All PCPs powders used for plates (serotypes 1, 3, 4, 5, 6 A, 6B, 7 F, 8, 9 V, 10 A, 11 A, 12 F, 14, 15B, 18 C, 19 A 19 F, 23 F) were courtesy provided by Pfizer®. Powders for Cell Wall Polysaccharide (CWPS, Ref 3459) and 22 F (Ref 76966) were provided Statens Serum Institut, Copenhagen, Denmark. In accordance with the product usage recommendations, the powders were reconstituted in type I reagent grade water, with a final concentration following reconstitution at 10 mg/mL. After complete dissolution under agitation at 4 °C using a shaker, the solutions were aliquoted and stored at −80 °C until use for adsorption.

We used the International Standard for Human Anti-pneumococcal capsule Reference Serum, lot 007SP as reference serum. The 007sp was prepared from 287 healthy volunteers following vaccination with 23-valent pneumococcal polysaccharide vaccine (PPV23 - PNEUMOVAX II®). Calibration for 24 serotype specific IgG concentrations was made in 2011 during an International Collaborative Study involving 5 laboratories [14]. These values were derived following double adsorption of 007sp with cell wall polysaccharide (CWPS) and Ps 22 F [6, 15]. This reference serum is distributed by the U.S. Food and Drug Administration, Bethesda, MD [14, 16].

The concordance between the 18-plex ECL assay and the WHO-SSA was made using sera from 82 adult patients explored before and after vaccination with PPV23, i.e. 164 samples documented in IgG for 7 serotypes (serotypes 4, 6B, 9 V, 14, 18 C, 19 F and 23 F) and of which 40 documented in IgG for 6 more serotypes (7 serotypes plus serotypes 1, 3, 5, 6 A, 7 F and 19 A). All specific anti-PCPs IgG levels were assessed using the WHO-SSA in the immune-monitoring facility at Cochin University Medical Center (Paris, France). These sera were collected from March 2015 to February 2020 [17].

Four samples provided by an external quality assessment service (samples references 19-Dec-2023 236-5/236-6 and 12-Mar-2024 241-5/241-6, UK NEQAS®) were tested with the 18-plex ECL. We used the mean results provided after the final analysis of all participating centers (n = 6 to 14 depending on the serotype) for comparison with the 18-plex ECL: as it merged results obtained with LUMINEX technology assay and WHO-SSA, this comparison with 18-plex ECL was made independently.

The diagnosis of SPAD was established in patients presenting with unexplained recurrent bacterial infections or at least one severe infection caused by encapsulated bacteria (mostly meningitis and/or pneumonia), in the presence of normal total immunoglobulin levels, normal IgG subclass levels, normal B cell counts, and an impaired anti-polysaccharide antibody response 4 to 8 weeks after immunization with PPV23. A lack of prior PPV23 vaccination within two years before testing was also required to avoid the potential influence of hypo-responsiveness.

The interpretation of anti-PCP antibody responses and the diagnostic criteria for SPAD were defined according to the recommendations of the American Academy of Allergy, Asthma & Immunology (AAAAI). Assessment of pre- and post-immunization anti-PCP antibody titers was performed using the third-generation WHO-SSA [9, 18, 19].

An inadequate response for a given serotype was defined as a post-vaccination antibody concentration below 1.3 mg/L or a failure to achieve a fourfold increase from baseline. A twofold increase was considered acceptable if the baseline level was already above 1.3 mg/L. An overall inadequate response to PCPs was defined as an insufficient response to at least 70% of the tested serotypes (ranging from 7 to 13), leading to the diagnosis of SPAD. The same criteria were applied to the ECL 18-plex assay to assess its diagnostic performance in comparison to the WHO-SSA, which was used as the reference method.

The MesoScale Discovery® (MSD) technology is based on ECL detection buffer as documented previously [20].

The multispot configuration used two 96-well MSD-multispot carbon microplates with 10 spots/well format. First plate (‘plate A’) was coated with PCPs 1, 3, 4, 5, 6 A, 6B, 7 F, 8 and 9 V, and second plate (‘plate B’) was coated with PCPs 10 A, 11 A, 12 F, 14, 15B, 18 C, 19 A 19 F and 23 F, each at a 100 µg/mL concentration. Each well also contained one CWPS spot 200 µg/mL concentration, which was used to assess and the effectiveness of adsorption with CWPS and 22 F and the background reactivity of the assay.

007sp was used at 8 different 3-fold dilutions (from 1:150 to 1:328,050) to generate a standard curve to determine antibody concentrations in the ECL assays. Sera were tested at a fixed dilution of 1:2,000.

