Introduction

The accurate diagnosis of Legionella pneumonia is important for the treatment of patients by health care providers and for the control of Legionella epidemics by public health officials. Rapid urinary antigen detection kits such as lateral flow immunochromatographic assays detecting L. pneumophila serogroup 1 are the first-line diagnostic tests for Legionnaires’ disease. In Europe, the 80 % of cases were diagnosed using urinary antigen kits in 2009–2010 [1].

Various immunochromatographic membrane kits requiring no specialized laboratory equipment are commercially available. These tests are easy to perform and provide results in a few minutes, but they have not demonstrated optimal sensitivity [2–4]. Two approaches have been proposed to increase the sensitivity of the commercialized kits that are available: additional readings after a longer incubation time [2, 4–8] and the concentration of urine samples by centrifugation using filter units. Several studies showed an increase in the sensitivity of the BinaxNOW® Legionella Urinary Antigen Card (UAC) by using concentrated urine samples [4, 9–11].

The Sofia Legionella Fluorescence Immunoassay (FIA; Quidel, San Diego, CA, USA) is a recently introduced rapid diagnostic test for Legionnaires’ disease (LD) that uses immunofluorescence technology coupled with an automatic reader to enhance its sensitivity. To date, no study has been published on the performance of this test compared with the BinaxNOW® Legionella UAC (Alere, Jouy-en-Josas, France), the kit that is most commonly used. The aim of this study was to evaluate the performance of Sofia Legionella FIA in comparison with BinaxNOW® Legionella UAC in concentrated and nonconcentrated urine samples at two National Reference Centers for Legionella.

Materials and methods

Clinical specimens

A total of 269 urine samples were tested in the French and Swiss National Reference Centers (centers 1 and 2 respectively: 199 in center 1 and 70 in center 2), corresponding to 179 prospective samples submitted for Legionella urinary antigen detection between January and March 2013 (150 in center 1 and 29 in center 2) and 90 repository urine samples collected from 2009 to 2013 from patients with known LD and stored at −20 °C (49 in center 1 and 41 in center 2). Previously, urine samples had been classified as positive based on a BinaxNOW® Legionella UAC result obtained from concentrated urine samples.

A case of LD was defined according to the European Legionnaires’ Disease Surveillance Network (ELDSNet) criteria [12], i.e., clinical and/or radiological evidence of pneumonia associated with positive urinary antigens and/or positive culture or PCR on respiratory samples. In our study, cases were diagnosed by positive urinary antigens using BinaxNOW® Legionella UAC on concentrated urine samples. Twenty-five cases were confirmed by culture on respiratory samples with Legionella pneumophila serogroup 1 strain and 7 of them were Mab3/1-negative.

Legionella urinary antigen detection

The BinaxNOW® Legionella UAC is a colorimetric immunochromatographic test. A sample was considered positive when both the control line and the test line were visible after 15 min of incubation at room temperature.

The Sofia Legionella FIA kit employs a lateral-flow immunofluorescence technique. The signal is only visible under UV light. The reading is performed after 10 min of incubation and interpreted using the SOFIATM analyzer, which automatically scans the test strip, collects and analyzes the fluorescence data, and then calculates and reports the result in about 1 min.

With the Binax Legionella Urinary Antigen EIA kit (Alere), a sample was considered positive when the mean absorbance value was three times the mean value of the negative control.

Method

The tests were performed according to the manufacturers’ instructions. In addition, some samples were concentrated and heat-treated before testing. Urine samples yielding discrepant results before and after heat treatment were checked by using the Binax Legionella Urinary Antigen EIA kit.

Nonconcentrated urine samples were tested using Sofia Legionella FIA and the results were compared with those of nonconcentrated (centers 1 and 2) and those of concentrated urine samples (center 1 only) tested simultaneously using BinaxNOW® AUC.

Urinary concentration was performed in center 1 by centrifugation at 4,000 g for 10 min using Amikon Ultra-4 Ultracel-10 k (Millipore Corporation, Bedford, MS, USA) [13].

In the two centers, urine samples yielding a positive result were retested after heating at 100 °C for 5 min and centrifugation for 15 min at 1,000 g to exclude false-positive results [14]. Any sample yielding a positive result after heating was considered true-positive. Any urine sample positive before heating and negative after heating was considered false-positive.

Results

Of the 269 nonconcentrated urine samples tested, the Sofia Legionella FIA test and the BinaxNOW® Legionella UAC yielded 172 and 181 negative results and 97 and 88 positive results respectively (Table 1). Of the 9 discrepant samples yielding a positive result with the Sofia Legionella FIA test, but a negative one with the BinaxNOW® Legionella UAC, 5 urine samples (4 in center 1 and 1 in center 2) yielded a negative result with the Sofia Legionella FIA test after heating. Four of them with a sufficient amount to be checked with the BinaxNOW® EIA showed a confirmed negative result. No respiratory samples were available for these 5 patients. The 4 other urine samples (2 in center 1 and 2 in center 2) remained positive with the Sofia Legionella FIA test after heating and yielded a positive result with the BinaxNOW® EIA before and after heating. For 2 of these patients, a Legionella PCR performed on respiratory samples was positive and in addition, for one of them a L. pneumophila serogroup 1-PCR was also positive. For the other one, the L. pneumophila serogroup 1-PCR was negative.

