A multiplex one-step real-time RT-PCR assay for influenza surveillance 17

For surveillance purposes real-time PCR assays for influenza viruses had to be adapted to the pandemic influenza A(H1N1)2009 strain. We combined published primers and probes for influenza A, influenza B and an internal amplification control with a detection system for influenza A(H1N1)2009 to set up a rapid, reliable, simple and cost-effective high-throughput multiplex one-step real-time RT-PCR. The workflow also includes automated sample preparation for high-throughput screening. The lower limit of detection of the multiplex assay was 3.5x10(2) RNA copies per PCR reaction. The diagnostic sensitivity of the multiplex assay was 87.7%, but increased to 99.4% for influenza-positive samples yielding C(t) values of less than 34 cycles in the respective diagnostic assay. High specificity was confirmed by sequencing and correct detection of 15 reference samples from two quality assurance studies. The multiplex PCR was introduced for surveillance of samples from a network of general practitioners and paediatricians in Bavaria, Germany during the influenza pandemic of 2009. Comparison with surveillance data from reported cases proved the reliability of the multiplex assay for influenza surveillance programmes.


Introduction
In April 2009, a novel influenza A(H1N1) virus emerged [1] that could not be detected by routine diagnostic assays for subtyping seasonal influenza A(H1N1) viruses.Therefore, accurate and reliable diagnostic tests for the new influenza A strain had to be established to screen patients with influenza-like illness (ILI) for the 2009 pandemic influenza virus [2][3][4][5][6][7][8][9].At the onset of the pandemic, public health control measures, namely the isolation of patients and suspected cases to limit the spread of the virus, were guided by the results of these tests [10].
In October 2009, mass vaccination programmes with different pandemic influenza vaccines were implemented globally.In Germany, about 6 million people were vaccinated from the end of October to the end of December 2009.At that stage of the pandemic the World Health Organization (WHO), the European Centre for Disease Prevention and Control (ECDC) and the Robert Koch Institute in Germany (RKI) recommended strengthening the influenza surveillance.This surveillance should persist throughout the whole year and include the new influenza strain as well as seasonal influenza strains, because co-circulation was reported and also expected in the future.At that time, no multiplex real-time RT-PCR assay was available for the simultaneous detection of seasonal influenza A, influenza B and pandemic influenza A(H1N1)2009 viruses.Published diagnostic assays focused more on subtyping of influenza viruses using microarrays and sequencing [11][12][13][14].However, these tests are not suitable for high-throughput routine diagnostic screening.
For large scale surveillance of ILI patients cost effective and time-saving methods for the detection of influenza viruses are needed.The multiplex real-time RT-PCR assay described here provides a diagnostic tool for the fast, simultaneous and reliable diagnosis of influenza A and B viruses with validated and well established real-time PCR protocols with minor modifications, and includes an officially recommended real-time PCR protocol for simultaneous subtyping of the pandemic influenza A(H1N1)2009 virus.

Specimen collection
For specificity and sensitivity testing as well as for the evaluation of the multiplex assay different panels of clinical samples and reference material were used in this study: The specificity of the PCR protocol for subtyping pandemic influenza A(H1N1)   The samples were provided for two external quality assurance studies (organised by INSTAND e.V., Germany in 2009/10.In addition, the oseltamivirresistant strain influenza A/Berlin/58/2008 (H1N1) was provided by the national reference centre for influenza at the RKI in Berlin.
The analytical sensitivity (limit of detection) of the multiplex real-time RT-PCR assay was determined using plaque-quantified influenza A/Hamburg/05/2009 (H1N1) virus with a concentration of 3.5x10 5 PFU/ml [15].A 10-fold dilution series of extracted RNA was generated from 3.5x10 5   we used clinical samples obtained from ILI patients during the influenza season 2008/09 and the 2009 influenza pandemic in Bavaria.ILI was defined by sudden onset with fever (>38.5°C),cough, sore throat and myalgia and/or headache.We had previously tested the samples with diagnostic real-time RT-PCR assays for seasonal influenza A and B [16,17] and for influenza A(H1N1)2009 [2,9].The panel consisted of 317 samples: 90 influenza-negative samples, 47 samples positive for seasonal influenza A(H3N2) and A(H1N1), 50 samples positive for influenza B viruses as well as 130 samples positive for influenza A(H1N1)2009 virus.Original specimens included nasopharyngeal and throat swabs in viral transport medium.After screening for influenza, the remaining RNA was stored at -80°C until testing with the multiplex assay.

