 pathways [30, 31], were found to be present in both patients, along with an increase of a segment on chromosome 5p containing telomerase reverse transcriptase (TERT) locus, which encodes the catalytic protein subunit of the telomerase [31]. The loss of PTEN, responsible for the negative regulation of the PI3K/AKT pathway [32] was found only in patient #30. In patient #37, two CMC populations (clone A and B) were identified (figure 5(b)). Mouse double minute 2 homolog (MDM2), an important negative regulator of the p53 tumor suppressor [33], was amplified in all CMCs from both clones and have more than 20 copies each. Amplification of BRAF [34], which regulates the MAPK signaling pathway, was identified in all CMCs from clone A. Kirsten rat sarcoma viral oncogene homolog (KRAS), involved primarily in regulating cell division, was only amplified in clone B. No chromosomal alterations were detected in the WBCs or 'excluded candidate cells' (supplementary figure S2 (stacks.iop.org/PB/12/016008/mmedia)).

Zoom In Zoom Out Reset image size

The number of patients in this study (n = 40) was not powered for survival analysis nor was the sampling of blood controlled for a specific line of therapy. Nevertheless, there was an association between the number of CMC ml⁻¹ of blood and the short survival observed in some patients (table 2). A ROC curve was constructed using the clinical outcome from melanoma patients (n = 39). A cutoff value of 8.7 CMC ml⁻¹ was determined from this cohort data. The mean overall survival time for patients with <8.7 CMC ml⁻¹ which was 315.9 days was significantly longer than that for those with ≥ 8.7 CMC ml⁻¹, which was 18 days.