Content uploaded by Reto Bale
Author content
All content in this area was uploaded by Reto Bale on Apr 30, 2016
Content may be subject to copyright.
CLINICAL INVESTIGATION
Stereotactic Radiofrequency Ablation (SRFA) of Liver Lesions:
Technique Effectiveness, Safety, and Interoperator Performance
Gerlig Widmann •Peter Schullian •
Marion Haidu •Reto Bale
Received: 28 February 2011 / Accepted: 22 May 2011
ÓSpringer Science+Business Media, LLC and the Cardiovascular and Interventional Radiological Society of Europe (CIRSE) 2011
Abstract
Purpose To evaluate technique effectiveness, safety, and
interoperator performance of stereotactic radiofrequency
ablation (SRFA) of liver lesions.
Methods Retrospective review including 90 consecutive
patients from January 2008 to January 2010 with 106
computed tomography-guided SRFA sessions using both
single and multiple electrodes for the treatment of 177
lesions: 72 hepatocellular carcinoma (HCC) and 105
metastases with a mean size of 2.9 cm (range 0.5–11 cm).
Technique effectiveness and 1-year local recurrence were
evaluated by computed tomographic scans. Complications,
mortality, and hospital days were recorded. The perfor-
mance between an experienced and inexperienced inter-
ventional radiologist was compared.
Results The overall technique effectiveness after a single
SRFA was 95.5% (93.1% for HCC and 97.1% for metas-
tases). Four of the eight unsuccessfully treated lesions
could be retreated (secondary technique effectiveness of
97.7%). Local recurrence at 1 year was 2.9%. Technique
effectiveness was significantly different for lesions \5cm
(96.7%) and [5 cm (87.5%) (P=0.044) but not for
lesions \3 cm (95.9%) and 3–5 cm (100%). Compared to
clear parenchymal property (97.3%), vessel vicinity
(93.3%) (P=0.349) and subcapsular (95.2%) (P=0.532)
had no, but hollow viscera vicinity (83.3%) had a signifi-
cantly lower technique effectiveness (P=0.020). Mortal-
ity rate was 0.9%. Major complications and hospital days
were higher for cirrhosis Child-Pugh B (20%, 7.2 days)
than Child-Pugh A (3.1%, 4.7 days) patients and for
metastases (5.1%, 4.3 days). There was no significant dif-
ference in interoperator performance.
Conclusion SRFA allowed for efficient, reliable, and safe
ablation of large-volume liver disease.
Keywords Ablation Liver tumor Hepatocellular
carcinoma (HCC) Liver/hepatic Navigation
Radiofrequency ablation Image fusion
Introduction
Percutaneous image-guided radiofrequency ablation (RFA)
is increasingly used as first-line treatment of hepatocellular
carcinoma (HCC) of \5 cm that is not suitable for liver
resection or transplantation; for bridging therapy before
liver transplantation; for therapy for recurrent and nonre-
sectable HCC; for alternative treatment for resectable HCC
sized 3 cm or less; and for local ablative therapy for liver
metastases [5–8,13,26,33]. RFA has proven to be a safe
procedure, with reported rates of major complications
ranging from 2 to 6% [19,21,27].
The success of RFA is excellent for small lesions sized
less than 3 cm [22,28]. In contrast, the results of large
(diameter [3–5 cm) and irregularly shaped (nonspherical)
lesions as well as lesions located subcapsularly or close to
major vessels are unsatisfying and may argue for excluding
such lesions from RFA treatment. Reasons for technical
G. Widmann P. Schullian M. Haidu R. Bale (&)
Department of Radiology, SIP-Department of Microinvasive
Therapy, Medical University of Innsbruck, Anichstr. 35,
6020 Innsbruck, Austria
e-mail: reto.bale@i-med.ac.at
G. Widmann
e-mail: gerlig.widmann@i-med.ac.at
P. Schullian
e-mail: peter.schullian@i-med.ac.at
M. Haidu
e-mail: marion.haidu@i-med.ac.at
123
Cardiovasc Intervent Radiol
DOI 10.1007/s00270-011-0200-4
failure may be poor visibility of the lesion and/or lesions
borders, imprecise electrode placement, insufficient abla-
tion size, the heat sink effect due to vicinity to large ves-
sels, and insufficient application of energy [15,18,28].
Lesion size and insufficient safety margin have been
identified as the most important influential factors [4,11,
22,24–26,37]. The recent introduction of stereotactic
radiofrequency ablation (SRFA) combines preinterven-
tional 3D imaging, computerized 3D planning of electrode
positions, and guided electrode placement at arbitrary
angulations and orientations [1–3]. SRFA may allow for
effective and safe treatment of even large lesions, lesions in
proximity to large vessels, or lesions in subcapsular loca-
tions [2]. Our purpose was to evaluate technique effective-
ness, safety, and interoperator performance of computed
tomography (CT)-guided SRFA.
Materials and Methods
The study was approved by the institutional review board.
According to the judgment of a multidisciplinary pro-
spective tumor board, clinical subjects with primary or
secondary liver tumors who were not eligible for surgical
treatment, who had medical comorbidities precluding sur-
gery, or who refused surgical treatment were referred for
SRFA. Informed consent was obtained from all patients.
Before treatment, clinical evaluation and general
anaesthetic evaluation including conventional liver bio-
chemical tests, prothrombin time, and complete blood cell
counts were performed. Exclusion criteria for SRFA were
lesions larger than 12 cm, more than 7 lesions, lesions
within 1 cm to the central bile duct, lesions infiltrating
adjacent hollow viscera, liver cirrhosis classified higher
than Child-Pugh class B, manifest ascites, prothrombin
time [23 s, prothrombin activity \40%, and platelet count
\40,000/ml. If the liver tumor had direct contact to hollow
viscera, laparoscopic liver packing with swabs immediately
before SRFA was performed in order to prevent thermal
damage (Bale et al., unpublished data).
Clinical Subjects
During the period January 1, 2008, to January 1, 2010, a
total of 90 consecutive patients (61 men, 29 women, mean
age 61.1 years, range 22–85 years) underwent a total of
106 CT-guided SRFA sessions for 177 lesions. A mean of
1.9 lesions per patient (range 1–7) were treated.
