Key words Shear wave elastography - Physical exertion - Alcohol consumption - Respiration -
ultrasound, Methods and techniques
Introduction
Liver elastography is established as an important noninvasive tool for the evaluation
of chronic liver diseases, allowing bedside assessment of liver elasticity as an estimation
of fibrosis [1 ]
[2 ] and offering an assessment of disease stage, progression, and prognosis. Liver stiffness
measurement by ultrasound elastography directly reflects a physical property of the
liver, in contrast to serological tests of fibrosis which are influenced by fibrosis
in other organs. However, various exposures (e. g. deep inspiration, recent meal intake)
and underlying conditions (e. g. right heart failure) other than liver fibrosis have
been demonstrated to affect liver stiffness measurements (LSM) [3 ]
[4 ]
[5 ].
In order to enable the comparison of LSM to established cut-off values, between centers,
and in repeated measurements over time in the follow-up of individual patients, current
international guidelines provide recommendations regarding the standardization of
the performance of liver elastography examinations and common quality criteria for
their interpretation. A minimum of 10 min rest prior to examination and a mid-respiratory
breath hold are explicitly recommended [2 ]. Furthermore, guidelines state that LSM decreases following 1–4 weeks of detoxification
in subjects with alcoholic hepatitis, and that there is insufficient data to evaluate
the use of shear wave elastography (SWE) in populations with chronic alcohol overuse
[2 ]. Data concerning recent alcohol exposure prior to LSM in healthy adults without
chronic alcohol overuse or alcoholic liver disease are scarce [2 ]
[6 ], and recommendations in international guidelines are lacking, giving rise to a lack
of certainty among clinicians and patients.
Breath hold has been identified as time-consuming and as an important limiting factor
for clinical implementation of liver elastography, as not all patients are able to
hold their breath [7 ]. Furthermore, the overall cost of cancelled appointments or invalid examinations
due to individuals not fulfilling the demanding and conceivably unnecessary instructions
might be reduced if evidence is provided to simplify preparations prior to LSMs [2 ]. Considering the benefits of time and cost effectiveness, we aimed to investigate
the impact of recent physical exercise, recent alcohol consumption, and calm free
breathing during LSM.
Materials and Methods
Subjects
This prospective cross-sectional study was performed at a single university hospital
between April and June 2019. 42 healthy volunteers (21 males, 50%) were enrolled.
The mean±SD age was 25.7±3.1 years. Structured patient interviews and a standardized
abdominal B-mode ultrasonography examination were performed in all subjects prior
to LSM. The exclusion criteria comprised a previous history of disease in the liver,
heart, or respiratory system, previous or ongoing malignancy, inflammatory bowel disease,
pregnancy, any liver pathology on B-mode ultrasound, or a BMI ≥30 kg/m2 . After excluding 1 subject, 41 (97.6%) subjects were included. Weight and height
were measured to calculate BMI. Some of the subjects participated in parts of the
study ([Fig. 1 ]).
Fig. 1 Flowchart for study participants. The flowchart shows inclusion of participants.
All underwent liver stiffness measurements (LSM), either in all stages including baseline,
physical exercise (PE), alcohol exposure (AE) and different respiratory phases; in
baseline and after PE/AE; or only LSM in different respiratory phases.
Investigation procedure
Liver stiffness measurement (LSM) by point shear wave elastography (pSWE) was performed
using Samsung RS80A with Prestige (Samsung Medison Co, Ltd, Seoul, Korea) with a CA1–7A
convex array probe (1–7 MHz). Measurements were obtained in the right liver lobe with
participants lying in a supine position with the right arm maximally abducted during
a short mid-respiratory breath hold. The transducer was placed perpendicularly in
an intercostal space, while applying minimal pressure. The region of interest (ROI)
had a fixed height of 10 mm and was placed in a homogeneous area 2–5 centimeters beneath
the liver capsule, avoiding visible biliary tracts and vessels. Values were expressed
in kilopascals (kPa). To obtain a valid measurement, 10 acquisitions were required,
with an interquartile range/median ratio (IQR/M) ≤30%. If the IQR/M exceeded 30%,
data were erased, and the examination was repeated once with 10 novel acquisitions
and kept if the IQR/M was ≤30% (n=2).
LSM was performed by three operators, all medical students (MR, VT, ST) who received
training from experienced liver elastography users (ABM, MV) prior to the start of
the study. The training period finished when LSM was adequately performed in a standardized
fashion. Supervision was maintained throughout the study period to ensure correct
technique. Two operators performed baseline examinations as well as LSM after exercise
or alcohol together: one (MR or VT) holding the probe, one controlling the ultrasound
scanner (MR or VT). For LSM evaluating the effect of breath hold versus free breathing,
LSM was performed by a third operator (ST).
