A flux-adaptable pump-free microfluidics-based self-contained platform for multiplex cancer biomarker detection

Abstract

Microfluidics drives technological advancement in point-of-care (POC) bioanalytical diagnostics towards portability, fast response and low cost. In most microfluidic bioanalytical applications, flowing antigen/antibody reacts with immobilized antibody/antigen at a constant flux; it is difficult to reach a compromise to simultaneously realize sufficient time for the antigen–antibody interaction and short time for the entire assay. Here, we present a pump-free microfluidic chip, in which flow is self-initialized by capillary pumping and continued by imbibition of a filter paper. Microfluidic units in teardrop shape ensure that flow passes through the reaction areas at a reduced flux to facilitate the association between antigen and antibody and speeds up after the reaction areas. By spotting different antibodies into the reaction area, four types of biomarkers can be measured simultaneously in one microfluidic chip. Moreover, a small-sized instrument was developed for chemiluminescence detection and signal analysis. The system was validated by testing four biomarkers of colorectal cancer using plasma samples from patients. The assay took about 20 minutes. The limit of detection is 0.89 ng mL⁻¹, 1.72 ng mL⁻¹, 3.62 U mL⁻¹ and 1.05 U mL⁻¹ for the assays of carcinoembryonic antigen, alpha-fetoprotein, carbohydrate antigen 125 and carbohydrate antigen 19-9, respectively. This flux-adaptable and self-contained microfluidic platform is expected to be useful in various POC disease-monitoring applications.

---

Introduction

The high incidence and mortality rate of colorectal cancer (CRC) makes it a huge burden worldwide.^1–3 The mortality of CRC ranks 2nd in the United States and the incidence rate increased by 1% annually among individuals aged 50 to 64 years and approximately 2% among those aged younger than 50 years, as reported by the American Cancer Society.⁴ In China, CRC is the 3rd most commonly diagnosed carcinoma and the 5th most common cause of cancer-related death.⁵ Early diagnosis and personalized treatment of CRC can improve the survival rate.^6,7 Serological testing is particularly helpful in the evaluation and treatment of CRC, aiding in the assessment of the degree of malignancy and monitoring of the advances of therapy as well as providing valuable prognostic information. A strong relationship between the clinical and pathological parameters of CRC and a group of glycoprotein and carbohydrate biomarkers, such as carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), carbohydrate antigen 125 (CA125) and carbohydrate antigen 724 (CA724), has been confirmed in many research studies.^8–10 CEA and CA19-9 have been suggested in the clinical management for monitoring patients who might have CRC and have been diagnosed with CRC.^11–13

Although each biomarker is correlated with a large proportion of patients with specific disease characteristics, it would be less meaningful when used alone as a screening tool owing to its limited specificity. Not one of the cancer biomarkers has presented good utility as an independent predictive biomarker.¹⁴ In practice, the use of a single biomarker is not recommended because the genome instability of cancers in mutually dependent and highly complex biological systems would make the assay not reliable.¹⁵ Instead, it is widely accepted that multiplex biomarkers are evaluated simultaneously for a comprehensive assessment.¹⁶

Microfluidics technology drives the development of bioassays in many aspects, including reduced volume of reaction system, shortened reaction time, enhanced detection limit, improved throughput and simplified operation.^17–22 Microfluidic chips can be flexibly designed to realize the detection of multiple biomarkers in immunoassays.^23–26 Chemiluminescence immunoassay (CLIA), in which a chemical probe generates light when reacted with an enzyme label, is a sensitive, simple approach; no excitation light sources or emission filters are needed. A microfluidic chip containing 1536 chambers was demonstrated to quantify 384 biomarkers by measuring luminescence in a microplate reader.²⁷ To improve the efficiency of the immunoassay for detecting two prostate cancer biomarkers, a three-dimensional (3D) printed chip was fabricated, in which the sample and reagents could be preloaded, and a programmable syringe pump was employed for driving the sample and reagents into a 3D mixer and a detection chamber according to the assay protocol.²⁸ Furthermore, an automated microfluidic platform was developed for both direct sandwich immunoassay (assay for C-reactive protein) and competitive immunoassay (assay for testosterone). The injection of preloaded reagents was controlled by six on-chip valves, a valve actuator and a peristaltic pump.²⁹ In previous research work, rapid, sensitive and high-accuracy CLIAs for detecting multiple biomarkers have been successfully demonstrated. Nevertheless, in these prototypes, a group of valve and pump systems is required for loading reagents, sample and wash buffer, making the whole system complicated.

Passive pumping microfluidics has the potential to address the problem of the utilization of external bulky pumping systems.^30,31 Usually, passive fluid infusion is based on gravity-driven or capillary-driven mechanisms. In the gravity-driven microfluidic chip, the fluid was driven into the chip by setting a hydraulic head difference between a reservoir and a drainage.^32–35 Viewed in terms of energy, the infusion is attributed to gravitational potential energy that requires enough difference in height. Gravity-driven microfluidic chips are suitable for cell culture applications in which a large amount of culture medium could be stored in the reservoirs for long-term cell culture. Microfluidic chips harnessing capillary action are appropriate for low-volume applications because capillary force plays a dominant role and could drive fluid to flow in a tiny space with dimensions much smaller than the capillary length. Capillary pump-based microfluidic manipulation has been realized by designing proper microstructures and controlling the surface wettability.^36–38 Capillary-driven microfluidics have been used in immunoassays.^39–42 However, only a few demonstrated quantitative multiplex detection of cancer biomarkers. Besides, microchannels in these microfluidic chips had a uniform size; thus, the flux of the flow through the entire microchannel was fixed. It was not possible to simultaneously satisfy both strong antigen–antibody interaction, in which low flux was preferred, and rapid assay, in which high flux was required.

