INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The causes of community-acquired pneumonia (CAP) are difficult to establish, among other reasons, because of the poor sensitivity of diagnostic tests in noninvasive respiratory samples to identify the causative pathogen (1). Blood cultures are infrequently positive and serologic diagnoses are too late to be of therapeutic use. Invasive sampling of the lower airways is reserved for selected cases and is considered unacceptable for epidemiological studies and for routine use in all patients with CAP. Streptococcus pneumoniae is the leading pathogen of CAP even among groups of subjects stratified by risk (4) but the methods available have failed to show uniform diagnostic sensitivity or specificity. The development of the polymerase chain reaction (PCR) for gene amplification has enabled detection of low numbers of pathogens in clinical samples; in two studies that used pneumolysin PCR as a diagnostic method for pneumococcal pneumonia (12, 13), acute phase sera from patients with blood culture-positive pneumococcal pneumonia were positive by the PCR. To our knowledge, however, no studies have been carried out to determine the clinical usefulness of PCR in patients with CAP. On the other hand, Chlamydia pneumoniae has emerged as a significant pulmonary pathogen in patients with CAP (14, 15) but given the high seroprevalence of C. pneumoniae in adults, the real implication of this organism in CAP has hitherto been unknown. This prompted us to perform a prospective study to assess the origins of CAP in patients admitted to a tertiary hospital using, among other techniques, pneumolysin PCR in noninvasive respiratory samples for the detection of S. pneumoniae in sera and PCR for C. pneumoniae and Mycoplasma pneumoniae in throat swab specimens.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study population consisted of 184 patients with the diagnosis CAP admitted to a tertiary hospital between January 20, 1996 and January 20, 1997. The hospital has a specialist in respiratory diseases on duty. The clinical diagnosis of CAP included evidence of a new chest radiographic infiltrate, and at least one of the major and two of the minor criteria as follows. Major criteria were: fever > 37.8° C, cough, or sputum production; minor criteria were: dyspnea, pleuritic chest pain, and leukocytosis 10,000/mm³ or leukopenia < 4,000/mm³. Patients were hospitalized following the recommendations of the American Thoracic Society (16). Patients with immunosuppression or those who had been hospitalized in the previous 15 d were excluded.

The following variables were recorded in each case: age, gender, underlying disease, smoking habit (> 20 cigarettes/d), alcohol abuse (> 80 g/d), antibiotic treatment prior to hospital admission, signs and symptoms at admission, results of laboratory tests, blood gas analysis, radiological findings, complications related to pneumonia episode, admission to the intensive care unit, length of hospital stay, and mortality. A follow-up visit was done 30 d after the patient's discharge.

On the first day of admission, blood, sputum, and throat swab samples were collected for microbiologic identification of ordinary causative pathogens of respiratory tract infection and, particularly, of S. pneumoniae, M. pneumoniae, C. pneumoniae, Chlamydia psitaci, Legionella pneumophila, Coxiella burnetii, influenza A and B, adenovirus, and parainfluenza 1, 2, and 3. Serum samples for the detection of DNA from S. pneumoniae were frozen at 20° C and codified until analysis. Detection of pneumococcal DNA was performed by a microbiologist (J.C.) who was blind to results of the other tests. Throat swab specimens were preserved in 1 ml transportation medium for chlamydias (Culturette; Becton Dickinson, Cockeysville, MD). Other invasive diagnostic procedures, such as protected specimen brush and bronchoalveolar lavage, were only performed by indication of the physician in charge of the patient.

Before initiating antibiotic treatment, two blood samples were drawn for aerobic and anaerobic blood cultures (Bac-Alert; Organon Teknika, Boxtel, The Netherlands). The plates were incubated for 7 d and identification of isolates and antimicrobial susceptibility testing were performed according to standard bacteriologic methods.

Blood samples for serological testing were obtained in the acute stage of illness and after 14 to 21 d of convalescence. Serum samples were tested in pairs. Complement fixation tests were performed for influenza viruses A and B, parainfluenza viruses 1, 2, and 3, and adenovirus. An indirect immunofluorescence antibody test was used to detect antibodies against C. burnetii and a microimmunofluorescence method to detect antibodies against C. pneumoniae and C. psittaci. The direct and indirect fluorescent antibody technique was used to test for L. pneumophila and an enzyme immunoassay for M. pneumoniae.

