- Technical advance
- Open Access
- Open Peer Review
High-recovery visual identification and single-cell retrieval of circulating tumor cells for genomic analysis using a dual-technology platform integrated with automated immunofluorescence staining
- Daniel E Campton†1,
- Arturo B Ramirez†1,
- Joshua J Nordberg1,
- Nick Drovetto1,
- Alisa C Clein6,
- Paulina Varshavskaya1,
- Barry H Friemel1,
- Steve Quarre1,
- Amy Breman2,
- Michael Dorschner3,
- Sibel Blau4,
- C Anthony Blau5,
- Daniel E Sabath6,
- Jackie L Stilwell1 and
- Eric P Kaldjian1Email author
© Campton et al.; licensee BioMed Central. 2015
- Received: 1 October 2014
- Accepted: 28 April 2015
- Published: 6 May 2015
Circulating tumor cells (CTCs) are malignant cells that have migrated from solid cancers into the blood, where they are typically present in rare numbers. There is great interest in using CTCs to monitor response to therapies, to identify clinically actionable biomarkers, and to provide a non-invasive window on the molecular state of a tumor. Here we characterize the performance of the AccuCyte® – CyteFinder® system, a comprehensive, reproducible and highly sensitive platform for collecting, identifying and retrieving individual CTCs from microscopic slides for molecular analysis after automated immunofluorescence staining for epithelial markers.
All experiments employed a density-based cell separation apparatus (AccuCyte) to separate nucleated cells from the blood and transfer them to microscopic slides. After staining, the slides were imaged using a digital scanning microscope (CyteFinder). Precisely counted model CTCs (mCTCs) from four cancer cell lines were spiked into whole blood to determine recovery rates. Individual mCTCs were removed from slides using a single-cell retrieval device (CytePicker™) for whole genome amplification and subsequent analysis by PCR and Sanger sequencing, whole exome sequencing, or array-based comparative genomic hybridization. Clinical CTCs were evaluated in blood samples from patients with different cancers in comparison with the CellSearch® system.
AccuCyte – CyteFinder presented high-resolution images that allowed identification of mCTCs by morphologic and phenotypic features. Spike-in mCTC recoveries were between 90 and 91%. More than 80% of single-digit spike-in mCTCs were identified and even a single cell in 7.5 mL could be found. Analysis of single SKBR3 mCTCs identified presence of a known TP53 mutation by both PCR and whole exome sequencing, and confirmed the reported karyotype of this cell line. Patient sample CTC counts matched or exceeded CellSearch CTC counts in a small feasibility cohort.
The AccuCyte – CyteFinder system is a comprehensive and sensitive platform for identification and characterization of CTCs that has been applied to the assessment of CTCs in cancer patient samples as well as the isolation of single cells for genomic analysis. It thus enables accurate non-invasive monitoring of CTCs and evolving cancer biology for personalized, molecularly-guided cancer treatment.
- Circulate Tumor Cell
- Buffy Coat
- SKBR3 Cell
- Whole Genome Amplification
Cancer metastasis accounts for 90% of cancer deaths . Circulating tumor cells (CTC) are malignant cells that migrate from a cancer into the bloodstream; most CTCs die, but some exit the circulation to develop into metastases . High numbers of CTC are associated with shorter overall and progression free survival [3-5]. CTCs, however, are rare – it is typical for one CTC to be present for every million white blood cells or more – and thus detecting and measuring CTC requires highly sensitive technology.
Platforms for CTC identification have been developed based on size, protein expression, or other physical characteristics (reviewed in ). Currently, the only FDA-cleared platform for CTC enumeration is the CellSearch® system (Veridex, Raritan, NJ, USA), and is used for monitoring CTC in patients with colorectal, breast, and prostate cancer. This system is based on automated immuno-magnetic capture of EpCAM expressing cells, followed by staining for DNA and cytokeratin to verify that captured cells are nucleated and epithelial in origin. An exclusionary stain for CD45 is included to prevent false positive identification of white blood cells that may be non-specifically captured. False negatives are an acknowledged weakness of immuno-magnetic capture, which will not identify CTCs that express low levels of the capture antigen. Other technologies for CTC analysis currently under development include other immunomagnetic positive or negative selection methods, microfluidic chips, filters, isolation based on cell deformability or cell density, and dielectrophoretic separation. Although there are advantages to each technology, there are also limitations. Microfluidic chips and filters that fractionate by size will not capture small CTCs. Most technologies do not provide high-resolution visualization of cells. Often sensitive technologies are not specific, and vice versa. Some require red blood cell lysis, which may damage cells. Finally, the ability to robustly retrieve individually identified cells within a practical workflow remains elusive.