The behavior of microfluid in the microchannels is dependent on the geometric structures of the microchannels. Microchannels with a special design can be developed into a variety of functional devices. Teardrop-shaped micro-cavities, which have a simple structure, have been used in many applications. In a microfluidic reflective multicolor display, a dot pixel matrix was formed by a sequence of teardrop-shaped micro-cavities.⁴³ Since the teardrop-shaped micro-cavities ensured that the water could be retained in the specific dot pixels, an image could be maintained in the display using dyed water droplets and air gaps with no energy consumption. In addition, teardrop-shaped cavities could be used to trap microbubbles. By oscillating microbubbles confined in a group of teardrop-shaped cavities upon acoustic actuation, complex flow patterns could be generated.⁴⁴ Teardrop-shaped cavities which were developed for cell trapping could be used to assemble precise constellations of cell clusters.⁴⁵

In this paper, we present a pump-free microfluidic chip and a CLIA serological analysis platform for measuring multiple CRC-related biomarkers, including CEA, alpha-fetoprotein (AFP), CA125 and CA19-9. The initial self-triggered start-up and the following continuous flow of plasma sample and reagents without using any peripheral pumping apparatuses make the entire assay simple and straightforward. The teardrop-shaped design of the reaction units allows the flow to slow down around the reaction areas and speed up in the non-reaction areas, which is beneficial to achieve sufficient antigen–antibody reaction within a short assay time. The microfluidic platform is validated and evaluated using clinical samples of patients.

Materials and methods

Chemicals

CEA antigen, AFP antigen, CA125 cancer antigen and CA19-9 cancer antigen, which were used as the standards, were purchased from BiosPacific Inc., USA. Human sera were purchased from Sigma-Aldrich, USA (S7023). Capture antibodies including mouse anti-CEA monoclonal antibody, mouse anti-AFP monoclonal antibody, mouse anti-CA125 monoclonal antibody and mouse anti-CA19-9 monoclonal antibody were obtained from BiosPacific Inc. Goat anti-AFP polyclonal antibody and goat anti-CEA polyclonal antibody as the detection antibodies were purchased from BiosPacific Inc. Then, horseradish peroxidase (HRP)-conjugated donkey anti-goat antibody (Jackson ImmunoResearch Laboratories, Inc., USA) was used as the secondary antibody against the detection antibodies of CEA and AFP biomarkers. HRP-conjugated mouse anti-CA19-9 monoclonal antibody and HRP-conjugated mouse anti-CA125 monoclonal antibody as the detection antibodies against CA19-9 and CA125 antigens were purchased from Wason Biotech Inc. Phosphate-buffered saline (PBS) with 0.05% Tween-20 (PBST) used as the wash buffer was purchased from Thermo Fisher Scientific. The response of human serum albumin (HSA) was used as a positive control in the assay. Mouse monoclonal antibody against HSA (Santa Cruz Biotechnology, Inc., USA) was used as the capture antibody. Goat anti-albumin polyclonal antibody (Sigma-Aldrich, USA) was used as the detection antibody. The secondary antibody for HSA detection was also HRP-conjugated donkey anti-goat antibody. Bovine serum albumin (BSA) was purchased from Solarbia, Co., China.

Preparation of the standards

Standard samples were prepared by spiking four kinds of antigens into the human sera (S7023, Sigma-Aldrich, USA) respectively with known concentration. The gradient concentration for CEA and AFP standards was from 1 ng mL⁻¹ to 1 μg mL⁻¹ and the concentration for CA125 and CA19-9 standards was from 2 U mL⁻¹ to 1250 U mL⁻¹.

Collection of blood samples

Blood samples were collected from 30 patients with colorectal cancer who were under treatment in Shanghai Cancer Center, Fudan University. All patients have signed informed consent, and samples were collected under ethical approval (certificate no.: 050432-4-1212B). Plasma was collected using pro-coagulation tubes.

Pump-free microfluidic chip

Fig. 1 shows the schematic diagram and photo of the pump-free microfluidic chip. The details of the fabrication process of the microfluidic chip are described in the ESI† and illustrated in Fig. S1. 0.5 μL of 0.2 mg mL⁻¹ monoclonal antibody in protein spotting buffer A (CapitalBio Technology, China) for each biomarker was first coated on a substrate whose surface was covalently modified to contain an aldehyde functional group (OPPolymerSlide™ D, CapitalBio Technology, China) by using CapitalBio SmartArrayer™ 136 (CapitalBio Technology, China). The CEA, AFP and HSA capture antibodies were coated on one side which would be later covered by a microchannel, while the CA125 and CA19-9 antibodies were coated on the other side which would be covered by another microchannel. The coating area for the each capture antibody was a spot of 1 mm diameter. Then, the substrate was soaked in 5% BSA-PBS for 1 hour at room temperature. The substrate coated with the capture antibodies was stored at 4 °C for at least 3 months.

Fig. 1 (a) Fabrication process of the pump-free microfluidic chip. (b) Photo of the microfluidic chip. (c) The flow is driven by the capillary force and the gravitational force in the initialization stage. (d) The continuous flow resulted from the suction of the filter paper during the post-initialization stage.