Only 134 patients were able to produce a sputum sample for culture. A gram stain was performed to assess the quality of the sample. The criteria for defining a good sputum sample included the presence of  25 polymorphonuclear leukocytes and less than 10 squamous epithelial cells (× 100 field). The culture results were accepted only as etiologically indicative if the growth from a good sputum sample was heavy or moderate (17).

For the extraction of pneumococcal DNA, 200 µl of lysis buffer (10 mM Tris-HCl [pH 8.1], 1 mM ethylenediaminetetraacetic acid [EDTA], 1 µg/µl of proteinase K) was added to 200 µl of serum, following incubation of 65° C for 90 min, DNA was extracted with phenol-chloroform-isoamyl alcohol (Sigma, St. Louis, MO), and DNA was precipitated with ethanol and dissolved in 100 µl of distilled water. Amplification was carried out in a final volume of 100 µl. The reaction mixture contained 50 mM KCl, 10 mM Tris-HCl (pH 9.0), 0.1% Triton X-100, and 1.5 mM MgCl₂; 0.2 mM deoxyribonucleoside triphosphate (dNTP); 50 pmol of primers; 0.5 U of Taq DNA polymerase (Promega, Madison, WI); and 20 µl of DNA extracted. Nested amplification was done using the outer primers Ia and Ib and the inner primers IIa and IIb that amplified a 348-bp and 208-bp region of the pneumolysin gene, respectively. Amplification was done with the following thermal cycles: one cycle at 94° C for 5 min, 30 cycles at 94° C for 1 min, annealing of the primers at 55° C for 1 min, synthesis at 72° C for 1 min, and final extension at 72° C for 6 min. Amplified products were visualized under ultraviolet light on agarose gel stained with 2% ethidium bromide. All samples were amplified twice and the results were the same. Each run of the test analyzed serum samples of the groups and two internal controls: a pneumococcal DNA preparation was used as positive control and sterile distilled water as negative control. General recommended procedures were followed to avoid contamination (18). The amplified products on the agarose gel detected in positive samples were confirmed by the nonradioactive hybridization assay (DNA Enzyme Immunoassay, GEN-ETI-K DEIA; Sorin Biomedica, Saluggia, Italy) (data not shown).

Before starting the study, the PCR assay was tested with several microorganisms. Following a protocol similar to that of Salo and coworkers (12), in order to evaluate the sensitivity and specificity of the test, pneumococcal strains, microorganisms that produced toxins similar to pneumolysin, such as Clostridium or Streptococcus group A, and other bacterial strains were tested. The PCR assay showed a band in the case of S. pneumoniae and negative results in the other bacteria (data not shown). During and after the study, sera from a control group were collected to further test the specificity of the PCR assay. This control group comprised 25 healthy blood donors and 77 patients with other respiratory disorders, without active infections during the last 2 mo, who attended the outpatient clinic for routine follow-up visits. Forty of the latter were recruited during the study, and after finishing this, in order to increase the size of the group, 37 additional patients were included. Their mean age was 60 ± 15 yr, and the underlying diseases were 21 chronic obstructive pulmonary disease (COPD), 7 asthma, 5 lung carcinoma, 10 cardiac disease, 8 tuberculosis, 8 CAP at least 2 mo before, and others. Their samples were coded and processed together with samples of pneumonia patients. The samples of the additional patients were also coded, and analyzed together with samples of other patients with infections not included in the study, to maintain blindness of the PCR assay reader (J.C.).

In the PCR assay for C. pneumoniae and M. pneumoniae, DNA extraction was carried out as described for pneumococcal DNA. In both cases, a single amplification was performed with the same reaction mixture using HM-1 and HR-1 primers derived from the region that codifies the 474-bp PstI restriction fragment (19) for C. pneumoniae, and MP-11 and MP-12 primers derived from the region that codifies the P1 adhesin gene (20) for M. pneumoniae. The thermal cycles included one cycle at 94° C for 5 min, 40 cycles (in case of C. pneumoniae) or 35 cycles (in case of M. pneumoniae) at 94° C for 1 min, annealing of the primers at 48° C (in case of C. pneumoniae) or at 55° C (in case of M. pneumoniae) for 1 min, synthesis at 72° C for 1 min, and final extension at 72° C for 6 min. Detection was performed by hybridization in a microtitration plate DEIA (Sorin Biomedica) according to the manufacturer's instructions.