The reference serum 007SP, quality controls, and sera were diluted at appropriate dilutions in phosphate-buffered saline containing 0.05% Tween 20 (PBS-T) and 1% Bovine Serum Albumin (BSA) for blockage, 20 µg/mL CWPS, 20 µg/mL PCPs 22 F for adsorption, and were incubated overnight at 4 °C on a stirring plate. Each antigen-coated plate was incubated at room temperature for 1 h on an orbital shaker at 700 rpm with 5% BSA and PBS-T. Plates were washed 3 times with PBS-T, and 50 µL per well of the preadsorbed and diluted test sera/reference serum/quality control was added and incubated for 2 h at ambient temperature on an orbital shaker. Plates were washed 3 times with PBS-T; and 50µL MSD Sulfo-tag-labeled-goat anti-human IgG secondary antibody diluted 1:1,000 in 1% BSA PBS-T was added to each well and incubated for 1 h at ambient temperature on an orbital shaker. Plates were washed 3 times with PBS-T, and 150 µl of MSD ECL read buffer was added to each well.

Plates were read with the Quick-Plex SQ 120 (MesoScale Discovery®). The concentrations of antibodies were determined using MSD Discovery Workbench© software, by referencing their ECL responses against a standard 4-parameter logistic (4PL) model curve generated from the serially diluted 007SP reference serum. This protocol is summarized in Fig. 1.

Graphical abstract illustrating the distribution of polysaccharides used on different Meso Scale Discovery® plates, the schematic workflow of the ECL assay, the generation of 4-PL calibration curves based on the 007SP calibrant to determine anti-PCPs concentrations for each serotype, the interpretation of these concentrations for patient global anti-PCPs response and diagnosis of SPAD

The WHO-SSA ELISA method used at the Cochin University Medical Center has been previously described and is available online [10]. Briefly, medium-binding 96-well microtiter plates were coated with various concentrations of PCPs including serotypes 1, 3, 5, 4, 6 A, 6B, 7 F, 9 V, 14, 18 C, 19 A, 19 F, and 23 F. Human serum samples and the international reference standard 007SP were pre-adsorbed with 5 µg/mL of CWPS and 5 µg/mL of serotype 22 F polysaccharide for 30 min at room temperature to block antibodies binding to CWPS and other impurities present in pneumococcal polysaccharide coating antigens.

Serial dilutions of the adsorbed sera were then applied to the antigen-coated plates. Serotype-specific antibodies were detected using an alkaline phosphatase-conjugated goat anti-human IgG secondary antibody, followed by the addition of the substrate p-nitrophenyl phosphate. The optical density of each well was measured at 405 nm with a reference at 620 nm using an ELISA plate reader. The optical density values of the samples were compared to those of the standards, and antibody concentrations in human serum were calculated using a 4PL curve-fitting model.

Correlation between ECL-assay and WHO-SSA was assessed using Spearman correlation test for 7 or 13 serotypes. Concordance between 18-plex ECL and WHO-SSA was assessed using Bland-Altman (difference between each method values versus average) and Deming regression method for 7 and 13 serotypes.

Aberrant values (more than ten times difference between the two methods) were excluded from per-protocol bridging analyses, representing between 1 and 3% of results depending on serotypes, but were kept for SPAD diagnostic performances of the assay.

ROC curves were used to determine the performance of 1.3 µg/mL threshold in ECL-assay compared to WHO-SSA. The combination criteria of 1.3 µg/mL threshold and 2- or 4-times fold change was assessed with percentage of agreement between the two methods, globally and serotype-by-serotype. The level of imbalance in the discordance between 18-plex ECL and UK NEQAS® sera assessment for 0.35 µg/mL threshold, or WHO-SSA for the diagnosis of SPAD was statistically assessed using a McNemar’s exact test. Cohen’s κ-coefficient was also calculated, providing an estimate of the agreement between assays beyond which might exist due to chance alone. The value of the κ-coefficient ranges from − 1.0 to 1.0. Agreement consistent with chance alone yields a κ-coefficient near 0, whereas agreement far exceeding chance alone yields a κ-coefficient approaching 1.0 (from 0.61 to 0.80 being seen as substantial agreement).

Comparisons between pre and post vaccinal response were performed using Mann and Whitney test.

Data analyses and graphs were performed using the GraphPad Prism software version 9.1.2 (GraphPad Software, La Jolla, CA, USA) and online tests provided by developers at https://www.graphpad.com/quickcalcs.