Performance of a Pneumolysin Enzyme-Linked Immunosorbent Assay for Diagnosis of Pneumococcal Infections

Patient groups.Samples were obtained from 481 patients admitted to three hospitals in Asturias, Spain (Hospital Universitario Central de Asturias, Hospital Monte Naranco, and Hospital San Agustín), between January 2003 and December 2006 and recruited as part of a prospective, controlled clinical study investigating novel diagnostic tools for serious community-acquired bacterial infections. Of the 383 adults, 245 were male and 138 were female (mean age, 60.03 years; range, 18 to 99 years), and of the 98 children, 57 were male and 41 were female (mean age, 4.8 years; range, newborn to 16 years). Samples were collected upon admission or during the first day of hospitalization and frozen at −70°C until use. Samples from healthy adults were obtained from volunteer subjects at the Faculty of Medicine. Samples from healthy children were obtained during routine pediatric check-ups at a health care center in Lugones (Asturias, Spain). Ethical approval was obtained from the Hospital Universitario Central de Asturias Research Ethics Committee. Informed consent was obtained from adult patients and from the parents or tutors of the children. Fifteen adult urine samples were obtained by M. Falguera at the Hospital Universitari Arnau de Vilanova (Lleida, Spain). In children, a nasopharyngeal swab sample was taken using a calcium alginate-tipped swab on a flexible aluminum wire (TRANSWAB; Medical Wire and Equipment Co., Ltd., Corsham, Wilts, England). The specimens were brought back to the laboratory within 4 h for processing. None of the healthy controls had been hospitalized for any reason during the previous month, nor they had received any antibiotic treatment during the previous week; no respiratory symptoms were present.

Adult patients included in the study were classified into five main groups (see Table 2). The pneumococcal infection patient group included 76 urine samples from patients with pneumonia (n = 47), probable pneumococcal pneumonia (n = 12), septicemia (n = 8), meningitis (n = 5), arthritis (n = 1), brain abscess (n = 1), breast abscess (n = 1), and puerperal infection (n = 1) due to S. pneumoniae. One patient diagnosed with pneumococcal pneumonia was also positive for Mycobacterium tuberculosis. The nonpneumococcal infection patient group included 56 urine samples from 31 patients with nonpneumococcal pneumonia caused by Mycoplasma pneumoniae (n = 7), Legionella pneumophila (n = 6), Chlamydia pneumoniae (n = 6), Pseudomonas aeruginosa (n = 5), Haemophilus influenzae (n = 1), Haemophilus parainfluenzae (n = 1), Staphylococcus aureus (n = 2), Pneumocystis jiroveci (n = 1), and Moraxella catarrhalis (n = 2); 5 patients with lung infections due to M. tuberculosis; 14 patients with urinary tract infections caused by Candida albicans (n = 1), Enterobacter cloacae (n = 1), Escherichia coli (n = 4), Enterococcus faecalis (n = 1), P. aeruginosa (n = 3), Proteus mirabilis (n = 1), Providencia stuartii (n = 1), Streptococcus agalactiae (n = 1), and Staphylococcus epidermidis (n = 1); 2 patients with hepatic infections caused by hepatitis C virus (HCV); 1 patient with endocarditis due to Streptococcus bovis; 1 patient with herpetic meningoencephalitis; and 1 patient with lung abscess caused by Staphylococcus aureus. The noninfectious disease patient group contained 12 samples from patients suffering lung cancer (n = 8), asthma (n = 2), and cardiac failure (n = 2). The unknown etiological disease patient group comprised 204 urine samples from patients with pneumonia of unknown etiology (n = 178), chronic obstructive pulmonary disease (n = 3), pleurisy (n = 2), fever of unknown origin (n = 6), and respiratory failure (n = 15). The healthy person group included 35 subjects without clinical or radiological signs of pneumonia or infections.

