Size-Selective Microcavity Array for Rapid and Efficient Detection of

Jun 28, 2010 - Folic acid modified electrospun poly(vinyl alcohol)/polyethyleneimine nanofibers for cancer cell capture applications. Zhang-yu ... Pro...
0 downloads 15 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 6629–6635

Size-Selective Microcavity Array for Rapid and Efficient Detection of Circulating Tumor Cells Masahito Hosokawa, Taishi Hayata, Yorikane Fukuda, Atsushi Arakaki, Tomoko Yoshino, Tsuyoshi Tanaka, and Tadashi Matsunaga* Department of Biotechnology, Tokyo University of Agriculture and Technology, 2-24-16 Naka-cho, Koganei, Tokyo 184-8588, Japan Circulating tumor cells (CTCs) are tumor cells circulating in the peripheral blood of patients with metastatic cancer. Detection of CTCs has clinical significance in cancer therapy because it would enable earlier diagnosis of metastasis. In this research, a microfluidic device equipped with a size-selective microcavity array for highly efficient and rapid detection of tumor cells from whole blood was developed. The microcavity array can specifically separate tumor cells from whole blood on the basis of differences in the size and deformability between tumor and hematologic cells. Furthermore, the cells recovered on the microcavity array were continuously processed for imagebased immunophenotypic analysis using a fluorescence microscope. Our device successfully detected approximately 97% of lung carcinoma NCI-H358 cells in 1 mL whole blood spiked with 10-100 NCI-H358 cells. In addition, breast, gastric, and colon tumor cells lines that include EpCAM-negative tumor cells, which cannot be isolated by conventional immunomagnetic separation, were successfully recovered on the microcavity array with high efficiency (more than 80%). On an average, approximately 98% of recovered cells were viable. Our microfluidic device has high potential as a tool for the rapid detection of CTCs and can be used to study CTCs in detail. Detection of CTCs has received a great deal of attention because detection of these cells could lead to earlier diagnosis of metastatic disease and the process is less invasive than conventional methods.1,2 The main techniques used to detect CTCs from blood are immunocytochemistry,3 polymerase chain reaction (PCR),4 and flow cytometry.5 However, since CTCs are extremely * To whom correspondence should be addressed. Phone: +81-42-388-7020. Fax: +81-42-385-7713. E-mail: [email protected]. (1) Paterlini-Brechot, P.; Benali, N. L. Cancer Lett. 2007, 253, 180–204. (2) Mostert, B.; Sleijfer, S.; Foekens, J. A.; Gratama, J. W. Cancer Treat Rev. 2009, 35, 463–474. (3) Cristofanilli, M.; Budd, G. T.; Ellis, M. J.; Stopeck, A.; Matera, J.; Miller, M. C.; Reuben, J. M.; Doyle, G. V.; Allard, W. J.; Terstappen, L. W.; Hayes, D. F. N. Engl. J. Med. 2004, 351, 781–791. (4) Smith, B. M.; Slade, M. J.; English, J.; Graham, H.; Luchtenborg, M.; Sinnett, H. D.; Cross, N. C.; Coombes, R. C. J. Clin. Oncol. 2000, 18, 1432–1439. (5) Moreno, J. G.; O’Hara, S. M.; Gross, S.; Doyle, G.; Fritsche, H.; Gomella, L. G.; Terstappen, L. W. Urology 2001, 58, 386–392. 10.1021/ac101222x  2010 American Chemical Society Published on Web 06/28/2010

