Va s c u l a r a n d I n t e r ve n t i o n a l R a d i o l o g y • O r i g i n a l R e s e a r c h

Haider et al.
Percutaneous Drainage of Liver Abscesses

Vascular and Interventional Radiology

Original Research

Liver Abscesses: Factors

That Influence Outcome of
Percutaneous Drainage
OBJECTIVE. The purpose of this study was to identify the details of percutaneous cath-
1
Steffen J. Haider
Massimo Tarulli2 eter drainage (PCD) of pyogenic liver abscesses, the etiologic factors, and the management
Nancy J. McNulty 3 techniques that contribute to successful treatment.
Eric K. Hoffer 3 MATERIALS AND METHODS. The records of 75 consecutively registered patients
who underwent PCD of 96 abscesses at a single institution between May 2009 and May 2014
Haider SJ, Tarulli M, McNulty NJ, Hoffer EK were retrospectively reviewed. Thirty-nine patients (52%) were oncology patients, and 36
(48%) had recently undergone abdominal surgery. Primary success was defined as abscess
American Journal of Roentgenology 2017.209:205-213.

Keywords: abscess, liver abscess, malignancy,
percutaneous abscess drainage
DOI:10.2214/AJR.16.17713
Based on a presentation at the Cardiovascular and
Interventional Radiological Society of Europe 2016
healing with the primary PCD intervention and 30-day postdrainage survival. Salvage suc-
cess was defined as abscess healing with follow-up secondary PCD placement for symptom-
atic hepatic satellite collections or for clinical recurrence. Catheter adjustments were per-
formed during follow-up to optimize existing drains. Univariate, multivariate, and general
linear mixed model analyses were performed. The median follow-up time after catheter re-
moval was 6 months (range, 2–62 months).
RESULTS. Drains were primarily successful in 54 patients (72%), and 17 patients (23%)
needed salvage PCD; thus, overall success was achieved in 71 patients (95%). The other four
patients (5%) died of sepsis. The primary success rate was reduced in patients with unresect-
able malignancies (p = 0.01), multiple abscesses (p = 0.01), and output ≥ 15 mL/d at catheter
endpoint (n = 7, p = 0.001). Only unresectable malignancies had slightly lower overall success.
Large abscesses (> 150 cm3) required more catheter adjustments and longer drainage duration
to reach abscess cavity closure. Successfully drained abscesses reached cavity closure a mean
of 23 days (95% CI, 20–27 days) after treatment.
CONCLUSION. PCD was effective first-line treatment of complicated pyogenic liver
abscesses, which often require catheter adjustment and salvage drainage procedures to reli-
ably achieve success.
utcomes of therapy for pyogenic The reported failure rate of primary PCD

annual meeting, Barcelona, Spain.
Received November 29, 2016; accepted after revision
January 15, 2017.
1
Department of Radiology, Columbia University Medical
Center, New York, NY.
2
Department of Medical Imaging, Faculty of Medicine,
University of Toronto, Toronto, ON, Canada.
3
Department of Radiology, Section of Interventional Radi-
ology, Dartmouth-Hitchcock Medical Center, 1 Medical
Center Dr., Lebanon, NH 03756. Address correspondence
to E. K. Hoffer (eric.k.hoffer@hitchcock.org).
This article is available for credit.
AJR 2017; 209:205–213
0361–803X/17/2091–205
© American Roentgen Ray Society
O
liver abscess (PLA) have improved
dramatically over the past 40 years
owing to earlier imaging diagno-
sis, more effective antibiotic coverage, and ad-
vances in percutaneous drainage techniques.
By the 1980s, percutaneous catheter drainage
(PCD) with antibiotic therapy was accepted
therapy for simple hepatic abscesses [1, 2]. Ow-
ing to advances in technique and management,
PCD has supplanted surgical drainage as the
first-line treatment of PLA. This development
has been accompanied by a reduction in the
need for general anesthesia and the complica-
tions of an open abdominal procedure [3–5].
Surgery remains first-line treatment of patients
who present with abscess rupture, peritonitis,
or a concomitant surgical condition but is oth-
erwise reserved for failure of PCD [3–7].
ranges widely, from 0% to 30% with an in-
fection-related mortality of 0–13.4% [7–16].
Several studies have shown that PCD is less
effective than surgical drainage in the treat-
ment of large loculated abscesses, variably
defined as greater than 3–6 cm in diameter
[13, 14]. In other series, however, no differ-
ence was found in drainage success rates of
large and small abscesses, regardless of loc-
ulation [3, 5, 9, 17–19]. Variation may be due
to differences in management given that fol-
low-up practices are not standardized [20].
Few studies have included detailed analy-
sis of percutaneous liver drain management,
making extraction of best practices difficult
[7, 20]. These include indications for addi-
tional drainage procedures and optimal du-
ration of drainage. In this study, we inves-

