Algorithms designed to identify influenza based upon clinical parameters show widely variable sensitivity and predictive values, since other respiratory viruses, such as adenovirus, parainfluenza virus, respiratory syncytial virus, rhinovirus, and even bacterial infections, may result in influenza-like symptoms (3
). The diagnosis of influenza is warranted for the administration of appropriate antiviral therapy, which has been shown to reduce secondary complications, including bacterial pneumonia and otitis media associated with influenza (2
). In addition, studies have shown that the diagnosis of influenza reduces inappropriate antibiotic use in several settings, including patients presenting to emergency departments (8
) and high-risk hospitalized patients (9
). Rapid antigen testing (RAT) for influenza and other respiratory viruses supports rapid clinical decisions made at or near the patient's bedside. RAT has specifically been shown to reduce unnecessary antibiotic usage (8
), and its use helps to effectively manage hospital bed utilization (1
). Thus, the rapid and accurate detection of influenza and differentiation from other etiologic agents of influenza-like illnesses are useful for several aspects of patient management (7
Currently, there are several rapid tests available in the marketplace that laboratorians and physicians alike may select for the diagnosis of influenza. The most popular formats for RATs are self-contained strips or cartridges that detect influenza antigens by immunochromatography (IC). The advantages of this format are that results are usually available rapidly (within 10 to 30 min) and that single specimens from individual patients can be analyzed without the need for test batching or extensive quality control. Indeed, many assays have been waived by the Clinical Laboratory Improvement Amendments for use in physician's offices where end users have minimal laboratory training. However, one significant disadvantage for most IC assays is that they are subjective, and unskilled readers may have difficulty interpreting weakly positive results.
The performance characteristics of RATs can vary significantly based on many factors, including the type of specimen received for analysis, the time of collection during the course of the illness, the prevalence of influenza in the population, and the gold standard comparator. Several studies have compared the sensitivity, specificity, and negative and positive predictive values of commercially available RATs (5
). In general, most assays are very specific for influenza A or B but lack sensitivity compared to cell culture.
The 3M rapid detection influenza A+B (3M flu) assay is a new IC assay designed to detect and differentiate between influenza virus A and B nucleoprotein in nasal swabs, aspirates, and nasopharyngeal swab specimens. The results of the assay are read and interpreted by the RAMP automated fluorescence reader, manufactured by Response Biomedical, Inc. (Burnaby, British Columbia, Canada). This assay can be completed within 15 min, and the formation of influenza virus antibody and antigen complexes is heralded by the generation of a fluorescent signal which is detected by the reader. Besides the objective readout, another significant advantage of this assay is the ability to connect the RAMP reader to laboratory information systems to facilitate accurate and rapid reporting of results. To date, no study has examined the performance characteristics of this assay in either an analytical or clinical setting.
In this report, we studied the clinical and analytical performance characteristics of the 3M rapid detection influenza A+B assay by comparing it to direct fluorescent antibody (DFA) detection, cell culture, and several other RATs. Influenza viruses were obtained from historical stocks collected, propagated, and subtyped using WHO reagents at the Centers for Disease Control and Prevention (Table ). Each virus was grown in primary rhesus monkey kidney cells and titrated using serial twofold dilutions, and endpoints were calculated by the Reed-Muench method (15
). The 50% tissue culture infective doses of the viruses ranged from 32,768 to 131,072. Twofold dilutions (ranging from 1:32 to 1:2,048) of each virus were made in phosphate-buffered saline and subsequently used to compare the 3M flu assay against the BinaxNOW flu (Inverness), QuickVue flu (Quidel), and Directigen EZ (BD) assays. Tests were performed in duplicate according to the manufacturer's directions. Subjective results were read by two independent readers (S. E. Dale and C. Mayer), and the 3M flu assay was adjudicated based on lot-specific, manufacturer-defined cutoffs. Results are reported as the reciprocal of the highest dilution for which at least one of the two replicates was positive (Table ). In most cases, the 3M flu assay was able to detect the entire range of dilutions. For influenza virus A strains, the 3M flu assay was more sensitive than the BinaxNOW flu assay. Compared to the QuickVue flu assay, the 3M flu assay was equally sensitive for four of five strains and equally sensitive to the Directigen EZ flu assay for one of five strains. For influenza virus B strains, the 3M flu assay was uniformly more sensitive than all of the other assays. The relative titers of the influenza virus A or B strains did not predict the analytical sensitivity of any of the rapid antigen assays, which is not surprising, given that viral infectivity is likely not related to the amount of antigen present for detection.
Stock influenza viruses used to demonstrate the analytical sensitivity of the influenza virus rapid antigen assays
To analyze the clinical sensitivity of the 3M flu assay, 249 specimens (nasal swabs or nasopharyngeal swabs) for respiratory virus culture were prospectively collected, and each specimen was placed in 3 ml of M5 viral transport medium (Remel) and submitted to the Clinical Virology Laboratory at the University of Rochester Medical Center between January and April 2007. The majority of specimens (80%) were labeled as nasopharyngeal swabs. After collection of the first 50 specimens and to enrich for positive specimens, the study was suspended until the overall number of positive influenza cultures reached ~10% in the clinical laboratory. The majority of specimens (212/249 [85%]) were from adult patients (≥18 years old). The results of the prospective clinical study are shown in Table .
