INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial meningitis is a serious and sometimes fatal infection affecting the central nervous system. Traditional laboratory diagnostic methods of culture for the identification of bacterial meningitis pathogens take up to 36 h or more. Furthermore, it has been observed that following an increase in the practice of starting antimicrobial therapy prior to clinical sample collection, the ability to confirm the pathogenic microorganisms of bacterial meningitis and septicemia has decreased by approximately 30% (8). This has been noted in Public Health Laboratory Service (PHLS) Communicable Disease Surveillance Centre (CDSC) data which show a growing discrepancy between the numbers of clinically suspected and culture-confirmed cases of bacterial meningitis and septicemia, with particular reference to meningococcal infection in England and Wales (17). To address this problem, nonculture methods like PCR have been employed (14) and shown to confirm additional cases of meningococcal disease (18).

The identification of all bacterial pathogens would be desirable, and to this end, amplification of conserved ribosomal nucleotide sequences has provided a strategy for universal detection of bacteria. Unfortunately, this approach is compromised by the presence of residual bacterial DNA contaminating the manufacture of commercially available reagents (4), which has frustrated attempts to exploit the 16S rRNA gene to develop a highly sensitive PCR assay for the universal detection of bacterial causes of meningitis (7).

The 7700 Sequence Detection System (ABI, Warrington, United Kingdom), known as TaqMan, enables amplification and detection to be carried out at the same time in a closed-tube system. Continuous real-time PCR monitoring permits the rapid throughput of large numbers of specimens in a highly standardized format. Multiplex PCR is particularly economical for small-volume samples such as cerebrospinal fluid (CSF) and pediatric blood specimens (10), which constitute a high proportion of samples referred to the PHLS Meningococcal Reference Unit (MRU) for testing. Furthermore, the closed-tube format reduces the chances of contamination. Recent developments in this laboratory have exploited the TaqMan platform to enable the introduction of sensitive and specific assays for the nonculture detection of Neisseria meningitidis (14). Advances in real-time PCR technology have now made possible the selective amplification of multiple genes in one reaction vessel by utilizing spectrally distinct phosphoramidite dye-labeled probes. Thus, multiple targets can be specifically identified in a single assay, obviating the need for repeated analysis.

The three major pathogens causing bacterial meningitis are N. meningitidis, Streptococcus pneumoniae, and Haemophilus influenzae type b (Hib); these three organisms accounted for 88.9% of all bacterial meningitis in England and Wales reported to the PHLS CDSC in 1998 (M. Ramsay, personal communication).

Prior to the introduction of Hib conjugate vaccination, more than 95% of invasive H. influenzae disease was caused by Hib (30). In countries where vaccination has been implemented, the incidence of invasive Hib disease has decreased by upwards of 90% (1). Despite the success of the Hib vaccination program that began in October 1992, laboratory reports of Hib disease continue to occur in England and Wales, with 20 bacteremia reports and 26 cases of meningitis and/or encephalitis in 1997 (PHLS CDSC). Furthermore, the reemergence of invasive Hib disease in a well-vaccinated population has been noted (13), emphasizing the necessity for continuous postvaccine surveillance.

N. meningitidis is a cause of meningitis and septicemia in adults and children and is now the major cause of bacterial meningitis in England and Wales, causing 71.9% of bacterial meningitis cases in 1998. These were predominantly caused by serogroups B and C. The annual incidence of meningococcal disease varies between 1 and 4 per 100,000 of the population, although recently the United Kingdom has experienced relatively high disease rates of 3 to 5 per 100,000 (25).

S. pneumoniae is the major cause of childhood invasive bacterial disease where Hib disease has been eliminated by vaccination (26) and is the second most frequently reported cause of septic meningitis (20, 24). In England and Wales the number of pneumococcal septicemia and meningitis cases reported annually increased substantially between 1982 and 1992 (2). Pneumococci were responsible for 19.1% of meningitis and/or encephalitis cases and for 9.6% of laboratory reports of bacteremia in England and Wales in 1997 (PHLS CDSC).

The developments in meningococcal and pneumococcal polysaccharide-protein conjugate vaccines have spurred the need for accurate laboratory confirmation of these infections in order to monitor the effect of vaccine implementation and continuing effectiveness.

