Key words elasticity imaging techniques - ultrasonography - liver transplantation - child -
pressure
Introduction
Ultrasound elastography is used widely to quantitatively assess liver stiffness, and
normal values for adults and children have been published [1 ]
[2 ]
[3 ]. Many centers have included ultrasound elastography in standardized protocols for
patient follow-up after liver transplantation [4 ]
[5 ], as this modality enables the noninvasive determination of the degree of fibrosis
and identification of acute allograft damage such as rejection, infection, and hepatic
outflow obstruction [5 ]
[6 ]
[7 ]
[8 ]
[9 ]
[10 ].
Available techniques for quantitative ultrasound elastography include transient elastography
(TE), point shear wave elastography (pSWE), and two-dimensional shear wave elastography
(2D-SWE) [11 ]
[12 ]. The appropriate method depends on the clinical situation. 2D-SWE is used widely
for the evaluation of liver transplantations, including left-lateral split transplantations
(LTXs), which account for the majority of pediatric transplants [6 ]
[7 ]
[13 ], and has been shown to be more accurate than pSWE for the evaluation of liver fibrosis
in children [14 ].
In contrast to the published recommendations for elastography measurement of whole
livers, the favored intercostal approach cannot be used in pediatric cases of left-lateral
LTX with midline grafting [1 ]. For a sufficient acoustic window in left lateral splits, the transducer needs to
be placed in an epigastric position [13 ]. Yet, unshielded placement of the probe directly on the transplanted liver raises
the concern that pressure loading via transducer compression of the liver could falsely
elevate measurements as suggested by the performance of ex-vivo measurements in piglet
livers and other superficially localized organs including the breast, thyroid, transplanted
kidneys, and skeletal muscle [15 ]
[16 ]
[17 ]
[18 ].
The aim of this study was to evaluate the effect of probe-induced abdominal compression
of left split LTXs in children on 2D-SWE values. We hypothesized that variable degrees
of transducer compression applied during free-hand clinical examinations of children
who had undergone LTX would be a source of unrecognized error.
Materials and Methods
Subjects
This retrospective study was legitimated by the local medical statute (Ethik-Kommission
der Ärztekammer Hamburg, WF-71/16), and the requirement for written informed consent
was waived. All examinations were conducted according to the Declaration of Helsinki.
All pediatric patients who had undergone LTX of segments 2/3 in January and February
2018 at our university hospital and received a standardized ultrasound examination,
including 2D-SWE examination with variable degrees of transducer pressure, were included.
The evaluated examinations were all clinically indicated. Two different modes of probe
placement and respective measurements were performed in the context of clinical introduction
of elastography to the pediatric transplantation scenario in the assessed two-month
period. The results of both settings were acknowledged in the findings and were considered
a means of internal quality control in order to avoid overestimation or false interpretation
of the results. After the two-month initiation period, the protocol was changed to
“low pressure” examinations only based on the initial data. Therefore, more patients
could not be included in this study.
The exclusion criteria were insufficient documentation and incomplete examination.
3 of 14 consecutive patients were excluded, due primarily to incomplete examination
caused by agitation during the examination. The final cohort thus included 11 children
(3 females and 8 males; mean age ± standard deviation [SD], 4.7 ± 4.8 years; [Table 1 ]). The primary diagnoses that led to liver transplantation were biliary atresia (n = 5), ornithine transcarbamylase deficiency (n = 2), hepatopathy of unknown etiology
(n = 2), arginase deficiency (n = 1), and type 2 progressive familial intrahepatic
cholestasis (n = 1).
Table 1
Curved transducer elastography measurements performed with no and low transducer-induced
pressure. Data are presented as means with standard deviations. IQR: interquartile range. Median liver stiffness with no vs. slight transducer pressure, p < 0.001; IQR of measurements with no vs. slight transducer pressure, p = 0.016; IQR/median with no vs. slight transducer pressure, p = 0.164.