Forty patients had 72 HCCs and 47 SRFA sessions.
Twenty-eight patients (70%) with HCC had Child-Pugh A
cirrhosis, nine (22.5%) Child-Pugh B, and three (7.5%) no
cirrhosis.
Fifty patients had 105 hepatic metastases (54 colorectal,
18 mammary, 9 ovarian, 7 melanoma, 6 renal, 4 neuroen-
docrine, 3 bronchial, 3 papillary carcinoma, 1 teratoma)
and 59 SRFA sessions.
Tumors were confirmed by abdominal CT scans that
were no older than 6 weeks. The CT protocol included
native helical and contrast-enhanced scans using 90–120 ml
iodixanol 320 mg J/mg. The slice thickness for the native
and the portal venous phase were 5 mm, and for the arterial
and late phase 3 mm, respectively. For HCC, a native scan
and an arterial, portal, and late phase with a flow rate of
4 ml/s were obtained. For neuroendocrine tumors, a native,
arterial, and portal phase and for all other entities a native
and a portal venous phase were acquired with a flow rate of
3 ml/s. The mean tumor size of all 177 lesions was 2.9 cm
(range 0.5–11 cm). The HCCs had a mean size of 3 cm
(range 0.5–9.5 cm). The metastases had a mean size of
2.8 cm (range 0.5–11 cm). A total of 123 lesions (69.5%)
were \3 cm, 30 lesions (16.9%) were 3–5 cm, and 24
lesions (13.6%) were[5 cm.
Seventy-three lesions (41%) were clear parenchymal
(lesions with[1 cm distance from the liver capsule, [1cm
distance to an adjacent hollow viscera, and [1 cm distance
vessels that sized [3 mm, 63 lesions (35.6%) were sub-
capsular (lesions \
1 cm close to the liver capsule), 30
lesions (16.9%) had vessel vicinity (\1 cm close to a
vessel that sized [3 mm), and 18 lesions (10.2%) had
hollow viscera vicinity (subcapsular lesions \1 cm close to
an adjacent hollow viscera). Eleven of the 18 lesions with
hollow viscera vicinity had direct contact with the hollow
viscera, requiring separation by liver packing.
Stereotactic Radiofrequency Ablation
All patients were admitted to the referring departments of
gastroenterology, general oncology, or surgery.
All SRFAs were performed at the department of radi-
ology in an interventional suite (class II surgical room,
according to national hospital standards) with a multislice
(40) CT Siemens Somatom Sensation Open (Siemens,
Munich, Germany) that has a 82-cm sliding gantry that
provided full diagnostic image quality.
More information on SRFA is available elsewhere
[1–3]. A flowchart of the procedure is provided in Fig. 1.
Patient Preparation and Respiratory Motion Control
After general anesthesia and muscle relaxation, patients
were rigidly fixed at the table with the BlueBAG patient
fixation system (Medical Intelligence, Schwabmu
¨nchen,
Germany) [38]. For lesions in direct contact with hollow
viscera, liver packing was performed in the intervention
G. Widmann et al.: SRFA of Liver Lesions
123
room by surgeons with extensive experience in laparo-
scopic surgery.
In lesions of the right liver lobe, patients were posi-
tioned in slight right anterior oblique position, and the right
arm was fixed to an adjustable mechanical arm in an ele-
vated position. A patient-covering bridge was fixed at the
table for placement of the dynamic reference frame of the
navigation system.
To allow for an image-to-patient registration with the
navigation system, self-attachable skin markers (Beekley
Spots, Beekley, Bristol, CT) were placed broadly around
the region of interest.
For respiratory motion control, planning, and control CT
scans, the registration procedure and the targeting were
performed at maximal expiration during temporary dis-
connection of the endotracheal tube, which was executed
by the attending anesthetist in the intervention room [38].
Planning CT
A contrast-enhanced (90–120 ml iodixanol 320 mg J/mg,
3 ml/s) CT scan of the liver with 3-mm slice thickness was
performed in the arterial or portal phase. Three-millimeter
slice thickness provides far better targeting accuracy than
5mm [34]. The CT data were immediately sent via
Intranet to the Treon Plus navigation system (Medtronic,
Louisville, KY).
3D Treatment Planning
According to the size and geometry of the lesion, one or
multiple electrode positions were planned on the naviga-
tion software as visualized on the 2D and 3D CT recon-
structions. By using the longitudinal and orthogonal cuts
along the planned path, potential penetration of vital
structures could be prevented [2].
The following scheme was used. Electrodes were placed
in and around the lesion with an interprobe distance of a
maximum of 2 cm. Electrodes at the lesion border were
placed at the interface between tumor and normal liver
tissue to assure the goal of a safety margin of 1 cm of
healthy tissue. In lesions that were close to large vessels,
the electrodes were planned to position approximately
5 mm close to the vessel to address the so-called heat sink
effect. If a visible feeding vessel entered the tumor, this
area was ablated first to reduce tissue cooling. In subcap-
sular lesions, the liver capsule was included in the radio-
frequency necrosis. In lesions located close to adjacent
Fig. 1 Flowchart of radiofrequency ablation (SRFA)
G. Widmann et al.: SRFA of Liver Lesions
123
hollow viscera, a safety distance of at least 1 cm of the
electrode position to the covering liver capsule was
respected.
Image-to-Patient Registration and Stereotactic
Placement of Coaxial Needles
A paired-point registration procedure was performed that
consisted of definition of coordinates of corresponding
Beekley Spots in the image (manually on the computer)
and on the patient (with the probe of the navigation system)
and calculation of the geometrical transformation that best
aligns these points. For safety reasons, the accuracy of the
navigation was checked by touching Beekley Spots on the
patient with the probe of the navigation system and com-
paring the position as indicated by the navigation system.
