A Novel Point-of-Care Rapid Diagnostic Test for Screening Individuals for Antibody Deficiencies

PATH constructed the prototype RIgGS as lateral flow test strips. Combinations of Protein L (PRO-1790, ProSpec, Rehovot, Israel), Protein A (IA000019-10–2001-0 M rSPA native recombinant S. aureus Protein A [animal free], Syd labs, Hopkinton, Massachusetts, USA), and human immunoglobulin G (16–16-090,707, Athens Research & Technology, Inc, Athens, Georgia, USA) were applied onto nitrocellulose membranes (90CNPH-N-SS40 and CNPC-SS12, MDI, Ambala Cantt, India) using a contact tip dispenser (Model XYZ3060, BioDot, Irvine, CA, USA). The lower membrane is 90CNPH-N-SS40 and is striped with one line of 2 mg/ml Protein A (blue prototype at 0.65 ul/cm, pink prototype at 0.55 ul/cm, and green prototype at 0.9 ul/cm) and 5 lines of 0.5 mg/ml Protein L (blue prototype at 0.65 ul/cm, pink prototype at 0.55 ul/cm, and green prototype at 0.8 ul/cm). The upper membrane is CNPC-SS12 and is striped with a test line of Protein L (0.25 mg/ml, 0.8 ul/cm) and procedural control line of human IgG (1.5 mg/ml, 0.8 ul/cm). A detection conjugate was prepared from 40 nm colloidal gold particles with adsorbed Protein A and applied to non-woven conjugate pad materials (PTR7, MDI). Test strips were prepared using the nitrocellulose and conjugate pad materials combined with GE Healthcare LF1 sample separation pad (LF1, 8121–6621, GE Healthcare, Marlborough, MA, USA), absorbent pads (R025 and 243, Ahlstrom, Helsinki, Finland), cover tape (7759, Adhesives Research, Glen Rock, PA, USA), and adhesive-laminated polystyrene backing cards (GL-57623, Lohmann, Orange, VA, USA). Test strips were cut into 5 mm strips using a Kinematic Matrix 2360 cutter. The test running buffer was prepared using phosphate-buffered saline with Tween-20 (Sigma-Aldrich P3563, St. Louis, MO, USA). Barrel-shaped housings with an integrated sample collection tip and prefilled foil-sealed running buffer pot were sourced from BioSure (Nazeing, Essex, England). One prototype was a basic strip test without a cassette housing (Figs. 1a, b), while two prototypes used barrel-integrated housings, assembled by placing test strips into them (Figs. 1c, d, e). The two barrel-integrated housing tests differed by slight variations in the test formulation to achieve the targeted threshold. To run the basic strip prototype, the strips were placed into flat-bottom multiwell plates (29,442–070 Corning 9017, Tewksbury, MA, USA) to draw up sample and buffer in the well by capillary action.

The sample mixed with running buffer enters the sample pad and then traverses the lower membrane by capillary action where it interacts with Protein A and Protein L. This first interaction is intended to reduce by sequestration the concentration of IgG in the sample based on the chosen threshold. If the concentration of IgG in the sample is below the chosen threshold, no IgG will progress to the rest of the test. If the concentration of IgG in the sample is higher than the chosen threshold, free IgG will continue up the strip to the conjugate pad. Free IgG binds the Protein A gold conjugate and travels up the strip to the upper membrane. The upper membrane contains the test line and control line. Only conjugate that has formed a complex with IgG will bind with the Protein L test line; free Protein A gold conjugate will bind the control line. The three prototypes differ by the quantity of striped Protein A and Protein L on the lower membrane. This allowed us to evaluate RIgGS prototypes with slightly different thresholds.

Three candidate prototypes were developed to detect approximately 3 g/L IgG, which was targeted as the initial threshold for identifying children at risk for PAD based upon literature review [9‐11]. Research and development were conducted using commercially available serum specimens with varying concentrations of IgG but not necessarily from PID patients. These thirty-three specimens were acquired from Discovery Life Sciences (Los Osos, CA, USA) with IgG concentration characterized by Siemens BN II System nephelometer (Siemens Healthcare GmbH, Erlangen, Germany). IgG levels in the chosen samples ranged from 0.12 to 5.76 g/L. No data were available for patient diagnosis, preexisting conditions, patient treatment, and IgA or IgM concentration.

