Fully automated, on-site isolation of cfDNA from whole blood for cancer therapy monitoring

Chi-Ju Kim a, Juhee Park b, Vijaya Sunkara a, Tae-Hyeong Kim b, Yongjin Lee c, Kyusang Lee d, Mi-Hyun Kim e and Yoon-Kyoung Cho *ab
aDepartment of Biomedical Engineering, School of Life Sciences, Ulsan National Institute of Science and Technology (UNIST), Ulsan 44919, Republic of Korea. E-mail: ykcho@unist.ac.kr
bCenter for Soft and Living Matter, Institute for Basic Science (IBS), Ulsan 44919, Republic of Korea
cLive Cell Instrument, Seoul 01788, Republic of Korea
dClinomics Inc., Ulsan 44919, Republic of Korea
eDepartment of Internal Medicine, Pusan National University School of Medicine and Biomedical Research Institute, Pusan National University Hospital, 179, Gudeok-ro, Seo-Gu, Busan 49241, Republic of Korea

Received 9th February 2018 , Accepted 9th April 2018

First published on 9th April 2018


The potential utility of circulating tumour DNA (ctDNA) in patient blood for cancer diagnostics and real-time monitoring of disease progression is highly recognized. However, the lack of automated and efficient methods for cell-free DNA (cfDNA) isolation from peripheral blood has remained a challenge for broader acceptance of liquid biopsy in general clinical settings. Here, we demonstrate a lab-on-a-disc system equipped with newly developed, electromagnetically actuated, and reversible diaphragm valves that allows fully automated and rapid (<30 min) isolation of cfDNA from whole blood (>3 ml) to achieve high detection sensitivity by minimizing the degradation of fragile ctDNA as well as contamination of wild-type DNA from abundant blood cells. As a proof of concept study, we used the lab-on-a-disc to isolate cfDNA from patients with non-small cell lung cancer and successfully detected epidermal growth factor receptor gene mutations (L858R, T790M) during targeted drug therapy. The proposed lab-on-a-disc enables a fully automated, rapid, and point-of-care cfDNA enrichment starting from whole blood to facilitate the wide use of liquid biopsy in routine clinical practice.


Introduction

Recent advances in genotyping and next-generation sequencing technologies have expedited the acceptance of liquid biopsy as a clinically viable alternative to tissue biopsy.1–3 Among emerging biomarkers for liquid biopsy, circulating tumour DNA (ctDNA), which comprises fragmented genomic DNA shed from tumour tissue circulating freely in the bloodstream, has been actively explored as a non-invasive ‘real-time’ biomarker for cancer diagnosis, prognosis, and treatment monitoring.4–8 Detection of the tumour-associated mutations found in ctDNA allows the characterization of heterogeneous cancers carrying heterogeneities as well as longitudinal surveillance of disease progression.9–11 For example, Bettegowda et al. reported that point mutations in genomic DNA were found in ctDNA from 75% of 640 patients with various advanced cancers and 82% of patients with metastatic cancer; these rates were higher than those reported for most conventional biomarkers.4

Detection of somatic mutations in epidermal growth factor receptor (EGFR) gene from ctDNA while monitoring the response to treatment has been demonstrated in patients with non-small cell lung cancer (NSCLC).12,13 According to a large number of studies, most NSCLCs with EGFR mutations respond to tyrosine kinase inhibitor (TKI) therapy such as gefitinib, erlotinib and afatinib.14–18 However, patients undergoing EGFR-TKI therapy usually develop resistance after approximately 10 months of the treatment through occurrence of the secondary point mutation of T790M.19 In particular, Zheng et al. reported that the EGFR T790M mutation occurred in 47% of 117 patients with NSCLC during TKI treatment with acquired EGFR-TKI resistance.13 In addition, a few recent studies demonstrated concordance between the level of T790M mutation in ctDNA and that in tissue samples.20,21 Recently, third-generation EGFR inhibitors have emerged as potential therapeutics to block the growth of EGFR T790M-positive tumours;22 thus, the early detection of T790M mutation has become critical to maximise the effect of third generation TKI therapy. However, invasive procedure of tumour tissue biopsy hinders frequent biopsy to facilitate timely detection. Therefore, the option of ‘real-time’ monitoring of the mutation level from blood plasma during the period of drug treatment is highly desired. Liquid biopsy is less invasive and therefore one can use it more frequently for the monitoring of the drug response. However, it is currently available only in big hospitals.

The commercially available kits widely used to purify ctDNA utilise packed silica columns or silica-coated magnetic beads. The basic mechanism of the technology relies on the properties of silica surfaces to bind nucleic acid in the presence of the chaotropic reagents.23 Although there has been significant improvement in terms of the yield and the purity of ctDNA isolation steps, most methods still possess limitations against their potential clinical use owing to the requirements of manual handling and peripheral apparatuses, as well as long processing time.

Furthermore, the largest challenges in ctDNA purification include the fragile nature of the relatively short fragmented DNA in addition to the intrinsic rareness of tumour-derived DNA in the peripheral blood. Previous studies have reported that the half-life of cell-free DNA (cfDNA) is between 16 minutes and 2.5 hours, and that it is important not only to minimise the degradation of ctDNA but also to remove the blood cell-originated wild-type genomic DNA.5,24–26

Therefore, it would likely be preferable for the cfDNA isolation to be performed on-site immediately following blood sample collection. However, the lack of efficient and automated cfDNA purification methods that can be directly employed in small local clinics has resulted in a considerable bottleneck for broader adaptation of the technology. If ‘on-site’ isolation of cfDNA is possible in local hospitals, it will provide not only more sensitive detection of cancer biomarkers but also offer the possibility of more frequent monitoring of cancer therapy. To maximise the potential use of cfDNA for personalised cancer therapy and early cancer diagnostics, it is thus important to perform cfDNA purification in local hospitals as soon as the blood samples are collected.