After registration, the treatment area was scrubbed and
draped. The sterilized Atlas aiming device (Medtronic,
Louisville, KY) was mounted to the draped attachment
bridge on the intervention table. The device is a lockable
jointed arm with a two joints head that enables the user to
align the probe of the navigation system with the planned
trajectory. After alignment, the depth from the aiming
device to the target point was automatically calculated and
marked on a 15-gauge, 17.2-cm coaxial needle (Bard,
Covington, GA). The navigation probe was removed from
the aiming device, and the adjustable breaks on the head
were narrowed to guide the coaxial needle. One by one,
each coaxial needle was advanced through the aiming
device to the preplanned depth during short (10 s) tempo-
rary endotracheal tube disconnections [39]. A mean of 4.1
(median 3, range 1–14) coaxial needles per lesion were
positioned. A single needle positioning required approxi-
mately 2 min.
Before ablation, a native control CT with 3-mm slice
thickness (similar scan protocol as for the planning CT)
was performed to verify the needle positions and to exclude
puncture-related complications such as pneumothorax or
bleeding. The data were fused to the planning CT and
allowed the actual needle position to be checked in relation
to the planned path and target lesion [39].
Radiofrequency Ablation
The unipolar Covidien Cool-Tip radiofrequency system
(Covidien, Mansfield, MA), including the Cool-Tip radio-
frequency switching controller and a 17-gauge Cool-Tip
electrode with a length of 25 cm and a tip exposure of
3 cm, were used. The electrodes were inserted through the
17.2-cm coaxial needles and placed tip to tip. Afterward,
the shorter coaxial needles were retracted in order to
uncover the 3-cm active electrode exposure. Thus, repeat
imaging for electrode positioning control was not required.
Following the recommendations of the manufacturer, all
ablations were performed using the switching control mode
for up to three electrodes during a 16-min ablation per
cycle. For a single electrode, the recommended 12 min
were used. Mean ±SD ablation time per treatment session
(n=106) was 64.2 ±44.3 min (median 50 min, range
12–210 min).
Tract ablation was performed during every probe
retraction (even if the electrode was not removed from the
liver capsule) in order to prevent tumor recurrence,
bleeding and potential tumor seeding. A final portal phase
contrast-enhanced control CT (90–120 ml iodixanol
320 mg J/mg, 3 ml/s) was performed to exclude treatment-
related complications. In the case that required laparo-
scopic liver packing, the surgical team immediately
removed all glycol-saturated drapes, and the laparoscopic
sites were closed.
After regular postoperative intensive care observation,
the patients were sent to the referring departments of gas-
troenterology, general oncology, or surgery. The time point
of discharge from the hospital was decided by the referring
physicians, depending on each patient’s individual postin-
terventional recovery and general physical state. There was
neither an intention for early discharge nor an intention for
an observation longer than necessary. The interventional
radiologist had no influence on the discharge time of the
patients.
Assessment of Technique Effectiveness and Local
Recurrence
Technique effectiveness and local recurrence were evalu-
ated per treated lesion.
The following parameters were evaluated: technique
effectiveness, local tumor progression, and complications.
The parameters were evaluated on postinterventional con-
trast-enhanced abdominal CT at 1, 3, 6, and 12 months. The
CT scans were performed according to the routine diag-
nostic protocol of our university. The evaluation was per-
formed by an independent reader who did not perform the
interventions by collecting the standard radiologic reports
during follow-up in the department of radiology. Two cases
of SRFA that were hard to treat with conventional tech-
niques in a single session are illustrated in Figs. 2and 3.
Technique effectiveness was defined as complete abla-
tion on contrast-enhanced imaging at 1 month after SRFA.
Local recurrence was defined as evidence of newly detec-
ted tumor at the site of ablation at contrast-enhanced
imaging at 3, 6, and 12 months after CT-guided SRFA.
Complete ablation of a treated lesion was defined as cir-
cumscribed nonenhancing region within and extended
beyond the tumor borders, eventually surrounded by a
reactive rim enhancement [5]. Remaining or newly
G. Widmann et al.: SRFA of Liver Lesions
123
developed contrast enhancement of the ablated lesion was
considered as vital tumor tissue [5].
Subanalysis of technique effectiveness rate was per-
formed for entity (HCC vs. metastases), lesion size (\3cm
vs. 3–5 cm vs. [5 cm; [5 cm vs. \5 cm; and [3cm
vs. \3 cm), properties (clear parenchymal vs. vessel
vicinity; clear parenchymal vs. subcapsular; clear paren-
chymal vs. hollow viscera vicinity), and location (liver
segments I to VIII).
Assessment of Complications and Hospital Days
Complications and hospital days were evaluated per
ablation.
Major complications related to needle insertion or the
ablation procedure were defined as if they were left untreated
without invasive intervention might threaten the patient’s
life, lead to substantial morbidity and disability, or result in
an extended hospital stay.
Fig. 2 A 54-year-old man with HCC showing three lesions (white
arrows) that were simultaneously treated by a single SRFA. (1)A
3-cm lesion subcapsular segment V (3 coaxial needles/3 electrode
positions). (2) A 1.2-cm lesion segment I (1 coaxial needle/1
electrode position). (3) A 1.3-cm lesion segment IVb (2 coaxial
needles/2 electrode positions). (10), (20), and (30) refer to the
corresponding postinterventional CT scans at 1 month. Note the
sufficient larger ablation zones (white arrowheads) confirming
technique effectiveness
Fig. 3 A 70-year-old woman with HCC showing a 4.5-cm lesion
(white arrows) in segment VIII in broad contact to the middle and
right hepatic vein (8 coaxial needles/18 electrode positions). (1)
Before CT-guided SRFA. (10) Postinterventional CT at 1 month
showing sufficient larger ablation zone (white arrowheads)
confirming technique effectiveness. (100) Postinterventional CT at
12 months showing shrinkage of the ablation zone (black arrow-
heads) confirming no local tumor progression. The slice positions of
the images were chosen to demonstrate the relationship of the ablation
and the adjacent vessels
G. Widmann et al.: SRFA of Liver Lesions
123
Minor complications were defined as complications and
relevant side effects that did not require invasive inter-
vention. Mortality rate was defined as the direct or indirect
procedure-related death rate within 6 months after
intervention.
Hospital days from the day of SRFA to the day of dis-
charge were calculated.
Complications and hospital days were compared for the
groups HCC in cirrhosis Child-Pugh A vs. Child-Pugh B
vs. metastases.