This study outlines the development and evaluation of a single-tube multiplex real-time PCR for the simultaneous detection of N. meningitidis, H. influenzae, and S. pneumoniae in clinical samples using the TaqMan system. The sensitivity and specificity for the detection of the three major meningitis-causing pathogens are assessed. The application of the multiplex PCR as an epidemiological tool for improved nonculture diagnosis and case ascertainment with a large collection (4,113) of culture-negative specimens referred for meningococcal PCR testing is presented.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and culture methods. (i) Sensitivity. The sensitivity of each of the primer sets was evaluated using samples from culture-confirmed cases of meningococcal disease (157 plasma, 36 CSF, 31 serum, 31 whole-blood-EDTA, and 8 throat swab samples), H. influenzae disease (6 CSF samples, 2 plasma samples, and 1 whole-blood-heparin sample), and pneumococcal disease (23 CSF samples and 13 serum or plasma samples) that had been obtained from diverse sources. For determination of comparative sensitivities of the primer sets in multiplex and individual PCR formats, serial dilutions (undiluted to 10⁴) of quantitated DNA preparations for N. meningitidis, H. influenzae, and S. pneumoniae were tested. Different concentrations of template DNA from each organism were tested in combination in order to ascertain the ability of the assay to coamplify multiple gene targets.

(ii) Specificity. The specificities of the three primer sets were determined using genomic DNAs from bacteria and viruses most likely to be present in CSF and blood samples and from other Neisseria species. The bacterial strains obtained from the Clinical Microbiology Laboratory at Manchester Public Health Laboratory had been isolated from blood or CSF samples and stored in Microbank vials (Pro-lab Diagnostics, Neston, Wirral, United Kingdom) at 80°C. Additional S. pneumoniae strains were obtained from the PHLS National Collection of Type Cultures (NCTC 11887, NCTC 11899, NCTC 11903, NCTC 11904, and NCTC 11908). These strains were cultured overnight on 5% (vol/vol) blood agar (Oxoid, Basingstoke, United Kingdom). The H. influenzae strains were obtained from the National Collection of Type Cultures (NCTC 8466, NCTC 8467, NCTC 8469, NCTC 8470, NCTC 8455, and NCTC 8473) and cultured on heated blood agar (Oxoid) at 37°C in 5% CO_2. Staphylococcal species, Klebsiella pneumoniae, and Pseudomonas aeruginosa were cultured under aerobic conditions. In addition, nine Escherichia coli isolates, six Staphylococcus aureus isolates, two Enterobacter cloacae isolates, one Streptococcus mitis isolate, and one group B streptococcus isolate from culture-confirmed whole-blood samples were tested. Herpes simplex virus and varicella-zoster virus isolated from tissue culture, cytomegalovirus from PCR-positive blood samples, and hepatitis B virus from antigen-positive serum samples were also tested in the system. Additionally, extracted human genomic DNAs from 46 healthy adults were tested.

Quantitation of DNA in bacterial cultures. A sweep of colonies from a pure culture obtained using a sterile cotton swab was emulsified in 2 ml of sterile injectable water in a microbiological class 2 safety cabinet. Using a spectrophotometer (Pharmacia, St. Albans, England) set at 650 nm, the bacterial suspension was standardized to an optical density of 0.1 and adjusted to a concentration of approximately 20,000 bacteria/ml, which represents 40 bacteria per 2 µl of inoculum.

DNA extraction. For the clinical isolates and samples, a 100-µl aliquot of standardized suspension or sample was added to 1 ml of DNAzol (Life Technologies, Paisley, Scotland), vortexed, and incubated for 5 min at 20°C. A 500-µl volume of 100% ethanol was added, and the tube was vortexed and incubated for a further 10 min at 20°C. Following centrifugation at 12,000 × g for 10 min, the supernatant was aspirated and a further 1 ml of 75% (vol/vol) ethanol was added to the tube, vortexed, and centrifuged at 12,000 × g for 5 min. The supernatant was aspirated (with care being taken to remove any residual ethanol), resuspended in 50 µl of sterile water added to the tube, and incubated for a minimum of 10 min in a Dri-bath (Barnstead, Dubuque, Iowa) at 50°C.

