Biosensors and Bioelectronics 49 (2013) 478–484

Contents lists available at SciVerse ScienceDirect

Biosensors and Bioelectronics

0956-56http://d

n CorrCairo, Afax: +20

E-mhazzazy

journal homepage: www.elsevier.com/locate/bios

Power-free chip enzyme immunoassay for detection of prostatespecific antigen (PSA) in serum

Heba Adel Ahmed a, Hassan M.E. Azzazy a,b,n

a Yousef Jameel Science and Technology Research Center (YJ-STRC), The American University in Cairo, AUC Ave., PO Box 74, New Cairo 11835, Egyptb Department of Chemistry, The American University in Cairo, AUC Ave, PO Box 74, New Cairo 11835, Egypt

a r t i c l e i n f o

Article history:Received 3 April 2013Received in revised form21 May 2013Accepted 31 May 2013Available online 11 June 2013

Keywords:Power-freeLab-on-chipEnzyme immunassay (EIA)Magnetic nanoparticlesProstate specific antigen (PSA)Cellphone imaging

63/$ - see front matter & 2013 Elsevier B.V. Ax.doi.org/10.1016/j.bios.2013.05.058

esponding author at: Department of ChemistrUC Ave., PO Box 74, New Cairo 11835,227957565.ail addresses: heba_adel@aucegypt.edu (H. Ad@aucegypt.edu (H.M.E. Azzazy).

a b s t r a c t

A power-free, portable “Chip EIA” was designed to render the popular Enzyme Linked ImmunosorbentAssay (ELISA) more suitable for point-of-care testing. A number of microfluidic platforms have enabledminiaturization of the conventional microtitre plate ELISA, however, they require external pumpingsystems, valves, and electric power supply. The Chip EIA platform has eliminated the need for pumps andvalves through utilizing a simple permanent magnet and magnetic nanoparticles. The magneticnanoparticles act as solid support to capture the target and are then moved through chambers harboringdifferent reagents necessary to perform a sandwich ELISA. The use of magnetic nanoparticles increasesthe volume-to-surface ratio reducing the assay time to 30 min. Changing the color of horseradishperoxidase (HRP) substrate to green indicates a positive result. In addition, a quantitative read-out wasobtained through the use of cellphone camera imaging and analyzing the images using Matlabs. Cellphones, including smart ones, are readily available almost everywhere. The Chip EIA device was used toassay total prostate specific antigen (tPSA) in 19 serum samples. The PSA Chip EIA was tested foraccuracy, precision, repeatability, and the results were correlated to the commercial Beckman Colter,Hybritech immunoassays for determination of tPSA in serum samples with a Pearson correlationcoefficient (R2¼0.96). The lower detection limit of the PSA Chip EIA was 3.2 ng/mL. The assay has 88.9%recovery and good reproducibility (% CV of 6.5). We conclude that the developed Chip EIA can be used fordetection of protein biomarkers in biological specimens.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Enzyme Linked Immunosorbent Assay (ELISA) is one of themost popular techniques that has been used at a large scale inclinical and other laboratories for detection of many antigens,antibodies, and haptens (Schuurs and Van Weemen, 1980). Theconventional 96 well plate ELISA is a sensitive quantitative detec-tion technique that gives out either colorimetric or fluorescentsignals (Windmiller et al., 2010). However, it is labor intensive as itinvolves several mixing, blocking, incubation, and washing stepsand requires skilled personnel (He et al., 2009). Furthermore, theconcentration of analyte requires the use of sophisticated instru-ments such as spectrophotometers or fluorometers. The draw-backs of ELISA represent a challenge toward its miniaturization aspoint-of-care (POC) testing devices (Moon et al., 2009).

ll rights reserved.

y, The American University inEgypt. Tel.: +20 226152559;

el Ahmed),

Current technological advances have allowed the miniaturiza-tion of conventional microtitre plate ELISA to a microfluidic ELISAplatform to overcome these drawbacks by capturing the analyte onthe surface of a microchannel or using nanoparticles entrappedinside the microchannels that increase the surface-to-volume ratioresulting in a decrease in assay time (He et al., 2009; Lin et al.,2010). The advantages of microfluidic devices are portability, smallvolume of samples and reagents, low chances of contamination,low cost, low power consumption, enhanced sensitivity, andreliability. These advantages have made microfluidics attractiveplatforms for POC testing. However, there are some limitationsconcerning their usage including bubble formation and deadvolume. Some microfluidic devices have reported the use ofpumps and valves to control the fluids flow into the microfluidicchambers, however, this complicates the design making it difficultto manufacture, requires the use of an external power source,pumping system, and increases the final cost of device (He et al.,2009; Okada et al., 2011).