AJR:209, July 2017 205


tigated management details that contribute
to optimal management of the percutaneous
PLA drainage catheter.
Materials and Methods
Patient Demographics and Study Design
This institutional review board–approved ret-
rospective review was conducted with a waiver of
informed consent. Between May 2009 and May
2014, 75 consecutively registered patients under-
went PCD of 96 PLAs at a single tertiary referral
academic medical center. The diagnosis of PLA
was confirmed by the finding of purulent aspirate
or positive microbiologic findings. Amebic, fun-
gal, and parasitic abscesses were excluded. Oth-
er than for the nonpyogenic abscesses, no patients
who underwent percutaneous liver abscess treat-
ment were excluded. Table 1 shows the demo-
graphics of the patient cohort. Patient data were
collected from the hospital electronic medical re-
cord (Epic, Epic Systems). The median length of
follow-up after catheter removal was 6 months
American Journal of Roentgenology 2017.209:205-213.
(range, 2–62 months). This study was performed
in accordance with the Strengthening the Report-
ing of Observational Studies in Epidemiology
guidelines [21].
Management Protocol
Percutaneous catheter drain placement—Pa-
tients received broad-spectrum antibiotic therapy,
which the primary team refined according to bacte-
rial culture results. PCD was performed with stan-
dard sterile technique after subcutaneous admin-
istration of anesthesia and IV moderate sedation.
The Seldinger technique was used to percutaneous-
ly introduce an 8- to 12-French catheter into the
abscess cavity under CT or ultrasound guidance.
Larger cavities and more viscous fluid prompted
the use of the 12-French catheters. The abscess
cavity was fully aspirated. Immediate reimaging
was performed to identify persistent collections or
loculations. Drainage of smaller satellite collec-
tions was performed at the discretion of the inter-
ventionalist. Drainage catheters were sutured to the
skin and left to bulb suction drainage. The catheter
was flushed twice daily with 5–10 mL normal sa-
line solution, and output was recorded.
Follow-up management—The intervention-
al radiology team assessed the patient clinical-
ly every other day for the first week, then week-
ly if the patient remained in the hospital. Patients
with continued drainage after hospital discharge
underwent follow-up in the interventional radiol-
ogy clinic. Follow-up imaging was performed for
pain, fever, or elevated WBC count after the 48-
hour postprocedural period [3, 4, 19, 20]. Drain
output that plateaued at  ≥ 15 mL/d or increased
also prompted follow-up imaging. Imaging mo-
206 AJR:209, July 2017
Haider et al.
dalities included fluoroscopic contrast study of the
drain (drain study), CT, and ultrasound.
Catheter adjustment was exchange or reposi-
tioning of an existing catheter under fluoroscopic
guidance to optimize catheter drainage. Common
indications included accidental removal, retrac-
tion, occlusion, and poor position of the catheter
with respect to residual abscess collection.
Salvage drainage was performed during fol-
low-up if a patient had symptoms and imaging
confirmed an undrained abscess. This could be
an initially undrained satellite collection or a sec-
ond abscess or recurrence that developed during
follow-up. Salvage PCD was performed with the
same protocol as for primary PCD.
Endpoint—A fluoroscopic drain study was
performed to ensure that the abscess cavity had
healed when the patient had no symptoms and out-
put was  < 15 mL/d for 2 days. A residual cavity
of < 3 cm and resistance to injection or leakage
of contrast material along the drain confirmed ab-
scess healing. However, the primary team occa-
sionally removed drains on the basis of clinical
findings alone.
Abscess management after catheter removal—
CT or ultrasound studies were obtained for 62 pa-
tients (85%) a median of 37 days (range, 2–252
days) after catheter removal. Routine follow-up
imaging accounted for 55% of the studies; 45% of
studies were prompted by symptoms. Clinical re-
currence was defined as worsening clinical symp-
toms and image confirmation of reaccumulation
of the originally drained collection. Salvage PCD
was performed on clinically recurrent abscesses.
Definition of Outcomes
Primary success was defined as abscess reso-
lution requiring only the primary PCD treatment
with 30-day survival after catheter removal. Pri-
mary failure was failure to meet the criteria for
primary success. Catheter adjustments to existing
drains are reported as a part of primary success.
Salvage success was defined as abscess healing
with 30-day posttreatment survival after salvage
PCD. Overall success was defined as either prima-
ry or salvage success.
Imaging Analysis
Radiographic abscess characteristics were re-
corded from primary radiographic reports by board-
certified diagnostic radiologists and then verified by
a board-certified interventional radiologist with 24
years of experience. Abscess volume was calcu-
lated with the ellipsoid formula at the greatest di-
mension in each of the axial, sagittal, and coronal
planes. If an ultrasound or drain study included only
two dimensions, the smaller of the two was used as
a surrogate for the third dimension.
Statistical Analysis
Statistical analyses were performed with SPSS
Statistics for Microsoft Windows software (ver-
sion 22.0, IBM). Significance was defined as p <
0.05. Data are expressed as mean and SD. Differ-
ences in subgroup mean drainage duration were
compared by nonparametric Mann-Whitney U test.
Subgroup association with failure of primary suc-
cess was analyzed at the univariate level with the
Fisher exact test, and relative risk was reported. A
multivariate logistic regression model was used to
test the independence of subgroups that had uni-
variate association with primary success. Odds ra-
tios were generated from the logistic regression.
The mean time to cavity closure was estimated
with general linear mixed models [22] by regres-
sion of repeated measurements of abscess volume
over the follow-up period. A random slope with
fixed intercept was assumed. We used a heteroge-
neous autoregressive structure to model covari-
ance [23]. We performed post hoc power analysis
at α = 0.05. There was approximately 65% power
for detecting a 30% difference in primary success
rate in subgroups of 15–20 patients.
Results
Demographics
The clinical and abscess characteristics of
the 75 patients with 96 PLAs are reported in
Tables 1 and 2. The most common under-
lying etiologic factors predisposing to PLA
were assessed. Recent hepatic oncologic sur-
gery (within 30 days of PCD) had been per-
formed on 25% of patients and recent nonhe-
patic abdominal surgery on 23%; 19% had
unresectable hepatopancreatobiliary malig-
nancies (Table 1). Hepatopancreatobiliary
malignancy was found in 52% of patients,
87% had abscesses > 5 cm in diameter, 35%
had multiple abscesses, and 56% had locu-
lated abscesses.
Outcomes
Patient outcomes are reported in Table 3.
Primary clinical success was achieved in 54
patients (72%), and primary drainage failed
in 21 patients (28%). Assisted salvage suc-
cess was achieved for 17 of the 18 patients
who underwent salvage PCD intervention.
This resulted in overall success for 71 pa-
tients (95%). The other four patients (5%)
died of sepsis within 30 days of catheter re-
moval. Adjustment to an existing catheter
was required in 17 patients (23%) to optimize
drainage. Three of the patients who died were
not surgical candidates because they had ad-
vanced malignant hepatobiliary disease, in-
cluding a case of multiorgan abscesses. The
 Percutaneous Drainage of Liver Abscesses

