Re-validation and update of an extended-specificity multiplex assay for detection of Streptococcus pneumoniae capsular serotype/serogroup-specific antigen and cell-wall polysaccharide in urine specimens

National surveillance of pneumococcal disease at the serotype level is essential to assess the effectiveness of vaccination programmes. We previously developed a highly sensitive extended-specificity multiplex immunoassay for detection of Streptococcus pneumoniae serotype-specific antigen in urine in the absence of isolates. The assay uses human mAbs that detect the 24 pneumococcal serotype/groups targeted by the pneumococcal conjugate vaccines (PCVs) and pneumococcal polysaccharide vaccine (PPV-23) plus some cross-reactive types and the pneumococcal cell-wall polysaccharide. However, the previous assay had some limitations, namely the reduced specificity of the serotype 7F, 20 and 22F assays, for which non-specific binding in urine samples was observed. Here we report on the further development and re-validation of a new version of the assay (version 2.1), which offers improved sensitivity towards serotypes 7F, 18C and 19F and increased specificity for serotypes 7F, 20 and 22F by replacement of some of the antibody clones with new clones. Using a panel of urine specimens from patients diagnosed with community-acquired pneumonia or pneumococcal disease, the overall clinical sensitivity of this version of the assay based on isolation of S. pneumoniae from a normally sterile site is 94.3 % and the clinical specificity is 93.6 %, in comparison with clinical sensitivity and specificity values of 96.2 % and 89.9 % in the previous assay.


InTrODucTIOn
Streptococcus pneumoniae can be classified into serotypes according to the structure of its cell-wall polysaccharide (CWP) capsule. The total number of serotypes described depends on the criteria and methodology used. Using commercial polyclonal typing sera, 92 are currently described by SSI Diagnostica [1], with a further two (35D and 7D) distinguishable using reactions to available typing sera [2,3]. Characterization using mAbs [4] and genotyping of the operon encoding capsule synthesis genes has also revealed a number of additional subtypes and variants [5,6]. This number is likely to increase as more whole genome sequence (WGS) data become available and analysis of the capsular polysaccharide (cps) loci is routinely performed. The introduction of the PCV-7 pneumococcal conjugate vaccine (PCV) in 2000 and later the PCV-10 and PCV-13 vaccines in 2010 has helped to reduce the overall incidence of pneumococcal disease (PD) caused by some of the most prevalent serotypes associated with disease and antibiotic resistance at the time of their development [7][8][9]. However, vaccine-associated serotypes 3 and 19A continue to cause significant numbers of pneumococcal infections whilst previously less common non-vaccine serotypes are starting to increase in prevalence [7][8][9][10][11][12][13]. It is therefore important to perform surveillance of PD including serotyping to continually assess the effectiveness of pneumococcal vaccination programmes taking place in countries around the world. Detection and serotyping of nonculturable pneumococci is a valuable tool in the surveillance and treatment of PD. This is particularly true in non-invasive PDs such as pneumonia, where it is reported that, despite being the most common cause of community-acquired pneumonia [14], pneumococci are only isolated from blood culture in less than 25 % of adult pneumococcal pneumonia cases [15,16].
We previously described the development of an extendedspecificity multiplex immunoassay for detection of S. pneumoniae serotype-specific antigen in urine using fully human, full-length pneumococcal polysaccharide mAbs (urinary antigen detection, UAD assay) [17,18]. Although the UAD assay is highly sensitive, the individual assays to detect serotype 7F, 20 and 22F demonstrated poor specificity as a result of non-specific binding to (an) unidentified substance(s) present in some urine samples. We present the development and validation of a new version of the assay (v.2.1) in order to improve the sensitivity and specificity for some of these target serotypes by the inclusion of new mAb clones.

