Note: This copy is for your personal, non-commercial use only.
To order presentation-ready copies for
distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
CT Appearance of Pyogenic
Liver Abscesses Caused by
Klebsiella pneumoniae1
Hind S. Alsaif, MD
Sudhakar K. Venkatesh, MD, FRCR
Douglas S. G. Chan, MD
Sophia Archuleta, MD
Materials and
1
From the Department of Diagnostic Imaging (H.S.A.,
S.K.V.), Division of Microbiology and Laboratory Medicine
(D.S.G.C.), and Division of Infectious Diseases, University
Medicine Cluster (S.A.), National University Health System,
5 Lower Kent Ridge Rd, Level 2, Main Bldg, National Uni-
versity Hospital, Singapore 119074. Received September 29,
2010; revision requested November 11; revision received
December 15; accepted December 22; final version
accepted January 6, 2011. Address correspondence to
S.K.V. (e-mail: sudhakar_k_venkatesh@nuhs.edu.sg).
q
RSNA, 2011
Radiology: Volume 260: Number 1—July 2011 n radiology.rsna.org
ORIGINAL RESEARCH
n GASTROINTESTINAL IMAGING
Purpose: To retrospectively compare the computed tomographic
(CT) features of liver abscesses caused by Klebsiella pneu-
moniae with those caused by other bacterial pathogens.
This retrospective study was approved by the institutional
Methods: review board, with waiver of informed consent. Hospital
records of all patients with a diagnosis of liver abscess be-
tween July 2003 and July 2010 were retrieved from an elec-
tronic hospital database. One hundred and thirty-one con-
secutive patients with confirmed pyogenic liver abscesses
were studied. Data on clinical presentation, comorbid con-
ditions, septic hematogenous complications, hospitalization
duration, and abscess-related mortality were obtained. CT
characteristics of abscesses including number, distribution,
unilocular or multilocular appearance, cystic or solid ap-
pearance, gas in cavity, pylephlebitis, thrombophlebitis,
and pneumobilia were reviewed. Etiology was established
by pus and/or blood culture. Patients were placed into a
monomicrobial K pneumoniae liver abscess group and a
comparison group. A comparison of the CT features and
clinical findings between the two groups was performed.
The x2 analysis or Fisher exact test was used for categorical
variables, and Student t and log-rank tests were used for
continuous variables. A P value of less than .05 was consid-
ered to indicate a significant difference.
Results: Monomicrobial K pneumoniae liver abscesses were pre-
sent in 92 cases (70.2%). On CT images, characteristics
more likely to be associated with monomicrobial K pneu-
moniae liver abscesses than other pyogenic liver abscesses
were a single abscess (79.3% vs 56.4%, P = .01), unilobar
involvement (82.6% vs 61.5%, P = .01), solid appearance
(57.6% vs 35.9%, P = .03), multilocular (94.6% vs 71.8%,
P = .01), association with thrombophlebitis (30.4% vs
5.1%, P , .01), and hematogenous complications (28.3% vs
7.7%, P , .01). Thrombophlebitis was associated with
higher incidence of hematogenous septic complications
(50.0% vs 13.9%, P , .001). Monomicrobial K pneumoniae
liver abscesses were associated with significantly shorter
duration of antibiotic treatment (P = .018) and hospital
stay (P = .005), but there was no significant difference in
incidence of septic shock and abscess-related mortality as
compared with other pyogenic liver abscesses.
Conclusion: Monomicrobial K pneumoniae liver abscesses appear as
single, solid, or multiloculated liver abscesses and are as-
sociated with thrombophlebitis and septic hematogenous
complications.
q
RSNA, 2011
Supplemental material: http://radiology.rsna.org/lookup
/suppl/doi:10.1148/radiol.11101876/-/DC1
129
GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses
L
iver abscess is a serious intraabdom-
inal infection that may be caused by
bacteria, fungi, or parasites. Until
the end of the last century, pyogenic liver
abscesses were predominantly caused by
mixed aerobic and anaerobic bacteria,
the most frequent isolate being Escheri-
chia coli (1). In the 1990s, some Asian
countries reported Klebsiella liver ab-
scess as an increasingly common infec-
tious disease (2), while accumulating
data from Taiwan and Korea suggested
that it has unique endemic features and
may be attributed to invasive Klebsiella
pneumoniae strains with capsular K1
subtype (3). In Western countries, K pneu-
moniae now accounts for a quarter of
all pyogenic liver abscesses (4). The ex-
act cause for increasing prevalence of
K pneumoniae liver abscesses in Asia
is not known but may be related to the
Asian population or emergence of a dif-
ferent strain of K pneumoniae with an
increased propensity to cause liver ab-
scesses (4). Another report (5) suggests
that K pneumoniae has emerged as the
most common isolate in the United
States and Taiwan. This distinctive epi-
demiologic change with a substantial
rise in the incidence of K pneumoniae
liver abscesses are now being widely re-
ported from other Asian countries, such
as Japan, Singapore, Korea, India, and
Hong Kong, as well as non-Asian coun-
tries, such as Spain, England, Trinidad,
Australia, and the United States (6).
Advances in Knowledge
n Compared with other pyogenic
liver abscesses, monomicrobial
Klebsiella pneumoniae liver
abscesses commonly appear as
single (79.3% vs 56.4%, P = .01),
