WO2009051734A1

WO2009051734A1 - Microchip-based devices for capturing circulating tumor cells and methods of their use

Info

Publication number: WO2009051734A1
Application number: PCT/US2008/011785
Authority: WO
Inventors: Sunitha Nagrath; Lecia V. Sequist; Ronald G. Tompkins; Daniel A. Haber; Mehmet Toner; Daniel Irimia; Shyamala Maheswaran
Original assignee: The General Hospital Corporation
Priority date: 2007-10-17
Filing date: 2008-10-16
Publication date: 2009-04-23

Abstract

In general, the invention provides devices and methods for capturing rare cells, e.g., CTCs from blood samples. The devices of the invention are capable of capturing large numbers of viable CTCs in a single step from whole blood without pre-dilution, pre-labeling, pre-fixation, or any other processing steps. The techniques described here and the broader application of microfluidic rare cell capture technology to cancer patients hold significant promise for identifying key biological determinants of blood-borne metastases, and for providing a robust platform aimed at early diagnosis and longitudinal monitoring of cancer.

Description

MICROCHIP-BASED DEVICES FOR CAPTURING CIRCULATING TUMOR CELLS AND METHODS OF THEIR USE

STATEMENT AS TO FEDERALLY SPONSORED RESEARCH

The invention was made with support from the National Institutes of Health under grants P41 EB002503 and U01HL080731. The government has certain rights in the invention.

BACKGROUND OF THE INVENTION

Viable tumor-derived epithelial cells (circulating tumor cells or CTCs) have been identified in peripheral blood from cancer patients and are likely the origin of intractable metastatic disease.^1"4 CTCs represent a potential alternative to invasive biopsies as a source of tumor tissue for the detection, characterization, and monitoring of non-hematologic cancers.^5"8 The ability to identify, isolate, propagate and molecularly characterize CTC subpopulations could further the discovery of cancer stem cell biomarkers and expand the understanding of the biology of metastasis. Current strategies for isolating CTCs are limited to complex analytic approaches that generate very low yield and purity.⁹ CTCs are considered to be rare, making up as few as 1 to 10 cells/mL (~1

CTC per 10 hematologic cells) in the blood of patients with metastatic cancer, hence their isolation has presented a tremendous technical challenge.⁷'⁹'^11"13 Microfluidic lab-on-a-chip devices provide unique opportunities for cell sorting and rare cell detection; they have been successfully used for microfluidic flow cytometry,¹⁴ continuous size-based separation,¹⁵'¹⁶ and chromatographic separation.¹⁷ Despite their success in manipulating microliter amounts of simple liquids in microscale channels,¹⁴'¹⁸'¹⁹ they have thus far had limited capability to deal with the cellular and fluid complexity of large volumes (milliliters) of whole blood samples.^20"22

Other approaches to enrich or sort CTCs from peripheral blood have been previously published including flow cytometry,²⁸ immunomagnetic beads, high throughput optical imaging systems,²⁹ and fiber optic array scanning.¹² Immunomagnetic bead purification¹'¹³'³⁰ is currently the lead technology in the clinical setting with reported success in identifying CTCs in a portion of tested patients with lung, prostate, colon, breast, and pancreas cancer.³⁰ However, this approach isolates small numbers of CTCs (4±24 (mean±s.d.) CTC/mL in lung, 1 l±l 18 in breast, 10±33 in prostate, 1±2 in both colorectal and pancreatic cancers)³⁰ with very low purity (0.01%-0.1%)²⁶, and low yield (-20-60% of patients).³⁰ The level of 'biological noise' associated with the low sensitivity, selectivity, and yield of magnetic bead-based technologies is prohibitive to their capacity to monitor response to treatment in a dynamic fashion and for early cancer detection. Hence, this method has thus far demonstrated clinical utility only as a gross prognostic tool, classifying patients into high- and low-risk categories.¹ Thus, there is a need for additional tools and methods for obtaining circulating tumor cells.

SUMMARY OF THE INVENTION

Accordingly, in one aspect, the invention features a device for capturing circulating, nonhemopoietic tumor cells including a microfluidic channel to which is bound a tumor specific binding agent; and a pump producing a continuous, unidirectional shear stress of 0.1 to 20 dyn/cm² in the channel. The specified shear stress may be localized to those regions where binding agents are located.

In a related aspect, the invention features a device for capturing circulating, nonhemopoietic tumor cells including a microfluidic channel having a plurality of obstacles to each of which is bound a tumor specific binding agent, wherein the obstacles are disposed in an equilateral triangular arrangement with a 10-100 μm distance between obstacles and a 10-100 μm shift between at least two rows. The shift may occur after at least three rows and may be repeated after every three rows. In certain embodiments, the distance between obstacles and the shift are about 50 μm. These arrangements of obstacles may also be employed in conjunction with other devices of the invention.

The invention also features a device for capturing circulating, nonhemopoietic tumor cells including a microfluidic channel to which is bound a tumor specific binding agent and a reservoir having a volume of less than 5 mL, wherein the reservoir and the channel are fluidically connected. The reservoir may have a volume of 10 μL, 20 μL, 50 μL, 100 μL, 250 μL, 500 μL, 1 mL, up to 5 mL. These reservoirs may also be employed in conjunction with other devices of the invention.

For any of the devices of the invention the channels may include a plurality of obstacles, e.g., to which are bound the tumor specific binding agent, or the walls of the channel may be substantially planar, unless otherwise stated. Further, tumor specific binding agent may be disposed in a region of the channel having a volume of 10 μL-20 mL, e.g., 100 μL-15 mL, 100 μL-10 mL, 100 μL-5 mL, 100 μL-1 mL, or 100 μL-0.5 mL. The shear stress produced in a channel, e.g., in conjunction with an attached pump, may in general be between the range of 0.1 to 20 dyn/cm², e.g., less than 15, 10, 5, 1, or 0.5 dyn/cm². Shear stress is not necessarily constant throughout a channel or region containing binding moieties, although it may be. hi other embodiments, sample may be transported through the channel at a rate of 0.1 mL to 30 mL/hr. Typical flow rates will be less than 20, 15, 10, 5, 1, or 0.5 mL/hr. hi general, samples will be passed through the channel at a constant rate, but a constant rate is not required. An exemplary binding agent binds to EpCAM, e.g., anti- EpCAM. Exemplary tumors producing CTCs are those of epithelial origin.

The invention also features kits including any device of the invention and a reagent for obtaining genetic information, e.g., presence or absence of a mutation, level of gene expression, presence of absence of a protein, or level of protein expression, from a circulating, nonhemopoietic tumor cell. Exemplary reagents lyse tumor cells, amplify nucleic acids from tumor cells, bind to a nucleic acid sequence in a tumor cell, or bind to a protein. Particular genetic information that may be obtained includes: for lung cancer, the presence or absence of an EGFR mutation; for prostate cancer, the presence or absence of a TMPRSS2-ERG fusion or level of PSA; and for breast cancer, genetic information on HER2. Exemplary EGFR mutations are a deletion in exon 19, T790M, L858R, L861Q, G719(S, A, or C), S768I, an insertion in exon 20, and a combination thereof. Exemplary TMPRSS2-ERG fusions are T3/E4, T2/E2, Tl/E3_5, T2/E5, Tl/E234_6, T1/E2, T1/E3, T2/E4, T1/E5, T1/E6, Tl/E_IIIa_4, Tl_I/E_IIIb_4, Tl/E_IIIc_4, T5/E4, T4/E4, T4/E5, and a combination thereof. CTCs may also be assayed for cancer stem cell markers, as described herein. Molecular diagnostics for other types of cancer are also encompassed by the invention.

In another aspect, the invention features a method for capturing circulating, nonhemopoietic tumor cells by introducing a blood sample into a device of the invention so that circulating tumor cells in the blood sample bind to the binding agent in the device. Another method of the invention obtains genetic information from a subject with a tumor by introducing a blood sample from the subject into a microfluidic channel to which is bound a tumor specific binding agent so that circulating tumor cells in the blood sample bind to the binding agent; lysing cells bound to the channel; and obtaining genetic information from the lysate, wherein the lysate is not purified prior to obtaining the genetic information. Particular genetic information that may be obtained includes: for lung cancer, the presence or absence of an EGFR mutation; for prostate cancer, the presence or absence of a TMPRSS2-ERG fusion or level of PSA; and for breast cancer, genetic information on HER2. Exemplary EGFR mutations are a deletion in exon 19, T790M, L858R, L861Q, G719(S, A, or C), S768I, an insertion in exon 20, and a combination thereof. Exemplary TMPRSS2-ERG fusions are T3/E4, T2/E2, Tl/E3_5, T2/E5, Tl/E234_6, T1/E2, T1/E3, T2/E4, T1/E5, T1/E6, Tl/E_IIIa_4, Tl_I/E_IIIb_4, Tl/E_IIIc_4, T5/E4, T4/E4, T4/E5, and a combination thereof. This method may further include treatment selection based on the obtained genetic information. The treatment may include, for example, tyrosine kinase inhibitors.

A further method of the invention for capturing circulating, nonhemopoietic tumor cells includes passing a blood sample of less than 4 mL through a microfluidic channel to which is bound a tumor specific binding agent so that circulating tumor cells in said blood sample bind to said binding agent. The invention also features a method of obtaining genetic information from a subject with a tumor by obtaining 1 to 1500 circulating tumor cells from a blood sample of 1 mL or less from the subject and assaying the cells for genetic information. For example, at 5, 10, 25, 50, 100, 200, 500, or 1000 CTCs may be obtained per mL. Yet another method of the invention is for diagnosing a carcinoma in a subject and includes introducing a blood sample from the subject into a microfluidic channel to which is bound a tumor specific binding agent so that circulating tumor cells in the blood sample bind to the binding agent, wherein the presence of one circulating tumor cell is diagnostic for the presence of the carcinoma, and the absence of any circulating tumor cells is diagnostic for the absence of the carcinoma.

For any of these methods, the blood sample assayed may have a volume of 50μL or more, e.g., 100 μL, 250 μL, 500 μL, 1 mL, or 5 mL. In other embodiments, the sample may have a volume of less than 4 mL. Preferred samples are anticoagulated whole blood. CTCs capture during any of the methods of the invention may be enumerated. This enumeration (and any assaying for genetic information) may be repeated over time, e.g., a period of days, weeks, months, or years. A change in the number of tumor cells over time may be indicative of the prognosis of the nonhemopoietic tumor. For example, a negative slope of the number of tumor cells as a function of time is indicative of a positive prognosis, and a positive slope or no slope of the number of tumor cells as a function of time is indicative of a negative prognosis.

CTCs obtained by the methods of the invention may be assayed for genetic information, for example, as described in connection with kits of the invention. Cells may be assayed for changes in genetic information over time as well as or in the alternative to enumeration, e.g., to monitor for the appearance of mutations that indicate a change in therapy is advisable.

In certain embodiments, the blood sample is from a subject at risk for a clinically localized tumor, wherein the presence of tumor cells bound in the device is diagnostic for the clinically localized tumor. Exemplary clinically localized tumors occur in prostate cancer, renal cancer, and bone cancer.

The methods of the invention may also include lysing CTCs bound to the binding agent. In these embodiments, the tumor specific binding agent may be disposed in a region of the microfluidic channel, and the lysing may include applying a lysing agent to a portion of the region where a majority of the circulating tumor cells are bound. In another embodiment, the tumor specific binding agent is disposed in a region of the microfluidic channel, and the lysing includes applying a lysing agent to the entire region. Methods of the invention may further include washing the bound tumor cells at a higher shear stress or volume than that used in the introducing step to increase purity, e.g., by reducing the number on weakly bound or non-specifically bound cells in the device compared to bound CTCs. Information on CTCs obtained using the methods of the invention may also in general be used for diagnosis, prognosis, therapy selection, triage, or long term surveillance of subjects. In particular, analysis of CTCs from prostate cancer patients may be used to determine whether the tumor is indolent or aggressive.

Further features and advantages will be apparent from the following description, the drawings, and the claims.