The cause of pneumonia was classified as definitive when the following criteria were met: (1) blood or pleural fluid yielded a pathogen or pneumococcal DNA was detected by the PCR assay in blood samples; (2) fourfold IgG titer rise for C. pneumoniae, C. psittaci, L. pneumophila, M. pneumoniae, C. burnetii, and respiratory viruses, or an initial titer  1/256 for L. pneumophila, an IgM-positive titer for M. pneumoniae, or rise in IgM titer ( 1/32) for C. pneumoniae; and (3) isolation of a pathogen in protected specimen brush cultures yielding growth  10³ colony-forming units per milliliter (cfu/ml), or a primary pathogen (e.g., L. pneumophila) from respiratory samples. When two or more respiratory pathogens were found, a mixed causation of CAP was accepted.

Statistical Analysis

The SPSS statistical package was used for the analysis of data. Qualitative variables were compared with the chi-square test and continuous variables with the Student's t test or the Mann-Whitney U test. The level of significance was set at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

A total of 184 patients attended at the emergency room with clinical and radiologic features of pneumonia and admitted to the hospital were included. Table 1 shows the main characteristics of the study population. Clinical manifestations were fever in 84% of cases, cough in 83%, dyspnea in 58%, pleuritic pain in 43%, chills in 44%, joint and muscular pain in 22%, and sudden onset in 19%. Basal arterial blood gases at admission were Pa_O2, 63 ± 16 mm Hg; Pa_CO2, 38 ± 9 mm Hg; and pH 7.43 ± 0.06. Hematologic and biochemical results obtained in the emergency room were leukocyte count 15.5 ± 7.7 × 10⁹/L, neutrophils 81%, hematocrit 40%, blood urea nitrogen 23 ± 15 mg/dl, and sodium 136 ± 4 mmol/L. Eighty-one percent of the patients presented consolidation homogeneous or inhomogeneous confined to one lobe, 5.9% bilateral pneumonia, 10% pleural effusion, and the remaining patients patchy infiltrates occupying more than one lobe or interstitial pattern.

Microbiology

Etiologic diagnosis was achieved in 78 (42%) patients (single pathogen, 72; various pathogens, six). The diagnoses were obtained by blood culture in 10 cases, serological tests in 32, and protected specimen brush in three (Table 2). Some clinical characteristics and concomitant illness according to the causative pathogen of CAP and in the case of S. pneumoniae according to the diagnostic method used, blood culture or PCR assay alone, are shown in Table 3. In 177 patients with negative blood culture, PCR assay was positive in 36 and negative in 141, whereas in seven patients with positive blood culture, the PCR assay was positive in five and negative in two. Sixty patients had been previously treated with antibiotics: only one in the group with positive blood culture for S. pneumoniae, and 11 in the group with positive PCR assay for S. pneumoniae, the latter with a mean duration of 3.8 d (four macrolides, four penicillins or cephalosporins, one quinolone, and two others).

Valid sputum samples were obtained in a few patients and were not taken into consideration for the definitive diagnosis. These cultures yielded S. pneumoniae in six (in two cases isolated concomitantly by PCR), Haemophilus influenzae in six, Streptococcus viridans in three, and Pseudomonas aeruginosa in nine. The blood culture was positive in 10 patients (S. pneumoniae, seven; H. influenzae, three). In 41 patients, serum PCR assay for S. pneumoniae was positive (the blood culture was positive in five and negative in 36); however, in two of them with positive blood cultures, the PCR was negative. One patient with positive PCR assay for S. pneumoniae also showed Mycobacterium tuberculosis in the Lowenstein-Jensen culture medium. In all controls, the PCR assay for S. pneumoniae was negative (specificity of 100%). Patients with blood culture positive for H. influenza showed a negative result for PCR assay for S. pneumoniae.

In nine patients, throat swab specimens showed positive PCR assays for C. pneumoniae. All patients, but one, had IgG titers between 1/16 and 1/64 without fulfilling criteria of seroconversion. Moreover, in four of these patients other etiologic diagnoses included M. pneumoniae in two, S. pneumoniae in one, and influenza A in one. The PCR assay for M. pneumoniae was positive in three patients (one with seroconversion).

Two serological samples were obtained in 117 patients and this technique demonstrated the causative pathogen in 32.

In 16 patients, a fiberoptic bronchoscopy was performed to obtain invasive samples and the etiologic microorganisms found were P. aeruginosa in three patients with the protected specimen brush and L. pneumophila in one patient with the bronchoalveolar lavage.

