Introduction

The European Influenza Surveillance Scheme (EISS) is a European network of 18 countries collecting epidemiological and virological data on a weekly basis (1). This surveillance scheme provides information about the influenza viruses detected, then issues an alert whenever there is a risk for an epidemic, and monitors the impact and the spread of the epidemic in participating countries. National reference centres (NRCs) in each country are in charge of the isolation and the characterisation of the type, subtype, and ideally, the variant of circulating influenza viruses. The viruses are compared to the vaccine prototype strains in order to verify the adequacy of the strains and select new variants for the next vaccine composition.

The respiratory syncytial virus (RSV) is also responsible for winter epidemics. This infection is often severe and the clinical manifestations can be similar to influenza. The aetiological viral diagnostic is therefore also of major importance for the control of influenza vaccine efficacy.

The harmonisation of methods used to collect clinical, epidemiological and virological data concerning influenza in Europe is necessary, and a virological quality control assessment (QCA) was initiated during the 2000-01 winter season. The QCA was designed to test the sensitivity and specificity of the diagnostic tests used for influenza and RSV viruses. All of the participating laboratories were volunteers. The Laboratory of virology in Lyons carried out the QCA, and all results were sent back to this laboratory for further analysis.

Methods

Post vaccination samples

A selection of four post vaccination sera taken from adults were sent for the QCA. The immune status of vaccinees must be estimated by the titration of anti-haemagglutinin antibodies. The haemagglutination inhibition (HI) data consisted of observed titres (reciprocal of serum dilution). The results were analysed by comparing the HI data with the expected titre. The results were considered correct if the reported titre differed by no more than twofold.

Respiratory samples

The simulated respiratory samples were a mixture of infected and non-infected cells, and uninfected MDCK and HEp2 cells were also included in the panels as negative controls.

The influenza viruses were the wild type prototypes, corresponding to the World Health Organization (WHO) recommended strains for the 2000-01 winter season vaccine: A/Panama/2007/ 99 (H3N2), A/New Caledonia/20/99 (H1N1), and B/Yamanashi/166/98. RSV was the type A Long strain. Influenza prototype strains were adapted on MDCK cells. The titration of viral infectivity was performed on the original virus stocks (50% tissue culture infective doses [TCID50] per milliliter).

Sensitivity and specifity

The sensitivity of the virological diagnostic tests was calculated by introducing various concentrations of influenza viruses types A and B in the samples, and the specificity by the preparation of coded samples containing or not containing viruses. The laboratory test sensitivity was measured as the lowest influenza virus concentration detected.

Questionnaire

Before the QCA began, a questionnaire was sent to all of the participating laboratories in order to determine the techniques, reagents and cell lines used in each laboratory.

Direct analysis

The viral direct test was based on the viral antigen detection by immunofluorescence (IF) or enzymatic assay (ELISA) and/or the viral genome detection after amplification by polymerase chain reaction (RT-PCR). The viruses could be isolated on cell culture. Samples of 2ml were coded. Samples were prepared and coded by the members of the QCA team, which was completely independent from the diagnostic technicians working in the Lyons NRC.

A standard form to report the results was sent with each panel.

The first panel was sent on 20 November and the second on 4 December. The two panel samples are listed in tables Ia and Ib. Each NRC were requested to test the samples using their usual serological and diagnostic techniques.

Table 1a. Panels components: simulated samples

^a Numéro du prélèvement / Sample number
^b VRS + virus grippal A H3N2 / RSV + Influenza virus A H3N2.

Table 1b. Panels components : sera from vaccinees

^a: Anticorps / Antibodies.

Participating laboratories

Sixteen influenza NRCs, including the reference laboratory in Lyons, participated in the QCA. The list of participants is given in table 2 (countries listed in alphabetical order). Fifteen laboratories received the first panel and sixteen the second one.

Table 2. European quality control – influenza viruses and RSV (QCA)

List of participants

Results

The laboratories participating in the QCA were anonymised by randomly allocating codes L1 to L16.

Influenza antibody titration

Nine laboratories (60%) performed the titration by HI test; two used the complement fixation test and four did not test the sera.

Five out of nine laboratories reported expected titres. One laboratory detected HI titre for B in the negative sample. One laboratory reported low HI titres for H3 (serum 2) and B (serum 3). Another laboratory reported low B HI titres (sera 3 and 4). One laboratory reported B HI titres in a range of five fold dilutions, depending on the antigens used in the test that derived from the same prototype strain.

Detection of influenza and RSV viruses

Four participating laboratories failed to follow the QCA protocol: one looked for influenza viruses in the first panel and for RSV in the second panel, two laboratories did not normally perform RSV testing and one NRC only received the second panel. Taking into account these features, each laboratory’s global score has been calculated on the number of samples fully tested. A total of 20 samples were sent but we recorded 21 items because we introduced a mixed sample (influenza virus A + RSV: 2 items) in the second panel and we expected two results for the same sample. The number of tested items was 12 for one laboratory, 19 for two laboratories, and 21 for twelve laboratories. Doubtful results were considered to be negative.