By "lysis" is meant disruption of a cell membrane sufficient to allow extracellular access to cellular nucleic acids. Lysis may occur by any means, e.g., chemical, thermal, optical, or mechanical.

By "clinically localized tumor" is meant a tumor confined to its tissue of origin, as determined by radiographic measures.

By "micro fluidic" is meant having at least one dimension of less than 1 mm.

By "nonhemopoietic cell" is meant any cell not of hemopoietic origin, that is excluding blood cells and immune cells. Examples of nonhemopoietic cells include epithelial cells, endothelial cells, neurons, hepatocytes, nephrons, glial cells, muscle cells, skin cells, adipocytes, fibroblasts, chondrocytes, osteocytes, and osteoblasts.

By "tumor specific binding agent" is meant any agent that binds to a nonhemopoietic cell that can form a tumor, either benign or malignant. The binding agent may bind to a cell surface marker that is specific for a type of cell that can form a tumor and that is not normally found in circulating blood. In an alternative, the binding agent may bind to a cell surface marker that is specific for a transformed cell. Such agents may also bind to healthy cells circulating in blood from non-pathogenic origins, e.g., venipuncture or trauma.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1. Isolation of CTCs from whole blood using a microfluidic device.

(a) The workstation setup for CTC separation. The sample is continually mixed on a rocker, and pumped through the chip using a pneumatic pressure regulated pump, (b) The CTC-chip with microposts etched in silicon, (c) Whole blood flowing through the microfluidic device, (d) SEM image of a captured NCI-Hl 650 lung cancer cell spiked into blood (pseudo colored with red). The inset shows a high magnification view of the cell.

Figure 2. CTC capture and enumeration, (a) Capture yield as a function of flow rate. Data shown represent measurements averaged over three devices, and each error bar represents the standard error of the mean, (b) Capture yields from buffer spiked with 100 cells/mL of four different cells lines: prostate (PC3-9) , breast (SkBr- 3), bladder (T-24), and NSCLC (NCI-H 1650). Each data point was repeated in at least 3 devices. The error bars represent standard deviations in measurements within each experiment, (c) Regression analysis of capture efficiency for various target cell concentrations, comparing whole blood to lysed blood samples. The plot represents number of cells spiked versus number of cells recovered, (d-k) Higher magnification (2Ox) images of captured CTCs and hematologic cells from NSCLC patients, stained with DAPI, CK, and CD45. Merged images identify CTCs in panels d-g and hematologic cells in panels h-k.

Figure 3. Enumeration of CTCs from cancer patients, (a) Summary of samples and CTC counts per 1 mL of blood in patients with various advanced cancers and localized prostate cancer, (b) Frequency of CTCs per 1 mL of blood, by diagnosis. The box plot presents the median, lower, and upper quartiles (25th, 75th percentiles). Data points that lie outside the 10th and 90th percentiles are shown as outliers, (c) Purity of captured CTCs (ratio of CTCs to total nucleated cells), by diagnosis, (d-i) Serial CTC assessment. CTC quantity (red), and tumor size (blue) measured as the unidimensional sum of all significant tumor sites on CT scan, are well correlated over the course of anticancer treatment for nine patients. Six of them are shown here, whose diagnoses and specific therapies were: (d) NSCLC: 2nd-line pemetrexed. (e) NSCLC: lst-line carboplatin, paclitaxel. (f) Colorectal: lst-line 5FU, irinotecan. (g) Pancreatic: lst-line gemcitabine, bevacizumab. (h) Pancreatic: lst-line gemcitabine. (i) Pancreatic: lst-line gemcitabine, erlotinib. Note, baseline CT scans were prior to therapy initiation and CTC measurements began at or shortly after the first treatment.

Figure 4. Characterization of CTCs with tumor specific molecular markers (a-b) CTCs from a prostate cancer patient stained positive for DAPI and PSA expression, (c) RT-PCR amplification of PSA transcript is seen in two patients with prostate cancer (PCa), but not in two patients with lung cancer (LuCa), and only in blood fractions enriched for CTCs as opposed to non-enriched fractions (non- CTC). (d-e) CTCs from a NSCLC patient stained for DAPI and TTF-I. (f) RT-PCR shows the expression of TTF-I in two patients with lung cancer (LuCa), but not in two patients with prostate cancer (PCa), and only when RNA was eluted from blood fractions enriched for CTCs as opposed to non-enriched fractions (non-CTC).

Figure 5. Microfluidic approach to isolate circulating tumor cells, (a) One- step process for point-of-care isolation of CTCs from peripheral blood, (b) Schematic of the manifold assembly. The microfluidic chip is sealed from above with a biological grade adhesive tape and placed in the manifold, (c) Scanning electron micrograph (SEM) image of the microposts array.

Figure 6. Design criteria and computational analysis of hydrodynamics in the microfluidic chip, (a) Comparison of the hydrodynamic efficiency of different post array arrangements by distance between the posts using square, diagonal square, and equilateral triangular arrays, (b) Computational analysis of the micropost array. Cell trajectories (solid lines) are based on particle tracings and the end positions of the cells are indicated by red dots. Cells starting from positions between the microposts followed the streamline and never came into contact with the microposts, an observation predicting detrimental impact on target cell capture. To address this concern, a vertical periodic shift of 50μm was incorporated after every 3 rows of microposts, forcing cells to change their trajectory and hence enhancing the probability of collision with microposts. (c) Flow profile for an equilateral triangular array: Velocity vectors and surface contours colored by magnitude of velocity, (d) Shear stress, x-component along the surface of the post. Figure 7. Shear stress studies using linear Hele-Shaw chambers, (a-c)

Phase contrast images of cells adhered to the anti-EpCAM coated surface at three different locations in the chamber. The area covered in each image is lmm x lmm. (d) Schematic and geometry of a Hele-Shaw chamber, (e) Linear shear stress profile along the chamber length, (f) The effect of shear stress on the cell adhesion through EpCAM antibody- antigen binding. Error bars denote standard errors for each point based on four repetitions.

Figure 8. Spiked cell experiments using cultured NCI-H1650 cells (n=3).

Spiked cells are pre-labeled with cell tracker dye to fluoresce in orange. The scale bar indicates 1 OOμm. (a) Absence of cell capture on a non-functionalized control device, (b) Cell capture on a device functionalized with anti-EpCAM. (c) Higher power magnification of the captured cancer cells, in orange.

Dynamics of cell capture from patient samples, (d) Fluorescent images of the CTCs captured from the whole blood of a NSCLC patient. White arrows indicate

CTCs, green arrows indicate leukocytes. The top left inset shows a magnified view of a CTC and the lower left inset shows a magnified view of a leukocyte, (e) Comparison of capture efficiency between whole blood and lysed RBC blood, (f) Concordance experiment to test experimental variability of split samples analyzed under identical conditions.

Figure 9. Gallery of CTC images captured from various metastatic epithelial cancers. The first column shows low magnification fluorescent images of CK+ cells in lung, prostate, pancreas, and colon cancers. The scale bar indicates 1 OOμm. The remaining columns show higher magnification images in which the scale bar indicates 1 Oμm. (b-e) NSCLC CTC stained DAPI+, CK+, CD45-, and the merged image of DAPI and CK fluorescent images, (g-j) Prostate cancer CTC cluster (2 cells) stained DAPI+, CK+, CD45-, and the merged image of DAPI and CK fluorescent images, (l-o) Pancreatic cancer CTC stained DAPI+, CK+, CD45-, and the merged image of DAPI and CK fluorescent images, (q-t) Colon cancer CTC cluster (3 cells) stained DAPI+, CK+, CD45-, and the merged image of DAPI and CK fluorescent images, (v-y) Fluorescent images of a leukocyte (PBMC) stained DAPI+, CK-, CD45+, and the merged image of DAPI and CD45.

Figure 10. Serial assessments of patients using both the CTC-chip and standard radiographic monitoring. CTC quantity (cells/mL) depicted in red, and tumor size (sum of measurable diameters in cm) in blue, are well correlated over the course of anti-cancer treatments for nine individual patients. Three of the patients are shown here, whose diagnoses and specific therapies are as follows: (a) NSCLC: lst- line carboplatin, paclitaxel, and an experimental agent, (b) Colorectal cancer: lst-line infusional 5FU, oxaliplatin, and bevacizumab. One of the samples had 0 CTC/mL, which may be due to insufficient volume, (c) Esophageal cancer: lst-line cisplatin and irinotecan.

Figure 11 is a series of images of gels and sequencing runs from prostate cancer molecular diagnostics. Figure 12 is a schematic depiction of lysis of cells in specific regions of a device of the invention.

Figure 13A is a schematic representation of CTC-chip analysis. Whole blood is collected from the patient and passed through the CTC-chip, containing 78,000 microposts coated with antibody to the epithelial surface antigen EpCAM. For CTC emumeration, captured cells are stained in situ using antibody to cytokeratin; for molecular studies, captured cells are lysed on the chip and eluted DNA undergoes the desired analysis.

Figure 13B is a fluorescent photomicrograph of CTCs captured against the sides of the functionalized microposts (dashed lines superimposed on images). DNA staining is used to identify all nucleated cells within a field. Cells here are also stained with rhodamine-conjugated antibody to cytokeratin (red) or fluorescein-conjugated antibody to EGFR (green); magnification 200X

Figure 13C is a scanning electron microscopic image of a single CTC captured from a patient with NSCLC (arrow).

Figure 14 A is a table showing allele-specific SARMS analysis of EGFR mutations in NSCLC tumor samples. Tumor specimens subjected to standard EGFR sequencing analysis were reanalyzed using the highly sensitive allele specific SARMS assay, and correlated with clinical outcome. Abbreviations: SARMS = Scorpion Amplification Refractory Mutation System, mo = months, Del = deletion, Und = undetected, SD = stable disease, PR = partial response, CR = complete response, PD = progressive disease, NA = not applicable

Figure 14B is a graph showing progression-free survival estimates for patients treated with gefitinib or erlotinib therapy. Figure 15A is a series of graphs showing serial analyses of CTC numbers, genotypes, and radiographic tumor burden. Dynamic quantification of isolated CTCs per milliliter (dotted) and radiographic tumor burden in centimeters (triangles) in four patients with EGFZ?-mutant NSCLC, measured at multiple time points during the course of treatment with gefinitib, chemotherapy or experimental agents (Expt). Time in days is displayed along the x-axis. Duration of each therapy is indicated by blue bars. CTC-genotypes determined by SARMS assay are shown in boxes at various time points. Bracketed mutations indicate those present at low allele frequencies. Figure 15B is a series of graphs showing SARMS analysis of EGFR genotypes in patient 9 demonstrating increased allelic abundance of the T790M drug resistance allele at the time of disease progression. Arrows denote the cutoff for amplification cycles (Ct) required for detection of the primary mutation (Deletion) and the T790M mutation, compared with the Exon 2 control. Δ Ct reflects the difference in allele frequency between the primary mutation and T790M in the tumor tissue biopsy, the CTCs isolated at the time of gefitinib-responsive disease, and the CTCs isolated at the time of disease progression.

Figure 15C are graphs showing nucleotide sequencing tracings from patient 2 in which the tumor tissue biopsy analysis demonstrates a T751_I759delinsS mutation that is distinct from the Del 746 750 mutation present in the CTC analysis. The 27 nucleotides deleted in the tumor (upper box, upper panel) are present in the CTC DNA (lower box, upper panel), while the 15 nucleotide deletion in the CTC DNA (upper box, lower panel) is present in the tumor DNA (lower box, lower panel). The CTC tracing represents direct nucleotide sequencing of DNA lysed from cells captured on the CTC-chip, indicating a high degree of captured tumor cell purity.