There are many commercial PSA detection assays such asAxSYM PSA assay and IMx PSA assay (Abbott Diagnostics), Hybri-tech Tandem-E assay (for total PSA) and Hybritech Tandem-R assay(for free PSA) (Beckman and Colter), Immulite 1000, 2000, and

Fig. 1. The PSA Chip EIA platform and a schematic representation of its assembly. a:A picture of the PSA Chip EIA platform. b: A schematic representation of theassembly of the PSA Chip EIA platform. The PSA Chip EIA measures 5�4 cm2 andhas 2 sample channels, 115 μL in volume, one for each sample and its respectivecontrol. It is composed of three layers of polymethyl-methacrylate (PMMA) ofdifferent thicknesses attached together using double-sided adhesive film (DSA).PSA Chip EIA is composed of a 3 mm PMMA layer-where the reaction chambers arecut–attached to a back layer of 1.5 mm thickness and a top layer (1.5 mm)containing “sample injection inlets”.

H. Adel Ahmed, H.M.E. Azzazy / Biosensors and Bioelectronics 49 (2013) 478–484 479

2500 PSA assays (Siemens and Diagnostic Products CooperationDPC), Elecsys free and total PSA assay (Roche Diagnostics), BayerACS:180 System for PSA testing (Bayer Diagnostic corporation),and Tosoh AIA-PACK PA 600 or 1200 (Tosoh Products), which areall immunoassays. Hybritech immunoassay, is a 2-site immunoen-zymatic sandwich assay that utilizes alkaline phosphatase mono-clonal antibody to capture the PSA and sandwich it withparamagnetic particles coated with another monoclonal antibody,and finally measuring the chemiluminescent signal after addingthe substrate (Lumi-Phos 530) (Laffin et al., 2001). However, thecommercially available assays are time consuming and costly.

Chikkaveeraiah et al. developed a microfluidic electrochemicalimmunoassay for detection of PSA in serum with a detection limitof 0.23 pg/mL. Superparamagnetic nanoparticles were conjugatedto horseradish peroxidase (HRP)-labeled antibodies for off-linecapture of the protein target, then injected into a microfluidic8 electrode array covered with 5 nm glutathione gold nanoparti-cles with immobilized antibodies for the protein target. Theamperometric signals are developed by flowing the antigen–magnetic particle complex on the electrode after injecting asolution of HRP substrate and H2O2 (Chikkaveeraiah et al., 2011).Amperometric biosensors based on the coupling of oxidaseenzymes and the final detection of H2O2 have the disadvantageof the high over potential needed for H2O2 oxidation. At this highpotential, many electroactive substances (i.e. ascorbic acid, uricacid etc.), that are usually present in patient samples, could also beoxidized to give interfering signals (Ricci et al., 2007).

Goluch et al. developed a bio-barcode biosensor on a chip fordetection of PSA with a detection limit of 500 mA. Magneticnanoparticles functionalized with PSA antibodies capture the PSAtarget in the separation area of the chip. Then gold nanoparticlesdecorated with polyclonal antibodies and barcode DNA are added.The barcode DNA is then released from the gold nanoparticles andtransported to the detection area of the chip, which containscapture DNA. Other gold nanoparticles with a complementaryDNA sequence to the barcode are then introduced into thedetection area for hybridization. The signal is then amplified usingsilver enhancement (Goluch et al., 2006). Although the biobarcodeassay is sensitive, this procedure is complicated and time con-suming (80 min).

Okochi et al. has reported the use of magnetic force for dropletmanipulation on a chip to perform cell lysis and reversetranscription-polymerase chain reaction (PCR). The chip usesmagnetic beads that carry cells in an aqueous drop, and then useto move, fuse, and coalesce two droplets one containing the lysisbuffer and the other contains the RT-PCR buffer in a chambercontaining an oil phase. This droplet-based device was used forsuccessful detection of WT1 gene, the prognostic factor for acuteleukemia, from a single cell (Okochi et al., 2010). In another effortto avoid using valves and pumps in chips, Berry et al., has reportedan enzyme immunoassay chip dubbed “Immiscible FiltrationsAssisted by Surface Tension (IFAST)” which utilizes paramagneticnanoparticles to capture the analyte and move it through cham-bers containing different reagents required for the sandwichimmunoassay. Different reagents were separated from each otherusing olive oil barriers, and the fluorescent signal was detectedusing a fluorescent microscope. The chip was used to detectrecombinant PSA in human plasma and conditioned medium fromPSA-secreting cells (Berry et al., 2012).

In this work, we have miniaturized the popular ELISA platformto a hand-held, low-cost, power-free, POC platform dubbed “ChipEIA”. The term “Chip EIA” refers to performing immunoassayswithout moving the reagents but rather by moving the solidsupport “magnetic nanoparticles”. The Chip EIA is made using softlithography technique from layers of polymethyl methacrylate(PMMA) of different thicknesses having 5 circular chambers