The use of information from CTCs for therapeutic decision-making is in its infancy. There is great interest in exploiting CTCs as a window on the molecular state of a tumor, since understanding the evolutionary path of a cancer may predict resistance before overt clinical progression, potentially allowing for the pre-emptive selection of a more effective therapy. An ideal CTC analysis platform would provide unambiguous morphology for definitive CTC identification, comprehensive CTC enumeration for monitoring a patient’s response to therapy, flexible characterization of biomarkers (including drug targets), and also enable isolation of CTCs for molecular analyses.
We characterize here the performance of the AccuCyte® – CyteFinder® system: a comprehensive, reproducible and highly sensitive dual-technology platform for collecting, identifying and analyzing CTCs, that employs two complementary technologies that surround a staining step using an automated immunohistochemistry instrument. The AccuCyte system – “front end” – is based fundamentally on the density of CTCs, which is within the range of the buffy coat. However, it is differentiated from existing density-based methods that separate the buffy coat from red blood cells and plasma by use of a unique separation tube and collector device, which allows virtually complete harvesting of the buffy coat into a small volume for application to a microscopic slide without cell lysis or wash steps, a potential source of CTC loss. The CyteFinder system – “back end” – is an automated scanning digital microscope and image analysis system that presents high-resolution images of candidate cells stained with well-characterized markers before definitive classification as a CTC. CyteFinder includes an integrated device (CytePicker™) for CTC retrieval that is mechanically precise and compatible with recently developed advanced genomic analysis methods for single CTCs.
Blood sample collection for spike-in experiments
Blood samples were collected from healthy volunteers at Rainier Clinical Research Center according to a protocol approved by Quorum Review institutional review board (IRB, Seattle, WA, USA). Approximately 40 mL was collected from healthy volunteers into anticoagulant EDTA Vacutainer® tubes (Becton-Dickinson) with a proprietary preservative (RareCyte, Seattle, WA, USA) and 20 mL was collected from cancer patients into CellSave® tubes (Veridex, Raritan, NJ, USA).
Tissue culture cells and model CTC (mCTC) spike-in experiments
LNCaP and PC3 (Prostate), A549 (lung), and MCF7 and SKBR3 (breast) cancer cell lines used as model CTCs (mCTC) were all obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). LNCaP, PC3, SKBR3, and A549 cell lines were maintained in RPMI 1650 medium and MCF7 cell lines were maintained in DMEM medium. Media were supplemented with 10% FBS.
For percent recovery determination, nuclei or mitochondria of live mCTCs were fluorescently labeled with Hoechst 33342 or Mitotracker Red (Life Technologies), respectively, and drawn into a glass capillary tube (VitroTube, Mountain Lakes, NJ, USA). The cells within the VitroTube were then scanned and counted using a DeltaVision fluorescent microscope (GE, Issaquah, WA, Additional file 1: Figure S1). Cells were expelled into 7.5 mL of blood by flushing the VitroTube with PBS and then rescanning the tube for cells that were not expelled to obtain the net precise count of the cells added to the blood. On the order of 100 cells (range ~70 – 200) from each mCTC cell line were spiked into 5 different blood samples and then the sample was processed as described in the next section.
For low mCTC detection experiments, freshly prepared Hoechst 33342 labeled PC3 cells were suspended at approximately 10,000 cells per mL and then pipetted into a well of a multi-chambered glass slide that allowed cells to remain in solution. The chambered slide was then imaged on the CyteFinder® fluorescent microscope (RareCyte, described below). Individual PC3 cells were drawn into a ceramic-tipped needle using the integrated CytePicker™ (RareCyte, described below) and deposited into a PCR tube. The contents of the PCR tube were then transferred into a blood sample by washing with PBS. Alternatively, the contents of the CytePicker needle were deposited into a separate sorting well on the chambered slide. The sorting well was then imaged to determine an accurate count of the number of PC3 cells deposited and the contents of the well were washed into a blood sample with PBS. From 1 cell to 6 cells were spiked into 7.5 ml blood samples.
