Vol.

96 (2002) 178 ^185

Lung abscess in adults: clinical comparison of

immunocompromised to
non-immunocompromised patients
N. MANSHARAMANI*, D. BALACHANDRAN*w, D. DELANEY*, J. D. ZIBRAK*,
R.C. SILVESTRI* AND H. KOZIEL*

*Division of Pulmonary and Critical Care Medicine, Department of Medicine, Beth Israel Deaconess Medical Center,
and Harvard Medical School, Boston, MA and wDepartment of Medicine, University of Chicago, Chicago, IL, U.S.A.

Abstract Information related to the clinical characteristics and isolated microbes associated with lung abscesses
comparing immunocompromised (IC) to non-immunocompromised (non-IC) patients is limited. A retrospective
review for1984^1996 identi¢ed 34 consecutive adult cases of lung abscess (representing 0
2% of all cases of pneumonia),
including 10 non-IC and 24 IC patients. Comparison of age, gender, tobacco use, pre-existing pulmonary disease
or recognized aspiration risk factors were not signi¢cantly di¡erent between the two groups.Upper lobe involvement
accounted for the majority of cases, although multi-lobe involvement was limited to IC patients. There were no
di¡erences in the need for surgical intervention, and mortality was very low for both groups. Anaerobes were the most
frequent isolates for non-IC patients (30%), whereas aerobes were the most frequent isolate for IC patients (63%). Im-
portantly, certain organisms were exclusively isolated in the IC group and multiple isolates were obtained only from the
IC patients.Thus, comparing non-IC to IC patients, clinical characteristics may be similar whereas important di¡erences
may exist in the microbiology associated with lung abscess. These ¢ndings have important implications for the clinical
management of these patient groups, and support a strategy to aggressively identify microbial agents in abscess
material.r 2002 Elsevier Science Ltd
doi:10.1053/rmed.2001.1247, available online at http://www.idealibrary.com on

Keywords lung abscess; immunocompromised; pneumonia.

