Nabatiyan, Arman PhD; Baumann, Michael A PhD; Parpia, Zaheer MS; Kelso, David PhD

INTRODUCTION

In 2007, 67% of all people living with HIV and 75% of the AIDS-related deaths were localized in sub-Saharan Africa. Of all children 15 years and younger, 90% of those living with HIV and 90% of the AIDS-related deaths were also localized in sub-Saharan Africa.¹ Like the prevalence of HIV infection, current practices in HIV diagnostic testing vary significantly depending on world geography. In high-income countries such as the United States and Western Europe, there is ready access to later generation HIV antibody or HIV antibody/antigen combination immunoassays in addition to nucleic acid test-based assays.^2,3 In low-income countries such as those in sub-Saharan Africa, access to later generation immunoassays and nucleic acid test assays is limited by: (a) their inherent cost; (b) the number of properly equipped clinical laboratories with trained personnel capable of performing the tests; and (c) the underdeveloped transportation infrastructure needed to move blood samples from distant collection sites to centralized laboratories.^4,5

For many resource-limited settings dealing with HIV, rapid self-performing tests for HIV antibodies are used because they address most of the limitations described above.^2-5 However, the public health benefit of even later generation HIV antibody assays is tempered by the window period between HIV infection and a detectable HIV antibody titer, which is on average around 22 days.⁶ During this time, individuals with acute infections (HIV antigen positive but HIV antibody negative) can significantly contribute to the spread of HIV among the population.⁷ Incorporating assays for the HIV p24 antigen can reduce the window period significantly to an average of around 17 days and help identify acutely infected individuals.⁶ Assays for HIV p24 are also useful for diagnosing HIV infection in children younger than 12 months as HIV antibody assays can yield false positives due to maternal inheritance of HIV antibodies during gestation. Although assays for proviral DNA or viral RNA are considered the reference assays for pediatric samples, their use in sub-Saharan Africa is limited by the same factors described above.^2,4,5 Hence, HIV p24 antigen assays can act as a viable alternative.

HIV p24 assays or HIV antigen/antibody combination assays are currently limited to enzyme-linked immunosorbent assay (ELISA)-based technology in resource-limited settings, which restricts their broad use in sub-Saharan Africa.^8-11 Furthermore, to achieve their published limits of detection for p24 antigen, these ELISAs require an amplification process known as enzyme-linked amplified sorbent test (ELAST) that adds several hours to the total assay time and requires meticulous handling.^12,13 Given the successful and widespread use of rapid self-performing HIV antibody assays in resource-limited settings, it is a logical next step to develop a rapid test for HIV p24. In this report, we describe the development of an initial prototype lateral flow assay that yields the same limit of detection for p24 antigen as the ELAST amplified ELISAs, but without the need for amplification steps, and has a total assay time of 30-40 minutes.

METHODS AND MATERIALS

Test Strip Configuration

A noncontact jetting system (XYZ series 3050; Biodot, Ervine, CA) was used to generate 3 lines on to 2.5 × 30-cm strip of backed nitrocellulose membrane (HF120; Millipore Corp, Billerica, MA). Neutravidin protein (Pierce, Rockford, IL), whole mouse IgG, and antimouse IgG (Biospacific, Emeryville, CA) was diluted to 2 mg/mL in 100 mM ammonium acetate and jetted at 0.6 μL/cm with 100 micron ceramic tips yielding 0.5-mm-thick lines. The mouse IgG line was positioned 15 mm from the bottom of the nitrocellulose membrane and the neutravidin test and antimouse IgG control lines were positioned at 19 mm and 23 mm, respectively. The nitrocellulose was transferred to a desicating chamber and dried at 37°C for 24 hours. Test strips were assembled using an adhesive polystyrene backing card 6 × 30 cm (#GL-49901; G&L Precision Die Cutting, San Jose, CA) over which was applied 2.5 cm of nitrocellulose membrane, 2 cm of sample pad (grade 8301; Ahlstrom, Holly Springs, PA), and 1.5 cm of blotter pad (grade 205; Ahlstrom). Strips were lightly pressed with a hand roller and cut into 3-mm strips.

Biotinylated Antibody and Latex Bead Conjugate Formation

One of the monoclonal antibodies targeting an epitope of the HIV p24 protein (115B151-423; Abbott Laboratories) was biotinylated. The antibody was diluted to 5 mg/mL in phosphate-buffered saline (PBS) and reacted with 85 μL of 10 mM Sulfo-NHS-LC-Biotin (Pierce) for 1 ml of antibody solution. Reactions were left at room temperature (RT) for 2 hours before being transferred to 4°C overnight. The following day purification of the antibody was performed over a PD10 column (PD 10; Sigma Chemical Co, St. Louis, MO), resulting in desalting into PBS. The final concentration of the antibody stock was 2.5 mg/mL with approximately 9-11 biotins per antibody as determined by the HABA dye (4'-hydroxyazobenzene-2-carboxylic acid) reaction (Pierce).