Meanwhile, two microchannels were patterned on a silicon wafer by photolithography. Polydimethylsiloxane (PDMS) with a mixture ratio of 10 : 1 (elastomer versus curing agent) was used to cast the pattern and solidified at 80 °C for 4 hours. 4 mm radius inlet holes were created by punching through the PDMS at the end of the microchannels. The thickness of the PDMS layer is 2 mm. The volume of the inlet hole is 100.48 μL. In each microchannel, there were three successive teardrop-shaped units. The inlet and the outlet were connected to the wide end of the first unit and the tapered end of the last unit, respectively, through two straight channels. Fig. S2† shows the structure of the microchannel in detail. Finally, the PDMS layer was placed on the substrate. Each unit was used as a reaction cavity for one biomarker and the wide end of every unit had a coating area for each antibody. The outlet hung over the edge of the substrate and opened downward.

Serological analysis platform (SAP)

The design and the photo of the SAP are shown in Fig. 2. An imaging system was inside the SAP to capture an image of the reaction products. The image was captured by a monochrome charge-coupled device (CCD) camera (FLIR Grasshopper®3, Edmund Optics, USA) via an aspherical lens (effective focal length = 50 mm) (Edmund Optics, USA) and an aluminum plane mirror (Edmund Optics, USA). The field of view was 16.2 mm × 20.3 mm. A chip holder was above the imaging system for holding the microfluidic chip. A disposable tray, well fitting into the chip holder, was designed to avoid cross-contamination. On one end of the tray, there was a groove filled with a stack of filter paper (GB002, Whatman, England). When the microfluidic chip was placed on the tray, the outlets, which were hung over the edge of the chip, were attached to the filter paper. The chip holder could be tightly closed by magnetic attraction. After the chip holder was closed, the imaging system aimed at the area of six teardrop-shaped units in the dark. The frame of the SAP was made of aluminum alloy and produced by computer numerical control (CNC) machining. The disposable tray was produced by 3D printing. The size of the platform was 183 mm × 123 mm × 70 mm and the weight was 1.2 kg.

Fig. 2 (a) Schematic diagram of the serological analysis platform (SAP). (b) Photo of the platform with the microfluidic chip on the disposable tray placed on the chip holder.

Assay protocol for detecting cancer biomarkers

The procedure of the assay is shown in Fig. 3 and demonstrated in Video S1.† In the assay, the microfluidic chip was placed on the disposable tray, whose groove was filled with filter paper. The disposable tray was placed on the chip holder. When loading sample and reagents, the chip holder was pulled out from the SAP. 20 μL of sample (10 μL plasma diluted with 10 μL PBS) was added into the inlet of each microchannel. After about 10 minutes and before the sample drains away in the inlet, 20 μL of mixed CEA, AFP and HSA detection antibodies and secondary antibody with a mixture ratio of 1 : 1 was added to the microchannel for the assay of CEA, AFP and HSA biomarkers, and 20 μL of CA125 and CA19-9 detection antibody was added to the other microchannel for the assay of CA125 and CA19-9 biomarkers. 4 minutes later, 10 μL of PBST was added per microchannel to flush the unbound antibodies away. After 4 minutes, 5 μL of chemiluminescent peroxidase substrate, i.e. luminol-based solution (catalog number: CPSOC, Sigma-Aldrich, USA), was added per microchannel. The imaging system was triggered after 2 minutes to acquire the image of the reaction products. The exposure time is set as 6 seconds. Six 20 × 20 pixel blocks were extracted from the reaction areas in the six units. The average chemiluminescence intensity was calculated for each pixel block. The values for cancer biomarkers of interest (AFP, CEA, CA125 and CA19-9), I_Biomarker, were calibrated with respect to the response of the HSA, I_Albumin, and the background intensity of the blank unit, I_Blank. Whatever the albumin level varies among patients, the response of the HSA measured in the microfluidic chip would be similar (Fig. S3†), because the dose of HSA in the plasma is high⁴⁶ and significantly in excess of the amount of the HSA capture antibody. Thus, the response of the albumin could be used as a positive control. The resultant response of a specific cancer biomarker can be expressed aswhere I_{Albumin_Control} is the reference intensity obtained in the preliminary experiment (Fig. S3†). Finally, the concentration of the biomarkers could be quantitatively figured out based on standard curves, which were established by the relation of a set of biomarker concentrations to the corresponding chemiluminescence intensity (additional details provided in the ESI†).

Fig. 3 Procedure of the assay for detecting multiple biomarkers.

Results and discussion

Initialization stage: flow triggering

In the pump-free microfluidic chip, the flow experiences two stages. Initially, the microchannel is empty and the fluid in the inlet slowly moves towards the outlet through the three teardrop-shaped units. The motion is driven by the gravitational force, F_Gravity, generated by the fluid itself in the inlet and the capillary force, F_Capillary, due to the interaction between the fluid and the microchannel walls. Once the fluid reaches the outlet where the filter paper is placed, the fluid is driven by the capillary suction, F_Suction, of the filter paper in addition to the gravitational force, as shown in Fig. 1c and d.

Fig. 4a shows the self-initialized flow in the successive teardrop-shaped units. The units are designed with a fixed volume, i.e. area of 5 mm² and height of 20 μm, and variable taper angle, θ. The fluid in the microchannel with a large taper angle flows faster and reaches the outlet earlier.

Fig. 4 Analysis of the flow initialization in the microchannel. (a) Montage of the flow in the microchannel. (b) Contact angles of the plasma on the PDMS and the substrate. (c) The calculated capillary pressure along the flow direction in a single teardrop-shaped unit when the taper angle is different. Inset: the structure of a single teardrop-shaped unit. (d) The calculated hydraulic resistance of the microchannels. (e) Time duration for filling up the three units in the initialization stage. Red curve: theoretical calculation. Dots: experimental measurement.

In the initialization stage, the fluid is driven by its own gravitational force as well as the capillary force. The pressure exerted by gravitational force, i.e. hydrostatic pressure, is

P_G = ρgh (2)

where ρ is the density of the plasma, g is the gravitational acceleration, and h is the height of the fluid in the inlet.

