The Etiology and Antimicrobial Susceptibility Patterns of Microorganisms in Acute Community-Acquired Lung Abscess

The Etiology and Antimicrobial Susceptibility Patterns of Microorganisms in Acute Community-Acquired Lung Abscess* Janet M.J. Hamrrwnd, MB, ChB, MD; Peter D. Potgieter, MB, ChB; David Hanslo, MB, ChB; Helen Scott, Dip Med Tech; and Denise Roditi, MB , ChB, M Med (Path) Objective: To detennine the spectrum and antibiotic susceptibility patterns of microorganisms causing acute community-acquired lung abscess. Design: A prospective survey. Setting: Medical emergency department and wards of a tertiary teaching hospital. Patients: Thirty-four adult patients with both clinical and radiologic features compatible with a diagnosis of acute community-acquired lung abscess who had received less than 48 h of antibiotic therapy. Interventions: Microbiologic specimens obtained by percutaneous lung aspiration and with a protected specimen brush via fiberoptic bronchoscopy were submitted for aerobic and anaerobic culture. Main outcome measures: Identification of all microorganisms, including anaerobes, and detennination of antibiotic susceptibility. Results: A mean of 2.3 bacterial species per patient was isolated, anaerobes alone being isolated in 44% of cases, aerobes alone in 19%, and mixed aerobic and anaerobic isolates in 22%. Aerobic Gram-negative pathogens were uncommon. In seven patients, Mycobacterium tuberculosis was identified; in two it was associated with other bacteria. In four patients, no organisms were isolated. All the nonmycobacterial isolates were susceptible to amoxicillin-clavulanate

Jn a third-world setting, acute community-acquired

lung abscess is still associated with dental caries, poorly controlled epilepsy, previously damaged lungs, and alcohol abuse; in more developed populations, however, this disease is now uncommon except in patients immunocompromised by drugs or disease or as a postobstructive complication. 1•2 Both anaerobic and aerobic organisms colonizing the oropharynx have been implicated as the major pathogens in this type of infection. In studies in which the lung abscess is nosocomial in origin or secondary to disease other than as*From the Respiratory Intensive Care Unit, Groote Schuur Hospital, and the D epartments of Medical Microbiology, Red Cross Children's Hospital and Groote Schuur Hospital, and University of Cape Town, South Africa. Supported by the South African Medical Research Council. Manuscript received February 7, 1995; revision accepted May 18.

and in addition the anaerobes were all susceptible to chloramphenicol and almost all to a combination of penicillin and metronidazole. Among the anaerobes, the level of resistance to penicillin, metronidazole, and clindamycin individually was 21%, 12%, and 5%, respectively. Conclusions: Community-acquired acute lung abscess is usually caused by multiple anaerobic and less frequently aerobic Gram-positive microorganisms, which should respond to empirical therapy with amoxicillinclavulanate, chloramphenicol, or a combination of penicillin and metronidazole. Tuberculosis, which may be indistinguishable from an acute lung abscess, occurred in 21% of patients in our study. Most bacterial pathogens are sensitive to conventional antimicrobial therapy and further investigation with percutaneous lung aspiration or bronchoscopy is indicated only when there is lack of early response to therapy or there is the presence of atypical clinical features. (CHEST 1995; 108:937-41) MIC=minimal inhibitory concentration; PLA=percutaneous lung aspirate; PSB=protected specimen brush

Key words: anaerobic microorganisms; diagnosis; lung abscess; lung aspirate; microbiology; therapy

piration, a greater number of aerobic organisms, particularly Enterobacteriaceae, have been found. 2-6 Anaerobic organisms have been identified as the most common pathogens in studies in which percutaneous lung or transtracheal aspiration techniques have been used. 7-9 Owing to the potential contamination by oropharyngeal organisms, the sputum from a patient with a lung abscess may give misleading culture results, and this has necessitated the use of more invasive procedures to obtain representative specimens for culture. 10 As a result of the clinical use of the sputum culture as guide to antibiotic prescribing, a combination of intravenous penicillin, metronidazole, and an aminoglycoside is widely used for the therapy of acute lung abscesses. Although effective, this regimen requires the CHEST I 108 I 4 I OCTOBER, 1995