DETAILED DESCRIPTION OF THE INVENTION We have developed a unique microfluidic platform (the "CTC-chip") capable of efficient and selective separation of viable CTCs from peripheral whole blood samples, preferably mediated by the interaction of target CTCs with antibody

(EpCAM)-coated microposts under precisely controlled laminar flow conditions, and without requisite sample pre-labeling or processing. Surprisingly, we have been able to isolate CTCs in high numbers from whole blood samples using the devices described herein. The present invention provides a novel and effective tool for accurate identification and measurement of CTCs in patients with cancer. It has broad implications in advancing both cancer biology research and clinical cancer management, including the detection, diagnosis, and monitoring of cancer.¹

The devices of the invention are unique in that they are able to sort rare cells directly from whole blood in a single step. From a technical perspective, this is possible because the device is the first microfluidic device that can successfully process milliliter volumes of whole blood, although, as described herein, the high number of CTCs recovered using the devices allows for the use of lower volumes of blood. This contrasts with magnetic bead-based systems ° that require multiple "bulk" semi-automated preparatory steps (centrifugation, washing, and incubation), resulting in loss and/or destruction of a significant proportion of the rare cells. In addition to its simplicity, the present devices are readily adaptable for potential use in various clinical scenarios, including changes in throughput and in the functionalized binding agent on the channel, allowing capture of other types of rare circulating cells. The invention's one-step potential and versatility make it conducive to point-of-care use and rapid integration into clinical practice.

The devices are also distinctive in that their gentle nature (e.g., maximum shear stress may be 0.4 dynes/cm²) allows for isolation of viable cells, whereas magnetic bead-based approaches can only isolate fixed, nonviable cells.³⁰ The stationary nature of the captured cells in the present invention allows wash-out of non-specifically bound cells, e.g., leukocytes, resulting in a 10⁶-fold enrichment, a level of purity that is two orders of magnitude higher than existing technologies. The capacity to isolate concentrated, viable CTCs makes the present invention an ideal tool for molecular access to rare CTC subpopulations such as metastatic precursor cells or cancer stem cells.

In particular, the devices and methods of the invention achieve capture of CTCs at high sensitivity (defined as the percentage of patients having a tumor identified as having CTCs); high specificity (defined as the percentage of patients not having a tumor identified as not having CTCs); and high purity (defined as the percentage of CTCs relative to other cells retained by the device). These levels of sensitivity, specificity, and purity are surprising in comparison to previous devices and methods for capturing CTCs. The devices and methods of the invention are also highly efficient, capturing on average 155 ± 236 (mean±s.d.) CTCs/mL for NSCLC, 16 to 292 (86 ± 78) for metastatic prostate, 25 to 174 (94 ± 63) among localized prostate cancer, 9 to 831 (196 ± 228) for pancreatic, 5 to 176 (79 ± 52) for breast, and 42 to 375 (121 ± 127) for colorectal, well above that typically obtained with other techniques, such as magnetic enrichment. In addition, the invention is capable of utilizing whole, anticoagulated blood (although not limited thereto) without any further sample treatment steps, such as dilution, centrifugation, red blood cell lysis, cell fixation, or cell labeling. Devices

In its simplest embodiment, a CTC-chip of the invention includes a microfluidic channel having tumor specific binding agents bound to the surface of the channel and which is capable of supporting fluid flow at the desired shear stress. As blood flows through the channel, CTCs in the blood come into contact with the tumor specific binding agents and become immobilized on the surface of the channel. The microfluidic channel may include posts or other interior structure to increase the surface area of the channel and, in some instances, increase the probability that a given cell passing through the channel will come into contact with a binding agent. Alternatively, the channel walls are substantially planar. When channel walls are substantially planar, the height of the channel may be designed so that CTCs readily contact the binding moieties.

The exact device geometry will be determined based on the assay. Devices may, or may not, include regions that allow for optical or visual inspection of the channels. Fluid pumps capable of producing desired shear stress in the device are also known in the art. Examples of pumps include syringe pumps, peristaltic pumps, and vacuum sources. Methods for coupling pumps to devices are known in the art. The device may be configured for substantially constant shear stress in any given channel or variable shear stress in a given channel. Exemplary devices are described herein.

An exemplary device that can efficiently and reproducibly isolate CTCs from the blood of patients with common epithelial tumors is shown in Fig. 1 and Fig. 5. The CTC-chip (Fig. Ib) includes an array of microposts (Fig. 5c) that are chemically functionalized with antiepithelial cell adhesion molecule (EpCAM) antibodies. EpCAM provides specificity for CTC capture from unfractionated blood as it is frequently overexpressed by carcinomas of lung, colorectal, breast, prostate, head and neck, and hepatic origin.²³"²⁴ A description of a manifold for use with a CTC chip is found in International Publication No. WO 2006/108101. Two essential parameters that determine the efficiency of cell capture on the

CTC-chip are (1) flow velocity, as it influences the duration of cell-channel (e.g., post) contact, and (2) shear force, which must be sufficiently low to ensure maximum cell-channel (e.g., post) attachment. In an example of the optimization of these parameters, we employed theoretical analyses characterizing the interaction of cells with microposts distributed within the flow path (Fig. 5c; Fig. 6). Briefly, the simulation indicated an equilateral triangular arrangement with a 50μm distance between microposts and a 50μm shift after every 3 rows to be the most efficient geometric arrangement. For a given volumetric flow rate of lmL/h through the device, the maximum shear stress experienced by a cell near the micropost surface was estimated to be 0.4 dynes/cm² at θ=68°, and the expected maximum velocity was 460μm/s (Fig. 6b-d), within the range facilitating maximum cell attachment according to linear shear stress chamber studies (Fig. 7). Based on the simulation results, we fabricated an exemplary CTC-chip containing an array of 78,000 microposts within a 970 mm surface. Other configurations may also be employed. For example, devices having low minima of stress near obstacle, but high shear stress at the boundary wall of device, which allows high throughput without impacting capture. The volume of the channel or the region having the binding agents may also be altered depending on the volume of the blood sample being employed. The volume of the channel (defined as that portion through which cells may pass) may be larger than the size of the sample. In such cases, a transporting fluid, which may be miscible or immiscible with the sample, may be employed to ensure that the sample comes into contact with the binding agents. Suitable transporting fluids include air and buffer. At least two variables can be manipulated to control the shear stress applied to the channel: the cross sectional area of the chamber and the fluid pressure applied to the chamber. Other factors may be manipulated to control the amount of shear stress necessary to allow binding of desired analytes and to prevent binding of undesired analytes, e.g., the binding moiety employed and the density of the binding moiety in the channel. Pumps that produce suitable shear forces in combination with channels of the invention preferably produce a unidirectional shear stress, i.e., there is no reversal of direction of flow, and/or substantially constant shear stress. Either unidirectional or substantially constant shear stress may be maintained only during the time in which a sample is passed through a channel. Washing or labeling steps after target cells have bound to the device may utilize reversals of flow or changes in shear stress. In another embodiment, the shear stress is not necessarily constant but is kept below a critical value for the duration of binding of target cells to a channel. The flow rate will typically be between 0.1 mL to 30 mL/hr. Dilution of blood may be employed at high flow rates, e.g., above 10 mL/hr.

The channel may include one or more binding agents (e.g., 1, 2, 3, 4, 5, or more). Multiple binding agents may bind to the same or different cells and may be placed in the same or different channels. For example, binding agents to multiple cell surface markers that occur on a desired cell may be disposed in one channel.

In another embodiment, channels are arranged in series, (e.g., 2, 3, 4, 5, or more channels). In this embodiment, each channel isolates one or more types of cells, which may or may not be the cells of interest. When multiple channels are arranged in series, the shear stress applied to each of the channels can be different (achieved for example by varying the cross sectional area of the channels) or the shear stress can be the same. Also, when multiple channels are arranged in series, each channel can contain binding agents that bind to different cell surface markers or the same cell surface markers. The methods may also be employed to isolate various types of analytes in parallel, e.g., by passing aliquots of the same sample through separate devices or one device including multiple channels in parallel. Different samples may also be assayed in parallel.

Devices of the invention may be fabricated using techniques known in the art.

The fabrication techniques employed will depend on the material used to make the device. Examples of fabrication techniques include molding, photolithography, electroforming, and machining. Exemplary materials include glass, polymers (e.g., polystyrene, silicones such as polydimethylsiloxane, epoxy, polymethylmethacrylate, and urethanes), silicon and other semiconductors, and metals.

Devices of the invention may be combined with pumps, detectors, and other laboratory components. The devices of the invention may include one or more inlets, e.g., to deliver two or more different fluids simultaneously or at different times.

Examples of fluids that may be introduced into a device include washing buffers, e.g., to remove nonspecifically bound cells or unused reagents, lysing reagents, or labeling reagents, e.g., extracellular or intracellular stains. In one embodiment, devices of the invention are designed to have removable covers to allow access to all of the region in which cell may be bound or a portion thereof. With these devices, it is possible to apply reagents, e.g., labeling reagents or lysing reagents, to specific regions. Individual cells may also be removed from such devices, e.g., using a pipette. In other embodiments, the device has more than one inlet and outlet to allow the introduction of more than one fluid to the device, typically at different times. By having multiple inlets and corresponding outlets, fluids may be introduced simultaneously in the device to lyse or otherwise manipulate bound cells in specified regions. The size of these regions may be controlled based on the location of the inlets and outlets and the relative volumetric flow rates from the inlets and outlets, under the principles of laminar flow occurring in microfluidic channels.

Circulating Cells and Binding Agents

The devices may in principle be employed for any rare cell separation that employs a selective binding agent, i.e., an agent that binds to the target cell and not to (or at least not to the same extent) to a non-target cell. A preferred rare cell is a circulating tumor cell of epithelial origin from peripheral blood. Other rare cells include organisms potentially found in peripheral blood (e.g., bacteria, viruses, protists, or fungi), other nonhemopoietic cells not normally found in blood (e.g., endothelial cells or fetal cells), and even cells of hemopoietic origin (e.g., platelets, sickle cell red blood cells, and subpopulations of leukocytes). The binding agent or agents employed will depend on the type of cell or cells being targeted. In general, specific binding agents for these cells are known in the art. Exemplary types of binding agents include antibodies, antibody fragments (e.g., Fc fragments), oligo- or polypeptides, nucleic acids, cellular receptors, ligands, aptamers, MHC-peptide monomers or oligomers, biotin, avidin, oligonucleotides, coordination complexes, synthetic polymers, and carbohydrates. Binding moieties may be attached to chambers using methods known in the art. The method employed will depend on the binding moiety and the material used to construct the device. Examples of attachment methods include non-specific adsorption to the surface, either of the binding moiety or a compound to which the binding moiety is attached or chemical binding, e.g., through self assembled monolayers or silane chemistry.

An exemplary binding agent is anti-EpCAM antibody, which is specific for epithelial cells. As described, circulating epithelial cells may provide clinical and diagnostic information relevant to tumors, even those considered clinically localized. Cancers that may be detected using the devices of the invention include prostate, lung, adenocarcinoma, adenoma, adrenal cancer, basal cell carcinoma, bone cancer, brain cancer, breast cancer, bronchi cancer, cervical dysplasia, colon cancer, epidermoid carcinoma, Ewing's sarcoma, gallbladder cancer, gallstone tumor, giant cell tumor, glioblastoma multiforma, head cancer, hyperplasia, hyperplastic corneal nerve tumor, in situ carcinoma, intestinal ganglioneuroma, islet cell tumor, Kaposi's sarcoma, kidney cancer, larynx cancer, leiomyoma tumor, liver cancer, malignant carcinoid, malignant hypercalcemia, malignant melanomas, marfanoid habitus tumor, medullary carcinoma, metastatic skin carcinoma, mucosal neuromas, mycosis fungoide, neck cancer, neural tissue cancer, neuroblastoma, osteogenic sarcoma, osteosarcoma, ovarian tumor, pancreas cancer, parathyroid cancer, pheochromocytoma, primary brain tumor, rectum cancer, renal cell tumor, retinoblastoma, rhabdomyosarcoma, seminoma, skin cancer, small-cell lung tumor, soft tissue sarcoma, squamous cell carcinoma, stomach cancer, thyroid cancer, topical skin lesion, veticulum cell sarcoma, or WiIm' s tumor.