Density enrichment and adherence of buffy coat to slides
After centrifugation the Separation Tube was removed from the centrifuge adaptors and placed into a CyteSealer® (RareCyte), which applies a brass ring clamp (CyteSeal) around the circumference of the tube at a position on the float below the buffy coat layer, to create a barrier seal between the tube and the float. After the seal was applied, the plasma was aspirated from the top of the float and approximately 4 ml of 1.793 gm/mL high-density retrieval (HDR) fluid was added to the tube. A collection device (EpiCollector®, RareCyte) was placed into the top of the Separation Tube. The EpiCollector has an inverted funnel that tapers to a 16 gauge needle oriented upwards. Excess HDR fluid was expelled from the needle as the EpiCollector was inserted, eliminating dead space within the EpiCollector. A Transfer Tube pre-filled with approximately 250 uL of HDR fluid was placed into the EpiCollector; the Transfer Tube has a rubber septum at its base that is pierced by the needle within the EpiCollector. The Separation Tube with inserted EpiCollector and Transfer Tube was centrifuged for 5 minutes at 500 RCF (Beckman Allegra® X-15R) resulting in the buoyant displacement of the buffy coat from the float into the Transfer Tube. The workflow is summarized in Additional file 2: Figure S2.
Adherence Solution (1000 ul, RareCyte) was added to the buffy coat in the collection tube and mixed. The sample was spread onto 8 SuperFrost® Plus slides (VWR) by pipetting 150 uL of the mixture onto a slide resting in a manual spreading device (CyteSpreader®, RareCyte) that was designed to evenly distribute the sample in a monolayer across a defined region of the slide without making contact with the slide and thus minimizing sample loss (see Figure 2).
Slides were dried for 30 minutes, fixed in 10% Neutral buffered formalin (NBF, Sigma Aldrich) for 1 hour, washed in PBS for 1 minute, and then incubated with 1 M Tris–HCl 10 minutes to neutralize the NBF. Slides were washed twice more with PBS and then stained using the Discovery Ultra automated slide staining system (Ventana Medical Systems, Tucson, Arizona, USA). Antigen retrieval was performed by heating the slides for 8 minutes at 90°C using buffer CC1. Slides were incubated with antibody to EpCAM (SPM491, Spring Bioscience, Pleasanton, CA, USA) diluted 1:100 for 32 minutes in a solution containing 2% goat serum and 2% BSA. Slides with A549 cells spiked into blood were incubated with EGFR antibody (Invitrogen, clone 31G7) at 1:100 in place of EpCAM. Goat anti-mouse secondary antibody conjugated to Alexa Fluor®647 (Life Technologies) was added at a 1:1000 dilution for 24 minutes in a 2% goat serum and 2% BSA solution. The slides were then incubated with Alexa Fluor® 488 labeled cytokeratin antibody (clones AE1 and AE3, 1:200 dilution, eBioscience, San Diego, CA, USA), Alexa Fluor® 488 labeled cytokeratin antibody (C11, 1:100 dilution, BioLegend, San Diego, CA, USA), and R-phycoerythrin (PE) labeled CD45 antibody (HI30, 1:100 dilution, BioLegend) for 48 minutes in a 2% mouse serum and 2% BSA solution. All antibodies and serum diluents were stored in Inline User-Fillable Dispensers (Ventana) at 4x working concentration and diluted into Reaction Buffer (Ventana). DAPI or Hoechst 33342 was also included in this last incubation at 5 ug/mL/mL. Washes were performed by the Discovery Ultra as per manufacturer’s protocol. After completion of staining slides were removed and placed in Reaction Buffer for 5 minutes and washed 5 times with distilled water, and once with PBS. Coverslips were applied using Fluoromount (Sigma Aldrich). Slides were dried for at least 1 hour at room temperature before scanning. For clinical samples, some slides were stained with Ki67 antibody (clone 7B11, 1:100 dilution, Invitrogen, Carlsbad, CA, USA ) using a similar protocol to that used for EpCAM staining, substituting Ki-67 for EpCAM.
Blood was collected from advanced breast, prostate and colorectal patients being followed at the Seattle Cancer Care Alliance according to a protocol approved by the Fred Hutchinson Cancer Research Center IRB. Blood was collected from a patient with triple-negative breast cancer as part of the ITOMIC study by the Center for Cancer Innovation at the University of Washington (clinicaltrials.gov identifier NCT01957514); the study protocol was approved by the Fred Hutchinson Cancer Research Center IRB. Appropriate informed consent was received from all cancer patients. Blood samples were processed onto slides and stained on the Discovery Ultra as described above.