solid (57.6% vs 35.9%, P = .03),
or multilocular (94.6% vs 71.8%,
P = .01) focal liver lesions on
CT images.
n Monomicrobial K pneumoniae
liver abscesses are significantly
associated with thrombophlebitis
(30.4% vs 5.1%, P , .01) and
hematogenous septic complica-
tions (28.3% vs 7.7%, P , .01)
as compared with other pyogenic
liver abscesses.
130
The clinical manifestation of K pneu-
moniae liver abscesses is similar to that
of other types of liver abscesses (4).
However, there are some notable dif-
ferences. While biliary disease is the
most likely predisposing condition for
pyogenic liver abscesses, K pneumoniae
liver abscesses are more often cryp-
togenic (7). Moreover, K pneumoniae
liver abscesses are more frequently as-
sociated with hematogenous spread of
infection as compared with pyogenic
abscesses caused by other bacteria (6).
The radiologic findings associated with
pyogenic liver abscess are well known
(8) but may sometimes be indistinguish-
able from a hepatoma with vascular
thrombosis (9). Portal and hepatic vein
thrombosis have been reported in as-
sociation with liver abscesses (10,11),
and in particular, thrombophlebitis of
the subhepatic veins in association with
K pneumoniae liver abscesses has been
reported (12). It is important to diag-
nose K pneumoniae liver abscesses,
since there is a higher chance (12% vs
0%) of distant spread and a lower inci-
dence (0.6% vs 95%) of associated in-
traabdominal pathologic findings, such
as biliary tree obstruction, intraabdomi-
nal infections, and malignancies, as com-
pared with other polymicrobial liver
abscesses (13).
We undertook a retrospective review
to compare the computed tomographic
(CT) features of liver abscesses caused
by K pneumoniae with those caused by
other bacterial pathogens.
Materials and Methods
Patients and Data
Our retrospective study was approved
by the institutional review board with
Implication for Patient Care
n CT demonstration of characteris-
tic features of monomicrobial
K pneumoniae liver abscesses
may aid in early institution
of appropriate antibiotics and
prompt investigation and man-
agement of hematogenous septic
complications.
radiology.rsna.org
Alsaif et al
waiver of informed consent. Hospital
records of patients with a diagnosis of
liver abscess over a 7-year period be-
tween July 2003 and July 2010 were re-
trieved from our institution’s electronic
hospital database. The database search
yielded 187 patients.
Inclusion criteria were (a) presence
of a focal lesion or lesions in the liver
on contrast material–enhanced CT im-
ages; (b) frank pus aspirated from
the abscess cavity through diagnostic,
therapeutic radiologic, and/or surgical
drainage procedures; and (c) positive
culture results from liver abscess and/or
blood cultures. Twelve patients who
only had an ultrasonographic (US) di-
agnosis of liver abscess were excluded.
Six patients who were only treated with
antibiotics were excluded (patient re-
fusal of abscess drainage, n = 3; abscess
considered too small to drain, n = 3).
Cases of amoebic liver abscess (n = 2)
and hydatid disease (n = 1) with serologic
or surgical confirmation were also ex-
cluded. Liver abscesses with no growth
on either pus or blood culture were also
excluded from analysis (n = 35).
The final study population consisted
of 131 patients. Hospital records were
reviewed for (a) demographic data; (b)
coexisting medical conditions, including
diabetes mellitus, biliary tract disease,
chronic renal failure, malignancy, and
cirrhosis; (c) clinical symptoms; (d) ini-
tial laboratory data, including total and
differential white blood cells, erythrocyte
sedimentation rate, and C-reactive pro-
tein; (e) the type of drainage procedure
Published online before print
10.1148/radiol.11101876
Radiology 2011; 260:129–138
Author contributions:
Guarantor of integrity of entire study, S.K.V.; study
concepts/study design or data acquisition or data
analysis/interpretation, all authors; manuscript drafting
or manuscript revision for important intellectual content,
all authors; approval of final version of submitted
manuscript, all authors; literature research, H.S.A.,
S.K.V., S.A.; clinical studies, H.S.A.; statistical analysis,
H.S.A., S.K.V.; and manuscript editing, H.S.A.,
S.K.V., S.A.
Potential conflicts of interest are listed at the end
of this article.
n Radiology: Volume 260: Number 1—July 2011
GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses
that was performed (ie, radiologic or
surgical); (f) microbiological cause or
causes of liver abscess; (g) the occur-
rence of any septic hematogenous compli-
cations elsewhere that were document-
ed with imaging or clinically during the
same admission; (h) antibiotic treat-
ment duration; (i) hospitalization du-
ration; and (j) mortality related to liver
abscess and its complications.
Etiology
The etiology of the pyogenic liver ab-
scesses was defined by the organism or
organisms that were recovered from
the liver abscess and/or blood cultures.
The organism cultured from blood was
considered to be the etiologic pathogen
in cases in which no growth was seen
in the culture of pus aspirated from
the liver abscess. Blood culture results
were included only if the cultures were
performed within 1 week of drainage.
At least three blood cultures were done
before considering the results to be