937

Table !-Underlying Risk Factors for Lung Abscess Risk Factor

No. of Patients

Alcohol abuse History of aspiration Dental cmies Old fibrocavitatmy pulmonary tuberculosis Epilepsy Drug abuse (not intravenous) Chronic obstructive lung disease Systemic lupus erythematosus Bronchial carcinoma Nil

13 10 9 7 3 l l

4

administration of intravenous antibiotics, thereby prolonging hospitalization with its attendant increased costs, the potential for drug toxic reactions, and the promotion of the development of antimicrobial resistance and bacterial superinfection. It may, in addition, provide unnecessary broad-spectrum cover for aerobic Gram-negative microorganisms. With the advent of newer (often more expensive) antibiotics and concerns about the emergence of penicillin resistance, even the conventional triple broad-spectrum therapy has been brought into question and it has been suggested that clindamycin might be the therapy of choice. 8 This study was designed to assess the etiology and antimicrobial susceptibility patterns of the microorganisms responsible for acute community-acquired lung abscesses in an attempt to provide guidelines for the most cost-effective investigation and antimicrobial therapy of these patients. METHODS

All adult patients with clinical and radiologic features compatible with a diagnosis of acute lung abscess and who required admission to the emergency unit or medical wards of our 1,200-bed teaching hospital from January 1992 until D ecember 1993 were included following informed consent. Patients who had been hospitalized within the past month or who had received antimicrobial therapy for more than 48 h before the special investigations could be performed were excluded. Patients who had sputum positive for Mycobacterium tuberculosis and those considered at high risk for complications of the diagnostic procedure, because of potential hypoxemia or an increased risk of hemorrhage, were also not enrolled in the study. The study was approved by the Ethics Committee of the University of Cape Town. A history of recent onset of malaise, fever, cough productive of sputum, compatible clinical examination, and a chest radiograph that showed an intrapulmonary cavity ;vith an air-fluid level were required for the diagnosis of lung abscess, which was confirmed in all patients by the investigators (J.M.J.H., P.D.P. ). The investigations were performed in the bronchoscopy-suite theater. The lesion was localized in two planes using fluoroscopy, and a percutaneous lung aspirate (PLA) was performed using a 22-gauge needle with the patient under local anesthesia. The aspirate (usually 0.5 to 2 mL) was injected into 2 mL sodium thioglycolate broth that was immediately transported to the laboratory. The patient then underwent fiberoptic bronchoscopy (Olympus BF-B2) under topical anesthesia. The airways were inspected and a sample taken using a protected specimen brush (PSB) and fluoroscopic localization of the lesion. The brush was then cut off into