Automated image capture and analysis
After staining, slides were placed onto the CyteFinder digital scanning microscope to acquire fluorescent images. The microscope is oriented with the objective positioned below the sample. For each slide, the CyteFinder acquired 4-channel fluorescent images of 2542 discrete fields of view to cover the area on the slide where the sample was spread (Additional file 3: Figure S3). Individual fields of view overlap by approximately 50 μm on all sides to prevent obtaining partial images of cells on the borders of adjacent fields. A solid-state, LED illuminator (Lumencor, Beaverton, OR) was used to excite the fluorophores. Images were captured using a Coolsnap® EZ CCD camera (Photometrics, Tucson, AZ). Filters for excitation and emission were from the Brightline® product collection (Semrock, Rochester, NY). Low magnification scan images were acquired with a Nikon 10X 0.3NA objective (Nikon Instruments, Melville, NY) with a lateral resolution of 1.06 um. The high resolution images of revisited points were acquired with a Nikon 40X 0.6NA objective with a lateral resolution of 529 nm. Revisited points were imaged with a “stack” of images through the Z plane with 1um steps. The images were presented to the reviewer as individual z planes rather than projection images.
Images were analyzed for the presence of signal above background for each channel (except nuclear dye channel) using Analyzer image analysis software (RareCyte) that employs an adaptive auto-threshold algorithm. The primary detection was performed on the fluorescent channel corresponding to the cytokeratin (CK) label. The objects identified by their CK signal were then analyzed to determine their correlation with the CD45 label (a negative marker). Highly correlative objects were rejected as this indicated the presence of CD45 label on CK positive objects. Objects that are determined by the algorithm to be CK positive and CD45 negative were presented to the reviewer for classification (see next section). Objects to be classified are termed “glyphs” and are highlighted by a 200 × 200 pixel box.
Review and cell classification
CyteMapper® is a review software system that presents glyphs to the reviewer as a row of 4 boxes showing each individual fluorescence channel as grayscale images with scalable brightness and contrast (Additional file 4: Figure S4). A later version of the viewer included a fifth box showing a color composite image of channels superimposed on one another. The reviewer can view the entire panel in which the glyph was found to determine its relationship to other cells in the sample and can zoom in on images to facilitate classification.
Objects were classified into three categories: (1) “Cell”, (2) “Not a Cell”, or (3) “Indeterminate” based on established criteria for cells of epithelial origin [7-9]. A “Cell” met all criteria for a CTC, including positive nuclear stain, a positive cytokeratin signal, and a negative CD45 signal. EpCAM or EGFR (for A549 mCTCs) were used as additional interpretive markers for classification of “Cell”. An “Indeterminate” object met a combination of criteria that may include positive signal in two of three channels and/or positive signal in the “negative” channel. “Not a Cell” is used for all other objects. A tally of the number of objects in each category was kept by the software and reported upon saving the reviewed file. Only objects classified as “Cell” were included in tallies of CTCs. The performance of CyteMapper review for the mCTC spike-in experiments was shared among three scientists with extensive experience in the investigation of CTCs and in the use of CyteMapper for the identification of epithelial cells.
CTC enumeration comparison
Blood from 10 patients with advanced breast, prostate or colorectal cancer was evaluated in a clinical feasibility study. Two 7.5 mL specimens of blood were drawn from cancer patients at the same time; one was given to the University of Washington (UW) Medical Center clinical laboratory for CTC evaluation by CellSearch and the other to RareCyte for CTC evaluation by AccuCyte – CyteFinder. CTCs were counted by CellSearch according to manufacturer’s instructions (Janssen Diagnostics, Raritan, NJ) and by AccuCyte – CyteFinder as described above. CTCs identified by AccuCyte – CyteFinder met CellSearch criteria: positive staining for cytokeratin and nucleus and negative staining for CD45. Investigators at RareCyte were blinded to the CellSearch counts until after the results from both assays were documented and delivered to investigators at UW.
Retrieval of individual mCTC from slides
Isolation of single cells from slides was performed with CytePicker that is integrated with CyteFinder (Additional file 5: Figure S5). CytePicker is a hydraulically controlled semi-automated single cell retrieval device that contains three critical parts: (1) needle with 22 um-bore ceramic tip, (2) pump capable of 200 pL droplet resolution, (3) precision Z-positioning system using a piezo-electric actuator. Imaging of the cells was performed with a 10x, 0.30NA objective through the slide (rather than through a coverslip) so that uncovered cells are accessible to the ceramic tipped needle above the slide. Chromatic aberrations are measured and compensated for in software prior to imaging so that all fluorescent channel images are appropriately co-registered.
