This report describes the clinical evaluation of a novel fluorescent immunoassay (FIA), Sofia Influenza A+B FIA (Quidel, San Diego, CA), for the rapid detection and differentiation of influenza A and B viruses. A total of 2,047 subjects provided nasal swabs and nasopharyngeal swabs or aspirates. The overall sensitivity and specificity for influenza A virus vs virus culture were 94% and 95%, respectively, and for influenza B virus were 89% and 96%, respectively. Fourteen hundred and sixty-one specimens were available for testing with reverse transcriptase–polymerase chain reaction (RT-PCR). The sensitivity of the Sofia Influenza A+B FIA for detecting influenza A and B viruses compared with the RT-PCR method was 78% and 86%, respectively. A high percentage of the positive specimens had low cycle threshold values, and almost all of these were positive with the Sofia test. This high level of sensitivity demonstrates that the Sofia influenza A+B FIA could improve the usefulness of rapid influenza virus testing.
Open in new tab
Materials and Methods
The accurate detection of influenza virus in respiratory tract specimens is an important component in the treatment of patients with influenza by health care providers and the control of influenza epidemics by public health officials. Many types of in vitro diagnostic tests are offered by laboratories to aid in the diagnosis of an influenza virus infection.1 Traditional culture methods are reasonably accurate but labor intensive and slow. Direct fluorescent antigen (DFA) methods are more rapid but also require a skilled technologist. Molecular diagnostic tests based on reverse transcriptase–polymerase chain reaction (RT-PCR) methods are highly sensitive but are generally expensive, often require highly skilled laboratory personnel, and have a rapid turnaround time of 2 hours or more.2 Rapid influenza antigen detection tests (RIDTs), such as lateral flow assays, are simple to perform and provide results in minutes but have demonstrated lower than desired sensitivity in many cases.3
The sensitivity problem of RIDTs was reported to be more pronounced with the 2009 H1N1 pandemic strain.4–6 Many studies and 2 recent meta-analyses reported a wide range of sensitivities for various RIDTs used in different clinical circumstances.3,7,8 Nevertheless, RIDTs remain a useful clinical tool, especially because the diagnosis of influenza cannot be made on the basis of clinical criteria alone.9 RIDTs have been shown to improve decision making in the emergency department.10 In particular, studies have also shown that RIDTs improve patient care by reducing the number of radiographs, unnecessary antibiotic usage, and emergency department length of stay.11,12 It also has been demonstrated that combining clinical findings with rapid tests can improve appropriate use of antiviral therapy.13 Effective and appropriate antiviral therapy depends on obtaining accurate results in a timely manner because antiviral therapy is most effective within 48 hours of onset of symptoms.14–16 One report emphasized the importance of RIDTs for travelers with influenza-like illness (ILI), and the value of RIDTs has been shown to be particularly high in resource-limited settings.17–19 There is a need, therefore, for a rapid influenza test that has greater sensitivity and does not rely on subjective interpretation.
Recently a novel fluorescent immunoassay (FIA) (Sofia Influenza A+B FIA, Quidel, San Diego, CA) was developed to detect and differentiate influenza A and B in patient specimens. The Sofia Influenza A+B FIA is a commercially available in vitro diagnostic device that detects viral nucleoprotein antigens. Although capable of detecting a broad range of influenza A virus subtypes, it was not designed to specifically identify the virus subtype. The test is a fluorescence-based lateral flow immunoassay in which results are analyzed objectively by a compact instrument (Sofia Analyzer) rather than visually. Instead of antibody-tagged colored microparticles, the Sofia technology uses a unique polystyrene microbead that has been impregnated with a chelate of europium. The europium compound yields a very efficient conversion of UV energy from 365 nm to a wavelength of 618 nm. This large Stokes shift avoids problems associated with many fluorescent compounds that may be present in the test samples. The use of europium-based immunofluorescence also provides for high analytic sensitivity.
In this report we describe a multicenter prospective clinical trial in which the Sofia Influenza A+B FIA test was compared with viral culture and RT-PCR for detecting influenza A and B viruses in respiratory specimens from patients with ILI.
This study was conducted at 16 distinct sites in the United States and involved a range of operator skill levels Table 1. All procedures were in accord with the ethical standards of the institution in which the experiments were performed, including obtaining approval from a local institutional review board (IRB) or a third-party IRB.