rare, as low as 1 CTC per 109 hematologic cells in the blood,3,6 separation and enrichment of them from whole blood is generally needed in order to increase detection sensitivity to an acceptable levels. Numerous methods have been developed to separate CTCs from whole blood. One type of CTC separation method targets cells with specific cell-surface markers such as EpCAM by using magnetic beads or structures coated with a monoclonal antibodies.7 CellSearch (VeridexTM, Warren, PA), a semiautomated technology, enriches whole blood for CTCs by adding ferrofluids coated with antibodies that target EpCAM.8,9 CTCs in the enriched population are stained for tumor markers and counted using an automated fluorescence microscope. In addition, a number of microfluidic devices have been developed to detect CTCs from whole blood, and these devices are more efficient than the beadbased method.10-13 Although EpCAM-based enrichment methods have been frequently used by many groups, several studies have shown that the presence of EpCAM on tumor cells varies with tumor type.14,15 Therefore, CTC separation methods based on an antigen-antibody reaction cannot achieve stable and reproducible recovery of CTCs from all tumor types. (6) Ross, A. A.; Cooper, B. W.; Lazarus, H. M.; Mackay, W.; Moss, T. J.; Ciobanu, N.; Tallman, M. S.; Kennedy, M. J.; Davidson, N. E.; Sweet, D.; et al. Blood 1993, 82, 2605–2610. (7) Allard, W. J.; Matera, J.; Miller, M. C.; Repollet, M.; Connelly, M. C.; Rao, C.; Tibbe, A. G.; Uhr, J. W.; Terstappen, L. W. Clin. Cancer Res. 2004, 10, 6897–6904. (8) Riethdorf, S.; Fritsche, H.; Muller, V.; Rau, T.; Schindlbeck, C.; Rack, B.; Janni, W.; Coith, C.; Beck, K.; Janicke, F.; Jackson, S.; Gornet, T.; Cristofanilli, M.; Pantel, K. Clin. Cancer Res. 2007, 13, 920–928. (9) Deng, G.; Herrler, M.; Burgess, D.; Manna, E.; Krag, D.; Burke, J. F. Breast Cancer Res. 2008, 10, R69. (10) Adams, A. A.; Okagbare, P. I.; Feng, J.; Hupert, M. L.; Patterson, D.; Gottert, J.; McCarley, R. L.; Nikitopoulos, D.; Murphy, M. C.; Soper, S. A. J. Am. Chem. Soc. 2008, 130, 8633–8641. (11) Nagrath, S.; Sequist, L. V.; Maheswaran, S.; Bell, D. W.; Irimia, D.; Ulkus, L.; Smith, M. R.; Kwak, E. L.; Digumarthy, S.; Muzikansky, A.; Ryan, P.; Balis, U. J.; Tompkins, R. G.; Haber, D. A.; Toner, M. Nature 2007, 450, 1235–1239. (12) Gleghorn, J. P.; Pratt, E. D.; Denning, D.; Liu, H.; Bander, N. H.; Tagawa, S. T.; Nanus, D. M.; Giannakakou, P. A.; Kirby, B. J. Lab Chip 2010, 10, 27–29. (13) Wang, S.; Wang, H.; Jiao, J.; Chen, K. J.; Owens, G. E.; Kamei, K.; Sun, J.; Sherman, D. J.; Behrenbruch, C. P.; Wu, H.; Tseng, H. R. Angew. Chem., Int. Ed. 2009, 48, 8970–8973. (14) Went, P. T.; Lugli, A.; Meier, S.; Bundi, M.; Mirlacher, M.; Sauter, G.; Dirnhofer, S. Hum. Pathol. 2004, 35, 122–128. (15) Sieuwerts, A. M.; Kraan, J.; Bolt, J.; van der Spoel, P.; Elstrodt, F.; Schutte, M.; Martens, J. W.; Gratama, J. W.; Sleijfer, S.; Foekens, J. A. J. Natl. Cancer Inst. 2009, 101, 61–66.

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6629

Figure 1. CTC recovery device equipped with the size-selective microcavity array (a) Schematic image of CTC recovery using the sizeselective microcavity array. (b) Photograph of the size-selective microcavity array. (c) Photograph of the CTC recovery device equipped with the array. (d) Scanning electron microscope image of MCF-7 cells trapped on the microcavity array. The microcavities are 9 µm in size, with a 60 µm pitch.