Assessment of Interoperator Performance
Interoperator performance between an experienced and
inexperienced interventional radiologist was calculated to
assess operator-dependent differences of SRFA. The
experienced radiologist (Author 1) had more than 10 years
of experience in percutaneous radiofrequency ablation and
stereotactic interventions. The inexperienced radiologist
(Author 2) had 4 years of residency training in radiology
and clinical routine in ultrasound (US)-guided and
CT-guided liver biopsies. Author 2 had no previous expe-
rience in RFA and started SRFA after a training period of
2 months. All treatment planning, electrode positioning,
and ablations were performed exclusively and completely
by either Author 1 or Author 2.
Statistical Analysis
Data analysis and descriptive statistics were performed by
PASW Statistics 17.0 (SPSS, Chicago, IL). Quantitative
data were described with mean values, standard deviation,
median, and range.
Pearson’s chi-square test was used to determine signif-
icant statistical differences between the groups ‘‘entity,’’
‘‘lesion size,’’ ‘‘properties,’’ and ‘‘location.’’ The same test
was used to determine statistically significant differences
between Author 1 and Author 2.
Student’s t-test was used to determine significant sta-
tistical differences in hospital days between HCC in cir-
rhosis Child-Pugh A vs. Child-Pugh B vs. metastases.
For all statistical tests, a P-value of less than 0.05 was
defined as statistically significant.
Results
Technique Effectiveness and Local Recurrence
All patients and lesions assigned to SRFA could be treated.
Technique effectiveness after one treatment session of
CT-guided SRFA was 95.5% per treated lesion (Table 1).
Of 177 lesions, 8 lesions in 7 patients were unsuccess-
fully treated, leading to technique effectiveness per patient
of 92.2%. Four lesions could be retreated by a second
CT-guided SRFA. One lesion showed local recurrence at
6 months after reintervention, and 4 lesions could not be
retreated as a result of progressive disease, other treatment,
or refusal of further treatment. The failed cases are pre-
sented in Table 2.
The failures of SRFA occurred in 4 lesions B2cmin
size using only one electrode and in 3 large lesions ranging
from 6 to 8 cm, one having vessel vicinity, one subcapsular
and one having subcapsular and vessel vicinity properties.
Local recurrence after technically successful SRFA was
0% at 3 months, 1.2% (2/173) at 6 months, and 2.9%
(5/173) at 12 months.
Subanalysis of technique effectiveness revealed the
following.
Entity
HCC showed no significant difference in technique effec-
tiveness rates (93.1%) compared to metastases (97.1%)
(P=0.198).
Table 1 Characteristics of 177 liver lesions in 90 patients
Characteristic Overall Author 1 Author 2 P-value
Patients (n)905535
Tumor size (cm), mean ±SD (range) 2.9 ±2.1 (0.5–11) 2.8 ±1.9 (0.5–9) 3 ±2.2 (0.5–11) 0.754
Lesions per patient (n), mean ±SD (range) 1.9 ±1.3 (1–7) 1.9 ±1.3 (1–7) 1.9 ±1.2 (1–5) 0.872
Needles per lesion (n), mean ±SD (range) 4.1 ±3 (1–14) 4 ±3 (1–14) 4.2 ±3.3 (1–14) 0.734
Technique effectiveness 95.5% (169/177) 95.9% (93/97) 95% (76/80) 0.780
Early local tumor progression (3 months, successfully treated) 0% 0% 0%
Local tumor progression (6 months, successfully treated) 1.1% (2/177) 2.1% (2/97) 0% 0.196
Major complications (intervention necessary) 6.6% (7/106) 6.6% (4/61) 6.7% (3/45) 0.982
Mortality 0.9% (1/106) 1.6% (1/61) 0% (0/45) 0.388
Hospital stay (days), mean ±SD (range) 4 ±2.4 (1–11) 4 ±2.4 (1–11) 4 ±2.5 (1–9) 0.749
G. Widmann et al.: SRFA of Liver Lesions
123
Lesion Size
The analyzed groups showed no significant difference
when divided into \3 cm (95.9%) vs. 3–5 cm (100%)
vs. [5 cm (87.5%) (P=0.083), and [3 cm (94.4%)
vs. \3 cm (95.5%) (P=0.669). Significant differences
were found for \5 cm (96.7%) vs. [5 cm (87.5%)
(P=0.044).
Properties
Statistical analysis revealed no significant difference of
clear parenchymal lesions (97.3%) compared to vessel
vicinity (93.3%) (P=0.349) and subcapsular (95.2%)
(P=0.532). In contrast, significant difference was found
for hollow viscera vicinity (83.3%) (P=0.020).
No significant differences were observed (P=0.125)
when comparing technique effectiveness of treatment of
lesions close to hollow viscera that did not require liver
packing (85.7%) and clear parenchymal location. How-
ever, significant differences were observed between the
treatment of lesions with clear parenchymal location and
lesions with direct contact to hollow viscera requiring
liver packing (81.8%) (P=0.025). There was no signif-
icant difference in the size of the tumors with hollow
viscera vicinity that were treated with (mean 4.2 cm,
median 4.5 cm, range 1.3–7 cm) or without (mean
3.9 cm, median 2.8 cm, range 1.3–11 cm) liver packing
(P=0.616).
Location
Statistical analysis of lesion location as a factor for tech-
nique effectiveness showed that there were no significant
differences among segments. Lesions located in segment II
or III did not show significant difference to the other seg-
ments (P=0.125).
Complications and Hospital Days
Mortality rate was 0.9% (1 patient, sepsis and multiorgan
failure). Major and minor complication rates were 6.6%
and 21.7%, respectively (Table 3).
In a subanalysis including both major and minor com-
plication rates related to entity, lesion size, properties, and
location, the following statistically significant differences
were found: HCC 34.7% (25/72) vs. metastases 19% (20/
105) (P=0.019), and [3 cm 37% (20/54) vs. \3cm
20.5% (25/122) (P=0.020). All other parameters did not
show statistically significant differences.
Of all 106 ablations, major complications occurred in
3.1% in HCC in cirrhosis Child-Pugh A (1/32), 20%
in HCC in cirrhosis Child-Pugh B (2/10), and in 5.1% in
metastases (3/59). (Note that 5 patients with HCC had no
cirrhosis.) Chi-square test showed P=0.131.
