OUP user menu

Evaluation of non-culture diagnosis of invasive meningococcal disease by polymerase chain reaction (PCR)

Georgina Tzanakaki, Maria Tsolia, Vasiliki Vlachou, Maria Theodoridou, Anastasia Pangalis, Maria Foustoukou, Themistocles Karpathios, C. Caroline Blackwell, Jenny Kremastinou
DOI: http://dx.doi.org/10.1016/S0928-8244(03)00175-5 31-36 First published online: 1 October 2003


Antibiotic treatment prior to transport or admission to hospital has reduced the proportion of cases of invasive meningococcal disease (IMD) from which Neisseria meningitidis can be isolated by standard microbiological techniques. Identification of meningococci by polymerase chain reaction (PCR) was assessed in relation to microbiological diagnosis for cases over a 4-year period between 1998 and 2001. A screening assay for the IS1106 gene was used to detect meningococcal DNA and five additional assays for siaD and orf-2 genes were performed to determine the serogroup. PCR results were compared with results of bacteriological culture, other laboratory test results and clinical data. The sensitivity of the PCR assay for culture-confirmed cases was 98.5%. The specificity of the assay was 96% based on test results for patients from whom other bacteria were isolated, children with viral meningitis and afebrile negative controls. The siaD B/C/W-135 and Y as well as the orf-2 gene for serogroup A PCR assays were able to determine the serogroup for 75.2% of cases that were positive by PCR screening assay. When isolates from patients with IMD were tested by both agglutination and PCR, the results agreed in all cases. PCR is a useful tool for diagnosis of IMD when Gram stain and culture tests are negative due to antibiotic treatment prior to collection of samples for microbiological analyses.

  • Polymerase chain reaction
  • Meningococcal disease
  • Neisseria meningitidis
  • Diagnosis

1 Introduction

Neisseria meningitidis is an important cause of morbidity and mortality around the world [1] and causes predominantly septicaemia and meningitis [24]. N. meningitidis is a leading cause of childhood septicaemia and meningitis and among the leading infectious causes of death in healthy children and young adults in developed countries [5,6].

Initial diagnosis of invasive meningococcal disease (IMD) is often based on clinical symptoms such as fever, vomiting, neck stiffness and skin rash. The diagnosis is confirmed by the isolation and biochemical identification of N. meningitidis from either cerebrospinal fluid (CSF), blood or other specimens. Blood and CSF cultures are often negative in patients with IMD, particularly if they have been treated with antibiotics before samples are collected for microbiological assessment. The performance of a lumbar puncture is frequently contraindicated since it has been associated with clinical deterioration [7]; therefore, the development and introduction into clinical practice of non-culture-based diagnostic methods such as the polymerase chain reaction (PCR) is urgently needed.

PCR-based methods have been extensively used for the diagnosis of IMD in Europe since 1996 [817]. Evaluation of this technique both in the laboratory and in clinical practice by a limited number of studies has shown high performance characteristics; however, additional validation is needed before it is widely accepted. A PCR-based method for the diagnosis of meningococcal infections was initiated by the National Meningococcal Reference Laboratory (NMRL) at the Greek National School of Public Health in 1998. The aim of this study was to evaluate the PCR test in a large number of samples submitted to the laboratory from different parts of the country over a 4-year period between 1998 and 2001.

2 Materials and methods

2.1 Sources of specimens

There were 1622 samples (CSF 669, blood 953) from 1041 patients admitted between 1998 and 2001 to different hospitals throughout the country. In each hospital, samples were collected from the wards or the ICU by a clinician. The microbiology laboratory records were cross-checked in each hospital by a participating microbiologist. These data were added to information obtained by the laboratory-based surveillance protocol established by the NMRL.

A standard questionnaire for each case of suspected IMD included the following: demographic data; patient's symptoms and physical findings; physician's evaluation of patient's general condition and of predominant clinical presentation; laboratory findings; complications; and outcome. Blood cultures were obtained from each patient. IMD was suspected if a patient had fever and at least one of the following: haemorrhagic rash; meningitis; or septic appearance. Meningitis was defined by the presence of ≥10 white blood cells mm−3 in the CSF and/or microscopic identification of Gram-negative diplococci in the CSF, detection of meningococcal antigens or culture. If a lumbar puncture was not performed, meningitis was suspected if the patient had depressed sensorium and/or nuchal rigidity and/or meningeal signs. The diagnosis of viral meningitis was based on the judgment of the clinician responsible for the patient.