fourth death occurred during salvage drain- TABLE 1: Demographic Characteristics, Underlying Pathologic Condition,
age. The one major complication, cholecys- and Abscess Size
tocutaneous fistula, required surgical repair Variable Result
after successful healing of a pericholecystic
abscess. No patient needed surgical abscess Demographic characteristics
drainage. The mean duration of drainage was PCD-treated patients 75
28 (SD, 22) days (range, 3–115 days). PCD-treated abscesses 96
Salvage percutaneous catheter drain—Fig-
Age (y)
ure 1 shows salvage PCD of a symptomatic
satellite collections and catheter adjustment Mean 62
in a patient with multiple abscesses. Eighteen Range 15–90
of the 21 patients with primary failure un- SD 13
derwent salvage PCD. The 17 patients who
Sex
healed after salvage PCD included five with
follow-up drainage of symptomatic satellite Men 50 (67)
collections, eight with clinical recurrence, Women 25 (33)
and four with both drainage of satellite col- Malignancya 39 (52)
lections and recurrence. The initial salvage
Pancreatic or ampullary malignancy 15 (20)
PCD for symptomatic satellite abscesses was
performed a mean of 24 (SD, 12) days (range, Colorectal carcinoma with hepatic metastases 13 (17)
5–46 days) after the primary PCD procedure, Gallbladder cancer or cholangiocarcinoma 6 (7)
resulting in healing of 13 of the 17 salvage
American Journal of Roentgenology 2017.209:205-213.

