Detection of Neisseria meningitidis in saliva and oropharyngeal samples from college students

Carriage of Neisseria meningitidis is an accepted endpoint in monitoring meningococcal vaccines effects. We have assessed N. meningitidis and vaccine-type genogroup carriage prevalence in college students at the time of MenACWY vaccine introduction in the Netherlands, and evaluated the feasibility of saliva sampling for the surveillance of carriage. For this, paired saliva and oropharyngeal samples collected from 299 students were cultured for meningococcus. The DNA extracted from all bacterial growth was subjected to qPCRs quantifying meningococcal and genogroup-specific genes presence. Samples negative by culture yet positive for qPCR were cultured again for meningococcus. Altogether 74 (25%) of students were identified as meningococcal carrier by any method. Sixty-one students (20%) were identified as carriers with qPCR. The difference between number of qPCR-positive oropharyngeal (n = 59) and saliva (n = 52) samples was not significant (McNemar’s test, p = 0.07). Meningococci were cultured from 72 students (24%), with a significantly higher (p < 0.001) number of oropharyngeal (n = 70) compared with saliva (n = 54) samples. The prevalence of genogroups A, B, C, W, and Y was none, 9%, 1%, 1% and 6%, respectively, and 8% of students carried MenACWY vaccine-type genogroup meningococci. Saliva is easy to collect and when combined with qPCR detection can be considered for meningococcal carriage studies.

Meningococcal carriage detection using culture. Upon arrival, NS-agar cultures were incubated for up to two days at 37 °C and 5% CO 2 . On both days cultures were screened for presence of meningococcus-like colonies (grey, round and smooth colonies with convex shape). When found, 1-3 colonies were re-plated on Columbia Blood agar (CBA, bioTRADING Benelux B.V., Mijdrecht, the Netherlands) and tested for species identification using Matrix-assisted Laser Desorption/Ionization Time-of-Flight mass spectronomy (MALDI-ToF, Bruker Daltonik GmbH, Bremen, Germany). Separately for oropharyngeal and saliva samples, a single isolate with a score ≥ 2.0 for Neisseria meningitidis (database BDAL V8.0.0.0 + SR1.0.0.0, Bruker Daltonik) was stored at − 70 °C in Brain Heart Infusion (BHI, Oxoid) supplemented with 0.5% Yeast Extract (YE, Oxoid) and 10% glycerol. NS-agar cultures displaying any microbial growth were harvested into 2 ml of Todd-Hewitt Broth (Oxoid) supplemented with 0.5% YE and 10% glycerol. These harvests were considered to be culture-enriched for meningococci, and 0.7 ml of it stored at − 70 °C.
Detection of meningococcal DNA with qPCR. DNA was extracted from 100 µl of harvest of cultureenriched samples using DNeasy Blood and Tissue kit (Qiagen, Hilden, Germany) as previously described 22 . DNA eluted into 100 µl sample volume was tested in quantitative-PCRs (qPCRs) using primers and probes (Eurogentec, Seraing, Belgium) targeting sequences within metA, a gene encoding for a periplasmic protein, and a capsule transporter gene ctrA 23,24 . The qPCRs were conducted using Probes Master 480 (Roche) mastermix, primers and probes concentrations are listed in Table S1, with 5.5 µl of DNA sample used in 12.5 µl reaction volumes. The qPCR assays were conducted on LightCycler480 (Roche) with programme as described in Table S2. A tenfold serial dilution of DNA from a meningococcal strain (Table S3) was used as standard curve. C T thresholds for positivity were determined with Youden index calculated using Receiving operating characteristic (ROC) curve analysis 25 . Recovery of live meningococcus from culture-enriched samples. To test whether lower sensitivity of conventional diagnostic culture could account for differences between qPCR and culture results, cultureenriched samples first classified as negative by culture yet positive by qPCR were revisited to recover viable meningococci. For this second culture guided by qPCR results, CBA plates were inoculated with 100 µl cultureenriched sample in 10 -2 -10 -4 dilutions, incubated at 37 °C and 5% CO 2 , and screened for meningococcus as described above.
Genogroup-specific qPCRs. Two microliters of DNA extracted from culture-enriched samples were tested in 12.5 µl of reaction volume in qPCRs targeting genogroups A, B, C, W or Y 24 . Primer and probe concentrations are listed in Table S1. These qPCRs were conducted on a LightCycler480, using SensiFast probe No-ROX mastermix (Bioline, London, United Kingdom) and with programme described in Table S2. Culture-enriched samples were regarded as positive for a genogroup when the C T was lower than the cut-off value set for metA and ctrA. Control strains are listed in Table S3.
Genotyping of meningococcal strains. DNA extracted from cultured strains was tested in metA, ctrA and genogroup-specific qPCRs. Since not all genogroups were covered by qPCRs, a simplified criterium of positivity for ctrA was also applied to classify strain as genogroupable.