negative. The abscesses were classi-
fied as monomicrobial or polymicrobial.
Monomicrobial liver abscesses were
those in which only one organism was
recovered from the liver abscess and/or
blood cultures. Polymicrobial liver ab-
scesses were those in which more than
one organism (including K pneumoniae)
was recovered from the liver abscess
and/or blood cultures.
CT Characteristics
All study patients underwent contrast-
enhanced CT before drainage of the liver
abscess. In some patients, CT studies
were also performed during follow-up
to assess the size of the abscess cavity,
to monitor complications, or to check
the position of the drainage catheter
in cases of poor or incomplete drain-
age. However, for the study purposes,
only the contrast-enhanced CT images
obtained before drainage of the liver
abscess were reviewed.
The CT examinations were per-
formed with one of the three different
scanners (Somatom Volume Zoom and
Somatom Sensation 64, Siemens Health-
care, Forchheim, Germany; PQ5000,
Philips, Amsterdam, the Netherlands)
available at our institution. CT examina-
Radiology: Volume 260: Number 1—July 2011
tions were performed by using a spiral
technique with 3–5-mm collimation. The
scans were acquired 60–75 seconds af-
ter the start of injection of 100 mL of
nonionic intravenous contrast medium
(iohexol, Omnipaque 300; Amersham
Health, Princeton, NJ) at a rate of
2–3 mL/sec. Axial sections of 5–7-mm
thickness were reconstructed, sent for
reporting, and archived.
The scans were reviewed by two
radiologists (S.K.V., an abdominal ra-
diologist with 10 years experience, and
H.S.A., a clinical research fellow with
5 years experience in general radiology)
in consensus. The following features
were recorded: (a) lobe involvement
(unilobar [right or left] or bilobar); (b)
number of abscesses (single or multiple);
(c) abscess size, with the largest abscess
measure when there were multiple; (d)
unilocular or multilocular (presence of
ⱖ1-mm-thick septations); (e) solid or
cystic appearance (.50% of the abscess
cavity appears hypodense or liquefied,
with an attenuation value of ⱕ20 HU)
in most of the sections showing the ab-
scess cavity; (f) gas within the abscess
cavity; (g) thrombophlebitis (hypodense
filling defects in the contrast-enhanced
hepatic veins, their tributaries, and/or the
inferior vena cava); (h) pylephlebitis
(hypodense filling defects in the contrast-
enhanced main portal vein and/or its
branches); and (i) pneumobilia (pres-
ence of gas in the bile ducts). Presence
of coexisting lesions in other organs,
including lung bases, covered in the CT
abdomen study was also noted.
Statistical Analysis
Statistical analyses were performed by
using software (SPSS 17; SPSS, Chi-
cago, Ill). The patients were grouped
into a monomicrobial K pneumoniae
liver abscess group and a comparison
group, comprising patients with non–
K pneumoniae monomicrobial or polymi-
crobial liver abscesses. Comparisons of
clinical and CT characteristics between
patients with monomicrobial K pneu-
monia liver abscess and the comparison
group were performed. The x2 analysis
or Fisher exact test was used for cate-
gorical variables, and the Student t test
was used for continuous variables. The
n radiology.rsna.org
Alsaif et al
log-rank test was used for time-to-event
variables. A P value of less than .05
was considered to indicate a significant
difference.
Results
The study group consisted of 131 patients
with a median age of 57 years (range,
21–89 years) (Table). There were more
men (median age, 55 years; range, 21–
89 years) than women (median age, 66
years; range, 30–86 years) in the study,
with a male-to-female ratio of 2.54:1.
The ethnicity of the study group was
predominantly Chinese, followed by Ma-
lay, Indian, other Asian, and others.
The most common presentation was fe-
ver and/or chills, followed by gastroin-
testinal symptoms (eg, gastrointestinal
upset, diarrhea, vomiting, nausea, dis-
comfort, pain), respiratory symptoms
(eg, cough, dyspnea, chest distress), and
jaundice. Septic shock was also pres-
ent in 24 patients. At admission, an el-
evated white blood cell count was found
in 127 patients (97%; reference range,
[3.3–9.3] 3 109/L). Diabetes mellitus
was the most common underlying medi-
cal condition, followed by biliary disease,
malignancy, and chronic renal failure.
Seven patients were receiving immuno-
suppressive treatment (two for lymphoma,
three following liver transplantation, and
two following renal transplantation).
Twenty-nine patients had evidence
of infectious lesions in other locations,
including lung abscesses, pyelonephritis,
splenic abscesses, soft-tissue abscesses,
prostatic abscess, pancreatitis, and oti-
tis media. There were no cases of en-
dophthalmitis or endocarditis.
Etiology of Liver Abscesses
Pus culture was positive in 119 patients
(monomicrobial in 104 and polymicrobial
in 15). The most common organism recov-
ered from pus was K pneumoniae (mo-
nomicrobial in 85, polymicrobial in six),
followed by E coli (monomicrobial in five,
polymicrobial in eight). Other organisms
cultured from pus included Enterococcus,
Burkholderia, Streptococcus , and
Staphylococcus species (Table). Blood
culture was positive in 67 patients and
K pneumoniae was the most common
131
GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses Alsaif et al