Tab. 1 Konvexschallkopf-Elastografiemesswerte mit wenig und nahezu ohne Ultraschallkopfdruck. Die Daten werden als Mittelwerte mit Standardabweichung angegeben. IQR, interquartile range. Mediane Leber-Stiffness mit vs. nahezu ohne Ultraschallkopfdruck, p < 0.001; IQR der Messwerte mit vs. ohne Ultraschallkopfdruck, p = 0.016; IQR/Median mit vs. ohne Ultraschallkopfdruck, p = 0.164.
Median liver stiffness
(kPA)
IQR
(kPA)
IQR/median
Patient age (y)/gender
Transducer
pressure
Transducer
pressure
Transducer
pressure
No
Low
No
Low
No
Low
1/8/m
8.8
16.28
1.85
2.15
0.21
0.13
2/0/m
5.46
12.26
0.93
1.43
0.17
0.12
3/0/m
5.94
10.73
0.6
2.80
0.10
0.26
4/3/m
9.87
17.11
0.77
3.68
0.08
0.22
5/16/f
9.04
13.64
1.20
1.37
0.13
0.10
6/3/f
6.42
12.03
0.60
1.49
0.09
0.12
7/3/f
6.6
15.14
0.81
1.97
0.12
0.13
8/6/m
7.7
18.27
1.21
7.94
0.16
0.43
9/0/m
4.52
8.45
0.67
0.71
0.15
0.08
10/8/m
7.39
11.82
1.13
1.16
0.15
0.10
11/5/m
5.47
11.42
0.64
2.00
0.12
0.18
All patients
7.02 ± 1.7
13.38 ± 3.00
0.95 ± 0.38
2.43 ± 2.0
0.13 ± 0.04
0.17 ± 0.1
Of the 11 patients, 7 had no fibrosis proven by biopsy (METAVIR F0) or low-grade fibrosis
(METAVIR F1). Of the remaining four patients without available biopsy, three patients
had elastography measurements during their early post-operative course after LTX (day
3–31 after LTX). For this subset of patients, absence of fibrosis can be safely assumed
soon after LTX also without histological confirmation.
Ultrasound Examination
A single experienced pediatric radiologist (J. H.) performed the ultrasound examinations
using a predefined protocol with prospective documentation and a commercial scanner
(GE Logiq E9 ultrasound system; GE Medical Systems, Milwaukee, WI, USA) with a C1–5 MHz
curved array transducer and an L9 MHz linear array transducer. The children were examined
in a supine position under free breathing after ≥ 2-hour fast. An epigastric abdominal
wall approach and two free-hand adjusted probe pressure modes were used. Measurements
were performed in the same location with both arrays using the lowest possible probe
pressure (“no pressure settings,” d
0 ), and then using slight abdominal wall compression (“low pressure settings,” d ) to achieve a closer connection. In the “no pressure setting”, the probe was placed
on the abdominal wall above the liver transplant with the lowest possible pressure
to just achieve acoustic coupling with the underlying tissue. With the curved array
transducer only acoustic coupling in the center of the window needed to be achieved
and coupling at the outer portions of the transducer was omitted as full coupling
of the more distant sides of the convex surface would require higher probe pressure.
With the low-pressure setting, full acoustic coupling needed to be achieved with slight
compression of the underlying tissues. Stronger compression of the underlying liver
transplant was not performed. A shear wave color map was positioned in the liver parenchyma
≥ 1–2 cm below the liver capsule. Vessels and focal lesions were omitted. The examinations
were stopped when 12 successful elastograms had been recorded.
To semiquantitatively assess the degree of probe-induced compression and its effect
on the underlying liver transplants, the methods of Barr et al. and Vachutka et al.
were adapted [16 ]
[17 ]. The distance between the cutis and the posterior margin of the liver transplant
was measured at the level of the elastogram, and mean values for each condition and
probe type were calculated. The degree of compression was calculated using the formula
(δ = 1 – d /d0
) ([Fig. 1 ]). Two experienced pediatric radiologists (J. H. and M. G.) then performed consensus
reading of the measurements on a PACS workstation (Ventricity Universal Viewer GE
Healthcare, Milwaukee, WI, USA). Anomalies and artifacts on the B-mode images and
embedded elastograms were noted when present.