Collection, characterization, and storage of the specimens used in this study were conducted within an ethically approved clinical research study reviewed and approved by Comité d’Ethique Médicale, Institut Pasteur de Tunis (IRB protocol number IPT/11/LR11IPT02, IPT/PCI/LR11IPT02, and IPT/23/I/LR11IPT02).

PATH’s Research Determination Committee reviewed this study and determined that it did not involve human subjects as defined in 45 CFR 46.102(e). The use of the stored specimens in this study was further confirmed to be acceptable as secondary research and within consent given by the patients and families.

One hundred and eighteen stored samples were blinded for the PAD diagnosis and prior immunoglobulin testing for the prototype evaluation and parallel reference assay testing (turbidimetry, SPAPLUS®, Binding Site, Birmingham, UK). The samples included 32 healthy donors and 86 PAD patients for whom a diagnosis of AG, CVID, and HIGM had been assigned according to the criteria of the Expert Committee of International Union of Immunological Societies [12]. These PAD included 57 patients prior to receiving treatment (n = 19, 24, and 14, respectively) and 29 undergoing intravenous immunoglobulin treatment (n = 8, 16, 5, and respectively) (Table 1).

Laboratory technicians were trained to correctly use the RIgGS prototypes, including reading and interpreting negative and positive results and observing test run characteristics. Each specimen was run on all three prototypes, with the basic strip prototype coded as green and the two barrel-integrated housing prototypes coded as blue and pink (Fig. 1). The prototype testing was conducted on the same day, by the same laboratory technician. Reference testing for quantitative IgG, IgA, and IgM concentrations was conducted in parallel to prototype testing. The reference method (turbidimetry, SPAPLUS, Binding Site) used was the standard of care for patients for determining immunoglobulin levels. Test users remained blinded to reference results until completion of all sample testing.

Green prototype tests were run by first vortexing to mix the sample (serum/plasma) for approximately 5 s. Then 1.5 µl of sample and 10 µL test running buffer were combined in a microplate well and mixed by pipetting at least 10 times. The strip was placed into the well with the diluted sample for 90 s. Then, the strips were moved into a new well with 90 µL of test running buffer. Results were read at 15 min.

Blue and pink prototype tests were run by first vortexing to mix the sample (serum/plasma) for approximately 5 s. Then 1.5 µl of sample was added into the capillary sample tip at the bottom of the barrel, making sure that the full sample volume was dispensed into the tip. The barrel tip was then slowly pushed into the buffer pot with a smooth, steady motion, piercing the buffer pot foil seal and sealing the barrel into the pot in an upright position. The test remained standing vertical in the buffer pot for the duration of the test. Results were read at 15 min.

Figure 1 shows the two methods. At the designated time, the laboratory technician visually read the prototypes. The presence of a control line was required for the test to be considered valid and interpreted. The presence of any line of any intensity at the test line location was interpreted as reactive, indicating an IgG level above the test visible threshold. Test line intensity was scored using a color intensity guide. The absence of any line at the test line location was interpreted as nonreactive, indicating an IgG level below the test threshold (i.e., at risk for PAD). Quality of the test runs was assessed through observations by test operators and compared to predetermined criteria established prior to the study. The primary criterion for run quality was a valid result indicated by the presence of a control line.

Secondary criteria for repeating tests included monitoring for improper sample or buffer flow in the test. These criteria included observation of excessive time for capillary flow (as measured by time for visible fluid to reach specific sections of the test) and observation of fluid movement within the barrel that did not transfer to the test strip. Tests were repeated if the results were not valid (no control line), and a portion of tests were repeated for secondary quality criteria. The first valid result for a sample was used for data analysis, unless repeated. Repeated test results were used so long as they were valid and secondary test run criteria were met better than the first run.

