Development and validation of a 6-plex Luminex-based assay for measuring human serum antibodies to group B streptococcus capsular polysaccharides

ABSTRACT

Six serotypes (Ia, Ib, II, III, IV, and V) cause nearly all group B streptococcal (GBS) disease globally. Capsular polysaccharide (CPS) conjugate vaccines aim to prevent GBS disease, however, licensure of a vaccine would depend on a standardized serological assay for measuring anti-CPS IgG responses. A multiplex direct Luminex-based immunoassay (dLIA) has been developed to simultaneously measure the concentration of serum IgG specific for the six prevalent GBS CPS serotypes. Assay validation was performed using serum samples obtained from human subjects vaccinated with an investigational 6-valent GBS CPS conjugate vaccine. Results for the assay are expressed as IgG concentrations (µg/mL) using a human serum reference standard composed of pooled sera from vaccinated subjects. The lower limits of quantitation (LLOQ) for all serotypes covered in the 6-plex GBS IgG dLIA fell within the range of 0.002-0.022 µg/mL IgG. Taken together, the 6-plex GBS IgG dLIA platform is specific for the six GBS serotypes included in Pfizer’s investigational vaccine, has a wide dilution adjusted assay range, and is precise (<18.5% relative standard deviation) for all serotypes, and, therefore, is suitable for quantitatively measuring vaccine-induced or naturally acquired serotype-specific anti-CPS IgG responses against GBS.

KEYWORDS:

• Streptococcus agalactiae
• group B streptococcus
• capsular polysaccharide
• serotype
• immunoassay
• IgG
• Luminex
• conjugate vaccine

Introduction

Group B streptococcus (Streptococcus agalactiae, GBS) are Gram-positive encapsulated bacteria that colonize the rectovaginal tract of ~ 25% of women and are the leading cause of neonatal sepsis and meningitis worldwide.^1–3 Though GBS disease is most prevalent in newborns within the first week of life, termed early-onset disease, infants are still highly susceptible to infection up to 90 days of life, during which they can develop late-onset disease. Older infants (>90 days of life) and adults can also experience severe invasive GBS infection.³ Intrapartum antibiotic prophylaxis of GBS-colonized pregnant women is standard-of-care in the United States and other (mainly high-income) countries to prevent early-onset disease, but such policies are not routinely implemented worldwide, especially in low- and middle-income countries.^4–6 Although implementation of intrapartum antibiotic prophylaxis has greatly reduced the incidence of meningitis and neonatal sepsis in the US,⁷ Intrapartum antibiotic prophylaxis is of little or no benefit to protect against late-onset disease.⁸

The capsular polysaccharide (CPS) that envelops the bacterial cell wall is a key virulence factor that facilitates immune evasion and thereby serves as a protective antigen for the pathogen.⁹ A multivalent maternal vaccine candidate consisting of six GBS CPS conjugated to CRM₁₉₇ (termed GBS6) is in clinical development at Pfizer.^10,11 GBS6 targets the six prevalent GBS serotypes (Ia, Ib, II, III, IV, and V) responsible for > 98% of infection¹² and aims to prevent GBS disease in young infants through active immunization of pregnant women. Further development of GBS6 and other GBS CPS vaccine candidates would be supported by a clear delineation of protective and non-protective anti-CPS immunoglobulin G (IgG) levels in infants, which could be related to vaccine-induced immune responses as an immunological endpoint for licensure.^13,14 Similar immune correlates have facilitated the licensure of other life-saving vaccines, such as meningococcal¹⁵ and pneumococcal vaccines.¹⁶

Published studies using a variety of immunoassays (ELISA- or Luminex-based) and reagents have shown a correlation between maternally-derived, naturally-induced anti-CPS serum IgG levels and reduced risk of GBS disease in newborns.^17–23 These studies have documented a wide range of IgG thresholds associated with protection, from 0.5 to 10 µg/mL,^{17–19,21–24} but these studies differed in geographical regions, variation in the definition of matched controls, statistical methodologies, serum source (maternal or infant), and the reference standard used to quantify anti-CPS IgG. For instance, Baker et al.¹⁷ used a monoplex direct-binding CPS ELISA and determined maternal Ia and III IgG concentrations >0.5 µg/mL corresponded to > 90% risk reduction in early-onset disease in infants born in the US. Furthermore, using a multiplex Luminex assay, Madhi et al.²² reported infant IgG concentrations of ≥ 1.04 and ≥1.53 µg/mL were associated with a 90% risk reduction for invasive GBS disease caused by serotype Ia and III, respectively. To date, the assays used to quantify anti-CPS IgG have not been standardized and did not include serotype IV.