Underlying pathologic condition predisposing to pyogenic liver abscess

successes. To achieve healing, four patients
Recent hepatic surgeryb 19 (25)
needed two or three salvage drainage proce-
dures 22–177 days after the initial PCD. Abscess of surgical bed after oncologic resection 18
Catheter adjustments—Between one and Abscess of radiofrequency ablation bed 1
five catheter adjustment procedures were Recent nonhepatic abdominal surgeryb 17 (23)
performed on 17 (23%) patients to optimize
Pancreatobiliary oncologic surgery 7
drainage during follow-up. This included 10
catheter exchanges, three caliber increases, Nononcologic abdominal surgery 10
two caliber decreases, and six catheter repo- Unresectable hepatopancreatobiliary malignancy 14 (19)
sitioning procedures. Catheter adjustments No malignancy or recent surgery 25 (33)
were performed in 10 of 28 patients (36%)
Abscess size
with abscess size ≥ 150 cm3 compared with
7 of 47 patients (15%) with smaller abscesses Volume (cm3)
(p = 0.05). Catheter adjustment was associ- Mean 153
ated with a longer mean drainage duration of Median 53
48 (SD, 31) days compared with 23 (SD, 15)
Range 3–1331
days for unrevised drains (p = 0.001).
SD 191
Associations With Primary Clinical Failure Median dimensions (cm)c
Univariate and multivariate relations are Length 4.7 (2.0)
reported in Tables 2 and 4. Three independent
Width 6.0 (2.6)
variables were associated with a significant-
ly lower rate of primary success in univar- Height 5.8 (2.5)
iate and multivariate analyses. These were Note—Unless otherwise indicated, values are number of patients with percentage in parentheses. PCD =
unresectable hepatopancreatobiliary malig- percutaneous catheter drainage.
aThe remaining five patients each had a different malignancy.
nancy as the predisposing pathologic condi- bAbscess risk factors refer to procedures within 30 days before abscess drainage.
tion (p = 0.01), presence of multiple abscess- cValues in parentheses are SD.

es (p = 0.01), and catheter removal despite

daily output ≥ 15 mL/d (p = 0.001). No other
independent variable was associated with pri-
mary success, including advanced age, diabe-
tes, postsurgical abscess, large abscess size,
or loculations. The primary success rate was
lower among the 10 patients with biliary fis-
tula in univariate analysis, but this did not
reach statistical significance (Tables 2 and 4).
Unresectable hepatopancreatobiliary malig-
nancy—Unresectable hepatopancreatobiliary
malignancy was associated with a 57% (8/14)
primary failure rate (Table 2). The risk of pri-
mary failure among patients with these malig-
nancies was 2.4 times as high as that among pa-
tients with other causes of abscess (RR 95% CI,
1.4–5.2). Three primary failures were due to
symptomatic satellite collections, six to clinical
recurrence, and one to both symptomatic satel-
lite collections and clinical recurrence. Salvage
PCD procedures resulted in overall success for
11 patients (79%). Three septic deaths account-
ed for a lower overall success rate (p = 0.02).
Multiple abscesses—Multiple abscesses
were associated with a 46% (12/26) prima-

AJR:209, July 2017 207

 Haider et al.

ry failure rate (Table 2). The risk of prima- TABLE 2: Univariate Associations Between Demographic Characteristics,
ry failure among patients with multiple ab- Abscess Features, and Follow-Up Management and Primary Failure
scesses was 2.5 times as high as that among and Overall Success Rates
patients with a single abscess (RR 95% CI, No. With Relative Risk of Primary Failure No. With
1.2–5.2). Salvage PCD procedures resulted Total No. of Primary Overall
in overall success for 25 (96%) of these pa- Characteristic Patients Failures Risk 95% CI p Success
tients. One patient died of sepsis. Figure 1 Demographics
shows a patient with multiple abscesses re-
Age ≥ 55 y 57 (76) 17 (30) 1.3 0.5–3.5 0.76 53 (93)
quiring salvage PCD for abscess healing.
Primary failures included seven symptom- Diabetes mellitus 16 (21) 4 (25) 0.9 0.3–2.2 1 15 (94)
atic satellite collections, seven clinical re- Recent hepatic surgerya 19 (25) 5 (26) 0.9 0.4–2.2 1 19 (100)
currences, and cases of both symptomatic Recent nonhepatic 17 (23) 4 (24) 0.8 0.3–2.1 0.77 16 (94)
satellite collections and clinical recurrence. abdominal surgerya
There was an 86% (6/7) primary success rate Unresectable hepatopan- 14 (19) 8 (57) 2.7 1.4–5.2 0.02 11 (79)b
when initial drainage of all diagnosed col- creatobiliary malignancy
lections was performed, compared with 42%
No malignancy or recent 25 (33) 4 (16) 0.5 0.2–1.3 0.17 25 (100)
(8/19) when only the largest collection was surgery
initially drained (p = 0.08).
Abscess features
Removal of drain despite high output—There
was a 78% (7/9) primary failure rate among Multiple abscesses 26 (35) 12 (46) 2.5 1.2–5.2 0.01 25 (96)
patients who underwent catheter removal and Abscess volume > 150 cm3 28 (37) 7 (25) 0.8 0.4–1.8 0.79 27 (96)
had ongoing daily output ≥ 15 mL/d (range,
American Journal of Roentgenology 2017.209:205-213.