Several previous studies have reported microfluidic chip-based DNA purification.27,28 For example, solid-phase extraction on a microfluidic chip was developed to purify the DNA from whole blood using silica-coated magnetic beads.29–31 In addition, centrifugal microfluidic devices for DNA isolation from bacteria and viruses were developed by integrating magnetic bead-based protocols.31–33 However, although these methods provide some level of benefit in terms of simple operation steps, cost, and time, they are not directly applicable for ctDNA purification owing to the lack of fully integrated plasma separation steps and capability of processing large volumes (>3 ml) of peripheral blood, which is essential because of the extreme rarity of the ctDNA.

Current isolation protocols require complex, time-consuming, and manual procedures including blood sample collection, shipping to central laboratories, plasma separation by centrifugation, and cfDNA purification as schematically illustrated in Scheme 1. Although several studies have reported that particular blood tubes prolonged the preservation times to obtain intact ctDNA, automatic isolation of cfDNA as soon as the blood sample is collected is highly desirable.34,35 Here, we have developed a lab-on-a-disc that can isolate cfDNA from whole blood (3 ml) in a short time (30 min) in a fully automated manner. The newly developed, reversible, and electromagnetically actuated diaphragm valves integrated on a disc enables the whole process of cfDNA purification including the plasma separation from peripheral blood, residual protein lysis, DNA binding, and elution using a table-top, custom-designed centrifuge system. As a proof of concept study, we demonstrated the utility of the proposed lab-on-a-disc for monitoring EGFR mutation from the isolated cfDNA during the course of EGFR-TKI therapy of patients with NSCLC. To our best knowledge, this constitutes the first demonstration of a point-of-care testing-based fully automated microfluidic device for cfDNA purification from whole blood that has the potential to be routinely practiced in clinical settings for potential liquid biopsy applications.


image file: c8lc00165k-s1.tif
Scheme 1 Workflow of cfDNA-based liquid biopsy.

Experimental

Instruments

As shown in Fig. 1a, the experimental set-up for the centrifugal microfluidic control was designed and used to control the linear stage motor (EDB2000-28V24-S, ERAETECH) for the positioning of the valve actuator as well as the main motor (EDB2000-56V24/48-S, ERAETECH) for the spinning of the disc.
image file: c8lc00165k-f1.tif
Fig. 1 Fully automated cfDNA isolation disc and its operation system. a) Schematic diagram (left) and photo image (right) of a point-of-care-type system to operate a lab-on-a-disc for fully automated cfDNA isolation. b) Illustrations showing the injection-molded ID valves when the valve actuator opens (left) or closes (right) the microfluidic channel. By setting the polarity of the electromagnet to be repulsive, the permanent magnet can move down to squeeze the elastomer to block the channel or reversibly be pulled back to the original position by attractive magnetic force by simply changing the polarity of the electromagnet. To open the channel, the valve actuator housing pushes the inner rim of the valve adaptor, releasing the push pin to the top position and thereby opening the microfluidic channel. c) Top view of a photo image of the lab-on-a-disc with full integration of the total process of cfDNA isolation from whole blood, i.e., plasma separation, protein lysis, cfDNA binding, multiple steps of washing, and elution of the cfDNA, performed within 30 min. d) Schematic illustration of the cross-sectional view showing the principles of cfDNA isolation using trapped silica beads.

The actuation principle of the reversible valve is schematically shown in Fig. 1b. To close the channel, the valve actuator equipped with an electromagnet (N pole facing down) in contact with the S pole of a neodymium magnet is aligned on top of the valve position, and then the polarity of the electromagnet is switched so that the repulsive magnetic force between the S poles of the electromagnet and the permanent magnet moves the push pin down to squeeze the elastomer such that it can block the fluidic channel. Then, the polarity of the electromagnet switches back to the N pole such that the attractive force between the electromagnet and the permanent magnet can return the permanent magnet back to the valve actuator housing. To open the microfluidic channel, the valve actuator moves down such that the bottom tip of the actuator housing is pushing the inner rim of the valve adaptor, and the pushpin is thereby moved upwards to a pre-fixed position.

Even though the reversible actuations of the elastomer-based valves were quite common in conventional microfluidic chips,36–39 it was extremely rare in centrifugal microfluidic devices.40,41 We previously reported the reversible and thermally stable actuation of 3D printed ‘push & twist’-type diaphragm valves integrated on a spinning disc, which have critical advantages of thermal stability, vapor-tightness, robustness, and reversible operation.42,43 In this work, the ID valves are modified, injection-molded (Fig. S1a) so that the actuation does not require ‘push & twist’ type action. Instead, alignment of the actuator on top of the valve and electromagnetic switch were enough to have reversible actuation of the valves. In addition, the fabrication step is simplified as shown in Fig. S1 and the operating system is upgraded and miniaturized as shown in Fig. 1a.

Device fabrication

The disc was designed using 3D CAD software, and is composed of two PC (I-Components Co. Ltd.) plates as the body and a PDMS (Dow Corning) layer as an adhesive. A CNC milling machine (Promill Smart 3530, Protek) was used to mill the PC plates according to the design. A 10[thin space (1/6-em)]:[thin space (1/6-em)]1 mixture of a PDMS pre-polymer and a curing agent was spun-coated onto the PC plate (0.2 mm in thickness), and cured in the oven at 65 °C for 4 h. A cutting plotter (CE3000-60, Graphtec) was used to generate the microfluidic channel on the PDMS sheet as the adhesive material for aminosilane-mediated bonding.44 3-Aminopropyltriethoxysilane (APTES, Sigma-Aldrich Corp.) was used to treat the PC plate for the disc assembly as shown in Fig. S1. An aqueous solution of 1% v/v APTES was prepared by mixing with deionised (DI) water and stirred at room temperature (RT) for 20 min. Both milled PC plates and patterned PDMS layers were cleaned with isopropyl alcohol and 70% ethanol, and then treated with oxygen plasma (60 W, Cute Plasma System) for 90 s. The oxygen plasma-activated surface of the PC plate was treated with 1% APTES solution for 20 min. The APTES-treated surface was then washed with DI water and dried by using N2 gas over a short time. Both the plasma-activated PDMS and ATPES-treated PC plates were kept in contact at RT for 10 min to complete the disc assembly.