Patient samples were classified into the following categories: (1) N. meningitidis was isolated from blood or CSF (n=66); (2) cultures were negative but meningococcal antigen was detected in the CSF or Gram-negative diplococci visualised in CSF or skin rash smear (n=38); (3) clinically suspected IMD but cultures and other tests were negative (n=425). Control samples were obtained from patients in the following categories: (4) cultures were positive for other bacteria (n=62); (5) patients with clinically diagnosed viral meningitis who received antibiotics (n=336); (6) patients with clinically diagnosed viral meningitis who did not receive antibiotics (n=58); (7) afebrile children with no symptoms or signs suggestive of IMD who were admitted to hospital for other reasons, e.g., bronchial asthma or seizure disorder (n=56). Examination of specimens from patients in categories 4 and 7 by PCR was carried without information regarding previous clinical or laboratory diagnoses.

A latex agglutination method (Pastorex®, Sanofi Diagnostics, Pasteur or Murex Wellcogen Bacterial Antigen Kit, Abbott) was used for antigen detection in the CSF.

2.2 Source and characterisation of bacterial isolates

Meningococcal strains cultured from patients from each referring hospital were sent to the NMRL for further identification and for serogroup/serotype analysis.

Bacterial strains isolated from the blood or CSF of patients with meningococcal disease were cultured on chocolate agar and grown at 37°C in the presence of 5% CO2. The suspected meningococcal colonies were characterised by Gram stain, oxidase test and rapid carbohydrate utilisation test (Gallerie Pasteur, Pasteur Merieux).

Serogroups were determined by slide agglutination with polyclonal antisera to serogroups A, B, C, W-135, X, Y and Z (Wellcome Diagnostics). Serotypes and subtypes were determined by whole-cell enzyme-linked immunoassay with monoclonal antibody reagents supplied by RIVM (Bilthoven, The Netherlands) [18].

2.3 PCR method for detection of N. meningitidis

Blood and CSF specimens obtained from patients with signs and symptoms suggestive of IMD were examined for N. meningitidis DNA by the reference laboratory.

For the isolation of target DNA in CSF, 150 µl of the sample obtained after centrifugation at 2000×g (Hettich, EBA 12) for 4 min was added to 150 µl of Cheelex detergent and 650 µl of distilled water [19]. The samples were heated at 100°C for 30 min and centrifuged again at 12,500×g for 8 min. A volume of 150 µl of the supernatant was used for the PCR assay.

For the isolation of DNA from the whole blood samples, the ISOQUICK kit (ORGA, USA) was used according to the manufacturer's instructions.

To identify N. meningitidis, the insertion sequence IS1106 was used. For serogroup prediction (A, B, C, W-135 or Y), the oligonucleotides in the siaD gene (serogroups B, C, W-135 and Y) and in orf-2 of a gene cassette required for the biosynthesis of the serogroup A capsule were used [13,20]. The sizes of the expected amplicons are shown in Table 1.

View this table:

Oligonucleotides used for the identification and serogroup determination of N. meningitidis

SequenceGene amplifiedAmplicon size (bp)
5′-attattcagaccgccggcag-3′IS1106N. meningitidis650
5′-cgcaataggtgtatatattcttcc-3′orf-2Serogroup A400
5′-ggatcatttcagtgttttccacca-3′siaDSerogroup B450
5′-tcaaatgagtttgcgaatagaaggt-3′siaDSerogroup C250
5′-cagaaagtgagggatttccata-3′siaDSerogroup W-135120
5′-ctcaaagcgaaggctttggtta-3′siaDSerogroup Y120

The PCR reaction mix (50 µl) contained the following components: 3 µl MgCl2 (20 mM); 5 µl of the PCR buffer composed of 750 mM Tris–HCl, 200 mM (NH4)2SO4 and 0.1% Tween 20; 0.5 µl of each deoxynucleotide triphosphate (200 µM); 0.5 µl of the corresponding oligonucleotides (100 µM); 0.2 µl of 250 U Taq Polymerase (AB gene); 10 µl of template DNA. The PCR reactions were performed in a PCR thermocycler (Robocycler, Gradient 96, Stratagene, USA) with 39 repetitions of the cycle: 25 s at 95°C, 40 s at 55°C and 1 min at 72°C.

The amplicons were electrophoresed in 2% agarose gel and were visualised with a UV transilluminator following staining with ethidium bromide. Negative controls consisting of distilled water were used in each assay. Positive controls included standard strains used for species identification as well as strains of serogroups A, B, C, W-135 and Y.