Another monoclonal antibody targeting a second epitope of the HIV p24 protein (108394-470; Abbott Laboratories) was used for coating latex particles as a detection conjugate. Two hundred fifty microliter of 500 nm beads (1% w/v) containing Nile Red dye (CFP-0556-2; Spherotech, Lake Forest, IL) was placed in a microfuge tube. The sample was sonicated with microtip probe sonicator for 3-5 seconds at 0-1 setting (Sonic Dismembrator Model 100; Fisher Scientific, Pittsburgh, PA). The beads were spun down at 1 K rpm for recollection. Fifty microliter of 108 antibody (5mg/ml) was added and the mixture incubated for 5 minutes after vortexing. Then 250 μL of 150 mM MES pH = 5.8 (Sigma Chemical Co) was added followed by vortexing and incubation for 10 minutes. Fifty microliter of 10 mM EDAC (Sigma Chemical Co) was added to the beads and vortexed. Reaction was carried out at RT for 2 hours, then transferred to 4°C overnight. For purification, beads are spun down at 8 k for 5 minutes. The first buffer change is with PBS/0.05% Tween-20. Three subsequent washes are performed in PBS/0.05% Tween-20/0.1% bovine serum albumin (BSA). Finally, the beads in are resuspended in 240 μL of PBS and stored at 4C.

Mix and Run Assay

Twenty-five microliter of plasma with or without p24 antigen is added to 10 μL of acid-shock buffer (200 mM citric acid/0.5% Triton X-100) in a 12 × 75-mm polypropylene tube and mixed. After 1 minute, the reaction is neutralized with 35 μL neutralization buffer (1M Tris HCl pH = 7.8 + 0.16% Triton X-100 + 0.03% sodium dodecyl sulfate) containing 5% BSA. After 1 minute of neutralization, 3 μL of biotinylated antibody is added (1:500 dilution of 2.5 mg/mL stock). Next, 3 μL of conjugates is added (1:10 dilution of 1% solids stock). The reaction is mixed for 1 minute and finally applied to a test strip.

Data Analysis

Dried test strips are mounted on a glass slide and analyzed using a standard epifluorescence microscope (Axiovert 135M; Zeiss, Thornwood, NY) fitted with a 4X objective. A 100 W mercury lamp was set at 20% power output, and the samples were analyzed at a fixed height with the following filters: excitation at 546 ± 12 nm (HQ546/12x; Chroma, Rockingham, VT) and emission at 607 ± 36 nm (# FF01-607/3625; Semrock Rochester, NY). Test lines were uniformly aligned using an optical reticle. Samples were exposed for 4 seconds. QCapture software (QImaging Corp, Surrey, British Columbia, Canada) exports average line intensities for the capture area containing the test line and regions upstream and downstream of it. Test line values were established through background subtraction. The background was assigned as the average value of the 350 most upstream lines, whereas the test line value was determined by averaging the 350 highest value lines within the test line region.

Clinical Sample Testing

Pathogen-free adult human citrate plasmas were obtained from Gulf Source Blood Bank (Houston, TX). Seroconversion panels PRB 930, 940, 941, 943, 944, 945, 947, 959, and 965 (SeraCare, Milford, MA) were thawed once and realiquot to avoid multiple freeze thaw cycles. For testing, an aliquot was thawed on ice before the sample volume was drawn and transferred to a tube containing acid-shock buffer. Virology Quality Assurance (VQA) HIV virus was obtained from Dr. James Bremer (contract #NO1-AI-50044) at the VQA Laboratory at Rush St. Luke's Medical Center, Chicago, IL. The VQA virus control standard (subtype B) was received as viral culture spiked in normal human plasma at a concentration of 1.5 million copies per milliliter determined in house by the VQA laboratories.¹⁴ HIV subtypes A (isolate UG273), B (isolate US2), C (isolate UG268), D (isolate UG270), F (isolate BZ163), and G (isolate HH8793) were kindly provided by Abbott Laboratories as viral cultures spiked in normal human plasma.¹⁵ VQA HIV and HIV subtypes were diluted into normal (HIV negative) plasma (Gulf Source Blood Bank; Houston, TX) before testing. Recombinant p24 protein was kindly provided by Dr. John Hackett at Abbott Laboratories at a stock concentration of 280 g/mL. The protein concentration was determined using an Abbott HIV-1 p24 microtiter enzyme immunoassay. Briefly, an Abbott HIV-1 p24 Primary Standard is used to establish a standard curve to calculate p24 concentration. The Abbott p24 Primary Standard is calibrated to the Agence Française de Sécurité Sanitaire des Produits de Santé (AFSSAPS) HIV-1 p24 standard.