The fabrication protocols used in this study is for the manufacturing of the lab-scale prototype device, which can be further improved in a large-scale production stage. One of the drawbacks of the current design is that the users need to add the reagents manually, which can be replaced by pre-stored reagents so that the users perform only one-step of manual operation; adding blood samples.45–47 In this study, the disc was CNC milled and bonded with the APTES bonding (Fig. S1b). Even though the bonding strength is good and does not give bonding failure, it needs to be improved for mass production. The valves are injection-molded and attached on the disc to simplify the manufacturing steps. However, we believe experts in industry can further improve the current fabrication method. Furthermore, the device may carry internal control DNA so that the DNA amplification data can be used as positive control for quality control purpose.

cfDNA purification

The fragmented short DNA was prepared by PCR with a pair of primers (5-ACA AAT TTA ACA GCT AAA GAG T-3 and 5-TAG ACA ACG ATG TTT TTA ACA-3) using genomic DNA of Staphylococcus aureus to mimic the 300 bp cfDNA. PCR amplification was performed using the following protocol: 95 °C for 2 min; 40 cycles of 95 °C for 15 s, 52 °C for 30 s, and 72 °C for 30 s, with 100 μl of final PCR volume: 10 μl of 10× PCR buffer (25 mM MgCl2, Solgent), 12 μl of dNTP mixture (10 mM, Solgent), 10 μl of the set of forward and reverse primers (0.5 μM, Macrogen), 20 μl of 5× PCR additives (Solgent), 37 μl of DI water, 6 μl of Taq polymerase (2.5 U μL−1, Solgent), and 10 μl of template DNA. Human serum was purchased from Sigma-Aldrich (H4522). The synthetic short DNA (300 bp) was spiked into human serum and used for further analysis.

For optimization experiments using synthetic short DNA spiked in human serum, the reagents such as proteinase K solution and buffers (AL, AW1, AW2, AE) were from a commercially available QIAamp DNA Blood Mini Kit (51106, Qiagen). All procedures followed the manufacturer's instructions.

For the clinical sample analysis, QIAamp Circulating Nucleic Acid Kit, cat. no. 55114 (Qiagen), was used and the same buffer solutions (proteinase K, ACL, ACB, ACW1, ACW2, and AVE) were used for the disc experiments with the same recommended volume. To perform the disc operation, reagents and samples were preloaded; 3 ml of whole blood, 100 μl of proteinase K, 800 μl of lysis buffer (ACL), 1800 μl of binding buffer (ACB), 600 μl of washing buffer 1 (ACW1), 750 μl of washing buffer 2 (ACW2), 750 μl of ethanol, 200 μl of elution buffer (AVE), and 400 mg of the silica beads (dia. 100 μm). The detailed operation steps used for the Disc operation and the recommended protocol for the commercial product are compared in Table S3, which requires 30 min and 78 min for the Disc and QIAamp Circulating Nucleic Acid Kit, cat. no. 55114 (Qiagen), respectively.

Clinical samples

This study was reviewed and approved by the Institutional Review Board (IRB) of Pusan National University Hospital (PNUH; PNUHIRB-017), and Ulsan National Institute of Science and Technology (UNIST; UNISTIRB-13-002-A), Republic of Korea. Signed informed consent was obtained from all participants and the peripheral blood samples were collected through an IRB consent process. The experiments were performed in accordance with the regulations and guidelines established by these committees. Blood samples were collected into vacutainer tubes (BD Vacutainer) with ethylenediaminetetraacetic acid to prevent blood coagulation. Collected blood samples were processed within 6 h.

Real-time qPCR

To quantify the yield of DNA eluted from the spiked experiment, real-time qPCR was performed using a TaqMan Gene Expression Master Mix kit (4369016, Applied Biosystems Inc.) with a real-time PCR instrument (QuantStudio 6 Flex, Applied Biosystems Inc.) using the following protocol: 50 °C for 2 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 30 s, with 10 μl of final PCR volume: 5 μl of TaqMan Gene Expression Master Mix, 1 μl each of a pair of primers (forward, reverse), 1 μl of probe, 2 μl of DI water, and 1 μl of eluted DNA sample. The primers and probe were purchased from Macrogen, and the concentrations of the stock solution of each primer and probe were 10 μM. The sequence for the forward primer was 5′-TAA AGA GTT TGG TGC CTT TAC AGA-3′, that for the reverse primer was 5′-TTA ACT CAT CAT AGT GGC CAA CAG TTT-3′, and that for the probe was 5′-TAG CAT GCC ATA CAG TCA TTT CAC GC-3′.

Droplet digital PCR

To detect the L858R and T790M mutations from the eluted cfDNA samples, ddPCR was performed using a ddPCR Supermix for Probes (BR186-3023, Bio-Rad) and a ddPCR instrument (QX200, Bio-Rad) with the following protocol: 95 °C for 10 min, followed by 45 cycles of 95 °C for 30 s and 56 °C for 1 min, with 20 μl of final PCR volume: 10 μl of ddPCR Supermix for Probe, no UTP, 1 μl of 20× primer/probe assay (FAM and HEX) for mutants and wild type, and 8 μl of eluted cfDNA sample. The target primers and probe were purchased from Bio-Rad. Analysis of ddPCR data to quantify the copy number of the mutants was performed using QuantaSoft software (Bio-Rad). Genomic DNA of A549 cells was used as a negative control and to decide the cut-off for mutant calling.