Assay standardization is important for regulatory acceptance of serotype-specific immune correlates in vaccine licensure.¹³ A shared serological GBS assay would ensure consistency of results by allowing comparison of IgG data across laboratories (interlaboratory comparison) and studies, including the ability to pool results for meta-analyses.²⁵ Pfizer initially developed and validated a 6-plex anti-GBS IgG direct Luminex-based immunoassay (dLIA) to quantify serotype-specific IgG levels against vaccine-relevant CPS in human trials.¹⁰ This assay was then adopted by an international GBS consortium²⁶ and is being used to measure CPS-specific IgG concentrations from both clinical and seroepidemiological studies worldwide.¹¹ The development of a human serum reference standard with weight-based IgG assignments was previously described using surface plasmon resonance (SPR) and permits the comparison of anti-CPS IgG across serotypes.²⁷ The present study describes the validation of the 6-plex GBS IgG dLIA, and the results presented here were generated using the cross-standardized reference standard.²⁷

Materials and methods

Reference standard

The human reference standard was formulated by Pfizer and is composed of a pool of sera from 12 vaccinated subjects in the first-in-human phase 1/2 study (NCT03170609) evaluating the GBS hexavalent CPS conjugate vaccine, GBS6.^10,28 This reference standard has weight-based assigned IgG antibody concentrations for all six serotypes and is self-calibrating, as previously described.²⁷ The reference standard uses log–log linear regression to interpolate test sample IgG concentrations (µg/mL) from median fluorescence intensity (MFI) values.

Serum sample panels and quality control samples

Serum sample panels and quality control samples used in assay validation and the long-term proficiency panel were derived from adult sera of subjects vaccinated with GBS6 that had been heat-inactivated. Samples used during assay development included vaccinated non-human primate samples, as well as both vaccinated and non-vaccinated (placebo) human sera from the phase 1/2 study for GBS6 (NCT03170609), where all subjects underwent informed consent.

Preparation of GBS CPS Poly-L-Lysine (PLL)

GBS CPS (Ia, Ib, II, III, IV, and V) PLL conjugates were prepared using 1-cyano-4-dimethylaminopyridinium tetrafluoroborate (CDAP) conjugation chemistry. Briefly, CPS preparations were desalted through buffer exchange to deionized water and then lyophilized. For the conjugation reaction, each CPS preparation was dissolved in deionized water and freshly prepared CDAP (Sigma) solution was added and stirred at room temperature. To the CDAP-activated CPS, 0.2 M aqueous triethylamine was added and stirred at room temperature. 0.2 M sodium bicarbonate buffer (pH 8.8) was added and rapidly mixed, and then PLL (Sigma) solution was immediately added. The reaction was then mixed for 20 h at 4°C. To quench the conjugation reaction, 2 M Glycine was added and the reaction was mixed for another 2 h at 4°C followed by stirring at 150 rpm for 1 h at room temperature. The unconjugated material was removed by diafiltration using a 100,000 MWCO PES XL membrane (Pellicon). The total saccharide content was characterized via Anthrone assay,²⁹ the PLL content was characterized by TNBS assay,³⁰ and free CPS content was analyzed via a fractogel approach.³¹ The efficiency of the PLL conjugation was > 80% and the typical purity of the final product was > 90%.

Coupling of GBS CPS-PLL conjugates to the magnetic carboxylated microspheres

To assess microsphere coating optimization, based on the method by Pavliakova, et al.³², CPS-poly L-lysine [PLL] conjugates were chemically coupled to spectrally distinct Luminex MagPlex® microspheres using a two-step carbodiimide reaction. Prior to coupling, the carboxyl groups on the surface of the polystyrene beads were activated with a carbodiimide derivative, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC; Pierce) and stabilized using sulfo-N-hydroxysulfosuccinimide (sNHS; Pierce). Bead coupling was optimized by evaluating varying concentrations of the antigen (CPS-PLL) and different CPS-PLL conjugate lots.