Fig. 1 2D-SWE of a lateral LTX (patient 4) measured under low (a ) and slight (b ) probe pressure. Indicated are the distances between the cutis and the posterior
margin of the liver, obtained with identical probe positioning. The degree of compression
was calculated using the formula δ = 1 – d/d0; for patient number 4: 0.07 = 1–4.30 cm/4.64 cm.
Abb. 1 2D-SWE eines lateralen LTX (Patient 4) nahezu ohne (a ) und mit wenig (b ) Ultraschallsondendruck. Eingezeichnet sind die Abstände zwischen der Kutis und dem
posterioren Leberrand mit identischer Sondenposition. Der Kompressionsgrad wurde mit
der Formel δ = 1 – d/d0 berechnet. Für Patient 4: 0,07 = 1–4,30 cm/4,64 cm.
Statistical Analysis
Elastography measurements obtained under the d
0 and d conditions were compared using Bland-Altman analysis and Student’s t test. All liver stiffness values (median liver stiffness, interquartile ranges (IQR),
and IQR/median ratio) and the distances between the cutis and lower transplant margin
measured under the d
0 and d conditions were compared using Studentʼs t test. Statistical analysis was performed using MedCalc Version 19.4.1 for Windows
(MedCalc Software Ltd, Ostend Belgium) and Excel (Microsoft Corporation, Redmond,
WA, USA).
Results
In total, 528 single hepatic 2D-SWE measurements on 11 children who had undergone
left-lateral LTX were performed and evaluated. Examination quality was good under
both conditions, with no anomalies or artifacts noted. All patients tolerated the
examination well.
Under the d
0 and d conditions, the mean distances between the cutis and lower LTX margin were 5.9 ± 1.3
and 5.0 ± 1.1 cm (p < 0.0001), respectively, for the curved array, and 5.3 ± 1.0 and 4.7 ± 0.9 cm (p < 0.0001), respectively, for the linear array. The degrees of abdominal compression
under d were 15 % ± 8 % with the curved transducer and 12 % ± 8 % with the linear transducer.
For examinations performed with the curved and linear transducers, the median liver
stiffness was significantly greater (by 6.6 and 9.8 kPa, respectively) under d than under d
0 (13.38 ± 3.0 vs. 7.02 ± 1.7 kPa and 18.53 ± 7.1 vs. 9.03 ± 1.5 kPa; p < 0.0001 and p = 0.003, respectively; [Fig. 2 ], [3 ]
[Table 1 ], [2 ]). The measurement variability was greater under d than under d
0 for examinations performed with both transducers ([Table 1 ], [2 ]). No significant difference in the IQR/median ratio was observed.
Fig. 2 2D-SWE of a 3-year-old boy (patient 4) who underwent LTX of liver segment 2/3, with
the probe placed on the abdominal wall directly above the transplant using curved
(a, b ) and linear (c, d ) transducers with low (a, c ) and slight (b, d ) probe pressure.
Abb. 2 2D-SWE eines 3-jährigen Jungen (Patient 4) mit Z. n. Lebersegmenttransplantation
der Segmente 2/3. Die Ultraschallsonde wurde unter Verwendung des Konvexschallkopfes
(a, b ) und des Linearschallkopfes (c, d ) direkt über dem Lebertransplantat auf der Bauchdecke positioniert nahezu ohne (a, c ) und mit wenig (b, d ) Ultraschallsondendruck.
Fig. 3 Bland-Altman plots of median liver stiffness (kPa) obtained with slight pressure
(x axis) and the least possible pressure with curved (a ) and linear (b ) transducers.
Abb. 3 Bland-Altman-Graph der medianen Leber-Steifigkeit (kPa) mit wenig (X-Achse) und dem
geringst möglichen Druck für den Konvex- (a ) und Linearschallkopf (b ).