INTRODUCTION scess comparing immunocompetent to immunosup-

The clinical and microbiological characteristics of pa-
tients with lung abscess are well described (1,2). Most
cases are associated with recognized risk factors includ-
ing alcohol abuse, periodontal disease, seizure disorders,
oesophageal disorders and neurological bulbar dysfunc-
tion (3). Historically, lung abscesses are associated with
anaerobic bacteria (1). Recent reports suggest, however,
that the spectrum of aetiological agents may be evolving
(2), and that changes may in part re£ect an increased
number of immunosuppressed patients (2,4). Although
a number of reports suggest an expanded spectrum of
aetiological agents in immunosuppressed patients (3^7),
the clinical and microbiological characteristics of lung ab-
Received 8 January 2001, accepted in revised form 22 October 2001and
published online 14 January 2002.
Correspondence should be addressed to: Henry Koziel, MD, Division of
Pulmonary and Critical Care Medicine, Beth Israel Deaconess Medical
Center; KBS-23, 330 Brookline Avenue, Boston, MA 02215, U.S.A. Fax:
617 667- 4849; E-mail: hkoziel@caregroup.harvard.edu
pressed patients has not been fully examined.
Recognizing the altered immune function and spec-
trum of lung infections in patients with chronic corticos-
teroid use (8 ^10), haematological or solid malignancies
(11,12), organ transplantation (12,13) and human immuno-
de¢ciency virus (HIV) infection (14 ^16), the clinical char-
acteristics and microbiology of lung abscess may be
di¡erent when compared to patients without underlying
immunosuppression. One report suggest fewer anaero-
bic isolates in immunocompromised patients (2). A
recent report of HIV-infected patients noted that
abscesses occurred in advanced HIV-related disease,
was associated with a broad spectrum of pathogens,
responded poorly to antibiotics and was associated
with a poor prognosis (4). Furthermore, the immune
dysfunction associated with conditions such as
diabetes mellitus (17,18) may predispose to certain pul-
monary infections (19), although the impact of diabetes
mellitus on lung abscess has not been fully investigated
(20 ^22).
LUNG ABSCESS IN ADULTS
Identifying di¡erences in the clinical manifes-
tations and microbiological isolates of lung abscesses
may have signi¢cant implications for guiding clinical diag-
nostic evaluations and empirical antimicrobial manage-
ment, especially in the immunocompromised host.
However, the clinical characteristics and aetiology of
lung abscess in immunocompromised hosts is not well
characterized, and studies comparing immunocompro-
mised to non-immunocompromised patients are limited.
The purpose of this study was to examine the clinical
characteristics and microbiology in cases of lung abscess
at a tertiary care urban medical center, comparing
immunocompromised to non-immunocompromised
patients.
METHODS
Setting
The Beth Israel Deaconess Medical Center, West
Campus, is a 375-bed tertiary care referral hospital
with general medical service, including infectious
disease and HIV clinical services, medical oncology and
hematology service, diabetic service, and pulmonary
and critical care services. Surgical services included gen-
eral surgical, thoracic surgical and solid organ transplant
service.
Identi¢cation of cases of lung abscess
Consecutive cases of lung abscess were identi¢ed
by a systematic search of all medical discharge summa-
ries with the appropriate International Classi¢cation
of Diseases-9 (ICD-9) discharge diagnosis, review of
radiological reports and surgical/anatomical pathology
reports identifying the diagnosis of lung abscess for
the period 1984 ^996. Only cases of lung abscess in
patients with clinical and radiographical evidence
(conventional chest radiograph, thoracic CT scan)
(23,24), or surgically obtained specimens were included
in the study. Lung abscess was de¢ned as a cavity
X2 cm in diameter. Patients with lung abscess secondary
to recent (o30 days) thoracic surgery, or patients with
cavitating, necrotic lung neoplasms were not included in
this study.
A retrospective review of medical records of all iden-
ti¢ed cases was performed, and all clinical data were re-
corded on a standardized form, with particular attention
to patient age, gender, ethnicity, underlying medical con-
ditions (including seizure disorder, CVA or neuromuscu-
lar disease), previous pulmonary infections, recognized
aspiration risks, alcohol and tobacco use. For the pa-
tients infected with HIV, additional information included
the most recent peripheral blood CD4+ T-lymphocyte
counts and use of antimicrobial chemoprophylaxis. For
each identi¢ed case, information was recorded for symp-
179
toms at time of presentation, prior antimicrobial use,
radiographic localization of abscess, mode of diagnoses
of lung abscess, microbiological agent identi¢ed from ab-
scess material, cultures of respiratory tract specimens
or blood, requirement for surgical intervention, duration
of hospitalization and mortality.
For the purpose of the study patients were divided
into two groups: (1) non-immunocompromised group,
which included patients with no clinically recognized un-
derlying medical conditions associated with immunosup-
pression; and (2) immunocompromised group, which
included patients with clinically recognized immunocom-
promised conditions, including human immunode¢ciency
virus type-1 (HIV-1) infection, organ transplant recipi-
ents, persons requiring chronic corticosteroids (de¢ned
as the equivalent of prednisone X20 mg daily for X2
months) or other immunosuppressive agents for in£am-
matory conditions, persons with underlying hematologi-
cal or solid malignancy, and persons with diabetes
mellitus.
Statistical analysis
Statistical analysis of non-continuous dichotomous data
were compared by the Chi-square test (with Yates cor-
rection) or the Fischer’s exact test as appropriate using
INSTAT2 statistical package (Graphpad Software; San
Diego, CA, U.S.A.) on an IBM PS/2 120 MB computer
(IBM Corp; Armonk, NY, U.S.A.). All P-values were
two-sided and statistical signi¢cance was accepted for
Po0
05.
RESULTS
Clinical characteristics
A total of 46 cases of lung abscess were identi¢ed, and
medical records for all 46 cases were reviewed. Twelve
cases were excluded because the identi¢ed abscess re-
presented lung necrosis complicating post-obstructive
pneumonia, cavitating lung neoplasm, or the available
medical records were incomplete. A total of 34 cases
were included in this study, which represented a 0
2%
incidence of lung abscess for all hospitalized cases of
pneumonia. The clinical characteristics for the 10 indivi-
duals in the non-immunocompromised group and 24 indi-
viduals in the immunocompromised group are presented
in Table 1. For the immunocompromised group, all HIV+
individuals and the organ transplant reciptient were con-
¢rmed HIV sero-negative. Although speci¢c serum HIV
antibody data were not available for the non-immuno-
compromised subjects, none had recognized HIV risk
factors, all had normal peripheral white blood cell counts
and none had lymphopenia.
Comparison of age, gender, tobacco use, pre-existing
pulmonary disease or recognized aspiration risk factors
180 RESPIRATORY MEDICINE