Excluding the patient who died and the ablations with
liver packing, a mean of 4 and a median of 3 hospital
days (range 1–11 days) were recorded from the day of
SRFA to discharge from the hospital. In a subanalysis for
Table 2 Details of the failed cases with residual tumor after SRFA
Sex Age Entity Location Property Lesion
size (cm)
Coaxial
needles
Supplemental information Repeat
SRFA
M 78 HCC III Subcapsular 1 1 Targeting imprecision Yes
M 78 HCC VII Clear parenchymal 1.2 1 Small residual tumor and multiple
disseminated new lesions
No
M 53 HCC I Vessel vicinity 1.5 1 Previous transarterial chemoembolization
and percutaneous ethanol injection due
to vicinity to central bile duct; SRFA
after size reduction; residual tumor was
at the bile duct
No
M 75 HCC II Hollow viscera vicinity 2 1 Insufficient ablation size No
M 69 Meta CRC II Hollow viscera vicinity 2.4 3 Small residual tumor close to the stomach Yes
M 74 HCC VIII Subphrenic 6 5 Insufficient ablation size Yes
M 59 Meta RCC VI Vessel vicinity, subcapsular,
hollow viscera vicinity
6 9 Small residual tumor close to the VCI and
lateral subcapsular
No
M 58 Meta CRC VIII
(main)
Vessel vicinity 8 12 Central lesion surrounding and attaching
right, middle and left hepatic vein, VCI,
and portal vein; small residual tumor at
the portal vein
No
RCC Renal cell carcinoma, CRC colorectal carcinoma
G. Widmann et al.: SRFA of Liver Lesions
123
longer hospital stays ([5 hospital days) related to the
above parameters, the following statistically significant
differences were found: \3 cm 23% (28/122) vs. 3–5 cm
53.3% (16/30) vs. [5 cm 62.5% (15/24) (P=0.000). All
other parameters did not show statistically significant
differences.
The corresponding hospital days for SRFA of HCC in
cirrhosis Child-Pugh A (n=32) showed mean 4.7 ±
3.8 days (median 4 days); for Child-Pugh B (n=10)
showed mean 7.2 ±6 days (median 5.5 days); and for
metastases (n=59) showed mean 4.3 ±2.5 days (median
4 days). The t-tests showed the following: HCC in cirrhosis
Child-Pugh B vs. metastases, P=0.187; HCC in cirrhosis
Child-Pugh B vs. Child-Pugh A, P=0.135; and HCC in
cirrhosis Child-Pugh A vs. metastases, P=0.538.
For the ablations with liver packing, a mean of 7.76 and
a median of 6 hospital days (range 4–20 days) were found,
which was significantly higher than without liver packing
(P=0.002).
Interoperator Performance
A total of 55 patients were treated by Author 1 and 35
patients by Author 2 (Table 1). There were no significant
differences in patient characteristics for age (P=0.822),
sex (P=0.898), lesion size (P=0.754), number of
lesions per patient (P=0.872), and number of coaxial
needles per lesion (P=0.734) for the two operators.
There were no significant differences in technique
effectiveness (P=0.780), major complications (P=
0.889), and hospital days (P=0.749).
Discussion
In the presented SRFA protocol, one or multiple cooled
monopolar electrodes were placed at one or multiple
electrode positions via coaxial needles at a maximum
interelectrode distance of 2 cm, depending on the tumor
size and shape. If multiple electrodes were used, they were
activated by the switching control mode, where the power
is automatically switched from one electrode to up to three
others depending on the impedance. Lee et al. [14] showed
that when using three cooled-tip monopolar electrodes, the
circularity (isometric ratio) of the ablation zone decreased
with increasing interelectrode distance, and that interelec-
trode distances of larger than 3 cm could not create con-
fluent coagulation necrosis. Multitined expandable
electrodes may not be useful in our approach to position
multiple electrodes very close to vessels, the diaphragm,
and the liver capsule [4,10,16,26,30]. Multipolar elec-
trodes are feasible but require strict parallelism of the
electrode positions to ensure confluent necroses [31]. In
conventional RFA, usually only one electrode or clusters of
three electrodes are used. For larger tumors, repeat abla-
tions are performed. The multielectrode/electrode position
approach in SRFA allowed for generation of arbitrary large
overlapping ablation necroses. Thus, unlike conventional
RFA, individual features and ablation sizes of different
electrode systems do not influence the success of SRFA.
However, in principle, the procedure may be adapted to
any radiofrequency electrode system, as well as to cryo-
ablation and microwave technologies.
By using conventional US-guided RFA for treatment of
HCC in 42 patients with 52 lesions B3 cm, with a mean of
Table 3 Details of mortality
and complications of 106
ablations in 90 patients
Characteristic Value Therapy
Mortality 0.9% (1/106)
Sepsis and organ failure 0.9% (1/106)
Major complications (intervention necessary) 7.5% (8/106)
Pneumothorax 2.8% (3/106) Drainage
Pleural effusion 0.9% (1/106) Drainage
Liver abscess 0.9% (1/106) Drainage and antibiotics
Perihepatic bleeding 0.9% (1/106) Coiling
Multiple organ dysfunction syndrome 0.9% (1/106) Intensive care unit
Minor complications (no intervention necessary) 21.7% (23/106)
Pleural effusion 12.3% (13/106)
Pneumothorax 2.8% (3/106)
Subcapsular hematoma 1.9% (2/106)
Perihepatic bleeding 1.9% (2/106)
Temporary blood pressure increase 0.9% (1/106)
Small pulmonary embolism 0.9% (1/106)
Skin burn 0.9% (1/106)
G. Widmann et al.: SRFA of Liver Lesions
123
1.2 treatment sessions, Livraghi et al. [17] achieved a
technique effectiveness rate of 90% (47/52). However, in
126 HCC lesions 3.1 cm or larger in diameter in 114
consecutive patients, the same authors reported technique
effectiveness of only 47.6% (60/126) [18]. In the series
of Seror et al. [31] with US-guided multipolar RFA of
HCC [5 cm (median size 5.7 cm, range 5.0–8.5 cm), a
technique effectiveness of 81% (22/27) was achieved after
a mean of 1.2 treatment sessions. Tumors larger than 8 cm
could not be successfully ablated.