Results and discussion

Fully automated cfDNA purification from whole blood on a disc

A custom-designed, table-top-sized operation system to control the spin program and valve actuation was developed (Fig. 1a and b). The actuation mechanism for multiple valves on the spinning disc is similar to our previously reported laser-actuated ferrowax microvalves48 or ‘push & twist’-type diaphragm valves.42 However, the current electromagnetic control of the injection-molded, individually addressable diaphragm (ID) valves significantly simplifies the thermally stable, robust, and reversible actuation of the valves, which enables full automation of complex biological reactions on a disc using a small, point-of-care-type operation system.

The disc is designed to have eleven liquid storage chambers connected by channels with ten reversible ID valves for automatic control of sequential transfer of liquid samples (Fig. 1c and S1). The total process of cfDNA enrichment from whole blood including plasma separation, protein lysis, binding cfDNA on silica beads, washing, and elution could be fully integrated on the disc. The silica beads with 100 μm diameter were trapped in the mixing chamber and used as the substrate for binding of cfDNA (Fig. 1d).

The total process of cfDNA enrichment from whole blood on the spinning disc was imaged by using a custom-designed visualization system equipped with a motor, a CCD camera, and a strobe light (Fig. 2, Movie S1).49,50 The reagents and operation steps are summarised in Tables S1 and S2, respectively. First, plasma samples (>1 ml) are obtained from whole blood (>3 ml) by spinning the discs at 3600 rpm for 5 min (Fig. 2a). For the robust operation, the plasma chamber volume was designed to accommodate broad range of haematocrit values of typical adults; males (42–54%) and women (38–46%). The typical plasma volume is larger than 46%, which is much larger than the current design of 30%. The plasma prepared in the chamber positioned near the centre of the rotation is sequentially transferred to chambers located in the radially outwards direction. Notably, the depth of the blood cell storage chamber, 5.5 mm, is much deeper than the interface, 500 μm, connecting to the plasma storage chamber such that the sediment blood cells remain in the storage chamber during the deceleration to minimise contamination (Fig. S2a and b). Furthermore, tilted groove walls with a slanted angle were introduced in the blood sample chamber to expedite the plasma separation process (Fig. S2c). The shortened sedimentation distance owing to the tilted wall generates regions with a higher cell fraction where cells move downward faster, whereas the clear liquid, plasma, flows upward in low cell density regions.51


image file: c8lc00165k-f2.tif
Fig. 2 Snapshot images of the spinning disc during the total process of cfDNA purification from whole blood. a) Disc spinning for plasma separation; b) valve #1 and #2 opening for proteinase K addition and mixing the plasma with lysis buffer; c) valve #3 opening and mixing the binding buffer with lysate; d) valve #4 opening and valve #5 closing, mixing of the binding mixture with silica beads for 1 min, and repeating for a total of six times to process 1 ml of plasma sample; e) opening valve #6 to wash the DNA-bound beads, and then opening valves #7 and #8 for additional sequential washing; f) closing valve #5 and opening valve #9 to transfer the elution buffer and mix the elution buffer with DNA-bound beads, and then opening valve #10 to transfer the eluted DNA sample. Whole blood and coloured dyes were used for easy visualization. Numbered circles indicate the status of the valve (blue: open, red: closed). Solid yellow arrows indicate the flow of fluids, and dashed yellow arrows indicate the mixing mode of the disc.

After the plasma separation step, the disc stops, valves #1 and #2 are opened, and then the disc rotates with spin step #2 to transfer 1 ml of plasma sample to chamber 3, pre-loaded with lysis buffer, through chamber 2 separately containing pre-loaded proteinase K solution (see details of the operation conditions in Table S2). The plasma solution is then well mixed with proteinase K and lysis buffer using the mixing mode (Fig. S3) and incubated at 60 °C for 5 min to promote the effective release of cfDNA by lysing the residual proteins, lipids, and DNases (Fig. 2b). Next, the lysate solution is mixed with the binding buffer (Fig. 2c) and then added into the bead chamber (Fig. 2d) for binding of cfDNA on silica beads. The detailed operation is as follows. Valve #3 is opened and the binding buffer stored in chamber 4 is added to the lysate to increase the absorption strength of nucleic acids on the surface of the silica beads. After valve #4 is opened, the lysate solution (approximately 650 μl) is transferred into the silica bead chamber and incubated for 1 min in mixing mode. After the binding step, the solution in the bead chamber is transferred to the waste chamber by spinning the disc in the closed and open states of valves #4 and #5, respectively. The binding reaction is repeated 5 more times until the total lysate solution is processed. We note that the reversible nature of the ID valves is essential to accomplish efficient binding of cfDNA from the large volume (3.7 ml) of plasma lysate solution. Next, the cfDNA-bound silica beads are washed by two washing buffers and ethanol to eliminate impurities (Fig. 2e). Finally, the cfDNA is eluted by adding the elution buffer into the bead chamber. After an agitation step, the eluted cfDNA sample is transferred to the eluent chamber (Fig. 2f). The total operation could be finished within 30 min.