Methods of Use

In addition to methods of isolating various cells from a sample, the invention provides methods in which the cells isolated may be used to provide additional information. In particular, cells isolated using the methods and devices of the invention can be further assayed using additional methods of the invention. In one embodiment, cells that are isolated using the methods and devices of the invention are counted. Cells can be counting by any method known in the art, including optical, e.g., visual inspection, automated counting, microscopy based detection, and FACS, and electrical detection, e.g., Coulter counters. Counting of the cells, or other analytes, isolated using the methods and devices of the invention can be useful for diagnosing diseases, monitoring the progress of disease, and monitoring or determining the efficacy of a treatment. Cell, or other analyte, counting may also be of use in non-medical applications, e.g., for determination of the amount, presence, or type of contaminants in environmental samples (e.g., water, air, and soil), pharmaceuticals, food, or cosmetics.

In another embodiment, cells isolated using the methods and devices of the invention can be lysed, and one or more properties of the cells, or portions thereof, can be measured. Examples of biological properties that can be measured in isolated cells include mRNA expression, protein expression, and DNA quantification. Additionally, the DNA of cells isolated by the methods of the invention can be sequenced, or certain sequence characteristics (e.g., polymorphisms and chromosomal abnormalities) can be identified using standard techniques, e.g., FISH or PCR. The chemical components of cells, and other analytes, may also be assayed after isolation. Cells may also be assayed without lysis, e.g., using extracellular or intracellular stains or by other observation, e.g., morphology or growth characteristics in various media. When lysis is employed, CTCs may be lysed while still bound to the chip, e.g., with any other cells nonspecifically retained. The ability to lyse CTCs on chip and obtain useful genetic information is made possible by the high purity of samples (typically greater than 50%) using the devices and methods of the invention.

Particular genetic information that may be obtained from a tumor cell captured by a CTC-chip includes identification or enumeration of particular genomic DNA, cDNA, or mRNA sequences; identification or enumeration of cell surface markers; and identification or enumeration of proteins or other intracellular contents that is indicative of the type or presence of a particular tumor. For example, CTCs may be analyzed to determine the tissue of origin, the stage or severity of disease, or the susceptibility to a particular treatment. For example, a diagnostic indicator for lung cancer and other cancers is the presence or absence of certain mutations in EGFR

(see, e.g., International Publication WO 2005/094357). Another example is assaying for FIER2 to determine prognosis or treatment regimen in breast cancer. (Gynecol Obstet Fertil. 2006 Jul-Aug;34(7-8):638-46. Epub 2006 JuI 28 ) also Breast Cancer Res. 2007 Oct 26;9(5):R74. We have also described herein that prostate specific antigen (PSA), a known marker for prostate cancer, can be detected in CTCs. It is detectable by protein expression staining using antibodies that measure PSA expression within cells captured on the chip (immunofluorescence and immunohistochemistry), and it is also detectable by reverse transcription polymerase chain reaction (RT-PCR) in RNA extracted from the CTCs. While measurement of PSA protein levels in blood is a standard marker for advanced prostate cancer, the ability to measure PSA levels within CTCs brings a significant measure of increased sensitivity. This increased sensitivity may be particularly important in the detection of invasive disease, since borderline PSA protein levels in the blood often lead to confusion, whereas presence of PSA-expressing CTCs in the blood may be an indicator of invasive disease.

In addition, the prostate cancer-specific chromosome translocation, TMPRSS2-ERG, has been identified in up to 50% of advanced prostate cancers, and a number of less frequent but related chromosome translocations have been identified in additional cases. The translocation results in the abnormal expression of a transcription factor (gene regulator), which, as a result of the translocation, is now driven by androgen. As such, expression of androgen (testosterone) activates the TMPRS S2 promoter and triggers the expression of the transcription factor (ERG) in prostate cancer, a characteristic that appears to contribute to invasive growth (see J. Clark, et al. Oncogene 2006; 1-7). We have successfully identified the presence of the expected TMPRSS2-ERG translocation in CTCs from patients with metastatic prostate cancer, using RT-PCR analysis of RNA from the CTCs. The significance of this observation is that expression of the translocation in CTCs is more common than it is in the original primary prostate cancer, and it serves as a marker for invasive prostate cancer. Since one of the most difficult clinical criteria in the treatment of prostate cancer is the differentiation of indolent prostate cancer from a more aggressive disease, our ability to detect this invasion-associated chromosome translocation in blood samples offers a powerful tool to distinguish invasive prostate cancer worthy of aggressive therapy (surgery, radiation and/or hormonal treatment) from indolent forms that may not require such treatments.

CTCs captured by the devices and methods described herein may also be assayed for the presence of markers indicative of cancer stem cells. Examples of such markers include CDl 33, CD44, CD24, epithelial-specific antigen (ESA), Nanog, and BMIl.

EXAMPLES

The following non-limiting examples provide further features and embodiments of the invention.

Example 1. Metastatic lung, prostate, pancreas, breast, and colon cancer

A CTC -chip successfully identified CTCs in the peripheral blood of patients with metastatic lung, prostate, pancreas, breast, and colon cancer in 115 of 116 (99%) samples, with a range of 5-1281 CTC/mL and approximately 50% purity. In addition, CTCs were isolated in 7/7 patients with early stage prostate cancer. Given the high sensitivity and specificity of the CTC-chip, we tested its potential utility in monitoring response to anti-cancer therapy. In a small cohort of patients with metastatic cancer undergoing systemic treatment, temporal changes in CTC numbers correlated reasonably well with the clinical course of disease as measured by standard radiographic methods.

Efficiency of Capture To determine the efficiency of capture, we spiked non-small cell lung cancer

(NSCLC) cells (NCI-Hl 650) into phosphate buffered saline (PBS) at 100 cells/mL and captured the spiked cancer cells using a CTC-chip, described below. NSCLC cells were visually evident about EpCAM-coated microposts, whereas no cancer cells were seen following flow through uncoated posts (Fig. 8a-c). The calculated capture efficiency was 65% and decreased significantly at flow rates above 2.5 mL/h (Fig. 2a), presumably due to increased shear stress, consistent with our simulation predictions. The efficiency of capture was not enhanced at flow rates less than 0.75mL/h, leading us to select a flow rate of 1-2 mL/h for subsequent studies.

To determine the effect of cellular EpCAM expression on efficiency of CTC capture, we compared capture rates among cancer cell-lines with varied EpCAM expression, including NSCLC NCI-H 1650 cells and breast cancer SKBr-3 cells with >500,000 antigens/cell, prostate cancer PC3-9 cells with approximately 50,000 antigens/cell, and bladder cancer T-24 cells with approximately 2,000 antigens/cell.²⁵ Each cell-line was spiked into PBS at a concentration of 100 cells/mL and analyzed, resulting in a mean capture yield of > 65% in all cases (Fig. 2b). Interestingly T-24 cells with relatively low EpCAM expression were captured as efficiently as high-level antigen expressing cells. We believe this is due to the augmented cell-substrate interactions inherent within a CTC-chip.

To evaluate cell capture under more physiologic conditions, we conducted a series of experiments in which NCI-H1650 cells were spiked into whole blood from healthy donors. Suspensions at concentrations ranging from 50 to 50,000 tumor cells/mL of whole blood were analyzed, yielding recovery rates of >60% in all cases

(Fig. 2c). To assess the potential steric hindrance of red blood cells in the flow path, these studies were repeated using lysed blood from healthy donors. Capture rates were comparable under both conditions (^=0.99, Fig. 2c). Similar results were obtained using whole blood and lysed samples from NSCLC patients (Fig. S4 d-e). We thus concluded that a CTC-chip does not require blood sample pre-processing.

Having optimized the CTC-chip with controlled quantities of cancer-derived cells, we tested its capacity to capture CTCs from whole blood samples donated by cancer patients. A total of 116 samples from 68 patients with epithelial cancers including NSCLC (n=55), prostate (n=26), pancreas (n=15), breast (n=10), and colon (n=10) were studied. The majority of patients had metastatic disease; however, 7 of 26 subjects with prostate cancer had untreated, clinically localized disease with specimens collected prior to prostatectomy with curative intent (Table 1). The average volume of blood analyzed was 2.7 mL/sample (range 0.9 to 5.1 mL). We also examined samples from 20 healthy individuals (3.0±0.4 mL (mean±s.d.) of blood/subject) as controls (Table 2).

Table 1. Quantification of circulating tumor cells per mL of blood among 116 samples from patients with epithelial cancers including NSCLC (n=55), prostate cancer (n=26), pancreatic cancer (n=15), breast cancer (N=IO) and colorectal cancer (n=10).

CTCs captured from a group of patient samples were identified using a comprehensive image analysis algorithm, consisting of staining with DAPI for DNA content, rhodamine-conjugated anti-cytokeratin (CK) antibodies for epithelial cells, and fluorescein-conjugated anti-CD45 antibodies for hematologic cells (Fig. 3d-k, Fig. 9) in order to confirm the validity of CK stain. Thereafter, cells captured by anti- EpCAM-coated microposts and staining for CK were scored as CTCs, while CD45- positive cells were scored as contaminating normal hematologic cells. The morphologic characteristics exhibited by the captured CTCs were consistent with malignant cells, including large cellular size with a high nuclear: cytoplasmic ratios and visible nucleoli (Fig. 2d-g). The mean viability of captured cells was 98.5 ± 2.3% (mean±s.d.), determined by assessing cell membrane integrity in 10 high-power fields of view/CTC-chip in 4 samples obtained from lung (n=2) and prostate (n=2) cancer patients. CTCs were identified in 115 of 116 (>99%) patient samples analyzed, with the single negative specimen being a small volume sample (0.9mL) from a colorectal patient. The number of CTCs isolated ranged from 5 to 1,281/mL for NSCLC (155 ± 236 (mean±s.d.) CTCs/mL), 16 to 292 (86 ± 78) for metastatic prostate, 25 to 174 (94 ± 63) among localized prostate cancer, 9 to 831 (196 ± 228) for pancreatic, 5 to 176 (79 ± 52) for breast, and 42 to 375 (121 ± 127) for colorectal (Fig. 3a, b). The identification of CTCs in subjects with clinically localized prostate cancer at numbers approximating those in metastatic prostate cancer patients is a novel finding enabled by the high sensitivity of our technique. The average purity of capture, as defined by the ratio of CK+ to CD45+ cells, was 52% in NSCLC, 49% in metastatic prostate, 53% in localized prostate, 53% in pancreatic, 60% in breast, and 67% in colon (Fig. 3c). None of the 20 healthy subjects had any identifiable CTCs (Table 2). Based on these results we calculated the sensitivity (99.1%) and specificity (100%) of a CTC- chip across all five cancers. Finally, we evaluated the reproducibility of CTC capture using split samples and showed high experimental reproducibility (r²=0.98, Fig. 8f).

Table 2. Quantification of circulating tumor cells per mL of blood in healthy subjects. None of the samples from healthy subjects had any detectable CTCs.

Molecular Analysis

To determine whether captured cells are suitable for subsequent molecular analyses, we tested expression of two tumor-specific markers: prostate specific antigen (PSA) in prostate cancer and thyroid transcription factor- 1 (TTF-I) in lung adenocarcinoma. Specific expression of these markers was evident by immunostaining (Fig. 4a-b,d-e), and confirmed by direct lysis of captured CTCs on the microchip and RT-PCR amplification of the individual transcripts (Fig. 4c,f). Considering the -50% purity of captured viable CTCs (two orders of magnitude higher than currently available technologies),²⁶ the CTC-chip provides a powerful opportunity for CTC-based molecular analyses.