Clinical and CT Findings in 131 Patients with Pyogenic Liver Abscesses

Monomicrobial K pneumoniae Liver
Characteristic All Patients (n = 131) Abscess Group (n = 92) Comparison Group (n = 39) P Value*

Age (y)
Median† 57 (21–89) 55 (21–89) 61 (33–77) .105
Mean‡ 57.5 6 14.9 56.5 6 15.8 59.7 6 12.4 .264
Sex .41
Male 94 (71.8) 67 (72.8) 27 (69.2)
Female 37 (28.2) 25 (27.2) 12 (30.8)
Ethnicity .94
Chinese 86 (65.6) 60 (65.2) 26 (66.7)
Malay 35 (26.7) 25 (27.2) 10 (25.6)
Indian 4 (3.1) 3 (3.2) 1 (2.6)
Other Asian 5 (3.8) 3 (3.2) 2 (5.1)
Other 1 (0.8) 1 (1.1) 0
Symptom at presentation
Fever 85 (64.9) 59 (64.1) 26 (66.7) .83
Gastrointestinal 68 (51.9) 43 (46.7) 25 (64.1) .08
Respiratory 41 (31.3) 29 (31.5) 12 (30.7) ..99
Septic shock 24 (18.3) 13 (14.1) 11 (28.2) .08
Weight loss 9 (6.9) 5 (5.4) 4 (10.3) .45
Jaundice 20 (15.3) 10 (10.9) 9 (23.1) .02
Underlying disease
Diabetes mellitus 67 (51.1) 45 (48.9) 22 (56.4) .45
Biliary tract disease 18 (13.7) 12 (13.0) 6 (15.4) .78
Chronic renal failure 6 (4.6) 2 (2.1) 4 (10.3) .06
Malignancy 13 (9.9) 9 (9.8) 4 (10.3) ..99
Immunosuppression 7 (5.3) 3 (3.3) 4 (10.3) .20
Diverticulosis 10 (7.6) 6 (6.6) 4 (10.3) .48
Liver cirrhosis 2 (1.5) 2 (2.1) 0 ..99
Chronic alcoholism 3 (2.3) 3 (3.3) 0 .55
Hematologic parameters§
White blood cells (3 109/L) 13.8 (14.7 6 7.0) 14 (15.2 6 6.9) 12.4 (13.5 6 7.3) .24
Neutrophils (3 109/L) 11.3 (12.6 6 6.7) 11.6 (12.9 6 6.5) 10.6 (11.6 6 7.0) .30
Lymphocytes (3 109/L) 1.0 (1.1 6 0.6) 1.1 (1.1 6 0.6) 0.9 (1.0 6 0.6) .20
Monocytes (3 109/L) 0.6 (0.7 6 0.6) 0.6 (0.7 6 0.4) 0.6 (0.6 6 0.3) .98
Erythrocyte sedimentation rate (mm/hr) 87 (83.7 6 29.2) 88 (87.9 6 30.8) 76 (74.5 6 23.5) .12
C-reactive protein (mg/dL)§ 162 (157.7 6 103.8) 163 (163 6 111.6) 155 (143 6 79.0) .37
Microbiologic characteristic
Pus culture
Negative 12 (9.2) 8 (8.7) 4 (10.3) .75
Positive 119 (90.8) 84 (91.3) 35 (89.7) .75
Monomicrobial 104 (79.4) 84 (91.3) 20 (51.3) ,.01
Polymicrobial 15 (11.5) 0 15 (38.5) ,.01
Microorganism recovered
K pneumoniae 91 (69.5) 84 7
E coli 13 (9.9) 0 13
Enterococcus species 6 (4.6) 0 6
Burkholderia pseudomallei 4 (3.1) 0 4
Streptocococcus species 4 (3.1) 0 4
Staphylococcus aureus 3 (2.3) 0 3
Bacteroides species 3 (2.3) 0 3
Enterobacter species 3 (2.3) 0 3
Proteus mirabilis 2 (1.5) 0 2
(continues)

132 radiology.rsna.org n Radiology: Volume 260: Number 1—July 2011

GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses Alsaif et al

(continued)
Clinical and CT Findings in 131 Patients with Pyogenic Liver Abscesses
Monomicrobial K pneumoniae Liver
Characteristic All Patients (n = 131) Abscess Group (n = 92) Comparison Group (n = 39) P Value*

Morganella morganii 1 (0.8) 0 1

Pseudomonas species 1 (0.8) 0 1
Citrobacter species 1 (0.8) 0 1
Lactococcus species 1 (0.8) 0 1
Stenotrophomonas maltophilia 1 (0.8) 0 1
Achromobacter species 1 (0.8) 0 1
Acinetobacter baumannii 1 (0.8) 0 1
Blood culture
Negative 64 (48.9) 43 (46.7) 21 (53.8) .57
Positive 67 (51.1) 49 (53.3) 18 (46.2) .57
Microorganisms recovered
K pneumoniae 53 (40.5) 49 4
E coli 4 (3.1) 0 4
Streptococcus species 3 (2.3) 0 3
Staphylococcus species 2 (1.5) 0 2
Bukholderia species 2 (1.5) 0 2
Bacteroides species 1 (0.8) 0 1
Enterococcus species 1 (0.8) 0 1
Proteus species 1 (0.8) 0 1
Septic hematogenous complications 29 (22.1) 26 (28.3) 3 (7.7) ,.01
Lung 17 (13) 15 (16.3) 2 (5.1) .09
Urinary tract 10 (7.6) 9 (9.8) 1 (2.6) .28
Spleen 7 (5.3) 5 (5.4) 2 (5.1) ..99
Soft tissues 2 (1.5) 2 (2.1) 0 ..99
Prostate 1 (0.8) 1 (1.1) 0 ..99
Pancreas 1 (0.8) 1 (1.1) 0 ..99
Ear 1 (0.8) 0 1 (2.6) ..99
Days between symptom onset and CT§ 3 (4.8 6 4.6) 4 (4.5 6 3.9) 3 (5.6 6 6.3) .35
Abscess size (cm)§ 7.0 (7.4 6 2.8) 7.0 (7.3 6 2.8) 8.0 (7.8 6 2.8) .35
No. of abscesses .01
1 95 (72.5) 73 (79.3) 22 (56.4)
.1 36 (27.5) 19 (20.7) 17 (43.6)
Appearance .03
Solid 67 (51.1) 53 (57.6) 14 (35.9)
Cystic 64 (48.9) 39 (42.4) 25 (64.1)
Septations within abscess .01
Unilocular 16 (12.2) 5 (5.4) 11 (28.2)
Multilocular 115 (87.8) 87 (94.6) 28 (71.8)
Lobar involvement .01
Unilobar 100 (76.3) 76 (82.6) 24 (61.5)
Bilobar 31 (23.7) 16 (17.4) 15 (38.5)
(continues)