Nucleic acid extraction
Viral nucleic acid was extracted using the QIAamp Virus Bio Robot 9604 kit (Qiagen) adapted to the robot Hamilton Microlab Star (Hamilton) for large numbers of samples or the Viral RNA mini kit (Qiagen) for small numbers of samples.From our routine diagnostic analyses we know that the extraction method has no influence on the results.

Internal amplification control
Commercially available heterologous in vitro-transcribed RNA (INTYPE IC-RNA Labordiagnostik, Leipzig, Germany) was used as PCR inhibition control.This in vitro transcript has proven its robustness in previous multiplex real-time RT-PCR assays [18].The stock solution (8x10 5 copies/µl) of the in vitro-transcribed RNA was stored at -80°C, and the working dilutions of 1x10 5 copies/µl were stored at -20°C.

Multiplex real-time RT-PCR assay
Different published primers and probes of real-time RT-PCR assays specific for influenza A, influenza B and pandemic influenza A(H1N1)2009 viruses were tested to determine whether they could be used together in a multiplex assay.We show here only those primer and probe sets that performed well when combined in preliminary tests.
In order to minimise the risk of PCR product contamination, we introduced one-step RT-PCR protocols using the commercially available QuantiTect Virus +ROX Vial kit (Qiagen) including QuantiTect Virus No Rox (NR) Mastermix and QuantiTect virus RT.For specific detection of influenza A, influenza A(H1N1)2009, influenza B and the internal amplification control (IAC), we used primers and probes of published or officially recommended real-time PCR systems (Table 1): a previously published real-time RT-PCR assay [16] for influenza A viruses targeting the matrix gene, with an optimised reverse primer (InfA M-124-mod) for reliable detection of pandemic influenza A(H1N1)2009 (recommended by the national reference centre for influenza at the RKI); an officially recommended real-time PCR system with primers and a TaqMan probe for the specific detection of influenza A(H1N1)2009 [9] targeting the HA gene, and a real-time RT-PCR assay for the detection of influenza B [17] targeting the matrix gene, with a slightly modified reverse primer that has been routinely applied for years for routine diagnosis in our laboratory.The detection system for the internal amplification control has been described previously for multiplex real-time PCR assays [18].
All primers and probes were synthesised by TIB Molbiol.For the three influenza single target real-time RT-PCR assays a 25 µl PCR reaction was prepared containing: 400 nM of each forward and reverse primer (see Table 1), 100 nM of TaqMan probe, 1x QuantiTect virus reverse transcription mix, 1x QuantiTect virus NR mastermix, 4U RNase inhibitor (Invitrogen) and 5 µl RNA extract.
For optimisation of the multiplex assay all primer concentrations were titrated from 100 to 500 nM and all probe concentrations from 100 to 300 nM.Fluorescence filter sets for 6-carboxyfluorescein (FAM), hexachloro-6-carboxy-fluorescein (HEX/VIC), Texas Red (TEX/ROX) and a cyanine dye (CY5) were used simultaneously.The influenza A-and B-specific probes were labelled with FAM and CY5, respectively.The probe specific for pandemic influenza A(H1N1)2009 was labelled with TEX.
The IAC probe was labelled with HEX.All four TaqMan probes were labelled with Black Berry Quencher (BBQ) as quencher dye.For the multiplex real-time PCR assay optimised probe concentrations were applied (see Table 1).
For single and multiplex real time PCR thermal cycling was performed on MX3000P and MX3005P real-time PCR instruments (Agilent Technologies) under the fol-

Efficiency of the multiplex assay
PCR efficiencies were determined for influenza A, influenza B and pandemic influenza A(H1N1)2009 in each single assay as well as the individual channels of the multiplex assay.PCR efficiency was calculated according to the PCR amplification formula E = 10 (1/slope) -1x100%, E being the PCR efficiency.