938

2 mL sodium thioglycolate broth, and the specimen was immediately transported to the laboratory. Processing of both specimens was commenced within 15 min of the specimens having been taken. On receipt of the specimens in the laboratory, 0.1-mL aliquots of the sample and of a 1:100 dilution were used for culture. Each aliquot was inoculated onto chocolate blood agar, blood agar with added gentamicin (5 mglmL), and MacConkey agar. Plates were incubated in air \vith 8% C0 2 . In addition, aliquots were inoculated onto two Wilkins-Chalgren agar plates supplemented with 10% horse blood-one with 10 mg!L nalidixic acid and the other with 2.5 mg!L vancomycin and 75 mg!L neomyci n added; these plates were placed in an anaerobic jar with palladium catalyst that was then evacuated and filled with a mixture of 10% H 2, 10% C0 2, and 80% N2 . The jar was transported to nearby laboratmy where plates were transferred to an anaerobic chamber (Jouan 440M) for further processing. Growth on all plates was assessed as +(scanty), ++(moderate), or +++( heavy) but all microorganisms cultured from PSB and needle aspirate specimens were considered pathogenic. Aerobic and facultative organisms were identified by standard procedures. Susceptibility testing on these organisms were performed according to national committee for clinicallaboratmy standards criteria. 11 The following breakpoint levels were used to determine susceptibility: penicillin, 4 mg!L; metronidazole, 16 mg!L; clindamycin, 4 mg!L; chloramphenicol, 16 mg!L; and amoxicillin/clavulanate, 8/4 mg!L, respectively. Anaerobic organisms were screened for their ability to grow under anaerobic conditions only. Each isolate was further subjected to gas liquid chromatography (Chrompak 437A) and to a battery of biochemical tests (Minitek, BBL). For each isolate, susceptibility testing was performed using an agar diffusion method of minimal inhibitory concentration (MIC) determination (E-test; Biodisk). The antibiotics tested were penicillin, metronidazole, clindamycin, chloramphenicol, and amoxicillin-clavulanate. Numeric data are shown as mean::'::standard deviation. H ESULTS

There were 27 male and 7 female patients with a median age of37 years (range, 20 to 73 years) enrolled into this study over a 24-month period. All patients had received antibiotics prior to microbiologic sampling of the abscess, with 30 patients having received antibiotics for less than 24 h and 4 for less than 48 h. During the study period, 52 patients had been referred for inclusion, of whom 18 were excluded (6 patients had already been receiving antibiotics for more than 48 h, 6 patients were not included for logistic reasons, 4 patients had clinically suggestive tuberculous infection that was subsequently proved, 1 patient had achalasia, and 1 had a hiatus hernia that had been incorrectly diagnosed on the chest radiograph). No patients were excluded because of the severity of their disease. The average duration of respiratory symptoms was 40.6 days (median, 21 days ), with nine patients having had symptoms for more than 30 days. Sputum production was a prominent feature; 13 patients were producing copious amounts (>1 cup/d), 16 moderate amounts, 5 minimal (d tsp/d), and 1 patient no sputum. The nature of the sputum was considered to be fetid in 5 patients, purulent in 24, and mucoid in 4; moderate hemoptysis occurred in 5 patients, and severe hemoptysis in l. A temperature greater than 38.5°C was present Clinical Investigations

Table 2-Spectrum of Bacterial Isolates From Acute Lung Abscesses

Anaerobes Prevotella sp Porphyromonas sp Unspeciated pigmented anaerobes Bacteroides sp Fusobacterium sp Anaerobic cocci Microaerophilic streptococci Veilonella sp Clostridium sp Nonsporing Gram-positive anaerobes "Mixed anaerobes" Total Aerobes Viridans streptococci Staphylococcus sp Corynebacterium sp Klebsiella sp Haemophilus sp Gram-negative cocci Total

No. of Isolates

%

17 7 4 4 4 4 7 1 1 9 1 59

22

7 5 3

9 5 5 5 5 9

11 1 74 9

6

2

4 3

2 20

26

3

in 11 patients, temperature less than 38.5°C was present in 16, and minimal fever occurred in 7 patients. The clinical features in the seven patients with M tuberculosis were similar to the other patients; median age was 35 years (range, 23 to 58 years) vs 37 years (range, 20 to 73 years); median duration of symptoms was 14 days (range, 3 to 60 days) vs 21 days (range, 2 to 365 days); and the spectrum of sputum production was similar, hemoptysis in two and five patients, high fever in three and seven patients with mycobacterial and nonmycobacterial disease, respectively, but importantly, two had additional bacterial microorganisms cultured from the abscess. Table 1 reflects the risk factors for lung abscess present in the patients. The PLA, was well tolerated under local analgesia and proceeded without complications. Fiberoptic bronchoscopy was less well tolerated and localization of the abscess with the PSB proved more difficult than by needle aspirate. Endoscopic drainage also generally proved to be unreliable because of edema of the involved bronchi, except in those patients in whom the abscess was already draining. In one hypoxemic patient, fiberoptic bronchoscopy was abandoned, and in another patient this procedure was not attempted because of respiratory distress. In four patients with large (>6 em diameter) abscess cavities that were not draining adequately, external percutaneous tube drainage using a large Silastic chest drain (Malingkrodt) was performed; the drain was inserted over forceps with blunt dissection, and this procedure was followed by rapid resolution of symptoms in all patients. Thirty-three patients responded to therapy and survived to hospital discharge;

