Development and analytical validation of real-time PCR for the detection of Streptococcus agalactiae in pregnant women

Background Group B Streptococcus (GBS) is the leading cause of invasive neonatal infection. In this study, we aimed to evaluate the analytical validation of qualitative real-time polymerase chain reaction (qPCR) as a means to detect GBS. Methods Genomic DNA (gDNA) was purified from 12 ATCC bacterial strains, two belonging to GBS and the remainder acting as negative controls. Additionally, gDNA was isolated from 21 strains of GBS from various serotypes (Ia, Ib and II-VIII). All gDNA was used to evaluate the analytical validation of the qPCR method employing a specific Taqman probe. Inclusivity, exclusivity, anticipated reportable range, the limit of detection and robustness were evaluated. The methods used are described in international guidelines and other existing reports. The performance of this qPCR method for detecting GBS was compared to other microbiological methods used with vaginal-rectal samples from pregnant women. Results Our qPCR method for detecting GBS was analytically validated. It has a limit of detection of 0.7 GE/μL and 100% analytical specificity. It detects all strains of GBS with the same level of performance as microbiological methods. Conclusion Data suggest that this qPCR method performs adequately as a means to detect GBS in vaginal-rectal swabs from pregnant women.


Background
Streptococcus agalactiae, also known as Group B Streptococcus (GBS), is a leading cause of sepsis and meningitis in newborns and young infants [1][2][3]. Colonisation in pregnant women is estimated to be 11 to 35%, and socioeconomic factors directly impact the percentage of carriers in the population [4]. Even though screening programs and intrapartum antibiotic prophylaxis has reduced early neonatal onset of disease, this bacterium remains the leading cause of infection in newborns in developing countries [2,5,6].
The Centers for Disease Control and Prevention (CDC) recommends that vaginal and rectal swabs be used to inoculate a selective enrichment medium such as Lim broth and subsequently subculture the product on blood agar plates [2]. For better sensitivity or test efficiency, an alternative such as Granada medium could be used. In this case, a color change indicates GBS growth [7]. Unfortunately, this approach has variable sensitivity [8][9][10] and is associated with long turnaround times (18-72 h). A more rapid test that retains high sensitivity and specificity compared to conventional culture-based methods is real-time polymerase chain reaction (qPCR) [11]. This method will not fail to detect non-hemolytic GBS, as do culture-based methods. Nonhemolytic and non-pigmented GBS are reportedly found in 1 to 4% of pregnant women [12].
It is crucial that sensitive and accurate qPCR strategies be developed so that the efficacy of intrapartum antibiotic prophylaxis can be determined [7]. Also essential is the analytical validation of the method by evaluating parameters such as analytical specificity (inclusivity and exclusivity), analytical sensibility (anticipated reportable range and limit of detection), linearity, efficiency and robustness, when used for detection. Quantitative methodologies require estimates of precision, accuracy and limits of quantification [13]. A validated analytical method confirms that GBS detection will be an accurate, precise and reliable measure of potency.
A surface immunogenic protein (SIP) is reportedly exposed at the surface of intact GBS cells no matter the serotype [14,15]. Because the sip gene is conserved in GBS, qPCR strategies have been used to successfully detect GBS in clinical studies [16][17][18][19][20]. However, an analytically validated qPCR assay for GBS using the target sip gene has not yet been reported.
We developed a qPCR method that uses the sip gene as its target (Intellectual Property PCT/IB2017/056506), and herein, we report the analytical validation and standardisation of that method using Clinical and Laboratory Standards Institute (CLSI) guidelines MM17-A (Validation and Verification of Multiplex Nucleic Acid Assays) and EP17-A (Protocols for Determination of Limits of Detection and Limits of Quantitation and Evaluation) [21][22][23].