Fistula 10 (13) 5 (50) 2.0 1.0–4.3 0.13 9 (90)

20–90 mL/d) (Table 2). The risk of primary
Multiloculated abscess 42 (56) 11 (26) 0.9 0.4–1.8 0.80 40 (95)
failure among these patients was 4.1 times as
high as that among patients with lower end- Follow-up management
point drain output (RR 95% CI, 2.2–7.7). Catheter adjustments 17 (23) 4 (24) 0.8 0.3–2.1 0.77 16 (94)
Salvage PCD that was continued until drain Drainage duration ≥ 3 wk 41 (55) 11 (27) 0.9 0.4–1.9 1 38 (93)
output was minimal resulted in overall suc-
Drain output ≥ 15 mL/d at 9 (12) 7 (78) 4.1 2.2–7.7 0.001 9 (100)
cess for all nine of these patients (Table 2). catheter endpoint
Figure 2 shows primary failure after cathe-
Note—Except for relative risk, values are numbers of patients with percentages in parentheses.
ter removal despite ongoing high drain out- a Abscess risk factors refer to procedures within 30 days before abscess drainage.
put. The nine high-output catheter removals bp < 0.05 for lower rate of overall success.

were performed to ameliorate catheter-relat-

ed discomfort in patients who were in clin- TABLE 3: Cohort Outcomes
ically stable condition. The general surgery
service independently performed seven of Outcome No. of Patients
these removals. The seven primary failures Primary success 54 (72)
were solely due to abscess recurrence after
Primary failure 21 (28)
removal of the catheter.
Salvage success 17 (23)
Natural History of an Adequately Drained Drainage of symptomatic satellite collection 5
Hepatic Abscess Recurrence of original abscess 8
The rate of abscess cavity closure was es-
Satellite drainage and clinical recurrence 4
timated for successfully drained abscesses.
This idealized model excluded drains with Catheter adjustmentsa 17 (23)
high output at removal or recurrences, be- Overall success 71 (95)
cause these did not reflect optimal treatment. Death within 30 d 4 (5)
Successfully drained abscesses reached cav-
Note—Values in parentheses are percentages.
ity closure in a mean of 23 days (95% CI, a Adjustment to an existing catheter including caliber change or repositioning.

20–27 days). Cavity closure of large ab-

scesses (≥ 150 cm3) was reached a mean of
30 days (95% CI, 25–38), compared with 20
days (95% CI, 17–26 days) for abscesses <
150 cm3 (p = 0.01) (Fig. 3). This correlated
clinically to mean drainage duration of 35
(SD, 25) days for abscesses ≥ 150 cm3 com-
pared with 25 (SD, 20) days for smaller ab-
scesses (p = 0.03). There was no significant
difference in time to cavity closure based on
abscess loculation (p = 0.28) or multiplicity
(p = 0.67).
Discussion
Successful percutaneous drainage of
PLAs was achieved in 95% of patients, al-
though 23% needed catheter adjustments and
another 23% needed a second drain. These
results compare favorably to published qual-
ity standards for percutaneous abscess drain-
age and reported success rates of 77–90% in
similar PLA populations [7, 9, 10, 14, 17, 18].
Success rates reported for the percutane-
ous drainage of complicated PLA (i.e., large,
loculated, multiple, or malignancy-associ-