SKBR3 mCTCs were spiked into blood, which was processed and stained as above for cytokeratin, EpCAM, CD45 and nuclear DNA. Samples that were used for individual cell retrieval were prepared without a coverslip. After CyteFinder scanning, the Imager3 software module used the data generated from the scan/analysis/review routine to create a list of coordinates of cellular locations on the slide. Individual cell locations were visited (and viewed at 40× objective magnification if desired) to verify that the candidate cell met CTC criteria described above. A droplet of PBS was deposited on the slide in the area of the cell of interest. Using the CytePicker software module, the needle was lowered to make contact with the sample surface. Using the piezo-actuated Z control, the operator directed the needle tip 20–30 μm past the surface of the sample to “cut” into the sample layer. A controlled circular movement (termed “wiggle”) with a diameter between 25 and 40 μm was directed by the Imager3 software to dislodge the cell from the surface of the slide into the needle tip. Removal of the cell was confirmed visually (see Additional file 6: Figure S6). The needle was then raised and the operator placed a PCR tube under the needle. A volume of 2 μL was then dispensed into the bottom of the PCR tube and the sample was immediately frozen at -80C.
AccuCyte – CyteFinder laboratory workflow (in minutes)
Image Review/CTC Confirmation
Total AccuCyte - CyteFinder
CytePicker cell retrieval (per cell)
2 - 3
Whole genome amplification and molecular analysis of mCTC
After thawing individually picked SKBR3 cells at room temperature, the cells were lysed and genomes amplified with the Ampli1 WGA procedure according to manufacturer’s instructions (Silicon BioSystems, Bologna, Italy). Approximately 1 μL of the WGA reaction product was used for amplification of the TP53 gene that encodes the region of the protein containing the p.R175H mutation. Nested PCR primers were designed from the NCBI human reference genomic sequence and amplified from ch17:7577987–7578592 for the outer primers (5′-CCCTGACTTTCAACTCTGTCTC-3′ and 5′-AGGCCCTTAGCCTCTGTAA-3′) and ch17:7578281–7578503 for the inner primers (5′-GTGCAGCTGTGGGTTGATT-3′ and 5′-GGGCCAGACCTAAGAGCAAT-3′) using Primer3 software [10,11]. The amplicon generated from the outer primer set was 606 bp and from the inner primer set was 224 bp. Approximately 1 μL of sample from the WGA product was transferred into a PCR tube with 2X PCR reaction mix (New England Biolabs, Ipswich, MA, USA), 0.5 μM of each primer, and water was mixed and placed into a thermal cycler (Thermo Fisher Scientific). Thermal cycling conditions were as follows: (1) incubation at 94°C for 7 minutes, (2) 30 cycles of 94°C for 30 seconds, 60°C for 30 seconds and 72°C for 30 seconds, (3) final extension at 72°C for 7 minutes. Samples were held at 4°C until they were analyzed by gel electrophoresis. After PCR, the presence of the 224 bp amplicon was confirmed by loading a portion of the reaction onto a 2% agarose gel, and staining with SYBR® safe (Invitrogen) and comparing its migration to a DNA size standard.
The resulting amplicon was purified from primers using the DNA Clean & Concentrator (Zymo Research, Irvine, CA, USA) according to manufacturer’s instructions. Approximately 1 ng of amplicon was mixed with sequencing primer (inner PCR primers) and BigDye® Terminator sequencing reactions (Life Technologies) were performed according to manufacturer’s directions. Reactions were run on a 3730XL DNA Analyzer (ThermoFisher Scientific). Sequences were analyzed for the presence of the nucleotide mutation that defines p.R175H (c.524G > A).
WGA products from single SKBR3 cells were analyzed by array CGH using oligonucleotide-based SurePrint G3 Human CGH 4x180K arrays from Agilent Technologies (Santa Clara, CA) as described previously . Briefly, one microgram of WGA DNA was labeled per hybridization. Since the WGA products ranged in size from 100 bp to 1 kb, it was not necessary to perform DNA fragmentation before labeling. Test DNAs were labeled with dCTP-Cy5 and reference DNAs were labeled with dCTP-Cy3, for 2 hours at 37°C using a Spectral Labeling Kit (Perkin Elmer, Boston, MA). Unincorporated nucleotides were removed using a MultiScreen-PCRμ96 Filter Plate (Millipore, Billerica, MA). Hybridizations were carried out at 65°C for 40–72 hours to enhance the binding of WGA DNA, after which they were washed and scanned using an Agilent Microarray Scanner (PN G2565BA). Data was extracted using Agilent’s Feature Extraction software (version 126.96.36.199) and was analyzed using Agilent CytoGenomics Edition 188.8.131.52. The DNA used as a reference for each single lymphoblast cell WGA product was a pool of WGA DNA from multiple (5–10 single cell) WGA reactions from either male or female lymphoblast reference cell lines. Gender-mismatched references were used unless otherwise indicated.