organism cultured. The causative mi-
croorganisms were cultured from pus
only in 64 patients, both pus and blood
in 55 patients and from only blood in
12 patients.
Overall 92 patients formed the mo-
nomicrobial K pneumoniae liver abscess
group (43 pus culture only positive, 41
both pus and blood cultures positive,
and eight blood culture only positive)
and 39 patients formed the others group
(21 pus culture only positive, 14 both
pus and blood culture positive, and four
blood culture only positive).
Treatment Received
All patients received antibiotics and had
one or more drainage procedures per-
formed (Table). Among the patients in
the monomicrobial K pneumoniae liver
abscess group, image-guided percuta-
neous drainage was performed in 82
patients, eight patients underwent sur-
gical drainage, and two patients under-
went image-guided drainage followed
by surgical drainage. The median dura-
tion of antibiotic treatment was 14 days,

Radiology: Volume 260: Number 1—July 2011 n radiology.rsna.org 133

GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses Alsaif et al

(continued)
Clinical and CT Findings in 131 Patients with Pyogenic Liver Abscesses
Monomicrobial K pneumoniae Liver
Characteristic All Patients (n = 131) Abscess Group (n = 92) Comparison Group (n = 39) P Value*

Thrombophlebitis 30 (22.9) 28 (30.4) 2 (5.1) ,.01

Pylephlebitis 3 (2.3) 1 (1.1) 2 (5.1) .21
Gas in abscess cavity 17 (13.0) 11 (28.2) 6 (15.4) .58
Pneumobilia 6 (4.6) 1 (1.1) 5 (12.8) .01
Treatment ..99
Image-guided drainage 115 (87.8) 82 (89.1) 33 (84.6)
CT 70 (53.4) 52 (56.5) 18 (46.1)
US 45 (34.3) 30 (32.6) 15 (38.5)
Surgical drainage 14 (10.7) 8 (8.7) 6 (15.4)
Laparoscopic 9 (8.2) 6 (6.5) 3 (7.7)
Open 5 (3.8) 2 (2.2) 3 (7.7)
Radiologic and surgical drainage 2 (1.5) 2 (2.2) 0
Days of antibiotic treatment† 14 (2–82) 14 (2–70) 16 (3–82) .018
Days of hospitalization† 14 (2–72) 14 (2–56) 16 (7–72) .005
Mortality from liver abscess 7 (5.3) 3 (3.3) 4 (10.2) .19
Note.—Unless otherwise specified, data are numbers of patients, with percentages in parentheses.
* P values for differences between monomicrobial K pneumoniae liver abscess group and comparison group.
†
Data are medians, with ranges in parentheses.
‡
Data are means 6 standard deviations.
§
Data are medians, with means 6 standard deviations in parentheses.

and the median duration of hospitaliza-
tion was also 14 days.
In the comparison group of patients
with other pyogenic liver abscesses, 33
had image-guided abscess drainage and
six had surgical drainage. The median
duration of the antibiotic treatment and
hospitalization were both 16 days.
Seven patients died, and all deaths
were attributed to abscess-related
septicemia.
CT Characteristics of Liver Abscesses
The median time interval between the
onset of fever and CT was 3 days (range,
1–24 days). The median size of the liver
abscess was 7 cm (range, 1.7–14 cm). A
single abscess was found in 95 patients,
and multiple abscesses were seen in
36 patients. Unilobar involvement was
seen in 100 patients, with the right lobe
affected more commonly (55.0%, 72
of 131) (Figs 1–3, E1 [online]). The
abscesses were predominantly solid
and masslike in 67 patients (Figs 1, 2,
E1 [online]) and cystic in 64 patients
(Figs 3 , E2 [online]). Multilocular
appearance of the abscesses was
present in 115 patients (87.8%) (Figs 2,
E1 [online]). Thrombophlebitis was pre-
sent in 30 patients (Figs 1, 2, E1 [on-
line]); pylephlebitis, in three; gas in
the abscess cavity, in 17; and pneu-
mobilia, in six (Fig 3). Three patients
had both gas in the abscess cavity and
pneumobilia.
Monomicrobial K pneumoniae Liver
Abscesses versus Others
There was no significant difference be-
tween the monomicrobial K pneumoniae
liver abscess group and the compari-
son group in terms of mean age, sex
ratio, ethnicity, or underlying medical
conditions (Table). There were also no
significant differences between the two
groups with respect to symptoms at
presentation except for jaundice, which
was significantly more common in the
comparison group (23.1% vs 10.9%,
P = .02). The median duration of anti-
biotic treatment was significantly lower
in patients with monomicrobial K pneu-
moniae liver abscesses than in those
with other pyogenic abscesses (14 vs 16
days, P = .018). Similarly, the median
duration of the hospital stay was signif-
icantly lower in patients with monomi-
crobial K pneumoniae liver abscess than
in those with other pyogenic abscesses
(14 vs 16 days, P = .005). Septic shock
and mortality were also higher in the
comparison group; however, this did not
reach statistical significance.
On CT images, monomicrobial
K pneumoniae liver abscesses were
more likely to appear as single abscess
cavity (Figs 1, 2, E1 [online]), to involve
single lobe of the liver, and to appear
solid and multilocular than comparison
abscesses (Table) (Figs 1, 2, E1 [online]).
There were no significant differences be-
tween monomicrobial K pneumoniae
liver abscesses and others with re-
spect to the size of the abscess cavity.
Thrombophlebitis (Figs 1, 2, E1 [online])
(30.4% vs 5.1%, P , .01) and septic
hematogenous complications (28.3% vs
7.7%, P , .01) occurred significantly
more often in the monomicrobial K pneu-
moniae liver abscess group than in the
comparison group. Septic hematogenous
complications were more common in
cases with thrombophlebitis (Figs 2,