harboring different reagents required for the EIA and separatedby rectangular chambers containing oil. The Chip EIA platform iseasy to manipulate with a simple magnet that moves the magneticnanoparticles from one chamber to the other through the oil phaseeliminating the need for external power supply. A qualitativeresult can be detected visually after 30 min through detection ofthe green color of the horse radish peroxidase substrate, and canbe quantified using a cell phone camera and picture analysis usingMatlabs software. The Chip EIA platform's concept was tested onprostate specific antigen (PSA), the prostate cancer (PCa) biomar-ker, in serum samples. PSA biomarker together with Digital RectalExamination (DRE) aids clinicians to distinguish between patientswith Benign Prostatic Hyperplasia (BPH) and those with malignanttumors. When the level of total PSA is between 4–10 ng/mL, thepatient is more likely to have prostate cancer rather than BPH. ThePSA Chip EIA platform was able to detect PSA in serum sampleswithin the clinical cut-off. Furthermore, the assay was validatedfor sensitivity, precision, recovery, and accuracy against the com-mercial EIA Beckman Colter Hybritech assay. Additionally, differ-ent cell phone cameras were evaluated for their performance asquantitative read-out devices. Fig. 1 shows the PSA Chip EIAplatform and a schematic representation of its assembly.

2. Materials and methods

2.1. Materials

Polymethyl-methacrylate (PMMA) (Spiroplastics, Cairo, Egypt)and double-sided adhesive film (iTapstore, Scotch Plains, NJ) werecut using a LASER cutter (VersaLaser™, Scottsdale AZ).

The antibodies for the enzyme-immunoassay were purchasedfrom K.HyTest Ltd. (K.HyTest Ltd., Finland) (see Supplementarydata).

Fig. 2. Assay principle of the PSA Chip EIA platform and pictures of PSA Chip EIAplatform with negative and positive samples for total PSA (tPSA). a: A schematicrepresentation of the different reagent chambers of the PSA Chip EIA platform: asample capture chamber (A) containing MNP coated with capture antibody+sample (antigen)+ secondary biotin conjugated antibody. Separating chambers(O) containing mineral oil. Wash chambers (C & C') containing wash buffers. Adetector Chamber (D) containing Horseradish peroxidase (HRP) conjugated detec-tion antibody. A signal chamber (E) containing the HRP-substrate [ABTS (2,2'-Azinobis [3-ethylbenzothiazoline-6-sulfonic acid]-diammonium salt)]. The PSAChip EIA principal utilizes antibody-coated magnetic nanoparticles to capture thePSA antigen. The secondary biotin-labeled antibody then binds the magneticnanoparticles with the captured antigen to form the sandwich. The detectorstreptavidin horseradish peroxidase (HRP)-antibody binds through biotin–strepta-vidin covalent bond to the immune complex, and finally the colorimetric ABTSsubstrate is added to produce a green color through the HRP enzyme catalyticreaction. b: PSA Chip EIA devices showing results for samples that are negative andpositive for tPSA, respectively. Antibody-coated magnetic nanoparticles capture thePSA antigen in the sample if present. The addition of the horseradish peroxidasecolorimetric substrate (ABTS) produces a green color through the HRP enzymecatalytic reaction in the presence of the tPSA antigen; hence a green color in thesignal chamber is seen after 30 min.

H. Adel Ahmed, H.M.E. Azzazy / Biosensors and Bioelectronics 49 (2013) 478–484480

Pictures of the PSA Chip EIA were taken using an ApplesiPhone 4. A comparison between different phone cameras ofdifferent resolutions was done using iPhone 4Gs (Apples)5 megapixel camera, Sony Ericsson i790 3.2 megapixel camera,and Blackberrys Bold 9650 Smartphone 5 megapixel camera. Theimages were analyzed for red color intensity per pixel usingMatlabs software.

2.1.1. Patient SamplesNineteen PSA-positive serum samples were obtained from the

National Cancer Institute (NCI), Cairo University for and wereanalyzed for free (fPSA) and total PSA (tPSA) using Beckman Colter,Hybritech immunoassays. The study has been approved by theInstitutional Reviewing Board (IRB) committee of the AmericanUniversity in Cairo (AUC); case number: 117 on 6 April 2012.

2.2. Methods

2.2.1. Synthesis of magnetic nanoparticlesMagnetic nanoparticles (Fe2O3) were prepared by the hydro-

thermal co-precipitation of ferric and ferrous ions in basic mediaas described elsewhere (Kouassi et al., 2005) (see Supplementarydata).

2.2.2. Magnetic nanoparticles surface functionalization with aminogroups

Amino functionalization of the prepared magnetic nanoparticleswas done by amino propyl trimethoxysilane (APTMS) as describedelsewhere (Kouassi and Irudayaraj, 2006) (see Supplementary data).

2.2.3. Characterization of magnetic nanoparticlesThe prepared magnetic nanoparticles were characterized by

scanning electron microscopy (SEM) for size and particles dis-tribution determination. Moreover, Fourier transform infraredspectroscopy (FT-IR) was used to record the IR spectra of thesamples using the potassium bromide (KBr) pellet technique (seeSupplementary data).

2.2.4. Magnetic nanoparticles conjugation to anti-PSA antibodyThe amine coated magnetic nanoparticles (1 mg/mL) were

conjugated with 1.5 μg/mL of PSA primary antibody (seeSupplementary data).

2.2.5. PSA chip EIA design and fabricationA non-lithographic technique was used for the fabrication of

the PSA Chip EIA platform as described in earlier publications(Fig. 1b) (see Supplementary data).