Other groups have reported CTC separation methods that are based on differences in the size and deformability between CTCs and hematologic cells. Isolation by size of epithelial tumor cells (ISET) can be achieved by using filtration to separate individual tumor cells on the basis of size, since tumor cells (>8 µm) are larger than leukocytes.16,17 ISET using a polycarbonate filter is an inexpensive, user-friendly method of enriching CTCs. However, the pores of polycarbonate filters, which are fabricated by track etching, are randomly placed with nonuniform density.18,19 To address this problem, some microfluidic devices include functional structures such as a membrane microfilter,20 crescent-shaped isolation wells,21 a serpentine channel lined with tall rectangular apertures,22 or an array of four successively narrow channels23 in order to isolate tumor cells from blood. Using such microfluidic devices, Zheng et al. recovered hematoxylin-stained LNCaP cells that had been used to spike whole blood, with an efficiency of 89.0 ± 9.5%.20 Tan et al. recovered MCF-7, MDA-MB231, and HT29 cells from PBS with more than 80% efficiency,21 and Kuo et al. recovered paraformaldehyde-fixed MCF-7 cells from a medium with 90% efficiency.22 Since pore size and geometry were precisely controlled by microfabrication, these devices recovered tumor cells with high efficiency via filtration; further design optimization could improve throughput and usefulness. Recently, we have developed a microfluidic device integrated with a high-density microcavity array with 100-10 000 microcavi(16) Vona, G.; Sabile, A.; Louha, M.; Sitruk, V.; Romana, S.; Schutze, K.; Capron, F.; Franco, D.; Pazzagli, M.; Vekemans, M.; Lacour, B.; Brechot, C.; PaterliniBrechot, P. Am. J. Pathol. 2000, 156, 57–63. (17) Zabaglo, L.; Ormerod, M. G.; Parton, M.; Ring, A.; Smith, I. E.; Dowsett, M. Cytometry, Part A 2003, 55, 102–108. (18) Fleischer, R. L.; Alter, H. W.; Furman, S. C.; Price, P. B.; Walker, R. M. Science 1972, 178, 255–263. (19) Rostagno, P.; Moll, J. L.; Bisconte, J. C.; Caldani, C. Anticancer Res. 1997, 17, 2481–2485. (20) Zheng, S.; Lin, H.; Liu, J. Q.; Balic, M.; Datar, R.; Cote, R. J.; Tai, Y. C. J. Chromatogr., A 2007, 1162, 154–161. (21) Tan, S. J.; Yobas, L.; Lee, G. Y.; Ong, C. N.; Lim, C. T. Biomed. Microdevices 2009, 11, 883–892. (22) Kuo, J. S.; Zhao, Y.; Schiro, P. G.; Ng, L.; Lim, D. S.; Shelby, J. P.; Chiu, D. T. Lab Chip 2010, 10, 837–842. (23) Mohamed, H.; Murray, M.; Turner, J. N.; Caggana, M. J. Chromatogr., A 2009, 1216, 8289–8295.

6630

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

ties (diameter, 2 µm).24,25 This device efficiently traps single cells by applying a negative pressure, enabling high-throughput microscopic analyses of the trapped cells. Under optimized conditions, on-chip fluorescence in situ hybridization (FISH) was possible.24 Moreover, the device successfully detected CD34+ hematopoietic stem cells and cytotoxic T lymphoma cells from peripheral blood mononuclear cells (PBMCs) and pathogenic protozoan oocysts from tap water.25-27 Through these studies, we have demonstrated that our microcavity array has potential as a tool for detecting and isolating specific cells, including stem cells, progenitor cells, and tumor cells. In this study, we modified the microcavity array to recover and detect CTCs from whole blood. The size of the microcavities was optimized in order to trap tumor cells on the microcavities while letting blood cells flow through (Figure 1a). Unlike the tracketched poly carbonate filters, the size, geometry, and density of the microcavities can be precisely controlled to achieve specific cell separation according to differences in cellular size and deformability. A size-selective microcavity array was integrated with a microfluidic device so that enrichment of CTCs from blood as well as the staining and washing processes in the microfluidic assay could be performed within one integrated device. This device selectively recovered tumor cells from blood on the array without relying on the expression of epithelial cell markers. The trapped cells could be easily counted by scanning specified areas with an automated fluorescence microscope. This device, therefore, has high potential as an inexpensive and efficient tool for CTC detection. EXPERIMENTAL SECTION CTC Recovery Device Fabrication. The size-selective microcavity array is made of nickel by electroforming (Optnics Precision Co. Ltd.). The microcavities of the array were fabricated (24) Matsunaga, T.; Hosokawa, M.; Arakaki, A.; Taguchi, T.; Mori, T.; Tanaka, T.; Takeyama, H. Anal. Chem. 2008, 80, 5139–5145. (25) Hosokawa, M.; Arakaki, A.; Takahashi, M.; Mori, T.; Takeyama, H.; Matsunaga, T. Anal. Chem. 2009, 81, 5308–5313. (26) Arakaki, A.; Ooya, K.; Akiyama, Y.; Hosokawa, M.; Komiyama, M.; Iizuka, A.; Yamaguchi, K.; Matsunaga, T. Biotechnol. Bioeng. 2010, 106, 311–318. (27) Taguchi, T.; Arakaki, A.; Takeyama, H.; Haraguchi, S.; Yoshino, M.; Kaneko, M.; Ishimori, Y.; Matsunaga, T. Biotechnol. Bioeng. 2007, 96, 272–280.