Optimisation of the multiplex assay
Primer titration from 100 to 500 nM as well as probe titration from 100 to 300 nM indicated an optimal primer concentration of 400 nM and an optimal probe concentration of 200 nM for all four assays in the multiplex real-time RT-PCR (see Table 1).Higher or lower concentrations did not alter the sensitivity of the multiplex assay significantly (results not shown).
The optimised multiplex real-time RT-PCR assay in a 25 µl PCR reaction volume was composed as follows: 400 nM of all primers and 200 nM of each of the four TaqMan probes, 1x QuantiTect virus RT mix, 1x QuantiTect virus NR mastermix, 4U RNase inhibitor, 0.25 µl IAC RNA (2.5x10 4 copies) and 5 µl RNA extract.Thermal cycling was performed on MX3000P and MX3005P under the same conditions as the individual single assays.The optimised multiplex real-time RT-PCR assay is shown in Figure 1 for 10-fold dilution series of viral RNA from 10 ng to 1 pg RNA.

Specificity of the multiplex assay
The specificity of the diagnostic assays for influenza A and influenza B has previously been tested and confirmed [16,17].Therefore it was not further tested during multiplex optimisation.We checked the specificity of the PCR for pandemic influenza A(H1N1)2009 virus that was unpublished at the time [9]

Analytical sensitivity of the multiplex assay
With plaque-quantified influenza A/Hamburg/05/2009 (H1N1) we found a linear dynamic range from 10 5 to 10 2 genome equivalents.The detection limit was below 3.5x10 2 PFU/ml for the matrix gene as well as the HA gene.Testing of each RNA concentration of the influenza A (Sydney/5/1997 (H3N2) and B (Brisbane/60/2008) reference material in triplicates yielded a sensitivity of 10 pg per PCR reaction for each detection system in the single assays as well as in the multiplex assay and detected RNA extracted from influenza-infected cell cultures (seasonal influenza A(H3N2) and B) and from egg cultures (influenza A(H1N1)2009) with equal sensitivity.

Efficiency of the multiplex assay
The  2. The PCR efficiency of each individual assay was determined as between 92% to 120% for the individual assays.The PCR efficiencies of the respective single assay were comparable to the PCR efficiency in the multiplex assay.The influenza B assay had a PCR efficiency of 92% in the single assay while in the multiplex assay the PCR efficiency was 120%, which was considered as acceptable for a screening assay.

Evaluation of the multiplex assay with samples of ILI patients
A total of 317 stored RNA samples from the respiratory tract of ILI patients that had previously been tested with diagnostic real-time RT-PCR assays, were retrospectively tested with the multiplex assay.The overall diagnostic sensitivity of the multiplex assay was 87.7%, specificity was 99.6% and positive (PPV) and negative predictive values (NPV) 99.5% and 90.6%, respectively, compared to the respective diagnostic assay.Ninety samples had been negative in all diagnostic assays.Of those 90, 89 were also negative when we tested them in the multiplex assay, but one sample yielded a positive result for pandemic influenza A(H1N1)2009 in the multiplex assay (C t value 35).The IAC was positive in all influenza-negative samples, indicating that failure to detect influenza virus was not due to inhibition.
Based on the detection rates of this evaluation we calculated that the multiplex assay would have correctly identified at least 1,238 of the 1,322 (93.6%) influenza A(H1N1)2009-positive samples (Figure 3), which were analysed at the Bavarian Health and Food Safety Authority between 27 April and 9 November 2009 using the diagnostic assays.The C t values were between 20 and 32 for 1,025 of these samples.
The The IAC was negative in five throat swabs which all tested negative for influenza viruses.After 10-fold dilution of the sample, the IAC was positive in all five samples.The negative results of three of these samples were confirmed negative when retested in dilution in the multiplex assay, while two were positive for pandemic influenza A(H1N1)2009.