TABLE 1. Clinical characteristics of patients with lung abscess

Non-immunocompromised Immunocompromised P-value

Total cases (n) 10 24
Age, yrs (mean7SD) 59
1717 50
1718 NS
Gender
Male (n) 7 20 NS
Female (n) 3 4 NS
Ethnicity
Caucasian 8 18
Black F 2
Hispanic F 3
Asian F 1
Unavailable 2 F
Immunocompromised condition
HIVdisease (n) F 7
Malignancy (n) F 7
Diabetes mellitus (n) F 5
Chronic corticosteroid use (n) F 4
Liver transplant recipient (n) F 1
Tobacco use, n (%) 7 (70%) 16 (67%) NS
Pre-existing pulmonary disease*, n (%) 3 (30%) 6 (25%) NS
Identi¢ed aspiration riskw, n (%) 5 (50%) 11 (46%) NS
Antibiotics during the1month prior to diagnosis 10 (100%) 24 (100%) NS
*COPD, asthma or bronchiectasis.
w
Ethanol use, seizure disorder, peridontal disease, bulbar neurological or oesophogeal disease.

for lung abscess were not signi¢cantly di¡erent between group were prescribed prophylactic antimicrobials (as
the two groups. For the non-immunocompromised noted above) prior to the diagnosis of lung abscess.
group, 20% of patients had prior remote bacterial pneu-
monia compared to 54% for the immunocompromised
group (P40
05). All subjects were prescribed antimicro- Clinical signs and symptoms of lung abscess
bials at the time of diagnosis of lung abscess. All subjects
Clinical signs and symptoms at the time of presentation
had used antimicrobials within 3 months of the diagnosis
are provided inTable 2.There were no statistically signi¢-
of lung abscess, and included cephalosporins (n=13),
cant di¡erences comparing non-immunocompromised
macrolides (n=10), semisynthetic penicillins (n=7) and
to immunocompromised patients. For the non-immuno-
combination penicillin/sulbactam (n=4). There were no
compromised patients, pleuritic chest pain (40%) was
di¡erences in the use of speci¢c antimicrobials compar-
the most common symptom at presentation. For the im-
ing non-immunocompromised to non-immunocompro-
munocompromised patients, fever (46%) was the most
mised subjects.
common symptom. A larger proportion of non-immuno-
For the immunocompromised group, in addition to
compromised patients were asymptomatic (30%) at the
the ¢ve patients with diabetes mellitus, diabetes mellitus
time of presentation, compared to immunocompro-
was a co-existing medical condition in two individuals re-
mised patients (4%), although this di¡erence was not sta-
quiring chronic corticosteroids, and in each of one pa-
tistically signi¢cant. Gastrointestinal symptoms
tient with HIV infection, haematological malignancy and
(diarrhoea, nausea and vomiting) were only reported in
liver transplant recipient. For the HIV-infected patients,
the immunocompromised group (0% vs. 29%, P=0
07).
all were con¢rmed HIV+, and the mean peripheral blood
CD4+ T-lymphocyte count was 66
2756 cells mm3
(range 20 ^157 cells mm3), 4/7 were prescribed anti-
Radiographic localization of lung abscesses
P. carinii prophylaxis (two oral trimethoprim-sulfa-
methoxozole, two aerosolized pentamadine), and 4/7 The radiographic localization of lung abscesses is pre-
had prior opportunistic infections.None of the other im- sented inTable 3. For each lobar distribution, there were
munocompromised patients had prior opportunistic in- no statistically signi¢cant di¡erences in the radiographic
fections. Only patients in the immunocompromised localization of the lung abscess comparing the non-
LUNG ABSCESS IN ADULTS 181

TABLE 2. Symptoms and signs attime of presentation for patients with lung abscess

Non-immunocompromised Immunocompromised P-value

(%) (%)
Fever 20 46 NS
Cough 20 42 NS
Gastrointestinal symptoms* 0 29 0
07
Dyspnoea 20 21 NS
Weight loss 10 21 NS
Pleuritic chest pain 40 12 NS
Haemoptysis 10 8 NS
Asymptomatic 30 4 NS
Total WBC 105 (cells mm3) 12
074 12
778 NS
*Nausea, vomitting, diarrhoea.
WBC=white blood cell count.