GBS CPS-PLL conjugates were coupled to the spectrally-unique magnetic carboxylated microspheres (Luminex) using a modification to a coupling kit (xMAP Antibody Coupling Kit, 40–50016) and a method published by Pickering et al.³³

6-plex GBS IgG dLIA procedure

The 6-plex GBS IgG dLIA was based on the Luminex platform and similar in principle to Pfizer’s 13-plex Pneumococcal Luminex-based assay.³² Each plate included an 11-point human serum reference standard (diluted initially 1:50 and then serially diluted 2.5-fold), quality control samples (QCS), and test serum samples. Additionally, two wells containing assay buffer alone acted as blank controls. All samples and controls were diluted in assay buffer (0.5% BSA in 10 mM PBS/0.05% Tween-20/0.02% NaN₃, pH 7.2) in 96-well microtiter plates (Costar) and incubated overnight (20 h) with GBS CPS-PLL microspheres (5×10⁴ microspheres/mL per serotype) at 2 to 8°C with shaking (MaxQ 2000 shaker at 300 rpm). Specifically, 50 µL/well of samples diluted into assay buffer were added to 50 µL/well of blocked beads and tested in duplicate at the following dilutions: 1:500, 1:5,000 and 1:50,000. The following day, the assay plates underwent three wash cycles using 100 µL/well wash buffer (0.02% NaN₃ and 0.05% Tween-20 in 1X PBS) in a Tecan HydroSpeed™ plate washer (with magnetic bead attachment) to remove non-bound components. Following the wash step, a R-Phycoerythrin-conjugated goat anti-human IgG secondary antibody (Jackson ImmunoResearch, 109-115-098) was diluted 1:500 in assay buffer and 50 µL/well was added to the plate for 90 ± 15 min at room temperature (18 to 25°C) with shaking at 300 rpm. The plates were washed again as above and 100 µL/well was added after the last wash to resuspend the beads. Following a final shake at 300 rpm for a minimum of 4 min and a maximum of 4 h, the assay plates were read on a Luminex reader. The signal output was expressed as MFI which were evaluated against the human serum reference standard curve with weight-based IgG assignments (in µg/mL) for each serotype, as previously described.²⁷

Assay specificity

The specificity of the 6-plex GBS IgG dLIA was assessed by competitive inhibition experiments. Briefly, homologous or heterologous competitors (1 µg/mL) were added to serum samples that had been diluted in assay buffer. Following a 2-h incubation, the samples were transferred to assay plates and tested in the dLIA. Specificity was expressed as the percentage of reduction in serotype-specific IgG concentration ([IgG]) relative to no competitor as follows:

$\mathrm{specificity}(\mathrm{\%})=100\mathrm{x}\frac{\mathrm{GBS}\mathrm{CPS}\mathrm{serotype}-\mathrm{specific}\mathrm{serum}\left[\mathrm{IgG}\right]\text{break}-\mathrm{GBS}\mathrm{CPS}\mathrm{serum}\left[\mathrm{IgG}\right]\mathrm{incubated}\mathrm{with}\mathrm{competitor}}{\mathrm{GBS}\mathrm{CPS}\mathrm{serotype}-\mathrm{specific}\mathrm{serum}[\mathrm{IgG}]}$

The assay was considered specific for the target GBS CPS when ≥ 80% homologous inhibition and ≤ 25% heterologous/unrelated inhibition were observed.

Assay validation

Assay validation consisted of a series of experiments to address standard curve bias, dilutional linearity, and precision,^34,35 and was similar to the approach taken to validate other Luminex assays.³² For the assessment of standard curve bias, reference standard curves for each plate were fitted by linear regression of the log₁₀(MFI) on the log₁₀(Expected Concentration). For each of the dilutions of the reference standard curve, sample concentrations were then calculated by interpolating the MFI off the fitted standard curves from the dilutional linearity and precision runs. Bias was calculated as the difference between the calculated and expected log-concentrations for each point on the standard curve. A total of 152 standard curves were generated from the validation experiments for each serotype.

Dilutional linearity of the 6-plex GBS IgG dLIA was established with a panel of 12 individual serum samples from GBS6-vaccinated adults that spanned the range (high, medium, and low) of GBS CPS serotype-specific IgG concentrations for each of the 6 serotypes. These samples were tested eight times across a series of 11 2.5-fold dilutions, and the results were incorporated into plots of relative bias versus expected well concentration (not adjusted by serum dilution). A smoothing spline curve was fit to the relative bias data. The lower and upper limits, based on dilutional linearity, were defined as the ranges of expected antibody titers with acceptable relative bias. This range was determined by the intersection of the mean relative bias, as defined by the smoothing spline, and the pre-defined acceptance limits of 80% to 125%.

Precision data describe the closeness of measurements for a sample tested multiple times and is a measurement of assay variability that includes both repeatability and intermediate precision. Repeatability measures the assay variability over a wide range of antibody concentrations usually within a single assay run and, in this work, is included in the residual estimates; whereas intermediate precision measures the within laboratory variability (within the defined assay range) including relevant sources of variability (e.g., different analysts, times, and coated microsphere lots), as well as the remaining residual variability.

