INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Since the initial description of Legionnaires' disease in 1976, Legionella pneumophila has been increasingly recognized as a pathogen causing both community-acquired and nosocomial pneumonia (23). Legionella pneumonia can be difficult to diagnose because the signs and symptoms are nonspecific and do not distinguish Legionella infection from other common causes of pneumonia (11).

Diagnosis by culture of respiratory-tract secretions requires 3 to 5 days of incubation (10). Many laboratories do not culture Legionella spp. or are unable to do so (7). Determination of antibody titers can provide only a retrospective diagnosis, since seroconversion usually occurs 2 to 9 weeks after the onset of infection (12). Direct fluorescent-antibody staining of respiratory secretions is a rapid test, but it has a low level of sensitivity and cross-reactions with other bacteria in the clinical laboratory.

The number of etiologic diagnoses could increase, providing therapeutic benefit, if a rapid, sensitive, and specific technique for detecting the soluble antigen of Legionella in urine were used. The detection of the soluble antigen of L. pneumophila serogroup 1 in urine by an enzyme immunoassay (EIA) from Binax (Portland, Maine) has proven rapid, sensitive, and specific (8, 9, 14, 18), and the use of concentrated urine improves the yield significantly (8, 9). Although L. pneumophila serogroup 1 is responsible for more than 80% of Legionnaires' disease cases (11), other species or serogroups can cause infection in humans (23, 27). The value of urinary antigen detection assays to diagnose Legionnaires' disease would be enhanced if Legionella infections other than L. pneumophila serogroup 1 infections could also be detected.

The aim of the present study was to evaluate a newly available kit for performing an EIA to detect all antigens of L. pneumophila serogroups with a relatively wide spectrum of cross-reactivity, as well as antigens of other Legionella species (Biotest AG, Dreieich, Germany) by comparing its sensitivity and specificity with those of the EIA to detect L. pneumophila serogroup 1 antigen (the Binax EIA). Both tests were performed with both concentrated and nonconcentrated urine samples in order to assess whether concentration improves the yield in the Biotest EIA as it does in the Binax EIA.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Patients and samples. Urine samples were obtained from four groups of patients. First, we studied 69 nonconcentrated and 45 concentrated urine samples from 65 patients (48 men and 17 women) with pneumonia caused by L. pneumophila. The mean age of the patients in this group was 61 years (age range, 28 to 86 years). Pneumonia was community acquired for 31 patients and nosocomial for 34 patients. Legionnaires' disease was diagnosed by isolating L. pneumophila from respiratory specimens for 19 patients, by seroconversion for 41 patients, and by culture plus seroconversion for 5 patients. L. pneumophila was isolated on selective buffered charcoal-yeast extract- medium (13), and identification was based on the usual criteria (31). Detection of specific antibodies was carried out by indirect immunofluorescence (13) for at least two serum samples, one obtained during the initial phase of disease and one obtained during convalescence (3 to 9 weeks); the criterion for seroconversion was a fourfold increase in antibody titer (13).

Sample treatment. The samples were boiled for 5 min and centrifuged at 1,000 × g for 15 min to prevent nonspecific reactions (6, 8, 9, 25, 28). The antigen present in the urine was concentrated 25-fold by selective ultrafiltration (8, 9) (Urifil-10 Concentrator; Millipore Corporation, Bedford, Mass.). Samples were tested by the Binax EIA immediately or after being stored from 1 to 12 months. The Biotest EIA results were compared to the prior results of the Binax EIA.

Soluble-antigen preparation. Soluble antigens of the American Type Culture Collection (ATCC) (Manassas, Va.) Legionella strains listed in Table 1 were prepared by a method adapted from that of Kohler et al. (22). Each Legionella strain was grown on buffered charcoal-yeast extract- agar at 35°C for 48 h in 3% CO₂ and harvested in 3 ml of 0.05 M phosphate-buffered saline (pH 7.4). The cell suspensions were centrifuged, and 0.5 ml of wet-packed cells was autoclaved at 100°C for 1 h and then left at 4°C for 10 days in order to extract soluble antigens. The suspensions were centrifuged at 1,200 × g for 10 min, and the supernatants were collected.

Binax Legionella urinary antigen EIA kit. The Binax EIA method is a direct sandwich assay that uses polyclonal rabbit immunoglobulin G specific for L. pneumophila serogroup 1 as the capture and detection antibody. The test was performed according to the manufacturer's instructions. Samples were considered positive when the absorbance units were triple those recorded for the negative control.

Biotest Legionella Urin Antigen EIA. The Biotest assay is a direct sandwich assay that uses polyclonal rabbit antibodies which react with antigens of all L. pneumophila serogroups as well as with antigens of other Legionella species as the capture and detection antibodies. The test was performed by following the manufacturer's instructions. The cutoff value was calculated as the mean absorbance of the negative controls plus 0.200. Urine samples with an extinction equal to or greater than the cutoff were considered positive. Testing of urine samples with absorbance values in the region of the cutoff plus 0.200 U had to be repeated for confirmation, and when absorbance values were again in this range or higher, the samples were considered positive.