TABLE 3. Radiographic location of lung abscesses at presentation

Non-immunocompromised Immunocompromised P-value

RUL, n (%) 5 (50%) 8 (33%) NS
LUL, n (%) 2 (20) 5 (21) NS
LLL, n (%) 1 (10) 3 (12) NS
RLL, n (%) 0 3 (12) NS
Multiple lobes, n (%) 0 3 (12) NS
RML, n (%) 2 (20) 2 (8%) NS
Total 10 (100%) 24 (100%)
RUL: right upper lobe; LUL: left upper lobe; LLL: left lower lobe;RLL: right lower lobe;RML: right middle lobe.

immunocompromised group to the immunocompro-
mised group. Collectively, the upper lobes accounted for
the majority of cases, with 70% of non-immunocompro-
mised patients and 54% of immunocompromised pa-
tients (P40
05). For both groups, the right upper lobe
(RUL) was more often involved than the left upper lobe
(LUL), although this di¡erence was not statistically signif-
icant. Multiple lung abscesses (Xtwo sites) were ob-
served only in the immunocompromised group,
occurring in 12% of cases.
Diagnostic approach to determine microbial
aetiological agent
All patients had at least one diagnostic procedure in an
attempt to identify a pathogen (Table 4). A surgical speci-
men most often provided material which identi¢ed a po-
tential pathogen for each group. Bronchoscopic
specimens, including site directed bronchoalveolar la-
vage (BAL) or trans-bronchial lung biopsy, identi¢ed a
pathogen in six cases, including one non-immunocom-
promised patient and ¢ve immunocompromised patients
(one included transbronchial biopsies). Quantitative cul-
tures were not performed on BAL specimens. For indivi-
duals with positive blood cultures, the same microbe was
isolated from the blood and respiratory tract specimens.
Only two patients (5
9%), both immunocompromised,
allowed only examination of sputum specimens and re-
fused more invasive investigation. For six individuals, no
potential pathogen was isolated despite sputum and
blood cultures, bronchoscopy, CT-guided needle biopsy
or surgical biopsy. All specimens were submitted for rou-
tine gram stain, routine bacterial culture, Legionella spp.
culture, fungal culture, acid fast bacillus (AFB) stain and
culture, and P. carinii stain. Anaerobic culture were per-
formed on 490% of the submitted specimens.
Microbial isolates associated with
lung abscesses
Identifying the most de¢nitive diagnostic procedure for
each case, the microbes isolated from the patients with
lung abscesses are presented inTable 5.Compared to the
non-immunocompromised group, a trend towards a sig-
ni¢cantly larger proportion of cases with isolated aero-
bic bacteria was observed for the immunocompromised
182 RESPIRATORY MEDICINE

TABLE 4. Diagnostic method* which identi¢ed a potential pathogen in patients with lung abscess

Non-immunocompromised Immunocompromised Total

Surgical material, n (%) 4 8 12 (35)
Bronchoscopy, n (%) 1 5 6 (18)
BAL (n) 1w 4
Transbronchial biopsy (n) 0 1
CT-guided needle aspiration (n) (%) 1 4 5 (15)
Blood cultures, n (%) 1z 2z 3 (8
8)
Sputum culture, n (%) 0 2} 2 (5
9)
No agent identi¢ed, n (%) 3 3 6 (18)
*Re£ectthe methods utilized to base clinical decisions regarding antimicrobial therapy.
w
Bronchoscopic ¢nding supported by serum immunoglobulin levels.
z
Same organisms isolated from blood cultures and respiratory tract specimens.
}
Both patients refused any further invasive investigations.

TABLE 5. Microbial isolates associated with lung abscesses*

Non-immunocompromised Immunocompromised
Aerobic bactera, n (%) 2 (20) 15 (63)w
S. aureus 1 5
P. aerugenosa F 3
Hemophilus spp. (non-in£uenza) F 2
P. maltiphilia 0 1
H. in£uenza F 1
b-Streptococcus 1 1
Enterobacter spp. F 1
K. oxytoca F 1
Anaerobic bacteria, n (%) 3 (30) 2 (8)
Actinomyces spp. 1 2
P. acnes 1 F
Peptostreptococcus spp. 1 F
Atypical bacteria, n (%) 0 (0) 2 (8)
Legionella spp. (non-pneumonphila) F 2
Mycobacterium, n (%) F 1 (4)
M. avium complex F 1
Fungi, n (%) 2 (20) 5 (21)
Aspergillus spp. 2 4
Candida spp. F 1
No organism identi¢ed, n (%) 3 (30) 3 (8)
*Some abscess specimens yielded multiple organisms.
w
P=0
057.