Participants were instructed not to consume alcohol 5 days prior to baseline measurements,
and fasted ≥3 h prior to every measurement, including caffeine and chewing gum. Water
consumption was restricted to one glass of 50 ml. Prior to measurements, subjects
rested in a chair for 15 min prior to measurements, except when tested for the effect
of physical exercise. Measurements after exercise and consumption of alcohol were
conducted on two separate days, following the same procedure.
On the first test day, baseline LSM was performed in all subjects. Participants were
then instructed to run up and down five flights of stairs. The pulse was manually
recorded immediately before and after the run. LSM was repeated when the candidate
was able to perform the necessary breath hold. On the second test day, chosen to be
a Monday as alcohol consumption is known to be higher during weekends, participants
registered the number of alcohol units consumed during the last 72 h. One unit of
alcohol was defined according to national guidelines: 12 grams of pure alcohol, corresponding
to a small glass (125 ml) of wine, a shot (40 ml) of strong liquor (35–40%) or 0.33
liters of beer. LSM was conducted as previously described. LSM was then performed
twice more in the same participants: the first LSM during calm free breathing, the
second with mid-expiratory breath hold.
Ethical aspects
The protocol was in accordance with the Declaration of Helsinki and approved by The
Regional Committee on Medical and Health Research Ethics. Written informed consent
was obtained from all subjects following both oral and written information.
Statistics
All statistical analyses were performed using IBM SPSS Statistics 25 Software (SPSS
Inc., 2016 Armonk, NY). The data were assessed using normality tests of Shapiro-Wilk.
Histograms and Q-Q plots were produced to visualize the distribution of the data.
For all normally distributed variables, the mean±standard deviation (SD) was calculated,
and for variables not normally distributed, the median [range] was calculated. For
comparison between groups or repeated measures, parametric (e. g. paired t-tests)
or nonparametric tests (Wilcoxon signed-rank test) were applied when appropriate.
Correlations and unequal distributions were tested by the Pearson correlation and
Bland-Altman plots. Statistical significance was defined as a p-value <0.05.
Results
A total of 41 healthy subjects were included in the final analysis (20 men [51.3%])
following exclusion of one female extreme outlier with consistently high baseline
LSM values at repeated measurements: although thorough investigations including viral,
hematological, and immunological screening, second opinion LSM on a different elastography
platform, and ultrasonography of the abdomen did not reveal any other signs of liver
disease, LSM values were normalized at control several weeks later and an intercurrent
liver affection could not be ruled out. LSM for the entire cohort was 3.8±0.8 kPa.
There was no difference in baseline LSM between females and males (3.8±0.7 vs. 3.7±0.9 kPa,
p=0.76).
Valid LSM values were obtained in all included subjects (n=39) for every different
measurement type: at baseline, following physical activity and alcohol exposure, and
during breath hold and free breathing ([Fig. 2 ]; [Table 1 ]). Baseline characteristics for participants are displayed in [Table 2 ].
Fig. 2 Liver stiffness measurements based on LSM measurement category. Baseline measurements
and after physical exercise (day 1) and after registration of alcohol exposure and
during different respiratory phases (day 2). Significant difference only between baseline
LSM and LSM post-exercise (p=0.01).
Table 1 Liver stiffness measurements for the entire panel (n=41) and across genders, in all
5 stages: baseline LSM; LSM after exercise; LSM after alcohol consumption registration;
LSM during breath hold; LSM during calm respiration.
Total panel
Males
Females
Liver stiffness measurement (LSM), kPa, mean±SD
Day 1: Baseline
3.8±0.8
3.8±0.7
3.7±0.9
Day 1: Post-exercise
4.1±0.8
4.2±0.5
4.0±1.1
Day 2: Post-alcohol
3.7±0.7
3.8±0.7
3.5±0.7
Day 2: Breath hold
4.0±0.8
3.9±0.9
4.1±0.7
Day 2: Calm respiration
3.9±0.8
3.9±0.8
4.0±0.8
Table 2 Background characteristics of all 41 participants. Data presented as mean±SD unless
otherwise specified.