The PSA Chip EIA has a total of 5 chambers of circular cross-sections containing the aqueous phase and 5 chambers withrectangular cross-section containing the oil phase (as shown inFig. 2a). The circular chambers design was found to be the bestdesign to avoid bubble formation and entrapment of air inside thechambers.

Chambers harboring the aqueous phases have a volume ofabout 115 mL with a radius of about 3 mm and height of 3 mm.Chambers harboring the oil phases have a volume of about 168 mLwith the dimensions of 14�4�3 mm3. Chambers with a circularcross-section contain the reagents for the assay and are separatedfrom one another by the chambers containing the oil phases.Chambers containing the aqueous phase have one inlet and thosecontaining the oil phase have one inlet and one outlet. This isrequired to remove air as the chamber was being filled up andhence avoid formation of bubbles. See Supplementary data andFig. 2b for details on contents of each chamber.

The assay essentially is based on Enzyme Immunoassay (EIA).The antigen was allowed to bind to the capture antibody on thesurface of the magnetic nanoparticles outside the chip. Then it wasadded to the sample chamber and allowed to form a sandwichwith the secondary biotin-labeled antibody. Magnetic nanoparti-cles were then moved by a magnet to the first wash chamber towash the beads and remove unbound antigen. The nanoparticleswere moved to the detector chamber where the antigen—alreadysandwiched between the capture antibody and the secondaryantibody—was allowed to bind to the detection antibody (con-jugated to HRP). After a wash, in the second wash chamber, toremove unbound secondary antibody, the beads were moved tothe signal chamber containing the ABTS substrate. The reactionwas allowed to proceed for sufficient time to allow the green colordevelopment. The color developed was imaged using a cell phonecamera and the image analyzed using Matlabs for RGB intensityper pixel. Fig. 2b shows an example of the PSA Chip EIA deviceresult for a negative sample and a positive one, respectively.

2.2.6. Manipulation of magnetic particles along the channel in thePSA chip EIA platform using a magnet

All reagents used in the PSA Chip EIA were aqueous reagentssuch as; the conjugation buffer, wash buffer, assay buffer, and the2,2'-Azinobis [3-ethylbenzothiazoline-6-sulfonic acid]-diammo-nium salt (ABTS) which could not pass through the oil phase. Onlythe magnetic nanoparticles carrying the immune complex crossedthe oil-filled chambers using the magnet (see Supplementary datafor details). Each chip was designed to run each sample in duplicate

Fig. 3. PSA Chip EIA for detection of serial dilutions of prostate specific antigen(PSA) spiked in serum samples. Serial dilutions of PSA recombinant protein (1.5–100 ng/mL) spiked in serum sample were run on the PSA Chip EIA device andimages were taken using cell phone camera. Average normalized red color intensityper pixel; 1-(R/255) values were plotted. The small caption shows the linear part ofthe graph for concentrations between 3.2–50 ng/mL of PSA. (For interpretation ofthe references to color in this figure legend, the reader is referred to the webversion of this article.)

Fig. 4. Accuracy correlation plot between samples concentration as determined byBeckman Colter Hybritech Assay vs. PSA Chip EIA device. Nineteen patient sampleswere tested for total prostate specific antigen (tPSA) concentration in serum usingBeckman Colter Hybritech assay and PSA Chip EIA device. Beckman ColterHybritech assay showed that 8 of the 19 samples had tPSA concentrations above4 ng/mL, while 11 samples were below the clinical cut-off. The PSA Chip EIA assayshowed that only 7 samples had tPSA concentrations above 4 ng/mL, and 12samples were below the clinical cut-off.

H. Adel Ahmed, H.M.E. Azzazy / Biosensors and Bioelectronics 49 (2013) 478–484 481

(Fig. 2b). The captured image was cropped and analyzed withMatlabs where R-values (intensity of red color in the image) wasdetermined.

2.2.7. Imaging apparatusAll imaging was done in a dark room using a cartoon box

having a stand made of PMMA covered with a white paper to actas a background. A white LED light source fixed underneath thestand was the sole source of light.

2.2.8. Quantitative image processingA cell phone (Apples iPhone 4) with a 5-megapixel camera

was used to image the developed color. The color intensity of red,green, and blue pixel values was measured using a customizedMatlabs (MathWorks, Natick, MA) code. With this code, red,green and blue pixel values of each channel were reported withinseconds as mean value7standard deviation. The red (R) pixelvalues were used for data analysis, since they demonstrated thewidest range of color intensity for the catalyzed ABTS substrate asmeasured using the cell phone (see Supplementary data).

2.2.9. Validation tests for PSA chip EIA assay2.2.9.1. Assay accuracy. The concentrations of total and free PSA(tPSA and fPSA) were determined using Hybritech PSAimmunoassays. A correlation between tPSA concentrationsobtained by Beckman Colter, Hybritech immunoassays andthose obtained using PSA Chip EIA was made.

