Sordé R, Falcó V, Lowak M, Domingo E, Ferrer A, Burgos J, Puig M, Cabral E, Len O, Pahissa A. Current and Potential Usefulness of Pneumococcal Urinary Antigen Detection in Hospitalized Patients With Community-Acquired Pneumonia to Guide Antimicrobial Therapy. Arch Intern Med. 2011;171(2):166-172. doi:10.1001/archinternmed.2010.347
Copyright 2011 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2011
According to current Infectious Diseases Society of America guidelines,1 hospitalized patients with community-acquired pneumonia (CAP) should receive an empirical broad spectrum antibiotic therapy to cover the most frequent etiologies. The same guidelines1 encourage investigation for specific pathogens that would significantly alter standard treatment. Rational use of antibiotics, using an appropriate pathogen-focused agent or narrowing empirical therapy, may decrease cost, drug adverse events, and the threat of antibiotic resistance.1- 4
The yield of traditional microbiological investigations is limited for several reasons: routine difficulties in obtaining good-quality sputum and the uncertainty of the value of its culture results, low sensitivity of blood cultures, and administration of antibiotics before samples collection.5,6
Tests based on urinary detection of bacterial antigens and nucleic acid amplification techniques try to solve these difficulties and, because of their rapidity, to minimize time in receiving microbiological results and provide an early and appropriate treatment.7
A quick and simple urinary antigen test, based on an immunochromatographic membrane technique, is widely available to detect the C-polysaccharide antigen of Streptococcus pneumoniae, the leading cause of CAP. This test has demonstrated reasonable sensitivity and good specificity in different studies.8- 13 Nevertheless, the clinical usefulness of this pneumococcal urinary antigen test is not well defined, and, consequently, current guidelines do not clearly recommend the situations in which testing should be performed. A prospective controlled study concluded that the pneumococcal urine test allows the targeted antibiotic therapy in young immunocompetent outpatients with nonsevere CAP.14 In hospitalized patients, a recently published randomized study15 found no benefit of targeted therapy based on urine antigen detection tests; however, its results were affected by design problems and the low number of patients included in the study.16
The aims of our study were to assess (1) the current use of pneumococcal urinary antigen detection in our setting, (2) its reliability for the etiologic diagnosis of pneumococcal pneumonia and its contribution to increase the etiologic diagnosis of CAP, and (3) the current and potential optimization of antimicrobial therapy according to pneumococcal urinary antigen detection results analyzing the outcomes of patients in whom treatment was modified according to the test result.
We performed a prospective study of all consecutive adult patients (≥16 years old) hospitalized with CAP from February 2007 through January 2008. The study was performed in the Hospital Universitari Vall d’Hebron (Barcelona, Spain), a 1200-bed teaching hospital that serves a population of about 500 000 people. The study was approved by the commission of medical ethics of our hospital.
Community-acquired pneumonia was defined as the presence of a new infiltrate on a chest radiograph with at least 1 of the following symptoms of acute respiratory illness: fever, new onset of cough, sputum production, dyspnea and/or tachypnea, pleuritic chest pain, and auscultatory findings consistent with pulmonary consolidation. Patients who were diagnosed as having diseases other than CAP during the follow-up (pulmonary tuberculosis, pneumonia caused by Pneumocystis jiroveci, lung cancer, or criptogenetic organizing pneumonia) were excluded.
The following criteria were used to classify a case of pneumonia as being of known etiology: (1) definite diagnosis proved by recovery of a pathogen from a normally sterile sample (blood or pleural fluid), positive result of a urinary antigen test for detection of Legionella pneumophila, 4-fold increase in the antibody titer for Mycoplasma pneumoniae, Chlamydophila pneumoniae, Coxiella burnetii, and L pneumophila between acute and convalescent serum samples and positive result in the polymerase chain reaction (PCR) for S pneumoniae in pleural fluid; (2) probable diagnosis shown by isolation of 1 predominant microorganism in a culture of a good-quality sputum sample with a concordant morphotype on Gram stain; (3) aspiration pneumonia, diagnosed on a clinical and radiological basis for patients who had a predisposing cause of aspiration (compromised consciousness, altered gag reflex, or dysphagia); and (4) mixed infections, diagnosed on the basis of the isolation of 2 or more microorganisms in a good-quality sputum sample with concordant morphotypes on Gram stain or more than 1 microorganism isolated in blood cultures.