however, one patient with extensive disease failed to respond and died while in hospital. Microbiologic investigation was suboptimal in seven patients owing to incomplete anaerobic culture in four and problems during specimen transport in three. Nevertl1eless, 16 patients (47%) had 3 or more bacterial species isolated from the PLA and/or PSB, with 7 of these yielding 5 or more different strains. Mycobacterium tuberculosis was identified in seven patients, two of whom had other bacteria isolated as well. Apart from the patients with tuberculosis, aerobes only were cultured in 5 patients (19%), anaerobes only in 12 (44% ), a mixed grovvth of aerobes and anaerobes in 6 (22%), while no organisms were isolated in 4 (15%). A total of 79 bacterial species were isolated; the spectrum of microorganisms is shown in Table 2. Forty-three strains of anaerobes (excluding the microaerophilic Streptococcus milleri isolates) were available for the evaluation of antimicrobial MICs by the E-test method, and the results are shown in Table 3. Resistance to penicillin was detected in 21% of the isolates tested, and all but one of these were pigmented Prevotella or Porphyromonas species. Metronidazole resistance was demonstrated in two isolates of Prevotella and one isolate each of Porphyromonas, Eubacterium, and Proprionibacterium. Two of these isolates (one Prevotella melanogenica and one Porphyromonas asaccharolytica) were resistant to both penicillin and metronidazole. Two isolates were resistant to clindamycin; one Prevotella sp and one Clostridium sporogenes. DISCUSSION

The pathogenesis of acute lung abscess has changed over the past decade and in a recent description of this disease, 1 pneumonia (40% of cases), severely immunocompromising disease, including pulmonary metastases (27%), and primary bronchial carcinoma (17%) were the underlying causes. In our study, however, the patient population was similar to earlier descriptions of community-acquired lung abscesses. Most patients were men who had a significant history of alcohol abuse, severe dental caries, or poorly controlled epilepsy. One third of the patients had a clear history of unconsciousness or seizures and aspiration. Many of our patients showed chest radiographic features compatible with previous pulmonary tuberculosis and some had previously been treated for pulmonary tuberculosis. Seven patients (21%) also had tuberculous lung abscesses; none had clinical features to suggest active tuberculosis rather than lung abscess, which probably reflects the high incidence of tuberculosis in our community. We have previously reported that 5% of patients with community-acquired pneumonia have active tuberculosis; likewise, a high incidence is reported from Singapore (21 %) and Hong Kong (12%), CHEST 1108 I 4 I OCTOBER, 1995

939

Table 3--The Antimicrobial Susceptibility Patterns of 43 Anaerobic Isolates From Acute Lung Abscess MIC Range, mg!L*

Prevotella!Porphyromonas sp (n =24) Fusobacterium sp (n=4 ) Other Gram-negative organisms' (n=3) Gram-positive bacilli! (n=9) Anaerobic cocci (n=3) All isolates (n=43)

Pe nicillin

Metronidazole

Clindamycin

Chloramphe nicol

Amoxicillin!Clavulanate

<0.01- >32 (8 ) <0.01-0.06

<0.01- >32 (3) <0.01-0.5

<0.01 -8 (1) <0.03-0.5 (0) <0.01-0.06 (0) <0.01-16 (1) <0.01-0.25 (0) <0.01-16 (2) 5%