For treatment of colorectal liver metastases using con-
ventional US-guided RFA in 88 consecutive patients with
134 lesions (mean size 2.1 cm, range 0.6–4 cm) and 119
ablations, Livraghi et al. [20] reported a technique effec-
tiveness of 63% (85/134), and Solbiati et al. [32] reported a
technical success of 98% (176/179) when treating 117
patients with 179 lesions (mean size 2.8 cm, range
0.6–9.6 cm). Chen et al. [4] used a mathematical model for
creation of overlapping necroses with US-guided RFA in a
total of 332 patients with 503 liver lesions (mean size
4.0 cm, range 1–8 cm) who were treated in a mean of
1.2–1.6 sessions; they reported technique effectiveness
rates of 95.8% (295/308) for HCC and 94.9% (185/195) for
liver metastases. The technique effectiveness in large
([3.5 cm) and small lesions was 91.3% (189/207) and
98.3% (291/296), respectively.
Livraghi et al. [17,18,20], Solbiati et al. [32], Seror
et al. [31], and Chen et al. [4] each performed repeat ses-
sions (up to 11) of conventional US-guided RFA, if pos-
sible, when residual tumor was identified on initial
posttreatment CT. Thus, reported technique effectiveness
often refers to secondary technique effectiveness [9].
SRFA provided a technique effectiveness after a single
treatment session (=primary technique effectiveness rate)
of 95.5% (169/177). For 54 lesions [3 cm, technique
effectiveness was 94.4%, and for the 24 lesions [5 cm,
87.5%. Compared to present techniques, SRFA allowed for
successful ablation of lesions [8 cm with a maximum
lesion size of 11 cm. Four of the 8 incompletely ablated
lesions could be treated by a second SRFA and were
completely ablated. Thus, SRFA finally showed a sec-
ondary technique effectiveness of 97.7%. One year after
successful CT-guided SRFA, a total of 2.9% (5/173) local
recurrences were found.
HCC vs. metastases did not show statistically significant
differences (93.1% vs. 97.1%). Remarkably, vessel vicinity
and subcapsular did not significantly influence technique
effectiveness compared to clear parenchymal, which is a
clear difference from a previous meta-analysis of RFA
[22]. Obviously, following the concept of placing elec-
trodes very close to vessels successfully overcame the heat
sink effects [2,3,35]. Only the hollow viscera vicinity
showed significantly lower technique effectiveness of
83.3% (15/18) compared to clear parenchymal with
97.3% (71/73). Compared to the mean overall tumor size
of 2.9 cm, the group of hollow viscera vicinity had a
mean tumor size of 4 cm. Remarkably, hollow viscera
vicinity that did not require liver packing (7/18) had no
significant different technique effectiveness compared to
clear parenchymal. However, lesions that required laparo-
scopic liver packing still were difficult to treat and had
worse results. In sum, the most important sources of failure
in the presented series of SRFA were insufficient targeting
and insufficient ablation size related to using only one
electrode, and very large tumor size with vessel vicinity
and subcapsular and hollow viscera vicinity. The impor-
tance of large ablation margins of healthy tissue sur-
rounding the tumors cannot be overemphasized.
In conventional RFA, the ablations were mostly treated
as outpatients or with one night’s stay. A multicenter study
of complications of percutaneous RFA of 3554 lesions in
2320 patients showed a mortality rate and major compli-
cation rate of 0.3 and 2.2%, respectively [19]. In the pre-
sented SRFA series, the mortality rate and major
complication rates of 0.9 and 6.6%, respectively, were
higher. The multineedle/needle position approach for sin-
gle-session treatment of up to 7 lesions per patient and a
mean of 4.1 coaxial needles per lesion (the largest lesions
with 14 coaxial needles) while the patient was under gen-
eral anaesthesia was certainly more aggressive than con-
ventional RFA techniques. With these considerations, the
complication rate and hospital days were still small,
and substantially smaller than that of surgical resection.
Most major complications, including pneumothorax, pleu-
ral effusion, liver abscess, and hepatic bleeding, could be
successfully treated by interventional therapy. Significantly
higher overall complication rates were found for
lesions [3 cm (37%) compared to lesions \3 cm (20.5%),
but remarkably, vessel vicinity, and subcapsular and hol-
low viscera vicinity did not show significant differences to
clear parenchymal. The overall complication rates were
significantly higher for HCC (34.7%) than for metastases
(19%), and major complications were substantially higher
for cirrhosis Child-Pugh B (20%) compared to Child-Pugh
A (3.1%) and metastases (5.1%). Patients were discharged
from the hospital after a mean and median of 4 and 3 days,
respectively. Patients needed a significantly longer hospital
stay of more than 5 days in correlation to tumor size in
62.5% of lesions [5 cm and only 23% of lesions \3 cm,
but not for other lesion properties. This may reflect the
consequences of the increasing aggressiveness that is
required for successful treatment of large lesions during
one treatment session. Cirrhosis Child-Pugh B had more
median hospital days (5 days) than Child-Pugh A and
metastases (4 days each). Obviously, patients who required
laparoscopic liver packing for treatment of lesions with
G. Widmann et al.: SRFA of Liver Lesions
123
contact to hollow viscera needed more hospital days than
patients without liver packing.
Real-time virtual sonography, synchronizing preinter-
ventional multiplanar reformatted CT, or magnetic reso-
nance imaging to B-mode US may improve targeting and
allows for the treatment of liver lesions that are undetect-
able by US [23,29].
However, conventional approaches may not be practi-
cable for lesions requiring multiple electrode positions as a
result of the three-dimensional problem considering opti-
mal ablation criteria, safety, and access routes surrounded
by anatomic obstacles [22,28]. In the presented series of
SRFA, all liver lesions in all locations could be treated with
no limitations concerning access or number and complexity
of electrode positions. In contrast to reports that enforce the
role of the experience of the interventional radiologist
performing RFA [5,28], SRFA did not reveal significant
differences of treatment success and morbidity rates if it
was executed by an experienced or inexperienced inter-
ventionalist, and the procedure seemed to greatly enhance
our outcomes. Because both operators had similar patients,
lesion characteristics, and technical characteristics, a bias
related to size and complexity could be ruled out.