Precision was established by evaluating 44 samples, per serotype, that had been obtained from vaccinated adults. Precision experiments were run over multiple days, with multiple operators using two independently prepared lots of coated microspheres. The total variability for each sample’s dilution, expressed as percent relative standard deviation (%RSD), was plotted against the observed IgG non-dilution adjusted IgG concentration, and a smoothing spline curve was fitted to each plot.³⁶ The lower and upper IgG concentration limits, based on precision, were defined as the lowest and highest IgG concentrations at which the smoothing spline curve remained equal to or below the pre-defined acceptable variability of 25%.

The assay range for each serotype was based on the most conservative values from the lower and upper IgG concentration limits with acceptable standard curve bias (bias ratio between 80% and 125%), dilutional linearity (relative bias ratio between 80% and 125%), and average precision (%RSD less than or equal to 25%). The LLOQ was defined as the most conservative value of 1) the acceptable precision results (<25% RSD) from the least diluted samples, which account for the highest matrix concentration or 2) the dilution-adjusted lower limit of the assay range.

The total %RSD (intermediate precision) was calculated by combining the estimates of variability due to analyst, day, and coated microsphere lot, as well as residual variability, from a variance component analysis (VCA).

Stability of GBS PS-PLL-coated microspheres

A 44-member non-vaccinated human serum panel was used to assess the stability of the GBS PS-PLL-coated microspheres. The panel was tested every 2 months. All serum samples used for the 44-member panel were purchased from BioIVT (Westbury, New York). Two microsphere lots (Lot A and B) were prepared and tested in the 6-plex GBS IgG dLIA. Each panel sample was tested by two analysts, who ran the dLIA on separate days, to yield two data points per sample (time 0 time point represents a total of four data points), per microsphere lot at each testing time point. A simple linear regression was fit to the sample panel geometric mean IgG concentration obtained for all time points tested.

Results

Assay optimization and well characterization

Optimization of the coupling of GBS CPS-PLL conjugates to microspheres

To determine the optimal concentration for each capsular serotype (Ia, Ib, II-V), microspheres (1.25 × 10⁷ microspheres/mL) were coupled with 0.5, 5, 10, and 15 µg/mL solutions of a CPS-PLL conjugate. A similar approach was taken previously.³² Coated microspheres were evaluated in the assay with a human serum reference pool. The reference curves of each serotype demonstrated a similar MFI signal when the coating concentration reached optimal conditions. The GBS CPS-PLL concentration that achieved acceptable sensitivity while maintaining robust assay performance was 10 µg/mL for each of the 6 serotypes (data not shown). Bead coating robustness based on the optimized 10 µg/mL coating concentration was assessed by evaluating IgG concentrations from a 44-member serum panel tested with microspheres coated at either 9, 10, or 11 µg/mL. A geometric mean ratio (GMR) was generated that compared data from the 9 and 11 µg/mL coated microspheres to the 10 µg/mL coated microspheres. The %GMR for data from the 9 and 11 µg/mL compared to the 10 µg/mL coated microspheres were contained within 80% to 125%, which indicated optimal bead coating robustness (Table 1). The optimal coupling time was 180 ± 30 min for all serotypes (data not shown).

Table 1. Geometric mean ratios (GMR) with variable coating concentrations.

Serotype	n^a	9 µg/mL vs. 10 µg/mL GMR (%)	90% CI	11 µg/mL vs. 10 µg/mL GMR (%)	90% CI
Ia	132	103.6	(101.6, 105.7)	102.1	(100.1, 104.1)
Ib	129	94.6	(92.7, 96.6)	97.2	(95.2, 99.2)
II	128	98.1	(96.0, 100.3)	97.4	(95.3, 99.6)
III	122	97.7	(95.5, 99.9)	98.8	(96.6, 101.1)
IV	132	95.6	(93.3, 98.1)	98.3	(95.9, 100.8)
V	129	105.7	(103.4, 108.0)	106.1	(103.8, 108.5)

^an indicates the number of total study observations.

Robustness of 6-plex GBS IgG dLIA

Design of Experiments were performed to evaluate optimal conditions for robust assay performance, as previously described.³² To achieve this, a panel of 11 positive human sera, a reference standard serum pool and 3 QCS, were tested in a variety of different assay conditions. For each serotype, a VCA was carried out to estimate the variability of each factor and selected interactions. The percent relative standard deviation (%RSD) for each variance component are shown in Table 2. For each serotype, %RSD was reported for both primary and secondary incubation times (16, 20 and 24 h, and 60, 90 and 120 min, respectively) and primary and secondary incubation temperatures (2, 5, and 8°C, 18, 21.5, and 25°C, respectively). Considering each factor, the total variability was less than ≤ 16% for all serotypes (Table 2). Assays were shown to be robust within the following ranges: primary and secondary incubation times between 16 to 24 h, and 60 to 120 min, respectively, and primary and secondary incubation temperatures of 2 to 8°C, and 18 to 25°C, respectively.