<0.01-8 (0) 0.03-1

<0.01-1 (0) <0.01-0.06 (0) 0.01-0.25 (0) <0.01-0.5 (0 ) 0.06-0.5

(O)

(O)

0 .01->32 (1) <0.01-0.12

0.5-8 (0) <0.01->32 (2) 0.06-8 (0) <0.01-.32 (5) 12%

(O) 0.01-2 (0) <0.01- >32 (9) 21 %

(O) 2-4 (0) 0.12-8 (0) 0.5-2 (0) <.01-8 (0)

(O) <0.01-1

(O)

*Numbe rs in parentheses are numbe r of resistant isolates. 1Bacteroides sp (2), Veillonella sp (1). !Eubacterium sp (3), Lactobacillus sp (3), Clostridium sp (1), Bifidobacte rium sp (1), Propionibacterium sp (1 ).

where a similar experience in the etiology of acute c.o~munity-acquired lung abscess might thus be antlclpated.1~-14 Only five patients in our study presented without any of the acknowledged risk factors for lung abscess. Four of these patients were young women with a short history of fever, cough, and halitosis, typical of a lung abscess. In three, acid-fast bacilli and in one Staphylococcus aureus were detected from the needle aspirate. All three had radiographic features typical of acute lung abscess with single air-fluid levels being found in a dependent segment of the lung with no other evidence of pulmonary tuberculosis. The HIV status of these patients was subsequently checked and in all cases was negative. The remaining patient, an elderly man, was found to have an obstructing carcinoma of the bronchus. Any atypical clinical findings associated with a lung abscess would thus support an early, aggressive diagnostic approach in our population; and depending on the clinical setting, either bronchoscopy or a PLA should be performed. In most abscesses in our study, multiple anaerobes and aerobes were isolated. Anaerobes accounted for 74% of the bacterial yield, and apart from the patients with tuberculosis, the bacterial growth obtained consisted of anaerobes alone in 52% and of aerobes alone in only 22%. This conforms to the microbiology of previously reported studies in which careful anaerobic culture of suitable specimens has been performed. 7•8 Marina et al 15 found multiple microorganisms with an average of 4.1 anaerobes and 3.0 aerobes per transtracheal aspirate in a retrospective analysis of 51 patients, with anaerobes the only pathogen in 33 specimens and of these, 16% were 13-lactamase producers. We found resistance to penicillin, clindamycin, and metronidazole among the anaerobes tested; however, 940

all were susceptible to amoxicillin-clavulanate and chloramphenicol, and almost all to a combination of penicillin and metronidazole. Accordingly, an empirical antibiotic regimen that provides cover for the most likely pathogens should be selected from one of the above antibiotics or antibiotic combinations. The use of amoxicillin-clavulanate would permit the early institution of oral antibiotic therapy, which could reduce the requirements for inpatient care. It is important, however, to be aware that lung abscesses with different mechanisms of pathogenesis may have a different spectrum of etiologic microorganisms. In a recent study using PLA in 49 patients, in which 20% of the patients had underlying carcinoma, an average of 3.25 bacterial species per patient was obtained, but only 58% were anaerobes, which suggests a different spectrum of organisms when the abscess is unrelated to aspiration. 2 This was confirmed in another study investigating the bacteriology of obstructive pneumonitis, using similar methods, in which a higher incidence of Gram-negative microorganisms was noted in association with postobstructive infections. 6 In our study, the low incidence of Gram-negative infections and rarity of an obstructing carcinoma causing a lung abscess were in contrast to the findings in these studies and a recent report evaluating 52 patients with lung abscess in which 36.5% had an associated bronchogenic carcinoma. 16 The causative organisms have also been noted to differ in children, where more aerobic microorganisms, particularly S aureus have been isolated, reflecting the different risk factors for abscess in this population. 3 In a study in which 40% of the patients had acquired their abscesses in hospital, most patients had aerobic organisms (predominantly Pseudorrwnas aeruginosa), suggesting that hospitalization may also have influenced their illness. 17 Clinical Investigations