PCR design. Oligonucleotide primers and dye-labeled probes (Table 1) were designed using the ABI Primer Express Software Package based on previously published ctrA (12), bexA (19), and ply (32) gene sequences. The ctrA sequence-specific probe was 6-carboxyfluorescein labeled, the bexA probe was tetrachloro-6-carboxyfluorescein labeled, and the ply probe was VIC (chemical name not disclosed by ABI at present) labeled.

PCR components and amplification profile. Based on a 25-µl reaction volume, the master mixture was prepared from the TaqMan Universal Master Mix kit (ABI). Briefly, this comprises a 300 nM concentration of each oligonucleotide primer; 25 nM 6-carboxyfluorescein-labeled probe; 100 nM (each) VIC and tetrachloro-6-carboxyfluorescein fluorescently labeled probes; 5.5 mM MgCl₂; 200 µM (each) deoxynucleoside triphosphates dATP, dCTP, dGTP, and dUTP; and 0.125 U of Taq DNA polymerase. A negative (no-template) control and control DNA preparations (2 µl) for each of the bacterial pathogens were included in every run. DNA was amplified with the TaqMan system using the following cycling parameters: heating at 95°C for 10 min followed by 45 cycles of a two-stage temperature profile of 95°C for 15 s and 60°C for 1 min. Real-time PCR results were based on the fluorescence readings taken by the TaqMan machine, which are used to calculate a baseline reading for each reaction. The cycle threshold (C_T) value is the PCR cycle number (out of 45) at which the measured fluorescent signal exceeds a calculated background threshold identifying amplification of the target sequence. If no increase in fluorescent signal is observed after 45 cycles, the sample is assumed to be negative.

Clinical evaluation of culture-negative samples. After establishing the specificity and sensitivity of the multiplex assay, it was used to assess the incidences of the three pathogens in 4,113 samples that had been submitted to the Meningococcal Reference Unit for meningococcal PCR testing. In addition, the sensitivity of the multiplex ctrA primers was compared with that of the set reported by Guiver et al. (14) that had been used to initially test the samples. The samples had been submitted between December 1998 and May 1999 and included plasma (n = 2,540), serum (n = 655), CSF (n = 451), whole-blood-EDTA (n = 398), and whole-blood-heparin (n = 9) samples. Miscellaneous specimen types (n = 4) included eye swab, urine, and pus. Patient ages ranged from 0 to 90 years. Clinical findings were suggestive of meningitis and/or septicemia being part of the differential diagnosis. These samples had been DNAzol extracted and stored at 20°C for between 2 and 8 months after initial meningococcal PCR testing. Before PCR screening for the three pathogens in the multiplex format, the samples were vortexed to resuspend any bacterial DNA. Any PCR-positive result was confirmed by testing with single primer sets for each gene target (ctrA, bexA, and ply). In addition, newly identified N. meningitidis ctrA-positive samples were tested with the serogrouping siaD PCR assay as a means of confirmation (3). Where it was not possible to confirm a multiplex result, the original specimen was reextracted and retested using the single primer in an attempt to confirm the initial result.

Statistical analysis. Differences between the results obtained with the N. meningitidis ctrA primer set and with the previously reported set were analyzed for statistical significance using McNenmar's test.

Nucleotide sequence accession numbers. The GenBank accession numbers for the sequences reported here are as follows: N. meningitidis ctrA, M80593; H. influenzae bexA, M19995; and S. pneumoniae ply, X52474.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Assay evaluation. (i) N. meningitidis ctrA PCR. The sensitivity of the meningococcal ctrA PCR was 88.4% when tested against samples from culture-confirmed cases of meningococcal disease (Table 2). There was no difference in the sensitivity of the ctrA primer set when compared in multiplex and single-primer-set formats using serially diluted N. meningitidis DNA (Table 3). The ctrA primer set amplified DNAs from meningococcal serogroups A, B, C, 29E, W135, X, Y, and Z and diverse serotypes and sero-subtypes. The primers did not amplify DNA from any of the other bacterial and viral DNA extracts tested (100% specificity) (Table 4). There was no cross-reaction with human genomic DNA.