Table 2. Results from assay robustness design of experiments.

Serotype	RSD (%)
	Primary Incubation		Secondary Incubation		Residual	Total
	Time intervals evaluated (16 h to 24 h)	Temperatures evaluated (2°C to 8°C)	Time intervals evaluated (60 min to 120 min)	Temperatures evaluated (18°C to 25°C)
Ia	2.3	0.5	0.3	0.0	7.1	7.5
Ib	7.8	0.0	2.6	1.4	13.6	16.0
II	2.6	0.8	1.6	0.0	12.2	12.6
III	1.7	3.2	2.3	1.2	9.9	10.9
IV	5.7	3.8	1.2	0.0	11.2	13.2
V	4.6	1.7	0.0	0.0	12.2	13.2

Specificity and interference

Specificity data is shown using the human serum reference standard in Table 3 and represents results from two independent experiments. The first experiment used homologous GBS CPS competitors, and a ≥ 93% reduction of the serotype-specific antibody concentration relative to the control (reference pool tested without competitor) was demonstrated for each of the serotypes. The second experiment used a heterologous GBS CPS and all serotypes showed ≤ 15% inhibition, with the exception of serotype Ia to Ib which resulted in ≤ 23% inhibition and may be attributed to the structural similarity between Ia and Ib CPS. Similarly, serum samples from multiple vaccinated adults also demonstrated acceptable specificity results (data not shown). To assess assay specificity for non-related bacteria, competition experiments were performed with Streptococcus pneumoniae serotypes 6B and 14 polysaccharide and demonstrated acceptable specificity with ≤ 13% inhibition for all GBS serotypes (Table 3).

Table 3. Specificity assessment of the 6-plex GBS IgG dLIA using an immunized serum reference pool.

Competitor 1 µg/mL	% Competition^a of Immunized Human Serum Reference Pool with Serotype-Specific GBS CPS as Competitor
	GBS Ia	GBS Ib	GBS II	GBS III	GBS IV	GBS V
GBS Ia	99	23	−11	14	−13	−9
GBS Ib	5	93	−9	−7	−9	−5
GBS II	−2	−5	93	2	−3	−5
GBS III	−1	−4	8	94	−7	−4
GBS IV	−3	−4	−7	−5	98	−3
GBS V	−6	15	−4	−8	3	94
Pn6B^b	−4	−2	−2	−4	−5	−4
Pn14^c	−2	−2	13	10	−2	−3

^aPercent (%) Competition was calculated as follows: [Baseline sample IgG result (human serum reference pools incubated in buffer only, no GBS CPS) – Serotype-specific IgG result of the human serum reference pool in the presence of GBS CPS competitors/baseline sample IgG result] x 100.

^bPn6B: Streptococcus pneumoniae serotype 6B polysaccharide was tested as a competitor in the 6-plex GBS IgG dLIA to assess assay specificity to other non-related bacteria.

^cPn14: Streptococcus pneumoniae serotype 14 polysaccharide was tested as a competitor in the 6-plex GBS IgG dLIA to assess assay specificity to other non-related bacteria.

The possibility of interference between the six GBS serotypes when tested in a multiplex assay was also evaluated by testing a panel of serum samples with multiplexed microspheres and compared to the same serum panel tested with single plex microspheres. The geometric mean ratio (GMR), expressed as a percentage, for performing the respective assay as multiplex vs. single plex were within 80–125% bias for all serotypes (Table A1 in Appendix A), indicating equivalence between the multiplex and single plex assays.

Validation and determination of assay range

A similar approach to that taken for a previously validated Luminex-based immunoassay³² was used for the validation study design and assay range determination for the 6-plex GBS IgG dLIA described here. Standard curves generated for the dilutional linearity and precision experiments were used for the standard curve bias analyses. A total of 152 standard curves were generated from the validation experiments for each serotype. Depending on the serotype, the lower and upper limits based on standard curve bias ranged from 3.00E–06 to 1.40E–05 µg/mL and from 4.23E–03 to 8.72E–03 µg/mL, respectively (Table 4).

Table 4. Assay range based on standard curve bias, dilutional linearity and precision.