Slides were scanned into image files using the Agilent G2565 Microarray Scanner. Scanned images were quantified using Agilent Feature Extraction software (v10.10.0.23). Text file outputs containing quantitative data were imported into the Agilent CytoGenomics software (version 184.108.40.206). Data were analyzed using the Aberration Detection Method 2 (ADM2) statistical algorithm at a threshold of 6.0 to identify genomic intervals with copy number changes. To reduce false positive calls, a filter was applied to define the minimum log2 ratio (0.25), the minimum size (100 kb) and the minimum number of probes (100) in a CNV interval. The Derivative Log Ratio Spread (DRLS), a measure of probe to probe noise calculated by the CytoGenomics software, was used as a performance measure for hybridization quality.
The karyotype of SKBR3 for reference comparison is found at this this website: http://old-www.path.cam.ac.uk/~pawefish/BreastCellLineDescriptions/sk-br-3.htm.
Whole exome sequencing
A DNA fragment library was constructed from WGA products from individual SKBR3 cells picked from whole blood spike-in samples using a modified version of the NEBNext (New England Biolabs) protocol. Libraries were enriched using the SeqCap EZ Exome v3 capture system (Roche NimbleGen) for the coding portion of the genome. The target includes all coding content from the CCDS, RefSeq and miRBase databases. Paired-end (100 base pair) sequencing of enriched libraries was performed using a HiSeq 2500 system with TruSeq v3 chemistry (Illumina) with a read depth of 15 – 30x. The resulting reads were aligned to the genome human reference (hg19) using BWA (Burrows-Wheeler Aligner)  and variants called with GATK (Genome Analysis Toolkit) [14,15].
Recovery of spiked-in mCTC from whole blood
Detection of single-digit numbers of spiked-in mCTC
Recovery of single-digit spike-in mCTCs
Number spiked-in mCTCs
Number mCTCs identified
CTC detection and characterization in breast cancer
AccuCyte-CyteFinder comparison to CellSearch
Retrieval and molecular analysis of individual CTCs
Whole exome sequencing and array CGH of SKBR3 mCTC
Here we have presented a dual-technology platform for the identification and characterization of CTCs that is comprehensive, reproducible, and sensitive. Across 4 different cancer cell lines, more than 90% of mCTCs spiked into blood were consistently recovered in replicate experiments. Spike-in experiments using single-digit numbers of mCTCs demonstrated reproducible detection of the mCTCs with minimal loss and an experimental limit of detection of a single cell in 7.5 mL of blood. CTC counts in advanced stage cancer patients either exceeded or were similar to CellSearch CTC counts. Finally, individually visualized mCTCs were isolated using a retrieval device for the performance of molecular genetic analyses – including Sanger sequencing, whole exome sequencing, and array-based comparative genomic hybridization – that employed preparative single-cell whole genome amplification.
Our mCTC recovery experiments suggest that the fraction of the buffy coat that is collected using the AccuCyte system approaches 100%, since there is likely some decrease in CTC yield due to the staining and image analysis steps. In contrast, CPT tube and Ficoll-Hypaque® density-based separation methods are reported to have a maximum white blood cell yield of 60 – 80% and can be highly variable [18,19]. Thus, this platform ensures that virtually all of the CTCs present within a blood sample are deposited onto slides for microscopic analysis, regardless of their size or the expression of specific surface molecules.
In our experience, cytokeratin is a more reliable epithelial marker than EpCAM, which has variable expression across cell lines and even within clinical CTCs in a single cluster. Low EpCAM expression or EpCAM downregulation in CTCs have been cited as reasons for the inefficiency of EpCAM capture methods in certain cancer types [20-22]. In our study, the PC3 cell line expressing very low levels of EpCAM was recovered from spike in experiments as efficiently as cell lines expressing higher EpCAM levels. CyteFinder incorporates high-resolution (40× objective) cell imaging as an important tool for definitively classifying CTCs. For the low number spike-in experiments this tool was used to exclude cytokeratin-positive cells that lacked mCTC morphology.