RESULTS

We adapted a neutravidin/biotin sandwich capture format to a lateral flow device utilizing fluorescent particles as detection labels of the HIV core antigen p24. In this configuration, a 25μL plasma sample is reacted offline with biotinylated antibody against one epitope of p24 and a conjugate coated with a second antibody against a second epitope of p24. The sample is then applied to a sample pad and allowed to flow through the nitrocellulose membrane where capture against a neutravidin test line occurs.

Initial dose-response testing using recombinant p24 antigen demonstrated limits of detection at approximately 2pg/ml of p24 (Fig. 1, left panel and Fig. 1C). Significant background noise was noted around the test line with this configuration. To improve the signal to noise ratio, a sacrificial line containing whole mouse IgG was introduced upstream of the test line as a means of specific signal enrichment at the test line. To this effect, the sacrificial line improved signal to noise ratios nearly 2-fold at the test line for analyte concentrations below 20 pg/mL (Fig. 1, right panel). With this modification, repeated dose-response testing resulted in a limit of detection of 0.2 pg/mL (Fig. 1C). It should be noted that we had initially tried adding mouse IgG directly into the reaction mixture with no effect on signal to noise enhancement (data not shown). The use of mouse IgG directly on the strip membrane demonstrated binding of fluorescent particles at that line (Fig. 1B). Because of the observed intensity of the sacrificial line, we speculate that it is functioning by absorbing defective latex particles and/or those uncoated/incompletely coated with antibody, which are “sticky” and would otherwise nonspecifically bind to the test line in the absence of the sacrificial line.

Figure 1

In many clinical HIV samples, the availability of analyte is limited by low viral titers in addition to the formation of immune complexes after seroconversion. So we decided to introduce an acid shock of the plasma to facilitate liberation of p24 analyte for detection. A study performed within our laboratory demonstrated improved antigen dissociation from an antibody-antigen complex with increasing acidity (data not shown) and was consistent with published reports for the release of p24 from immune complexes.^16,17 Initially, acetic acid was used to lower the pH of human plasma, but because of the volatile nature of the acid and its ability to lyze red blood cells,¹⁸ we substituted citric acid because it was more compatible with whole blood and additionally is nonvolatile. Fixed volumes of citric acid at concentrations of 0.1-1 M were added to human plasma and pH readings taken (data not shown). Citric acid concentrations above 0.3 M caused visible aggregation of the plasma proteins and as such were excluded from further consideration (data not shown). Given that protein aggregates can limit or inhibit lateral flow performance, we decided on an upper limit of 0.2 M citric acid for pretreatment so as to avoid possible reaction interference.

Because our assay requires pH neutralization for binding reactions to occur optimally, we used a simple Tris buffer for restoration of the acid shocked plasma back to neutrality. To this buffer was also added detergent (Triton X-100 and sodium dodecyl sulfate) to minimize aggregation of the sample, lyze the HIV, and facilitate proper wetting and laminar flow through the nitrocellulose membrane.

Our pretreatment step was evaluated through direct comparison of a seroconversion panel consisting of 6 serial blood samples (members) with moderate viral copy numbers and antibody reaction (Fig. 2). A comparison was performed between no pretreatment, 0.1 M citric acid, and 0.2 M citric acid. Expectedly, the highest number of positive samples was detected with the higher 0.2 M citric acid treatment, and we chose this concentration for our acid shock.

Figure 2

Next we tested the efficacy of our overall assay including the pretreatment step by examining the dose response of live cultured virus. VQA control standard¹⁴ was diluted in healthy human plasma, subjected to acid shock, and processed accordingly (Fig. 3A). We found that at least 10 k virus copies per milliliter could be discriminated with our assay. We also wanted to examine the subtype antigenic reactivity of our assay and thus subjected the 5 most common subtypes (A, B, C, D, F, and G) to standard conditions. We diluted viral stocks in HIV-negative human plasma at a concentration of 140 k copies per milliliter and quantified the outputs. We found that at this concentration, all subtypes generated robust signals although subtypes C and F showed nearly 2-fold higher signals than the other 4 (Fig. 3B). Similar differences in subtype antigenic reactivity were observed with the Abbott Architect HIV Ag/Ab Combo assay,¹⁹ which uses the same antibody pair but under different reagent and reaction conditions. This suggests that our assay exhibits a lower limit of detection for 2 of the most globally prevalent subtypes (C and F), particularly relevant in developing countries.