Furthermore, the Young–Laplace equation can be applied to describe the capillary pressure over the interface of plasma and air in the microchannel. The pressure drop in the tear-shaped unit with a rectangular cross section can be expressed as^36,47,48

where γ is the surface tension of the plasma, h is the height of the microchannel, and W(z) is the position-dependent width of the microchannel along the flow direction, as follows:φ_PDMS and φ_Substrate are the contact angles of the plasma on the PDMS and the substrate. PDMS has a hydrophobic surface, forming a contact angle, φ_PDMS, of about 107°, as shown in Fig. 4b. On the glass substrate, an aldehyde group is introduced to activate the surface. Therefore, the glass substrate has a hydrophilic surface and the contact angle, φ_Substrate, is 41°. α(z) is related to the tangential angle of the side walls, which is

Fig. 4c shows the capillary pressure along the flow direction in the unit. The high wettability of the glass substrate allows the fluid to spread over the bottom surface, while the hydrophobic characteristics of the PDMS top and side walls adversely affect the flow of the fluid. Since the microchannel has a low aspect ratio (height-to-width ratio), the interaction of the fluid and the top and bottom walls dominates the capillary pressure. Moreover, the surface hydrophilicity of the glass substrate is relatively stronger than the hydrophobic effect on the top PDMS wall. Thus, the hydrophilicity of the glass substrate provides sufficient driving force to pull the fluid forward.

Considering the capillary effect as well as hydrostatic pressure in Hagen–Poiseuille's law, the volumetric flow rate, Q, of the flow overcoming hydraulic resistance, R, can be written as

where is the average capillary pressure over the microchannel, and the hydraulic resistance of the three teardrop-shaped units can be estimated aswhere μ is the viscosity of the plasma.

In the calculation, the density and surface tension of the plasma are 1.06 × 10³ kg m⁻³ and 55.89 × 10⁻³ N m⁻¹,⁴⁹ respectively. 20 μL plasma is in the 4 mm radius inlet. The slight change of the liquid level in the inlet is neglected. Fig. 4d depicts the hydraulic resistance of the microchannel with different taper angles. The hydraulic resistance decreases with the increase of the taper angle because the microchannel becomes wide and the total length is shortened.

Then, the volumetric flow rate can be calculated using eqn (6). Furthermore, the time for filling up the three units from the moment when the plasma was added into the inlet to the moment it reaches the filter paper can be estimated. The estimated time and the measured results are shown in Fig. 4e. The measured time has the same tendency as that in the calculation. The deviation between the theoretical estimation and the measured results could be attributed to the fabrication error of the microchannels. Fig. S4† shows the scanning electron microscope (SEM) images of the microchannels. The height of the microchannel is 16 μm, which is slightly lower than that in the design. The reduced height leads to higher hydraulic resistance and lower volumetric flow rate. As a result, it requires a little bit more time to fill up the microchannels. Since all the microchannels were replicated from the patterns on the same silicon wafer, the height is identical and the consistency can be ensured.

In the initialization stage, the flow which is free from any power-driven pump could be self-triggered and fill up the microchannel rapidly. The time duration for the initialization in the unit with a large taper angle is short because of high capillary pressure and low hydraulic resistance. In the following analysis and the assay, microfluidic chips consisting of two microchannels with a taper angle of 7° are used. The dimensions of the microfluidic chips are illustrated in Fig. S5.† Fig. S4† shows the SEM images of the chip.

Post-initialization stage: continuous flow with adaptable flux

When the fluid reaches the outlet, the capillary suction of the filter paper, F_Suction, and the gravitational force, F_Gravity, of the fluid in the inlets work together to drive the fluid and the capillary suction plays a dominant role. The capillary suction of the filter paper was investigated by dripping 20 μL of plasma, chemiluminescent peroxidase substrate and PBST on three pieces of filter paper and measuring the spreading diameters after 30 seconds. The process of the spreading is recorded in Video S2.† It is obvious that PBST spread much faster. It could be attributed to the presence of the surfactant, i.e. Tween-20. The surfactant reduced the surface tension, diminishing the attraction of liquid molecules to each other, i.e. cohesion. As a result, when the cohesive force was weaker than the adhesive force, the capillary suction of the filter paper became significant and thus the filter paper imbibed PBST quickly.

Furthermore, the volumetric flow rate of the plasma and reagents was evaluated by measuring the time duration for the imbibition of 20 μL of fluid in the microchannels. 20 μL of plasma, chemiluminescent peroxidase substrate and PBST were added into the microchannels. The time duration was counted from the moment when the flow reached the filter paper to the moment when they completely drained away in the microchannels, as depicted in Fig. 5a. The measurement was repeated four times and plasma samples from six patients were used.

Fig. 5 (a) Time duration for blotting up 20 μL of plasma, chemiluminescent peroxidase substrate and PBST from the moment when the flow reached the filter paper to the moment when they completely drained away in the microchannels. (b) The calculated flux along the flow direction in a single teardrop-shaped unit. The shadow area represents the antibody-coating area.

The volumetric flow rate is proportional to the driving force of the flow and inversely proportional to the viscosity of the liquid. Since the plasmas were slightly viscous, the volumetric flow rate was relatively low, about 1.8 μL min⁻¹. The volumetric flow rate of the chemiluminescent peroxidase substrate was 2.3 μL min⁻¹. Low-viscosity PBST, which underwent a strong capillary suction of the filter paper, had a high volumetric flow rate of 3.8 μL min⁻¹.