Total panel
Males
Females
Age, years
25.7±3.1
26.3±3.6
25.1±2.4
Body mass index (BMI), kg/m
2
23.8±3.0
25.4±2.2
22.1±2.9
Alcohol consumption last month, units
26.6±23.3 (0–120)
23.7±18.8 (2–80)
29.6±27.4 (0–120)
Potentially harmful monthly alcohol use, n (%)
12 (30%)
7 (35%)
4 (21%)
Alcohol consumption in the last 72 h prior to LSM day 2
8.7±7.0 (0–27)
10.1±7.0 (0–24)
7.2±6.8 (0–27)
Number of participants (average number of units) drinking alcohol <24 h prior to LSM
day 2
14 (4.3)
7 (3.6)
7 (4.9)
Number of participants (average number of units) drinking alcohol 24–47 h prior to
LSM day 2
14 (5.8)
7 (8.4)
7 (3.3)
Number of participants (average number of units) drinking alcohol 48–72 h prior to
LSM day 2
23 (8.6)
12 (9.8)
11 (7.3)
Effects of physical exercise prior to LSM
Compared to baseline, the mean LSM value after physical strain was significantly higher
for the entire population 4.1±0.8 vs. 3.8±0.8 kPa (p=0.01). This effect seemed to
be gender-specific, as the significant increase only concerned males (4.2±0.5 kPa
vs. 3.8±0.7 kPa, p=0.02), and not females 4.0±1.1 vs. 3.7±0.9, p=0.18). Clinically
significant differences (≥1 kPa) were found in 8/39 (20.5%), the maximum difference
was +2.7 kPa. Seven of these were positive, including three females.
Participants (n=9) with an exercise-induced increase less than 40 beats per minute
(bpm) did not show a change in LSM (4.0±0.5 vs. 3.9±0.8 kPa, p>0.7), while LSM increased
in subjects with a pulse increase ≥40 bpm (4.2±0.9 vs. 3.7±0.7 kPa, p=0.01). Pulse
change was equal across genders.
Effect of alcohol 72 h prior to LSM
36 participants consumed alcohol during the registration period, reporting a median
[range] amount of 7.5 [1.5–27] units, while 3 subjects remained abstinent. The historical
median weekly alcohol consumption reported among alcohol consumers, was 5.6 units,
possibly suggesting that the alcohol consumption during the three days prior to test
day 2 was higher than normal, but the difference was not significant (p=0.13). Alcohol
was consumed by 14, 14, and 23 participants on day one, day two, and day three before
LSM, respectively. The corresponding number of subjects consuming ≥5 units of alcohol
on a single day was 4, 8, and 15, respectively.
LSM on test day 2 (following alcohol registration) was not significantly different
from baseline among participants who consumed alcohol during the last 72 h (n=36)
(3.7±0.7 vs. 3.8±0.8 kPa, p=0.58). Dot plot did not indicate any tendency for correlation
of LSM with recent alcohol consumption ([Fig. 3 ]). Clinically significant differences (≥1 kPa) were found in 11/39 (28.2%), with
a maximum difference of −2.9 kPa.
Fig. 3 Scatter plot with alcohol consumption in units plotted against LSM difference from
baseline, with all differences converted to positive values. No tendency of increasing
LSM discrepancy after heavy alcohol intake, with almost no LSM change in individuals
having consumed ≥7 units the last 48 h. The seeming tendency of less LSM difference
with increasing alcohol consumption was not significant (rho=−0.27, p=0.1).
Participants (n=11) reporting potential harmful alcohol use, defined as ≥14 units/week
for men and ≥9 units/week for women, had similar baseline LSM values compared to other
participants (3.8±0.7 vs. 3.7±0.8, p=0.82) and showed no alterations following alcohol
consumption when compared to baseline (3.8±0.5 vs. 3.8±0.7 kPa, p=0.91).
LSM during calm respiration vs. breath hold
There was no difference in mean LSM for the total panel between measurements during
breath hold and calm respiration (4.0±0.8 vs. 3.9±0.8 kPa, p=0.34). There was a tendency
toward higher LSM during breath hold in females, but this was not significant (4.1±0.7
vs. 4.0±0.8, p=0.06). The correlation between LSM in the two breathing states in the
individual subject was good (rho=0.82, p<0.001), with only 1/39 (2.6%) differences
being clinically significant (≥ 1.0 kPa), the maximum difference noted was 1.0 kPa.
Analyzing the effect of breathing pattern on LSM quality criteria, we found an increased
IQR/M in men during calm respiration, as compared to measurements during breath hold
(17.8±5.3% vs. 13.7±5.4%; p=0.03), whereas no such difference was demonstrated in
females or in the total cohort (p=0.68 and 0.07, respectively).
Baseline LSM and LSM change
Baseline LSM was the main determinant factor affecting change in LSM following exposure
to exercise or various breathing patterns. When the baseline LSM was low, LSM often
increased on subsequent measurements as illustrated in [Fig. 4 ], showing a strong correlation (rho 0.723, p<0.001), and [Fig. 5a-b ], suggesting regression towards the mean as an explanatory factor.