Figure 3

To estimate the clinical efficacy of our assay, we evaluated a set of 9 seroconversion panels. But first, to determine the cut-off value for assigning a positive value to a sample, we tested 100 HIV-negative plasmas and assigned a cut-off value of 347 relative fluorescence units by adding 3 times the standard deviation to the mean (mean sample relative fluorescence units = 136; standard deviation = 71). In this way, we would expect a false positive rate of 1 in 1000 samples. The seroconversion panels we commercially obtained had viral loads ranging from 900 copies per milliliter to in excess of 1 million per milliliter with a mean load of 350 k copies per milliliter (Fig. 4 and Table 1). For the 52 samples that were HIV positive according to the Roche Standard 1.5 RNA PCR test, 37 were positive by the lateral flow assay compared with 32 samples registering as positive by the Coulter ELISA assay (Fig. 4; Tables 1, 2). The lateral flow assay sensitivity can be estimated from these data to be 71% ± 12% (95% confidence). To estimate the specificity of the lateral flow assay, 52 HIV-negative samples were tested and all yielded HIV-negative results. The specificity was estimated at 90% ± 10% (95% confidence). Compared with the results provided by the supplier for the Coulter ELISA on the same seroconversion panels, the lateral flow assay performs equally or better based on this limited sample size.

Table 1

Table 2

Figure 4

DISCUSSION

Despite the declining costs of HIV drugs accessible to the developing world, the costs of HIV laboratory tests remain high and this impedes successful establishment of treatment programs. Affordable HIV diagnostics that can be performed at regional and local health centers in developing countries are urgently needed, particularly in HIV screening, so as to diagnose individuals at risk and those newly infected who carry the highest infectious potential. To address the reduction of the window period, ultrasensitive HIV p24 antigen assays are needed. Here, we have developed and tested a laboratory prototype of an assay capable of high analytical sensitivity and clinical performance comparable to more time-consuming and laboratory-based ELISAs.

With our system, we have been able to show subpicogram detection of recombinant p24 protein spiked into human plasma and the detection of as little as 10 k viral copies per milliliter. When we tested for detection of the different viral subtypes, all subtypes were reactive although subtypes C and F that are respectively most prevalent across Africa and South America exhibited higher than average reactivity. As for our testing of clinical samples, we were able to detect 90% of specimens with viral loads equal to or higher than 10 k copies per milliliter with 100% specificity. This caliber of sensitivity performs equally or better than ELISA-based systems such as the Coulter assay and demonstrates the versatility of lateral flow devices for high analytical sensitivity coupled to inexpensive test units and fast turn-around times.

In order for this prototype assay to be fully evaluated, additional work is necessary. In particular, (a) incorporation of plasma separation to allow the use of whole blood specimens; (b) incorporating reagents in dry form on the test strip to reduce the amount of user involvement; (c) the development of a low-cost fluorescent reader device; and finally (d) performing a blinded prospective study using patient samples that are representative of the target populations in sub-Saharan Africa. Work is already underway addressing these issues, particularly the development of a hand-held reader utilizing light emitting diodes (LEDs) for illumination of the test strip and detection.

Although our current method of data capture relies on using a fluorescent microscope set at low power output and magnification, a battery-powered, hand-held reader employing off-the-shelf LEDs for illumination and diodes for detection is very feasible and is being pursued by us. This would result in a test combining a platform utilizing a single low-cost disposable test and an inexpensive multiuse reader. As such, lateral flow-based HIV diagnostics can make affordable laboratory testing in resource-poor settings a possibility.

Preliminary tests within our laboratory have demonstrated the ease with which the reaction reagents can be efficiently dried down on the sample pad or lyophilized into a pellet for rapid reconstitution upon contact with the plasma sample. Moreover, the use of citric acid shock for facilitating immunocomplex disruption can be easily incorporated within the plasma collection device so as to reduce operator steps to a simple 2-step assay. We envisage a configuration of the test where the user applies whole blood from a finger stick to a plasma separation pad; adds a neutralization chase buffer; and within 30 minutes, performs a readout of the results. Given the availability and scale of the reagents used, such a test would cost well below US $1.

In conclusion, we have developed and tested a prototype version of a mix-and-run lateral flow assay with ultrasensitive analytical sensitivity capable of integration as a self-performing point-of-care device. These parameters of high sensitivity/specificity, low cost, minimal operator input, and fast turn-around time all meet the specifications necessary for an HIV screening diagnostic of high clinical impact in resource-limited settings.

ACKNOWLEDGMENTS

We would like to thank Drs. Gerald Schochetman, John Hackett and John Russell of Abbott Laboratories for their technical advice and support in providing the antibodies used in this report. We would also like to thank Dr. David Charlton of Inverness Medical Innovations for his technical advice in lateral flow assay design.

REFERENCES