GBS serotype	Well concentration (µg/mL)						Assay Range (µg/mL)
	Standard Curve Bias		Dilutional Linearity		Precision
	Lower limit	Upper limit	Lower limit	Upper limit	Lower limit	Upper limit	Lower limit	Upper limit
Ia	3.00E–06	5.24E–03	1.00E–06	1.10E–02	4.00E–06	9.15E–03	4.00E–06	5.24E–03
Ib	9.00E–06	5.67E–03	3.00E–06	7.82E–03	8.00E–06	1.88E–03	9.00E–06	1.88E–03
II	1.40E–05	8.72E–03	8.00E–06	2.08E–02	1.60E–05	1.79E–02	1.60E–05	8.72E–03
III	3.00E–06	4.23E–03	1.80E–05	3.41E–03	3.00E–06	4.48E–03	1.80E–05	3.41E–03
IV	8.00E–06	4.58E–03	3.00E–06	5.07E–03	8.00E–06	5.60E–03	8.00E–06	4.58E–03
V	1.30E–05	7.62E–03	2.10E–05	2.88E–03	6.00E–06	8.74E–04	2.10E–05	8.74E–04

Dilutional linearity describes the ability of the assay to consistently measure IgG concentrations over a specified range of sample dilutions. Twelve serum samples that span the expected assay range were selected for each of the six serotypes (Ia, Ib, II-V). Depending on the serotype, the lower and upper IgG concentration limits based on the dilutional linearity of 12 human serum samples, per serotype, ranged from 1.00E–06 to 1.80E–05 and from 2.88E–03 to 2.08E–02 µg/mL, respectively (Table 4).

Estimates of assay variability due to the day of run, and within run, were assessed for each of the 44 samples spanning the expected assay range, per serotype, at each of the assay dilutions. Depending on the serotype, the lower and upper limits based on precision ranged from 3.00E–06 to 1.60E–05 µg/mL and from 8.74E–04 to 1.79E–02 µg/mL, respectively (Table 4).

The most conservative lower and upper IgG concentration limits based on standard curve bias, dilutional linearity, and precision constituted the final assay range for each serotype and are shown in context of the human serum reference standard in Figure 1a–f. In addition, the lower and upper assay ranges at the well concentration level (Table 4) were multiplied by the dilution factors of 500 and 50,000, respectively, to calculate assay limits at the sample concentration level. Table 5 shows the dilution-adjusted assay range by serotype, as well as the LLOQ of the assay quantitation limits for each serotype. Samples with values above the upper limit of quantitation (ULOQ) may be prediluted and retested; therefore, no formal ULOQ was defined for the sample concentration.

Figure 1. Dynamic ranges of the reference standard curves for the 6-plex GBS IgG dLIA. The reference standard serum dilution profiles for each of the 6 GBS CPS serotypes are shown: (a) Ia, (b) Ib, (c) II, (d) III, (e) IV, and (f) V. Median fluorescence intensity (MFI) signals for the reference standard curves are on the y axis, and specific IgG concentrations (µg/mL) are on the x-axis. The vertical dotted lines indicate the lower and upper assay limits determined from the validation study. Error bars represent the standard deviation from independent reference standard curves (n = 3).

Table 5. Final assay range and lower limit of quantitation (dilution adjusted).

GBS serotype	Dilution adjusted assay range (µg/ml)
	Lower limit	Upper limit^a	LLOQ (µg/mL)
Ia	0.002	262.140	0.002
Ib	0.005	93.950	0.005
II	0.008	436.100	0.022^b
III	0.009	170.650	0.009
IV	0.004	229.075	0.004
V	0.010	43.700	0.010

^aUpper limit of quantitation (ULOQ) was calculated by multiplying the well concentration upper limit by 50,000.

^bSerotype II LLOQ was restricted based on the acceptable precision results (<25% RSD) from the least diluted samples, which account for the highest sample matrix concentration. All other serotype LLOQs were defined by the dilution-adjusted lower limit of the assay range.

To evaluate the impact of analyst, assay day, and coated microsphere lot on the overall variability of the assay, a VCA was performed whereby sample IgG concentrations for intermediate assay precision were derived from those IgG concentrations that fell within the established final assay range (Table 6). This is in contrast to precision analyses, which were performed to establish the assay range (Tables 4 and 5). The residual %RSD represented the amount of variability that cannot be ascribed to the analyst, day, or bead lot and includes assay repeatability. The total %RSD was the estimated intermediate assay precision, which was below 18.5% RSD for all serotypes (Table 6).

Table 6. Intermediate assay precision.