Antimicrobial therapy was prescribed by the attending physician according to hospital protocol. In brief, empirical treatment for CAP consists in amoxicillin plus clavulanic acid or a third-generation cephalosporin. In those patients with severity criteria (based on clinical findings and current prediction rules) or with clinical suspicion of an atypical microorganism, azithromycin is added. Levofloxacin is the alternative agent in case of β-lactams allergy. If the patient has risk factors for Pseudomonas aeruginosa, piperacillin-tazobactam or cefepime is used.
The following baseline data were recorded from each patient: age, sex, smoking (≥5 cigarettes/d at CAP presentation or ≥1 month before), harmful alcohol consumption (≥60 g/d in men or ≥40 g/d in women), underlying diseases (chronic obstructive pulmonary disease [COPD], diabetes mellitus, chronic liver disease or cirrhosis, chronic heart disease, chronic renal failure, cerebrovascular disease, and cancer), immunosuppressive condition (long-term corticosteroid use defined as consumption of ≥10 mg/d of prednisone or equivalent during ≥130 days, chemotherapy for solid tumor, hematological malignant disease, human immunodeficiency virus [HIV] infection, and solid organ or bone-marrow transplantation), severity-of-illness scores (Pneumonia Severity Index [PSI] and CURB-65 [scoring by confusion, uremia, respiratory rate, low blood pressure, and age ≥65 years]) and, finally, previous and empirical antimicrobial therapy.
During follow-up we collected data on time to apyrexia, complications (respiratory failure, complicated pleural effusion, septic shock, and intensive care unit [ICU] admission), antimicrobial modifications (including timing and medical reason for these modifications), treatment duration, length of stay, mortality during the first 48 hours after admission, and, finally, 30-day mortality.
Diagnostic workup included (1) 2 sets of blood cultures in aerobe and anaerobe medium; (2) sputum sample collected when available (acceptable if it contained >25 polymorphonuclear leukocytes and <10 epithelial cells per high-power field); (3) culture and PCR findings for S pneumoniae in pleural fluid in patients requiring a thoracocentesis; (4) serologic determinations to detect L pneumophila, M pneumoniae, C pneumoniae, and C burnetii when they were requested by the attending physician; (5) urinary antigen detection for L pneumophila performed if there was clinical or epidemiological suspicion and in all cases of severe CAP; and (6) pneumococcal urinary antigen detection performed according to the attending physician.
The immunochromatographic assay BinaxNOW S pneumoniae urinary antigen test (Binax Inc, Portland, Maine) was used to detect the C-polysaccharide antigen from the cell wall of S pneumoniae. This test, performed in accordance with the manufacturer's instructions, is used in unconcentrated urine samples and gives a result in an average of 15 minutes.
Susceptibility of S pneumoniae to penicillin was analyzed according to 2008 Clinical and Laboratory Standards Institute revised breakpoints for infections other than meningitis: susceptible, minimal inhibitory concentration (MIC) = 2 mg/mL or lower; intermediate, MIC = 4 mg/mL; and resistant, MIC = 8 mg/mL or higher.17
To evaluate the reliability of pneumococcal urinary antigen detection we calculated its sensitivity and positive and negative predictive values, using 3 different reference groups of patients as the gold standard: (1) in definite pneumococcal pneumonias (CAP with S pneumoniae isolated in blood or pleural fluid culture); (2) in probable pneumococcal pneumonias (CAP with S pneumoniae as the predominant morphotype on Gram stain or culture of good-quality sputum); and (3) globally in all pneumococcal pneumonias (definite plus probable).