Statistical analysis. Data was analyzed using Prism (GraphPad Software, v8.4.1) and R (version 4.0.0). A p value of < 0.05 was considered significant. ROC curve analysis was performed using "cutpointr" package 25 , and Cohen's Kappa (κ) was determined in analysis of methods agreement. Youden index values were determined via bootstrapping (n = 1000) on metA qPCR data from saliva and oropharyngeal samples to determine the optimal cut-off value for qPCR detection 25

Results
All samples were collected in October and November 2018. Of 300 students that consented to participate, one person refused to have the oropharynx swabbed and was excluded from the study. Paired saliva and oropharyngeal samples from the remaining 299 (61% female; median age 20 years, range 16-28 years) students were analyzed (Fig. 1). Bacterial strains classified with MALDI-ToF as meningococcus were cultured from 72 students (24% of 299) of which 70 had strains isolated from the oropharynx and 54 from saliva (Table 1). Sixty-five (93%) of 70 oropharyngeal samples positive by culture had meningococcus isolated from the first culture and the remaining five strains were recovered when samples positive by qPCR yet initially culture-negative for meningococcus were revisited. For saliva, the same procedure resulted in fourteen samples positive for meningococcal strains in the first culture (26% of 54) and the remaining 40 in cultures guided by qPCR showing that initial diagnostic cultures displayed vastly reduced sensitivity for saliva when compared with oropharyngeal samples (14 vs. 65 strains cultured from 299 students, McNemar's test, p < 0.0001). The differences also remained significant after qPCR-guided culture (54 vs. 70, p < 0.001). Genogroupable meningococci were cultured from 62 students (21% of 299). Here too, the number of culture-positive samples was significantly higher for oropharyngeal swabs compared with saliva (58 vs. 45, p < 0.01) ( Table 2).
The study criterium for classification of a sample as positive for N. meningitidis by qPCR was detection of both metA and ctrA in DNA extracted from a culture-enriched sample, and was derived by calculating the optimal  (Table S4). Using qPCR detection we identified 61 students (20% of 299) as a meningococcal carrier. The difference in proportion of carriers detected by qPCR between oropharyngeal samples and saliva samples was not significant (59 or 20% vs. 52 or 17%, McNemar's test; p = 0.0704), both methods showed high agreement (96%; κ 0.88). Samples classified as positive for meningococcus by qPCR displayed significant correlation between metA and ctrA, supporting high specificity of molecular detection (Fig. 2). Detection by culture and by qPCR resulted in 71 (24% of 299) and 62 (21%) students identified as a meningococcal carrier in oropharyngeal and saliva samples, respectively. Altogether 74 (25%) students were identified as a meningococcal carrier and 62 (21%) as carrier of genogroupable meningococci by any method used (Fig. 3). When comparing methods and specimen types used for detection of meningococcal carriage overall, all evaluated procedures displayed comparable specificity of detection (Table 1) and primarily varied in performance for sensitivity and for positive predictive value (PPV).
The criterium based on both ctrA and metA was expected to impact negatively the sensitivity of meningococcal carriage detection by qPCR when compared with culture due to the presence of non-genogroupable meningococci that were likely to be ctrA-negative. Therefore, we compared methods and specimen types on samples containing genogroupable meningococci (Table 2), which were supposed to be positive for both ctrA and metA. The PPV and sensitivity of the evaluated methods were highest for detection by qPCR, whereas using saliva samples resulted in decreased negative predictive values (NPV) when compared with oropharyngeal samples. Detection of genogroupable meningococci using qPCR and saliva displayed increased PPV and comparable sensitivity when compared with detection of meningococcus in initial oropharyngeal cultures.
Next, we determined with genogroup-specific qPCRs the prevalence of genogroup A, B, C, W and Y carriage. The specificity of these qPCR assays was tested using culture-enriched samples negative for meningococcal carriage by culture and qPCR and none of the samples negative for ctrA generated a signal below 25 C T for a genogroup-specific gene (Fig. S1). Altogether, 51 (83.6%) of 61 students identified as carriers of meningococci with qPCR were positive for any of the genogroups targeted in group-specific qPCRs (Fig. 4). Genogroup-specific Table 1. The accuracy of Neisseria meningitidis detection in oropharyngeal and saliva samples collected from 299 students and tested with culture and using molecular methods applied to DNA extracted from culture-  www.nature.com/scientificreports/ C T s of almost all samples positive for any of the tested genogroups corresponded strongly to the C T for ctrA. The exception was a single oropharyngeal sample for which results were indicative of potential co-carriage of a genogroup Y strain with another ctrA-positive meningococcal strain of unidentified group (Fig. 4E). The prevalence of genogroup B (8.7% of 299) and Y (6.4% of 299) was highest while genogroups C (0.7% of 299) and W (1.3% of 299) were less prevalent. None of the samples were positive for genogroup A. MenACWY VT genogroups accounted for 8.4% (95% CI 5.7-12.1) carriage prevalence or 41.0% of meningococcal identified by qPCR. Results between specimen types were highly concordant for genogroups (Table 3).