208 AJR:209, July 2017

 Percutaneous Drainage of Liver Abscesses
TABLE 4: Multivariate Associations Between Primary Failure Rate by
Subgroup and Failure Rate in Univariate Analysis
Subgroup No. of Patients
Unresectable hepatopancreatobiliary 14
malignancy
Multiple abscesses 26
Drain output ≥ 15 mL/d at catheter 9 42
endpoint
Fistula 10
No malignancy or recent surgery 25
ated) range from 33% to 90% [7, 12–14, 16,
24–28]. This has led to controversy regard-
ing the role of PCD versus surgical drainage,
several nonrandomized comparisons favor-
ing surgical drainage [13, 14, 24]. Some au-
thors have argued that an open surgical ap-
proach allows the breakdown of loculations
and that the placement of large (28-French)
American Journal of Roentgenology 2017.209:205-213.
surgical catheters improves drainage of ne-
crotic debris [24, 29]. In the current study,
however, we found no evidence of a lower
success rate of PCD for loculated or large ab-
scesses. Catheter adjustments were more of-
ten required in patients with large abscesses,
but this did not compromise the success rate.
This finding is concordant with the finding
of Liu et al. [9] that PCD-treated single and
multiple abscesses had equivalent success
rates regardless of loculation.
In several reports, persistent symptoms af-
ter drain placement were considered failures
of PCD that could lead to surgical drainage
[12–14, 24, 25, 30]. Tan et al. [14] reported
that 40% of PCD failures were due to cath-
eter blockage or dislodgments in a cohort
comparable to that in our study. Hope and
colleagues [13] also found PCD inferior to
surgical drainage for loculated abscesses > 3
cm, at 33% (8/24) versus 100% (27/27).
Proactive drain management may differ-
entiate the current results from reports of
high PCD failure rates [12–14, 25]. Our prac-
tice had a low threshold for performing a
contrast-enhanced drain study for drain dys-
function with the intent of providing early
catheter adjustment or salvage drainage. This
may have prevented overall PCD failure that
would require escalation of care to an open
surgical procedure.
Although patients with unresectable hepa-
topancreatobiliary malignancies had a pri-
mary success rate of only 43%, secondary
procedures improved the overall success rate
to 79%, which compared favorably with the
AJR:209, July 2017 209
Multivariate Regression of Primary Failure
Odds Ratio 95% CI p comparable studies [3–6, 9, 11, 14, 15, 26].
14.4 2.1–98 0.01
8.0 1.5–43 0.01
4.4–393 0.001
4.3 0.2–4.3 0.16
0.7 0.1–5.7 0.73
results of similar advanced oncology sub-
groups, in which abscess survival ranged
from 15% to 64% [12, 26].
Patients with multiple abscesses also had
a lower primary PCD success rate owing to
symptomatic satellite collections or primary
abscess recurrence. Salvage PCD for initial-
ly undrained satellite collections was com-
monly required if all collections were not
initially drained. Concordant to this logic, a
small subgroup of seven patients in whom
all collections were initially drained did
have a higher primary success rate. Persis-
tence of satellite collections may have led to
reseeding of the nearly healed primary ab-
scess cavity in abscess recurrences [29]. Pa-
tients with multiple abscesses may benefit
from cross-sectional imaging before drain
removal. Persistent satellite collections
could be drained to potentially avert clinical
abscess recurrence.
Catheter removal despite high drain output
represented a correctable factor and empha-
sized the need for communication between
the interventional radiology and the prima-
ry teams on the appropriate drain endpoint.
Ahmed et al. [19] reported that an output end-
point < 10 mL/d resulted in a 90% success rate
for large abscesses, which is concordant with
our experience. Although a drain may be un-
comfortable, and there is often a desire among
patients and referring clinicians to remove the
drain as soon as possible, we found no evi-
dence to support earlier removal of drains.
Successful drainage of larger abscesses and
those that required catheter adjustments was
associated with longer drain dwell times. This
is further supported by the idealized mean
time to abscess healing of 3–4 weeks.
Selection bias was a potential limitation of
this retrospective study because only patients
referred for percutaneous drainage were in-
cluded. There was no attempt to compare
PCD to surgical drainage or medical treat-
ment alone. Although the patient population
was heterogeneous in terms of underlying
pathologic condition, the demographics and
abscess complexity were similar to those in
We report on primary success in each of the
etiologic groups. The cohort was fairly small,
which may have underpowered detection of
small subgroup differences in primary fail-
ure rate. Patient care was multidisciplinary,
and this resulted in premature drain remov-
al in some patients. Although this may have
compromised the primary PCD success rate,
it reflects the challenges of true clinical prac-
tice. The general linear model of abscess vol-
ume over time was limited by a lack of im-
aging at the time of drain removal in some
patients. Furthermore, we assumed a linear
relation between abscess volume and time,
and although some patients deviated from
this relation (i.e., times to 50% reduction
may not be accurate), the time to cavity clo-
sure was well approximated.
Conclusion
In the study cohort, PCD was effective
first-line treatment of complicated PLAs
with overall success in 95% of patients. Sal-
vage drainage procedures were more often
required to achieve success in patients with
unresectable hepatopancreatobiliary malig-
nancies and multiple abscesses. Recurrence
was associated with catheter removal when
output was  ≥ 15 mL/d. Patients with large
abscesses needed more catheter adjustments
and had longer drainage duration, but this
did not compromise the success rate. Percu-
taneous PLA drainage effectiveness is opti-
mized by diligent drain follow-up and inter-
vention and multidisciplinary coordination
regarding drainage endpoints.
Acknowledgment
We thank Todd A. MacKenzie of the Dart-
mouth Clinical and Translational Science In-
stitute (National Institutes of Health award
UL1TR001086) for reviewing the statistical
methods in this study.
References
1. vanSonnenberg E, Mueller PR, Ferrucci JT Jr.
Percutaneous drainage of 250 abdominal abscess-
es and fluid collections. Part 1. Results, failures,
and complications. Radiology 1984; 151:337–341
2. Gerzof SG, Johnson WC, Robbins AH, Nabseth DC.
Intrahepatic pyogenic abscesses: treatment by percu-
taneous drainage. Am J Surg 1985; 149:487–494
3. Ferraioli G, Garlaschelli A, Zanaboni D, et al.