Specificity was calculated using 2 different control groups: (1) exclusively nonpneumococcal pneumonias (CAP with definite or probable microbiological results different from S pneumoniae, excluding CAP caused by aspiration; CAP of mixed etiology in which pneumococcus was isolated; and CAP with unknown etiology) and (2) all patients without a diagnosis of pneumococcal pneumonia, including those with aspiration pneumonia and pneumonia of unknown etiology. We also calculated positive and negative likelihood ratios (LRs) as a measure of the extent to which the pretest odds were altered by the test results; low negative LR (<0.1) and high positive LR (>10) are considered useful for ruling out and ruling in decisions, respectively.18
We collected the modifications of antimicrobial therapy according to the pneumococcal urinary antigen results and classified this modification as (1) optimal: narrowing the antimicrobial spectrum to intravenous penicillin or ampicillin or switch to oral route with amoxicillin; (2) improved: withdrawal of the macrolide in patients empirically treated with β-lactam and macrolide combination or partial reduction of antimicrobial spectrum (eg, changing from piperacillin-tazobactam to amoxicillin-clavulanate or ceftriaxone instead of to penicillin or ampicillin); (3) inappropriate therapy modifications; or (4) no changes in empirical therapy. We considered that the change in antibiotic treatment was due to the pneumococcal urinary antigen result when it was specifically recorded in the clinical chart of the patient and confirmed by the attending physician.
Finally, clinical outcomes of patients who received treatment adjustments according to urinary antigen test results were assessed. Statistical calculations were performed using SPSS software (version 15.0 for Windows; SPSS Inc, Chicago, Illinois).
A total of 474 episodes of CAP in 464 patients were included. There were 317 men (66.9%) and 157 women (33.1%), with a mean (SD) age of 64 (19.5) years. Baseline demographic and clinical characteristics of patients are summarized in Table 1. Forty-nine patients (10.3%) died during follow-up, 15 of whom (30.6%) died during the first 48 hours after admission.
Blood cultures were performed in 382 patients (80.6%) and in 68 cases (17.8%) a microorganism was isolated. A culture of pleural fluid was analyzed in 47 patients (9.9%), and in 10 cases the findings were positive. Legionella pneumophila urinary antigen test was performed in 386 cases (81.4%) with 14 positive results. Finally, in 45 patients (9.5%), paired serum samples were tested for atypical microorganisms with positive results in 10 cases.
Causal microorganisms are detailed in Table 2. Diagnosis of pneumococcal pneumonia was performed in 171 cases (36.1%). In 75 cases, diagnosis was made only by positive pneumoccoccal urinary antigen detection. Microbiological tests that provide the diagnosis of pneumococcal pneumonia are detailed in Table 3.
Susceptibility to antibiotics was available for 78 isolates of S pneumoniae, 50 of them from blood samples and 28 from sputum cultures. There was not any strain resistant to penicillin or cephalosporins.
Pneumococcal urinary antigen assay was performed in 383 cases of CAP (80.8%), and findings were positive in 136 (35.5%). The test was performed in 153 of 171 patients (89.5%) with pneumococcal pneumonia, and it was positive in 130 (85.0%). The other 6 positive results were found in the following situations: 3 positive results in mixed infections in which pneumococcus was isolated, 1 in aspiration pneumonia and the remaining 2 in pneumonias due to Escherichia coli. These diagnosis of pneumonia caused by E coli were based on isolation in blood culture in 1 case and isolation of E coli as the only bacterial isolation in a good-quality sputum sample in the other. Both patients were severely immunosuppressed, 1 with an extensive lung cancer in chemotherapy and the other with AIDS. The result of pneumococcal antigen test was considered false positive in both cases.
In the group with definite pneumococccal pneumonias, the pneumococcal urinary antigen test was performed in 50 cases, and it was positive in 39. In the group with probable pneumococcal pneumonias, the antigen test was performed in 28 cases and was positive in 16. Taking into account both groups together, the pneumococcal urinary antigen test was performed in 78 cases, and its result was positive in 55. In the nonpneumococcal pneumonias group, the test was performed in 51, and the findings were positive in 2 cases, which were considered to be false-positive results. Calculations of sensitivity, specificity, positive and negative predictive values, and positive and negative LRs by the different methods described are shown in Table 4.
The test allowed the diagnosis of pneumococcal pneumonia in 75 additional cases (43.8% of pneumococcal pneumonias). In consequence, the number of patients with CAP in whom an etiologic diagnosis was achieved increased from 194 (40.9%) to 269 (56.7%).
Positive results from the pneumococcal antigen test led the clinicians to reduce the spectrum of antibiotic treatment in 41 patients (8.6%). These reductions consisted in improved modification of empirical therapy in 18 cases and optimal modification of empirical therapy in 23. The median time to these modifications was 1 day (interquartile range [IQR], 1 day) in the improved adjustments group and 3 days (IQR, 3 days) in the optimal group. Globally, the median time to a treatment optimization due to the pneumococcal urinary antigen test result was 2 days (IQR, 2.5 days).