GBS serotype	n^a	Analyst (%RSD)	Day (%RSD)	Bead lot (%RSD)	Residual (%RSD)	Total (%RSD)
Ia	574	7.0	9.0	3.7	13.5	18.2
Ib	494	4.7	7.3	4.4	12.1	15.6
II	676	5.9	9.2	2.6	14.5	18.4
III	446	5.0	7.9	0.0	12.5	15.6
IV	552	4.9	8.2	0.9	12.8	16.0
V	551	5.1	8.2	1.7	14.6	17.7

^an indicates all sample results collected during assay validation using ≥ 2 laboratory analysts, ≥4 days, and 2 microsphere lots.

Long-term assay performance and microsphere stability

Long-term performance of the 6-plex GBS IgG dLIA beyond validation was monitored in two ways. One was by trending run data from the three QCS which are included on each assay plate and tested in all runs. QCS results collected from each assay run were < 16.2% RSD over time and across all serotypes (data not shown). The second approach was by periodic testing of proficiency panel serum samples that were selected to include low, medium, and high concentration samples (n = 44 samples/serotype) to gauge long-term performance across the assay range by plotting the IgG concentrations along with the GMC (with 95% CI) over time. With comparable IgG concentrations and overlapping CI within a given serotype across all time points tested, it was determined that the assay performed consistently throughout the >1 year of the testing period (Figure 2a–f).

Figure 2. Proficiency panel performance in the 6-plex GBS IgG dLIA. A proficiency panel with IgG concentrations ranging from low to high was evaluated post-assay validation, quarterly, for >1 year. Each panel shows the reported IgG concentrations (µg/mL; y-axis) across the quarterly timepoints tested (x-axis) for the following GBS serotypes: (a) Ia, (b) Ib, (c) II, (d) III, (e) IV, and (f) V. The geometric mean concentration with 95% CI is shown within a given time point and each black dot represents a proficiency panel sample. Of note, while 44 samples were used for the proficiency panel for all six serotypes, not every sample yielded a reportable result for every serotype. Therefore, the designated samples that constitute each of the six serotypes’ proficiency panel were slightly variable but lie between 41–44 samples.

The stability of the coated microspheres was monitored pre-assay validation by periodic testing of a panel of serum samples (n = 44) on two microsphere lots stored at 4°C over a 12-month period. As shown by the linear regression, the IgG geometric mean concentration (GMC) remained stable over the 12-month testing duration (Figure 3a–f). The slopes of the regression lines were not statistically different from zero for the GBS serotypes (p-values > .29).

Figure 3. Microsphere stability for the 6-plex GBS IgG dLIA. The geometric mean (GM) IgG concentration (y-axis) for two independently prepared microsphere lots (red circles, lot A; blue squares, lot B) across 12 months (x-axis) were tested by two laboratory analysts. Each panel depicts stability of the CPS PLL-coated microspheres for the following GBS serotypes: (a) Ia, (b) Ib, (c) II, (d) III, (e) IV, and (f) V. In each panel, the predicted line (black line) represents the linear regression line (or the line of best fit) for all bead lots (A and B) recorded over the 12-month period with 95% confidence intervals shown as the dashed lines.

Discussion

A standardized assay that quantitatively measures CPS-specific IgG across the six GBS serotypes of disease significance would permit the comparison of serological data across studies, an important step toward licensure of a GBS vaccine. This study outlines the validation of Pfizer’s 6-plex GBS IgG dLIA, which has been used previously to quantify anti-CPS IgG levels in GBS6-vaccinated humans^10,11 and animals.²⁸ The approach taken to develop and validate the assay described here was similar to that taken previously by Pavliakova, et al.³² This assay has been adopted as the standardized assay (termed GASTON [Group B Streptococcal Assay STandardizatiON]³⁷) by an international consortium and was transferred to multiple laboratories around the world. The primary objective of the GBS consortium is to develop and maintain a standardized IgG assay for use by the global GBS research community, for example, in epidemiological studies examining serocorrelates of infant GBS disease risk, or in clinical trials evaluating the immunogenicity of GBS vaccine formulations. While other 5-plex GBS anti-CPS Luminex assays have been reported previously,³⁸ the assay described here is the first immunoassay to measure serotype-specific IgG for the six prevalent GBS serotypes (Ia, Ib, II, III, IV, and V) responsible for > 98% of infection.¹²

Carol Baker and colleagues first documented the inverse relationship between serotype-specific antibody titers and risk of invasive GBS disease in newborns.^39,40 Their seminal study used a radioactive antigen-binding assay with purified CPS from serotype III GBS to assess total antibody levels in maternal and umbilical cord sera.³⁹ Since that founding study, numerous other studies have yielded similar findings with further attempts to derive IgG protective thresholds for the most common capsular GBS serotypes.^{17–19,21–24} Proposed protective thresholds were described for serotype Ia, III and V and ranged from 0.5 to 10 µg/mL IgG in maternal sera and, where applicable, 0.5–7 µg/mL IgG in cord/infant sera.²⁴ These studies were conducted across continents with varied assays, serum sources, and statistical analyses to propose a protective threshold value. Consequently, this critical value has remained only theorized, or predicted, based on limited datasets.