2.2.9.2. Assay precision. The inter- and intra- assay precision wasassessed by assaying 5 standards on 5 different chips in5 consecutive days. Images were taken using iPhone 4s (AppleInc.) and analyzed using Matlabs for the intensity of red color (R-value). Normalized red color intensity per pixel (average 1-R/255)was calculated.

2.2.9.3. Recovery study. 3 serum samples (samples: 1, 2, and 3)were each spiked with 0, 3.2, or 50 ng/mL of recombinant PSA. Thespiked samples were assayed and average (1-R/255) values weremeasured. The percentage recovery was determined bysubstituting the obtained normalized red color intensity valuesin the equation obtained from the standard curve of recombinantPSA protein in assay buffer.

2.2.9.4. Cell phone camera evaluation. Signals recorded using cellphone cameras were compared using different cell phones. Greenfood dye was serially diluted (1:1250, 1:2500, 1:5000, 1:10,000)and 5 consecutive images were taken using 3 different cellphones: iPhone 4Gs (Apples) with a 5 megapixel camera, SonyEricsson i790 with a 3.2 megapixel camera, and Blackberrys Bold9650 Smartphone with 5 megapixel camera. In addition, signalreproducibility for the iPhone 4Gs (Apples) was assessed bytaking 10 images of blue food dye diluted (1:5000) everyday for5 consecutive days.

3. Results

The PSA Chip EIA assay was used to measure serial dilutions ofrecombinant PSA protein (1.5–100 ng/mL) spiked in female serumsamples in duplicates. Images were captured with cell phonecamera after 5 min of the development of the green color of theABTS substrate. The captured images were cropped and analyzedwith Matlabs and R-values (intensity of red color in the image)were determined. The normalized red color intensity per pixel (1-R/255) was calculated, and the average 1-R/255 was obtained for

each concentration (Fig. 3). A calibration curve was constructed byplotting the average normalized red values (average 1-R/255)against the respective PSA concentrations (Fig. 3). The plot showsan exponential curve within the concentration range of (1.5–100 ng/mL) of PSA. However, a linear curve was obtained for PSArange (3.2–50 ng/ml).

3.1. Assay accuracy

The correlation plot showed a linear correlation between the2 methods with a Pearson correlation coefficient R2 of 0.96 (Fig. 4).Beckman Colter Hybritech assay showed that 8 of the 19 sampleshad tPSA concentrations above 4 ng/mL, while 11 samples werebelow the clinical cut-off. The PSA Chip EIA assay showed that only

H. Adel Ahmed, H.M.E. Azzazy / Biosensors and Bioelectronics 49 (2013) 478–484482

7 samples had tPSA concentrations above 4 ng/mL, and 12 sampleswere below the clinical cut-off where sample number 13 wasshown to have tPSA concentration below 4 ng/mL using PSA ChipEIA contrary to the result obtained by Beckman Colter Hybritechassay (data not shown).

3.2. Assay precision

For assay precision, results of the PSA standards determinationtrials performed on the PSA Chip EIA on 5 consecutive days wereobtained to assess the reproducibility of this assay technique. The%CV of the 5 trials showed to be 6.5% for inter-assay precision.

3.3. Recovery study

To assess the interference of the sample matrix and over allassay efficiency, 3 serum samples were spiked with 0 ng/mL,3.2 ng/mL or 50 ng/mL, respectively. The amounts of PSA spikedin the 3 samples were determined using the PSA Chip EIAplatform. Nearly 88.9% of the PSA spiked into the serum sampleswere recovered with the PSA Chip EIA (table 1).

Fig. 5. Evaluation of cell phone camera. a: Plot for evaluation of different cellphonecameras. Green food dye was serially diluted (1:1250, 1:2500, 1:5000,1:10,000) and5 consecutive images were taken with 3 different cell phones: iPhone 4Gs(Apples) with a 5 megapixel camera, Sony Ericson i790 with a 3.2 megapixelcamera, and Blackberrys Bold 9650 Smartphone with 5 megapixel camera. Thenormalized red color intensity per pixel [average 1-(R/255)] was plotted againstserial dilutions of food dye. b: Evaluation of Apple iPhone 4 with a 5-megapixelcamera. 10 images of 1:5000 dilution of green food dye were taken and normalized

3.4. Cell phone camera evaluation

Several cell phone cameras were evaluated for measuring thesignal generated by the PSA Chip EIA. Green food dye was seriallydiluted (1:1250, 1:2500, 1:5000,1:10,000) and 5 consecutiveimages were taken with 3 different cell phones: iPhone 4Gs(Apples) with a 5 megapixel camera, Sony Ericsson i790 with a3.2 megapixel camera, and Blackberrys Bold 9650 Smartphonewith 5 megapixel camera. A linear relationship between serialdilutions of the food dye and the normalized red color intensityper pixel was obtained using all 3 cell phones (Fig. 5a). The iPhone4Gs and Blackberry phones with the 5-megapixel cameras wasshown to give higher normalized red color intensity per pixel(average 1-R/255) than the Sony Ericsson i790 with a 3.2 mega-pixel camera indicating better camera quality.