group (20% vs. 63%; P=0
057), whereas no signi¢cant dif-
ferences were observed for the isolation of anaerobic,
Legionella spp., mycobacteria or fungi.For the non-immu-
nocompromised group, anaerobes were most frequently
isolated, followed equally by aerobic bacteria and fungi.
For the immunocompromised patients, aerobic bacteria
were most commonly isolated, representing a broad
spectrum with P. aerugenosa, Hemophilis spp. (non-in£u-
enza), X. maltophilia, H. in£uenza, Enterobacter spp. and
K. oxytoca exclusively isolated in the immunocompro-
mised group.Other micro-organisms exclusively isolated
in specimens from immunocompromised patients in-
cluded non-pneumophilia Legionella spp., M. avium com-
plex and Candida spp. The patient with Candida spp. had
yeast isolated from surgical specimens of multiple pul-
monary abscesses and high-grade fungaemia.
Multiple isolates (Xtwo micro-organisms) were ob-
tained in seven cases, all in immunocompromised pa-
tients.Whereas anaerobes were isolated in 20% of cases
for non-immunocompromised patients, no anaerobes
LUNG ABSCESS IN ADULTS
TABLE 6. Clinical management and outcomes of patients with lung abscess
Non-immunocompromised
Medical management only, n (%)
Surgical management, n (%)
Discharged, n (%)
Hospital mortality, n (%)
were isolated for the immunocompromised group,
although this did not achieve statistical signi¢cance.
There were no cases with Mycobacterium tuberculosis
identi¢ed in either group. Although a larger proportion
of non-immunocompromised cases failed to identify a
potential pathogen (30% vs. 8%), this did not achieve sta-
tistical signi¢cance.
Recognizing the limitations of isolating microbes from
specimens of sputum and BAL as the aetiology of lung
infections, the current study revealed the following: (1)
for the non-immunocompromised group, only one pa-
tient had BAL as the most de¢nitive diagnostic proce-
dure. The BAL specimen yielded Aspergillus spp. and this
patient also had positive serology for Aspergillus antigen;
(2) for the immunocompromised group, the two patients
who had only sputum specimens as the most de¢nitive
diagnostic procedure, one yielded Klebsiella oxytoca and
Hemophilus spp. (non-in£uenza), and one yielded Actino-
myces and Hemophilus spp. (non-in£uenza). For the ¢ve
patients with BAL as the most de¢nitive diagnostic pro-
cedure, the isolated organisms included E. coli (co-iso-
lated from three specimens), X. maltophilia, S. aureus
and Enterobacter spp. For both groups, all sputum and
BAL specimens also co-isolated C. albicans, but none of
the isolates were considered pathogenic unless C. albi-
cans was also identi¢ed in blood cultures, or specimens
from surgical material or CT-guided needle aspiration.
Overall, for the non-immunocompromised patients, a
potential pathogen was identi¢ed in 70% of all cases. For
the immunocompromised patients, potential microbial
diagnoses were identi¢ed in 88% of all cases. Excluding
sputum cultures, and recognizing the possible limitations
of BAL specimens (even in immunocompromised pa-
tients), the yield was 63^79% (excluding or including
BAL results, respectively).
Clinical management and outcomes
All patients were prescribed antimicrobial agents, includ-
ing speci¢c agents with activity directed against isolated
microorganisms, for a period of 4 ^ 6 weeks.For the non-
immunocompromised group, medical management
alone was employed for 50%. All cases receiving surgical
treatment represented patients speci¢cally referred to
the Beth Israel Deconess Medical Center (BIDMC) for
183
Immunocompromised
5 (50) 15 (63)
5 (50) 9 (37)
10 (100) 22 (92)
0 (0) 2 (8)
surgical evaluation, and indications included concern for
possible cavitating malignancy (n=3) and haemoptysis
(n=2).Overall for the group, all patients were discharged
from the hospital.
For the immunocompromised group, medical man-
agement alone was employed for 63%. For patients re-
ceiving surgical treatment, most were initially referred
to the BIDMC surgical service, and indications included
progressive abscess enlargement despite antimicrobial
treatment (n=4), suspicion for cavitating malignancy
(n=2), haemoptysis (n=2) and pneumothorax (n=1). Over-
all mortality during the hospitalization was 8% (one pa-
tient with Aspergillus abscess and one patient with C.
albicans abscess died) and the remaining 92% of patients
were discharged.
Three-month follow up data were available for 17
(50%) patients. For the four in the non-immunocompro-
mised group, two had complete resolution of the ab-
scess, whereas for two other cases there was no
resolution (one of these two required thoracoplasty for