At the center of the platform workflow is a staining step employing an automated IHC instrument (the Ventana Discovery® Ultra). Automated IHC instruments are common in histopathology laboratories worldwide, and their use simplifies workflow and reduces hands-on time requirements for sample processing. Moreover, they allow the application of various antibody reagents, providing for an “open” platform for CTC evaluation. These reagents may be to identify drug targets, such as Her2 for breast cancer, or non-epithelial phenotypic markers, such as CD146 and NG2 for melanoma, (data not shown) or physiologically meaningful biomarkers, such as Ki67, as demonstrated above. Markers for mesenchymal transformation or cancer differentiation are equally possible to incorporate. Recently we have demonstrated that CTC identification by the AccuCyte – CyteFinder system is independent of automated staining instrument; spike-in recovery of PC3 cells on the Dako Autostainer® Link 48 averaged 93%, and single digit spike-in limit of detection was also one cell in 7.5 mL (data not shown).
In a clinical feasibility cohort of advanced breast, prostate and colorectal cancer patients, AccuCyte – CyteFinder enumeration of CTCs compared favorably to the only FDA-cleared system for counting CTCs (CellSearch). This is consistent with the understanding that not all CTCs express sufficient EpCAM to be collected by immunomagnetic bead capture. Clinical application of the AccuCyte – CyteFinder system was also demonstrated in an evaluation of CTCs in a patient with advanced triple-negative breast cancer; here we demonstrated application of biomarkers for proliferation (Ki-67) and drug targeting (Her2), and observed cell clusters, which have been reported to be indicative of aggressive disease [16,17].
False positive CTC identification by AccuCyte – CyteFinder appears to be very low. In the spike-in recovery study of 20 samples, recovery rate was never greater than 100% (unlike some other CTC platforms), and in the single-digit spike-in experiments, only one sample had a higher CTC count than spike-in number, and these cells could be morphologically distinguished as non-malignant. Furthermore, in the ongoing comparison with CellSearch, we have evaluated numerous samples in which no cells have been found (data not shown). This is circumstantial evidence that false-positive identification is likely to be extremely rare. Formal studies of false-positive rate are important and will be performed in the future.
CTCs are increasingly regarded as windows through which to observe dynamic changes in the molecular biology of solid tumors. Retrieval of CTCs for molecular analysis will thus likely be an important aspect of future CTC technologies. We have demonstrated the use of an integrated single cell retrieval device, the CytePicker, that can routinely collect individual cells that are adherent to microscopic slides after identification with CyteFinder. The process is compatible with whole genome amplification of single cells, which then can be followed by various molecular genetic analysis methods. Here we have shown that both nested PCR followed by Sanger sequencing and whole exome sequencing identifed a known TP53 mutation in SKBR3 mCTCs, and that array-based comparative genomic hybridization confirmed the reported SKBR3 karyotype. Similar investigations are currently being undertaken in single CTCs from cancer patient samples.
We have developed a comprehensive and sensitive dual-technology platform for flexible identification and characterization of CTCs on microscopic slides using established histopathology staining instruments. The platform has been successfully applied to longitudinal investigation of a patient with breast cancer on a clinical trial protocol and it can readily isolate single cells for sequencing and other genomic analyses. It thus permits the non-invasive and repeated accurate monitoring of therapeutic response and evolving cancer biology to enable personalized, molecularly-guided cancer treatment.
The authors gratefully recognize Elizabeth Mahen and Kimberly Burton for support in the breast cancer clinical study, and Drs. David Parkinson and John Rasko for critical reading of the manuscript.