Collection of clinical samples
Vaginal-rectal swabs were collected from pregnant women at a private health center and sent to the Health Public Institute of Chile ISPCH for GBS detection. All samples (n = 103) were collected between the 35th and 37th weeks of pregnancy. The study protocol was approved by the Ethics Committee of Clínica Dávila, and written informed consent was obtained from all study participants, agreeing that this work can be published. Samples were collected and submitted to Stuart transport medium at room temperature at the health center. After 2 hours, samples were cooled to 4°C with ice packs, collected in a tertiary container and sent for analysis.

Collection strains
To perform standardisation and to determine the analytical parameters of the methodology, specific reference strains were obtained from the American Type Culture Collection (

Pretreatment of vaginal-rectal swabs
All vaginal swabs were inoculated in 3 mL selective Todd-Hewitt broth and supplemented with nalidixic acid (15 μg/mL; Sigma-Aldrich, St. Louis, MO, USA) and gentamicin (8 μg/mL; Sigma-Aldrich), then were incubated for 18 to 24 h at 37°C in an atmosphere of 5% CO 2 . This procedure was duplicated; one was analysed using qPCR and the other using a microbiological method (Granada agar, Christie, Atkins and Munch-Peterson [CAMP] test and latex agglutination; Fig. 1).

DNA extraction from clinical samples
Bacterial cultures were centrifuged at 10,000 g for 10 min. Then, from each pellet, a portion with a sterile tip was resuspended in 200 μL 5% Chelex-100 (Sigma-Aldrich), 1 μL commercial numan DNA (10 ng/μL) and 10 μL lysozyme (5 mg/mL, Biological, Salem, MA, USA). The mixture was incubated at 37°C for 15 min, then, 10 μL proteinase K (2 mg/mL, Merck) was added. Incubation at 56°C for 15 min was followed by a one-step inactivation (95°C for 15 min). Finally, the lysate bacterial culture was centrifuged at 10,000 g for 3 min then stored at − 20°C until qPCR analysis.

DNA isolation from each strain
All strains used for the determination of analytical parameters (previously characterised strains, both ATCC and non-ATCC) were cultured in blood agar, and gDNA was then extracted using E.Z.N.A. A manual bacterial DNA kit (OMEGA Bio-Tek, Norcross, GA, USA) was applied following manufacturer instructions. Genetic material was eluted using 100 μL eluent, then samples were screened for GBS using qPCR. The gDNA was quantified using a DS-11 spectrophotometer (Denovix, Wilmington, DE, USA). To determine the number of genomic equivalents, one GBS genome (2.16 Mbp) was obtained from the NCBI Reference Sequence (GenBank accession no. NC_004116.1), and its mass was calculated using: bp*M*N A , where bp is the size of the genome in bp, M is its molar mass in g/mol and N A is Avogadro's number in mol − 1 . One genomic equivalent of GBS was found to have a mass of 2.4 fg.

Real-time PCR procedure
Assays were performed using 1 μL gDNA template and a Stratagene Mx3000P thermocycler (Agilent Technologies, Santa Clara, CA, USA). To detect GBS, 1 μL of each primer (4 μM stock), 1 μL probe (2 μM stock) and 10 μL Brilliant Multiplex Real-Time PCR Master Mix (Agilent Technologies, Santa Clara, CA, USA) were used. The optimal reaction conditions were found to be 95°C for 10 min followed by 40 cycles at 95°C for 15 s, then 60°C for 1 min. A cycle threshold (Ct) value of 35 was used as the cut-off for positive fluorescence detection of the target.

Analytical specificity
Analytical specificity was evaluated in terms of selectivity, given the responsiveness of the method to selectively

Analytical sensitivity
Analytical sensitivity, the lowest amount of analyte that can be detected with (a stated) probability was evaluated. However, exact values were not the focus. This feature includes:

Anticipated reportable range
Serial dilutions of gDNA from S. agalactiae ATCC 12403 resulted in final concentrations between 4.8E04 and 0.48 fg/μL (20,000 to 0.2 GE/μL). Each dilution was amplified six-fold in four qPCR assays, providing 24 data points for the standard curve. Assigned (theoretical) versus measured values were converted to log 10 (GE/μL) and were then plotted for linear regression analysis.