A drawback of the optical navigation system was the line-
of-sight problem, which may constrain navigation after
insertion of a large number of needles and which may
require repositioning of the stereoscopic camera. In contrast,
electromagnetic navigation has no line-of-sight problem, but
position measuring is less stable and significantly influenced
by metallic distortion [36,40,41].
Disadvantages of this high-end guidance technology are
the increased requirements on infrastructure, staff, and
specialized training in 3D-navigated stereotaxy [1,2,12].
A multifunctional intervention room with full anesthesia
equipment is required. Large gantry diameters for inter-
ventional radiology are not widely available but should be.
Sliding gantries allow moving the CT between two sepa-
rable rooms and may not block the CT scanner during the
entire intervention.
Limitations of the study are as follows. The study is a
retrospective review of consecutive patients treated with
SRFA in a single center. Unfortunately, worldwide, no
other center performs SRFA or any other form of stereo-
tactic ablative therapy, and we could not compare our
results to others. We could not conduct a prospective
randomized, controlled study comparing SRFA with
US-guided RFA because at our university hospital,
US-guided RFA is no longer accepted, the result of the
substantial technical advantages of SRFA.
Although data from consecutive patients during a
period of 2 years were reviewed, the number of patients
(n=90) and number of treated lesions (n=177) are
small—perhaps too small to show significant differences
in some of the parameters for subanalysis. Related to the
inhomogeneous group of metastases and small numbers, a
differentiated subanalysis of metastases was not per-
formed. The wide range of sizes (0.5–11 cm) makes
definite conclusions difficult, and it is hard to draw major
conclusions from one experienced vs. one less experi-
enced interventional radiologist. The results should be
confirmed by others at different medical centers.
The main purpose of the study was to evaluate tech-
nique effectiveness and safety of liver lesions treated by
SRFA. We did not perform a statistical multivariable
evaluation for technical failure and local recurrence
because we had only 8 failures and 5 local recurrences.
The reasons for the technical failures were discussed in
detail. We did not evaluate the potential influence of
systemic therapy on technique effectiveness and local
recurrence. The local recurrence during 1 year was
included in our evaluation; however, Kaplan–Meier
analysis on overall and recurrence-free survival was not
the purpose of this study.
It is difficult to draw a clear line between incomplete
ablation determined as ‘‘technically ineffective’’ and ‘‘local
recurrence.’’ In the presented study, 1 month’s control was
used to determine technique effectiveness, and the sub-
sequent 3, 6, and 12 months’ control was used to determine
local recurrence. Local recurrence (not distant recurrence
of a new clone near the ablated lesion) may always be the
result of an insufficient ablation margin and thus incom-
plete ablation. However, at 3 months, local recurrence after
technical successful SRFA was not noted.
In conclusion, compared to conventional image-guided
percutaneous RFA, SRFA is a technically advanced pro-
cedure that integrates interventional 3D imaging, 3D
computerized treatment planning, and 3D stereotactic
guidance. Technique effectiveness after a single session of
SRFA for even multiple lesions per patient was 95.5%.
There were no significant performance differences between
an experienced radiologist and a less experienced radiolo-
gist. Because of careful 3D planning and stereotactic
placement of multiple electrodes/electrode positions,
SRFA allowed for safe and effective treatment of large-
volume liver disease. Long-term clinical results of SRFA
remain to be determined.
Acknowledgments The authors thank the radiation technicians:
Bob Lang, RT, Martin Fasser, RT, Martin Knoflach, RT, Julia
Mahlknecht, RT, and Florian Schanda, RT.
Conflict of interest Gerlig Widmann, Peter Schullian, and Marion
Haidu have no conflict of interest. Reto Bale is a coinventor of the
stereotactic aiming device used in this study (Atlas, Medical Intelli-
gence GesmbH, Schwabmu
¨nchen, Germany) and a coshareholder in
its financial returns.
G. Widmann et al.: SRFA of Liver Lesions
123
References
1. Bale R, Widmann G (2007) Navigated CT-guided interventions.
Minim Invasive Ther Allied Technol 16:196–204
2. Bale R, Widmann G, Stoffner DI (2010) Stereotaxy: breaking the
limits of current radiofrequency ablation techniques. Eur J Radiol
75:32–36
3. Bale R, Widmann G, Haidu M (2010) Stereotactic radiofrequency
ablation. Cardiovasc Intervent Radiol. doi:10.1007/s00270-010-
9966-z
4. Chen MH, Wei Y, Yan K et al (2006) Treatment strategy to
optimize radiofrequency ablation for liver malignancies. J Vasc
Interv Radiol 17:671–683
5. Crocetti L, de Baere T, Lencioni R (2010) Quality improvement
guidelines for radiofrequency ablation of liver tumours. Cardio-
vasc Intervent Radiol 33:11–17
6. Dupuy DE, Goldberg SN (2001) Image-guided radiofrequency
tumor ablation: challenges and opportunities—part II. J Vasc
Interv Radiol 12:1135–1148
7. Gazelle GS, Goldberg SN, Solbiati L, Livraghi T (2000) Tumor
ablation with radio-frequency energy. Radiology 217:633–646
8. Goldberg SN, Gazelle GS, Mueller PR (2000) Thermal ablation
therapy for focal malignancy: a unified approach to underlying
principles, techniques, and diagnostic imaging guidance. AJR
Am J Roentgenol 174:323–331
9. Goldberg SN, Grassi CJ, Cardella JF et al (2009) Image-guided
tumor ablation: standardization of terminology and reporting
criteria. J Vasc Interv Radiol 20(7 Suppl):S377–S390
10. Head HW, Dodd GD III, Dalrymple NC et al (2007) Percutaneous
radiofrequency ablation of hepatic tumors against the diaphragm:
frequency of diaphragmatic injury. Radiology 243:877–884
11. Hori T, Nagata K, Hasuike S et al (2003) Risk factors for the local
recurrence of hepatocellular carcinoma after a single session of
percutaneous radiofrequency ablation. J Gastroenterol 38:977–981
12. Jacob AL, Regazzoni P, Steinbrich W, Messmer P (2000) The
multifunctional therapy room of the future: image guidance,
interdisciplinarity, integration and impact on patient pathways.