Bacterial strains
The same bacterial isolates and control strains used to validate the previous version of the UAD assay (v.2.0) [17] were used for v.2.1 together with an additional eight clinical Streptococcus oralis and eight Streptococcus parasanguinis isolates obtained from the Respiratory and Vaccine Preventable Bacteria Reference Unit (RVPBRU), Public Health England (PHE) National Infection Service, Colindale, UK. This made a total of 13 non-pneumococcal streptococcal species (n=58 strains), and 111 strains of various non-streptococci bacteria associated with respiratory infections or the urogenital tract, comprising 21 genera and 59 distinct species. Pure cultures of each isolate were suspended in PBS. Bacterial suspensions of >1.2×10 9 c.f.u. ml -1 were heat-killed or treated with 2 % formalin and used to spike 25 mM HEPES (Sigma) buffered pneumococcal antigen-negative urine samples as previously described [17].

clinical specimens
In total, 1995 urine samples were obtained from patients diagnosed with community-acquired pneumonia (CAP) upon admission to the Nottingham University NHS Trust hospitals in the UK between 2008 and 2018. Diagnosis of CAP was defined in those patients admitted to hospital with the presence of new or progressive infiltrates on chest radiograph and at least one symptom of acute lower respiratory tract infection, such as cough, fever, dyspnoea, sputum or pleuritic chest pain. These urine samples were frozen at −80 °C prior to transportation to PHE Colindale for testing with the UAD assay. On arrival at Colindale the urine specimens were held at ≤−70 °C for long-term storage and refrigerated at 2-8°C prior to testing. A further panel of 42 urine samples from patients with suspected PD based on clinical symptoms, contact links with confirmed outbreak cases and/or a positive BinaxNOW S. pneumoniae test (Alere) were obtained from RVPBRU at PHE Colindale. The RVPBRU urine samples were stored at 2-8 °C.

Multiplex serotype/serogroup-specific antigen detection assay
The UAD assay was performed as previously described [17] with the exception that the assay diluent no longer contained the negative human reference serum and comprised 2 % (w/v) BSA (Sigma) diluted in PBS at pH 7.4 (Sigma, P3813). The serotype 7F, 18C, 19F, 20 and 22F targeting clones used in version 2.0 of the assay were replaced with new clones. The antibody clones used in both versions of the assay are listed in Table 1 alongside the spectrally distinct Luminex MagPlex bead regions to which they were coupled.

repeatability
To test the repeatability of the UAD assay, three standard curves, consisting of a mixture of purified capsular polysaccharide (American Type Culture Collection, and SSI Diagnostica) for each of the 25 antibody targets at individual concentrations of 10, 3, 1, 0.3, 0.1, 0.03, 0.01, 0.003, 0.001 and 0.0003 ng ml −1 were tested over 5 days, giving 15 results in total. A pneumococcal antigen-negative control urine and 21 additional urine samples were included in the repeatability runs. Fifteen of these were clinical samples obtained between 2008 and 2013, from patients with culture-confirmed or suspected PD based on clinical symptoms, contact links and/or BinaxNOW S. pneumoniae antigen results as previously described. One sample was a urine specimen donated from a healthy donor which tested BinaxNOW S. pneumoniae antigen-negative and the remaining five samples consisted of BinaxNOW antigen-negative urine specimens that were spiked with purified polysaccharide antigens to produce a final concentration just above the estimated sensitivity cutoff as follows: serotype 2 polysaccharide at 0.04 ng ml −1 , 17F at 0.1 ng ml −1 , 20 at 0.04 ng ml −1 , 22F at 0.04 ng ml −1 , and a combination of serotypes 23F and 33F at 0.04 ng ml −1 . Each urine sample was tested in triplicate across the five days.

Interpretation of results
The test sample to negative control (t/n) ratio of the median fluorescence intensity (FI) reported by the BioPlex was calculated and used to determine the presence of pneumococcal antigen within a sample. For each sample these t/n ratios were normalized using a method similar to that described by Sheppard et al. [19] to account for any samples producing elevated background signals when compared to the negative control. This normalization involves dividing the t/n ratios by the 80 th percentile of the t/n ratios across the serotypes for that sample and therefore assumes that there are no more than four positive results (excluding the CWP) for each sample. A second normalization step was then performed by dividing each of the ratios reported for the individual serotype assays by the median of the normalized t/n ratios for all samples in that same serotype assay. This second normalization can only be used in cases where the majority of samples are negative and therefore the median would represent a negative result. Any results above or equal to the normalized t/n ratio of 2.0 were considered positive.