Serological data could not be combined to derive a single serum IgG threshold that is predictive of protection against invasive disease in infants, as the IgG assays were not standardized or poorly characterized.²⁴ The serotype-specific nature of IgG antibodies targeting the envelope CPS also makes it difficult to provide a basis for protective thresholds for less common capsular GBS serotypes.²⁵ The 6-plex GBS IgG dLIA permits direct cross-serotype comparison of CPS-specific IgG,²⁷ and in combination with the validation presented in this work, positions the assay as a powerful tool to serve as the primary serological readout for global seroepidemiology studies and vaccine clinical trials.

The validation of a serological ELISA and its associated standard reference serum (89-SF) was a critical step in the development and licensure of later pneumococcal conjugate vaccines.⁴¹ Unlike Streptococcus pneumoniae, the World Health Organization (WHO) has not yet adopted a serological assay for GBS. The WHO has stated that immune correlates for GBS can be derived from demonstrating a strong association between a validated immunoassay and protection against disease.⁴² As with S. pneumoniae, collaborative efforts toward a GBS assay would greatly contribute to comparability assessments and generation of a potentially regulatory acceptable serocorrelate, thus accelerating the pathway to vaccine licensure.⁴² The performance of the 6-plex GBS IgG dLIA assay was verified by the collaborating GBS consortium laboratories via an interlaboratory study.⁴³ Using a common set of reagents, the dLIA demonstrated reproducible and precise results across the five laboratories on three continents.⁴³ The validation experiments described in this study confirmed the standard curve bias, precision, dilutional linearity and specificity of this assay. Furthermore, assay robustness was exhibited for more than 1 year using a panel of human immune serum and demonstrated consistent assay performance.

For a GBS vaccine to be licensed using a serological approach, a direct comparison of vaccine-induced titers to titers from natural history studies is warranted. Such a comparison is possible if the serological assay in use is standardized across studies. After a vaccine is licensed, future effectiveness studies can help validate the anti-CPS IgG threshold that is likely to confer broad protection against GBS disease in newborns. Creation of a large volume GBS international reference standard for potential use with the 6-plex GBS IgG dLIA and a long-term proficiency panel to monitor the performance of the dLIA over time are key activities that are ongoing to support GBS research worldwide. Maintaining a standardized anti-CPS IgG assay and facilitating the development of key reagents to monitor the assay, an integral step taken for S. pneumoniae, would move the GBS scientific research field one step closer to bridging natural history data with vaccine immunogenicity data, and licensure of a GBS vaccine.

Author contributions statement

M. A. G. contributed to overall study design, conception of the work, data analysis, and original writing of the manuscript. M. L, D.G., D.P., and C. D. G. performed dLIA experiments, optimizations, data analysis and contributed to conception of the work; S. S. prepared the GBS-CPS poly lysine conjugates. H. H. N., A. M., and C. Y. T. performed statistical evaluation; P. G., W. V. K., and A. S. A. contributed to study conception, interpretation of data, and funding acquisition. All authors contributed to the development of the manuscript.

Acknowledgments

The authors would like to thank Pfizer colleagues Natalie Silmon de Monerri for critically reviewing the manuscript, Christina D’Arco for editorial support, and laboratory analyst William Manzo.

Disclosure statement

All authors are current or past employees of Pfizer and may, as a consequence, be shareholders. Pfizer was involved in the design, analysis, and interpretation of the data in this research study, the writing of this report, and the decision to publish.

Data availability statement

All raw and processed data files are available upon request to the corresponding author. The data are not publicly available due to privacy restrictions.

Additional information

Funding

This work was supported by Pfizer Inc.

Appendix A

Table A1 Comparison of the 6-plex GBS IgG dLIA run as multiplex versus single plex.

GBS Serotype	Multiplex versus Single plex
	N^a	GMR^b	90% CI
Ia	28	115.8%	(108.9, 123.2)
Ib	28	103.0%	(96.8, 109.5)
II	28	96.6%	(92.8, 100.5)
III	28	88.4%	(84.1, 92.9)
IV	28	93.6%	(90.9, 96.3)
V	28	109.4%	(106.7, 112.1)

^aN = total number of samples.

^bGMR = geometric mean ratio, expressed as a percent.