(ii) H. influenzae bexA PCR. The sensitivity of the bexA PCR when tested against nine samples from culture-confirmed cases of Hib disease was 100%.There was no difference in the sensitivity of the bexA primer sets when compared in multiplex and single-primer-set formats using serially diluted H. influenzae DNA (Table 3). The bexA primer set amplified DNAs from H. influenzae Pittman types b and c. The primers did not amplify Pittman type a, d, e, or f. There was no cross-reactivity with any of the other bacterial and viral DNA extracts tested (100% specificity) (Table 4) and no cross-reaction with human genomic DNA.

(iii) S. pneumoniae ply PCR. The S. pneumoniae ply PCR sensitivity was assessed using 36 samples from culture-confirmed cases of S. pneumoniae disease and was determined to be 91.3% for CSF and 92.3% for serum or plasma, giving an overall sensitivity of 91.8%. There was no difference in the sensitivity of the ply primer set when compared in multiplex and single-primer-set formats using serially diluted S. pneumoniae DNA (Table 3). The primer set amplified the 23 pneumococcal serotypes tested (serotypes 1, 2, 3, 4, 5, 6, 7, 8, 9, 10A, 11A, 12, 14, 15B, 17F, 18C, 19, 20, 22, 23, 24, 31, and 33). There was no cross-reactivity with any of the other bacterial and viral DNA extracts tested (100% specificity) (Table 4) and no cross-reaction with human genomic DNA.

Multiple target amplification. Two different gene targets were tested using culture extracts and were shown to coamplify without a loss in sensitivity (Table 5). Three samples from which one organism had been cultured and one sample previously PCR positive were simultaneously positive with two of the three gene targets. These results were reproducible with the respective individual PCR assay. Of the three samples PCR positive for both N. meningitidis and S. pneumoniae, two had been confirmed as N. meningitidis infections by culture. Hib was cultured from the H. influenzae and S. pneumoniae PCR-positive sample (Table 6). These results were reproducible upon reextraction of the original sample.

Culture-negative samples. By testing a large number of culture-negative samples, the N. meningitidis primers developed for use in the multiplex assay were found to be significantly more sensitive (P < 0.0001) than the previously reported ctrA primer set (14). This represented an improvement in the meningococcal detection rate of 2.9% (13.0 versus 15.9% of total ctrA PCR-positive samples), or 87 additional cases identified by PCR alone. Of these samples, 53.8% were confirmed as serogroup B or C by siaD PCR. Conversely, 62 previously ctrA PCR-positive results were not detected using the multiplex ctrA primer set on repeat testing. Of these, 24.2% had been confirmed by siaD PCR. The total number of specimens ply PCR reactive by the multiplex assay was 73, of which 48 (65.7%) were confirmed using the ply primer set only. This represents an additional 46 cases confirmed by PCR alone, as two patients were ply positive for CSF and plasma samples. One sample not previously Hib culture positive was identified with the bexA primers in the multiplex assay (data from the PHLS CDSC).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Due to the problems associated with the development of a sensitive and specific universal PCR assay (7), a single-tube multiplex PCR was developed based on N. meningitidis, S. pneumoniae, and H. influenzae, which are responsible for upwards of 80% of cases of bacterial meningitis in developed and developing countries (11).

The ctrA gene is unique to N. meningitidis, and parts of the gene are highly conserved and common to all meningococcal serogroups (12). A previous study has demonstrated the rapid PCR amplification of the ctrA gene by continuous monitoring on the TaqMan system using samples from normally sterile sites (14). The limitation of this assay was the design of the forward primer near the 5' end of the gene, which contains sequence variation between different meningococcal serogroups, particularly those that contain sialic acid (B, C, Y, and W135) and those that do not (12). The ctrA primer set reported here amplified sialic acid-containing and non-sialic acid-containing meningococcal serogroups and was found to be significantly more sensitive than the previously described meningococcal ctrA PCR (14). Of the discrepant results between the two ctrA assays, a higher proportion of the samples positive with the new primer set were confirmed by siaD PCR (53.8%) than were confirmed with the previous ctrA primer and probe set (24.2%). The high proportion being confirmed enhances the degree of confidence in the results obtained with the new ctrA primer set. Increased sensitivity is most likely due to improved primer design, a characteristic that has been noted by Guiver et al. (14). In addition, amplification of all meningococcal serogroups may account for some of the additional positive samples; for example, a case of serogroup A disease not detected with the previous primer set was identified using the new set. The new ctrA primers failed to amplify DNA from some samples that had been previously positive; however, almost all of these samples had been weak positives in earlier analyses. These samples remained negative upon repeat testing with the original ctrA primer set and also after the sample was reextracted and retested. This failure to repeatedly detect DNA in specimens indicates that degradation of DNA is likely to be occurring and is likely to be caused by long-term storage and/or repeated freezing and thawing of samples. In addition, sampling error is more likely to be associated with previously low-level-positive samples. The gene target is present in low numbers as indicated by the high C_T value, and sample variation would lead to nonreproducible amplification.