The performance of Apple iPhone 4s, used throughout thisproject was evaluated for inter and intra-assay variability that mayarise from the built in camera, or user error. 10 images of 1:5000dilution of the green food dye were taken over 5 consecutive daysand average 1-(R/255) were reported (Fig. 5b). The variation in theaverage 1-(R/255) was small from day to day (intra-assay error) withaverage %CV of 2.55 and also between the 10 images takenwithin thesame day (inter-assay error) with a %CV not exceeding 4.06%.

Table 1Recovery of 3 serum samples (unspiked, spiked with 3.2 or 50 ng/mL of recombi-nant PSA) tested by the PSA Chip EIA platform.

Serum Un-spikedSpiked with 50 ng/mL

Spiked with 3.2 ng/mL

S1 average redcolora

0.4(SD70.1)

0.7 (SD70.5) 0.5 (SD70.7)

S1 tPSA (ng/mL) 0.6 50.0 3.9S2 average redcolora

0.4(SD70.7)

0.7 (SD70.2) 0.5 (SD70.2)

S2 tPSA (ng/ml) 0.7 45.3 3.0S3 average redcolora

0.4(SD70.1)

0.7 (SD70.4) 0.5 (SD70.4)

S3 tPSA (ng/mL) 0.7 56.0 4.4Mean recovery (%) 88.4 85.4 86.5

a S1, S2 and S3 are samples (1, 2, and 3). Average red color is the averagenormalized red color intensity of duplicates (1-R/255).

red color intensity per pixel, average 1-(R/255), were reported. (For interpretationof the references to color in this figure legend, the reader is referred to the webversion of this article.)

4. Discussion

Several hand-held microfluidic lab-on-chip devices have beendesigned for POC testing which are based on immunoassay or PCRtechniques. Microfluidic devices offer the advantages of ease ofhandling, multiplexing, and disposability; furthermore, manydevices have an incorporated read-out system eliminating theneed for external instruments (Bhattacharyya and Klapperich,2007; Moon et al., 2009). In spite of these advantages, many ofthe microfluidic devices employ valves and pumps to control thefluid flow through the micro channels; however this complicatesthe design, increases the total cost, and makes it more difficult tomanufacture on mass scale, and often requires the use of anexternal power source.

H. Adel Ahmed, H.M.E. Azzazy / Biosensors and Bioelectronics 49 (2013) 478–484 483

Okochi et al. showed droplet manipulation through coalescenceof aqueous phases in the form of droplets for RNA extraction andperforming RT-PCR. The PSA Chip EIA is based on moving the solidphase (the magnetic nanoparticle carrying the immune complex)using a permanent magnet. The Okochi's chip is made of alumi-num with single channel containing oil with aqueous dropletscontaining the cells, lysis buffer, and RT-PCR mixture (Okochi et al.,2010). On the other hand, the Chip EIA has 5 aqueous phasechambers (containing the primary antibody, wash buffer, detectorantibody, wash buffer, and finally the enzyme substrate) separatedby 4 chambers containing mineral oil.

The developed PSA Chip EIA is based on the movement ofmagnetic nanoparticles in a similar setting to that of the IFASTchip, however, it offers several advantages. First, the PSA Chip EIAis made of PMMA, thermoplastic sheets, cut using a laser cutter,which is much more suitable for low cost mass production. On theother hand, the IFAST chip reported by Berry et al. is made of PDMSin multiple steps (Berry et al., 2012). Second, the PSA Chip EIA hasbeen used for detection of recombinant PSA spiked in buffer andhuman serum samples as well as clinical specimens obtained frompatients. The results obtained were concordant with thoseobtained using the commercial Hybritechs PSA immunoassay.Third, the signals generated by the PSA Chip EIA were detectedusing a cellphone-imaging technique, which is more suitable forpoint of care testing as it eliminates the need for spectrophot-ometers and fluorescent microscopes.

A PSA Chip EIA platform was developed as a miniaturizedpower-free point of care platform which does not employ valvesand pumps. The new chip is designed to allow miniaturization ofthe ELISA assay to make it more robust and available as POC.

The chip fabrication and design is relatively simple, and thePMMA material used is suitable for mass manufacturing with alow-cost and good durability for ease of transportation. We choseprostate specific antigen (PSA) the tumor biomarker for prostatecancer as a proof of concept for the PSA Chip EIA platform. Serialdilutions of the recombinant PSA antigens (1.5–100 ng/mL) inserum samples were tested on the PSA Chip EIA and pictureswere taken with cell phone camera.

There are many variables that could affect the performance ofthe Chip EIA including the concentrations of the 3 differentantibodies (capture, secondary, and detection antibody), the incu-bation time within each chamber, and the concentration ofmagnetic nanoparticles. The same concentrations of secondaryand detector antibodies that were used in the PSA ELISA forsuccessful detection of PSA were employed in the Chip EIA. Severalconcentrations of capture and magnetic nanoparticles were tested.It was shown that PSA capture Abo1.5 μg/mL would lead toundetectable color signal at the lower PSA concentration, whilehigher concentration of the PSA capture Ab (42.0 μg/mL) led to adark signal in samples containing high PSA concentrations whichwas difficult for the cellphone camera to quantify. Capture anti-body concentration of 1.5 μg/mL was found optimal for this assay.Further optimization of other test parameters will be consideredwhile developing the next generation of this test.