The pediatric population included in the study was classified into five main groups (see Table 3). The pneumococcal infection patient group included 32 urine samples from patients with pneumonia (n = 11), probable pneumococcal pneumonia (n = 6), septicemia (n = 1), meningitis (n = 2), otitis (n = 8), sinusitis (n = 2), conjunctivitis (n = 1), and a finger infection (n = 1) due to S. pneumoniae. The nonpneumococcal infection patient group included two cases of pneumonia caused by M. pneumoniae (n = 1) and H. influenzae (n = 1), gastroenteritis caused by Salmonella species (n = 3), tonsillitis caused by Streptococcus pyogenes (n = 3), septicemia caused by Enterococcus faecalis (n = 1), common cold (n = 3), and urinary infection caused by Escherichia coli (n = 2). The noninfectious diseases patient group included samples from children with allergic rhinitis (n = 6) and jaundice (n = 1). The unknown etiological disease patient group was made up of 31 urine samples from patients with pneumonia of unknown etiology (n = 17), respiratory failure (n = 8), otitis (n = 3), sinusitis (n = 1), septicemia (n = 1), and fever of unknown origin (n = 1). The healthy person group included 12 subjects without clinical or radiological signs of pneumonia or infections. Pediatric subjects were divided into S. pneumoniae nasopharyngeal carriers and noncarriers, and the carrier status was determined to be positive when S. pneumoniae was isolated from the nasopharyngeal swab sample culture.

Microbiological analysis.Culture of blood (ESP system; Difco Laboratories, Detroit, MI), cerebrospinal fluid, peripheral venous catheter, tracheal aspirates, sputum, urine, ascites, pleural fluid, pus from paracentesis, and joint effusion samples was performed according to standard microbiological methods (21). Identification of microorganisms was based on the usual criteria. Human immunodeficiency virus identification was performed by enzyme immunoassay and Western blotting, and HCV identification was done by a recombinant immunoblot assay. Influenza A virus, influenza B virus, parainfluenza virus types 1, 2, and 3, adenovirus, and respiratory syncytial viruses were detected by culture and shell vial assay. Complement fixation tests were performed for influenza viruses A and B, parainfluenza virus types 1, 2, and 3, adenovirus, and respiratory syncytial virus. A microimmunofluorescence method was used to detect antibodies against C. pneumoniae, Coxiella burnetii, Chlamydia psittaci, and L. pneumophila. An enzyme immunoassay was carried out to detect anti-M. pneumoniae immunoglobulin M (IgM); the diagnosis of acute infection criteria was retained when a positive IgM or a fourfold or greater rise in IgM and IgG titers was obtained on sequential sera sampled at 3-week intervals. Identification of M. tuberculosis was performed by Ziehl-Nielsen staining, culture, and specific hybridization probes. Nasal swabs were inoculated onto 5% horse blood agar supplemented with gentamicin (5 μg/ml) and incubated in 5% CO₂ for 16 to 24 h. All S. pneumoniae isolates were identified using colony morphology, Gram stain, optochin susceptibility, and bile solubility. All urine samples were cultured and tested with the Binax Now Legionella pneumophila test. Cases with incomplete microbiologic results or for which such results were unavailable were included in the groups of “unknown etiological diseases” and discarded from test performance calculations.

Diagnostic criteria.Patients were classified as having pneumonia if they presented with an acute illness with the presence of new or progressive infiltrate on chest radiograph plus at least two of the following symptoms: fever, cough, shortness of breath, and pleuritic chest pain. All patients were classified as having pneumococcal pneumonia when S. pneumoniae was isolated from blood samples, pleural fluid, or peripheral venous catheter. Pneumococcal pneumonia was considered to be probable when S. pneumoniae was isolated from tracheal aspirates (≥10⁶ CFU/ml) or sputum (ratio of polymorphonuclear neutrophils/squamous epithelial cells of ≥2 or Q [quality] score of ≥+1 on Gram stain). The diagnosis of urinary tract infection was considered to be positive if >10⁵ CFU/ml of a single bacterium was isolated in urine cultures. Noninfectious diseases such as asthma, lung cancer, and cardiac failure were diagnosed by standard criteria.