Statistical analysis. Data were analyzed by the SAS statistics program (24). Concentrated and nonconcentrated urine samples were compared by a Student t test. Results by the Binax EIA and the Biotest EIA were compared, and specificity was evaluated, by means of McNemar's test.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Detection of Legionella antigens in patients' urine. By using the Binax EIA, soluble antigen was detected in 44 nonconcentrated samples and in 40 concentrated samples from group 1 patients. By using the Biotest EIA, Legionella antigen was detected in 46 nonconcentrated samples and in 39 concentrated samples from group 1 patients. The sensitivity of the Binax EIA was 63.76% with nonconcentrated urine and 88.88% with concentrated samples. The sensitivities of the Biotest EIA with nonconcentrated and concentrated urine samples were 66.66 and 86.66%, respectively (Table 2). Urine samples from patients with no clinical symptoms or signs of pneumonia or with pneumonia of other etiologies were all negative, whether they were concentrated or not. The specificity for both EIAs was 100% with both nonconcentrated and concentrated urine samples.

Detection of Legionella soluble antigens from culture extracts. The Binax EIA and Biotest EIA were tested for their capabilities to detect soluble antigens of Legionella. For the 14 serogroups from L. pneumophila and for the Legionella bozemanii ATCC strain, both tests were positive; for the L. longbeachae ATCC strain, both tests were negative.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The EIA for detection of L. pneumophila serogroup 1 antigens has been shown to be a sensitive and specific test when applied to urine specimens (4-6, 19, 21, 23, 25, 29). Several authors (9, 14, 18) have confirmed the usefulness of the EIA kit available from Binax for detecting antigenuria in Legionnaires' disease. Our previous study (9) indicated that the Binax EIA was rapid, sensitive, and specific and was comparable in sensitivity and specificity to the Binax radioimmunoassay (RIA). Hackman et al. (14) also compared the Binax EIA with the Binax RIA. The sensitivities were equal (77%), and the EIA was as useful as the RIA for detecting L. pneumophila serogroup 1. Kazandjian et al. (18) evaluated the Binax EIA in cases of suspected or proven legionellosis, detecting antigen in samples from 37% of patients with suspected legionellosis and from 83% of those with proven L. pneumophila serogroup 1 infection.

The recently available EIA for detecting antigen from all L. pneumophila serogroups from Biotest was evaluated recently by the manufacturer (26). Biotest developed the EIA to detect Legionella antigen in urine, and it found antigen in nonconcentrated urine from 102 patients with a specificity of 99.8% and a sensitivity of 100%. Tang et al. (28) also developed an EIA that detected soluble antigens of L. pneumophila serogroups 1, 3, 4, 6, and 8, Legionella micdadei, and L. longbeachae serogroup 1 in 25 of 35 patients with legionellosis with 100% specificity.

In our study we found no significant differences in sensitivity between the Binax EIA and the Biotest EIA, although nonconcentrated urine gave a slightly higher level of sensitivity for the Biotest EIA, and the sensitivity of the Binax EIA was slightly higher than that of the Biotest EIA when concentrated urine was used. The sensitivities of the Binax EIA and the Biotest EIA for detecting antigenuria, whether in nonconcentrated or concentrated urine, were higher for patients diagnosed by positive culture than for those diagnosed by seroconversion alone. Culture-positive patients may excrete more antigen than do culture-negative patients (20).

Our study indicated that boiling samples and concentrating the antigen is useful with both EIAs. Concentrating the antigen present in urine by selective ultrafiltration increases sensitivity with no decrease in specificity. Urine concentration by this method is thus a simple procedure which contributes substantially toward increasing the utility of techniques for diagnosing Legionnaires' disease. Given that soluble antigen is stable at high temperatures (22), boiling urine for 5 min and centrifuging the sample at 1,000 × g is recommended to prevent nonspecific reactions (8, 9, 23, 25, 28).

Antigenuria is detectable from the onset of disease, and the results have been unaffected by prior administration of antibiotics in several studies (11, 21, 22). Other authors have reported that soluble antigen can be detected in urine many days after the start of symptoms: from 10 to 15 days (1, 23, 25), 60 days (6), and 42 days or longer (21). In the present study, we were able to detect antigen from the onset of symptoms up to 50 days later by both the Binax EIA and the Biotest EIA. Birtles et al. (6) obtained a sensitivity of 77% when testing urine samples collected from the start of symptoms up to 28 days afterwards and a sensitivity of 86% when only urine from the first 7 days of illness was used.

The coefficient by the Binax EIA for samples from patients not suffering from Legionnaires' disease never produced ratios over 2.0, and the results by the Biotest EIA never were in the range of the cutoff plus 0.200, whether urine was concentrated or nonconcentrated.