Limit of detection
According to the CLSI guidelines, the LOD is the lowest DNA concentration that provides at least 95% detectable results. Four concentrations of purified gDNA were adjusted to 9.6, 19.2, 38.4 and 76.8 fg/μL (32 to 4 GE/μL) and were analysed using qPCR. A blank (pooled negative samples) was also included. A total of 60 points and blanks were obtained over the course of five consecutive days. According to the CLSI protocol, to determine LOD, the limit of blank (LOB), the highest qPCR measures expected in samples devoid of analyte, must be determined.

Robustness
To evaluate a test's robustness, combinations between factors in standardised and alternative conditions were generated. The factors analysed were: master mix, concentration of primers, concentration of probe, upper alignment temperature, lower alignment temperature, equipment and operator. (Table 1).

Precision
Three concentrations of purified gDNA from S. agalactiae ATCC 12403 were adjusted to 4.8, 9.6 and 240 fg/μl (2, 4 and 100 GE/μL or 0.3, 0.6 and 2 log 10 GE/μL). Precision was found using CLSI recommendations (the 20x2x2 method, performing 2 daily runs of duplicates on 20 different days). The estimates of within-laboratory precision standard deviation (S WL ) and repeatability standard deviation (S R ) were calculated using: S WL = [Ѵ day + Ѵ run + Ѵ error ] 1/2 ; S R = [Ѵ error ] 1/2 , respectively, where Ѵ day , Ѵ run and Ѵ error correspond to variance components of the day, run, and error, respectively. The variance component was calculated using MS, the mean of the square: Ѵ error = MS error ; Ѵ run = MS run -MS error/ n rep (number of replicates per run); Ѵ day = MS day -MS run / n run n rep (number of runs per day; number of replicates per run). The sum of the squares, provided by the twoway nested analysis of variance, was used to calculate MS, and DF, the degrees of freedom (for each source of variation), was obtained as: DF day = n -1; DF run = (n run -1)n day ; DF = Nn daynrun ; and DF total = N − 1, where n day = 20 days; n run = 2 runs per day, n rep = 2 replicates per run and N = 80 results per sample.
Two-sided 95% confidence intervals (CIs) were calculated using the standard deviations of the degrees of freedom associated with repeatability (df R ) and the degrees of freedom associated with the laboratory (df WL ), where α day = 0.25, α run = 0.25 and α error = 0.5, using the following formulas, respectively: df R = N -n daynrun and df WL = (α day MS day + α run MS run + α error MS error ) 2 / (α day MS day ) 2 / DF day + (α run MS run ) 2 / DF run + (α error MS error ) 2 / DF error . The CIs of repeatability were then computed using: S R [df R / χ21-α/2] 2 and S R [df R / χ2α/2] 2 . The CIs of within-laboratory precision were computed using: S WL [df WL / χ21-α/2] 2 and S WL [df WL / χ2α/2] 2 . Values obtained were re-expressed as %CV using the data (Supplementary Fig. 3).  Quality assurance of the results Parallel to the clinical samples, commercial exogenous internal controls were run to rule out errors in the DNA purification process: A TaqManRNase P Detection Reagent Kit (Applied Biosystems, CA, USA) was applied following manufacturer instructions. The thermal profile conditions were the same as those used for amplification of the sip gene.

Statistical analysis
The GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA) was used to carry out ANOVA analysis to determine the precision of the method and normal data distribution tests for choose the equation that determined the LOB and LOD, according to guidelines provided by CLSI documents EP05A3 and EP17-A, respectively.