Percutaneous and surgical treatment of pyo-
genic liver abscesses: observation over a 21-
year period in 148 patients. Dig Liver Dis 2008;
40:690–696
4. Seeto RK, Rockey DC. Pyogenic liver abscess:
changes in etiology, management, and outcome.
Medicine (Baltimore) 1996; 75:99–113
5. Alvarez Pérez JA, González JJ, Baldonedo RF, et
al. Clinical course, treatment, and multivariate
analysis of risk factors for pyogenic liver abscess.
Am J Surg 2001; 181:177–186
6. Stain SC, Yellin AE, Donovan AJ, Brien HW.
Pyogenic liver abscess: modern treatment. Arch
Surg 1991; 126:991–996
7. Lo JZ, Leow JJ, Ng PL, et al. Predictors of therapy
failure in a series of 741 adult pyogenic liver ab-
scesses. J Hepatobiliary Pancreat Sci 2015;
22:156–165
8. Rahimian J, Wilson T, Oram V, Holzman RS.
Pyogenic liver abscess: recent trends in etiology
and mortality. Clin Infect Dis 2004; 39:1654–1659
9. Liu CH, Gervais DA, Hahn PF, Arellano RS, Up-
American Journal of Roentgenology 2017.209:205-213.
pot RN, Mueller PR. Percutaneous hepatic abscess
drainage: do multiple abscesses or multiloculated
abscesses preclude drainage or affect outcome? J
Vasc Interv Radiol 2009; 20:1059–1065
10. Singh S, Chaudhary P, Saxena N, Khandelwal S,
Poddar DD, Biswal UC. Treatment of liver ab-
scess: prospective randomized comparison of
catheter drainage and needle aspiration. Ann Gas-
troenterol 2013; 26:332–339
11. Alkofer B, Dufay C, Parienti JJ, Lepennec V,
Dargere S, Chiche L. Are pyogenic liver abscesses
still a surgical concern? A Western experience.
HPB Surg 2012;2012:316013
12. Chen SC, Lee YT, Tsai SJ, et al. Clinical out-
210 AJR:209, July 2017
Haider et al.
comes and prognostic factors of cancer patients
with pyogenic liver abscess. J Gastrointest Surg
2011; 15:2036–2043
13. Hope WW, Vrochides DV, Newcomb WL, Mayo-
Smith WW, Iannitti DA. Optimal treatment of
hepatic abscess. Am Surg 2008; 74:178–182
14. Tan YM, Chung AY, Chow PK, et al. An appraisal
of surgical and percutaneous drainage for pyo-
genic liver abscesses larger than 5 cm. Ann Surg
2005; 241:485–490
15. Pang TC, Fung T, Samra J, Hugh TJ, Smith RC.
Pyogenic liver abscess: an audit of 10 years’ expe-
rience. World J Gastroenterol 2011; 17:1622–1630
16. Lai KC, Cheng KS, Jeng LB, et al. Factors associ-
ated with treatment failure of percutaneous cath-
eter drainage for pyogenic liver abscess in patients
with hepatobiliary-pancreatic cancer. Am J Surg
2013; 205:52–57
17. Zerem E, Hadzic A. Sonographically guided per-
cutaneous catheter drainage versus needle aspira-
tion in the management of pyogenic liver abscess.
AJR 2007; 189:[web]W138–W142
18. Mangukiya DO, Darshan JR, Kanani VK, Gupta
ST. A prospective series case study of pyogenic
liver abscess: recent trends in etiology and man-
agement. Indian J Surg 2012; 74:385–390
19. Ahmed S, Chia CL, Junnarkar SP, Woon W, She-
lat VG. Percutaneous drainage for giant pyogenic
liver abscess: is it safe and sufficient? Am J Surg
2016; 211:95–101
20. Lewandowski RJ, Desai SB, Nemcek AA Jr.
Treatment of effusions and abscesses. In: Mauro
MA, Murphy KP, Thomson KR, Venbrux AC,
Morgan RA, eds. Image-guided interventions,
2nd ed. Philadelphia, PA: Saunders Elsevier
Health, 2014; 1137–1144
21. von Elm E, Altman DG, Egger M, et al. The
Strengthening the Reporting of Observational
Studies in Epidemiology (STROBE) statement:
guidelines for reporting observational studies.
J Clin Epidemiol 2008; 61:344–349
22. Zeger SL, Liang KY, Albert PS. Models for longi-
tudinal data: a generalized estimating equation
approach. Biometrics 1988; 44:1049–1060
23. Littell RC, Pendergast J, Natarajan R. Modelling
covariance structure in the analysis of repeated
measures data. Stat Med 2000; 19:1793–1819
24. Barakate MS, Stephen MS, Waugh RC, et al. Pyo-
genic liver abscess: a review of 10 years’ experi-
ence in management. Aust N Z J Surg 1999;
69:205–209
25. Chen SC, Tsai SJ, Chen CH, et al. Predictors of
mortality in patients with pyogenic liver abscess.
Neth J Med 2008; 66:196–203
26. Mezhir JJ, Fong Y, Jacks LM, et al. Current man-
agement of pyogenic liver abscess: surgery is now
second-line treatment. J Am Coll Surg 2010;
210:975–983
27. McDonald MI, Corey GR, Gallis HA, Durack DT.
Single and multiple pyogenic liver abscesses: nat-
ural history, diagnosis and treatment, with empha-
sis on percutaneous drainage. Medicine (Balti-
more) 1984; 63:291–302
28. Pearce NW, Knight R, Irving H, et al. Non-opera-
tive management of pyogenic liver abscess. HPB
(Oxford) 2003; 5:91–95
29. Farges O, Leese T, Bismuth H. Pyogenic liver ab-
scess: an improvement in prognosis. Br J Surg
1988; 75:862–865
30. Johnson RD, Mueller PR, Ferrucci JT Jr, et al. Per-
cutaneous drainage of pyogenic liver abscesses.
AJR 1985; 144:463–467
(Figures start on next page)
 Percutaneous Drainage of Liver Abscesses