haemopytsis). For the 13 immunocompromised patients,
12 had complete resolution of the abscess and one pa-
tient was admitted with a new cavity in the opposite lung
at 3 months.
DISCUSSION
This retrospective review of consecutive cases of lung
abscess in adults at tertiary care medical center suggests
that certain important clinical similarities as well as im-
portant clinical di¡erences exist comparing immuno-
compromised patients to patients without recognized
immunocompromised states. For both non-immuno-
compromised and immunocompromised patients, the
identi¢ed cases were predominantly male patients
(70 ^ 83%), with a mean age of 50 ^59 years, with a his-
tory of tobacco use (67^70%) and with underlying pul-
monary disease (25^30%). Recognized aspiration risks
were identi¢ed equally in both groups, present in 46 ^
50% of cases.
Several important clinical di¡erences were suggested
in this study.The observation that twice as many cases of
lung abscess were identi¢ed in the immunocompromised
group compared to the non-immunocompromised
group suggests that primary lung abscess may be more
184
common in immunocompromised hosts, although data
to allow independent incidence determination for each
group was not available. Whereas most non-immuno-
compromised patients were asymptomatic or experi-
enced pleuritic chest pain at the time of presentation,
most immunocompromised patients experienced fever
and cough, although these di¡erences did not achieve
statistical signi¢cance when the two groups were com-
pared. Gastrointestinal symptoms were exclusively ob-
served in the immunocompromised group.
The radiographical localization of lung abscess demon-
strated that both non-immunocompromised and immu-
nocompromised groups had predominantly upper lobe
involvement (70% vs. 54%, respectively), similar to other
reports (2,4,25). Although lower lobe involvement was
observed more frequently with immunocompromised
patients (24% vs.10%), this di¡erence did not achieve sta-
tistical signi¢cance. Multiple-lobe involvement was iden-
ti¢ed only in immunocompromised patients.
A potential aetiological agent was identi¢ed in the ma-
jority of cases for both non-immunocompromised (70%)
and immunocompromised (63^ 88%) patients. For the
non-immunocompromised group aerobic bacteria (30%)
were most frequently isolated microbes, followed
equally by anaerobic bacteria (20%) and fungi (20%). Simi-
larly, for the immunocompromised group aerobic bacter-
ia (71%) were the most frequent isolates, followed by
fungi (21%), ¢ndings similar to immunocompromised pa-
tients in other reports (2,4). Importantly, immunocom-
promised patients did not have anaerobic isolates, were
more likely to have multiple pathogens isolated from ab-
scess material, and exhibited a broad spectrum of poten-
tial pathogens such as P. aerugenosa, Hemophilis spp. (non-
in£uenza), X. maltophilia, H. in£uenza, Enterobacter spp.,
K. oxytoca, non-pnuemophilia Legionella spp., M. avium
complex and Candida spp., which were exclusively iso-
lated in the immunocompromised group.
We observed a high rate of recovery of microbial iso-
lates from clinical specimens, where potential pathogens
were identi¢ed in 82
4% of all cases included in the study.
The high rate of recovery may in part re£ect the aggres-
sive approach to identify potential pathogens at our insti-
tution. For the bronchoscopic specimens, although
quantitative cultures were not performed, isolation of
potential pathogens from the airways of immunocom-
promised individuals often are considered pathogenic
(26). Recognizing that microbes isolated from sputum
specimens may not re£ect true pathogens, exclusion of
these cases yielded an overall rate of 76
5% isolation of
potential pathogens. Although the isolated microbes
may represent colonization rather than true infection,
the observation that individuals clinically responded to
speci¢c antimicrobial therapy directed against the iso-
lated organisms in part supports a causative role. Finally,
we observed a broad spectrum of isolated microbes,
especially from the immunocompromised group, many
RESPIRATORY MEDICINE
isolates which would not respond to most empirical anti-
microbial regimens (27).These observations suggest the
importance of identifying a potential causative agent,
and support a strategy to aggressively identify aetiologi-
cal agents from abscess material.
The low rate of recovery of anaerobes for both groups
in the current study contrasts the ¢ndings of prior re-
ports (1,28,29) where anaerobes were isolated in up to
93% of cases (exclusively in 56% of cases). As 490% of