Each reaction cavity was in a teardrop shape, whose non-uniform cross section leads to a variation of flux along the flow direction, i.e. J(z) = Q_FS⁻¹(z), where the subscript F stands for various liquids, i.e. plasma, chemiluminescent peroxidase substrate and PBST, and S(z) is the cross-sectional area of the microchannel. The calculated flux for plasma, chemiluminescent peroxidase substrate and PBST is shown in Fig. 5b.

The flux is low when the flow passes through a wide end where reaction happens, while the flux around the tapered end after the reaction area increases. It has been reported that the association rate of flowing antigen/antibody with immobilized antibody/antigen could increase with the decrease of the volumetric flow rate.^50,51 In these previous studies, the microchannel was straight and had a uniform cross section along the flow direction. Thus, the flow had a constant flux through the entire microchannel. Herein, it is worth clarifying that it is flux that affects antigen–antibody interaction. Precisely speaking, low flux is beneficial for association of flowing antigen/antibody with immobilized antibody/antigen. Therefore, in the teardrop-shaped unit, the association between antigen and antibody would be high at the wide end where the flux drops to the minimum.

In addition, if the width of the reaction area (the wide end) is fixed, the tapered end, contributing to a relatively high hydraulic resistance, can be used to control the volumetric flow rate over the entire microchannel. Last but not least, the tapered end takes only a small portion of the reaction cavity, reducing to some extent the size of the units and avoiding the waste of sample/reagents on the non-reaction areas. The design of the teardrop-shaped unit ensures a proper condition for the reaction and an efficient way for the entire assay.

Validation of the assay for multiple-biomarker detection

First, standards of gradient concentrations from 1 ng mL⁻¹ to 1 μg mL⁻¹ (CEA and AFP antigens) and 2 U mL⁻¹ to 1250 U mL⁻¹ (CA125 and CA19-9 antigens) were tested. The intensity of the reaction products was read out. The raw data and the calculated values for the 5 samples are listed in Table S1.† Standard curves for the four biomarkers were established by fitting the data to a four-parameter logistic (4PL) equation, as shown in Fig. 6a–d. The data used for establishing standard curves were obtained from three parallel experiments.

Fig. 6 (a)–(d) Established standard curves, (e)–(h) linear regression of the concentration measured by the microfluidic platform and the commercial testing and (i)–(l) B&A plots of the difference between the microfluidic platform and the commercial testing. (a), (e) and (i) Assay for CEA biomarker. (b), (f) and (j) Assay for AFP. (c), (g) and (k) Assay for CA125. (d), (h) and (l) Assay for CA19-9.

Then, as a proof-of-principle test for clinical applications, 30 plasma samples from CRC patients were tested using the pump-free microfluidic chip and the SAP. Fig. S6† shows the images of the reaction products captured by the SAP. The units for albumin present identical brightness, while the blank units have a dark background. The coating areas in the units for CEA, AFP, CA125 and CA19-9 emit light with different brightness. The intensity was recorded and the concentration of the antigens in the plasma samples could be derived from the intensity based on the inverse functions of the corresponding 4PL equations.

The testing results measured by our microfluidic platform was compared with those obtained in the clinical testing, in which electrochemiluminescence immunoassays (ECLIAs) based on commercial testing kits (Elecsys CEA, Elecsys AFP, Elecsys CA125II and Elecsys CA19-9, Roche Diagnostics GmbH, Germany) were adopted and measured using a commercial immunoassay analyzer (Cobas e801, Roche Diagnostics GmbH, Germany), as shown in Fig. 6e–h. The coefficient of determination (R²) for the curve fitting is higher than 0.9990, implying a highly linear correlation between the SAP and the commercial immunoassay analyzer. The results for 5 out of 30 samples are listed in Table S1.† The difference percentage, defined as the ratio between the difference and the average of the concentration obtained in our measurement and the clinical testing,⁵² was calculated based on the results for all the 30 samples. The low difference percentage (<6.6%) confirmed high consistency between the results obtained in our scheme and the clinical testing.

Furthermore, Bland–Altman (B&A) analysis was conducted. All difference data points, except for the case of AFP biomarker testing (96.67%), are within the limits of agreement, i.e. δ ± 1.96s, where δ and s are the mean and the standard deviation of the differences, respectively, as shown in Fig. 6i–l. The bias between the SAP and the commercial immunoassay analyzer is less than 2.23%. The B&A analysis indicates that the CLIAs for CEA, AFP, CA125 and CA19-9 multiple biomarkers conducted in our microfluidic platform have a perfect agreement with the ECLIAs in the clinical testing.

Reproducibility and sensitivity of the assay

The reproducibility of the assay was evaluated by testing three samples repeatedly. Each sample was tested in the three microfluidic chips three times. The standard deviation was no more than 0.034, 0.025, 0.034, and 0.031 for CEA, AFP, CA125 and CA19-9, respectively (Table S2†), indicating that the reproducibility of the assay could be guaranteed.

The limit of detection of the microfluidic platform for the assays of CEA, AFP, CA125 and CA19-9 is 0.89 ng mL⁻¹, 1.72 ng mL⁻¹, 3.62 U mL⁻¹ and 1.05 U mL⁻¹, respectively. Additional details about the calculation of the 4PL equation and limit of detection are provided in the ESI.†

Influence of the coating order on the assay

In the pump-free microfluidic chip, a multiplex assay for different biomarkers is conducted simultaneously in the connected units. The flow is unidirectional from the inlet to the outlet via the three reaction areas in series. Since the capture antibodies immobilized on the chip are all monoclonal antibodies, they are solely specific to the antigens of interest. To verify the irrelevance of the sequential order to the assay, the coating positions for immobilizing capture antibodies were swapped. Differentiated from the coating sequence as illustrated in Fig. 3, two microfluidic chips were specially prepared with capture antibodies for AFP and CA19-9 biomarkers immobilized in the units next to the inlets and capture antibodies for CEA and CA125 biomarkers immobilized in the middle units. The coating position for HSA was unchanged and the amount of each capture antibody immobilized on the chip remained the same. Two plasma samples were tested. The procedure of the entire assay followed the same protocol. Fig. S7† shows the captured images of the reaction products for the two samples tested in the microfluidic chips using different coating strategies and Table S3† lists the concentration measured in the assays. The almost identical results obtained from the microfluidic chips with different coating sequence indicate that the antigen–antibody reaction is specific and not affected by the coexistence of other irrelevant antibody molecules in the reagents. Therefore, the simultaneous multiplex assay of cancer biomarkers in the connected units is feasible and the sequential order has no influence on the assay.