Results
Determining the analytical parameters of the qPCR assay of GBS Analytical validation of the qPCR assay was obtained using the parameters specified. The first to be determined was necessarily the anticipated reportable range. The high and low concentrations were measured in GEs per microliter. The anticipated reportable range, determined by constructing two standard curves, showed a linear correlation between the expected and obtained values (R 2 = 0.9937) and between the assigned values for Ct and those found by the analytical software (R 2 = 0.9997) when the analyte was between 20E04 and 2 GE/ μL (Fig. 2). The sixth point on the curve (0.2 GE/μL) was not amplified in any of the 24 replicates run during the four tests designed to determine this parameter. The Ct values at the lowest detectable dilution (2 GE/μL) ranged from 33.06 to 35.36. The LOD (0.7 GE/μL) correlated with the results of the anticipated reportable  range, which lacked a detection signal when dilution was between 2 and 0.2 GE/μL. Exclusivity tests demonstrated that qPCR did not detect DNA in the range of 1 to 1000 pg/μL for S. pyogenes ATCC 21547, S. aureus ATCC 25923, E. faecalis ATCC 29212, S. pneumoniae ATCC 49619, E. coli ATCC 25922, K. pneumoniae ATCC 700603, K. aerogenes ATCC 13048, S. saprophyticus, Proteus spp. or S. marcescens. The 21 strains that corresponded to 9 of 10 available serotypes (Ia, Ib and II to VIII) and the γhemolytic strain S. agalactiae ATCC 13813 (serotype II) were detected at low DNA concentrations (24 fg/μL or 10 GE/μL), revealing that qPCR is inclusive and able to independently detect the various strains of the serotype. Each parameter was found to have an analytical specificity of 100% (Table 2; Fig. 3). Robustness was shown: Every combination produced an amplification signal, and Ct values remained between 32 and 34.
Comparative analysis between the microbiological method and qPCR for detecting GBS in vaginal-rectal swabs from pregnant women In total, 103 vaginal-rectal swabs (clinical samples) were analysed to confirm the presence of GBS. Samples were analysed blindly, and the results of the qPCR assays were compared to those found using a microbiological method. With the latter (a Granada agar plate, CAMP test and latex agglutination), GBS was found in 15 samples. The same was true for qPCR (Table 3; Fig. 4).