Evolution and Outcome

The evolution of patients after the pneumonia episode was improvement in 169 (92%) cases and death in 15 (8%). The following antibiotics or combinations of them were given: third-generation cephalosporin + macrolide, 85.3%; macrolide alone, 8.2%; macrolide + ciprofloxacin or aminoglycoside, 2.7%; third-generation cephalosporin alone, 1.1%; and others. Antibiotic treatment was modified in 21 patients due to: microbiological results, 10; poor tolerance, three; therapeutic failure, five; interaction with other treatments, one; and due to unknown causes in two. Fever disappeared in 2.8 ± 2.4 d in the whole group and the mean length of hospital stay was 13 ± 9 d. Complications included septic shock (6%), pleural effusion (3.3%), adult respiratory distress syndrome (3.3%), radiological progression (2.7%), and superinfection by another microorganism (1%). Death was due to evolution of pneumonia and/or complications in 11 patients and to the underlying disease in four. The evolution of patients in relation to the causative pathogen is shown in Table 3. Evolution and outcomes in the group with positive blood culture for S. pneumoniae and the group of patients with a positive PCR assay alone are depicted separately and compared. The initial treatment in both groups was similar: third-generation cephalosporin + macrolide was the most frequent combination, 85.7% in positive blood culture versus 86% in the group with positive PCR assay alone, and the rest received macrolide or cephalosporin alone or other combination. Patients with positive blood cultures for S. pneumoniae showed a higher rate of complications, lower Pa_O2 at admission, more multilobar radiographic infiltrates, longer hospital stays, and required admission to the intensive care unit more frequently than patients with negative blood culture and positive PCR assay. In the group with positive PCR for S. pneumoniae alone fever disappeared in 2.3 ± 1.6 d (6.5 ± 2.4 d in patients with positive blood culture). Complications appeared in eight patients, only one of whom required admission to the intensive care unit, and causes of death were: septic shock, one; hepatic failure, one; respiratory distress, one; and cardiac arrest in one patient with COPD and ischemic heart disease.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The aim of this study was to determine the cause of CAP in patients admitted to the hospital and to assess the usefulness of the PCR in noninvasive respiratory samples for diagnosis of CAP in clinical practice. Because criteria for hospital admission were based on international recommendations, our study population is comparable to that of other studies (8, 9, 11, 15). S. pneumoniae was the most common causative microorganism followed by M. pneumoniae and, in contrast to previous studies carried out in the Mediterranean area (7), L. pneumophila was the etiologic agent in only one patient. The cause of CAP was multiple in six patients.

The main findings of this study are related to the use of the PCR assay for the detection of S. pneumoniae. Definitive diagnosis of pneumococcal pneumonia in noninvasive respiratory samples requires the isolation of this pathogen in blood samples, but blood cultures are infrequently positive (21). On the other hand, the sensitivity of pneumococcal capsular antigen detection in sputum and pneumococcal serology is low (24). The PCR assay for the detection of S. pneumoniae DNA in blood allowed us to increase the etiologic diagnosis of streptococcal pneumonia from seven to 41 cases. Recently, Salo and coworkers (12) and Rudolph and coworkers (13) in samples from patients with culture-proven pneumococcal bacteremia showed a sensitivity of 60 to 70% and a specificity of 94% for pneumolysin PCR. We decided to amplify the gene for pneumolysin because this toxin is species-specific, it is produced by all pneumococcal strains isolated from clinical samples, and because amplification of this gene was found more sensitive than the autolysin gene (12, 13). The six cases of false-positive findings in the study of Salo and coworkers (12) were attributed to laboratory contaminants. In our study, the PCR assay for S. pneumoniae increased the diagnostic yield from seven cases with positive blood culture to 41 patients. In all control subjects without evidence of pneumococcal infection, the PCR assay was negative (specificity of 100%). The unexpected finding of two cases of culture-positive pneumococcal pneumonia and negative results in the PCR assay may be due to degradation of the pneumococcal DNA associated with improper sample preparation, delayed performance of the assay, or the presence of specific inhibitors. In fact, the most appropriate methods for sample processing and amplification protocols have not yet been clearly established.