Conclusions

We have developed a flux-adaptable and self-contained microfluidic platform including a pump-free microfluidic chip on which CLIAs for multiple biomarkers can be conducted simultaneously and a SAP in which the chemiluminescence emission of reaction products can be detected. In the immunoassay, the flow of the samples and reagents can be triggered by gravitational force and capillary force and continued by the suction of the filter paper. The design of the teardrop-shaped units contributes to retardant flow with low flux through the reaction area and a time-saving process for the entire assay. We have demonstrated the CLIAs of the four representative CRC biomarkers by using the microfluidic platform. Only 20 μL plasma per clinical sample is used. The entire assay is easy to operate and cost-effective. The assay can be simply accomplished within 20 minutes. The total assay cost per sample testing for the quantitative multiplex detection of the four cancer biomarkers is US$ 8.84 (Table S4†). The outcomes of the testing for 30 patient samples have a good agreement with those obtained in the clinical testing that was based on commercial testing kits and instruments. The platform can be further developed to include enhanced features. Clocked control schemes for reagent loading and image acquisition are desired in order to realize an automated assay. Besides, in an improved version, a dense arrangement of the reaction units and a large field of view of the imaging system are necessary to achieve multiple-sample assay. In addition, the testing cost could be further reduced by bulk purchasing in the mass production. The high-efficiency microfluidic platform has proven to be a promising candidate in the field of serological immunoassays.

Author contributions

Conceptualization: B. D., L. Z., Y. F., D. Z.; project administration: B. D., F. L., Y. F., D. Z., S. Z.; methodology: B. D., J. W., L. Z., Y. F. ; data curation: B. D., C. Y., W. L., X. H., Y. F.; formal analysis: B. D., C. Y., W. L., L. Z., Y. F.; investigation: B. D., J. W., L. Z., Y. F.; visualization: C. Y., W. L.; validation: B. D., J. W., L. Z., Y. F.; writing – original draft: B. D., Y. F.; writing – review and editing: B. D., J. W., L. Z., F. L., Y. F., D. Z., S. Z.; funding acquisition: B. D., Y. F., D. Z.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The work is financially funded by the National Key Research and Development Program of China (2016YFD0500604, 2016YFD0500603), the National Natural Science Foundation of China (61775140), the Shanghai Science and Technology Commission (18142200800) and the Shanghai Rising-Star Program (20QA1407000).