Number of individual specimens positive for influenza A virus by each assay tested in the prospective clinical study
Specimens were inoculated (0.2 ml) onto primary rhesus monkey kidney cell monolayers and incubated for 72 h at 33°C on a rolling drum. The remaining quantity of specimen was used for the 3M flu, BinaxNOW flu, and QuickVue flu assays, for DFA detection of influenza virus A and B, and frozen at −70°C for PCR analysis, if required. For DFA testing, primary specimens (1 ml) were centrifuged, and the cell pellet was used for detecting influenza virus A or B with the Light Diagnostics influenza A+B DFA kit, according to the manufacturer's recommendations. Specimens were scored as positive or negative for influenza A or B only if >20 epithelial cells were present on the slide. The 3M flu and BinaxNOW flu assays were performed simultaneously on specimens within 24 h of collection. The QuickVue flu assay was performed upon receipt of the specimen (also within 24 h of collection) and scored by the clinical laboratory staff of the University of Rochester Medical Center. All assays were performed in accordance with the package insert for each product.
The overall positive rate for influenza virus A and B was 16% (40/249) and 5% (13/249), respectively, in cell culture. The overall performance of each assay for influenza virus A is shown in Table . The 3M flu assay and the QuickVue flu assay showed similar sensitivities for influenza virus A; however, four specimens were falsely positive for influenza virus A by the 3M flu assay as determined by both cell culture and PCR detection (data not shown). The 3M flu assay was not repeated on these specimens.
Overall clinical performance of the influenza virus A rapid antigen tests
Overall, we isolated only 13 influenza virus B isolates from cell culture. Of these, the 3M flu assay detected only four isolates (sensitivity of 31%), whereas the BinaxNOW flu and QuickVue assays each detected two isolates (sensitivity of 15%). Due to the low number of influenza virus B isolates recovered in the clinical study, we retrieved archived specimens from the University of Rochester Medical Center Clinical Virology Laboratory collection. These specimens were previously shown to be both antigen positive (QuickVue or Directigen) and culture positive during the 2004 to 2006 respiratory virus seasons. All of these archived specimens (n = 30) were also positive by the 3M flu assay.
The sensitivity of the 3M flu assay was further investigated by analyzing lot-specific cutoff values for culture-positive specimens compared to the measured relative fluorescence for each specimen. As shown in Fig. , the distribution of fluorescence units for specimens that were positive for influenza virus A or B ranged over 2 orders of magnitude. Moreover, influenza A specimens that were falsely negative by the 3M flu assay all clustered just below the positive cutoff value. This suggested that the level of antigen in these specimens was not sufficient to be detected by the assay. In the commercially available 3M flu assay, results are reported as positive or negative for influenza virus A or B, and the user will not have access to the numerical value assigned for each specimen. Therefore, negative specimens may be verified by virus culture or another detection method, if required.
FIG. 1. Analysis of the relative fluorescence units calculated by the RAMP reader for positive tissue culture specimens of influenza virus A or B. The horizontal lines depict the lot-specific cutoff values, and the asterisks depict the mean relative fluorescence (more ...)
The specificity for each rapid antigen assay for both influenza virus A and B ranged from 98 to 100% using fresh clinical specimens. One sample gave a positive result for both influenza virus A and B using the QuickVue assay but was negative by culture, DFA, the 3M flu assay, BinaxNOW flu assay, and PCR analysis for influenza virus A and B (data not shown). For the 3M flu assay, seven samples could not be evaluated due to a cartridge error resulting in high background. The DFA technique was not interpretable for 86 specimens due to the low number of epithelial cells recovered. Specimens that could not be adequately evaluated were not included in the overall assay performance calculations shown in Table . To further assess specificity, we performed the 3M flu assay on 30 primary specimens that were negative by both culture and rapid influenza virus antigen testing from frozen archives (collected during the 2004 to 2006 respiratory virus seasons), and all specimens tested negative for influenza virus A and B.
All of the influenza virus RATs that we tested demanded much less technical expertise than either cell culture or DFA detection; however, since the sensitivity of each rapid assay was inferior compared to the sensitivity of cell culture, confirmatory testing of negative antigen assays may be warranted. Taken together, we present data that compares the clinical and analytical sensitivity of the 3M flu, QuickVue flu, BinaxNOW flu, and Directigen EZ flu assays for the detection and differentiation of influenza virus A and B from nasal and nasopharyngeal swabs. We report that the 3M flu assay, which has recently received FDA 510(k) approval, is substantially equivalent to other rapid influenza virus antigen diagnostic kits that are currently available in the marketplace. The 3M flu assay is an attractive platform to consider for influenza diagnosis, since it combines the ease of use of manual RATs but does not require subjective interpretation. Moreover, the RAMP reader allows connectivity with laboratory information systems to streamline and verify the accuracy of reporting.
(This study was presented in part at the 45th Annual Meeting of the Infectious Diseases Society of America, San Diego, CA, 4 to 7 October 2007.)