134 radiology.rsna.org n Radiology: Volume 260: Number 1—July 2011

GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses Alsaif et al

Figure 1

Figure 1: Contrast-enhanced CT images of a 63-year-old man with diabetes

who presented with fever and abdominal discomfort of 5 days duration and
was found to have (c) a single large monomicrobial K pneumoniae liver
abscess (dotted outline) in the right liver lobe, which is predominantly solid
in appearance with some areas of hypodensity representing liquefaction.
(a, b) Hypodense filling defect (arrows) in right hepatic vein is thrombophlebitis.
Both pus and blood cultures were positive for K pneumoniae. Abscess was
drained under CT guidance, patient received antibiotics for 12 days, and he
recovered and was discharged after 21 days.

Our results support those of a pre-

E1 [online]), as compared with those with
no evidence of thrombophlebitis (50.0%
[15 of 30] vs 13.9% [14 of 101], P ,
.001). Monomicrobial K pneumoniae
liver abscesses showed a trend toward
higher incidence of gas in the abscess
cavity, whereas pylephlebitis was more
common in the comparison abscesses,
but these results did not reach statis-
tical significance. Pneumobilia (Fig 3)
was more commonly found in pyogenic
liver abscesses of other causes (Table).
Discussion
Adequate antibiotics and drainage are
standard treatment for patients with
pyogenic liver abscesses. The suspected
abscess etiology and underlying co-
morbidities should be considered when
choosing empirical antimicrobial agents
while awaiting culture results. Imag-
ing characteristics suggestive of a par-
ticular liver abscess etiology might be
useful in guiding appropriate antibiotic
selection.
Our study results show that mono-
microbial K pneumoniae liver abscesses
commonly present as a single unilobar
abscess with solid or multilocular ap-
pearance as compared with liver ab-
scesses caused by other bacteria. Mono-
microbial K pneumoniae liver abscesses
are also associated with an increased
incidence of thrombophlebitis and he-
matogenous septic complications.
vious study (14) that found that mono-
microbial K pneumoniae liver abscesses
appeared predominantly as a solid lesion
at US. A possible explanation for the
solid appearance of the liver abscesses
is failure of liquefaction owing to high
prevalence of a phagocytosis-resistant
capsular serotype K pneumoniae strain
associated with liver abscesses in the
Chinese population (15,16). Hui et al
(14) found an association between solid
appearance of monomicrobial K pneu-
moniae liver abscesses and diabetes;
however, we did not find any similar asso-
ciation, which may be related to a differ-
ence in population. Hui et al (14) also
found that the amount of pus aspirated at
the initial drainage was lower in K pneu-
moniae abscesses than in other pyogenic
abscesses, supporting the finding that
these liver abscesses were not fully liq-
uefied. We were not able to examine
data on pus volume because it was not
available in all cases.
Thromboses of the portal and he-
patic veins in association with liver
abscesses have been reported before

Radiology: Volume 260: Number 1—July 2011 n radiology.rsna.org 135

GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses Alsaif et al

Figure 2

Figure 2: Contrast-enhanced axial (a, c) and (b) coronal reconstruction CT

images in an 83-year-old man with diabetes who presented with 4-day history
of fever, abdominal discomfort, and cough show a single multiloculated liver
abscess in the right lobe (short black arrows). Patient had thrombophlebitis
extending from the abscess to the right hepatic vein (straight white arrows)
and septic lung complications. Note also consolidation of the right lower lobe
(curved arrow) with a small lung abscess (long black arrow) in lung base. Liver
abscess was drained under US guidance. Both abscess and blood cultures
were positive for K pneumoniae. Patient received antibiotics for 7 days, made a
good recovery, and was discharged after 11 days.

necrotizing fasciitis, epidural abscess,

(10,11). However, to our knowledge, there
has not been a report on correlation of
pylephlebitis and thrombophlebitis with
the etiology of the abscesses and as-
sociated septic hematogenous complica-
tions. Two cases of thrombophlebitis of
the hepatic vein in association with a
K pneumoniae liver abscess have been
reported by Maffiolo et al (12). They
raised the possibility that thrombosis
facilitated the diffuse and severe pulmo-
nary dissemination found in their cases.
In our series, thrombophlebitis was asso-
ciated with nearly one-third of the mono-
microbial K pneumoniae liver abscesses
as compared with 5.1% of the other
pyogenic organisms. To the best of our
knowledge, there are no other reports
with imaging findings on thrombophlebi-
tis associated with liver abscesses.
Hematogenous infections and severe
pulmonary complications often result in
a grave prognosis. Previous reports have
shown that hematogenous septic com-
plications are more common in K pneu-
moniae liver abscesses than in non–
K pneumoniae liver abscesses (3,17,18).
Yang et al (2) found a significant differ-
ence in the incidence of hematogenous
septic manifestations. Many distant le-
sions have previously been described as
complications, including endophthalmi-
tis, brain abscess, septic pulmonary em-
bolism, septic arthritis, psoas abscess,
osteomyelitis, and endocarditis (19–23).
Hematogenous seeding has been pro-
posed as the mechanism for hematog-
enous septic complications. Both the
virulence of K pneumoniae and the as-
sociation with underlying diabetes play
important roles in hematogenous com-
plications (24,25). K pneumoniae K1
and K2 capsular serotypes are the most
prevalent serotypes identified in mono-
microbial K pneumoniae liver abscesses
and hematogenous infections, includ-
ing endophthalmitis (26). Our study
results show that thrombophlebitis is
a characteristic feature of monomicro-
bial K pneumoniae liver abscesses and
that more than half of the monomicro-
bial K pneumoniae liver abscesses with
thrombophlebitis were associated with
hematogenous septic complications, sug-
gesting that thrombophlebitis is likely
to be involved in the dissemination of
the infectious processes.
The duration of antibiotic treatment
and hospital stay were both significantly