References

1. M. Arnold, M. S. Sierra, M. Laversanne, I. Soerjomataram, A. Jemal and F. Bray, Global patterns and trends in colorectal cancer incidence and mortality, Gut, 2017, 66, 683–691.
2. L. A. Torre, F. Bray, R. L. Siegel, J. Ferlay, J. Lortettieulent and A. Jemal, Global cancer statistics, 2012, Ca-Cancer J. Clin., 2015, 65, 87–108.
3. F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre and A. Jemal, Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries, Ca-Cancer J. Clin., 2018, 68, 394–424.
4. R. L. Siegel, K. D. Miller, A. G. Sauer, S. A. Fedewa, L. Butterly, J. C. Anderson, A. Cercek, R. A. Smith and A. Jemal, Colorectal cancer statistics, 2020, Ca-Cancer J. Clin., 2020, 70, 145–164.
5. R. Feng, Y. Zong, S. Cao and R. Xu, Current cancer situation in China: good or bad news from the 2018 Global Cancer Statistics?, Cancer Commun., 2019, 39, 22.
6. N. Hawkes, Cancer survival data emphasise importance of early diagnosis, Br. Med. J., 2019, 364, 1408.
7. A. Mahasneh, F. Al-shaheri and E. Jamal, Molecular biomarkers for an early diagnosis, effective treatment and prognosis of colorectal cancer: current updates, Exp. Mol. Pathol., 2017, 102, 475–483.
8. M. Thomsen, E. Skovlund, H. Sorbye, N. Bolstad, K. J. Nustad, B. Glimelius, P. Pfeiffer, E. H. Kure, J. S. Johansen and K. M. Tveit, Prognostic role of carcinoembryonic antigen and carbohydrate antigen 19-9 in metastatic colorectal cancer: a BRAF-mutant subset with high CA 19-9 level and poor outcome, Br. J. Cancer, 2018, 118, 1609–1616.
9. D. S. Thomas, E. Fourkala, S. Apostolidou, R. Gunu, A. Ryan, I. Jacobs, U. Menon, W. Alderton, A. Gentry-Maharaj and J. F. Timms, Evaluation of serum CEA, CYFRA21-1 and CA125 for the early detection of colorectal cancer using longitudinal preclinical samples, Br. J. Cancer, 2015, 113, 268–274.
10. Y. Gao, J. Wang, Y. Zhou, S. Sheng, S. Y. Qian and X. Huo, Evaluation of serum CEA, CA19-9, CA72-4, CA125 and Ferritin as diagnostic markers and factors of clinical parameters for colorectal cancer, Sci. Rep., 2018, 8, 2732.
11. G. Y. Locker, S. Hamilton, J. Harris, J. M. Jessup, N. Kemeny, J. S. Macdonald, M. R. Somerfield, D. F. Hayes and R. C. Bast Jr, ASCO 2006 update of recommendations for the use of tumor markers in gastrointestinal cancer, J. Clin. Oncol., 2006, 24, 5313–5327.
12. T. Tanaka, M. Tanaka, T. Tanaka and R. Ishigamori, Biomarkers for colorectal cancer, Int. J. Mol. Sci., 2010, 11, 3209–3225.
13. American Cancer Society, Tests to Diagnose and Stage Colorectal Cancer, https://www.cancer.org/cancer/colon-rectal-cancer/detection-diagnosis-staging/how-diagnosed.html.
14. C. V. Poznak, M. R. Somerfield, R. C. Bast, M. Cristofanilli, M. P. Goetz, A. M. Gonzalez-Angulo, D. G. Hicks, E. G. Hill, M. C. Liu, W. Lucas, I. A. Mayer, R. G. Mennel, W. F. Symmans, D. F. Hayes and L. N. Harris, Use of biomarkers to guide decisions on systemic therapy for women with metastatic breast cancer: american society of clinical oncology clinical practice guideline, J. Clin. Oncol., 2015, 33, 2695–2704.
15. C.-S. Ang, J. Rothacker, H. Patsiouras, P. Gibbs, A. W. Burgess and E. C. Nice, Use of multiple reaction monitoring for multiplex analysis of colorectal cancer-associated proteins in human feces, Electrophoresis, 2011, 32, 1926–1938.
16. S. Mahboob, S. B. Ahn, H. R. Cheruku, D. Cantor, E. Rennel, S. Fredriksson, G. Edfeldt, E. J. Breen, A. Khan, A. Mohamedali, M. G. Muktadir, S. Ranganathan, S.-H. Tan, E. Nice and M. S. Baker, A novel multiplexed immunoassay identifies CEA, IL-8 and prolactin as prospective markers for Dukes' stages A-D colorectal cancers, Clin. Proteomics, 2015, 12, 10.
17. X. Li and X. Liu, A microfluidic paper-based origami nanobiosensor for label-free, ultrasensitive immunoassays, Adv. Healthcare Mater., 2016, 5, 1326–1335.
18. C. D. Chin, T. Laksanasopin, Y. K. Cheung, D. Steinmiller, V. Linder, H. Parsa, J. Wang, H. Moore, R. Rouse and G. Umviligihozo, Microfluidics-based diagnostics of infectious diseases in the developing world, Nat. Med., 2011, 17, 1015–1019.
19. B. Dai, S. Chen, W. Li, L. Zheng, X. Han, Y. Fu, J. Wu, F. Lin, D. Zhang and S. Zhuang, Fully-functional semi-automated microfluidic immunoassay platform for quantitation of multiple samples, Sens. Actuators, B, 2019, 300, 127017.
20. H. Fu, P. Song, Q. Wu, C. Zhao, P. Pan, X. Li, N. Y. K. Lijessen and X. Liu, A paper-based microfluidic platform with shape-memory-polymer-actuated fluid valves for automated multi-step immunoassays, Microsyst. Nanoeng., 2019, 5, 50.
21. P. Jin, C. Jiao, Z. Guo, Y. He, S. Zhu, H. Tian and W. Zhu, Rational design of a turn-on fluorescent sensor for α-ketoglutaric acid in a microfluidic chip, Chem. Sci., 2014, 5, 4012–4016.
22. M. Zhao, X. Li, Y. Zhang, Y. Wang, B. Wang, L. Zheng, D. Zhang and S. Zhuang, Rapid quantitative detection of chloramphenicol in milk by microfluidic immunoassay, Food Chem., 2021, 339, 127857.
23. S. T. Sanjay, G. Fu, M. Dou, F. Xu, R. Liu, H. Qi and X. Li, Biomarker detection for disease diagnosis using cost-effective microfluidic platforms, Analyst, 2015, 140, 7062–7081.
24. Y. Wang, J. Luo, J. Liu, S. Sun, Y. Xiong, Y. Ma, S. Yan, Y. Yang, H. Yin and X. Cai, Label-free microfluidic paper-based electrochemical aptasensor for ultrasensitive and simultaneous multiplexed detection of cancer biomarkers, Biosens. Bioelectron., 2019, 136, 84–90.