Table 1 Comparison of the results of the Sofia Legionella fluorescent immunoassay (FIA) and the BinaxNOW® Legionella urinary antigen card (UAC) using nonconcentrated urine samples (n = 269)
Full size table

As several studies demonstrated that urine sample concentration increased the sensitivity of urinary tests, the results of the BinaxNOW® Legionella UAC on concentrated urine samples were compared with those of the Sofia Legionella FIA on nonconcentrated urine samples in center 1 (n = 199; Table 2). The two reagents detected 49 positive samples, confirming the presence of Legionella urinary antigen. Among them, 2 samples had not been detected by the BinaxNOW® Legionella UAC before concentration. Four discrepant results were observed, yielding a positive result with the Sofia Legionella FIA and a negative one with the BinaxNOW® Legionella UAC. These samples were confirmed as negative by the Sofia Legionella FIA after heating, corresponding to false-positive results as previously described in Table 1.

Table 2 Comparison of the results of the Sofia Legionella fluorescent immunoassay (FIA) using nonconcentrated urine samples and the BinaxNOW® Legionella urinary antigen card (UAC) using concentrated urine samples in center 1 (n = 199)
Full size table

Although the manufacturer of the Sofia Legionella FIA does not recommend urine concentration, we evaluated its impact on 199 concentrated urine samples in center 1. Eleven urine samples were detected as being positive with use of the Sofia Legionella FIA, but negative with use of the BinaxNOW® Legionella UAC. Among these samples, 9 turned out to have negative results after heating, confirming false positivity. In 1 case there was not enough of the urine sample available for retesting after heating; thus, it was classified as invalid. The last sample remained positive after heating, but was detected as being negative by the BinaxNOW® EIA. Legionella spp. and L. pneumophila PCR performed on respiratory samples were negative for this patient.

Discussion

In this work, we evaluated the Sofia Legionella FIA, a recently introduced rapid diagnostic test for the detection of L. pneumophila serogroup 1 urinary antigens, which uses immunofluorescence technology to enhance its sensitivity.

Although the manufacturers do not recommend urine concentration, the centrifugal ultrafiltration method for rapid concentration of L. pneumophila urinary antigen was evaluated. The Legionella urinary antigen is stable at 100 °C for 30 min [15]; therefore, all positive samples were retested as described previously after boiling to suppress nonspecific reactions and confirm positive results [9, 14, 16, 17].

Our results demonstrate that the Sofia Legionella FIA test exhibits higher sensitivity than the BinaxNOW® Legionella UAC, since it allowed the detection of all the LD-positive patients with nonconcentrated urine samples. This study also confirms that the sensitivity of BinaxNOW® Legionella UAC is enhanced by the concentration of urine without any decrease in specificity. As a matter of fact, 2 nonconcentrated urine samples that tested negative with the BinaxNOW® Legionella UAC yielded a positive result after concentration. Thus, these two tests demonstrated similar performance in terms of sensitivity only if urine samples are concentrated for the BinaxNOW® Legionella UAC.

The major limitation of the Sofia Legionella FIA is the lower specificity compared with the BinaxNOW® Legionella UAC that was detected in both concentrated and nonconcentrated urine samples. As a matter of fact, Quidel recommends using the Sofia Legionella FIA in nonconcentrated urine samples. Our results confirm this important point, since the number of false-positive results increases from 4 before to 11 after concentration. Moreover, 1 out of the 11 false-positive results in concentrated urine samples remained positive after heating.

These results of sensitivity and specificity were further confirmed in 214 new prospective urine samples tested between March and June 2014 in center 1 with Sofia Legionella FIA performed on nonconcentrated samples and the BinaxNOW® Legionella UAC on concentrated samples. Three positive urine samples were equally detected by the two kits and boiling urine samples eliminated 6 false-positive results with use of the Sofia Legionella FIA.

Finally, without heat treatment, a lack of specificity was observed for 2.3 % of all samples tested in this study (11 out of 483). Because of the consequences of false-positive results on the patient outcome and on outbreak investigation [17, 18], the recommendation of boiling any positive urine to differentiate a false-positive result from a true one should be recommended in the manufacturer’s instructions.

The Sofia Legionella FIA may represent a next step in the evolution of immunoassays, combining lateral flow and fluorescent antibody detection formats. The choice of fluorescent rather than colorimetric detection is at least partially responsible for the improved sensitivity demonstrated by the Sofia Legionella FIA, and this improvement has also been demonstrated for Influenza virus detection in comparison with several colorimetric lateral flow devices [19–22]. Because of its automated reading rather than subjective readings of the colorimetric bands, this assay produces objective data. Moreover, the Sofia analyzer can be connected to laboratory information systems to facilitate the accurate and rapid reporting of results.