Clinical Examples

To demonstrate the unique clinical potential of our approach, we evaluated the ability of changes in CTC burden to predict changes in tumor volume in cancer patients undergoing anti-cancer treatments. Patients with advanced epithelial-based malignancies that were commencing a first or second line systemic treatment regimen were eligible. Blood samples were collected at baseline and subsequent clinic visits; the exact follow-up schedule varied between patients. Computed tomograms (CT scans) were performed at baseline and at regular intervals according to standard clinical practice. For each CT scan, we calculated the sum of the unidimensional size in centimeters of all significant and measurable tumor sites, as per the RECIST standardized system.²⁷ All patients with baseline and at least one follow-up CTC sample and CT scan were analyzed, including patients with NSCLC (n=3), colorectal (n=2), pancreatic (n=3), and esophageal cancer (n=l) (Fig. 3d-i). The absolute number of CTCs captured did not necessarily correspond with tumor size, and may be influenced by other factors. The correlation between percent change in CTC quantity and percent change in tumor size, from first to last measurement, was analyzed over all 9 patients, and yielded a Pearson's correlation coefficient of 0.68 (p= 0.03). These results from a small cohort of patients indicate that CTC quantity correlates reasonably well with clinical response and clinical non-response to treatment (Fig. 3 d- i, Fig. 10), suggesting that monitoring treatment response using a CTC-chip may be a powerful tool enabling accurate, early decision-making. The clinical impact of such an approach could be large, and provide the ability to decrease patient exposure to toxicities from ineffective therapies and guide them toward the treatments most active against their specific tumor. The high sensitivity (1 target cell in 1 billion blood cells), selectivity (>47% purity), and yield (99%) of a CTC-chip makes it ideally suited for real-time monitoring of response to cancer therapy. Methods summary

The microfluidic system described in these examples (Fig. Ia) consists of a microfluidic chip etched in silicon (Fig. Ib), a manifold to enclose the chip (Fig. Ic, Fig. 5b), and a pneumatic pump (Fig. 1 a) to establish the flow through the chip (Fig. Ic). The schematic of the microfluidic system is depicted in Figure 5b. The dimensions of the chip are 25mm x 66mm, with an active capture area of 19mm x 51mm. It contains an equilateral triangular array of microposts, lOOμm tall and lOOμm in diameter with an average 50μm gap between microposts (Fig. 5c). For increased hydrodynamic efficiency, the repeated patterns of equilateral triangular arrays were shifted vertically by 50μm for every three rows throughout the chip to maximize the interactions between micropost structures and cells. This array incorporates 78,000 microposts within a surface area of 970 mm². Microposts were fabricated with deep reactive ion etching (DRIE) by Silex (Stockholm, Sweden).

Methods

Blood specimen collection and processing

Blood samples were drawn from patients with advanced stage solid tumors before, during, and after chemotherapy at Massachusetts General Hospital under an IRB-approved protocol. Blood specimens were also drawn from healthy donors after obtaining informed consent. All specimens were collected into vacutainer tubes containing the anticoagulant EDTA and were processed within 24hrs. Between sample collection and sample processing, whole blood specimens were stored at 4°C on a rocking platform to prevent cell settling. For experiments using lysed blood, NH₄Cl was added to whole blood in 10:1 v/v ratio and mixed for 15-20 minutes at room temperature. Following centrifugation at 1050 rpm (10°C) for 5 minutes, the supernatant was removed, and the pellet re-suspended in an equivalent volume of buffer and stored on a lab mixer at 4°C.

Macro-Micro coupling

To establish flow through the chip, the following methods were adopted to couple the microchip with a macrofiuidic pumping system. The silicon chips were purged with nitrogen and sealed with pressure-sensitive adhesive tape (3M, St. Paul, MN.)- The sealed microfluidic devices were then placed in a transparent 2 inch x3 inch plastic manifold consisting of a base, top cover and a spacer (Fig. 5b). The base has inlet and outlet ports for fluid handling. The manifold base also has six guiding metal pillars, each lmm in height, to hold the device in place and in alignment with the inlet and outlet ports. A metal spacer placed between the base and the top cover prevents mechanical stress on the device. The base and top cover attach by screws, providing a leak-proof assembly with minimum dead-volume. For ease of operation the port dimensions are such that standard Luer fittings can be used.

A pneumatic macrofluidic drive system was specifically designed to control flow through the microfluidic CTC-chip, as shown in Figure Ia. It uses a pneumatic pump, pressure regulators, and a digital pressure display to control the pressure of the air used to drive blood from a sealed sample container into the CTC-chip. A rocker assembly provides sample mixing throughout the experiment. Prior to running samples through the chip, the device was purged with 3.0 mL of buffer. A 5 mL aliquot of blood sample was measured into a conical tube, sealed, placed on the rocker unit, and connected to the chip with low dead volume fittings (Fig Ia). The sample was allowed to mix on the rocker for at least 5 minutes before running the experiment. The pneumatic pump was turned on, and the pressure adjusted according to the required flow rate. After the experiment, the CTC-chip was flushed with 10.0OmL PBS at lOmL/hr to remove any non-specifically bound cells.

Identification and enumeration of CTCs by fluorescence microscopy

An Olympus SZX (Olympus America Inc., NY) upright reflective microscope with an automated ProScan stage (Prior Scientific Inc., MA) was used to image the microfluidic chip. The microchip was scanned automatically in a lmm x lmm grid- format using the programmable stage and Qcapture Pro software (Media Cybernetics, MD). Captured images (at 10Ox total magnification) were carefully examined, and the objects that met predetermined criteria were counted. Color, brightness and morphometric characteristics such as cell size, shape, and nuclear size were considered in identifying potential CTCs and excluding cell debris and non-specific cells. Cells that stained positive for cytokeratin (CK+) and met the phenotypic morphological characteristics were scored as CTCs. Cell counts were expressed as the number of cells per 1 mL of blood (CTCs/mL). Cell counts by CK+ were also confirmed in a group of patients (n=8) using triple stain (CK+, DAPI+, CD45-). To evaluate inter-operator variability, the same set of scanned images from each chip (n=12) was analyzed by two different operators blinded to each other's results. The regression analysis of CTC counts by multiple operators demonstrated correlation coefficient (r²) of 0.992 with a slope of 1.04, indicating that the percentage of error in counting CTCs from the images is low and that enumeration of cell counts is highly reproducible between operators.

Cell Viability Cell viability was determined with the LIVE/DEAD viability assay kit. This assay is based on intracellular esterase activity of live cells and plasma membrane integrity of dead cells. Briefly, captured CTCs were incubated at room temperature for 30 minutes in a solution of 2 μM calcein AM and 4 μM ethidium bromide prepared in PBS. At the end of the incubation period, the chip was washed with ImL of 1 x PBS and visualized under microscope.

Molecular Analysis

Total RNA was isolated from cells using the Picopure RNA isolation kit according to manufacturer's instruction and subjected to linear amplification using the TransPlex Whole Transcriptome Amplification Kit (Rubicon Genomics). The 509 base pair human PSA coding region was amplified from circulating prostate tumor cell cDNA using the following primers pairs (sense and antisense, 5 '-3'): primary PCR: (TTGTGGGAGGCTGGGAGTG and CCTTCTGAGGGTGAACTTGCG; SEQ ID NO: 1), secondary PCR: (GGCAGGTGCTTGTGGCCTCTCGTGG and GTCATTGGAAATAACATGGAGGTCC; SEQ ID NO: 2). The TTF-I transcript was amplified using the following primer pairs (sense and antisense, 5 '-3'): primary PCR: (CTGCAACGGCAACCTGGGCAACATG ; SEQ ID NO: 3 and CAGGTACTTCTGTTGCTTGAAGCG; SEQ ID NO: 4), secondary PCR: (CAGGACACCATGAGGAACAGCGCCTC; SEQ ID NO: 5 and CAGGTACTTCTGTTGCTTGAAGCG; SEQ ID NO: 6).

Modeling and theoretical device optimization Micropost geometry and the arrangement of the micropost array were systematically explored in the process of designing the CTC-chip. Three different micropost distributions and arrangements were tested: a square array, a diagonal square array, and an equilateral triangular array. The area occupied by the microposts for the square and triangular distribution is given by: For a square array,

*-^- ω

For a triangular array,

2πa² ...

SP where a is the radius of the micropost and / is the center-to-center distance between adjacent microposts. Hence, for a given spacing between microposts, an increase in the radius of the micropost increases the micropost density, resulting in higher capture area. The hydrodynamic efficiency for each distribution is calculated based on the analytical solution derived by Drummond and Tahir³¹ and is determined as the ratio of the effective capture length to the mean spacing between the microposts. The effective capture length is based on the limiting trajectory for classical interception by a cylinder calculated from stream function. Accordingly, for an equilateral triangular distribution of microposts, the flow distribution is given by:

J

Where U is the mean velocity between microposts, F is the force per unit length acting on the cell, μ is the fluid viscosity, and ε is as given by Eq (2).

We evaluated the efficiency of cell capture using the three different micropost array distributions and found that the calculated hydrodynamic efficiency of capture as a function of the spacing between the microposts was greatest with the equilateral triangular micropost arrangement, and improved more so the shorter the center-to- center spacing between the microposts (Fig. 5a). However, excessive reduction in micropost spacing might lead to physical trapping of cells between the microposts; therefore, we selected a 50μm distance between microposts as the calculated optimal spacing, which yields a theoretical capture efficiency of 65%. Consequently, a detailed computational analysis of the hydrodynamics (flow distribution and cell trajectories) for an equilateral triangular array distribution of microposts was performed using commercially available finite element microfluidics solver COMSOL (Boston, MA). To ensure this CTC-chip would be sensitive and allow for high-throughput, we employed numerical analyses using COMSOL to characterize the interaction of cells with the microposts distributed within the flow path. Two factors essential to achieve high CTC capture are (1) optimization of flow velocity to maximize frequency of contact between cells and the chemically-modified microposts, and (2) optimization of shear forces to ensure that they are lower than those favoring cell attachment to the posts. Micropost size, spacing and the distribution along the streamlines are the critical variables that determine flow velocity and shear stress. Hence, we modeled the flow properties and cell-micropost interactions for an equilateral triangular array (Fig. 5b). We found that the triangular micropost arrangement resulted in non-linear streamlines, facilitating cell contact with micropost surfaces. For a volumetric flow rate of lmL/h through the device (Figure 2a), the expected maximum velocity was 460μm/s and occurred halfway between microposts (Fig. 5c). We also observed that the anticipated maximum shear stress experienced by a cell near the surface of a micropost is 0.4 dynes/cm² at θ=68 Fig. 5d), well below the levels physiologically deleterious and within the range at which maximum cell attachment would be expected to occur, as determined experimentally by linear shear stress chamber studies (Fig. 6). In summary, the analysis indicated an equilateral triangular micropost arrangement with a 50μm distance between microposts and with a 50μm shift after every 3 rows of microposts to be the most efficient micropost geometric arrangement and spacing. Applying a volumetric flow rate of lmL/h through the device, yields high throughput efficient capture of cells with low non-specific binding

Surface functionalization

The CTC-chip surface was functionalized with EpCAM antibodies using Avidin-Biotin chemistry.³² The surface of the chip was modified with 4% (v/v) 3- mercaptopropyl trimethoxysilane in ethanol at room temperature for 45 min, then treated with the coupling agent N-γ-maleimidobutyryloxy succinimide ester (GMBS, lμM) resulting in GMBS attachment to the microposts. Next, the chip was treated with lOμg/mL of Neutravidin at room temperature for 30 min leading to immobilization onto GMBS, and then flushed with PBS to remove excess Avidin. Finally, biotinylated EpCAM antibody at a concentration of lOμg/mL in phosphate buffered solution (PBS) with 1% (w/v) BSA and 0.09% (w/v) sodium azide was allowed to react for 15-30 minutes before washing with PBS. The chip was air dried and stored at ambient temperature for up to three weeks until use.

Cell-line experiments

The human non-small-cell lung cancer (NSCLC) cell line NCI-H1650 was maintained and grown to confluence in RPMI- 1640 medium containing 1.5mM L- glutamine supplemented with 10% fetal bovine serum at 37°C in 5% CO₂ with humidity. Growth medium was aspirated, and cells incubated with trypsin for 10 minutes. A protein buffer was added to quench protease activity. Cells were then pre- labeled with cell tracker orange using the standard protocol provided by the manufacturer. The cell titer was determined by counting with a hemocytometer. The desired concentration of cells was then prepared by serial dilution of the original cell suspension in PBS. Labeled cells were spiked into whole blood.