Optimization of cfDNA enrichment on the disc

The key variables to achieve efficient solid-phase DNA extraction were optimised using synthetic short DNA (300 base pairs) spiked into the human serum to mimic cfDNA, with the recovery rate quantified by real-time PCR. To achieve a high recovery rate, lysis time was found to have a minimal effect (Fig. 3a), whereas a binding time longer than 1 min per 650 μl of lysate solution was critical (Fig. 3b). In addition, a minimum of 333 mg of silica beads per 1 ml of plasma samples (Fig. 3c) and efficient mixing with a high acceleration rate of 30 rad s−2, (Fig. 3d) provided a high recovery rate.49,53 Finally, comparison tests with two commercially available blood DNA extraction kits (QIAamp DNA Blood Mini Kit (Q-BM) and QIAamp Circulating Nucleic Acid Kit (Q-cfNDA), Qiagen) showed that the recovery rate achieved by the lab-on-a-disc operation was as good as those obtained by manual protocols recommended by the manufacturer as shown in Fig. 3e. It is note to worthy that the process time used for the disc was shorter and the same volume of the reagents were used for the disc and manual operation. The automatic operation by disc resulted in better reproducibility. Furthermore, the lab-on-a-disc was also used to detect EGFR mutation in ctDNA isolated from blood samples of patients with lung cancer. As shown in Fig. 3f the results were as good as those obtained using a commercially available kit specialised for ctDNA isolation (Qiagen QIAamp Circulating Nucleic Acid Kit).
image file: c8lc00165k-f3.tif
Fig. 3 Optimization of DNA binding reaction conditions. Major parameters of the DNA binding reaction were optimised using human serum samples spiked with synthetic fragmented DNA (300 bp); eluted DNA was quantified by real-time PCR. a) Time for protein lysis; b) incubation time for the DNA binding reaction on the surface of silica beads; c) bead amount per plasma sample volume; and d) angular acceleration of the mixing mode were optimised. Except for the test variables, all other parameters were set at the maximum condition (lysis time: 30 min, binding time: 1 min, bead concentration: 666 mg ml−1, angular acceleration: 30 rad s−2). 3 discs were tested for each data point. e) The DNA recovery rates achieved in our device were compared with those obtained by manual operation using two commercial DNA purification kits (QIAamp DNA Blood Mini Kit (Q-BM) and QIAamp Circulating Nucleic Acid Kit (Q-cfNDA), Qiagen)). 3 discs were tested for each data point. f) The EGFR L858R mutant fraction in ctDNA purified by using both lab-on-a-disc (Disc) and a commercial kit (QIAamp Circulating Nucleic Acid Kit, Qiagen) from plasma samples of lung cancer patients previously determined to be positive for the mutation from tissue samples was measured by droplet digital (dd)PCR. One disc is used per patient due to the limitation of the blood sample volume.

Detection of mutant alleles in ctDNA purified from blood samples of patients with NSCLC

To validate the feasibility of our device for clinical testing, we performed droplet digital (dd)PCR with cfDNA purified from 15 blood samples of patients with lung cancer whose cancer tissue carried the EGFR L858R mutation as well as 7 healthy donor samples as a negative control. Detailed patient information is summarised in Table S4. It was shown that 9 of 15 (60%) samples from cancer patients were positive, and all samples from healthy donors (0 of 7 positive, 0%) were negative for the L858R mutation (Fig. 4).
image file: c8lc00165k-f4.tif
Fig. 4 Detection of lung cancer mutation by ddPCR using cfDNA prepared using the lab-on-a-disc. The number of mutant copies in the cfDNA purified by the lab-on-a-disc and quantified by ddPCR from 22 plasma samples (7 normal individuals and 10 patients with NSCLC confirmed as carrying the EGFR L858R mutation in tumour tissue) was determined. One disc is used per patient due to the limitation of the blood sample volume.

Monitoring of cancer therapy via lab-on-a-disc

As a proof-of-concept study, we performed serial monitoring of cfDNA levels confirmed by ddPCR for both EGFR L858R and T790 mutations (Fig. S4) from blood samples together with computerised tomography (CT) images and the size of the target lesion during chemotherapy in a patient with NSCLC (Fig. 5). The patient responded to erlotinib for approximately 3 months as shown in the CT images and by the size of the target lesion (29% decrease). The EGFR L858R level in cfDNA purified by the lab-on-a-disc from 3 ml of whole blood was well correlated with disease progression from the starting point of the blood test to the third sample collection on the 76th day (D+76). However, during the following days from D+76 to D+160, the tumour burden suddenly increased and L858R levels also increased concomitantly. In particular, the T790M mutation was first detected at D+123 and the level of T790M also increased at D+152. These findings demonstrate that the tumour had newly developed resistance to the EGFR TKI treatment. At that time, the anti-cancer drug was changed from erlotinib to osimertinib,54 which is known to be a superior choice for patients with T790M-positive advanced NSCLC on or following EGFR TKI therapy.22,55,56 Upon treatment with the new drug, the tumour burden was decreased and the level of both L858R and T790M mutations also decreased. Notably, the mutational evolution in tumour tissue could be tested at D+145 when the tissue biopsy samples were available because typically, response evaluation is performed after 8–12 weeks of chemotherapy. In contrast, the ctDNA analysis already showed the occurrence of T790M mutant cfDNA at D+123, which clearly demonstrates that the longitudinal ctDNA analysis by liquid biopsy constitutes a potential tool for treatment monitoring during cancer therapy.
image file: c8lc00165k-f5.tif
Fig. 5 Serial monitoring of the level of EGFR mutations during cancer therapy. The copy number of the EGFR mutations (L858R: opened blue circles, T790M: closed red circles) in response to chemotherapeutic treatment (erlotinib: orange stars, osimertinib: green stars) in a patient with NSCLC, along with the tumour burden (CT images) and the size of the target lesion (closed black squares), changes as a function of the time point of blood collection from −21 days to +235 days. The number of days was counted from the first collection of blood samples for the cfDNA analysis. SD: stable disease, PD: progressive disease according to the response evaluation criteria in solid tumour (RECIST).52

Conclusions

We consider that the proposed on-site cfDNA isolation platform may provide an unprecedented tool that may simplify and expedite the total workflow of the cfDNA enrichment process and thereby overcome the instability issues of ctDNA, which is crucial for successful downstream molecular analysis.24,57 To the best of our knowledge, the present study represents the first demonstration of the full integration of cfDNA enrichment starting from whole blood. Using a stand-alone, table-top-sized operation system, the full process of plasma separation, lysis, DNA binding, several washing steps, and elution was fully integrated within 30 minutes. This efficiency was attributed to the robust actuation of reversible ID valves and efficient mixing and washing operations offered by centrifugal microfluidics.