In our study, we considered a positive PCR assay for S. pneumoniae (without a positive blood culture) as diagnostic for several reasons: patients had CAP without other etiological agents; S. pneumoniae is the etiological agent most frequently found in CAP; they were treated as suggested by international guidelines and showed an evolution accordingly; patients with other etiological agents isolated from blood had negative results for PCR assay; and detection of pneumolysin in serum is possible from very few bacteria, even partially degraded by treatment. The nested PCR assay employed has a great sensitivity, it is able to detect 24 femtograms (fg) of DNA, corresponding to 10 bacteria (12, 13). Confirmation of the amplified fragments of DNA was demonstrated by a hybridization assay similar to that employed by Salo and coworkers (12). It is plausible that patients with prior infection with S. pneumoniae might show pneumolysin in serum, but we did not find this toxin in our control group, in which we selected patients visited as outpatients in our Pneumology Service, in order to recruit a group of similar characteristics to those of the group with pneumonia but without active infection during the past 2 mo.

The role of previous antibiotic therapy on the recovery rate of S. pneumoniae has been extensively documented. Marston and coworkers (25) found that infection with S. pneumoniae was demonstrated less frequently among patients with previous antibiotic treatment. We also found a lower percentage of previous antimicrobial therapy in patients with positive blood culture (14%) than in patients with positive PCR assay (31%), which further stresses the value of the PCR assay for the diagnosis of pneumococcal pneumonia, based on its higher ability to detect bacteria not viable for culture. A similar finding has been described with other tests to investigate the presence of antigens of S. pneumoniae, which remain positive after some days of antibiotic treatment (24).

Bacteremic pneumococcal pneumonia appears predominantly in patients with chronic or severe concomitant illness. In our study, patients with positive blood cultures for S. pneumoniae were older, presented more chronic underlying conditions (COPD and heart diseases), showed a higher rate of complications and a longer hospital stay, and required admission to the intensive care unit more frequently than patients with negative blood culture and positive PCR assay. These findings may suggest a low number of bacteria entering the circulation or rapid clearance of pneumococci by polymorphonuclear leukocytes in relation to a better immunological status as corresponding to a younger age group with a lower percentage of underlying cardiorespiratory disorders.

C. pneumoniae has become a frequent respiratory pathogen, accounting for 5 to 15% of episodes of CAP (6, 10, 14). Routine culture of C. pneumoniae from clinical specimens is difficult (26, 27) and the technique is not routinely used in many laboratories. Interpretation of serologic data may be problematic given the high prevalence of antibody titers in the adult population. It has been shown that a modified PCR assay that detects amplified DNA by enzyme immunoassay is more sensitive than culture for detecting C. pneumoniae in nasopharyngeal swab specimens from hospitalized patients with CAP (28). However, up to 2 to 5% of healthy populations may be asymptomatically infected with C. pneumoniae (29). In our study, DNA of C. pneumoniae in throat swab specimens was detected in nine patients, although criteria of seroconversion were met in only one; the remaining eight patients had variable titers and in four of them other etiologic diagnoses included M. pneumoniae in two, S. pneumoniae in one, and influenza A in one. Other studies have also documented a high rate of coinfection with other pathogens suggesting that C. pneumoniae may be present as a copathogen or as an opportunist. C. pneumoniae and S. pneumoniae as a dual infection results in more severe illness requiring longer hospital treatment than either C. pneumoniae or S. pneumoniae infection alone (5, 15).

M. pneumoniae causes a relatively large proportion of pneumonias among young persons but is also an important cause of pneumonia in older age groups (25). In our study, the use of the PCR assay in throat swab specimens showed a much lower sensitivity compared with conventional serologic methods (30). On the other hand, DNA detection of M. pneumoniae without fulfilling criteria of seroconversion criteria may be found in healthy carriers or in cases of previous infection.

In conclusion, the etiological diagnosis of pneumonia caused by S. pneumoniae, traditionally associated with a poor sensitivity in noninvasive samples, shows a fivefold increase using PCR in serum with respect to blood culture, although there were two false-negative cases. The main advantages of PCR assay are its rapidity as compared with conventional methods, and its higher sensitivity, even in patients treated with antibiotics, mainly due to the fact that it can detect nonviable pneumococci or their toxin. It may be useful in children in whom it is very difficult to obtain respiratory samples. However, its potential clinical usefulness is limited by complexities of sample preparation and detection methods. Identification of C. pneumoniae by PCR in throat swab specimens without fulfilling criteria for acute infection, may be considered a prior infection. The PCR assay in throat swab samples for the diagnosis of M. pneumoniae has a lower sensitivity than serologic methods.