specimens in the current study were submitted for anae-
robic culture, the low incidence of anaerobic isolation in
the current study does not re£ect omission of anaerobic
cultures, but may re£ect sampling methods (no trans-
tracheal aspirations were performed in the current
study), lapses in appropriate handling of respiratory tract
specimens, or changes in the microbiological spectrum in
our study populations, perhaps re£ecting the state of
immunocompetence. Alternatively, the spectrum of mi-
crobes may be altered due to the high rate of premorbid
antibiotic prescriptions for both groups. Importantly, a
recent report of lung abscess in HIV-infected individuals
failed to identify anaerobes (4), which taken together
with the ¢ndings in the current study suggests that the
microbial aetiology for immunocompromised patients
maybe distinct from non-immunocompromisedpatients.
The observed mortality associated with lung abscess
was low for both groups, and lower than the reported
mortality rates of 15^28% (2,30 ^32). The favourable
outcomes in the current study may in part re£ect the
patient population, the aggressive management (employ-
ing surgical resection in many cases), and the use of anti-
microbial agents directed against all isolated microbial
agents.The only observed deaths occurred in the immu-
nocompromised patients, and both cases involved fungal
abscess. Although de¢nitive conclusions are not possible,
other investigators have reported high mortality rates
associated with fungal lung abscess (25,33).
Other limitations of this study include the retrospec-
tive nature and the relatively small number of cases iden-
ti¢ed in each group. The limited number of cases
identi¢ed despite a comprehensive review of medical re-
cords over a 13-year period suggests that the develop-
ment of lung abscess may be infrequent, and may re£ect
a change in the evolution of lung abscess, or may re£ect
antimicrobial use patterns, clinical practices or patient
populations speci¢c to our institution. The absence of
statistical di¡erences comparing the two groups may re-
£ect the fact that di¡erences do not exist, or perhaps
the small number the non-immunocompromised pa-
tients likely limited the statistical analysis comparing
these groups. The relatively limited number of each of
the immunocompromised patients did not allow for sub-
group analysis, such as determining the in£uence of dia-
betes mellitus. The inclusion of only cases requiring
hospitalization may under-estimate the total number of
actual cases of lung abscess at our institution, as ambula-
LUNG ABSCESS IN ADULTS
tory patients were not included. In the absence of a stan-
dardized approach to the diagnostic and therapeutic
management of patients included in this retrospective
study, conclusions or recommendations regarding the
appropriate management practice is not possible.
Although a broad spectrum of immunocompromised
states were represented, including HIV disease, malig-
nancy, diabetes mellitus, chronic corticosteroid use, and
organ transplantion, the ¢ndings may not apply to other
immunocomromised patients. Finally, the experience at
our institution may not re£ect the experience of other
medical centers, although a broad spectrum of medical
conditions were represented in this study.
In summary, this retrospective review of lung ab-
scesses comparing non-immunocompromised patients
to immunocompromised patients notes that the clinical
characteristics of lung abscess are remarkably similar
comparing the two groups, with a favourable short-term
prognosis. However, several observations suggest that
important di¡erences may exist, as immunocompro-
mised patients with lung abscess may present more often
with gastrointestinal symptoms, more often have multi-
ple abscesses, more often have aerobic bacteria isolated
from abscess material, and may have exclusive isolation of
certain microbes or multiple microbes, although addi-
tional larger studies are needed to further assess for
these di¡erences. The observed broad spectrum of mi-
crobial isolates associated with lung abscesses for each
group has important implications for the clinical and em-
pirical management of these patients, and supports a
strategy to aggressively identify causative agents
Acknowledgements
These data were presented in part at the1998 American
Thoracic Society/American Lung Association Interna-
tional Conference in Chicago, IL, U.S.A.
REFERENCES
1. Bartlett J, Finegold S. Anaerobic infections of the lung and pleural
space. Am Rev Respir Dis 1974; 110: 56^77.
2. Pohlson EC, McNamara JJ, Char, C et al. Lung abscess: a changing