Clinical Study Subjects
Patients enrolled in the study had 1 or more of the following signs and symptoms: fever (temperature ≥38°C either at the time of visit or onset within the previous 2 days), nasal congestion, rhinorrhea, sore throat, cough, headache, myalgia, or malaise.
Open in new tab
All enrolled subjects or a parent or guardian gave their informed consent before enrollment.
Open in new tab
R-Mix Too Cluster Plates (DHI) were inoculated with a 0.2-mL aliquot of the specimen and incubated at 35°C to for 40 to 48 hours. The inoculated cells were 37°C, 5% CO2 recovered from tissue culture and tested for influenza A and B with DFA staining using D3 Duet DFA Influenza A/Respiratory Virus ID Kit (DHI) for identifying influenza A and the D3 UItra DFA Respiratory Virus ID Kit (DHI) for identifying influenza B.
Specimens were transported daily from Twelve Corners Pediatrics, Rochester, NY, to the nearby virus laboratory at the University of Rochester Medical Center, Rochester, NY. Madin Darby Canine Kidney cell culture tubes (DHI) were inoculated with a 0.26-mL aliquot of the specimen and incubated at 34°C. The inoculated cells were recovered from tissue culture on day 7 and tested for influenza A and B with the Light Diagnostics Influenza A and B DFA Kit (EMD Millipore, Billerica, MA).
R-Mix Too Shell Vials (DHI) were inoculated with a 0.25- to 0.4-mL aliquot of the specimen and incubated at 35°C to 37°C (non-CO2) for 48 to 72 hours. The inoculated cells were recovered from tissue culture and tested for influenza A and B using D3 FastPoint L-DFA Influenza A/Influenza B (DHI).
The Sofia analyzer contains a microprocessor-controlled optics unit that scans the length of the nitrocellulose test strip, thereby exposing the strip to UV light from a UV light-emitting diode. A positive result for either analyte is determined by the detection of a fluorescent signal at levels above a signal threshold set on scanning a “negative-control” line. This process is controlled by a specific algorithm embedded in the Sofia analyzer. The test strip is processed in the Sofia analyzer, which automatically scans the test strip, collects and analyzes the fluorescence data, and then calculates and reports the result in about 1 minute. There is no visual signal, and the laboratory technologist must use the Sofia analyzer to obtain a test result. These features eliminate the subjectivity involved in visually read lateral flow assays. The test result can be printed directly from the instrument and/or transmitted to a laboratory information system.
The Sofia Influenza A+B FIA testing was performed in accordance with the manufacturer’s recommendations. Study sites were instructed to follow the Sofia Influenza A+B FIA package insert, and the Sofia analyzer was used in compliance with the manufacturer’s user manual (Quidel). After the sample was prepared and added to the cassette, it was incubated for 15 minutes, either on the bench top or in the Sofia analyzer, before being analyzed.
Positive and negative external controls were provided with the Sofia FIA kit. These noninfectious swab specimens were tested as instructed in the package insert. Positive and negative external controls were run at each site on each day of use of the Sofia Influenza A+B FIA during the clinical study. The negative control contains a lysate of group C Streptococcus. The positive control contains recombinant influenza nucleoprotein and uniformly gives a positive signal if the analyzer is functioning properly.
Influenza A and B virus RNA was extracted from the specimens using the EasyMag (bioMérieux, Lyon, France). Viral RNA was detected with RT-PCR with the Quidel Molecular Influenza A+B kit (Quidel) and the ABI 7500 thermocycler instrument (Life Technologies, Carlsbad, CA).
Detection of Influenza Virus Using Sofia Influenza A+B FIA
A total of 2,152 subjects from 16 sites (Table 1) were enrolled in this study. Because of technical problems, deviation from the protocol, or lack of culture results, 105 specimens were excluded from the study. Data from 2,047 subjects were included in the analyses. Six hundred sixty-five subjects provided an NS; 733 provided an NP swab, and 649 provided an NPA/wash specimen. Table 2 shows the gender and age distribution of the 2,047 subjects enrolled in the study.
External Positive and Negative Controls
Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for Sofia Influenza A+B FIA compared with culture were calculated for each virus and each sample type Table 3 and Table 4. The overall sensitivity of the Sofia Influenza A+B FIA for detecting influenza A virus vs culture was 94% (292/310). The overall specificity was 95% (1,650/1,737). The sensitivity for each specimen type was 90%, 97%, and 99% for NS, NP swabs, or NPA, respectively (Table 3). The specificity ranged from 95% to 96% across all sample types.