25. Y. Fu, W. Li, B. Dai, L. Zheng, Z. Zhang, D. Qi, X. Cheng, D. Zhang and S. Zhuang, Diagnosis of mixed infections with swine viruses using an integrated microfluidic platform, Sens. Actuators, B, 2020, 312, 128005.
26. X. Tan, Q. Chen, H. Zhu, S. Zhu, Y. Gong, X. Wu, Y. Chen, X. Li, M. W.-H. Li, W. Liu and X. Fan, Fast and reproducible ELISA laser platform for ultrasensitive protein quantification, ACS Sens., 2020, 5, 110–117.
27. X. Zhao and T. Dong, Design and fabrication of low-cost 1536-chamber microfluidic microarrays for mood-disorders-related serological studies, Sensors, 2013, 13, 14570–14582.
28. C. K. Tang, A. Vaze and J. J. F. Rusling, Automated 3D-printed unibody immunoarray for chemiluminescence detection of cancer biomarker proteins, Lab Chip, 2017, 17, 484–489.
29. B. Hu, J. Li, L. Mou, Y. Liu, J. Deng, W. Qian, J. Sun, R. Cha and X. Jiang, An automated and portable microfluidic chemiluminescence immunoassay for quantitative detection of biomarkers, Lab Chip, 2017, 17, 2225–2234.
30. B. Cho, S. H. Lee, J. H. Song, S. Bhattacharjee, J. Feng, S. G. Hong, M. S. Song, W. Kim, J. Lee, D. Bang, B. W. Wang, L. W. Riley and L. P. Lee, Nanophotonic cell lysis and polymerase chain reaction with gravity-driven cell enrichment for rapid detection of pathogens, ACS Nano, 2019, 13, 13866–13874.
31. E. Berthier, A. M. Dostie, U. N. Lee, J. Berthier and A. B. Theberge, Open microfluidic capillary systems, Anal. Chem., 2019, 91, 8739–8750.
32. A. Tourovskaia, X. Figueroa-Masot and A. Folch, Differentiation-on-a-chip: a microfluidic platform for long-term cell culture studies, Lab Chip, 2005, 5, 14–19.
33. T. Kim and Y. Cho, A pumpless cell culture chip with the constant medium perfusion-rate maintained by balanced droplet dispensing, Lab Chip, 2011, 11, 1825–1830.
34. M. Marimuthu and S. Kim, Pumpless steady-flow microfluidic chip for cell culture, Anal. Biochem., 2013, 437, 161–163.
35. J. Wu, D. Tomsa, M. Zhang, P. Komenda, N. Tangri, C. Rigatto and F. Lin, A passive mixing microfluidic urinary albumin chip for chronic kidney disease assessment, ACS Sens., 2018, 3, 2191–2197.
36. A. Olanrewaju, M. Beaugrand, M. Yafia and D. Juncker, Capillary microfluidics in microchannels: from microfluidic networks to capillaric circuits, Lab Chip, 2018, 18, 2323–2347.
37. A. O. Olanrewaju, A. Ng, P. DeCorwin-Martin, A. Robillard and D. Juncker, Microfluidic capillaric circuit for rapid and facile bacteria detection, Anal. Chem., 2017, 89, 6846–6853.
38. P. Novo, F. Volpetti, V. Chu and J. P. Conde, Control of sequential fluid delivery in a fully autonomous capillary microfluidic device, Lab Chip, 2013, 13, 641–645.
39. G. A. Posthuma-Trumpie, J. Korf and A. V. Amerongen, Lateral flow (immuno)assay: its strengths, weaknesses, opportunities and threats. a literature survey, Anal. Bioanal. Chem., 2009, 393, 569–582.
40. L. Gervais and E. Delamarche, Toward one-step point-of-care immunodiagnostics using capillary-driven microfluidics and PDMS substrates, Lab Chip, 2009, 9, 3330–3337.
41. J. Wang, H. Ahmad, C. Ma, Q. Shi, O. Vermesh, U. Vermesh and J. R. Heath, A self-powered, one-step chip for rapid, quantitative and multiplexed detection ofproteins from pinpricks of whole blood, Lab Chip, 2010, 10, 3157–3162.
42. R. Gao, Z. Lv, Y. Mao, L. Yu, X. Bi, S. Xu, J. Cui and Y. Wu, SERS-based pump-free microfluidic chip for highly sensitive immunoassay of prostate-specific antigen biomarkers, ACS Sens., 2019, 4, 938–943.
43. K. Kobayashi and H. Onoe, Parallel droplet microfluidics for high throughput cell encapsulation and synthetic microgel generation, Microsyst. Nanoeng., 2018, 4, 17076.
44. A. Hashmi, G. Heiman, G. Yu, M. Lewis, H. J. Kwon and J. Xu, Oscillating bubbles in teardrop cavities for microflow control, Microfluid. Nanofluid., 2013, 14, 591–596.
45. N. Fatsis-Kavalopoulos, P. O'Callaghan, B. Xie, R. H. Vera, O. Idevall-Hagren and J. Kreuger, Formation of precisely composed cancer cell clusters using a cell assembly generator (CAGE) for studying paracrine signaling at single-cell resolution, Lab Chip, 2019, 19, 1071–1081.
46. Y. Hong, C. Feng, Y. Yu, J. Liu, J. W. Y. Lam, K. Luo and B. Z. J. Tang, Quantitation, visualization, and monitoring of conformational transitions of human serum albumin by a tetraphenylethene derivative with aggregation-induced emission characteristics, Anal. Chem., 2010, 82, 7035–7043.
47. E. Delamarche, A. Bernard, H. Schmid, A. Bietsch, B. Michel and H. Biebuyck, Microfluidic networks for chemical patterning of substrates: design and application to bioassays, J. Am. Chem. Soc., 1998, 120, 500–508.
48. H. Gu, M. H. Duits and F. Mugele, Interfacial tension measurements with microfluidic tapered channels, Colloids Surf., A, 2011, 389, 38–42.
49. E. Hrnčíř and J. Rosina, Surface tension of blood, Physiol. Res., 1997, 46, 319–321.
50. J. Lin, J. Andrade and I. N. Chang, The influence of adsorption of native and modified antibodies on their activity, J. Immunol. Methods, 1989, 125, 67–77.
51. G. A. Wemhoff, S. Y. Rabbany, A. W. Kusterbeck, R. A. Ogert, R. Bredehorst and F. S. Ligler, Direct visualisation and quantification of cellular cytotoxicity using two colour flourescence, J. Immunol. Methods, 1992, 156, 223–230.
52. L. J. Wang, R. Sun, T. Vasile, Y. C. Chang and L. Li, High-throughput optical sensing immunoassays on smartphone, Anal. Chem., 2016, 88, 8302–8308.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0lc00944j

‡ These authors contributed equally to this work.

---