136 radiology.rsna.org n Radiology: Volume 260: Number 1—July 2011

GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses Alsaif et al

Figure 3

Figure 3: Contrast-enhanced CT images in a 68-year-old man with diabetes who presented with fever, jaundice, cough, and breathlessness (a) show a single large
unilocular liver abscess in the right liver lobe of the liver (black arrows) that is cystic in appearance. (b) There is also pneumobilia (white arrow). Note also mild
bilateral pleural effusions and basal atelectases. Abscess was drained under CT guidance. Abscess culture was positive for E coli; however, blood culture grew
Streptococcus anginosus. Patient received antibiotics for 19 days and was hospitalized for 28 days.

shorter in patients with monomicrobial
K pneumoniae liver abscesses as com-
pared with those with other etiologies.
Previous studies have shown either no
significant difference (2) or longer hos-
pital stay (27) between K pneumoniae
liver abscesses and other pyogenic or-
ganisms. Similarly, there have been no
significant differences reported in the
duration of antibiotic therapy for K pneu-
moniae liver abscesses as compared
with other pyogenic liver abscesses. In
our study, the duration of antibiotic
treatment was shorter in monomicrobial
K pneumoniae liver abscesses, likely
because of broad-coverage antibiotics
given to the patients and probably re-
lated to the population studied. As a
result of shorter duration of antibiotic
treatment, the hospital stay was also
shorter. The mortality rate in the mono-
microbial K pneumoniae liver abscess
group in our study was lower compared
than that in the comparison group but
did not reach significance. A study by
Yang et al (2) found a statistical dif-
ference, whereas other studies have
reported no significant differences in
mortality rate (12,27). These differences
in mortality rates in pyogenic liver ab-
scesses may be related to variability in
the populations studied and the spec-
trum of pathogenic organisms.
Our study had limitations. First, it
was retrospective in design. Second,
we excluded patients who received only
antibiotic treatment without drainage,
which is an accepted treatment method
for liver abscesses at many institutions.
However, the number of patients who
received only antibiotics at our institu-
tion was small. Third, the causation of
hematogenous septic infections by liver
abscesses is speculative, since many of
the subjects had either clinical features
or findings on CT images at the time
of the diagnostic CT scan. The possi-
bility of the infection developing else-
where and then spreading to the liver
cannot be entirely ruled out. However,
the most common source of K pneu-
moniae and other organisms that cause
liver abscesses is from the gastrointesti-
nal tract. The gut organisms first enter
the portal circulation and pass through
the liver before entering the systemic
circulation. Therefore, we believe that
the infective focus first developed with-
in the liver, followed by hematogenous
spread elsewhere.
In summary, our study results show
that monomicrobial K pneumoniae liver
abscesses manifest most commonly as
single, solid-appearing, and multilocu-
lated focal hepatic lesions and are as-
sociated with thrombophlebitis of the
hepatic veins and hematogenous septic
complications as compared with other
types of pyogenic liver abscesses.
Acknowledgments: Authors thank Vasanthi Pe-
rumal, SEN, (Diagnostic Radiology, National Uni-
versity of Singapore) for her assistance with data
collection and Chan Yiong Huak, PhD, (Biosta-
tistics Unit, Yong Loo Lin School of Medicine,
National University of Singapore) for his assis-
tance with the statistical analysis.
Disclosures of Potential Conflicts of Interest:
H.S.A. No potential conflicts of interest to dis-
close. S.K.V. No potential conflicts of interest
to disclose. D.S.G.C. No potential conflicts of
interest to disclose. S.A. No potential conflicts of
interest to disclose.
References
1. Brook I, Frazier EH. Microbiology of liver
and spleen abscesses. J Med Microbiol 1998;
47(12):1075–1080.
2. Yang CC, Yen CH, Ho MW, Wang JH. Com-
parison of pyogenic liver abscess caused by
non-Klebsiella pneumoniae and Klebsiella
pneumoniae. J Microbiol Immunol Infect
2004;37(3):176–184.
3. Chung DR, Lee SS, Lee HR, et al. Emerging
invasive liver abscess caused by K1 serotype
Klebsiella pneumoniae in Korea. J Infect
2007;54(6):578–583.
4. Rahimian J, Wilson T, Oram V, Holzman RS.
Pyogenic liver abscess: recent trends in
etiology and mortality. Clin Infect Dis 2004;
39(11):1654–1659.
5. Lee K, Wong S, Sheen P. Pyogenic liver ab-
scess: an audit of 10 years’ experience and