Eur Radiol 10:1763–1769
13. Lau WY, Lai EC (2009) The current role of radiofrequency
ablation in the management of hepatocellular carcinoma: a sys-
tematic review. Ann Surg 249:20–25
14. Lee JM, Han JK, Kim HC et al (2007) Switching monopolar
radiofrequency ablation technique using multiple, internally
cooled electrodes and a multichannel generator: ex vivo and in
vivo pilot study. Invest Radiol 42:163–171
15. Lencioni R, Della Pina C, Bartolozzi C (2005) Percutaneous image-
guided radiofrequency ablation in the therapeutic management of
hepatocellular carcinoma. Abdom Imaging 30:401–408
16. Lin SM, Lin CC, Chen WT et al (2007) Radiofrequency ablation
for hepatocellular carcinoma: a prospective comparison of four
radiofrequency devices. J Vasc Interv Radiol 18:1118–1125
17. Livraghi T, Goldberg SN, Lazzaroni S et al (1999) Small hepa-
tocellular carcinoma: treatment with radio-frequency ablation
versus ethanol injection. Radiology 210:655–661
18. Livraghi T, Goldberg SN, Lazzaroni S et al (2000) Hepatocellular
carcinoma: radio-frequency ablation of medium and large lesions.
Radiology 214:761–768
19. Livraghi T, Solbiati L, Meloni MF et al (2003) Treatment of focal
liver tumors with percutaneous radio-frequency ablation: compli-
cations encountered in a multicenter study. Radiology 226:441–451
20. Livraghi T, Solbiati L, Meloni F et al (2003) Percutaneous
radiofrequency ablation of liver metastases in potential candidates
for resection: the ‘‘test-of-time approach’’. Cancer 97:3027–3035
21. Mulier S, Mulier P, Ni Y et al (2002) Complications of radio-
frequency coagulation of liver tumours. Br J Surg 89:1206–1222
22. Mulier S, Ni Y, Jamart J et al (2005) Local recurrence after
hepatic radiofrequency coagulation: multivariate meta-analysis
and review of contributing factors. Ann Surg 242:158–171
23. Nakai M, Sato M, Sahara S et al (2009) Radiofrequency ablation
assisted by real-time virtual sonography and CT for hepatocel-
lular carcinoma undetectable by conventional sonography.
Cardiovasc Intervent Radiol 32:62–69
24. Ng KK, Poon RT, Lo CM et al (2008) Analysis of recurrencepattern
and its influence on survival outcome after radiofrequency ablation
of hepatocellular carcinoma. J Gastrointest Surg 12:183–191
25. Okuwaki Y, Nakazawa T, Shibuya A et al (2008) Intrahepatic
distant recurrence after radiofrequency ablation for a single small
hepatocellular carcinoma: risk factors and patterns. J Gastroen-
terol 43:71–78
26. Rhim H, Goldberg SN, Dodd GD III et al (2001) Essential
techniques for successful radio-frequency thermal ablation of
malignant hepatic tumors. Radiographics 21(Spec No):S17–S35
27. Rhim H, Dodd GD III, Chintapalli KN et al (2004) Radiofre-
quency thermal ablation of abdominal tumors: lessons learned
from complications. Radiographics 24:41–52
28. Rhim H, Lim HK, Kim YS et al (2008) Radiofrequency ablation
of hepatic tumors: lessons learned from 3000 procedures. J Gas-
troenterol Hepatol 23:1492–1500
29. Rhim H, Choi D, Kim YS et al (2010) Ultrasonography-guided
percutaneous radiofrequency ablation of hepatocellular carcino-
mas: a feasibility scoring system for planning sonography. Eur J
Radiol 75:253–258
30. Sakuhara Y, Shimizu T, Abo D et al (2007) Influence of surgical
staples on radiofrequency ablation using multitined expandable
electrodes. Cardiovasc Intervent Radiol 30:1201–1205
31. Seror O, N’Kontchou G, Ibraheem M et al (2008) Large
(C5.0-cm) HCCs: multipolar RF ablation with three internally
cooled bipolar electrodes—initial experience in 26 patients.
Radiology 248:288–296
32. Solbiati L, Livraghi T, Goldberg SN et al (2001) Percutaneous
radio-frequency ablation of hepatic metastases from colorectal
cancer: long-term results in 117 patients. Radiology 221:159–166
33. Stang A, Fischbach R, Teichmann W et al (2009) A systematic
review on the clinical benefit and role of radiofrequency ablation as
treatment of colorectal liver metastases. Eur J Cancer 45:1748–1756
34. Stoffner R, Augscholl C, Widmann G et al (2009) Accuracy and
feasibility of frameless stereotactic and robot-assisted CT-based
puncture in interventional radiology: a comparative phantom
study. Rofo 181:851–858
35. Tacke J, Mahnken A, Roggan A, Gunther RW (2004) Multipolar
radiofrequency ablation: first clinical results. Rofo 176:324–329
36. Wagner A, Schicho K, Birkfellner W et al (2002) Quantitative
analysis of factors affecting intraoperative precision and stability
of optoelectronic and electromagnetic tracking systems. Med
Phys 29:905–912
37. Widmann G, Bodner G, Bale R (2009) Tumour ablation: tech-
nical aspects. Cancer Imaging 9(Spec No A):S63–S67
38. Widmann G, Schullian P, Haidu M et al (2010) Respiratory
motion control for stereotactic and robotic liver interventions. Int
J Med Robot 6:343–349
39. Widmann G, Schullian P, Haidu M et al (2011) Targeting
accuracy of CT-guided stereotaxy for radiofrequency ablation of
liver tumours. Minim Invasive Ther Allied Technol. doi:10.3109/
13645706.2010.533923
40. Wood BJ, Zhang H, Durrani A et al (2005) Navigation with
electromagnetic tracking for interventional radiology procedures:
a feasibility study. J Vasc Interv Radiol 16:493–505
41. Zhang H, Banovac F, Lin R et al (2006) Electromagnetic tracking
for abdominal interventions in computer aided surgery. Comput
Aided Surg 11:127–136
G. Widmann et al.: SRFA of Liver Lesions
123