We observed no cross-reactions with other microorganisms in urine samples from patients with urinary-tract infections and high bacterial counts, unlike other authors (1, 22, 28). Although cross-reactions among species of the genus Legionella have been described (5, 20, 23), Kazandjian et al. (18), using the Binax EIA, observed no cross-reactions of L. pneumophila serogroup 1 with other species or serogroups (L. longbeachae, and L. pneumophila serogroups 2, 3, 4, and 10). Our results for the single patient with pneumonia due to L. longbeachae were negative by both the Binax EIA and the Biotest EIA.

The major advantage of the Biotest EIA over the Binax EIA is that the first detects the antigens from all L. pneumophila serogroups with a wide spectrum of cross-reactivity as well as antigens from other Legionella species, but with regard to the detection of Legionella soluble antigens from culture extracts, both EIAs were able to detect antigen from all strains tested, except for L. longbeachae. Further studies involving urine samples from patients with legionellosis due to Legionella species or serogroups other than L. pneumophila serogroup 1 are required in order to determine the real usefulness of the Biotest EIA, given that Legionella soluble antigens from culture extracts are not an adequate substitute. Kohler et al. (20) reported cross-reactive urinary antigens among patients with legionellosis caused by L. pneumophila serogroups 1 and 4 and the Leiden 1 strain. Kohler and Sathapatayavongs (19) have reported on the detection of antigen in the urine of patients infected with L. pneumophila serogroup 4 and Legionella dumoffii with an L. pneumophila serogroup 1 direct system. Bibb et al. (5) found that extracts of cells of the reference strains of L. pneumophila serogroups 2 through 8 cross-react with serogroup 1 antibodies in their direct enzyme-linked immunosorbent assay (ELISA).

Recently Benson et al. (3) evaluated the usefulness of the Binax RIA for detection of serogroups and species of Legionella other than L. pneumophila serogroup 1. They tested 34 urine samples from non-L. pneumophila serogroup 1 patients diagnosed by a broad-spectrum ELISA (28) and found that 14 of the 34 samples were positive with the Binax RIA.

The antigen detected by both the Binax EIA and the Biotest EIA is now believed to be a lipopolysaccharide (LPS) (30). Although the reference strain for each serogroup of L. pneumophila possesses at least one LPS-specific epitope not found on any other reference strain and therefore designated the serogroup-specific epitope (15), when antiserum prepared against serogroup 5 was used to probe the LPS from L. pneumophila serogroups 1 to 14, the antibodies recognized a common epitope harbored by all L. pneumophila serogroups (16, 17). Barthe et al. (2) also detected an epitope that was common to serogroups 1 to 8 of L. pneumophila and was attached to the LPS.

The EIA manufactured by Binax was initially developed by Kohler and coworkers. Given that the Binax EIA still produces polyclonal antibody using whole cells from L. pneumophila serogroup 1, as in Kohler's assay (22), it is more probable that the immune sera obtained contain antibodies against common epitopes harbored by all L. pneumophila serogroups. This fact could explain why the Binax EIA detects antigens from L. pneumophila serogroups other than serogroup 1. However, antigens of other species of Legionella and other serogroups of L. pneumophila may or not be excreted in urine. An antigen may be excluded from passage across the walls of the glomerular capillaries by virtue of its molecular weight or charge and thus may not be present in urine. Further studies are required in which multiple urine samples from patients with legionellosis due to species or serogroups other than L. pneumophila serogroup 1 are tested in order to determine the usefulness of the Binax EIA.

The fact that the Biotest EIA is able to detect all L. pneumophila serogroups does not imply a decrease in sensitivity in the detection of L. pneumophila serogroup 1. In cases of outbreaks of nosocomial pneumonia due to L. pneumophila serogroup 1, we found no significant differences between the Binax EIA and the Biotest EIA. Furthermore, the Biotest EIA could be especially useful in geographic areas where a minority of proven Legionella infections are caused by L. pneumophila serogroup 1.

The Biotest EIA is rapid, sensitive, and specific, comparable in sensitivity and specificity to the Binax EIA. The drawback of the Biotest EIA lies in its inability to identify the particular Legionella species and/or serogroup that is causing the infection. The Binax EIA contains 12 strips with eight separable wells each. When a unique sample is tested, it uses only four wells (one positive control, one negative control, one blank, and the sample). However, in the Biotest EIA the eight wells from a strip are inseparable. Thus, when a unique sample is tested, the Biotest EIA has to be performed with one strip using five wells (for one positive control, two negative controls, one blank, and the sample), wasting the three left, since they cannot be stored for further determinations. Consideration should be given to designing Biotest EIA strips with separate wells so that it will not be necessary to wait until there are enough samples to use all the wells of one strip, since the major interest of these soluble-antigen detection techniques is the rapid determination of the urine antigen. Both the Binax EIA and the Biotest EIA are useful tools for assisting physicians in the diagnosis and treatment of patients with Legionnaires' disease.