The analyte concentration and the time of incubation in thesignal chamber were among the determining factors which affectthe dynamic range of the assay. It was found that 5 min incubationtime (in the signal chamber) was optimal for the desired dynamicrange of the assay (1.5–100 ng/mL of recombinant PSA) where athigher target concentration, the cell phone camera was unable todistinguish the different intensities of green color developed after5 min. As in regular ELISA experiment, the green color of substratestarted to fade after 15 min.

The conjugation efficiency of the capture antibody to themagnetic nanoparticles (MNP) was assessed by measuring theabsorbance at 280 nm of the capture antibody before the addition

of magnetic nanoparticles and of the supernatant collected after 2successful washes of the magnetic nanoparticles after conjugationto the capture antibody. The conjugation efficiency was calculatedusing the following equation: Conjugation efficiency (%)¼[Absor-bance at 280 nm for the capture Ab—(Absorbance at 280 nm forwash 1+Absorbance at 280 nm for wash 2)/Absorbance at 280 nmof the capture Ab]�100%. The conjugation efficiency was found tobe 67%.

The PSA Chip EIA was shown to have a linear range from 3.2 to50 ng/mL, which involves the clinically significant values for PCa(4.0–10 ng/mL). In addition, several incubation times in differentchambers and different concentrations of the antibodies andsubstrate were evaluated. It was found that 7 min incubation witheach of the primary, secondary, and detector antibody and 5 minincubation with the substrate were optimal for detectable colorsignal in the linear range of 3.2–50 ng/mL.

Nineteen sero-positive samples for prostate cancer were ana-lyzed for total PSA (tPSA) concentration using Beckman Colter,Hybritech immunoassays and with the PSA Chip EIA. The PSAChip EIA showed comparable levels of tPSA as indicated by a linearcorrelation between the 2 techniques with a Pearson correlationcoefficient (R2 value) of 0.96 (Fig. 3). Eight of the nineteen sampleswere shown to have tPSA concentration above 4 ng/mL using theBeckman Colter Hybritech assay, while only 7 samples wereshown to have tPSA concentrations above 4 ng/mL. Sample num-ber 13 was shown to have a concentration of 4.9 ng/mL usingBeckman Colter Hybritech assay, and 3.5 ng/mL using the PSA ChipEIA. This might be due to degradation in the protein during storageand transportation or due to the fact that the PSA Chip EIAplatform has a percentage recovery of 88.9% that can be improvedin future studies. The reported lower limits of PSA detection usingstandard ELISA (GmbH, 2011) is 0.2 ng/mL and 0.008 ng/mL forHybritech PSA assay (Laffin et al., 2001). The lower limit ofdetection of the PSA Chip EIA was found to be 3.2 ng/mL

The precision results of the PSA Chip EIA were within theacceptable range of 6.5%. To further improve the performance ofthe assay and to eliminate errors that result from changing theangle of the camera from 1 experiment to the other which changesthe intensity of the incident light, we suggest adjusting theimaging set-up with a fixed slot for the cell phone to fix the angleof transmitted light and decrease the inter- and intra-experimentvariation.

The PSA Chip EIA showed good recovery results (88.9%) forserum samples spiked with recombinant PSA. A small residualamount of magnetic nanoparticles remains at the interfacebetween the aqueous and oil phases between each chamberleading which may be responsible for loss of tPSA. This might bedue to the small size of the magnetic nanoparticles used (50–100 nm). A balance has to be achieved where the magneticnanoparticles are large enough to increase the percentage recov-ery, yet not so large to decrease the surface area available for theconjugation, which may lead to an increase in the overall turn-around time of the assay. In addition, increasing the time ofexposure of the magnetic nanoparticles to the magnet andincreasing the magnet strength, would increase the amount ofmagnetic nanoparticles recovered in the last chamber, suggestingthat future automation of the platform and using a strongermagnet might improve the percentage recovery.

The magnet was placed immediately under the chip and movedmanually. The percent recovery of magnetic nanoparticles in thesignal chamber may be affected by many factors such as themagnetic field intensity, the size of the magnetic nanoparticlesused, the distance between the magnet and the magnetic particlesand the speed by which the magnet is moved. The magnet used inthe assay was a permanent alnico magnet; however using a rareearth metal magnet such as neodymium magnets, with stronger

H. Adel Ahmed, H.M.E. Azzazy / Biosensors and Bioelectronics 49 (2013) 478–484484

magnetic field, will increase the percent recovery of the magneticnanoparticles and consequently the color intensity in the detec-tion chamber. The magnet should be placed immediately underthe chip and moved slowly to allow it to attract and move themaximum number of magnetic particles.