with diameters of 8, 9, 10, or 11 µm at the top surface. The distance between each microcavity was 60 µm, and a total of 10 000 cavities were arranged in each 100 × 100 array (Figure 1b). Poly(dimethylsiloxane) (PDMS) structures were fabricated to integrate with the size-selective microcavity array. The upper substrate consists of a microchamber, sample inlet, and an outlet. A vacuum line was fabricated in the lower substrate beneath the size-selective microcavity array to produce a negative pressure, enabling cell entrapment. Master mold substrates comprising poly(methyl methacrylate) (PMMA) were prepared by a computer-aided modeling machine (PNC-300, Roland DG Corp., Shizuoka, Japan), and silicone tubes (i.d. ) 500 µm) were then connected to the inlet, outlet, and vacuum lines of the respective molds. Both the upper and lower PDMS layers were then fabricated by pouring a mixture of Sylpot 184 silicone elastomer (Dow Corning Asia Ltd., Tokyo, Japan) and curing agent (10:1) onto either the master molds or a blank wafer, followed by curing for at least 20 min at 85 °C. Upon curing, the PDMS substrates were carefully peeled off the molds. The CTC recovery device was constructed by assembling the size-selective microcavity array, the upper and lower PDMS layers using spacer tapes. The sample inlet was connected to a reservoir, while the vacuum microchannel was connected to a peristaltic pump. Then, the cell entrapment setup was placed on a computer-operated motorized stage of an upright microscope (Figure 1c). Cell Culture and Labeling. NCI-H358, AGS, SW620, SNU-1 cells were cultured in RPMI 1640 medium containing 2 mM L-glutamine (Sigma-Aldrich, Irvine, UK), 10% (v/v) FBS (Invitrogen Corp., Carlsbad, CA), and 1% (v/v) penicillin/streptomycin (Invitrogen Corp.) for 3-4 days at 37 °C with 5% CO2 supplementation. Immediately prior to each experiment, cells grown to confluence were trypsinized and resuspended in PBS. MCF-7 cells were cultured in EMEM medium (ATCC, Manassas, VA) containing 2 mM L-glutamine (Sigma-Aldrich), 0.1 mM nonessential amino acids (GIBCO), 0.01 mg/mL bovine insulin (Invitrogen Corp.), 1.5 mg/mL sodium bicarbonate (Invitrogen Corp.), and 10% (v/v) FBS and maintained in a manner similar to that mentioned above. Hs578T cells were cultured in DMEM medium (ATCC, Manassas, VA) containing 0.01 mg/mL bovine insulin and 10% (v/v) FBS and also maintained as described above. To measure the sizes of the tumor cells, cell size distribution was measured using CASY CASYcell counter + Analyzer System Model TTC (Scha¨rfe System GmbH, Reutlingen, Germany). To evaluate device performance, the tumor cells were labeled with CellTracker Red CMTPX (Molecular Probes, Eugene, OR). Labeling was achieved by incubating the cells with a tracking dye (5 µM) for 30 min. The cells were then pelleted by centrifugation (200g for 5 min), the supernatant was decanted, and the cells were washed twice with PBS to remove any excess dye. Then, the cells were resuspended in PBS containing 2 mM EDTA and 0.5% BSA. Cell Entrapment Operation. Blood samples or cell suspensions were introduced into the reservoir. Subsequently, a negative pressure was applied to the cell suspension using a peristaltic pump, connected to the vacuum line. The sample (1 mL) was passed through the microcavities at a flow rate of 100-2000 µL/ min for 0.5-10 min. To remove blood cells that remained on the array, PBS containing 2 mM EDTA and 0.5% BSA (2 mL) was