Fixation and staining of captured cells Captured cells were fixed by flowing 0.9mL of 1 % PFA in PBS, through the device at 3.0mL/hr for 20 minutes. The chip was subsequently washed with a solution of 0.9mL of 0.2% Triton X-100 in PBS for 10 minutes to induce cellular permeability and allow for intracellular staining. To identify any bound lymphocytes, 0.9mL of anti-CD45 stock solution (50μL of antibody stock solution in ImL of PBS) was passed through the chip at 3mL/hr for 30 minutes, followed by a PBS wash to remove excess antibody. To identify epithelial cells, 0.9 mL of anti-cytokeratin stock solution (50μL of antibody stock solution in ImL of PBS) was passed through the chip at 3mL/hr for 30 minutes, followed by a PBS wash. Finally, to permit the identification of cellular nuclei, 0.9mL of DAPI solution (lOμl of DAPI reagent in ImL of DI water) was passed through the chip at 3mL/hr, for 15 minutes followed by a PBS wash. The chip was removed from the manifold, wiped dry near the fluid ports and stored in the dark at 4°C until imaging. Shear stress studies using linear shear Hele-Shaw chambers

Shear stress plays an important role in cell capture. An optimum shear stress should be applied such that one can capture maximum number of cancer cells at high enough flow rates. To find optimal flow rate, we studied the effect of shear stress on cell capture using Hele-Shaw microfluidic chambers.² The geometry of these chambers (Fig. 6d) is such that the shear stress varies linearly along the chamber length (Fig. 6e), permitting the study of a wide range of shear stresses for a given flow rate. Cultured lung cancer cells were spiked into PBS solution, and then passed through the Hele-Shaw chambers functionalized with EpCAM Ab at a constant flow rate. As the shear stress decreased along the channel, the density of the cells adhering to the micropost surface increased (Fig. 6, a-c). The effect of shear stress on cell adhesion through EpCAM antibody-antigen binding (Fig. 6f) indicated that 8 dyn/cm² was the optimum shear rate, resulting in the capture of 200 cells/mm² of functionalized capture surface.

Testing CTC capture from patient blood

Prior to the more extensive clinical testing described above, we performed initial clinical experiments using blood samples from patients with advanced NSCLC. First, we evaluated CTC capture efficiency among whole blood samples and red blood cell-lysed samples in 6 NSCLC patients. Isolated cells are shown in Figure 7a. The white arrows point to CTCs and the green arrows to leukocytes. The inset at the top left of Figure 7a shows a high-magnification view of a cytokeratin positive (CK+) CTC and the inset at the lower left presents a CD45+ leukocyte. The total number of cells captured for each case was analyzed to calculate captured CTCs per mL of blood (Fig. 7b). Consistent with the spiked cell experiments, we observed no significant difference in the CTC counts of cancer patients from whole blood or lysed blood confirming that the CTC -chip can be used to directly isolate CTCs from whole blood without any need for pre-processing. The reproducibility of CTC measurements between split samples was tested with a Wilcoxon matched-pairs signed-ranks two- sided test.

Materials 3-Mercaptopropyl trimethoxysilane was obtained from Gelest (Morrisville, PA). Ethanol (200 proof), tissue culture flasks, a hemocytometer, serological pipettes, were purchased from Fisher Scientific (Fair Lawn, NJ). Fetal bovine serum (FBS) and 0.5M ethylene diamine tetra acetic acid (EDTA) were purchased from Gibco (Grand Island, NY). Dimethyl sulfoxide (DMSO), sodium azide, lyophilized bovine serum albumin (BSA), and a glovebag for handling the moisture-sensitive silane were obtained from Sigma Chemical Co. (St. Louis, MO). The coupling agent GMBS (N-γ- maleimidobutyryloxy succinimide ester), NHS-LC-LC-biotin (succinimidyl-6¹- [biotinamido]-6-hexanamido hexanoate), and fluorescein-conjugated NeutrAvidin were obtained from Pierce Biotechnology (Rockford, IL). Biotinylated mouse anti- human anti-EpCAM was obtained from R&D Systems (Minneapolis, MN). Human non-small-cell lung cancer line NCI-H1650, prostate cell line PC3-9, breast cancer cell line SKBr-3 and bladder cancer cell line T-24 were purchased from American Type Culture Collection (Manassas, VA), and RPMI- 1640 growth medium was purchased from Invitrogen Corporation. Orange [5- (and 6-)-(((4-chloromethyl)- benzoyl) amino) tetrarnethyl-rhodamine, CMTMR] and green [5- chloromethylfluorescein diacetate, CMFDA] cell tracker dyes were obtained from Molecular Probes (Eugene, OR). Anti-Cytokeratin PE (CAM 5.2, conjugated with phycoerythrin), CD45 FITC, the fluorescent nucleic acid dye nuclear dye 4',6- diamidino-2-phenylindole (DAPI) and 1OmL vacutainer tubes was purchased from BD Biosciences (San Jose, CA).

Molecular Analysis of Prostate Cancer CTCs

We have also performed molecular analysis on CTCs captured from prostate cancer patients. In these experiments, RNA was produced from genomic DNA followed by conversion to cDNA. The cDNA was amplified and assayed for TMPRSS2-ERG translocations. Figure 11 shows the results of RT-PCR of prostate CTCs run on a gel and the sizes of different isoforms. Most bands were one size (Tl :E4), but in the GU34 fraction one band was shorter, corresponding to Tl :E5. Sequencing chromatograms below the gel illustrate the different breakpoints (Tl :E4 and Tl :E5) at the DNA level.

Example 2. EGFR mutations in NSCLC The development of effective molecular therapeutic targeting strategies in epithelial cancers requires the ability to sample tumor tissue for markers predictive of drug response and ideally to do so repeatedly and noninvasively during the course of therapy. Hence blood-based assays to detect epidermal growth factor receptor (EGFR) mutations in patients with advanced non-small cell lung cancer (NSCLC) are very attractive. Using a CTC-chip, we isolated CTCs with greater than 50% purity. Hence the CTC-chip presents an opportunity to detect EGFR mutations from genomic DNA extracted from CTCs. hi addition, by focusing on the portion of the chip where CTCs were preferentially enriched, discussed below, the purity was further increased. We isolated genomic DNA and RNA from captured cells in four fractions that cover the entire surface area of the device (Figure 12). The individual fractions are analyzed based on the CTC capture profiles observed for each tumor type; depending on the tumor type, CTCs may be enriched at different parts of the chip. For example, lung CTCs were enriched near the inlet, whereas prostate CTCs were uniformly captured throughout the chip. This method of fractionating led to successful detection of EGFR mutations in lung CTCs.

After the experimental run (capture followed by wash), the chip was taken out of the manifold, and the tape sealing the chip was gently removed. Then a protease (lOOμL) was directly pipetted onto the chip on each region of the chip, sequentially, as indicated in Figure 11. From each region and with little mixing, about 1 OOμL of cell lysate was collected into individual eppendorf tubes.

The genomic DNA extracted from CTCs was amplified twice using the TransPlex nucleic acid amplification kit (Rubicon Genomics) according the manufacturer's protocol. The EGFR mutations were detected in the amplified material using the EGFR Mutation Test Kit from DxS Ltd. This assay can detect EGFR mutations in a background of large quantities of wild type genomic DNA.

We demonstrated that the EGFR mutations detected in CTCs correlate well with that in the primary tumor. In several NSCLC patients, we have detected mutations that were expected based on the primary tumor analysis, such as deletions in exon 19 and L858R substitutions in exon 21. More significantly, we have also identified the mutation that confers resistance to EGFR inhibitors, T790M, which resides in exon 20. Table 3 summarizes the results found in advanced lung cancer patients. In 21/22 patients, the analysis of CTCs showed the original mutations that were expected, based on the primary tumor analysis. In addition, we have also detected common resistant mutations in 14/22 patients. We observed that CTCs are more reliable compared to plasma. Furthermore, 6 of the patients with no detectable T790M in CTCs showed response to treatment. In addition for the first time, we have shown the evolution of T790M mutation during therapy which is associated with resistance to therapy.

Table 3

In summary, the present invention in combination with the fractionation method to isolate genomic DNA and RNA offers a blood-based molecular diagnostics that provides a new and exciting approach to monitor genetic lesions in circulating tumor cells, possibly circumventing the requirement for serial biopsies of inaccessible solid tumors. Hence this approach has tremendous potential to provide an ideal tool that will enable cancer biologists unprecedented molecular access to rare CTC subpopulations.

Identification of CTCs in patients with metastatic NSCLC

Blood samples were obtained from 23 patients with £GFi?-mutant advanced NSCLC, including 5 treatment-naϊve patients, 10 previously treated with either erlotinib or gefitinib, and 8 previously treated with chemotherapy or multiple regimens including TKIs and chemotherapy. The strategy used for microfluidic CTC isolation from whole blood is schematically depicted, along with representative images of captured cells in Figs. 13A-C. CTCs were identified in all patients, with a mean of 131 cells/ml (range 5 to 771 cells/ml), which was not different from the quantity identified in patients with EGFR wild-type NSCLC (Table 4). Matched radiographic tumor burden measurements close to the time of CTC analysis (median 8 days, range 0-38 days) demonstrated that CTC quantity measured at a single time point was not well correlated with radiographic tumor burden (P= 0.9), suggesting that additional tumor characteristics, including invasiveness and vascularity, are likely to contribute to the shedding of CTCs.

Abbreviations: mo = months, cm = centimeters, M = male, F = female, Adeno = adenocarcinoma, Adeno/BAC = adenocarcinoma with bronchioloalveloar features, NSCLC = non-small-cell lung cancer not otherwise specified, C = chemotherapy, E = erlotinib, G = gefitinib, O= other/experimental agent.

¹ Prior systemic therapies are listed in the order received. Tumor burden was measured by unidimensional diameter as per RECIST. CTC numbers/ml were calculated based on analysis of l-5ml of whole blood per case. Tumor burden and CTC number are not correlated, Spearman correlation coefficient of -.02 (p=0.9).

Allele-specific detection of EGFR mutations in paraffin-embedded tumor specimens

To establish a reliable assay capable of detecting EGFR mutations in rare cell populations, we adapted the allele-specific SARMS assay, which is designed to detect 7 types of EGFR mutations, including the multiple in-frame exon 19 deletions (collectively analyzed as "Del" mutations) and the L858R missense mutation, which together account for 90% of sensitizing EGFR mutations. The assay also detects the recurrent T790M mutation associated with acquired TKI resistance. Since the SARMS assay has not previously been compared to standard sequencing analysis of paraffin-embedded tumor samples, we first analyzed 26 NSCLC tumors previously identified as EGFi?-mutant and 8 tumors reported as wild-type by sequencing analysis (Fig 14A). SARMS analysis and nucleotide sequencing identified the same mutation in 25 cases, while all 8 wild-type cases were confirmed negative yielding a sensitivity of 96% and a specificity of 100%. The single discrepancy was due to a unique deletion mutation that is not within the detection capacity of the SARMS assay.

Use of the highly sensitive SARMS assay to analyze tumor specimens also led to identification of rare EGFR mutant alleles below the detection limit of standard sequencing. In addition to the known primary EGFR mutation, low levels of the TKI resistance-associated T790M EGFR allele detected in 10/26 (38%) pre-treatment tumor samples. The relatively high number of amplification cycles required to demonstrate T790M suggests that it is present in a small number of cells, and indeed, sequencing of cloned PCR products from one of these tumors identified only one T790M mutation in 500 EGFR alleles. Presence of the drug resistance mutation at such low frequency did not preclude significant responses to TKI therapy, but it was associated with a striking difference in progression-free survival (PFS) with a median PFS of 7.7 months in cases with a detectable T790M allele, compared to 16.5 months in cases lacking T790M (PO.001) (Fig 14B). The preexisting drug resistance allele may be rapidly selected following TKI therapy, possibly accounting for some of the known variation in response duration among EGFi?-mutant NSCLC.

ΕGFR mutation detection in CTCs

Having established the reliability of the SARMS assay in paraffin-embedded tumor specimens, we adapted it to analysis of CTCs isolated from peripheral blood. We first compared the EGFR mutations detected in CTCs by SARMS assay with those reported for the tumor specimen using either standard sequencing or SARMS analysis. Of 20 cases available for molecular analysis of CTCs, SARMS analysis identified activating mutations in 19 (95%) cases. Tumor EGFR mutation results were available for 14 of these, which were concordant with CTC analyses in 13 (93%)

(Table 5). In addition to the primary EGFR activating mutation, the T790M allele was detectable in CTCs from 2/6 patients responding to TKI therapy and 9/14 patients with clinical progression.