- Wittekind C, Neid M. Cancer invasion and metastasis. Oncology. 2005;69 Suppl 1:14–6.View ArticlePubMedGoogle Scholar
- Kim MY, Oskarsson T, Acharyya S, Nguyen DX, Zhang XH, Norton L, et al. Tumor self-seeding by circulating cancer cells. Cell. 2009;139(7):1315–26.View ArticlePubMedPubMed CentralGoogle Scholar
- Cristofanilli M, Budd GT, Ellis MJ, Stopeck A, Matera J, Miller MC, et al. Circulating tumor cells, disease progression, and survival in metastatic breast cancer. N Engl J Med. 2004;351(8):781–91.View ArticlePubMedGoogle Scholar
- de Bono JS, Scher HI, Montgomery RB, Parker C, Miller MC, Tissing H, et al. Circulating tumor cells predict survival benefit from treatment in metastatic castration-resistant prostate cancer. Clin Cancer Res. 2008;14(19):6302–9.View ArticlePubMedGoogle Scholar
- Cohen SJ, Punt CJ, Iannotti N, Saidman BH, Sabbath KD, Gabrail NY, et al. Relationship of circulating tumor cells to tumor response, progression-free survival, and overall survival in patients with metastatic colorectal cancer. J Clin Oncol. 2008;26(19):3213–21.View ArticlePubMedGoogle Scholar
- Parkinson DR, Dracopoli N, Petty BG, Compton C, Cristofanilli M, Deisseroth A, et al. Considerations in the development of circulating tumor cell technology for clinical use. J Transl Med. 2012;10:138.View ArticlePubMedPubMed CentralGoogle Scholar
- Meng S, Tripathy D, Frenkel EP, Shete S, Naftalis EZ, Huth JF, et al. Circulating tumor cells in patients with breast cancer dormancy. Clin Cancer Res. 2004;10(24):8152–62.View ArticlePubMedGoogle Scholar
- Allard WJ, Matera J, Miller MC, Repollet M, Connelly MC, Rao C, et al. Tumor cells circulate in the peripheral blood of all major carcinomas but not in healthy subjects or patients with nonmalignant diseases. Clin Cancer Res. 2004;10(20):6897–904.View ArticlePubMedGoogle Scholar
- Riethdorf S, Fritsche H, Muller V, Rau T, Schindlbeck C, Rack B, et al. Detection of circulating tumor cells in peripheral blood of patients with metastatic breast cancer: a validation study of the Cell Search system. Clin Cancer Res. 2007;13(3):920–8.View ArticlePubMedGoogle Scholar
- Koressaar T, Remm M. Enhancements and modifications of primer design program Primer3. Bioinformatics. 2007;23(10):1289–91.View ArticlePubMedGoogle Scholar
- Untergasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC, Remm M, et al. Primer3–new capabilities and interfaces. Nucleic Acids Res. 2012;40(15):e115.View ArticlePubMedPubMed CentralGoogle Scholar
- Bi W, Borgan C, Pursley AN, Hixson P, Shaw CA, Bacino CA, et al. Comparison of chromosome analysis and chromosomal microarray analysis: what is the value of chromosome analysis in today’s genomic array era? Genet Med. 2013;15(6):450–7.View ArticlePubMedGoogle Scholar
- Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60.View ArticlePubMedPubMed CentralGoogle Scholar
- McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The genome analysis toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20(9):1297–303.View ArticlePubMedPubMed CentralGoogle Scholar
- DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43(5):491–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Hou JM, Krebs MG, Lancashire L, Sloane R, Backen A, Swain RK, et al. Clinical significance and molecular characteristics of circulating tumor cells and circulating tumor microemboli in patients with small-cell lung cancer. J Clinical Oncol. 2012;30(5):525–32.View ArticleGoogle Scholar
- Divella R, Daniele A, Abbate I, Bellizzi A, Savino E, Simone G, et al. The presence of clustered circulating tumor cells (CTCs) and circulating cytokines define an aggressive phenotype in metastatic colorectal cancer. Cancer causes & control : CCC. 2014. doi:10.1007/s10552-014-0457-4. PubMed PMID: 25135616.Google Scholar
- Schlenke P, Kluter H, Muller-Steinhardt M, Hammers HJ, Borchert K, Bein G. Evaluation of a novel mononuclear cell isolation procedure for serological HLA typing. Clin Diagn Lab Immunol. 1998;5(6):808–13.PubMedPubMed CentralGoogle Scholar
- Nilsson C, Aboud S, Karlen K, Hejdeman B, Urassa W, Biberfeld G. Optimal blood mononuclear cell isolation procedures for gamma interferon enzyme-linked immunospot testing of healthy Swedish and Tanzanian subjects. Clin Vaccine Immunol. 2008;15(4):585–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Rao CG, Chianese D, Doyle GV, Miller MC, Russell T, Sanders Jr RA, et al. Expression of epithelial cell adhesion molecule in carcinoma cells present in blood and primary and metastatic tumors. Int J Oncol. 2005;27(1):49–57.PubMedGoogle Scholar
- Krebs MG, Sloane R, Priest L, Lancashire L, Hou JM, Greystoke A, et al. Evaluation and prognostic significance of circulating tumor cells in patients with non-small-cell lung cancer. J Clinical Oncol. 2011;29(12):1556–63.View ArticleGoogle Scholar
- Thurm H, Ebel S, Kentenich C, Hemsen A, Riethdorf S, Coith C, et al. Rare expression of epithelial cell adhesion molecule on residual micrometastatic breast cancer cells after adjuvant chemotherapy. Clin Cancer Res. 2003;9(7):2598–604.PubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.