The incidence of pneumothorax reported from the studies evaluating PLA has ranged from 3 to 14%, but in this study, we had no complications using either PLA or fiberoptic bronchoscopy and a PSB. 2.1 8 We found the PLA to be quick and easy to perform, the procedure less uncomfortable and more acceptable to the patients, the abscess to be more easily localized, and a greater microbiologic yield was obtained from the specimens than with PSB. The selection of appropriate empirical antibiotics that are likely to cover all the multiple anaerobic and aerobic microorganisms that cause community-acquired lung abscess with the preceeding caveats is therefore possible. Invasive microbiologic investigation of acute community-acquired lung abscess is not indicated on a routine basis unless there are atypical features or the patient fails to respond to therapy. REFERENCES

1 Pohlson EC, McNamara JJ, Char C, et al. Lung abscess: a changing pattern of the disease. Am J Surg 1985; 150:97-101 2 Peiia Griiian N, Munoz Lucena F, Vargas Romero J, et al. Yield of percutaneous needle lung aspiration in lung abscess. Chest 1990; 97:69-74 3 Tumwine JK. Lung abscess in children in Harare, Zimbabwe. East Afr Med J 1992; 69:547-49 4 Neild JE, Eykyn SJ, Phillips I. Lung abscess and empyema. Q J Med 1985; 57:875-82 5 Mori T, Ebe T, Takahashi M, et al. Lung abscess: analysis of 66 cases from 1979 to 1991. Intern Med 1993; 32:278-84

6 Liaw Y, Yang P, Wu Z, et al. The bacteriology of obstructive pneumonitis. Am J Respir Crit Care Med 1994; 149:1648-53 7 Beerens H , Tahon-Castel M. Infections humaines a bacteries anaerobies non toxigenes. Brussels: Presses Academiques Europeenes, 1965; 91-114 8 Gudiol F, Manresa F, Pallares R, et al. Clindamycin vs penicillin for anaerobic lung infections: high rate of penicillin failures associated with penicillin-resistant Bacteroides melanogenicus. Arch Intern Med 1990; 150:2525-29 9 Bartlett JG, Gorbach SL, Finegold SM. The bacteriology of aspiration pneumonia. Am J Med 1974; 56:202-07 10 Bartlett JG. Anaerobic bacterial infections of the lung and pleural space. Clin Infect Dis 1993; 16(suppl 4):S248-55 ll National Committee for Control of Laboratory Standards. Antimicrobial susceptibility testing. 3rd ed. Villanova, Pa: NCCLS, 1991; 10:7 12 Prout S, Potgieter PD, Forder AA, et al. Acute communityacquired pneumonias. S Afr Med J 1983; 64:443-46 13 Chan CHS, Cohen M, Pang J. A prospective study of community-acquired pneumonia in Hong Kong. Chest 1992; 101:442-46 14 Hui KP, Tan WC, Chan TB, et al. Community-acquired pneumonia in the Far East [abstract]. Chest 1993; 103:1637 15 Marina M, Strong CA, Civen R, et al. Bacteriology of anaerobic pleuropulmonary infections: preliminary report. Clin Infect Dis 1993; 16(suppl 4):S256-62 16 Sosenko A, Glassroth J. Fiberoptic bronchoscopy in the evaluation of lung abscesses. Chest 1985; 87:489-94 17 Irwin RS, Garrity FL, Erickson AD, et al. Sampling lower respiratOJy tract secretions in primary lung abscess: a comparison of the accuracy of four methods. Chest 1981; 79:559-65 18 Chen CH, Kuo ML, Shih JF, et al. Etiologic diagnosis of pulmonary infection by ultrasonically guided percutaneous lung aspiration. Chung Hua IHsueh Tsa Chih 1993; 51:333-39

CHEST I 108 I 4 I OCTOBER, 1995

941