Examination of the positivity cut-off
The UAD v.2.0 assay uses a normalized t/n ratio of 2.5 as the positivity cut-off, calculated as previously [17]. Analysis of the FI data produced from the repeatability runs in UAD v.2.1 revealed that an equivalent t/n ratio of 1.8 is ≥5 sd above the mean FI of the negative controls in all 25 serotype assays and therefore, based on these data, would indicate >99.9 % certainty of positivity compared to the negative control urine sample (Table 2). Additionally, a normalized t/n ratio of 1.8 is ≥4 sd above the average of the normalized ratios for the negative (t/n<2.0) clinical samples (n=1194) tested in all 25 assays. This demonstrates that results above or equal to a normalized t/n ratio of 2.0 are discernibly above the negative control and, consequently, a normalized t/n ratio of 2.0 was set as the general positivity cut-off for the analysis of the urine results. *CWP refers to the cell-wall polysaccharide antigen target. *CWP refers to the cell-wall polysaccharide antigen target.

Analytical sensitivity
The limit of detection for the purified capsular polysaccharides for each assay was estimated using the standard curve dilutions as 3 sd above the mean FI of the negative controls in the repeatability runs. Using the value at 3 sd above the mean of each serotype assay negative control as a cut-off, it was estimated that serotypes 1, 2, 3, 4, 5, 8, 9V, 11A, 14, 18C and 20 can be detected at or below 0.3 pg purified polysaccharide per ml (0.3 pg ml −1 ), serotypes 6A, 10A, 15B, 19A, 19F and 22F can be detected at concentrations as low as 1 pg ml −1 , serotypes 7F, 12F and 33F at concentrations as low as 3 pg ml −1 , serotypes 6B, 9N and 23F as low as 10 pg ml −1 , and serotype 17F can be detected as low as 30 pg ml −1 .

Analytical specificity
Similar cross-reactions to those observed in v.2.0 of the UAD were seen in v.2.1 [17], with the exceptions that the new serotype 18C targeting mAb clone no longer cross-reacted with serotypes 35C and 42, the new clone targeting serotype 19F cross-reacted with all the serogroup 19 serotypes, and the clone targeting 22F no longer cross-reacted with serotypes 22A and 43 but did cross-react with serotypes 6C and 6D if present at high concentrations. The clone targeting serotype 22F also occasionally cross-reacted with serotype 20. The false-positive reporting observed for the previous clones targeting serotypes 7F, 20 and 22F as a result of non-specific binding with an unknown substance in some urine samples [17] was not observed with the new clones. Taking the crossreactions into account, positive serotype 6A, 6B, 7F, 9V, 10A,  11A, 12F, 15B,   The 16F antigen is detectable in the UAD assay due to a crossreaction with the 11A targeting clone (Table 3) and in the case of this particular urine sample, on initial testing an 11A/C/E or serogroup 16 antigen was detected at the positivity cut-off along with the serotype 5 antigen. However, on repeat testing the 11A/C/E or serogroup 16 antigen was detected below the positivity cut-off whilst the serotype 5 antigen was detected just above the cut-off, and therefore the sample was reported as positive for the serotype 5 antigen only. The BinaxNOW pneumococcal test can detect pneumococcal cell-wall C polysaccharide in urine and cerebrospinal fluid (CSF) samples with reported sensitivities of 74-90 % and specificities of 71-97 % [20][21][22][23], although sensitivities may vary in relation to the serotype [23]. In total, 412 urine samples tested positive for the pneumococcal antigen with the BinaxNOW test and 374 (90.8 %) of these were positive for a pneumococcal serotype and/or CWP antigen with the BioPlex assay.
Using the method described by Huijts et al. [24], which calculates the specificity of a pneumococcal serotype-specific urine antigen test (