Discussion
A qPCR method to directly identify GBS from vaginalrectal swabs after an enrichment process in Todd-Hewitt broth was successfully developed. This methodology provides results that are the same as those obtained using the microbiological method routinely used in clinical laboratories. The qPCR methodology is quick and easy to interpret by technical health personnel, and it can be incorporated into a working algorithm to identify GBS. Analytical validation was performed by extracting gDNA using proteinase K, lysozymes and Chelex- Fig. 3 Amplification curves of measured parameters. Positive, negative and no-template controls were included in all reactions. a Standard curve (2E04-2 GE/μL) constructed to determine the anticipated reportable range. b Amplification of curves (32-4 GE/μL) used to determinate the limit of detection. The pool of samples lacking GBS strains did not produce an amplification signal. c Curves showing the amplification signals from all GBS serotypes tested, used to determine the inclusivity parameter. d Curves for negatives, samples from non-GBS strains, used to determinate the exclusivity parameter. Note. PAC: positive amplification control for GBS 100, reducing costs compared to a genomic DNA purification kit. This methodology is advantageous compared to others because it can detect low concentrations of genomic DNA (2 GE/μL) and can potentially be used as a GBS screening tool at delivery directly from a vaginal swab. A disadvantage is that the CDC does not recommend this practice. Moreover, molecular methods are limited in GBS detection because antibiotic resistance genes have not yet been incorporated into the PCR primer set. Doing so could enable the development of a rapid test that will provide additional information such as antibiotic susceptibility, for example, in the context of a penicillin allergy [29]. Meanwhile, microbiological analyses are recommended for determining susceptibility. Several molecular diagnostic tools are emerging as potential candidates for more rapid identification of invasive GBS disease or to rapidly identify women colonised with GBS [29]. Rapid diagnostic tests for invasive diseases include the MinIon®, loop-mediated isothermal amplification [17] and optical immunoassays. Those for rapid intrapartum colonisation screening include PCRbased methods [30]. However, publicly available data for the analytical validation of these methods for GBS screening in a clinical setting are lacking. Our analytical validation method assessed the performance of qualitative qPCR strategies based on the TaqMan probe for GBS detection, and it was conducted according to international guidelines. The sip gene from GBS was used as a molecular target, and inclusivity and exclusivity were estimated using gDNA from stocks of various bacterial strains. The analytical value of this methodology is in its anticipated reportable range (20,000 to 2 GE/μL) and its LOD (0.7 GE/μL). Additionally, it has a high level of precision in repeatability conditions, reinforcing its applicability ( Supplementary Fig. 3, Supplementary Table 1  and Supplementary Table 2). During pregnancy, rectovaginal colonisation of GBS is not persistent; it is characterised by a high rate of new acquisitions and a loss of colonisation over the course of pregnancy [31]. A culture-based method applied to samples taken between the 35th and 37th weeks of gestation might fail to detect all colonisations at delivery because GBS could infect between sampling and the onset of labor. For this reason, an intrapartum screening strategy using a qPCR method to identify and quantify GBS would help avoid riskfactor-based screening [32]. Compared to GBS cultures in enriched media, PCR has better sensitivity and specificity, ranging from 62.5-100% and 84.6-100%, respectively [31].
To date, the gold standard for detecting GBS is the microbiological method, which is inexpensive and easy to perform. However, it fails to detect non-hemolytic GBS strains, and β-hemolysis could be difficult to observe [7]. Therefore, negative samples should be further tested for other GBS strains using either a subculture on Granada agar plates, direct latex agglutination testing or (increasingly) molecular methods such as qPCR. Several clinical studies have been described for GBS detection by qPCR targeting the sip gene and have reported sensitivity between 95.4-97% and specificity between 84.6-99% (Supplementary Table 3); high sensitivity and specificity previously describes may be explained by the fact that the sip gene is conserved among all serotypes. On the other hand, none have declared analytical validation of the qPCR methodology, which could considerably improve future clinical validation studies. Although positive and negative controls are important when validating a result, a greater understanding of the methodology used, gained through analytical validation, is imperative. Our analytical validation parameters describe a trend that is similar to those seen with pathogens such as Trypanosoma cruzi (LOD, 0.23 and 0.70 parasite equivalents/ mL) [13], Plasmodium knowlesi (analytical sensitivity, 10 copies/μL) [33] and the American H7 virus (10E3 and 10E4 gene copies per reaction) [34]. Finally, the availability of this standardised and validated qPCR method provides new possibilities for screening women in labor and helps to improve the detection of GBS strains using microbiological methods.
Using qPCR to detect the sip gene has been reported elsewhere, and sensitivity and specificity were acceptable [16,19,20,35,36] (Supplementary Table 3). However, we established the limitations of this amplification process using analytical validation. Furthermore, our methodology, when compared with traditional methods currently considered to be the gold standard for GBS detection, shows 100% correlation. Nevertheless, the potential of this test should be explored because it could reduce swab incubation time or possibly allow direct extraction of genetic material. The technique developed here allows GBS to be detected in polymicrobial samples, and its specificity is superior compared to the microbiological method, which requires 3 to 5 days for incubation.

Conclusion
This kit accurately detects the bacterial agent GBS, it avoids cross-reactions with related bacteria and it provides 100% analytical specificity. It has a low LOD (0.7 GE/μL or 14 genomes per reaction) and optimal performance, and it meets analytical needs, as shown in its linearity and efficiency data, which correlates with current screening methods. These preliminary results are favorable and provide the basis for a study with a greater sample size, which could confirm the effectiveness of this test using clinical validation.