Global laboratory score

Whatever the test used, the global laboratory scores were 100% for four laboratories, 95-90% for seven laboratories, 85% for one laboratory and less than 80% for four laboratories (table 3).

Table 3. Global score per laboratory

Twelve laboratories (75%) gave a false test result for at least one of the samples, ten laboratories (62.5%) gave a false result for influenza, and six laboratories (43%) gave a false result for RSV. Five laboratory reported one false result, four participants reported two errors, and three laboratories gave more than two false results.

Sensitivity

The false negative results mainly concerned the A(H1N1) strain which was not detected by six laboratories at the 1 TCID50 dose ; for influenza A(H3N2), two laboratories failed to detect 100 TCID50 , and for influenza B, two laboratories did not detect 1000 TCID50.

RSV was detected by 86% of the participants, and by 71% of the participants when the sample contained both influenza and RSV viruses.

Virus identification

Out of 16 laboratories, 14 (87%) performed influenza A virus subtyping, and five fully achieved it (31%). The variant characterisation was carried out by five laboratories and three correctly identified all samples. Five (36%) of the laboratories subtyped the RSV sample as type A.

Genome detection

RT-PCR and in-house reagents were used by nine laboratories for influenza virus diagnosis, and by six laboratories for RSV. One laboratory reported a false influenza B diagnosis on a sample containing uninfected MDCK cells. The inter-laboratory influenza A sensitivity varied from 1 to 10 000 TCID50, and from 1 000 to 10 000 TCID50 for influenza B. Two reference laboratories detected 1 TCID50 influenza A virus on cell culture while RT-PCR was negative. All six laboratories using RT-PCR for RSV diagnosis detected the virus in the two samples. 

Discussion

The NRC in Lyons organized the QCA for 16 laboratories in Europe. Each parti-cipating NRC individually received the decoding of the samples after all the QCA results arrived in Lyons. The NRCs performed the laboratory tests using their own techniques and reagents.

The sensitivity of the test results varied widely among the participating laboratories: false negative results were reported at least once for the A(H3N2), A(H1N1) and B samples containing 100 to 1 000 TCID50.

For the A(H1N1) sample containing 1 TCID50, the concentration was too low to be regularly detected on cell culture, but should be detected by RT-PCR.

The antigenic detection techniques are known to be of low sensitivity, and they have to be associated with other techniques such as the growth of the virus in cell cultures. As recommended by a number of studies (2-4), all the participants isolated viruses on MDCK cell culture. The performance of this technique greatly varied from one laboratory to another. To improve virus growth, it is necessary to check the sensitivity of the cell line, the quality of the medium, the techniques for sample inoculation, and virus detection.

The panels in the QCA were prepared with cell adapted influenza viruses, which could be regularly isolated within four days of incubation at 33°C. The positive cell culture could be tested either for NP nucleoprotein antigen in ELISA or for HA hemagglutinin using chicken, human or guinea pig red blood cells. It is the duty of NRC to isolate the viruses for further antigenic characterisation which might be performed by HI test using post infection ferret antisera. This identification allows a comparison of the circulating viruses with the current vaccine strains. Six laboratories also subtyped the genome by RT-PCR: both antigenic and genetic subtyping were in agreement. All NRC sent characterised isolates to the international reference centre which will confirm the identification and select variants as possible candidates for the vaccine of the following winter season.

RT-PCR techniques and in-house reagents gave specific results: one false positive out of 189 tests. The introduction of high virus concentrations in the quality control showed that laboratories correctly controlled the risk of contamination. The sensitivity of the RT-PCR varied widely (40 to 100% for influenza, 71 to 86% for RSV) depending on the laboratory, but within the same laboratory it was identical to the sensitivity of cell culture as previously reported in the literature (5, 6).

We were surprised that only 60% of the NRC performed the HI test for evaluating the immune status of vaccinees. The HI titres also varied greatly from one laboratory to another, mainly on the B HI titres. It is most probable that at least one laboratory used detergent treated influenza B antigen, but did not report this. To improve the serological test it would be necessary to organise a specific QCA with a large panel of various coded samples and to investigate the procedures in each laboratory.

To establish the score that the assessment considered to be acceptable for a NRC, we considered that the sample with A(H1N1) at 1 TCID50 was weakly positive, and could be excluded from the score. Since there are 20 items left, one error would reduce the score by 5%. Altogether a score of 90% or more was considered to be good. Eleven out of 16 NRC laboratories (69%) obtained a score of 90% or more. The results of this study did not undermine the detection or description of an influenza epidemic, or the characterisation of the virus.

Following this first QCA, it appeared necessary to perform other quality controls, not only for NRCs which had to improve their practices, but also to improve the sensitivity and the rapidity in the detection of new variants. Within the framework of the global pandemic preparedness programme, the flu survey requires an early warning (alert), which is the duty of the NRCs.

The Lyons NRC will organise the next QCA on influenza and RS virus detections and identifications during the winter of 2002-03. This assessment will be proposed to the same national reference laboratories and to the new members of EISS (26 laboratories in 20 countries are expected to participate in this assessment). 

Acknowledgements :

The EISS QCA was partly funded by the European Commission. We would like to thank John Paget for reviewing the article, and all of the national reference centres who participated in the QCA.