A B
American Journal of Roentgenology 2017.209:205-213.

C D

Fig. 1—59-year-old previously healthy man with multiple abscesses who presented with 1 month of abdominal
pain, fever, and weight loss.
A, Contrast-enhanced CT image shows anterolateral (long white arrow), posterolateral (short white arrow), and
medial (black arrow) collections. Primary percutaneous catheter drainage of anterolateral and posterolateral
collections was performed with single catheter.
B, Contrast-enhanced CT image obtained 5 days after A because of persistent symptoms shows anterolateral
cavity improvement (long white arrow) but persistence of posterolateral (short white arrow) and medial (black
arrow) collections. Catheter was repositioned to better drain posterolateral collection, and multiple aspirations
of medial cavity were performed.
C, Follow-up fluoroscopic drain study image obtained because of persistent high output shows biliary fistula
without evidence of distal obstruction. Drainage was continued for 7 weeks until output was minimal and drain
study showed cavity closure. Oral antibiotics were continued for additional 2 months.
D, Follow-up CT image shows abscess resolution. There were no further complications.

AJR:209, July 2017 211

 Haider et al.

A B
American Journal of Roentgenology 2017.209:205-213.

C D

Fig. 2—57-year-old man with fever and abdominal pain due large single abscess recurrence caused by
premature drain removal 3 weeks after partial hepatectomy for colorectal cancer metastatic to liver.
A, Contrast-enhanced CT image shows unilocular 9-cm abscess in hepatic resection bed. Primary
percutaneous catheter drainage (PCD) was performed. Catheter was removed after 22 days despite 20-mL/d
output.
B, Patient presented with right upper quadrant pain 1 month after A. CT image shows persistent abscess.
Salvage PCD was performed.
C, Output was minimal after 11 days of salvage PCD. Fluoroscopic drain study shows residual 2.5-cm3 cavity.
Catheter was removed.
D, CT image obtained 2 months after C shows abscess resolution. There were no further complications.

212 AJR:209, July 2017

 Percutaneous Drainage of Liver Abscesses

110
Percentage of Original Abscess Volume

100 110
90 100

Percentage of Original Abscess

80 90

Volume With 95% CI

70 80
With 95% CI

70
60
60
50
50
40
40
30
30
20 20
10 10
0 0
0 7 14 21 28 35 0 7 14 21 28 35
Days Since Percutaneous Drain Placement Days Since Percutaneous Drain Placement

A B
Fig. 3—Idealized model of time to abscess cavity healing. Models were composed of each appropriately managed collection that did not recur.
A, Graph shows successfully drained small abscesses (< 150 cm3) reached cavity healing by 2.5–3.5 weeks (n = 58 abscesses, which excludes high-output drain removal
and recurrence). Of note, 75% were still > 5 cm in long axis. General linear mixed model, p < 0.001 for trend with cavity closure in 20 days (95% CI, 17–26 days).
B, Graph shows successfully drained large abscesses (≥ 150 cm 3) reached cavity healing by 5–6 weeks (n = 22 abscesses, which excludes high-output drain removal and
American Journal of Roentgenology 2017.209:205-213.

recurrence). This time was significantly longer than for smaller abscesses (p = 0.012). General linear mixed model, p < 0.0001 for trend with cavity closure in 30 days (95%
CI, 25–38 days).

F O R YO U R I N F O R M AT I O N
This article is available for CME and Self-Assessment (SA-CME) credit that satisfies Part II requirements for
maintenance of certification (MOC). To access the examination for this article, follow the prompts associated with
the online version of the article.

AJR:209, July 2017 213