then introduced into the reservoir and passed through the microcavities at a flow rate of 200 µL/min for 10 min. To optimize the flow rate and size of the microcavity for CTC recovery, cell recovery efficiencies were compared among the tumor cell lines MCF-7, NCI-H358, SW620, AGS, SNU-1, and Hs578T. The cells were preliminarily labeled with CellTracker Red, and the actual number of cells in the suspension was determined as follows: Cell concentrations were first determined by manual counting using a hemocytometer. Next, the cells were diluted to ideal concentrations of 10-100 cells/50 µL by statistical sampling of a serial dilution. Subsequently, 50 µL of the diluted cell suspension was transferred to a 96-well microtiter plate, using a micropipet, and then centrifuged. The number of cells in the transferred suspension was counted under a microscope. Counting was repeated three times in order to determine the actual number of cells in the suspension. After repeating these operations, the cell suspension volume used to spike 1 mL whole blood or buffer was adjusted to prepare tumor cell-spiked samples. The intersample variation in cell number was 5-20%. Human blood samples were collected from healthy donors; the samples were collected in a collection tube with EDTA to prevent coagulation and used within 24 h. Scanning Electron Microscopy (SEM) Imaging of Cells Trapped on the Array. To perform SEM imaging of cells trapped on the microcavity array, the PDMS structures were dismounted from the array, and the SEM sample was prepared as follows: Cells trapped on the microcavity array were fixed with 30 mM HEPES buffer (pH 7.4) containing 2% glutaraldehyde and 2% paraformaldehyde for 60 min at 4 °C. Subsequently, the cells were fixed with 30 mM HEPES buffer containing 1% osmium tetroxide for 60 min at 4 °C. After washing with HEPES buffer for 5 min and Milli-Q water for 1 min, the cells were dehydrated through ethanol solutions of 25, 50, 75, 90, 99.5, and 100% (3 ×), with 5 min in each solution. After dehydration, 100% ethanol was replaced with 1,1,1,3,3,3-hexamethyldisilazane for 5 min, and the cells were then air-dried. Finally, the cells were coated with gold by sputtering using an E-1010 ion sputter (Hitachi Ltd., Tokyo, Japan). Following this, the cells were observed under a scanning electron microscope (VE-9800; Keyence Corp., Osaka, Japan). Fixation and Staining of Trapped Cells for Identification of CTCs. A cell fixation solution and cell staining solution were introduced into the reservoir and passed through the microcavities using a peristaltic pump soon after washing. To stain the CTCs with anticytokeratin antibody, trapped cells were fixed by flowing 400 µL of 1% PFA in PBS through the microcavity array at a flow rate of 20 µL/min for 20 min. After washing with 100 µL of PBS, the cells were subsequently treated with 200 µL of 0.2% Triton X-100 in PBS at a flow late of 20 µL/min for 10 min. The abovedescribed operations were only needed to enable cytokeratin staining. To identify CTCs and leukocytes, 600 µL of cell staining solution containing Hoechst 33342 (Molecular Probes), FITClabeled anti-CD45 antibody (BD Biosciences, San Jose, CA), and PE-labeled anti-EpCAM antibody (BD Biosciences) or PE-labeled anticytokeratin (CAM 5.2) antibody (BD Biosciences) was flowed through the microcavities at a flow rate of 20 µL/min for 30 min. The cell staining solution was prepared by adding 0.5-3.0 µg/ mL of Hoechst 33342 and 50 µL of the antibody stock solution in 600 µL of PBS containing 2 mM EDTA and 0.5% BSA. Finally, Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6631

the array was washed with 1 mL of PBS containing 2 mM EDTA and 0.5% BSA to remove excess dye. Identification and Enumeration of CTCs by Fluorescence Microscopy. After recovery of tumor cells, a whole image of the cell arrayed area was obtained using a fluorescence microscope (BX61; Olympus Corporation, Tokyo, Japan), integrated with a computer-operated motorized stage, WU, NIBA, and WIG filter sets, and a cooled digital camera (DP-70; Olympus Corporation). The Lumina Vision acquisition software (Mitani Corporation, Tokyo, Japan) was used to acquire the images. Image scanning of the microcavity array at three fluorescent wavelengths was completed within 45 min by using a 10× objective lens and the motorized stage. Subsequently, image analysis was performed and objects that satisfied predetermined criteria were counted. Fluorescent intensities and morphometric characteristics such as cell size, shape, and nuclear size were considered when identifying CTCs and excluding nontumor cells. Cells identified as CTCs had round to oval morphology and a visible nucleus (Hoechst 33342 positive) and were positive for cytokeratin/EpCAM and negative for CD45. Sensitivity Test. Samples of peripheral blood to which known amounts of tumor cells had been added were used in sensitivity tests. Varying numbers of tumor cells were spiked into blood, and tumor cell recovery was evaluated using our microcavity array. Briefly, samples of peripheral blood (1 mL) from healthy volunteers were spiked with predetermined numbers of NCI-H358 cells (0, 10, 25, 50, and 100 cells). Cell counts for seeding experiments conducted with whole blood were performed in a manner similar to that mentioned above. The CTC count was performed within 24 h. RESULTS Optimization of the Size-Selective Microcavity Array. In this study, arrays with microcavities of four different diameters within the range of 8-11 µm were produced. The diameters of the cavities were measured by microscopic observation. The diameters were 8.4 ± 0.3 µm, 9.1 ± 0.3 µm, 10.0 ± 0.3 µm, and 10.6 ± 0.3 µm. First, the passage of blood cells through the microcavities of the array was evaluated. Human whole blood (1 mL) was introduced into a reservoir and flowed through the microcavities. The blood sample successfully flowed through without clogging the microcavities. On the other hand, when using a previously developed array with microcavities having a diameter of 2 µm to filter whole blood, blood cells immediately clogged the microcavities. Second, the microcavity size suitable for recovery of various tumor cell lines was determined. Supporting Information (SI) Figure S-1 shows the size distribution of the tumor cell lines used in this study. The typical diameters of tumor cells were 22.5 µm for MCF-7, 18.1 µm for NCI-H358, 14.9 µm for AGS, and 11.6 µm for SW620 cells. To evaluate the effect of cell size on tumor cell recovery rate, MCF-7 and SW620 cells were stained with CellTracker Red and spiked into whole blood at 100 cells/mL. Then, the cells were recovered using different size-selective microcavity arrays. Recovery of tumor cells was completed within 15 min, and successful entrapment was confirmed using a fluorescence microscope. Whole images of the microcavity array were acquired by scanning using the computer-operated motorized stage, while sectional images (8 × 10 ) 80 images) were captured using a 10× 6632