DIscussIOn
Pneumococci continue to be a major cause of morbidity and mortality worldwide in the era of pneumococcal conjugate and polysaccharide vaccines. To assist with the detection and surveillance of PD, a UAD assay capable of detecting 24 pneumococcal serotype/groups plus some cross-reactive types and the pneumococcal CWP in the urine samples of patients with non-invasive diseases such as pneumonia was developed [17]. The assay uses 25 fully human, full-length pneumococcal polysaccharide mAb clones that target the pneumococcal serotypes/groups. Each antibody clone has different affinities and avidities for its target and this is reflected in the different sensitivities observed towards the various serotypes. We have now improved upon the previously published version of the assay by replacing some of the mAb clones to increase the sensitivity of the assay towards serotypes 7F, 18C and 19F. The replacement 19F targeting clone, however, is no longer specific for serotype 19F and instead targets all the serogroup 19 serotypes. The crossreactions observed with the 19F targeting clone, as well as some of the other cross-reactions described in Table 3 may be due to specific similarities in the polysaccharide structures between the serotypes, with certain clones targeting these in-common structures. Indeed, this is often the case for serotypes within the same serogroup [25]. The specificity of the assay for serotypes 7F, 20 and 22F was also improved as previous clones were observed to report false-positive results in some urine samples due to non-specific binding, meaning that the UAD v.2.1 assay can now detect serotypes 7F, 20 and 22F at lower concentrations with more certainty. The negative human reference serum was included in the assay diluent of UAD v.2.0 to reduce any non-specific binding that may occur when testing clinical samples in immunoassays using human mAbs. We decided to investigate the possibility of removing the negative human reference serum from the assay diluent as this represented an additional variable for which batch to batch variation would be difficult to control. Experiments after removal from the assay demonstrated that non-specific binding is not a significant problem with UAD v. As with the previous version of the assay, we observed crossreactivity of some mAb clones to other mitis group streptococci. This may be due to these non-pneumococcal streptococci carrying similar genes to those found in the pneumococcal capsular operon, which express similar polysaccharides to those found in pneumococci, including the pneumococcal cell wall [26][27][28][29][30][31]. However, these streptococci are associated far less with CAP or urine infections and are therefore less likely to be detected or isolated from the urine specimens of patients with clinical disease [32][33][34]. Additionally, the high sensitivity of the UAD assay means that it may be possible to detect pneumococcal antigen in urine samples of individuals who are carrying pneumococci due to colonization of their nasopharynx. Indeed, BinaxNOW-positive results have been reported in healthy children who have been colonized with pneumococci [35][36][37][38]. Author contributions SE: conceptualizsation, methodology, validation, formal analysis, investigation, writing -original draft preparation, review and editing preparation, visualization and project administration. CS: conceptualization, methodology, validation, investigation and writing -review and editing preparation and supervision. SR: investigation and writing -review and editing preparation. KS: resources and writing -review and editing preparation. NA: methodology, validation, formal analysis and writing -review and editing preparation. WSL: resources and writing -review and editing preparation. DJL: conceptualization, validation, writingreview and editing preparation and supervision. NKF: conceptualization, validation, writing -review and editing preparation and supervision.

Conflicts of interest
The Public Health England National Infection Service Vaccine Preventable Bacteria Section (VPBS) conduct contract research for pharmaceutical industries on behalf of Public Health England. No personal remuneration is received. The Public Health England National Infection Service Immunisation and Countermeasures Division has provided vaccine manufacturers with post-marketing surveillance reports, which Marketing Authorisation Holders are required to submit to the UK Licensing authority in compliance with their Risk Management Strategy. A cost recovery charge is made for these reports. WSL's institution has received unrestricted investigator-initiated research funding from Pfizer for a multicentre cohort study in which WSL is the Chief Investigator. WSL's work is supported by the National Institute for Health Research (NIHR) Nottingham Biomedical Research Centre.

Ethical statement
The diagnostic samples used in this research were obtained with approval from the Nottingham Research Ethics Committee (REC reference 08/H0403/80).