As a proof of concept test, we assessed the proposed lab-on-a-disc for its ability to purify cfDNA from blood samples of patients with NSCLC and EGFR L858R mutations and confirmed that the performance was as good as that of a commercially available kit specialised for ctDNA isolation (Qiagen QIAamp Circulating Nucleic Acid Kit). In addition, the lab-on-a-disc was utilised for the real-time monitoring of disease progress during the EGFR TKI treatment of a patient with NSCLC. Notably, we could confirm that the changes of EGFR mutant (L858R, T790M) copies measured from cfDNA isolated from 3 ml of whole blood using the lab-on-a-disc were correlated with the changes in tumour burden measured by CT image analysis during chemotherapy. Furthermore, we detected the occurrence of resistance to the EGFR TKI treatment mediated by T790M earlier than the mutational analysis performed upon the second tissue biopsy. In addition, we confirmed the dramatic therapeutic effect of the new treatment, T790M-targeted TKI therapy (osimertinib), upon the change of the drug. We propose that a better patient response may have been obtained if the T790M-targeted therapy were given to the patient one month earlier when the levels of the L858R and T790M mutations first increased together with the tumour burden, as indicated by the lab-on-a-disc analysis. We expect that the proposed on-site cfDNA purification system will enable more frequent monitoring of disease status during cancer therapy even in local hospitals, which will maximise the benefits of personalised therapy.

Conflicts of interest

C. J. K., T. H. K., Y. L., K. L., and Y. K. C. filed patents based on the results presented in this paper.

Acknowledgements

The work by J. Park, T. Kim, and Y. Cho was partially supported by IBS-R020-D1. The work by C. Kim, V. Sunkara, and K. Lee was supported by a grant from the Small and Medium Business Administration (SMBA) (S2334629). The work by J. Park, T. Kim, M. Kim, and Y. Cho was supported by a grant from the Korean Health Technology R&D Project of the Ministry of Health & Welfare (HI12C1845).