Discussion
We report on a multiplex one-step real-time RT-PCR assay for the simultaneous detection of seasonal influenza A and B as well as influenza A(H1N1)2009 viruses.
The assay was optimised for multiplex real-time PCR from published, validated and well established PCR protocols with minor modifications.The multiplex assay proved to be as specific as the respective diagnostic PCR assay.Only one sample tested negative in the diagnostic assays but positive for influenza A(H1N1)2009 in the multiplex assay in two replicates.We ran out of patient material and could not retest the sample with the diagnostic assay.As we detected only a low positive signal, neither a false positive result of the multiplex assay due to contamination, nor a false negative result of the diagnostic assay could be ruled out.A PCR inhibition control was successfully integrated into the assay for accurate interpretation of negative results.Interestingly, two samples positive for pandemic influenza A(H1N1)2009 would have been missed without the IAC.Dilution of the RNA before PCR successfully abolished the inhibitory effect.As we used the in vitro-transcribed RNA as an amplification control we could not control for inhibitory effects due to the extraction protocol.
We consider our multiplex assay that has shown its functionality in a high number of patient samples a useful tool for general public health laboratories.In contrast to the evaluation of other published assays [5,7,19], we have tested our multiplex assay on a very large number of clinical samples, including a high number of positive samples.In our analysis of patient samples, the diagnostic sensitivity of the multiplex PCR was slightly lower than that of the respective diagnostic assays, even if RNA dilution series of reference material showed equal sensitivity when determining the detection limit of the multiplex assay in comparison to the single assays.This might be explained by degradation due to storage of weakly positive patient RNA samples for up to one year, whereas dilution series were performed with freshly isolated RNA from reference material for the single as well as the multiplex assays.The overall sensitivity was 87.7%, but was 99.4% for samples with moderate and high viral loads (C t value>34).In a situation with population-wide screening in which patients with acute ILI yielding high viral loads are tested, we consider the slightly lower sensitivity acceptable.The assay has been validated for routine diagnosis of influenza and is used for large scale surveillance of influenza activity.While the pandemic subtype was reliably recognised during the 2009 pandemic, specificity and sensitivity of the multiplex assay was also shown for seasonal, avian and an oseltamivir-resistant virus.The assay is used to monitor influenza viruses throughout the whole year.By introducing the multiplex assay we were able to lower costs by saving reagents and working time.Furthermore we reduced sample turnaround time in comparison to the diagnostic PCR assays.
Diagnostic tools for surveillance are applied for the general identification of influenza viruses.Although mutation of the pandemic influenza A(H1N1)2009 virus was rare in the 2009 pandemic [20], we also addressed this possibility by including conserved regions (matrix genes) as PCR target.Our multiplex assay is capable to both identify the circulating pandemic strain (HA gene) and screen for other influenza A and B viruses (matrix genes).These should be further subtyped to confirm other seasonal influenza A subtypes or to detect changes in the circulating strain.
Chen et al. [21] also published a multiplex real-time RT-PCR assay for the simultaneous detection and subtyping of influenza viruses including the pandemic influenza AH1N1(2009), that has been evaluated on a high number of patient samples.Compared with our one-step real-time RT-PCR assay, this assay is based on a two-step real-time RT-PCR.
The 2009 pandemic is a reminder for public health laboratories to monitor influenza activity not only during the season of influenza circulation, but during the whole year.Our assay proved to be a convenient, rapid, reliable and cost effective way to meet this requirement.

Figure 1
Figure 1 Typical RT-PCR amplification curves for influenza viruses A: Seasonal influenza A virus detected in the FAM channel B: Pandemic influenza A(H1N1)2009 virus detected in the TEX channel

Figure 3 C
Figure 3 C t values of samples positive for pandemic influenza A(H1N1)2009 in the diagnostic PCR, Bavaria, 27 April− 9 November 2009 (n=1,322) real-time PCR runs of the sensitivity tests for influenza A (Sydney/5/1997 (H3N2), influenza A/Hamburg/05/2009 (H1N1) and influenza B (Brisbane/60/2008) were applied for the determination of the PCR efficiencies in the multiplex real-time PCR compared to the individual single real time PCR assays.The PCR efficiencies of the single real-time PCR assays in comparison to the individual channels of the multiplex PCR assay are shown in Table

Table 1
Primers and probes used in the multiplex one-step real-time RT-PCR assay for the detection of different influenza virus strains

Table 3
Sensitivity, specificity, positive predictive value and negative predictive value of the multiplex assay NPV: negative predictive value; PPV: positive predictive value.Numbers are also shown for samples with Ct-values below and above 34.