Testing of Diagnostic Methods for Detection of Influenza Virus for Optimal Performance in the Context of an Influenza Surveillance Network

Influenza disease is subjected to surveillance worldwide by national networks that predict the epidemic threshold by reporting clinical and virological data (6). Epidemiological and virological surveillance is carried out by sentinel physicians and virologists, respectively. The clinical criteria for sampling and laboratory testing of influenza viruses vary among different countries. Indeed, a laboratory-confirmed influenza case is defined for each country by the national laboratory network, within the national influenza surveillance network. In Spain, the national laboratory network is currently reporting a laboratory-confirmed influenza case when the virus is isolated in cell culture. Although reverse transcription (RT)-PCR assays are more sensitive than cell culture (5), only those cases detected by virus isolation are being reported as positive for influenza in the surveillance network (7). Many regional laboratories within the national network use the rapid shell-vial assay (SV) (viral culture by SV) because it shortens the time to obtain positive cases. Screening of influenza viruses from the SV is usually done by direct fluorescence assay (DFA), which also allows the differentiation of influenza viruses A and B (10). Subsequently, antigenic characterization by serological or molecular methods is conducted on the isolates.

Our laboratory carries out regional influenza surveillance in Andalusia (in the south of Spain) as part of the national influenza laboratory network. During a two-year period, from October 2004 to May 2005 and October 2005 to May 2006, we evaluated several methods for influenza virus detection to determine which was the most reliable and sensitive. For this purpose, we compared three diagnostic methods: multiplex RT-nested-PCR directly on the specimen for the detection and differentiation of influenza A subtypes H1 and H3 and influenza B viruses (DD) (11), SV followed by DFA (SV-DFA), and SV followed by RT-PCR (11) on the cell culture supernatant (SV-PCR).

The clinical specimens were throat and nasal swabs from outpatients (with clinical evidence of acute respiratory infection over a duration of ≤72 h and a body temperature of ≥38°C) collected by the sentinel physicians. Specimens were sent to the laboratory within the first 6 h following their collection, at 4°C in minimal essential medium supplemented with 1% bovine serum albumin. A 140-μl aliquot of the specimen was used for RNA extraction with a QIAmp viral RNA kit (QIAGEN, Hilden, Germany) for use in DD. Other 200-μl aliquots were inoculated into MDCK cells (Vircell SL, Granada, Spain) for SV and traditional tube culture (10). The remaining specimen was frozen at −80°C. After a 48-h incubation, cell monolayers from the SV were subjected to DFA with monoclonal antibodies against influenza A and B viruses (Dako Diagnostics Ltd., Cambridgeshire, United Kingdom). The SV supernatant was used for RNA extraction and RT-PCR as described above. The traditional tube culture was examined daily during 14 days to observe the appearance of cytopathic effect. Before being discarded as negative, tubes without cytopathic effect were subjected to a hemagglutination test as previously described (9). All positive results by RT-PCR were confirmed by repeating the test with a frozen aliquot of the specimen or cell culture supernatant. The national reference laboratory (National Center for Microbiology, Majadahonda, Madrid, Spain) carried out further characterization of the strains.

The data were statistically analyzed with SPSS 13.0.1 software (SPSS Inc., Chicago, IL). Univariate analysis was conducted on the results by the χ² test. A P value of <0.05 was considered significant.

A total of 565 specimens were analyzed within the two periods, 299 in the 2004-to-2005 period and 266 in the 2005-to-2006 period. Influenza viruses were detected in 152 specimens (26.9%) by any of the three methods. The distribution of the results by method and influenza virus type and subtype is shown in Table 1. The most sensitive method was SV-PCR (95.4%). The two methods that allowed recovery of the virus strain, i.e., SV-DFA and SV-PCR, were compared. Whereas influenza B and H1 were similarly detected by both methods, influenza H3 was detected in 64.7% and 95.3% of samples by SV-DFA and SV-PCR, respectively (P, 0.016). Positive results were concordant between DD and SV-DFA in 70% of cases, between DD and SV-PCR in 82.8% of cases, and between SV-DFA and SV-PCR in 80.2% of cases (Table 2). By combining both PCR methods, DD and SV-PCR, only one positive sample gave a false-negative result.

Direct detection, typing, and/or subtyping can be rapidly carried out by several methods, such as nucleic acid techniques, immunofluorescence assay, or enzyme-linked immunosorbent assay (3, 8, 11). This step can be carried out in each regional laboratory within the national surveillance network. Subsequently, laboratories belonging to the Community Network of Reference Laboratories in Europe must be able to replicate growth in cell lines for isolation of influenza viruses. A selection of the virus isolates is sent to the WHO Collaborating Centre for Reference and Research on Influenza in London for complete characterization (5). Thus, even though multiplex RT-PCR and antigen detection methods can rapidly give a result which eases the management of patients with influenza (1, 5), virus isolation is required in the context of the annual influenza surveillance.

In the 2005 to 2006 season, our regional network covered a population of 77,000 individuals, which is 1% of the surveillance coverage (7). We conduct both the detection and isolation, and only the viral isolates are sent to the national reference laboratory. Any delay in sample processing negatively influences virus isolation. Moreover, nasal/throat swabs are not the optimal specimens for influenza virus recovery.

Previous reports have used this diagnostic tool, i.e., integrated cell culture-PCR, for improving the recovery of other viruses (2, 4). But few data are available for influenza virus detection in the context of surveillance systems. We have found that the most reliable method for influenza virus detection is SV-PCR. This method recovered 95.4% of the total positive results by any of the three methods at 48 h of sample processing. Only one sample (0.6%) was positive by SV-DFA and negative by SV-PCR. The remaining samples negative by SV-PCR were positive only by the DD method. This may be a consequence of delayed sending or processing and/or a suboptimal sampling, which lead to nonviable viruses, in which case only viral RNA can be detected in the specimen. SV-DFA is the method currently recommended for the rapid diagnosis of influenza viruses in different influenza networks. However, our results show that this method may not be the most appropriate, especially when the circulating strain is the H3 subtype.

In conclusion, within an influenza surveillance network, the best diagnostic method would be the combination of rapid culture by SV with multiplex RT-PCR on the cell culture supernatant. It reaches the highest sensitivity, recovers the viral strain for further characterization, and is able to simultaneously carry out the detection, typing, and subtyping of influenza viruses.

• Copyright © 2007 American Society for Microbiology