Fig. 4 Scatter plot showing the correlation between baseline LSM and change in LSM on day
2 (after registration of alcohol consumption), demonstrating that individuals with
a low baseline LSM often had an increase in LSM on day 2, while the opposite was true
for individuals with a high baseline LSM.
Fig. 5 Boxplot with baseline liver stiffness measurements (LSM) divided into ≤4 and ≥4 kPa,
with change in LSM on the y-axis. a Difference between LSM after alcohol exposure and baseline LSM. b difference between LSM after physical activity and baseline LSM.
Discussion
The introduction of liver elastography has improved the follow-up of patients with
chronic liver disease, offering an easily accessible tool for real-time, noninvasive
diagnosis and monitoring of liver fibrosis development. International guidelines,
aiming to standardize examination procedures and quality criteria in an important
effort to secure reliability and validity of results, are continuously being updated
in response to the technological development in the field. However, for applicability
as well as cost-effectiveness purposes, all recommendations with implications for
practical application in clinical practice should be evidence-based, avoiding precautions
complicating the procedure if not strictly necessary. Thus, it is important to thoroughly
evaluate the clinically significant effect on LSM of the various procedure-related
recommendations to identify the preparations truly required for performing valid liver
elastography.
We found that LSM increased significantly following physical exercise, observing clinically
significant differences (≥1 kPa) in 1 in 5 subjects.
Our study thus supports previous findings, strengthening the hypothesis that physical
strain prior to LSM results in significantly higher LSM values and should be avoided
prior to measurement. To our knowledge, only one other study has previously investigated
the impact of exercise on LSM [8 ], but includes only 7 participants. The study found a significant LSM increase 0
and 5 min after activity, while it returned close to baseline values after 10 min.
Pulse increase during physical strain has been linked to increased blood flow through
the liver [9 ], further suggesting that blood redistribution may increase LSM [8 ]. In line with this, we observed that the increase in exercise-induced LSM was restricted
to subjects with a pulse increase >40 bpm, whereas individuals with a low pulse increase
(≤40 bpm) did not show signs of increasing LSM after activity, suggesting that only strenuous exercise
inducing a high heart rate affects LSM values. Hence, 10 min of rest prior to LSM
may be superfluous in most subjects if strenuous exercise is avoided. The low number
of individuals (n=9) and short observation time following exercise preclude definitive
conclusions, and further studies are warranted to strengthen recommendations regarding
rest prior to LSM. Gender differences found in our study, i. e., significant LSM increase
observed in males and not in females, also warrant further study to allow conclusions.
To the best of our knowledge, this is the first prospective study examining the effects
of recent alcohol consumption on LSM in healthy individuals. We found no significant
effect on LSM following alcohol consumption 24–72 h prior to measurements, even in
individuals with high recent consumption. Whether alcohol intake affects LSM in patients
with chronic liver disease remains unknown and further investigations are needed.
Data on liver elastography and alcohol are limited and mostly retrospective [6 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ]. There is insufficient evidence to evaluate the role of pSWE in alcoholic liver
disease, as there are only three small studies on the use of pSWE for assessing alcoholic
liver fibrosis, with inconsistent cut-off values for distinguishing absent or mild
fibrosis from significant or severe fibrosis and cirrhosis [2 ]
[12 ]
[13 ].
The lack of a generally accepted definition of potentially harmful weekly alcohol
use is challenging in research regarding the effects of alcohol on the liver. Participants
in our study meeting the criteria for potential harmful alcohol intake, applying Swedish
guidelines citing a limit of 14 units per week for men and 9 units per week for women
[15 ], did not show a higher baseline LSM or increasing LSM with recent alcohol exposure.
Our findings may indicate that recent alcohol intake in healthy individuals does not
affect LSM. However, since there were few subjects binge drinking the day prior to
LSM (n=4), we cannot exclude that a very high alcohol intake the night before LSM
could cause transient liver inflammation with a corresponding LSM increase, or affect
LSM through the induction of body water imbalance or dehydration. Our findings do
not automatically transfer to chronic liver patients, and investigations in chronic
liver patients are warranted.