To compare the use of different cellphone cameras withdifferent resolutions and imaging technologies, green food dyewas serially diluted and 5 consecutive images were taken with 3different cellphones. A linear relationship was obtained using all 3cellphones (Fig. 5a). The data showed that phones with higherresolution produce comparable results and give higher average redcolor intensity per pixel than the phones with lower resolutioncameras. The results showed that commercially available phonecameras with the available resolution can be used alternativelygiving consistent results within the same experiment providedthat the same phone will be used through-out the sameexperiment.

Evaluation of the cell phone used to quantify the signalindicated that the variations in the average 1-(R/255) were smallbetween different days (inter-experimental errors) and alsobetween the 10 images taken within the same day (intra-experi-mental errors) (Fig. 5b). These results suggest that using the samephone throughout the experiment is essential to generate repro-ducible results.

5. Conclusions

Developing an inexpensive and simple POC version of ELISAtechnique would be an invaluable addition to the clinical labora-tory. The PSA Chip EIA has a power-free design and is amenable forremote areas and limited-resources settings. It uses cell phonecamera to generate quantitative results. The estimated cost of thechip for detection of tPSA (including reagents for running 4-pointcalibration curve, controls, and the unknown sample and chipmanufacturing cost) is estimated at $6.00.

Some future improvements can be done to the developedprototype to improve the performance of the chip especially itssensitivity at low target concentration. This includes improvingthe quality of pictures taken by improving the current set-up forimage capturing such as the use of a stronger LED light and usingcell phones with high-resolution cameras. The PSA Chip EIAplatform is also amenable for automation and multiplexing.

The PSA Chip EIA platform shows a great promise for improvingthe current ELISA technique making it portable with its hand-heldsize, and more robust. It does not require a trained operator,external power source, or external specialized read-out

instruments, and shows a potential for automation. The PSA ChipEIA simple design makes it suitable for mass-production whileretaining its low cost. All those advantages render the PSA Chip EIAattractive for point of care testing in limited resources settings.

Acknowledgments

We would like to thank Mr. Kamel Eid (YJ-STRC, AUC) forpreparing the magnetic nanoparticles. This work was funded by agrant from YJ-STRC to Dr. Hassan Azzazy.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.bios.2013.05.058.

References

Berry, S.M., Maccoux, L.J., Beebe, D.J., 2012. Analytical Chemistry 84 (13),5518–5523.

Bhattacharyya, A., Klapperich, C., 2007. Biomedical Microdevices 9 (2), 245–251.Chikkaveeraiah, B.V., Mani, V., Patel, V., Gutkind, J.S., Rusling, J.F., 2011. Biosensors

and Bioelectronics 26 (11), 4477–4483.GmbH, D.I., 2011. Total PSA ELISA, EIA-3719. In: DRG Instruments GmbH, G. (Ed.),

Marburg.Goluch, E.D., Nam, J.-M., Georganopoulou, D.G., Chiesl, T.N., Shaikh, K.A., Ryu, K.S.,

Barron, A.E., Mirkin, C.A., Liu, C., 2006. Lab on a Chip 6 (10), 1293–1299.He, H., Yuan, Y., Wang, W., Chiou, N.R., Epstein, A.J., Lee, L.J., 2009. Biomicrofluidics 3

(2), 22401.Kouassi, G., Irudayaraj, J., McCarty, G., 2005. Biomagnetic Research and Technology

3 (1), 1.Kouassi, G.K., Irudayaraj, J., 2006. Analytical Chemistry 78 (10), 3234–3241.Laffin, R.J., Chan, D.W., Tanasijevic, M.J., Fischer, G.A., Markus, W., Miller, J.,

Matarrese, P., Sokoll, L.J., Bruzek, D.J., Eneman, J., Nelson, J., Bray, K.R., Huang,J., Loveland, K.G., 2001. Clinical Chemistry 47 (1), 129–132.

Lin, C.-C., Wang, J.-H., Wu, H.-W., Lee, G.-B., 2010. Journal of Laboratory Automation15 (3), 253–274.

Moon, S., Keles, H.O., Ozcan, A., Khademhosseini, A., HÊggstrom, E., Kuritzkes, D.,Demirci, U., 2009. Biosensors and Bioelectronics 24 (11), 3208–3214.

Okada, H., Hosokawa, K., Maeda, M., 2011. Analytical Sciences 27 (3) 237–237.Okochi, M., Tsuchiya, H., Kumazawa, F., Shikida, M., Honda, H., 2010. Journal of

Bioscience and Bioengineering 109 (2), 193–197.Ricci, F., Moscone, D., Palleschi, G., 2007. In: Alegret, S., Merkoçi, A. (Eds.),

Comprehensive Analytical Chemistry. Elsevier, New York, United States,pp. 559–584, Chapter 24.

Schuurs, A.H.W.M., Van Weemen, B.K., 1980. Journal of Immunoassay 1 (2),229–249.

Windmiller, J.R., Chinnapareddy, S., Santhosh, P., Hal �mek, J., Chuang, M.-C.,Bocharova, V., Tseng, T.-F., Chou, T.-Y., Katz, E., Wang, J., 2010. Biosensors andBioelectronics 26 (2), 886–889.