References

  1. C. Alix-Panabières and K. Pantel, Cancer Discovery, 2016, 6, 479 CrossRef PubMed .
  2. K. R. Chi, Nature, 2016, 532, 269–271 CrossRef CAS PubMed .
  3. G. Siravegna, S. Marsoni, S. Siena and A. Bardelli, Nat. Rev. Clin. Oncol., 2017, 14, 531–548 CrossRef CAS PubMed .
  4. C. Bettegowda, M. Sausen, R. J. Leary, I. Kinde, Y. Wang, N. Agrawal, B. R. Bartlett, H. Wang, B. Luber, R. M. Alani, E. S. Antonarakis, N. S. Azad, A. Bardelli, H. Brem, J. L. Cameron, C. C. Lee, L. A. Fecher, G. L. Gallia, P. Gibbs, D. Le, R. L. Giuntoli, M. Goggins, M. D. Hogarty, M. Holdhoff, S.-M. Hong, Y. Jiao, H. H. Juhl, J. J. Kim, G. Siravegna, D. A. Laheru, C. Lauricella, M. Lim, E. J. Lipson, S. K. N. Marie, G. J. Netto, K. S. Oliner, A. Olivi, L. Olsson, G. J. Riggins, A. Sartore-Bianchi, K. Schmidt, I.-M. Shih, S. M. Oba-Shinjo, S. Siena, D. Theodorescu, J. Tie, T. T. Harkins, S. Veronese, T.-L. Wang, J. D. Weingart, C. L. Wolfgang, L. D. Wood, D. Xing, R. H. Hruban, J. Wu, P. J. Allen, C. M. Schmidt, M. A. Choti, V. E. Velculescu, K. W. Kinzler, B. Vogelstein, N. Papadopoulos and L. A. Diaz, Sci. Transl. Med., 2014, 6, 224ra224 Search PubMed .
  5. F. Diehl, K. Schmidt, M. A. Choti, K. Romans, S. Goodman, M. Li, K. Thornton, N. Agrawal, L. Sokoll, S. A. Szabo, K. W. Kinzler, B. Vogelstein and L. A. Diaz Jr, Nat. Med., 2008, 14, 985–990 CrossRef CAS PubMed .
  6. J. Phallen, M. Sausen, V. Adleff, A. Leal, C. Hruban, J. White, V. Anagnostou, J. Fiksel, S. Cristiano, E. Papp, S. Speir, T. Reinert, M.-B. W. Orntoft, B. D. Woodward, D. Murphy, S. Parpart-Li, D. Riley, M. Nesselbush, N. Sengamalay, A. Georgiadis, Q. K. Li, M. R. Madsen, F. V. Mortensen, J. Huiskens, C. Punt, N. van Grieken, R. Fijneman, G. Meijer, H. Husain, R. B. Scharpf, L. A. Diaz, S. Jones, S. Angiuoli, T. Ørntoft, H. J. Nielsen, C. L. Andersen and V. E. Velculescu, Sci. Transl. Med., 2017, 9, eaan2415 CrossRef PubMed .
  7. H. Schwarzenbach, D. S. B. Hoon and K. Pantel, Nat. Rev. Cancer, 2011, 11, 426–437 CrossRef CAS PubMed .
  8. M. Lim, C.-J. Kim, V. Sunkara, M.-H. Kim and Y.-K. Cho, Micromachines, 2018, 9, 100 CrossRef .
  9. S.-J. Dawson, D. W. Y. Tsui, M. Murtaza, H. Biggs, O. M. Rueda, S.-F. Chin, M. J. Dunning, D. Gale, T. Forshew, B. Mahler-Araujo, S. Rajan, S. Humphray, J. Becq, D. Halsall, M. Wallis, D. Bentley, C. Caldas and N. Rosenfeld, N. Engl. J. Med., 2013, 368, 1199–1209 CrossRef CAS PubMed .
  10. A. M. Newman, S. V. Bratman, J. To, J. F. Wynne, N. C. W. Eclov, L. A. Modlin, C. L. Liu, J. W. Neal, H. A. Wakelee, R. E. Merritt, J. B. Shrager, B. W. Loo Jr, A. A. Alizadeh and M. Diehn, Nat. Med., 2014, 20, 548–554 CrossRef CAS PubMed .
  11. A. R. Thierry, F. Mouliere, S. El Messaoudi, C. Mollevi, E. Lopez-Crapez, F. Rolet, B. Gillet, C. Gongora, P. Dechelotte, B. Robert, M. Del Rio, P.-J. Lamy, F. Bibeau, M. Nouaille, V. Loriot, A.-S. Jarrousse, F. Molina, M. Mathonnet, D. Pezet and M. Ychou, Nat. Med., 2014, 20, 430–435 CrossRef CAS PubMed .
  12. T. K. Sundaresan, L. V. Sequist, J. V. Heymach, G. J. Riely, P. A. Jänne, W. H. Koch, J. P. Sullivan, D. B. Fox, R. Maher, A. Muzikansky, A. Webb, H. T. Tran, U. Giri, M. Fleisher, H. A. Yu, W. Wei, B. E. Johnson, T. A. Barber, J. R. Walsh, J. A. Engelman, S. L. Stott, R. Kapur, S. Maheswaran, M. Toner and D. A. Haber, Clin. Cancer Res., 2016, 22, 1103–1110 CrossRef CAS PubMed .
  13. D. Zheng, X. Ye, M. Z. Zhang, Y. Sun, J. Y. Wang, J. Ni, H. P. Zhang, L. Zhang, J. Luo, J. Zhang, L. Tang, B. Su, G. Chen, G. Zhu, Y. Gu and J. F. Xu, Sci. Rep., 2016, 6, 20913 CrossRef CAS PubMed .
  14. S. Jenkins, J. C. H. Yang, S. S. Ramalingam, K. Yu, S. Patel, S. Weston, R. Hodge, M. Cantarini, P. A. Jänne, T. Mitsudomi and G. D. Goss, J. Thorac. Oncol., 2017, 12, 1061–1070 CrossRef PubMed .
  15. S. Kobayashi, T. J. Boggon, T. Dayaram, P. A. Jänne, O. Kocher, M. Meyerson, B. E. Johnson, M. J. Eck, D. G. Tenen and B. Halmos, N. Engl. J. Med., 2005, 352, 786–792 CrossRef CAS PubMed .
  16. T. J. Lynch, D. W. Bell, R. Sordella, S. Gurubhagavatula, R. A. Okimoto, B. W. Brannigan, P. L. Harris, S. M. Haserlat, J. G. Supko, F. G. Haluska, D. N. Louis, D. C. Christiani, J. Settleman and D. A. Haber, N. Engl. J. Med., 2004, 350, 2129–2139 CrossRef CAS PubMed .
  17. T. Mitsudomi and Y. Yatabe, Cancer Sci., 2007, 98, 1817–1824 CrossRef CAS PubMed .
  18. J. G. Paez, P. A. Jänne, J. C. Lee, S. Tracy, H. Greulich, S. Gabriel, P. Herman, F. J. Kaye, N. Lindeman, T. J. Boggon, K. Naoki, H. Sasaki, Y. Fujii, M. J. Eck, W. R. Sellers, B. E. Johnson and M. Meyerson, Science, 2004, 304, 1497 CrossRef CAS PubMed .
  19. K. Suda, R. Onozato, Y. Yatabe and T. Mitsudomi, J. Thorac. Oncol., 2009, 4, 1–4 CrossRef PubMed .
  20. C. Karlovich, J. W. Goldman, J.-M. Sun, E. Mann, L. V. Sequist, K. Konopa, W. Wen, P. Angenendt, L. Horn, D. Spigel, J.-C. Soria, B. Solomon, D. R. Camidge, S. Gadgeel, C. Paweletz, L. Wu, S. Chien, P. Donnell, S. Matheny, D. Despain, L. Rolfe, M. Raponi, A. R. Allen, K. Park and H. Wakelee, Clin. Cancer Res., 2016, 22, 2386 CrossRef CAS PubMed .