2.4 Detection of antibodies to meningococcal outer membrane proteins (OMP)

The enzyme-linked immunosorbent assay (ELISA) published previously for detection of IgM and IgG antibodies to meningococcal OMP was used to screen sera from some children in whom the results for conventional and PCR methods did not agree [21].

3 Results

The results of the PCR tests for samples from the different categories are shown in Table 2.

View this table:

Detection of meningococcal DNA among samples classified by patient diagnosis

Patient categoryBlood or CSF cultureAntigen detected in CSFCSF Gram stain (+)PCR
1 (n=66)660NANA65 (blood 57, CSF 27)1 (blood 1, CSF 1)
2 (n=38)03863230 (blood 20, CSF 17)8 (blood 4, CSF 6)
3 (n=425)0425NANA288 (blood 238, CSF 105)137 (blood 126, CSF 55)
4 (n=62)0624 (blood 4, CSF 1)58 (blood 54, CSF 8)
5 (n=336)0336001 (CSF 1)335 (blood 335, CSF 336)
6 (n=58)058002 (CSF 2)56 (blood 58, CSF 56)
7 (n=56)0561 (blood 1)55 (blood 55)
Total390 (blood 320, CSF 153)651 (blood 633, CSF 516)
  • NA, not available because results of CSF antigen detection and Gram stain tests were not reported from all centres but only for the last year of the study (2001).

  • Numbers in parentheses in these columns denote numbers of samples, all other numbers are numbers of patients.

  • Peripheral blood PCR was positive in this patient.

  • Cultures negative for N. meningitidis but positive for other bacteria.

3.1 Assessment of sensitivity of the PCR assay

Sensitivity of the PCR assay was evaluated with data from categories 1 and 2 (Table 2). For category 1, the PCR assay was positive in 65/66 (98.5%) patients and 84/86 (97.7%) samples. The sensitivity of the PCR test compared to culture (category 1) was 98.5% (95% C.I. 91.9–99.7%). It should be noted that the one CSF sample which was negative in the PCR assay was obtained from a patient whose blood and CSF cultures were positive for N. meningitidis. The PCR assay for the blood sample of the same patient was positive. For another patient, the blood culture was positive and the PCR was negative, but the blood specimen tested in the PCR assay was obtained after 48 h of antibiotic treatment.

For category 2, the PCR assay was positive for 30/38 (79%) of patients and 37/47 (78.7%) samples. PCR was positive in 2/6 (33%) patients in whose CSF meningococcal antigens were detected and 28/32 (87.5%) patients in whose CSF the bacteria were identified by Gram stain. In category 2, there were 11 samples from children for whom culture and latex agglutination assays were negative. For one child, a serum sample collected 1 week after admission was demonstrated to have IgM antibodies against meningococcal OMP. Sera from 10 children collected 3–4 weeks after admission had IgG antibodies to the meningococcal OMP. There were four serum samples from patients with negative results for identification of meningococci by both conventional and PCR assays; each of the samples was negative for IgG and IgM.

For category 3, patients with symptoms of IMD but for whom there was no confirmation by conventional methods, the PCR assay was positive for 288/425 (67.8%) patients and 343/524 (65.4%) samples. The proportion of category 3 patients who were classified as positive by the PCR assay did not differ significantly from that of category 2 (79%) (χ2=2.08, P=0.154).

3.2 Assessment of specificity of the PCR assay

The specificity of the PCR method was evaluated in relation to data from patients classified in categories 4–7.

The following micro-organisms were isolated from blood (58 samples) and/or CSF (nine samples) from the 62 patients in category 4: Streptococcus pneumoniae (n=27); Escherichia coli (n=10); Staphylococcus aureus (n=5); Staphylococcus epidermidis (n=3); Klebsiella pneumoniae (n=6); Enterococcus faecalis (n=3); Citrobacter koseri (n=1); Salmonella typhimurium (n=1); Haemophilus influenzae type b (n=1); Group B Streptococcus (n=1); Group A Streptococcus (n=1); Acinetobacter baumannii (n=1); Enterobacter cloacae (n=2).

For category 4, the PCR assay was positive for 4/62 (6.4%) patients and 5/67 (7.5%) specimens. A positive PCR result was obtained for blood samples from three patients with E. coli bacteraemia and from both the blood and CSF sample from a patient with S. aureus meningitis.