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

Figure 2. Relationships between the average diameters of microcavities and the recovery rate of tumor cells. CellTracker Red-stained MCF-7 and SW620 cells were spiked into whole blood at 100 cells per milliliter and recovered using various size-selective microcavity arrays. The average number of cells detected is presented (n ) 3).

Figure 3. CTC recovery rate as a function of flow rate. Data shows NCI-H358 cell recovery rate from a buffer spiked with 100 NCI-H358 cells per milliliter. (n ) 3).

objective lens and reconstructed to form a single image. Subsequently, the cells recovered on the microcavity array were enumerated. The recovery rates of MCF-7 cells were almost constant when the sizes of microcavities changed. The cells were recovered at a rate of more than 90% (Figure 2). Additionally, the SEM image clearly shows that the cells were trapped and retained on the microcavities (Figure 1d). In contrast, the recovery rate of SW620 cells varied according to the size of microcavities, and these cells were obtained at a maximum recovery rate of 88% when the size-selective microcavity array with microcavities with an average diameter of 9.1 µm was used. When the average diameter of the microcavities increased over 9.1 µm, some of the cells passed through the microcavities, resulting in a decrease in the number of recovered cells. Therefore, for precise and reproducible measurement, subsequent tumor cell recovery experiments were conducted using the size-selective microcavity array with microcavities having average diameters of 8.4-9.1 µm. Optimization of the Flow Rate. To evaluate the effect of flow rate on tumor cell recovery rate, NCI-H358 cells were stained with CellTracker Red and spiked into whole blood at 100 cells/mL. Then, the cells were recovered using the size-selective microcavity arrays with microcavities with an average diameter of 9.1 µm. Whole blood was flowed through the array at rates ranging from 100 to 2000 µL/min for 0.5-10 min. When the blood samples were flowed at a flow rate less than 1000 µL/min, over 90% of NCIH358 cells were successfully recovered (Figure 3). In contrast, when the flow rate was more than 1000 µL/min, some of the tumor cells passed through the microcavities, resulting in lower tumor

cell recovery rate. Therefore, we used a flow rate of 200-1000 µL/min for subsequent experiments. Under this optimized condition, 1000-3000 leukocytes were trapped on the size-selective microcavity array when 1 mL of whole blood was flowed through the microcavity array. The number of leukocytes trapped on the size-selective microcavity array was not obviously changed when the flow rate was increased or if blood was diluted with buffer. The number of leukocytes depends on the degree of degradation of blood sample such as the degree of platelet aggregation. This problem can be addressed by further optimization of array design, such as optimizing the total number and density of microcavities, or blood sample storage condition. Recovery of EpCAM-negative CTCs. To demonstrate our device’s ability to recover EpCAM-negative tumor cells from whole blood, CellTracker Red-labeled Hs578T cells and SNU-1 cells were spiked into whole blood at 100 cells/mL, and then the blood samples were analyzed using the microfluidic device integrated with the size-selective microcavity array. These cells cannot be recovered and detected by CellSearch because CellSearch uses antibodies targeting EpCAM on the surface of CTCs for cell separation and these cells are EpCAM-negative.15 However, recovery with our device does not depend on the expression of cell surface markers such as EpCAM. Thus, 96 ± 5% of Hs578T cells and 93 ± 3% of SNU-1 cells were recovered from the whole blood when our size-selective microcavity array was used. Cell Viability Test. Recovery of viable CTCs is required to understand metastasis. In this study, NCI-H358 cells stained with Hoechst 33342 were spiked into whole blood and recovered on the microcavity array under optimized conditions. Cell viability testing using calcein-AM and ethidium homodimer-1 (EthD-1) was performed after trapping the cells. A solution containing calceinAM and EthD-1 was then flowed through the microcavities, and trapped cells were observed under a fluorescent microscope (SI Figure S-2). Average cell viability was approximately 98%, as determined by assessing membrane integrity. Evaluation of the Process for CTC Enumeration. After the size-selective microcavity array was optimized, the microcavity array was then applied to detect and enumerate model CTCs from a human blood sample. To demonstrate the detection process, unstained tumor cell lines (MCF-7, NCI-H358, AGS, and SW620) were spiked into whole blood at 100 cells/mL. Then, the blood samples were introduced into the reservoir connected to the sizeselective microcavity array (average diameter, 9.1 µm) and flowed through the microcavities. CTC staining solution containing Hoechst 33342, FITC-labeled anti-CD45 antibody and PE-labeled anti-EpCAM or cytokeratin antibody was then flowed through the microcavities. After washing, whole images of the microcavity array were acquired, and cellular immunophenotypes of the trapped cells were determined. Recovery and staining of the tumor cells was completed within 75 min, and image scanning of the size-selective microcavity array, using three fluorescence wavelengths was completed within 45 min by using a 10× objective lens and a motorized stage. Figure 4 shows the fluorescence images of the stained cells that were recovered on the microcavity array. Recovered cells that had round to oval morphology and a visible nucleus (Hoechst 33342-positive) and were positive for EpCAM or cytokeratin and negative for CD45 were identified as tumor cells. CD45-positive cells were identified as contaminating