PLY-ELISA.Prior to analysis, samples were thawed at room temperature, an aliquot was spun at 15,000 × g for 1 min, and the supernatant was removed for assay. PLY quantification was performed by ELISA as previously described (5). Triton X-100 was added to urine samples to a concentration of 2% (wt/vol); the samples were then incubated at 37°C for 30 min and centrifuged at 3,000 × g for 15 min. Flat-bottomed polystyrene Combiplate White Breakable plates (Labsystems, Helsinki, Finland) were coated with 1 μg/well of PLY-7 (IgG1κ, anti-PLY mouse monoclonal antibody) in phosphate-buffered saline (PBS) at 37°C for 3 h. Plates were washed at each step with PBS plus 0.1% Tween 20 and blocked with blocking buffer prepared according to the instructions provided by the manufacturer of the ELISA-Light chemiluminescent detection system (Tropix; Applied Biosystems, Bedford, MA). Samples were then added to the wells and incubated at 37°C for 1 h with shaking. After washing the plates six times, they were incubated with rabbit IgG polyclonal anti-PLY antibody diluted in blocking buffer and incubated at 37°C for 30 min. Anti-rabbit IgG alkaline phosphatase conjugate secondary antibody (Sigma Chemicals Co.) was diluted to 1:5,000 in blocking buffer and incubated as described above. Plates were loaded onto a Luminoskan RS (Labsystems) luminometer, and the wells were automatically filled with substrate/enhancer solution (0.4 mM chloro-5-substituted adamantyl-1,2-dioxetane phosphate with 1× Sapphire-II) and incubated at 37°C for 10 min. Triplicate samples of urine from each subject were studied on a single plate. Each plate contained six twofold dilutions of a known concentration of PLY. Some control wells on each plate had all reagents added except diluted PLY and a pool of negative urine samples obtained from 20 healthy subjects. All ELISAs were carried out with two types of control wells: blank wells had all reagents added except diluted PLY, whereas negative wells were loaded with a pool of negative urine samples obtained from 20 known healthy subjects. No significant differences were observed between relative light units (RLU) obtained from said controls. A cutoff value was calculated from the mean concentration plus twice the standard deviation of the control groups (healthy donors, noninfectious diseases, and nonpneumococcal infections). The cutoff value was 11.13 pg PLY/well. The PLY-ELISA test was judged positive only if the final RLU exceeded twice that for the control wells or the cutoff point.

Bacterial strain lysates.Cell lysates of L. pneumophila serotype 1 (OLDA, Bellingham, and Philadelphia strains), serotype 3, serotype 6, and serotype 10 were kindly provided by J. H. Helbig, Technical University of Dresden, Dresden, Germany. M. tuberculosis H37Rv and CSU 93 cell lysates were obtained from Colorado State University (Mycobacteria Research Laboratories TB Vaccine Testing and Research Materials Program). C. pneumoniae CWL 029 and M. pneumoniae FH cell lysates were purchased from Microbix Biosystems Inc., Toronto, Canada. Clinical isolates of S. agalactiae, P. aeruginosa, P. mirabilis, E. faecalis, E. coli, P. stuartii, C. albicans, E. cloacae, and S. aureus were obtained from patients at the microbiology laboratory of the Hospital Universitario Central de Asturias. They were harvested from solid cultures with a cotton swab, and suspensions containing 10⁷ CFU/ml were prepared in PBS, boiled for 5 min, and centrifuged at 3,000 × g for 15 min before being tested for antigen detection by PLY-ELISA.

Binax Now S. pneumoniae test.The Binax Now S. pneumoniae test was performed for nonconcentrated urine samples obtained from all enrolled patients according to the manufacturer's instructions. The results were read visually after 15 min. The test was interpreted by the presence or absence of visually detectable pink to purple lines.