Categories 5 and 6 consisted of samples from patients in whom viral meningitis was diagnosed. The diagnosis in most cases was presumptive and was based on clinical and epidemiological data but not on viral culture or PCR assay for enterovirus. Most of these specimens were submitted to the NMRL during an outbreak of enteroviral meningitis that took place in the country during 2001. Antibiotics were administered to a large proportion of these patients (category 5) because of a lack of laboratory confirmation of the suspected viral aetiology and also because of differences in management policies for clinically diagnosed viral meningitis in different centres.

For category 5, the PCR assay was positive for 1/336 (0.003%) patients and 1/336 (0.003%) specimens. For category 6, the PCR assay was positive for 2/58 (3.4%) patients and 2/58 (3.4%) specimens. Since meningococcal infection cannot be excluded in patients with clinical diagnosis of viral meningitis who had received antibiotics and for whom there was no confirmation of virus infection, only patients who were not treated with antibiotics (category 6) were included in the specificity estimation.

For category 7, the PCR assay was positive in one blood sample among the 56 tested (1.8%). Based on the above findings the specificity of PCR method for the diagnosis of IMD was 96% (169/176) (95% C.I. 92–98%).

The positive predictive value (estimated from categories 1, 2, 4, 6 and 7) was 93% (95% C.I. 86.5–96.6%). The negative predictive value was 94.9% (95% C.I. 90.7–97.3%).

3.3 Determination of serogroup by the PCR assays

Serogroup was examined by PCR in all samples if the sample was positive in the screening assay with the IS1106 oligonucleotide. Serogroup was determined for 288/383 (75.2%) samples. The serogroup distribution of samples from patients in categories 1–3 is shown in Table 3. The assays for serogroup determination were negative for samples that were negative in the screening assay for meningococcal DNA. N. meningitidis isolates from patients in category 1 were assessed by the NMRL by agglutination with serogroup reagents. For each of the isolates, the serogroup determined by PCR was identical to that obtained by conventional methods. For samples from patients in category 2, 33% of those that were positive in the screening assay were not serogroupable with the probe available, as were 30% of the samples from patients in category 3 that were positive in the screening assay.

View this table:

Serogroup distribution of N. meningitidis identified by PCR in different patient categories

Patient categorySerogroup, n (%)
1 (n=65)1 (1.5)46 (71)17 (26)1 (1.5)00
2 (n=30)7 (23)9 (30)3 (10)01 (3)10 (33)
3 (n=288)33 (11.4)115 (40)21 (7)12 (4)22 (7.6)85 (30)
Total (n=383)41 (10.7)170 (44.4)41 (10.7)13 (3.4)23 (6.0)95 (24.8)
  • These samples were positive in the PCR screening assay but were not groupable in the assays for determination of serogroup.

4 Discussion

Bacteriological confirmation and serogroup determination of clinically suspected IMD is important for contact management, outbreak recognition and for the detailed epidemiological surveillance of these infections. PCR-based methods evaluated in this study included an initial screening test using the IS1106 insertion sequence and tests for serogroup prediction using the orf-2 gene for serogroup A and the siaD genes for serogroups B, C, W-135 and Y. Samples from patients with clinically suspected or confirmed IMD were compared with controls with no evidence of IMD. To avoid bias, the diagnosis of IMD or other infections was based not only on clinical criteria but also on culture and non-cultural test results. Patients with clinically diagnosed viral meningitis were included in the analysis for specificity only if they had received no antibiotic treatment. The PCR method evaluated was shown to have good sensitivity, specificity, positive and negative predictive values.

The sensitivity of the method tested was high (98.5%) compared with cases confirmed by culture (category 1). The blood sample from one patient which was negative in the PCR assay was collected after 48 h of antibiotic treatment. For a second patient, the PCR test was negative for the CSF sample, but a blood sample collected from the same patient at the time was positive. False negative PCR results have been noted previously with different protocols. The reported percentages of false negative results have ranged between 3 and 74%[1215,22]. The proportion of false negatives in this study was 1.5%; however, the PCR assays correctly identified meningococcal DNA in patients from category 1.

It has been reported that meningococcal DNA was detected up to 72 h after antibiotic treatment, but the collection of the sample as soon as possible after administration of the antibiotic will optimise detection of DNA. It has been suggested that the presence of PCR inhibitors might affect the assay [13]. More accurate results might be obtained if both a blood and a CSF sample are examined if the patient's clinical condition allows.

For category 2, the non-cultural tests are not as specific for diagnosis as isolation of meningococci for documentation of IMD. The PCR assay was positive for 30/38 (79%) patients. Although no meningococci were cultured from the samples from patients in this group, the assay was positive for 87.5% of patients in whose CSF the bacteria were identified by Gram stain but only a third of samples in which only antigen detection was positive. This might reflect preservation of DNA in the intact bacteria and its degradation in specimens in which only the latex agglutination test was positive.