Figure 4. Fluorescence images of cells trapped on the size-selective microcavity array. Human whole blood spiked with 100 NCI-H358 cells per milliliter was flowed through the array at 200 µL/min. The tumor cell recovery experiments were conducted using the size-selective microcavity array, which had microcavities with an average diameter of 9.1 µm. The cells were stained with Hoechst 33342 (a) and fluorescent-labeled antibodies that target CD45 (b) and EpCAM (c). The merged image (d) identifies CTCs and hematologic cells. Scale bar; 50 µm.

Figure 5. Tumor cell detection efficiency from 1 mL of whole blood spiked with 100 cells of four different cell lines: breast (MCF-7), lung (NCI-H358), gastric (AGS) and colon (SW620) (n ) 3 for each cell line).

normal hematologic cells. The images showed a distinct immunophenotype of epithelial cell marker-positive tumor cells. Although a number of leukocytes were retained on the array, tumor cell enumeration was not complicated because individual cells were trapped on the precisely aligned microcavities. The average detection efficiency for tumor cells was >80% in all cases (Figure 5). In order to evaluate the detection sensitivity of the device, NCIH358 cells that were spiked into whole blood at 10, 25, 50, 100 cells/mL were recovered and detected using the device. The calculated detection efficiency was constant and over 90% when 10-100 tumor cells were present per milliliter of blood (Figure 6). Regression analysis of the number of observed tumor cells versus the number of expected tumor cells yielded a slope of 0.97 Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6633

Figure 6. Evaluation of CTC detection efficiency at various target concentrations. A known number of NCI-H358 cells (10-100 cells) was spiked into 1 mL whole blood and detected using the CTC recovery device integrated with the size-selective microcavity array. The plot represents the number of cells spiked versus the number of cells detected (n ) 3).

[95% confidence interval (CI), 0.91-1.02], an intercept of 0.03 (95% CI, -3.02-3.08), and a correlation coefficient (R2) of 0.99. DISCUSSION To achieve highly efficient recovery of tumor cells using the size-selective microcavity array, the tumor cell recovery conditions were optimized. First, we optimized the diameters of the microcavities for recovery of tumor cells from whole blood without clogging the microcavities. Although the diameter of erythrocytes is in the range of 7.5-8.5 µm, their deformability allows them to traverse capillaries with diameters of 4 µm.28,29 Furthermore, a study of filtration of blood cells reported that 95% of leucocytes with diameters in the range of 6-20 µm can pass through 5 µm pores.30 Based on these understanding, we evaluated arrays with microcavities of four different diameters within the range of 8-11 µm. There is no evidence that all CTCs are larger than 8 µm. However, the size-selective microcavity array is an attractive tool with potential for recovering tumor cells that cannot be recovered from the whole blood of cancer patients by conventional separation methods that use antigen-antibody reactions. Second, we optimized the flow rate at which blood samples were introduced into the microfluidic device equipped with the microcavity array. Blood samples were introduced into previously reported CTC recovery devices at flow rates of 10-30 µL/min.11,21 Blood samples should be analyzed as rapidly as possible to prevent the sample from deteriorating. Since our size-selective microcavity array processes samples rapidly (200-1000 µL/min), stable recovery of CTCs from human blood is possible. Then, the process for enumeration of tumor cells performed within the microfluidic device was evaluated. Tumor cells recovered on the size-selective microcavity array were successfully stained with fluorescent-labeled antibodies that target tumor cell markers. Staining and washing were considered to have little or no effect on the retention of tumor cells on the microcavities. Because individual cells were trapped on the microcavities, the cells could be easily enumerated at the single-cell level by image(28) Shapiro, H. M.; Schildkraut, E. R.; Curbelo, R.; Laird, C. W.; Turner, B.; Hirschfeld, T. J. Histochem. Cytochem. 1976, 24, 396–401. (29) Chien, S.; Usami, S.; Dellenback, R. J.; Gregersen, M. I. Am. J. Physiol. 1970, 219, 136–142. (30) Cook, A. M.; Evans, S. A.; Lane, I. F.; Jones, J. G. Br. J. Haematol. 1998, 102, 952–956.