  21. K. S. Thress, R. Brant, T. H. Carr, S. Dearden, S. Jenkins, H. Brown, T. Hammett, M. Cantarini and J. C. Barrett, Lung Cancer, 2015, 90, 509–515 CrossRef PubMed .
  22. S. Wang, S. Cang and D. Liu, J. Hematol. Oncol., 2016, 9, 34 CrossRef PubMed .
  23. R. Boom, C. J. Sol, M. M. Salimans, C. L. Jansen, P. M. Wertheim-van Dillen and J. van der Noordaa, J. Clin. Microbiol., 1990, 28, 495–503 CAS .
  24. Q. Kang, N. L. Henry, C. Paoletti, H. Jiang, P. Vats, A. M. Chinnaiyan, D. F. Hayes, S. D. Merajver, J. M. Rae and M. Tewari, Clin. Biochem., 2016, 49, 1354–1360 CrossRef CAS PubMed .
  25. J. C. M. Wan, C. Massie, J. Garcia-Corbacho, F. Mouliere, J. D. Brenton, C. Caldas, S. Pacey, R. Baird and N. Rosenfeld, Nat. Rev. Cancer, 2017, 17, 223–238 CrossRef CAS PubMed .
  26. W. Yao, C. Mei, X. Nan and L. Hui, Gene, 2016, 590, 142–148 CrossRef CAS PubMed .
  27. C. W. Price, D. C. Leslie and J. P. Landers, Lab Chip, 2009, 9, 2484–2494 RSC .
  28. S. J. Reinholt and A. J. Baeumner, Angew. Chem., Int. Ed., 2014, 53, 13988–14001 CrossRef CAS PubMed .
  29. G. R. M. Duarte, C. W. Price, B. H. Augustine, E. Carrilho and J. P. Landers, Anal. Chem., 2011, 83, 5182–5189 CrossRef CAS PubMed .
  30. G. R. M. Duarte, C. W. Price, J. L. Littlewood, D. M. Haverstick, J. P. Ferrance, E. Carrilho and J. P. Landers, Analyst, 2010, 135, 531–537 RSC .
  31. K. R. Jackson, J. C. Borba, M. Meija, D. L. Mills, D. M. Haverstick, K. E. Olson, R. Aranda, G. T. Garner, E. Carrilho and J. P. Landers, Anal. Chim. Acta, 2016, 937, 1–10 CrossRef CAS PubMed .
  32. O. Strohmeier, A. Emperle, G. Roth, D. Mark, R. Zengerle and F. von Stetten, Lab Chip, 2013, 13, 146–155 RSC .
  33. O. Strohmeier, S. Keil, B. Kanat, P. Patel, M. Niedrig, M. Weidmann, F. Hufert, J. Drexler, R. Zengerle and F. von Stetten, RSC Adv., 2015, 5, 32144–32150 RSC .
  34. C. Pérez-Barrios, I. Nieto-Alcolado, M. Torrente, C. Jiménez-Sánchez, V. Calvo, L. Gutierrez-Sanz, M. Palka, E. Donoso-Navarro, M. Provencio and A. Romero, Transl. Lung Cancer Res., 2016, 5, 665–672 CrossRef PubMed .
  35. J. L. Sherwood, C. Corcoran, H. Brown, A. D. Sharpe, M. Musilova and A. Kohlmann, PLoS One, 2016, 11, e0150197 Search PubMed .
  36. C.-Y. Chen, C.-H. Chen, T.-Y. Tu, C.-M. Lin and A. M. Wo, Lab Chip, 2011, 11, 733–737 RSC .
  37. X. Li, P. Zwanenburg and X. Liu, Lab Chip, 2013, 13, 2609–2614 RSC .
  38. J. C. Harper, J. M. Andrews, C. Ben, A. C. Hunt, J. K. Murton, B. D. Carson, G. D. Bachand, J. A. Lovchik, W. D. Arndt, M. R. Finley and T. L. Edwards, Lab Chip, 2016, 16, 4142–4151 RSC .
  39. M. Rahbar, L. Shannon and B. L. Gray, J. Micromech. Microeng., 2016, 26, 055012 CrossRef .
  40. R. Gorkin, J. Park, J. Siegrist, M. Amasia, B. S. Lee, J.-M. Park, J. Kim, H. Kim, M. Madou and Y.-K. Cho, Lab Chip, 2010, 10, 1758–1773 RSC .
  41. O. Strohmeier, M. Keller, F. Schwemmer, S. Zehnle, D. Mark, F. von Stetten, R. Zengerle and N. Paust, Chem. Soc. Rev., 2015, 44, 6187–6229 RSC .
  42. T.-H. Kim, V. Sunkara, J. Park, C.-J. Kim, H.-K. Woo and Y.-K. Cho, Lab Chip, 2016, 16, 3741–3749 RSC .
  43. T.-H. Kim, C.-J. Kim, Y. Kim and Y.-K. Cho, Sens. Actuators, B, 2018, 256, 310–317 CrossRef CAS .
  44. V. Sunkara and Y.-K. Cho, ACS Appl. Mater. Interfaces, 2012, 4, 6537–6544 CAS .
  45. J. Hoffmann, D. Mark, S. Lutz, R. Zengerle and F. von Stetten, Lab Chip, 2010, 10, 1480–1484 RSC .
  46. S. Lutz, P. Weber, M. Focke, B. Faltin, J. Hoffmann, C. Muller, D. Mark, G. Roth, P. Munday, N. Armes, O. Piepenburg, R. Zengerle and F. von Stetten, Lab Chip, 2010, 10, 887–893 RSC .
  47. T. van Oordt, Y. Barb, J. Smetana, R. Zengerle and F. von Stetten, Lab Chip, 2013, 13, 2888–2892 RSC .
  48. J.-M. Park, Y.-K. Cho, B.-S. Lee, J.-G. Lee and C. Ko, Lab Chip, 2007, 7, 557–564 RSC .
  49. B. S. Lee, J.-N. Lee, J.-M. Park, J.-G. Lee, S. Kim, Y.-K. Cho and C. Ko, Lab Chip, 2009, 9, 1548–1555 RSC .
  50. B. S. Lee, Y. U. Lee, H.-S. Kim, T.-H. Kim, J. Park, J.-G. Lee, J. Kim, H. Kim, W. G. Lee and Y.-K. Cho, Lab Chip, 2011, 11, 70–78 RSC .
  51. T.-H. Kim, H. Hwang, R. Gorkin, M. Madou and Y.-K. Cho, Sens. Actuators, B, 2013, 178, 648–655 CrossRef CAS .
  52. C. M. Costelloe, H. H. Chuang, J. E. Madewell and N. T. Ueno, J. Cancer, 2010, 1, 80 CrossRef PubMed .
  53. M. Grumann, A. Geipel, L. Riegger, R. Zengerle and J. Ducree, Lab Chip, 2005, 5, 560–565 RSC .
  54. T. S. Mok, Y.-L. Wu, M.-J. Ahn, M. C. Garassino, H. R. Kim, S. S. Ramalingam, F. A. Shepherd, Y. He, H. Akamatsu, W. S. M. E. Theelen, C. K. Lee, M. Sebastian, A. Templeton, H. Mann, M. Marotti, S. Ghiorghiu and V. A. Papadimitrakopoulou, N. Engl. J. Med., 2016, 376, 629–640 CrossRef PubMed .
  55. O. Juan and S. Popat, Ther. Adv. Med. Oncol., 2017, 9, 201–216 CrossRef CAS PubMed .
  56. J. Ni, L. Weng, Y. Liu, Z. Sun, C. Bai and Y. Wang, Oncol. Lett., 2017, 13, 4549–4557 CrossRef PubMed .
  57. N. Normanno, M. G. Denis, K. S. Thress, M. Ratcliffe and M. Reck, Oncotarget, 2017, 8, 12501–12516 CrossRef PubMed .

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c8lc00165k

This journal is © The Royal Society of Chemistry 2018