Table 2
Linear transducer elastography measurements performed with no and low transducer-induced
pressure. Data are presented as means with standard deviations. IQR: interquartile range. Median liver stiffness with low vs. slight probe pressure, p < 0.001; IQR of measurements with low vs. slight transducer pressure, p = 0.013; IQR/median with no vs. slight transducer pressure, p = 0.374.
Tab. 2 Linearschallkopf-Elastografiemesswerte mit wenig und nahezu ohne Ultraschallkopfdruck. Die Daten werden als Mittelwerte mit Standardabweichung angegeben. IQR, interquartile range. Mediane Leber-Stiffness mit vs. nahezu ohne Ultraschallkopfdruck, p < 0.001; IQR der Messwerte mit vs. nahezu ohne Ultraschallkopfdruck, p = 0.013; IQR/Median mit vs. ohne Ultraschallkopfdruck, p = 0.374.
Median liver stiffness
(kPA)
IQR
(kPA)
IQR/median
Patient no./ age (y)/sex
Transducer
pressure
Transducer
pressure
Transducer
pressure
No
Low
No
Low
No
Low
1/8/m
10.51
25.92
1.02
5.02
0.10
0.19
2/0/m
9.44
15.24
0.91
1.19
0.10
0.08
3/0/m
8.59
16.00
0.51
1.51
0.06
0.09
4/3/m
9.83
19.31
1.28
1.05
0.13
0.05
5/16/f
6.58
15.93
0.94
1.37
0.14
0.09
6/3/f
7.95
11.78
1.38
2.30
0.17
0.20
7/3/f
10.14
17.24
1.63
2.73
0.16
0.16
8/6/m
11.86
22.47
2.92
7.83
0.25
0.35
9/0/m
7.93
10.57
1.07
0.53
0.13
0.05
10/8/m
7.28
35.24
0.95
6.67
0.13
0.19
11/5/m
9.18
14.14
1.86
2.78
0.20
0.20
All patients
9.03 ± 1.55
18.35 ± 7.1
1.32 ± 0.65
3.00 ± 2.4
0.14 ± 0.05
0.15 ± 0.09
Discussion
This study evaluated the effect of probe-induced abdominal compression on 2D-SWE values
obtained for children who had undergone left-lateral LTX. With epigastric probe positioning,
even slight compression of the abdominal wall and underlying liver transplant resulted
in the acquisition of significantly higher stiffness values compared to the use of
the lowest possible pressure. This finding is in accordance with one experimental
ex-vivo study examining the use of 2D-SWE in piglet livers, in which slight tissue
compression altered the quantitative elastography results [16 ]. Similar observations were reported for 2D-SWE in skeletal muscle, thyroid, breast,
and transplanted kidneys [15 ]
[16 ]
[17 ]
[18 ].
With ultrasound elastography, the estimation of liver stiffness is based on the measurement
of shear wave speed; commercial systems apply simplified equations assuming that tissues
are homogeneous, linear, isotropic, and incompressible (Palmeri et al.) [19 ]. However, in biological tissues and clinical situations, these assumptions may be
violated to various degrees [20 ]. Probe-induced pressure on the abdominal wall can induce tissue compression, as
indicated by the reduced distance between the cutis and the posterior margin of the
liver transplant acquired under the d condition in this study. SWE measurement with slightly more pressure led to mean
increases in elastography results with the curved and linear transducers, respectively.
Depending on the clinical context, these increments could lead to misinterpretation
of the results. Six of our patients were assessed in the early post-operative period
after transplantation (< 1 month after LTX), when elevation of liver stiffness can
be a sign of acute rejection, inflammation, or hepatic outflow obstruction [5 ]
[6 ]
[7 ]
[8 ]
[9 ]
[10 ]. Four children prone to chronic changes like post-transplant fibrosis or chronic
rejection associated with increased elastography values were evaluated during the
later follow-up phase [5 ]
[6 ]
[7 ]
[8 ]
[9 ]
[10 ]. Quantitative longitudinal measurements of liver stiffness are seen as especially
useful for monitoring disease activity in a variety of disorders including liver transplantation.