Table 5: EGFR mutation analysis in captured CTCs and free plasma

DNA

Abbreviations: SARMS = Scorpion Amplification Refractory Mutation System, Del = deletion, na = sample not available for analysis, NA= not applicable due to unavailable tumor for comparison, Und = undetected The detection of T790M by SARMS assay is indicated by "+" or "-". The detection of other mutations by SARMS assay are listed only when present. Concordance between the presence of the primary mutation in the tumor specimen and in the analysis of CTCs (C) and free plasma DNA analysis (P) is indicated in the final column. Clinical response to EGFR TKIs was classified as responding or progressing at the time of CTC analysis. In all cases, mutations listed as "other" were present at lower abundance than the primary mutation.

"Mutation in the primary tumor was detected by HPLC analysis but below detection by standard sequencing analysis. ² Mutation of unknown significance in the primary tumor (S885L) is not included in the SARMS assay. Additional tumor specimen was not available for SARMS analysis.

Recent studies have reported detection of EGFR mutations using SARMS in free plasma DNA from patients with metastatic NSCLC. We therefore compared the accuracy of mutational analysis in purified CTCs versus free DNA isolated from plasma using paired blood samples from 18 patients with EGFR-nmtant NSCLC and 3 with wild-type NSCLC (Table 5). SARMS analysis identified EGFR mutations in 17 (94%) CTC specimens, but in only 7 (39%) plasma samples. When detectable, the primary EGFR mutations identified in plasma were identical to those in CTCs. In addition to the primary mutation, the sensitive SARMS assay identified rare secondary EGFR activating mutations in a subset of primary tumors, CTCs, and plasma samples (Table 5 and Fig. 14A). To determine their significance, along with that of detectable T790M during response and progression of NSCLC, we monitored CTC -genotypes and CTC numbers over time in a subset of patients. The SARMS assay groups all variant breakpoints of the in-frame EGFR deletion mutations as a single "Del" mutation, and all mutations at codon 719 as "G719X". The detection of T790M by SARMS assay is indicated by "+" or "-". The detection of other mutations by SARMS assay is listed only when present. The mutation identified by sequencing in patient 2 is not within the detection capacity of the SARMS assay, and therefore SARMS analysis would be expected to be negative. In patients 30 and 31 , an infrequent mutant EGFR allele (other) was detected by SARMS in addition to the prevalent mutation detected by both SARMS and standard nucleotide sequencing analysis. EGFR TKI therapy was either gefitinib or erlotinib and was administered as l^st-line therapy for advanced disease except where indicated by " " (EGFR TKI therapy was given 2^nd- or 3^rd-line), or by "none" (no EGFR TKI was given). Duration of therapy was measured in months, and when preceded by ">" indicates ongoing therapy. Best clinical response is defined per RECIST.

Serial measurements of CTCs in patients following initiation of therapy Detailed serial analyses of CTC quantity and genotype were available in four cases of £Gi*7?-mutant NSCLC, following initiation of gefitinib therapy (Figure 15A). In all cases gefitinib was associated with a profound decline in CTC numbers. The interval between blood sample analyses was dictated by previously scheduled clinical encounters, therefore we cannot ascertain the precise timing of decline in CTC counts; however, in one case (patient 9), a 50% decline in CTC counts was evident within a week of initiating therapy, with the nadir at three months. The development of resistance to the inhibitors was associated with an increase in CTC counts, and in one case a subsequent decline as the tumor responded to chemotherapy. In all cases, blinded radiographic assessment of tumor volume was concordant with CTC numbers during the course of therapy.

CTC genotypes evolved during the course of therapy, with consistent presence of the primary EGFR activating mutation, but emergence of the T790M drug resistance mutation. While this mutation was present at very low allele frequency in the initial tumor specimen, as determined by the relative number of cycles required for amplification (ΔCt), serial analysis indicated increased prevalence over time consistent with acquisition of clinical resistance (Fig 15B). Remarkably, some cases also showed the emergence of additional EGFR activating mutations. While in most cases, these additional mutations were less prevalent than the primary mutation, at least one case clearly demonstrates the potential for evolution in the dominant tumor genotype (Fig. 15C). In this case, sufficient DNA was isolated from captured CTCs to allow direct nucleotide sequencing of EGFR, confirming that the dominant mutation in CTCs differs from that present in the original tumor specimen.

By studying patients with advanced EGFZ?-mutant NSCLC, we have shown that the CTC-chip technology reproducibly identifies CTCs in sufficient quantity and with sufficient purity to allow molecular analyses that are relevant to clinical management. CTCs were readily identified in all cases, in numbers that are approximately 100-fold higher than with currently other technology. In cases studied longitudinally, CTC numbers over time showed a significant decline associated with tumor response to EGFR TKIs, with rising numbers as drug resistance emerged. CTC genotypes were highly concordant with the mutational status of tumor biopsy specimens, and provided additional information as they were serially repeated during the course of the disease. Taken together, these molecular studies provide novel insight into the progression of EGi*7?-mutant NSCLC, and illustrate the potential impact of CTC-based serial noninvasive monitoring in epithelial cancers.

The use of the allele-specific SARMS assay to identify activating EGFR mutations in rare cell populations is made possible by the recurrent nature of these mutations, with only two cases in our study having mutations not represented within the panel detectable by the assay. Together with the CTC-chip, this molecular assay allows noninvasive genotyping in NSCLC. Given the use of fine needle aspirates to diagnose lung cancer, which provide minimal amounts of tumor material for molecular analysis, CTC-genotyping provides a reliable approach that may be repeated as therapeutic decision-making points emerge during a patient's course. CTC-genotyping appears to be more sensitive than analysis of free plasma DNA, and the concomitant quantification of CTCs provides an important context in which to interpret genotyping results.

In addition to the primary activating EGFR mutation, we also identified the secondary T790M mutation associated with acquired TKI-resistance. Consistent with results from serial biopsy and autopsy studies, the T790M mutation was commonly observed in CTCs from patients progressing on ΕGFR TKI therapy. Unexpectedly, use of the highly sensitive allele-specific assay showed that a subset of NSCLCs harbor rare T790M alleles prior to TKI exposure. The T790M allele is thought to emerge through selective pressure during therapy, although it has been reported in rare cases without prior drug exposure and has been shown to encode additional transforming properties when combined in cis with the more common EGFR activating mutations. Thus, the T790M allele may initially arise by virtue of its oncogenicity, and rapidly emerge as a dominant allele following drug treatment. Presence of rare T790M alleles in pre-treatment tumor specimens did not preclude dramatic clinical responses to TKIs, but did have a very significant impact on the progression-free survival. This molecular marker may therefore be a major determinant distinguishing patients likely to have a prolonged response to single agent erlotinib or gefitinib, from those whose response is likely to be short-lived and who may be appropriate candidates for second-generation irreversible TKIs or combination targeted therapy regimens. Amplification of the gene encoding the growth factor receptor MET has recently been reported as a second mechanism of acquired resistance to EGFR TKIs¹⁹' ²⁰. However, we did not identify significant (>2- fold) amplification of the MET gene in either pretreatment NSCLC or in CTCs, using a quantitative PCR assay.

Methods

Patients and clinical specimens Patients with advanced NSCLC (incurable stage III or stage FV) were recruited under one of two IRB-approved clinical trial protocols. Group A patients were treated at the Massachusetts General Hospital Cancer Center and donated 10ml of peripheral blood on one or more occasions during the course of their disease for CTC-chip analysis. Blood samples were analyzed for CTC quantity and/or EGFR mutations using the Scorpion Amplified Refractory Mutation System (SARMS) technology (DxS Delivery Pharmacogenomics). In a subset of cases, concurrently isolated free plasma DNA was also analyzed for EGFR mutations. Available archived paraffin- embedded tumor biopsy specimens, previously subjected to standard EGFR sequencing analysis, were reanalyzed by SARMS assay. The number of tumor biopsy specimens available for comparison of EGFR sequencing and SARMS analysis was extended by inclusion of Group B patients, who had participated in a multi-center phase II clinical trial utilizing first-line gefitinib therapy. Medical charts of all patients were reviewed for demographics and clinical history, with tumor burden at various time points quantified as the sum of the unidimensional size of all measurable tumor sites (as per RECIST) via central review. Patients who had received therapy with an EGFR TKI (gefitinib or erlotinib) were assessed for length of therapy and best clinical response using RECIST. The correlation between CTC quantity and computed tomography (CT scan) measurements was analyzed using the Spearman correlation coefficient. The relationship between progression-free survival on EGFR TKI therapy (time from start of therapy until tumor progression by RECIST or death, whichever was sooner) was analyzed using the Kaplan-Meier method and the Log-Rank test. CTC capture and enumeration

Whole blood was passed through the CTC-chip at a flow rate of l-2ml/hr allowing capture of CTCs, followed by saline wash (10ml/hr) to remove non- specifically bound leukocytes. For enumeration, cells were fixed on the CTC-chip (1 % paraformaldehyde, 0.2% Triton X- 100, 1 % BSA, all in PBS) and stained with Hoechst to identify DNA content, rhodamine-conjugated antibody to cytokeratin (CK) to identify epithelial cells and fluorescein-coηjugated antibody to CD45 to identify leukocytes. The number of CTCs/ml was determined via comprehensive image analysis, scanning the entire chip (Olympus SZX microscope, Olympus America Inc., NY) and identifying CTCs based on cell size, morphology, and fluorescence staining (Hoechst, CK positive). For demonstration of EGFR expression, captured cells were stained with a mouse monoclonal antibody (Vector Laboratories). Molecular analysis of CTCs DNA was eluted from captured CTCs using the Pico Pure DNA Extraction Kit (Molecular Devices) and subjected to two rounds of linear amplification using the Transplex amplification kit (Rubicon Genomics). Free plasma DNA was isolated using Vacutainer PPT Plasma Preparation Tubes and the QIAmp DNA Blood Midi Kit (Fisher Scientific/ DNA was prepared from paraffin-embedded tumor blocs using standard Proteinase K isolation. For identification of EGFR mutations using the SARMS assay (DxS Delivery Pharmacogenomics), 1.5 ng of DNA was analyzed using the ABI 7500 Detection System. The assay detects grouped deletions within exon 19, insertions within exon 20 and mutations affecting codon 719 (G719X), as well as the individual mutations T790M, L858R, L861Q, and S768I. The rate of amplification of these mutant alleles is compared with a control amplification of EGFR exon 2. Standard bidirectional nucleotide sequencing was performed using a Capillary ABI 3100 sequencer (Applied Biosystems, Foster City, CA).

REFERENCES

1. Cristofanilli, M. et al. Circulating tumor cells, disease progression, and survival in metastatic breast cancer. N Engl J Med 351, 781 -91 (2004).

2. Steeg, P. S. Tumor metastasis: mechanistic insights and clinical challenges. Nat Med 12, 895-904 (2006). 3. Gupta, G. P. & Massague, J. Cancer metastasis: building a framework. Cell 127, 679-95 (2006).

4. Reya, T., Morrison, S. J., Clarke, M. F. & Weissman, I. L. Stem cells, cancer, and cancer stem cells. Nature 414, 105-11 (2001). 5. Mocellin, S., Hoon, D., Ambrosi, A., Nitti, D. & Rossi, C. R. The prognostic value of circulating tumor cells in patients with melanoma: a systematic review and meta-analysis. Clin Cancer Res 12, 4605-13 (2006).

6. Smerage, J. B. & Hayes, D. F. The measurement and therapeutic implications of circulating tumour cells in breast cancer. Br J Cancer 94, 8-12 (2006). 7. Rolle, A. et al. Increase in number of circulating disseminated epithelial cells after surgery for non-small cell lung cancer monitored by MAINTRAC(R) is a predictor for relapse: A preliminary report. World J Surg Oncol 3, 18 (2005).