Five of these 41 patients (12.2%) were admitted to an ICU setting, 24 (58.5%) had 1 or more underlying disease, and 20 (48.8%) were classified as having a high mortality risk according to the PSI. Community-acquired pneumonia was cured in all of them. Just 1 patient died of progression of an underlying oncologic disease in the 18th day of admission. In this patient, the antibiotics modification was from piperacillin-tazobactam plus azithromycin to amoxicillin-clavulanate due to the positive antigen test result.
There were no cases of inappropriate adjustment of antimicrobial therapy. Despite the positive result of urinary antigen, in the remaining 89 cases the treatment was not modified.
The urinary detection of pneumococcal antigen was widely used in our setting for the etiologic diagnosis of CAP. Sensitivity of the test varied according to the reference group used as the gold standard. It changed from 78.0% in pneumococcal pneumonias with definite diagnosis to 57.1% in the group of probable pneumococcal pneumonias. Taken together, the sensitivity of the test was 70.5%. These results are similar to those reported in previous studies.8- 13
It is known that in the pediatric population specificity is below 70% because children have high rates of nasopharyngeal colonization.19 In contrast, the number of false-positive test results in adults is lower even in patients with COPD, those with respiratory infections different from pneumonia, patients infected with HIV, and those with bacteremias caused by Streptococcus species different from pneumococcus.9- 11,13,20 We attempted to analyze this parameter using patients with nonpneumococcal pneumonia as the control group. Specificity was high and reached 96%, findings similar to those of other studies.9- 11 According to this specificity, the high positive predictive value and the high positive LR obtained in all reference groups, we think that the test, when the findings are positive, has a great value for the diagnosis of pneumococcal pneumonia. There were 2 patients with a urinary antigen test classified as false positive. They could have had a polymicrobial pneumonia in which S pneumoniae was involved as we documented in 3 other mixed infections with isolation of pneumococcus. Mixed infections represent approximately 5% to 10% of CAP in several studies.21,22 Most coinfective pathogens in these cases are viral agents that do not usually require specific antimicrobial therapy. These microorganisms are not routinely investigated, but the underestimation of its role in CAP seems to have little importance in therapeutic matters.22,23
We also calculated the diagnostic accuracy of the test using all patients without a diagnosis of pneumococcal pneumonia, including those with aspiration pneumonia and pneumonia of unknown etiology. With this method, the specificity, negative predictive value, and positive LR are even higher. Nevertheless, it is possible that some cases of aspiration pneumonia or CAP of unknown etiology could be caused by an undiagnosed S pneumoniae. In addition, if we include all patients with CAP of unknown etiology, because they had a negative result in the pneumococcal urinary antigen detection, the number of cases with a negative test result is higher, so it increases, maybe erroneously, the specificity, the negative predictive value, and positive LR of the test, causing an overestimation in its accuracy. For these reasons, we think that evaluating the accuracy using only the nonpneumococcal pneumonias as a control group is a more reliable approach.
Few studies reflect the usefulness of the pneumococcal urinary antigen detection in the current clinical practice. In a recent open-labeled controlled trial14 in immunocompetent young outpatients with positive pneumococcal urinary antigen test results, the results of a targeted therapy with amoxicillin were as effective as therapy with broader spectrum antibiotics.
In hospitalized patients, there is scarce information about pneumococcal urinary antigen-guided therapy. In these patients this test would allow physicians to use therapy directed against pneumococcus (pneumococcus-directed therapy if the test results are positive) and would enable them to switch to the oral route with amoxicillin when clinical stability would be achieved.