However, as the change of liver stiffness over time (delta) can also be a result of
measurement variability, a standardized setup for the follow-up of individual patients
is important [21 ].
In adult patients with whole livers, the problem of tissue deformation as a result
of transducer compression seems to be less relevant. When performing shielded elastography
measurements via the right intercostal space with the right arm elevated as recommended
by EFSUMB guidelines [1 ], even a high transducer pressure can be exerted to reduce the skin-to-capsule distance
in obese patients without changing the tests’ diagnostic performance [22 ]. Maximum pre-compression in the cited study led to slightly lower SWE values than
normal probe pressure settings (5.02 vs. 5.49 kPa) with the benefit of an improved
technical success rate (87.6 % vs. 98 %) [22 ]. The small opposite effect can be explained by the lower attenuation of the ultrasound
waves which can better traverse compressed than non-compressed subcutaneous fatty
tissues. Interestingly, another experimental study performed in healthy adult volunteers
also found that increased transducer force applied on the epigastric abdominal wall
slightly reduced ARFI (acoustic radiation force imaging) values measured in the left
lobe [23 ].
The anatomical conditions, however, found in adult patients with whole livers are
not fully applicable to the pediatric population. Children have substantially narrower
intercostal spaces than adults, and higher technical failure rates have been reported
for TE using an intercostal approach (up to 17 % for patients aged < 2 years and up
to 10 % for all other ages) [14 ]
[24 ]. Children with segment 2/3 transplants can generally only be examined in the epigastric
position with the graft located in the midline position in the limited space of the
abdominal cavity above the spine [13 ]. Contrary to the situation in adults, the abdominal wall in young children is thin-layered
and contains only small amounts of subcutaneous fat and muscle. This explains why
even the low transducer pressure applied in our study was relatively directly passed
on to the graft, thereby inducing a substantial SWE value increase due to tissue deformation.
Since a variable degree of transducer compression in free-hand examinations can be
a significant source of error, the applied transducer force needs to be controlled
in comparative as well as in longitudinal studies evaluating disease progression.
A semi-quantitative method to indirectly capture the degree of transducer compression
during follow-up examinations is to measure the distance from the skin to a lower
anatomical landmark beneath the probe [19 ]. This method is also feasible in children with split liver transplants. For the
breast it has been postulated that a transducer pre-compression < 1 % of the respective
organ depth will give the most reproducible imaging results [25 ]. A possible future alternative to free-hand SWE examinations is to technically control
the probe pressure with the help of a robotic device. A system has been developed
and successfully tested in phantoms and human volunteers under breath-hold conditions
[26 ].
This study has several limitations. First, it was retrospective and involved the comparison
of two transducer pressure modes adjusted subjectively by a single sonographer. Due
to this clinical situation, inter-rater variability could not be assessed. Second,
we did not precisely measure the probe pressure applied to the abdominal wall. To
our knowledge, no scale to control for probe pressure in quantitative SWE performed
with a commercial system is available. Third, which measurement better reproduces
actual liver stiffness remains unclear. We consider the floating probe position with
the lowest possible probe pressure to be representative, as the ultrasound system’s
internal quality control validated the measurements obtained under this condition.
In addition, the elastography values were within the range of normal stiffness values
reported for liver transplants [13 ]
[27 ]. Histological confirmation of the absence of clinically significant fibrosis (METAVIR
≤ F1) was available for seven of our patients and can be postulated for three further
cases without available biopsy who were evaluated within one month after the transplantation.
In conclusion, SWE measurement of split liver transplants performed via the abdominal
wall can be altered by even slight transducer pressure, indicating the need for careful
pressure control. Differences in free-hand compression among examiners remains a concern
for quantitative SWE, as a proper B-mode image can almost never be obtained without
exerting some pressure on the underlying transplant [19 ]. A technical solution, such as commercial systems’ measurement and registry of the
probe pressure applied during quantitative SWE, would enhance the accuracy of the
examination of split liver transplants and other indications involving direct probe
impact.