8. Braun, S. & Marth, C. Circulating tumor cells in metastatic breast cancer — toward individualized treatment? N Engl J Med 351, 824-6 (2004). 9. Zieglschmid, V., Hollmann, C. & Bόcher, O. Detection of disseminated tumor cells in peripheral blood. Crit Rev Clin Lab Sci 42, 155-96 (2005).

10. Bell, D. W. & Haber, D. A. A blood-based test for epidermal growth factor receptor mutations in lung cancer. CHn Cancer Res 12, 3875-7 (2006).

11. Kahn, H. J. et al. Enumeration of circulating tumor cells in the blood of breast cancer patients after filtration enrichment: correlation with disease stage.

Breast Cancer Res Treat 86, 237-47 (2004).

12. Krivacic, R. T. et al. A rare-cell detector for cancer. Proc Natl Acad Sci U S.4 101, 10501-4 (2004).

13. Racila, E. et al. Detection and characterization of carcinoma cells in the blood. Proc Natl Acad Sci USA 95, 4589-94 (1998).

14. Fu, A. Y., Spence, C, Scherer, A., Arnold, F. H. & Quake, S. R. A microfabricated fluorescence-activated cell sorter. Nat Biotechnol 17, 1109-11 (1999).

15. Davis, J. A. et al. Deterministic hydrodynamics: taking blood apart. Proc Natl Acad Sci USA 103, 14779-84 (2006). 16. Huang, L. R., Cox, E. C, Austin, R. H. & Sturm, J. C. Continuous particle separation through deterministic lateral displacement. Science 304, 987-90 (2004). 17. Chang, W. C, Lee, L. P. & Liepmann, D. Biomimetic technique for adhesion based collection and separation of cells in a microfluidic channel. Lab Chip 5, 64-73 (2005).

18. Whitesides, G. M. The origins and the future of microfluidics. Nature 442, 368-73 (2006).

19. Hong, J. W. & Quake, S. R. Integrated nanoliter systems. Nat Biotechnol 21, 1179-83 (2003).

20. Toner, M. & Irimia, D. Blood-on-a-chip. Annu Rev BiomedEng 7, 77-103 (2005). 21. Dittrich, P. S. & Manz, A. Lab-on-a-chip: microfluidics in drug discovery.

Nat Rev Drug Discov 5, 210-8 (2006).

22. El-AIi, J., Sorger, P. K. & Jensen, K. F. Cells on chips. Nature 442, 403-11 (2006).

23. Went, P. T. et al. Frequent EpCam protein expression in human carcinomas. Hum Pathol 35, 122-8 (2004).

24. Balzar, M., Winter, M. J., de Boer, C. J. & Litvinov, S. V. The biology of the 17-1 A antigen (Ep-CAM). JMoI Med 11, 699-712 (1999).

25. Rao, C. G. et al. Expression of epithelial cell adhesion molecule in carcinoma cells present in blood and primary and metastatic tumors. Int J Oncol 27, 49-57 (2005).

26. Smirnov, D. A. et al. Global gene expression profiling of circulating tumor cells. Cancer Res 65, 4993-7 (2005).

27. Therasse, P. et al. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92, 205-16 (2000).

28. Terstappen, L. W. et al. Flow cytometry—principles and feasibility in transfusion medicine. Enumeration of epithelial derived tumor cells in peripheral blood. Vox Sang 74 Suppl 2, 269-74 (1998). 29. Kraeft, S. K. et al. Reliable and sensitive identification of occult tumor cells using the improved rare event imaging system. Clin Cancer Res 10, 3020-8 (2004). 30. Allard, W. J. 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 10, 6897-904 (2004).

31. Drummond, J. E. & Tahir, M. I. Laminar viscous flow through regular arrays of parallel solid cylinders. Int. J. Multiphase Flow 10, 515-539 (1984).

32. Murthy, S. K., Sin, A., Tompkins, R. G. & Toner, M. Effect of flow and surface conditions on human lymphocyte isolation using microfluidic chambers. Langmuir 20, 11649-55 (2004).

OTHER EMBODIMENTS

All publications, patents, and patent applications mentioned in the above specification are hereby incorporated by reference. Various modifications and variations of the described method and system of the invention will be apparent to those skilled in the art without departing from the scope and spirit of the invention. Although the invention has been described in connection with specific embodiments, it should be understood that the invention as claimed should not be unduly limited to such specific embodiments. Indeed, various modifications of the described modes for carrying out the invention that are obvious to those skilled in the art are intended to be within the scope of the invention. What is claimed is:

Claims

1. A device for capturing circulating, nonhemopoietic tumor cells, said device comprising: a. a microfluidic channel to which is bound a tumor specific binding agent; and b. a pump producing a continuous, unidirectional shear stress of 0.1 to 20 dyn/cm² in said channel.

2. The device of claim 1, wherein said channel comprises a plurality of obstacles to which are bound said tumor specific binding agent.

3. The device of claim 1, wherein the walls of said channel are substantially planar.

4. The device of claim 1, wherein said tumor specific binding agent is disposed in a region of said channel having a volume of 10μL-20mL.

5. The device of claim 1, wherein said channel and pump are configured to produce a flow rate of 0.1 mL to 30 mL/hr.

6. A device for capturing circulating, nonhemopoietic tumor cells, said device comprising a microfluidic channel comprising a plurality of obstacles to each of which is bound a tumor specific binding agent, wherein said obstacles are disposed in an equilateral triangular arrangement with a 10-100 μm distance between obstacles and a 10-100 μm shift between at least two rows.

7. The device of claim 6, wherein said shift occurs after at least three rows.

8. The device of claim 6, wherein said distance between obstacles and said shift are about 50 μm.

9. A device for capturing circulating, nonhemopoietic tumor cells, said device comprising a microfluidic channel to which is bound a tumor specific binding agent and a reservoir having a volume of less than 5 mL, wherein said reservoir and said channel are fluidically connected.

10. The device of claim 9, wherein said reservoir has a volume of less than 1 mL.

11. A kit comprising the device of any of claims 1-10 and a reagent for obtaining genetic information from a circulating nonhemopoietic tumor cell.

12. The kit of claim 11, wherein said reagent lyses said tumor cell.

13. The kit of claim 11, wherein said reagent amplifies a nucleic acid from said cell.

14. The kit of claim 11, wherein said reagent binds to a specific nucleic acid sequence from said cell.

15. The kit of claim 11, wherein said reagent determines protein expression.

16. The kit of claim 11, wherein said reagent is a labeled antibody.

17. The kit of claim 11, wherein said circulating tumor cell is from a lung cancer and said genetic information comprises the presence or absence of an EGFR mutation.

18. The kit of claim 17, wherein said EGFR mutation is a deletion in exon 19, T790M, L858R, L861Q, G719(S, A, or C), S768I, an insertion in exon 20, or a combination thereof.

19. The kit of claim 11, wherein said circulating tumor cell is from a prostate cancer and said genetic information comprises the presence or absence of a TMPRSS2-ERG fusion.

20. The kit of claim 19, wherein said TMPRSS2-ERG fusion is T3/E4, T2/E2, Tl/E3_5, T2/E5, Tl/E234_6, T1/E2, T1/E3, T2/E4, T1/E5, T1/E6, Tl/E_IIIa_4, Tl_I/E_IIIb_4, Tl/E_IIIc_4, T5/E4, T4/E4, T4/E5, or a combination thereof.

21. The kit of claim 11 , wherein said CTC is from a breast cancer and said nucleic acids assayed encode HER2.

22. A method for capturing circulating, nonhemopoietic tumor cells, said method comprising introducing a blood sample into a device of any of claims 1-10 so that circulating tumor cells in said blood sample bind to said binding agent in said device.

23. A method of obtaining genetic information from a subject with a tumor, said method comprising introducing a blood sample from said subject into a microfluidic channel to which is bound a tumor specific binding agent so that circulating tumor cells in said blood sample bind to said binding agent; lysing cells bound to said channel; and obtaining genetic information from said lysate, wherein said lysate is not purified prior to said obtaining.

24. The method of claim 23, further comprising treatment selection, wherein said treatment is selected based on said genetic information

25. The method of claim 24, wherein said treatment is a tyrosine kinase inhibitor.

26. A method for capturing circulating, nonhemopoietic tumor cells, said method comprising passing a blood sample of less than 4 mL through a microfluidic channel to which is bound a tumor specific binding agent so that circulating tumor cells in said blood sample bind to said binding agent.

27. A method of obtaining genetic information from a subject with a tumor, said method comprising obtaining 1 to 1500 circulating tumor cells from a blood sample of 1 mL or less from said subject and assaying said cells for genetic information.

28. A method of diagnosing a carcinoma in a subject, said method comprising introducing a blood sample from said subject into a microfluidic channel to which is bound a tumor specific binding agent so that circulating tumor cells in said blood sample bind to said binding agent, wherein the presence of one circulating tumor cell is diagnostic for the presence of said carcinoma, and the absence of any circulating tumor cells is diagnostic for the absence of said carcinoma.

29. The method of any of claims 22-28, wherein said blood sample has a volume of 50μL or more.

30. The method of any of claims 22-26, wherein said blood sample has a volume of less than 4 mL.

31. The method of any of claims 22-28, wherein said blood sample is anticoagulated whole blood.

32. The method of any of claims 22-28, further comprising the step of enumerating the tumor cells bound to said binding agent.

33. The method of claim 32, further comprising the step of repeating said introducing and enumerating steps after a predetermined interval of time, wherein the change in the number of tumor cells over time is indicative of the prognosis of said nonhemopoietic tumor.

34. The method of claim 33, wherein a negative slope of the number of tumor cells as a function of time is indicative of a positive prognosis.

35. The method of claim 33, wherein a positive slope or no slope of the number of tumor cells as a function of time is indicative of a negative prognosis.

36. The method of any of claims 22 and 26-28, further comprising the step of obtaining genetic information from at least one tumor cell bound in said introducing step.

37. The method of any of claims 22-28, wherein said blood sample is from a patient at risk for a clinically localized tumor, wherein the presence of tumor cells bound in said introducing step is diagnostic for said clinically localized tumor.

38. The method of any of claims 22-28, wherein said introducing or passing occurs at a rate of 0.1 mL to 30 mL/hr.

39. The method of any of claims 22 and 26-28, further comprising lysing circulating tumor cells bound to said binding agent.

40. The method of any of claims 23 or 39, wherein said tumor specific binding agent is disposed in a region of said microfluidic channel, and said lysing comprises applying a lysing agent to a portion of said region where a majority of said circulating tumor cells are bound.

41. The method of any of claims 23 or 39, wherein said tumor specific binding agent is disposed in a region of said microfluidic channel, and said lysing comprises applying a lysing agent to the entire region.

42. The method of any of claims 23 or 36, further comprising assaying nucleic acids of said circulating tumor cells.

43. The method of claim 42, wherein said circulating tumor cells are from a lung cancer, and said nucleic acids assayed encode EGFR.

44. The method of claim 43, wherein said nucleic acid is assayed for a deletion in exon 19, T790M, L858R, L861Q, G719(S, A, or C), S768I, an insertion in exon 20, or a combination thereof.

45. The method of any of claims 23 or 36, wherein said circulating tumor cells are from a prostate cancer, and said nucleic acids assayed encode a TMPRS S2-ER.G fusion.

46. The method of claim 45, wherein said nucleic acid is assayed for T3/E4, T2/E2, Tl/E3_5, T2/E5, Tl/E234_6, T1/E2, T1/E3, T2/E4, T1/E5, T1/E6, Tl/E_IIIa_4, Tl_I/E_IIIb_4, Tl/E_IIIc_4, T5/E4, T4/E4, T4/E5, or a combination thereof.

47. The method of any of claims 23 or 36, wherein said CTC is from a breast cancer and said nucleic acids assayed encode HER2.

48. The method of any of claims 22-28, further comprising washing said bound tumor cells at a higher shear stress or volume than that used in said introducing step to increase purity.

49. The method of any of claims 22-28, further comprising assaying for protein expression in said circulating tumor cells.

50. The method of claim 49, wherein said circulating tumor cells are from a prostate cancer and said protein assayed is PSA.