Discussion
In this cross-sectional study, we evaluated the application of saliva for meningococcal carriage detection using both culture and qPCR-based methods. Our goal was to optimize meningococcal detection for future carriage studies assessing the impact of the meningococcal vaccines on carriage. Although meningococcal detection with culture resulted in fewer meningococcal carriers identified in saliva when compared with oropharyngeal samples, qPCR detection of meningococcus did not result in significant differences between these two sample types. Based on culture, we observed an overall carriage prevalence of 24.1%. The difference in positivity for meningococcus between oropharyngeal and saliva samples was likely caused by a greater difficulty to isolate meningococci from saliva cultures. In the saliva, a higher abundance of commensal species capable of growth on the culture media was observed and described first by Gordon in 1916 18 . Using qPCR detection we observed an overall carriage prevalence of 20.4%, and no significant differences were observed between oropharyngeal and saliva samples in positivity for meningococcus. Importantly, when detecting carriage of genogroupable meningococci, the method of testing culture-enriched saliva with a qPCR performed at least equally well compare with conventional diagnostic culture of oropharyngeal swab. Although numerous studies have implicated oral fluids in meningococcal transmission [8][9][10][11][12][13][14][15][16] , very few that tested saliva as specimen for assessing meningococcal carriage [18][19][20] report on seemingly contradictory results. In this context, our findings are in line with a meningococcal carriage study conducted recently by Rodrigues et al. 20 but in opposition to findings by Orr et al. describing a virtual absence of viable meningococci in saliva of N. meningitidis carriers 19 . However, with saliva collected by Orr et al. by swabbing gingiva, the volume of oral fluids cultured was likely to be lower than in our study, reducing sensitivity of carriage detection. Moreover, although we do not dispute the bactericidal properties of saliva, we believe that Gordon attributed the failure to culture N. meningitidis primarily to the highly polymicrobial nature of the sample rather than meningocidal properties of oral fluids per se 18 . This limitation can be addressed with   www.nature.com/scientificreports/ the introduction of antibiotics as supplements in media selective for meningococci, and the application of qPCR detection methods 26 .
Among 299 students, we observed an overall meningococcal carriage rate of 24.7%, a prevalence that is in line to what has been reported previously with pharyngeal swabs for this age group 2,15 . The prevalence of meningococcal carriage among young adults is considered to be higher than other age groups due to increased social interactions which facilitate meningococcal transmission 6 . In addition, age-related alterations in the microbiota of the URT may prime individuals for meningococcal colonization 27 .
VT serogroups targeted in the MenACWY vaccine accounted for 41.0% of meningococci detected in carriage, corresponding to a prevalence of 8.4%. Of these VT genogroups, genogroup Y was most frequently detected. While an outbreak of serogroup W was ongoing in the Netherlands during the fall of 2018, the prevalence of genogroup W in carriage was low (1.3%). The prevalence of genogroup C was also low, possibly reflecting reduced circulation since implementation of menC vaccine in the Netherlands in 2002 that included catch-up for 1-18 year-olds 28 . Although we did not collect information on participants vaccination status, with an approximately 94% vaccination coverage in the teenagers in the Netherlands we assume the great majority of study participants had received the menC vaccine 28 . Serogroup A was not detected in our study, its circulation appears to be limited in the Netherlands 15, 29 . The most prevalent genogroup among carriers was B. Genogroups B and Y have both been described to be most commonly detected genogroups among young adults 26 . The timing of our study coincided with start of a menACWY vaccine campaign among 14-18 years-olds 3 . However, at the time of our study individuals aged 16-18 years were not yet invited for the menACWY vaccine campaign, therefore we do not believe the menACWY vaccine had any substantial impact on the study findings.
Our study has a number of limitations. Firstly, MALDI-ToF may have identified more samples of students positive for meningococcus than qPCR detection with metA and ctrA carriage criterium as MALDI-ToF also takes non-genogroupable, unencapsulated meningococci into account, and is susceptible to misidentification 30 . To avoid misidentification by MALDI-ToF, we have only included bacterial strains for which identification displayed high confidence (≥ 2.0). Another limitation is false-positivity of qPCR tests. To minimize this issue, we have conducted ROC curve analysis and used the Youden index to determine a cut-off value for qPCR detection. Considering that the majority of qPCR positive samples facilitated successful recovery of viable meningococci, we conclude that false-positive results have had no significant impact on our conclusions.
One of the strengths of our study was the paired comparison of saliva and oropharyngeal samples in detection of meningococcal carriage. Furthermore, we have used selective media and inoculated plates immediately after samples collection. Fast processing of samples may be crucial for the sensitivity of meningococcal detection. Moreover, the combined use of two meningococcal qPCR targets for specific meningococcal detection in polymicrobial samples has allowed us to detect meningococcus with high specificity.
In conclusion, our findings show that the detected prevalence of meningococcal carriage between oropharyngeal and saliva samples was nondifferent with qPCR detection, the results for saliva were highly concordant with oropharyngeal swabs, and that the majority of samples positive with qPCR were shown to contain viable meningococci. Since the collection of saliva is easy, well tolerated and can be performed without professional assistance, we propose that saliva combined with qPCR-based surveillance can be considered for future meningococcal carriage studies.