The specificity of the PCR method used was calculated after testing a large number of samples from patients with other infections and from afebrile controls. False positive PCR results have previously been reported when the IS1106 gene was used in a screening assay for the diagnosis of meningococcal infection [15]. These results were attributed to the fact that insertion sequences are mobile and have the ability to spread among different bacterial species and even genera. Among 5000 specimens examined in that study, a false positive result was obtained in four patients with blood cultures positive for S. aureus (n=2), E. coli (n=1) and S. pneumoniae (n=1); however, the total number of patients with cultures positive for other bacteria was not reported and an overall estimation of the sensitivity and specificity was not made. The current study was carried out with no information relating to the clinical diagnosis of the controls and the specificity was estimated at 96%. There were 67 samples from 62 patients and a false positive result was obtained from 2/58 (3.4%) patients with viral meningitis and from 1/56 (1.8%) afebrile patients with no symptoms or signs suggestive of IMD. A false positive result was found in 6.5% of patients in category 4 with blood culture confirmed infection caused by other bacteria: three blood specimens from patients with E. coli bacteraemia; one blood and 1 CSF sample from a patient with S. aureus infection. In comparison with other studies, all 30 specimens from 27 patients with cultures positive for S. pneumoniae were negative.

The PCR method employed in the present study differed in several aspects from those used in previous studies in which the IS1106 insertion sequence was also utilised [15,17,22]. The annealing temperature was decreased from 61°C to 55°C. The PCR protocol was different. In this study, DNA was extracted from whole blood as well as CSF while others used serum or plasma [10]. Whole blood has been used as a source of meningococcal DNA for Taqman PCR assays. Whole blood was used as a source of meningococcal DNA for a Taqman PCR assay recently with considerable improvement in sensitivity [17], compared with methods such as ctrA ELISA and Taqman ctrA for which sensitivities of 26% and 64% respectively compared to cultures have been reported [15]. The sensitivity of the method assessed in this study approached 100% compared with a large number of cases diagnosed by isolation of the organism. It is known that cross-reactions between N. meningitidis and several other species were reported in antigen detection assays [23]. Although frequent cross-reactions of Neisseria lactamica isolates have been reported [15], this species is unlikely to cause diagnostic problems in cases in which meningococcal infection is clinically suspected.

The positive and negative predictive values of the PCR assay used were high and were estimated at 93% and 95% respectively. To maintain high performance the test has to be applied in cases in which the diagnosis of meningococcal infection is strongly suspected.

The serogroup was determined in 288/383 (75%) specimens, which were positive in the screening assay. This was significantly higher than the proportion of 95/383 (36.5%) strains from carrier studies that were serogroupable by the PCR method (χ2=102.5, P<0.001) (Kremastinou et al., submitted for publication). Similar results have been reported in previous studies and could be attributed to the higher sensitivity of the screening PCR assay compared to the serogroup-specific assays [12,14,22]. The serogroup was determined by PCR for all isolates from patient in category 1. For each strain, the serogroup predicted by PCR was identical to that determined by antigen assays. The serogroup assays were not positive for any of the false positive samples; however, this does not exclude the possibility that non-serogroupable strains were detected by the IS1106 sequence. The specificity of a positive PCR result when the serogroup was predicted was 100%.

PCR was the only laboratory test which provided evidence for diagnosis of IMD in a large number of cases in this study. During the last year of the study, 62% of cases were confirmed by PCR only. In studies from other laboratories using PCR methods, confirmation of diagnosis by PCR only ranged between 23% and 50%[12,14,22]. Differences in use of other non-culture methods could account in part for the observed differences. Other factors include possible differences in sensitivity of the PCR methods used and in the extent to which it is being used in different settings.

In conclusion, systematic evaluation of the PCR method used by our reference laboratory over a 4-year period revealed that this test was an extremely valuable diagnostic tool. The high sensitivity and specificity of the method presented in this paper led the authors to consider that although a small number of apparently false positive results were encountered, the assay is a rapid and reliable method for confirmation of the diagnosis of meningococcal disease. Improvements related to modifications of the PCR methods by other groups during the last 4 years [15,16] imply that further refinement of the methods is desirable.

The development of PCR assays for non-cultural detection of meningococcal DNA has enabled confirmation of the cases, in particular those in which early antibiotic treatment precludes detection by culture. This is an important step towards more accurate recognition and surveillance of meningococcal infections.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
View Abstract