To compare the target threshold to the analytical performance with clinical specimens, the data for the blue prototype was modelled as a continuous function of probability of detection versus IgG concentration for the study specimens. Due to interference of excess IgM, results obtained from HIGM specimens were not included in the model. The logistic regression model was fitted using a Hamiltonian Monte Carlo procedure implemented in Stan via the brms package in software R [13, 14]. The test binary result (0, 1) was the dependent variable, and the untransformed IgG concentration was the independent variable. Informative Gaussian priors with a mean of 0 and standard deviation of 3 were used for the model parameters. Two chains were each run for 10,000 iterations after a burn-in of 5,000 iterations. Convergence was visually assessed from the plots of both parameters. The fitted line and shaded area were plotted to identify median and the 95% credible interval from the logit transformed posterior distributions from the linear predictor.

The study included 32 individuals identified as healthy controls. The IgG reference test results ranged from 4.5 to 15.4 g/L (Fig. 2). Blue, green, and pink prototypes were reactive 100% (32 of 32), 100% (32 of 32), and 90.6% (29 of 32), respectively, in this category (Fig. 3).

The study included 19 individuals identified as AG and not yet receiving intravenous IgG substitutive treatment. The IgG reference test results ranged from 0.1 to 3.1 g/L (Fig. 2). Blue, green, and pink prototypes were nonreactive 89.5% (17 of 19), 84.2% (16 of 19), and 84.2% (16 of 19), respectively, in this category (Fig. 3).

The study included 8 individuals identified as AG receiving intravenous IgG treatment. The IgG reference test results ranged from 5.7 to 12.4 g/L (Fig. 2). Blue, green, and pink prototypes were reactive 100% (8 of 8), 100% (8 of 8), and 100% (8 of 8), respectively, in this category (data not shown).

The study included 24 individuals identified as CVID and not yet receiving intravenous IgG substitutive treatment. The IgG reference test results ranged from 0.2 to 4.8 g/L (Fig. 2). Blue, green, and pink prototypes were nonreactive 29.2% (7 of 24), 37.5% (9 of 24), and 29.2% (7 of 24), respectively, in this category (Fig. 3).

The study included 16 individuals identified as CVID receiving intravenous IgG substitutive treatment. The IgG reference test results ranged from 3.8 to 19.4 g/L (Fig. 2). Blue, green, and pink prototypes were reactive 87.5% (14 of 16), 100% (16 of 16), and 93.8% (15 of 16), respectively, in this category (data not shown).

The study included 14 individuals identified as HIGM and not yet receiving intravenous IgG substitutive treatment. The IgG reference test results ranged from 0.1 to 3.5 g/L (Fig. 2) and IgM ranged from 1.0 to 20.8 g/L, whereas the upper end of the normal IgM reference interval in the literature is 2.9 g/L [15]. Blue, green, and pink prototypes were nonreactive 35.7% (5 of 14), 35.7% (5 of 14), and 35.7% (5 of 14), respectively, in this category (Fig. 3).

The study included 5 individuals identified as HIGM and receiving intravenous immunoglobulin treatment. The IgG reference test results ranged from 3.5 to 10.8 g/L (Fig. 2). Blue, green, and pink prototypes were reactive 100% (5 of 5), 100% (5 of 5), and 100% (5 of 5), respectively, in this category (Fig. 2).

A single prototype was chosen for data analysis, based on both highest agreement with reference and preferred form-factor which was the blue prototype. Analysis focused on healthy controls and PAD cases not receiving immunoglobulin treatment, as would be encountered in screening. Agreement between the reference assay result and the blue prototype for clinical case definitions of healthy control, and AG, CVID, and HIGM not receiving treatment was 96.9% (31 of 32), 84.2% (16 of 19), 33.3% (8 of 24), and 35.7% (5 of 14), respectively, while total agreement across all case definitions was 67.4% (60 of 89) (Table 2).

The IgG reference test value was also compared to the age-related reference interval of normal IgG (supplemental file). Samples with reference test values below the normal range were identified as having low IgG. Using these criteria, the IgG reference test agreement with the clinical case definitions of healthy controls and PID subtypes not under treatment of AG, CVID, and HIGM was 96.9% (31 of 32), 94.7% (18 of 19), 95.8% (23 of 24), and 100% (14 of 14), respectively. Ten of the 24 specimens with CVID had a reference IgG value above the RDT prototype targeted threshold of 3 g/L. There was no correlation between the patient age and their IgG reference value (Fig. 4).