Open in new tab
The overall sensitivity of the Sofia Influenza A+B FIA for detecting influenza B virus compared with culture was 90% (247/274). The overall specificity was 96% (1,707/1,771). The sensitivity for each type of specimen was 89%, 90%, and 88% when testing NS, NP swabs, or NPA, respectively (Table 4). The specificity ranged from 96% to 97% for all sample types.
Open in new tab
The overall sensitivity of the Sofia Influenza A+B FIA for detecting influenza A virus compared with RT-PCR was 78% (260/333; confidence interval, 73.6–82.5) Table 5. The overall specificity was 100%. The sensitivity for each specimen type was 85%, 69%, and 71% for NS, NP swabs, or NPA, respectively (data not shown). The NPV and PPV were 93.8% and 97.5%, respectively.
Detection of Influenza Virus RNA With RT-PCR
The sensitivity of Sofia Influenza A+B FIA compared with RT-PCR for detecting influenza A and B viruses was dependent on the level of RNA in the sample as assessed by the PCR cycle threshold value Figure 1. Seventy-seven (77%) percent of the patients who were PCR positive for influenza A had cycle threshold values less than or equal to 30, and 96% of these were detected as positive on the Sofia Influenza A+B FIA. Similarly, about 78% of the patients who were PCR positive for influenza B had cycle threshold values less than or equal to 30, and 99% of these tested positive on the Sofia assay.
In this study, the Sofia Influenza A+B FIA, performed by hospital and physician office personnel using NS, NP swabs, or NPA specimens, yielded good sensitivity and specificity in comparison with cell culture and RT-PCR. These performance results, combined with some of the practical advantages of this technology, demonstrate that the Sofia Influenza A+B FIA could be significantly useful in many health care delivery settings. This test has all the advantages of traditional RIDTs but offers an objective readout. In addition, the sensitivity seen in this study is higher than all reports on RIDTs.3
Although cell culture was traditionally considered the gold standard for sensitivity assessments, recent molecular methods have been shown to be the most sensitive methods for detecting these viruses. One problem with using RT-PCR as a comparator is that the clinical significance of low levels of viral RNA remains controversial. The ability to detect viral RNA continues after symptoms have waned and certainly after antiviral therapy has been effective.16 In the present prospective study, most of the samples yielded high levels of viral RNA for which the sensitivity of the Sofia FIA was almost 90%. This suggests that the sensitivity is highest in circumstances in which the results are clinically most useful to support effective antiviral therapy.
This study focused primarily on children with ILI and included a comparatively low percentage of elderly patients. Further studies are needed to demonstrate the clinical performance of this test in older patients, especially because RIDTs have been reported to perform less effectively in hospitalized, elderly patients.20 Also, although analytical studies showed excellent performance for detecting numerous influenza A virus subtypes, viral subtyping was not performed in this study.
Comparison of Sofia Influenza A+B FIA With Virus Culture
The treatment of patients with ILI has been shown to be significantly benefited by the availability of a rapid result of an influenza virus detection test. The ideal RIDT should be rapid, sensitive, specific, simple to perform, economically feasible, and, perhaps most importantly, provide actionable results for the physician. When used within 48 hours of onset of symptoms, the Sofia FIA will yield highly accurate, objective results within 15 minutes. Indeed, of the patients with PCR-confirmed influenza A virus infection in this clinical study, 80% had relatively high viral loads (cycle threshold values less than 30); of these, 96% tested positive with Sofia. The Sofia assay system invites more clinical studies to firmly establish the benefits it can bring to patient care, hospital bed management, and reduction of nosocomial spread of influenza infections. The Sofia FIA’s clinical performance in comparison with culture and PCR was remarkably good and assures the delivery of actionable results to physicians within minutes of their patients’ visits. RIDTs generally have had very good specificity, which provides clinicians the confidence to act appropriately when a positive result is obtained. In this study, the higher sensitivity achieved with Sofia and the high NPV demonstrates that Sofia can reduce reflex testing of negative samples and reduce the number of cases managed as presumptive influenza while clinicians wait for the results of confirmatory testing. Additional clinical studies are needed to more fully delineate the contribution that this test could make to improved treatment of patients with ILI in various health care settings.
We thank the health care providers at the trial sites, including the physicians, nurses, technologists, and laboratory supervisors.