pattern of the disease. Am J Surg 1985; 150: 97^101.
3. Wiedemann HP, RiceTW. Lung abscess and empyema. SeminThor-
acic Cardiovasc Surg 1995; 7: 119^128.
4. Furman AC, Jacobs J, Sepkowitz KA. Lung abscess in patients with
AIDS.Clin Infect Dis 1996; 22: 81^ 85.
5. Ernst A,Gordon FD, Hayek J, et al. Lung abcess complicating Legio-
nella micdadei pneumonia in an adult liver transplant recipient.
Transplantation 1998; 65: 130 ^134.
6. Ruutu P,ValtonenV, Elonen E, et al. Invasive pulmonary aspergillosis:
a diagnostic and herapeutic problem.Clinical experience with eight
haematologic patients. Scand J Infect Dis 1987; 19: 569^575.
7. Bauling PC, Weil R, Schroter GP. Legionella lung abscess after renal
transplantation.J Infect 1985; 11: 51^55.
8. Slade JD, Hepburn B. Prednisone-induced alterations of circulating
human lymphocyte subsets.J Lab Clin Med 1983; 101: 479^ 487.
185
9. Walzer PD, LaBine M, RedingtonTJ, et al. Lymphocyte changes dur-
ing chronic administration of and withdrawal form corticosteroids:
relation to Pneumocystis carinii pneumonia. J Immunol 1984; 133:
2502^2508.
10. Ten Berge RJM, Sauerwein HP, Young SL, et al. Administration of
prednisolone in vivo a¡ects the ratio of OKT4/OKT8 and the
LDH-isoenzyme pattern of humanT lymphocytes.Clin Immunol Im-
munopathol 1984; 30: 91^103.
11. Siminski J, Kidd P, Phillips GD, et al. Reversed helper/suppressor T-
lymphocyte ratio in bronchoalveolar lavage £uid from patients with
breast cancer and Pneumocystis carinii pneumonia. Am Rev Respir Dis
1991; 143: 437^ 440.
12. Collin BA, Ramphal R. Pneumonia in the compromised host includ-
ing cancer patients and transplant patients. Infect Dis Clin N Am
1998; 12: 781^ 805.
13. Hughes W, Smith B. Provocation of infection due to Pneumocystis
carinii by cyclosporin A.J Infec Dis 1982; 145: 767.
14. Murray J, Mills J. Pulmonary infectious complications of human im-
munode¢ciency virus infection. Part I. Am Rev Respir Dis 1990; 141:
1356^1372.
15. Murray J, Mills J. Pulmonary infectious complications of human im-
munode¢ciency virus infection. Part II. Am Rev Respir Dis 1990; 141:
1582^1598.
16. Bartlett JG. Pneumonia in the patient with HIV infection. Infect Dis
Clin N Am 1998; 12: 807^ 820.
17. Glass EJ, Stewart J, Matthews DM, et al. Impairment of monocyte
‘lectin-like’ receptor activity in Type I (insulin-dependent) diabetic
patients. Diabetologia 1987; 30: 228 ^231.
18. Geisler C, Almdal T, Bennedsen J, et al. Monocyte functions in dia-
betes mellitus. Acta Pathol Microbiol Immunol Scand1982; 90: 33^37.
19. Koziel H, Koziel MJ. Pulmonary complications of diabetes mellitus:
pneumonia. Infect Dis Clin N Am 1995; 9: 65^96.
20. Cheng DL, Liu YC, Yen MY, et al. Septic metastatic lesions of pyo-
genic liver abscess.Their association with Klebsiella pneumonia bac-
teremia in diabetic patients. Arch Intern Med 1991; 151: 1557^1559.
21. Jerng JS, Hsueh PR, Teng LJ, et al. Empyema thoracis and lung ab-
scess caused by viridans streptococci. Am J Respir Crit Care Med
1997; 156: 1508 ^1514.
22. LimTK, ChanTB. Sudden death from primary lung abscess in mid-
dle-aged diabetic menFtwo case reports. Singapore Med J 1989;
30: 102^104.
23. Bartlett JG, Gorbach SL, Tally FP, et al. Bacteriology and treatment
of primary lung abscess. Am Rev Respir Dis 1974; 109: 510 ^518.
24. Stark DD, Federle MP, Goodman PC, et al. Di¡erentiating lung ab-
scess and empyema: radiography and computed tomography. Am J
Roentgenol 1983; 141: 163^167.
25. Mori T, EbeT,Takahashi M, et al. Lung abscessFanalysis of 66 cases
from 1979 to1991. Intern Med 1993; 32: 278 ^284.
26. Stover DE, Zaman MB, Hajdu SI, et al. Bronchoalveolar lavage in the
diagnosis of di¡use pulmonary in¢ltrates in the immunosuppressed
host. Ann Intern Med 1984; 101: 1^7.
27. Bartlett JG. Antibiotics in lung abscess. Semin Respir Infect 1991; 6:
103^111.
28. Bartlett JG. Anaerobic bacterial infections of the lung and pleural
space.Clin Infect Dis 1993; 16: S248 ^S255.
29. Hammond JMJ, Potgieter PD, Hanslo D, et al.The etiology and anti-
microbial susceptability patterns of microorganisms in acute com-
munity acquired lung abcess.Chest 1995; 108: 937^941.
30. Hirshberg B, Sklair-Levi M, Nir-Paz R, et al.Factors predicting mor-
tality of patients with lung abscess. Chest 1999; 115: 746 ^750.
31. Harber P, Terry PB. Fatal lung abscesses: review of 11 years experi-
ence. South Med J 1981; 74: 281^283.
32. Hagan JL, Hardy JD. Lung abscess revisted: a survey of 184 cases.
Ann Surg 1983; 197: 755^762.
33. Grinan NP, Lucena FM, Romero JV, et al.Yield of percutaneous nee-
dle lung aspiration in lung abscess.Chest 1990; 97: 69^74.