Statistical analysis.The confidence intervals (CIs) were calculated by Wilson's method. Frequencies were compared using the χ² test with Yates correction. Statistical differences in RLU were calculated by one-way analysis of variance followed by the Bonferroni test. Statistical analyses were performed using GraphPad PRISM software (v.4.00 for Windows; GraphPad Software, San Diego, CA). The limit of statistical significance was a P value of 0.05.

Antigen detection in lysates of nonpneumococcal bacteria.The possibility of positive results in the PLY-ELISA due to cross-reactions with antigens from other microorganisms that might be present in the samples was assessed. Bacterial lysates (Table 1) containing 10⁷ CFU/ml from 19 nonpneumococcal strains did not produce significant immune reactions, with the exceptions of S. agalactiae and P. aeruginosa.

Detection of pneumococcal antigens in urine samples.The sensitivity of the PLY-ELISA was calculated for patients with S. pneumoniae pneumococcal infections proven by culture. Test specificity was calculated in diverse subject groups used as controls, including patients with a definitive etiologic diagnosis of nonpneumococcal infections, patients with noninfectious diseases, and healthy subjects. There were no significant differences in ages and genders between patients diagnosed with pneumococcal infections and the groups of subjects used as controls.

Table 2 shows the results of PLY-ELISA and the Binax Now S. pneumoniae urinary antigen test in adults. In order to know if the presence of nasopharyngeal S. pneumoniae could determine the performance of PLY-ELISA in children, 35 pneumococcal nasopharyngeal carriers and 63 noncarriers were selected (Table 3). The concentrations of PLY in urine samples of different subject groups were obtained (Table 4). The descriptive statistics for PLY-ELISA and Binax Now S. pneumoniae urinary antigen test of the same urine samples are shown in Table 5.

Table 6 presents the etiologic diagnoses of patients with potential false-positive results by the PLY-ELISA. Three patients were shown to have legionellosis, two of which were diagnosed by antigen detection in urine using the Binax Now Legionella pneumophila test; the samples were also positive for pneumococcal antigen using the Binax Now S. pneumoniae test. Two patients presented pneumonia due to M. pneumoniae, and one patient presented pneumonia caused by C. pneumoniae. Patient 7 was diagnosed with pneumonia by H. influenzae, and urine samples were positive for pneumococcal antigen according to the Binax Now S. pneumoniae test. Patient 8 was diagnosed with a urinary infection, although she developed respiratory symptoms while hospitalized. Two false-positive results were obtained from the group of noninfectious disease patients. The patient with cardiac failure was an elderly woman and died some hours later. Patient 10 was a premature infant diagnosed with jaundice. In all cases, the PLY-ELISA was positive, presenting RLU values close to the established cutoff (data not shown).

Persistence of PLY in urine.To examine the hypothesis that free PLY in urine was quickly eliminated, the persistence of detectable antigens was tested in 11 adults with pneumococcal pneumonia. Patients were diagnosed on the bases of a blood culture positive for S. pneumoniae (n = 3) as well as a positive sputum or tracheal aspirate culture (n = 8). Sequential urine samples were collected every day for the first week after hospital admission. The Binax Now S. pneumoniae test remained positive in samples taken over the first 6 days, while significantly decreased RLU values by PLY-ELISA were obtained after 4 to 6 days of hospital admission (P < 0.05) (Fig. 1).

• Open in new tab
• Download powerpoint

FIG. 1.

Persistence of PLY in urine samples from 11 pneumococcal pneumonia patients. Sequential urine samples were collected every day for the first week after hospital admission. The presence of PLY was determined by PLY-ELISA and compared with a negative urine sample pool (NU). A significant decrease in RLU values was obtained after 4 to 6 days of hospital admission. *, P ≤ 0.05.