The bexA gene encodes the capsulation-associated BexA protein present in all capsulated H. influenzae strains. These strains express one of six capsular polysaccharides (types a to f) (22). The amplification of the bexA gene for the detection of H. influenzae in CSF samples has previously been reported, and the gene was shown to be amplified in all six H. influenzae types. Van Ketel et al. used a relatively insensitive, gel-based detection system and experienced problems with contamination (31). Optimal primer and probe sets for use in the TaqMan system were developed using the criteria in the Primer Express software (ABI) from the available Hib sequence. This primer and probe set amplified types b and c only, and the inability to detect other serotypes was assumed to be because of nucleotide sequence variation.

The S. pneumoniae pneumolysin gene encodes the hemolysin species-specific protein toxin produced intracellularly by all clinically relevant pneumococcal serotypes (21). PCR amplification from clinical material is indicative of invasive pneumococcal infection. There have been several reports of pneumococcal PCR utilizing amplification of the pneumolysin gene (16, 27, 29), with a report of PCR for the detection of S. pneumoniae DNA in culture-negative samples where meningitis was the diagnosis (5). This assay utilized the autolysin gene and was evaluated using only a small number of culture-negative clinical samples (n = 11).

The pneumococcal PCR developed here was specific for the 23 pneumococcal serotypes tested while simultaneously offering a high level of sensitivity. The 4,113 samples tested were from patients clinically suspected as having meningococcal disease; 48 samples from 46 cases were confirmed as pneumococcal PCR positive. These had not been identified by laboratory culture, emphasizing the beneficial impact of including pneumococcal PCR in the routine diagnostic testing strategy.

Data suggest that 1% of all cases of meningitis are due to more than one pathogen (9), and such cases have recently been reported in the literature (6, 23). However, traditional laboratory methods may not always identify multiple pathogens in a single clinical sample, as identification from culture is based on the predominating organism and may be influenced by the use of selective culture media. PCR assays have been shown to amplify multiple pathogens (15, 28), but these assays relied on a nested PCR approach for improved sensitivity. The multiplex PCR in this study coamplified gene targets in a single-round PCR with a correlation between the organism identified by laboratory culture and the one with the lowest C_T value (Table 6). In all cases, the cultured organism has the lower C_T value, suggesting that this was the predominant organism in the specimen. The possibility of cross contamination of extracts was ruled out by reproducing the original results using another extract of the sample. The evidence confirms other observations that on some occasions, more than one organism may be present in a clinical sample and these may be underdetected by traditional laboratory methods.

In cases where it was impossible to confirm the multiplex PCR-positive results, the cycle number to reach the baseline threshold (C_T) value was greater than 34 in 93.9% of specimens. Plasmid titration experiments for the generation of a standard curve have demonstrated that the detection of samples around cycle 35 represents a target input of fewer than 10 copies (14). This is therefore approaching the limits of detection of the PCR, and it would be expected that a positive result would not always be obtained on repeat testing due to sampling error.

By utilizing the available TaqMan technology, the introduction of a three-in-one multiplex PCR enables rapid identification and a high throughput of samples (130 min for 96 specimens), with a modest additional cost for primers and probes in each reaction. The multiplex PCR demonstrated that testing a large number of previously culture-negative specimens provides information on the incidence of meningococcal, H. influenzae, and pneumococcal infections in clinical specimens originally referred for meningococcal PCR testing. The potential for guiding clinicians towards the most appropriate antimicrobial therapy and patient management is improved. The inclusion of the multiplex PCR in the routine molecular diagnostic screening regimen would provide a rapid and robust assay for the improved nonculture diagnosis and case ascertainment of meningitis and septicemia.