6634

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

based immunophenotyping. This advantageous feature was used to successfully detect various tumor cells such as breast, lung, gastric, and colon cancer cells from whole blood with high efficiencies. The results obtained from the detection sensitivity test showed that our device enables more efficient and reproducible detection of CTCs from whole blood as compared to immunomagnetic separation. Zheng et al. developed a microfluidic device that separates tumor cells on the basis of differences in cellular size and deformability by using a membrane microfilter comprising microfabricated pores on a flat substrate.20 The membrane microfilter has 10 µm diameter cylindrical holes with a 20 µm centerto-center distance between adjacent pores. In order to provide the fluidic accesses, the membrane microfilter was sandwiched between two pieces of PDMS substrates and integrated with two syringes having needles that penetrated the PDMS pieces. In the case of our device, microcavities were electroformed to produce conical apertures with diameters of 8-11 µm at the top surface and so that adjacent cavities were 60 µm apart. Our experiments with this array confirmed that single cells were neatly trapped on single microcavities and that single cells were not trapped across multiple microcavities.25,26 The conical shape of the microcavities of our array might have prevented the clogging of cavities better than cylindrical pores. Furthermore, for stable continuous flow through the microcavities, a negative pressure was applied to the samples by using a peristaltic pump connected to a vacuum line. These specific features of our device contribute to its better performance in terms of CTC recovery as compared with conventional devices. Recently, we developed an on-chip FISH technique to detect specific mRNAs expressed in cells trapped on the microcavity array.24 This technique would be useful as an additional or alternative method for detecting CTCs that do not express epithelial cell markers.31,32 Furthermore, we have already developed a procedure for isolating single cells from the array by using microcapillaries.25,26 Consequently, recovered tumor cells can be picked up from the size-selective microcavity array and cultured for subsequent studies of CTCs.21,33 This technology will, therefore, not only serve as a rapid and efficient tool for CTC detection but also as a efficient analytical tool in cancer research, and contribute to our understanding of the properties of CTCs and cancer metastasis. CONCLUSIONS In this study, the size selective microcavity array was designed for the size-based separation of CTCs from whole blood. We then integrated the array with a PDMS microfluidic device to enable the recovery, staining, washing, and detection of CTCs via a microfluidic assay. This device enabled rapid and highly efficient recovery of viable tumor cells from whole blood. In addition, EpCAM-negative tumor cells that cannot be isolated by conven(31) Ntouroupi, T. G.; Ashraf, S. Q.; McGregor, S. B.; Turney, B. W.; Seppo, A.; Kim, Y.; Wang, X.; Kilpatrick, M. W.; Tsipouras, P.; Tafas, T.; Bodmer, W. F. Br. J. Cancer 2008, 99, 789–795. (32) Swennenhuis, J. F.; Tibbe, A. G.; Levink, R.; Sipkema, R. C.; Terstappen, L. W. Cytometry, Part A 2009, 75, 520–527. (33) Maheswaran, S.; Sequist, L. V.; Nagrath, S.; Ulkus, L.; Brannigan, B.; Collura, C. V.; Inserra, E.; Diederichs, S.; Iafrate, A. J.; Bell, D. W.; Digumarthy, S.; Muzikansky, A.; Irimia, D.; Settleman, J.; Tompkins, R. G.; Lynch, T. J.; Toner, M.; Haber, D. A. N. Engl. J. Med. 2008, 359, 366–377.

tional immunomagnetic separation were successfully recovered using our device. Thus, our device has potential as an inexpensive yet efficient tool for rapid detection and more detailed studies of CTCs. ACKNOWLEDGMENT This work was supported in part by a Grant-in-Aid for Scientific Research on Young Scientists (B) (No. 20760535) from the Ministry of Education, Culture, Sports, Science and Technology

of Japan. M.H. thanks the Japan Society for the Promotion of Science for financial support. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review May 9, 2010. Accepted June 17, 2010. AC101222X

Analytical Chemistry, Vol. 82, No. 15, August 1, 2010

6635