Radiology: Volume 260: Number 1—July 2011 n radiology.rsna.org 137

GASTROINTESTINAL IMAGING: CT Appearance of K pneumoniae Liver Abscesses
analysis of risk factors. Dig Surg 2001;18(6):
459–465; discussion 465–466.
6. Lederman ER, Crum NF. Pyogenic liver ab-
scess with a focus on Klebsiella pneumoniae
as a primary pathogen: an emerging disease
with unique clinical characteristics. Am J
Gastroenterol 2005;100(2):322–331.
7. Pope J, Teich D, Clardy P, McGillicuddy D.
Klebsiella pneumoniae liver abscess: an
emerging problem in North America. J Emerg
Med doi:10.1016/j.jemermed.2008.04.041.
Published online November 10, 2008. Ac-
cessed June 15, 2010.
8. Mortelé KJ, Segatto E, Ros PR. The infected
liver: radiologic-pathologic correlation. Ra-
dioGraphics 2004;24(4):937–955.
9. Brown KT, Gandhi RT, Covey AM, Brody LA,
Getrajdman GI. Pylephlebitis and liver ab-
scess mimicking hepatocellular carcinoma.
Hepatobiliary Pancreat Dis Int 2003;2(2):
221–225.
10. Lee KH, Han JK, Jeong JY, et al. Hepatic
attenuation differences associated with ob-
struction of the portal or hepatic veins in
patients with hepatic abscess. AJR Am J
Roentgenol 2005;185(4):1015–1023.
11. Syed MA, Kim TK, Jang HJ. Portal and he-
patic vein thrombosis in liver abscess: CT
findings. Eur J Radiol 2007;61(3):513–519.
12. Maffiolo C, Novellas S, Chevallier P, Brunner
P, Mourou MY, Bruneton JN. Thrombophle-
bitis of the hepatic veins: complication of
a Klebsiella liver abscess. Clin Imaging
2006;30(1):63–65.
138
13. Wang JH, Liu YC, Lee SS, et al. Primary
liver abscess due to Klebsiella pneumo-
niae in Taiwan. Clin Infect Dis 1998;26(6):
1434–1438.
14. Hui JY, Yang MK, Cho DH, et al. Pyogenic
liver abscesses caused by Klebsiella pneu-
moniae: US appearance and aspiration find-
ings. Radiology 2007;242(3):769–776.
15. Lau YJ, Hu BS, Wu WL, Lin YH, Chang HY,
Shi ZY. Identification of a major cluster of
Klebsiella pneumoniae isolates from patients
with liver abscess in Taiwan. J Clin Micro-
biol 2000;38(1):412–414.
16. Lee CH, Leu HS, Wu TS, Su LH, Liu JW.
Risk factors for spontaneous rupture of liver
abscess caused by Klebsiella pneumoniae.
Diagn Microbiol Infect Dis 2005;52(2):79–84.
17. Chang FY, Chou MY. Comparison of pyo-
genic liver abscesses caused by Klebsiella
pneumoniae and non-K. pneumoniae patho-
gens. J Formos Med Assoc 1995;94(5):
232–237.
18. Chen SY, Lin CH, Chang HM, et al. Education
and imaging: hepatobiliary and pancreatic—
gas-forming pyogenic liver abscess. J Gas-
troenterol Hepatol 2007;22(1):133.
19. Hu BS, Lau YJ, Shi ZY, Lin YH. Necrotiz-
ing fasciitis associated with Klebsiella pneu-
moniae liver abscess. Clin Infect Dis 1999;
29(5):1360–1361.
20. Cheng DL, Liu YC, Yen MY, Liu CY, Wang
RS. Septic metastatic lesions of pyogenic
liver abscess: their association with Kleb-
siella pneumoniae bacteremia in diabetic
radiology.rsna.org
Alsaif et al
patients. Arch Intern Med 1991;151(8):
1557–1559.
21. Kuramochi G, Takei SI, Sato M, Isokawa O,
Takemae T, Takahashi A. Klebsiella pneu-
moniae liver abscess associated with septic
spinal epidural abscess. Hepatol Res 2005;
31(1):48–52.
22. Yang PW, Lin HD, Wang LM. Pyogenic liver
abscess associated with septic pulmonary
embolism. J Chin Med Assoc 2008;71(9):
442–447.
23. Rivero A, Gomez E, Alland D, Huang DB,
Chiang T. K2 serotype Klebsiella pneu-
moniae causing a liver abscess associated
with infective endocarditis. J Clin Microbiol
2010;48(2):639–641.
24. Lin JC, Chang FY, Fung CP, et al. High preva-
lence of phagocytic-resistant capsular sero-
types of Klebsiella pneumoniae in liver ab-
scess. Microbes Infect 2004;6(13):1191–1198.
25. Lin JC, Siu LK, Fung CP, et al. Impaired
phagocytosis of capsular serotypes K1 or K2
Klebsiella pneumoniae in type 2 diabetes
mellitus patients with poor glycemic con-
trol. J Clin Endocrinol Metab 2006;91(8):
3084–3087.
26. Fung CP, Chang FY, Lee SC, et al. A global
emerging disease of Klebsiella pneumoniae
liver abscess: is serotype K1 an important
factor for complicated endophthalmitis?
Gut 2002;50(3):420–424.
27. Chen SC, Wu WY, Yeh CH, et al. Com-
parison of Escherichia coli and Klebsiella
pneumoniae liver abscesses. Am J Med Sci
2007;334(2):97–105.
n Radiology: Volume 260: Number 1—July 2011