In a study of 59 patients with CAP, positive results of this test did not involve any modification in the empirical treatment; however, only 9 patients with a positive test result were included.24 Kobashi et al25 evaluated prospectively the use of the test in 168 patients admitted with CAP to a university hospital. Empirical treatment was modified in the 44 patients in whom the test results were positive, and they had acceptable outcomes. However, just 10 received a targeted antipneumococcal antibiotic like ampicillin. The remaining patients received broader spectrum antibiotics, such as carbapenems and respiratory quinolones, which makes it difficult to evaluate the usefulness of targeted therapy. The only randomized trial that tried to focus on this question failed to demonstrate a clinical benefit of targeted therapy based on urinary antigen test results. This study was designed to compare an empirical vs a targeted treatment on the basis of urine antigen results in hospitalized patients with CAP. The authors15 did not find differences between these 2 strategies in terms of clinical outcomes or economic benefits. Moreover, they observed some clinical relapse in patients who received a targeted therapy that might be due to uncovered agents in some potential mixed infections. However, the main limitations of this study were the low number of patients with positive pneumococcal antigen test results included (25 patients in the targeted treatment arm) and the 2 to 6 days of broad spectrum antibiotic therapy given before the targeted therapy that makes it difficult to assess the effect of this strategy.16
In our study, patients started the targeted therapy during the first to third day after admission, and the good clinical outcomes assessed in the 41 patients who received a favorable adjusted therapy suggest that a targeted therapy for pneumoccoccus is a valid strategy to treat inpatients with CAP. We also did not observe any relapse in the 23 patients with an optimal adjustment.
Taking into account that current usefulness of the pneumoccoccal antigen test allowed the clinicians to use a narrow antimicrobial spectrum with good clinical outcomes, potentially the antimicrobial optimization could have been possible in the 75 patients diagnosed exclusively by the pneumococcal urinary antigen test. Moreover, the adjustment could have also been performed in all 130 patients with pneumococcal pneumonia in whom the test results were positive because it provides results faster than conventional microbiological methods.
Patients with immunosuppressive conditions are often excluded from studies of adult CAP. In our study we included all patients with CAP, even those with immunosuppressive conditions (20.3% of the total population). Because of the high proportion of these patients, we think that our results could also be applicable to this high-risk population.
Our study had some limitations. We did not study the duration of test positivity after pneumococcal infection. Previous studies have shown that 50% to 70% of patients have positive test results 4 to 6 weeks after the pneumonia episode.10- 20 In our cohort, this matter could have little relevance because there were just 10 patients who experienced more than 1 episode of CAP, and the time between episodes was over 3 months in all cases. Another possible limitation of our study is that specific diagnostic tests were performed according to the attending physician, and consequently we did not have complete microbiological data for all patients included in the study. Because patients with missing data are not included in the analysis of the accuracy of the pneumococcal urinary antigen test and urinary antigen detection for S pneumoniae was performed in a very high proportion of the patients (80.8%), we think that the possible bias should not significantly affect the final results of our study. Although our study was not designed as a randomized clinical trial, the prospective analysis of an important number of patients with CAP allows us, in our opinion, to suggest the clinical effectiveness of the pneumococcal urinary antigen detection.
In conclusion, because of its high specificity, positive predictive value, and positive LR, we think that the urinary detection of pneumococcal antigen is a useful tool in the treatment of adult inpatients with CAP. When findings are positive, it allows clinicians to optimize antimicrobial therapy with good clinical outcomes. In our opinion, this test should be incorporated into clinical guidelines at the same level as classic microbiological studies because it can supplement, but not replace, their results.
Correspondence: Roger Sordé, MD, Department of Infectious Diseases, Hospital Universitari Vall d'Hebron, Universitat Autònoma de Barcelona, Passeig de la Vall d’Hebron 119-129, 08035 Barcelona, Spain (firstname.lastname@example.org).
Accepted for Publication: July 29, 2010.
Published Online: September 27, 2010. doi:10.1001/archinternmed.2010.347
Author Contributions: Drs Sordé and Falcó had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Sordé, Falcó, Lowak, and Pahissa. Acquisition of data: Sordé, Falcó, Lowak, Domingo, Ferrer, Burgos, Puig, and Cabral. Analysis and interpretation of data: Sordé, Falcó, and Len. Drafting of the manuscript: Sordé and Falcó. Critical revision of the manuscript for important intellectual content: Sordé, Falcó, Lowak, Domingo, Ferrer, Burgos, Puig, Cabral, Len, and Pahissa. Statistical analysis: Sordé and Len. Study supervision: Falcó and Pahissa.
Financial Disclosure: None reported.
Funding/Support: This study was institutionally supported by the Spanish Network for Research in Infectious Diseases (REIPI), RD06/008, from the Ministry of Science and Innovation, “Instituto de Salud Carlos III.”
Previous Presentation: This study was presented at the 49th Interscience Conference on Antimicrobial Agents and Chemotherapy; September 13, 2009; San Francisco, California (abstract L1-997).