In our study, we found no significant impact of breath hold on LSM values. Only a
few studies have previously assessed the effect of respiration on LSM, and results
are inconsistent regarding direct impact. Furthermore, studies have been inconsistent
concerning methodology, systems used, and study population demography, precluding
definitive conclusions regarding any impact of normal respiration [7 ]
[16 ]
[17 ]
[18 ]
[19 ]. Some studies suggest that for certain elastography methods, deep inspiration and
factors that decrease hepatic venous return (e. g. heart congestion, Valsalva maneuver),
falsely increase LSM values, thereby overestimating cirrhotic changes [5 ]
[20 ]. It has been reported that different disease entities like hepatitis, alcoholic
liver disease, and cirrhosis may respond differently regarding LSM changes when exposed
to respiratory motion [17 ]
[20 ].
Some studies question whether displacement of the liver may yield increased or inaccurate
LSM values, or if the fast speed of the shear waves can somehow compensate for the
motion effect [2 ]
[16 ]
[17 ]
[21 ].
In a study investigating 123 adults with chronic liver disease using TE, free-breathing
values in the expiration phase were higher compared to the inspiration phase, suggesting
that liver decompression during free breathing may have a false-positive effect on
LSM. However, free breathing was not compared to breath hold [17 ]. In contrast, two studies investigating 2D-SWE during calm free expiratory breathing
versus breath hold in children, found no significant difference concerning respiratory
motion and LSM changes [16 ]
[22 ]. This corresponds well with our results using pSWE in healthy adults.
Breath hold is challenging for some patients and is not always applicable in clinical
practice [2 ]. Previous studies adopting a free breathing pattern during examination [7 ]
[23 ] have shown the free breathing approach to be beneficial regarding both applicability
and time effectiveness. This resonates with our experience that measurements taken
in free respiration are more time efficient and less demanding for the subjects, especially
regarding apneal recovery.
We also describe an example of regression towards the mean, a statistical phenomenon
arising when a random variable is extreme on the first observation but closer to the
mean on the second observation ([Fig. 4 ]
[5 ]
a -b ) [24 ]. LSM acquisitions will often have a wide dispersion and vary even in controlled
circumstances. In the case of a cohort of subject with a true LSM of 5.0 kPa, with
LSM values varying between 4.5 and 5.5 kPa, subjects with a first LSM of 4.5 kPa are
likely to have a higher second LSM, while the opposite is true for those with a first
LSM of 5.5 kPa. This must be taken into account when performing similar studies.
Limitations
The study design enabled the subjects to function as their own controls, allowing
us to accept a smaller sample size, although a larger sample size might have been
optimal for subanalyses per gender.
We considered the chance of unknown hepatic disease to be very low, as the study population
consisted largely of young, non-obese, healthy students, with no history of serious
illness and from an area (Norway) with a very low prevalence of viral hepatitis in
the general population. Thus, the lack of relevant blood tests is a relatively minor
limitation. Furthermore, all included participants showed LSM results within the normal
range.
Alcohol intake was not standardized (for ethical reasons) and there is also a possibility
of recollection bias regarding the subjects’ recollection of the number of alcohol
units consumed during the period of 72 h prior to LSM. Furthermore, fluid balance
is a possible confounder after alcohol consumption, and measuring blood pressure and
weight both before and after alcohol exposure would strengthen the study. This is,
however, less likely given our negative findings.
The effect of breath hold was studied on day 2, after alcohol exposure. Although we
made no observations pointing to difficulties during the study on day 2, we cannot
know for certain that this did not have an impact.
Our study population included only healthy young adults and cannot be directly extrapolated
to chronic liver patients. The normal range is wide in our material as well as in
the published literature, and it is likely that day-to-day and intra-individual variations
of LSM affected our measurements.
Conclusion
We found that physical exercise led to increased LSM. However, this was restricted
to individuals with a considerable heart rate increase following exercise, suggesting
that current general recommendations for rest might be modified to advise patients
to avoid strenuous exercise prior to liver elastography. Furthermore, our findings
suggest that point shear wave elastography may be performed during calm respiration
without inducing clinically significant affection of LSM in healthy adult livers.
Further studies in chronic liver patients are warranted to decide whether recommendations
regarding breath hold should be modified. Alcohol exposure in the 24–72 h before examination
did not affect LSM in healthy adults.
Victoria Taraldsen, Sunneva Tomasgard, Margrethe Thune Rudlang, Odd Helge Gilja, Mette
Vesterhus, Anders Batman Mjelle.
Point Shear Wave Elastography and the Effect of Physical Exercise, Alcohol Consumption,
and Respiration in Healthy Adults.
Ultrasound Int Open 2020; 06 (03): E54–E61 DOI: 10.1055/a-1298-9642
In the above article, names of three co-authors were indicated incorrectly.
Notice
This article was changed according to the Erratum on Dec 23, 2020.
Erratum
In the above article, names of three co-authors were indicated incorrectly.
Correct:
