Impact of controlled attenuation parameter on detectingfibrosis using liver stiffness measurement

T. Karlas1 | D. Petroff1 | M. Sasso2 | J.-G. Fan3 | Y.-Q. Mi4 | V. de L�edinghen5 |

M. Kumar6 | M. Lupsor-Platon7 | K.-H. Han8 | A. C. Cardoso9,10 | G. Ferraioli11 |

W.-K. Chan12 | V.W.-S. Wong13 | R. P. Myers14 | K. Chayama15 | M. Friedrich-Rust16 |

M. Beaugrand17 | F. Shen3 | J.-B. Hiriart5 | S. K. Sarin6 | R. Badea7 | H.W. Lee8 |

P. Marcellin10 | C. Filice11 | S. Mahadeva12 | G. L.-H. Wong13 | P. Crotty14 | K. Masaki15 |

J. Bojunga16 | P. Bedossa2,17 | V. Keim1 | J. Wiegand1

1Leipzig, Germany

2Paris, France

3Shanghai, China

4Tianjin, China

5Pessac, France

6New Delhi, India

7Cluj-Napoca, Romania

8Seoul, Korea

9Rio de Janeiro, Brazil

10Clichy, France

11Pavia, Italy

12Kuala Lumpur, Malaysia

13Hong Kong, China

14Calgary, Canada

15Hiroshima, Japan

16Frankfurt, Germany

17Bondy, France

Correspondence

Dr. J Wiegand, Division of Gastroenterology

and Rheumatology, University Hospital

Leipzig, Leipzig, Germany.

Email: johannes.wiegand@medizin.uni-leipzig.

de

Funding information

German Federal Ministry of Education and

Research, Grant/Award Number: FKZ

01EO1001, unrestricted grant from

Echosens

Summary

Background: Liver fibrosis is often accompanied by steatosis, particularly in patients with

non-alcoholic fatty liver disease (NAFLD), and its non-invasive characterisation is of

utmost importance. Vibration-controlled transient elastography is the non-invasive

method of choice; however, recent research suggests that steatosis may influence its

diagnostic performance. Controlled Attenuation Parameter (CAP) added to transient elas-

tography enables simultaneous assessment of steatosis and fibrosis.

Aim: To determine how to use CAP in interpreting liver stiffness measurements.

Methods: This is a secondary analysis of data from an individual patient data meta-

analysis on CAP. The main exclusion criteria for the current analysis were unknown

aetiology, unreliable elastography measurement and data already used for the same

research question. Aetiology-specific liver stiffness measurement cut-offs were

determined and used to estimate positive and negative predictive values (PPV/NPV)

with logistic regression as functions of CAP.

Results: Two thousand and fifty eight patients fulfilled the inclusion criteria (37% women,

18% NAFLD/NASH, 42% HBV, 40% HCV, 51% significant fibrosis ≥ F2). Youden optimised

cut-offs were only sufficient for ruling out cirrhosis (NPV of 98%). With sensitivity and speci-

ficity-optimised cut-offs, NPV for ruling out significant fibrosis was moderate (70%) and could

be improved slightly through consideration of CAP. PPV for significant fibrosis and cirrhosis

were 68% and 55% respectively, despite specificity-optimised cut-offs for cirrhosis.

Conclusions: Liver stiffness measurement values below aetiology-specific cut-offs are

very useful for ruling out cirrhosis, and to a lesser extent for ruling out significant fibro-

sis. In the case of the latter, Controlled Attenuation Parameter can improve interpreta-

tion slightly. Even if cut-offs are very high, liver stiffness measurements are not very

reliable for ruling in fibrosis or cirrhosis.

The Handling Editor for this article was Professor Stephen Harrison, and it was accepted for publication after full peer-review.

T. Karlas and D. Petroff contributed equally to this work.

The names and full addresses of the International CAP-IPDMA group are listed in Appendix 1.

Received: 16 October 2017 | First decision: 19 November 2017 | Accepted: 1 January 2018

DOI: 10.1111/apt.14529

Aliment Pharmacol Ther. 2018;47:989–1000. wileyonlinelibrary.com/journal/apt © 2018 John Wiley & Sons Ltd | 989

1 | INTRODUCTION

Liver stiffness measurement (LSM) has quickly grown to become an

established first-line tool for assessment of liver fibrosis, especially

vibration-controlled transient elastography (VCTE).1,2 It is considered

quite accurate for ruling out advanced stages of fibrosis or cirrhosis,

but less so for ruling them in.3 The diagnostic performance is known

to be affected by a number of factors, and protocols have been

devised for minimising their influence, eg, measurements should be

made after fasting4 and B-mode ultrasound should be performed to

exclude patients with biliary obstruction or congestive hepatopathy.3

In addition, aetiology and inflammatory activity must be taken into

account when interpreting liver stiffness measurements.

Fatty liver is commonly related to prevalent aetiologies and may

drive disease progression.5,6 This is particularly relevant for non-alco-

holic fatty liver disease or steatohepatitis (NAFLD/NASH), with dra-

matically increasing numbers as a result of the obesity pandemic.7,8

Current research suggests that fatty liver can lessen LSM accuracy9

and modify cut-offs.10 Non-invasive surrogates of steatosis staging

such as bright echo pattern in conventional ultrasound might help in

interpreting results.11 With the advent of Controlled Attenuation

Parameter (CAP) – an ultrasound-based quantitative estimate of hep-

atic fat incorporated in VCTE – consideration of steatosis in real-

time is facilitated.12 A CAP-based algorithm was explored by Petta

et al for use in patients with non-alcoholic fatty liver disease

(NAFLD).13 There is some debate, however, regarding various

sources of bias,14 patient selection and methodological issues.15

To clarify these issues, and explore aetiology-specific aspects of

the putative connection between LSM and CAP, we performed a

secondary analysis of individual patient data provided by authors of

19 papers that had originally been collected for determining opti-

mised CAP cut-offs.16

2 | PATIENTS AND METHODS

2.1 | Paper and patient selection

This study is a secondary analysis using the database established for

conducting an individual patient data meta-analysis (IPDMA) of Con-

trolled Attenuation Parameter accuracy. The analysis was approved

by the local ethics committee (331/16-ek) and registered prior to

starting the analysis (PROSPERO 2016: CRD42016046271). Studies

published up to May 2015 were included if CAP measurements were

performed with the Fibroscan M-probe and liver biopsies taken.

Measurements with the Fibroscan XL-probe were not considered.

Roughly speaking, patients were excluded if the body mass index

(BMI) was > 35 kg/m² and if the time interval between biopsy and

CAP measurement was > 30 days.16 In the current analysis, patients

were further excluded if they had been used for publications on a

similar research question, if fibrosis staging was unavailable, if the

liver stiffness measurements were unreliable according to established

criteria ( < 10 valid measurements, ratio of median to interquartile

range ≥ 0.3),1,3 if aminotransferases were equal to or above 5 times

the upper limit of normal, or the aetiology was unknown.

2.2 | Steatosis grading and fibrosis staging

As already described,16 steatosis was defined according to the num-

ber of affected hepatocytes: S0 ( < 5% or 10% depending on the

trial), S1 (5% or 10%-33%), S2 (34%-66%) and S3 ( > 66%). Fibrosis

was staged from F0 to F4 according to the Metavir or Kleiner score,

as appropriate.17,18 If the original publication used fibrosis staging

according to Ishak,19 then the following transformation was used:

Ishak F0 = Metavir F0; Ishak F1 = Metavir F1; Ishak F2/3 = Metavir

F2; Ishak F4 = Metavir F3; Ishak F5/6 = Metavir F4.

2.3 | Objectives of the study

The primary objective is to analyse the influence of steatosis either

defined by histology or Controlled Attenuation Parameter on the per-

formance of liver stiffness measurements for distinguishing significant

liver fibrosis ( ≥ F2) from mild or non-existent stages. Secondary

objectives are determination of disease-specific cut-offs for detection

of significant fibrosis, advanced fibrosis ( ≥ F3) and cirrhosis (F4) and

analysis of ruling out and ruling in fibrosis/cirrhosis. The initial goal of

analysing inflammatory activity could not be performed due to lack of

harmonised histological data, but high aminotransferases, indicating

inflammatory activity, were part of the exclusion criteria.

2.4 | Statistical methods

All analyses were performed with the software R version 3.3.3.20

Receiver operating characteristic (ROC) analyses made use of the

pROC package21 and optimal cut-offs were determined by maximis-

ing the Youden index or applying the minimum criterion of 90% for

sensitivity or specificity. Since very wide ranges of cut-offs have

been published,3 we chose optimisations resulting in particularly low

ones when optimising negative predictive values (NPV) and high

ones for positive predictive values (PPV). LSM was always treated

on a logarithmic scale and P-values from linear models comparing

multiple groups were corrected using the Bonferroni-Holm method.

A linear mixed-model was used with liver stiffness measurement

results as the dependent variable and fibrosis staging, aetiology,

steatosis grading, CAP, age, sex, BMI and inflammation status as

fixed effects with the trial (ie, “centre effects”) as a random term and

where P-values for the fixed effects are taken from a Wald chi-

square test.

Logistic regression was used to provide estimates and confidence

intervals for predictive values and for correct classification of fibrosis

stages. In particular, the histologically determined fibrosis category

was taken as the dependent variable with the aetiology-specific

LSM-based classification and the CAP value as independent vari-

ables, including a quadratic term for CAP when necessary. The

regression model yields estimates that can be used to determine a

990 | KARLAS ET AL.

probability given a positive/negative LSM result and a particular CAP

value, which can be interpreted as a PPV/NPV. Prevalence was

taken into account using Bayes’ theorem and the observed values of

sensitivity and specificity. Smoothing splines were used as a sensitiv-

ity analysis to the curves provided by logistic regression.

3 | RESULTS

Of the initial 2735 patients included in the primary analysis of the

CAP-IPDMA, 2058 remained for the current analysis after applying

additional exclusion criteria, Figure 1. Note that 140 patients used

by Petta et al13 to address similar questions to those of the pre-

sent work were excluded from the analysis. Patient characteristics

are presented in Table 1. Of note is the high prevalence of patients

with significant fibrosis or cirrhosis (51%). Patient characteristics

according to geographic location of the study centre can be found

in Table S2.

3.1 | Aetiology-specific liver stiffness measurementcut-offs optimising the Youden index

Significant fibrosis (F2-4) was distinguished from absent or mild

fibrosis (F0-1) at the optimal cut-offs of 8.4 kPa (AUC 0.80, SE 0.68,

SP 0.80) for NAFLD/NASH, 8.2 kPa (AUC 0.77, SE 0.59, SP 0.83)

for HBV and 7.2 kPa (AUC 0.83, SE 0.70, SP 0.82) for HCV.

Cirrhosis (F4) was distinguished from the remainder (F0-3) at the

optimal cut-offs of 13.2 kPa (AUC 0.87, SE 0.84, SP 0.88) for

NAFLD/NASH, 10.8 kPa (AUC 0.86, SE 0.77, SP 0.80) for HBV and

10.2 kPa (AUC 0.94, SE 0.89, SP 0.87) for HCV.

See Table S1 for further details including optimisation with other

criteria and the incorporation of ALT and AST and the cut-offs for

F0-2 vs F3-4, which can be relevant for prognosis.

3.2 | Dependence of liver stiffness measurementson biopsy-based steatosis

As a first step, the dependence of LSM values on biopsy-based

steatosis grading was analysed. Small, but significant elevations in

LSM were seen for higher steatosis stages. For absent or mild fibro-

sis, LSM values in patients with steatosis stages S2 or S3 were a

Unique for this analysis (n = 2595)

Fibrosis staging unavailable (n = 48)

Complete histology (n = 2547)

Aminotransferases ≥5 ULN (n = 31)

Reliable TE (n = 2333)

Analysed (n = 2058)

Unknown etiology (n = 144)

Unreliable LSM measurement (n = 214)

Aminotransferases <5 ULN (n = 2202)

CAP-IPDMA data base (n = 2735)♦ valid CAP measurement with M-probe♦ histologically determined steatosis grading available♦ ≤30 days between TE and biopsy♦ BMI ≤35 kg/m2 or skin-to-liver capsule distance ≤25 mm

Used for similar research question (n = 140)

F IGURE 1 Flowchart of the patients included in the finalanalysis. CAP, Controlled Attenuation Parameter; IPDMA, individualpatient data meta-analysis; LSM, liver stiffness measurement; TE,transient elastography; ULN, upper limit of normal

TABLE 1 Patient characteristics

NAFLD/NASH(n = 372)

HBV(n = 871)

HCV(n = 815)

Female 127 (34%) 295 (34%) 337 (41%)

Age (y) 47 � 13 39 � 12 48 � 12

BMI (kg/m2) 27 � 4 24 � 4 25 � 4

Fibrosis (biopsy)

F0 85 (23%) 104 (12%) 52 (6%)

F1 144 (39%) 308 (35%) 318 (39%)

F2 67 (18%) 294 (34%) 228 (28%)

F3 51 (14%) 81 (9%) 94 (12%)

F4 25 (7%) 84 (10%) 123 (15%)

Steatosis (biopsy)

S0 33 (9%) 515 (59%) 495 (61%)

S1 128 (34%) 257 (30%) 205 (25%)

S2 155 (42%) 74 (8%) 87 (11%)

S3 56 (15%) 25 (3%) 28 (3%)

VCTE (kPa) 7.2 [5.4, 10.4] 7.0 [5.3, 11.0] 8.8 [5.1, 10.2]

cut-offa

(F0-1 vs F2-4)

8.4 8.2 7.2

cut-offa

(F0-3 vs F4)

13.2 10.8 10.2

CAP (dB/m) 307 [262, 338] 229 [196, 262] 222 [193, 260]

Numbers are counts (percentages), mean � standard deviations or med-

ian [interquartile range].

BMI, body mass index; HBV, hepatitis B; HBC, hepatitis C; CAP, con-

trolled attenuation parameter; NAFLD/NASH, non-alcoholic fatty liver

disease/steatohepatitis; VCTE, vibration-controlled transient elastogra-

phy.aCut-offs found by optimising the Youden index.

KARLAS ET AL. | 991

factor 1.14 (95% CI 1.08 to 1.20, P < 0.001) higher than those with

steatosis stages S0 or S1. The corresponding difference in patients

with more advanced fibrosis stages was 1.15 (1.04 to 1.27,

P = 0.0050), Figure 2A. When considering NAFLD/NASH alone, a

similar result was observed for absent or mild fibrosis, but not for

more advanced stages, Figure 2B. These results also hold upon taking

into account the slightly different relative proportions of individual

fibrosis stages between the S0-S1 and the S2-S3 groups. Linear

mixed models, however, do not indicate that steatosis has a signifi-

cant influence on LSM after taking covariates other than CAP into

account (P = 0.056), though the effect becomes significant when

comparing S0-S1 with S2-S3 (P = 0.023). Upon including CAP, the

effect of steatosis vanishes (P = 0.71) and, although CAP is significant

(P = 0.0063), the incremental increase in LSM values is only by a fac-

tor 1.0067 (1.0019 to 1.0116) for an increase in CAP by 10 dB/m.

Similarly small elevations in LSM values were seen in HBV and

HCV patients with steatosis grades S2 and S3 compared to S0 and

S1, though only significantly so for F2-4 HCV patients.

This qualitative behaviour is also seen when detecting advanced

fibrosis or cirrhosis. Significant differences may, sometimes, be seen

between steatosis grades, but they are always very small (Figure S1

and Figure S2).

3.3 | Effect of CAP on correct LSM-classification offibrosis/cirrhosis

Clearly, interpretation of a non-invasive determination of liver fibro-

sis should not rely on a biopsy-based characterisation of steatosis.

Therefore, as a second step, we analysed whether CAP can be used

to increase the diagnostic performance of LSM. For the entire popu-

lation, LSM correctly classified 1509 (73%) of the patients as F0-1

vs F2-4 using aetiology-specific, Youden index optimised cut-offs,

Figure 3. The overall result was very similar for NAFLD/NASH

patients, where 279 (75%) were correctly classified. The probability

of correct classification is independent of the CAP value (P = 0.45

for all patients, P = 0.66 for NAFLD/NASH patients). For example,

12

510

2050

100

Live

r st

iffne

ss (

kPa)

12

510

2050

100

Live

r st

iffne

ss (

kPa)

F0−F1 F2−F4

S0−S1S2−S3 punivariate < 0.001

(A)

(B)

pmultivariate =0.0098

punivariate = 0.005

pmultivariate = 0.18

n = 798 213 835 212

All patients

F0−F1 F2−F4

S0−S1S2−S3

punivariate = 0.028

pmultivariate = 0.076

punivariate = 0.26

pmultivariate = 0.18

n = 91 138 70 73

NAFLD/NASH patients

F IGURE 2 Impact of histologicallydetermined steatosis grade on liverstiffness according to fibrosis stages. Theunivariate analysis is based on a t test andthe multivariate analysis on a linear mixedmodel with aetiology, age, sex and bodymass index as covariates and the study asa random effect. A Bonferroni-Holmcorrection was used. NAFLD/NASH, non-alcoholic fatty liver disease/steatohepatitis

992 | KARLAS ET AL.

the probability of correct classification in all patients is 73% at a

CAP value of 200 dB/m compared to 75% at 350 dB/m.

The probability of correct classification of cirrhosis is 84% and

does not depend on CAP values (P = 0.81), Figure S3. Since only 25

NAFLD/NASH patients have cirrhosis, we refrained from performing

the analogous analysis. In contrast, there was a significant impact of

CAP on classification of F0-2 vs F3-4 (P = 0.0072), though the

impact is small: at CAP values of 200 dB/m, 75% are correctly classi-

fied compared to 82% at 350 dB/m.

3.4 | Effect of CAP on ruling out fibrosis/cirrhosisusing LSM

Ruling out fibrosis or cirrhosis is of particular clinical relevance,

implying that the negative predictive value must be high.

For ruling out fibrosis in our cohort, NPV was found to be 69%

(67% to 72%) at the observed prevalence of 50%, but would reach

88% and 95% at a prevalence of 25% and 10%, respectively. A fairly

strong and nonlinear dependence on CAP was seen (P < 0.001), where

NPV increased from 66% at 200 dB/m to 81% at 350 dB/m and

where a noticeable increase in NPV begins at about 250 dB/m, Fig-

ure 4A. For NAFLD/NASH alone, 211 patients can be analysed, who

have CAP values compatible with presence of steatosis ( > 200 dB/

m). NPV was 79% for 200 dB/m and 85% at 350 dB/m, but the CAP

dependence was not significant (P = 0.33), Figure S4A. Fibrosis can be

ruled out more effectively by choosing cut-offs that require high sensi-

tivity (Table S1), but this results in values that exclude most patients,

eg, in our data only 342 patients (17%) are below this cut-off. Using

ALT and AST optimised cut-offs, 781 (38%) of the patients are below

the cut-off and, at our prevalence, we find an NPV of 70%.

To rule out cirrhosis in our cohort, NPV is 98% (97% to 98%).

There is a nonsignificant dependence on CAP (P = 0.21), which

remains high for all CAP values, eg, it is 97.8% at 200 dB/m and 98.2%

at 350 dB/m, Figure 4B. Even for those above the optimised cut-off, a

large proportion of patients may not have cirrhosis. Introducing speci-

ficity-optimised cut-offs results in a “grey zone” for LSM values in

100 150 200 250 300 350 4000.

00.

20.

40.

60.

81.

00.

00.

20.

40.

60.

81.

0

Controlled attenuation parameter (dB/m)

100 150 200 250 300 350 400Controlled attenuation parameter (dB/m)

Pro

babi

lity

of c

orre

ct

clas

sific

atio

n (F

0−F

1 vs

F2−

F4)

Fibrosis estimation with TE (aetiology specific cut-offs)

Correctly classifiedIncorrectly classifiedLogistic regressionSmoothing spline

All patients(A)

(B)

Pro

babi

lity

of c

orre

ct

clas

sific

atio

n (F

0−F

1 vs

F2−

F4)

Fibrosis estimation with TE(cut-off 8.4 kPa)

Correctly classifiedIncorrectly classifiedLogistic regressionSmoothing spline

NAFLD/NASH patients

F IGURE 3 Impact of ControlledAttenuation Parameter on liver stiffnessbased fibrosis classification. NAFLD/NASH,non-alcoholic fatty liver disease/steatohepatitis, TE, transient elastography

KARLAS ET AL. | 993

which the NPV for ruling out cirrhosis is 80% and increases from 77%

at a CAP value of 250 dB/m, but reaches 85% for 350 dB/m.

The NPV for ruling out advanced fibrosis (F3-4) is 94% (93% to

95%) in our cohort. Although the CAP dependence is nominally sig-

nificant (P = 0.018), NPV changes only from 94.3% at a CAP value

of 200 dB/m to 95.5% at 350 dB/m.

3.5 | Effect of CAP on detecting fibrosis/cirrhosisusing LSM

In another common scenario, LSM is used to “diagnose” fibrosis or

cirrhosis and one must contend with false-positive findings. This too

is highly dependent on prevalence. PPV for detecting fibrosis was

found to be 79% (76% to 82%) at the observed prevalence of 50%,

but falls dramatically to 55% and 29% at a prevalence of 25% and

10% respectively. There is a significant linear dependence on CAP

(P < 0.001) with PPV of 82% for 200 dB/m and 74% at 350 dB/m,

Figure 5A. For NAFLD/NASH patients, the CAP dependence is not

significant (P = 0.33) although PPV falls a similar amount from 72%

to 63% when progressing from 200 to 350 dB/m.

Cirrhosis detection for all aetiologies has a PPV of only 40% (36

to 44)% and does not depend significantly on CAP (P = 0.21) with

PPV changing from 42% to 36% at 200 and 350 dB/m respectively,

Figure 5C. Examples of aetiology-specific PPV at high LSM values

can be found in Figure 5B and 5D. These results do not change

meaningfully if patients with BMI > 30 kg/m2 are excluded or if

models with BMI as a covariate are considered. PPV for detecting

F3-4 is only 53% (50% to 57%) in our cohort.

Even specificity-optimised cut-offs for cirrhosis, which can be as

high as 17.7 kPa for HBV patients (Table S1), result in PPV of 55%

for detecting cirrhosis and is almost independent of CAP, ranging

from 55% at a value of 200 dB/m to 53% at 350 dB/m. Considering

patients with LSM values above the ALT and AST optimised cut-off

for fibrosis and below the specificity cut-off for cirrhosis, the PPV

100

(B)

(A)

150 200 250 300 350 400

0.0

0.2

0.4

0.6

0.8

1.0

CAP (dB/m)

100 150 200 250 300 350 400

CAP (dB/m)

NP

V

Fibrosis estimation with TE (aetiology specific cut-offs)

True negativeFalse negative

All patients, fibrosis estimation0.

00.

20.

40.

60.

81.

0

NP

V

Cirrhosis estimation with TE (aetiology specific cut-offs)

True negativeFalse negative

All patients, cirrhosis estimation

F IGURE 4 Impact of ControlledAttenuation Parameter (CAP) on negativepredictive value (NPV) for ruling outsignificant fibrosis (F2-4) or cirrhosis. Thegrey shaded area denotes a 95%confidence band. TE, transientelastography

994 | KARLAS ET AL.

100 150 200 250 300 350 400

0.0

0.2

0.4

0.6

0.8

1.0

(A) (B)

(C) (D)

CAP (dB/m)

100 150 200 250 300 350 400

CAP (dB/m)

100 150 200 250 300 350 400

CAP (dB/m)

100 150 200 250 300 350 400

CAP (dB/m)

PP

V

0.0

0.2

0.4

0.6

0.8

1.0

PP

V

0.0

0.2

0.4

0.6

0.8

1.0

PP

V

0.0

0.2

0.4

0.6

0.8

1.0

PP

V

Aetiology specific fibrosis estimation

True positiveFalse positive

All patients, fibrosis detection

NAFLD/NASHHBV HCV

Example for fibrosis detection at 12 kPa

Aetiology specific cirrhosis estimation

True positiveFalse positive

All patients, cirrhosis detection

NAFLD/NASHHBV HCV

Example for cirrhosis detection at 20 kPa

F IGURE 5 Impact of Controlled Attenuation Parameter (CAP) on positive predictive value (PPV) for detecting significant fibrosis (F2-4) orcirrhosis. The grey shaded area denotes a 95% confidence band. HBV, hepatitis B; HCV, hepatitis C; NAFLD/NASH, non-alcoholic fatty liverdisease/steatohepatitis

KARLAS ET AL. | 995

for detecting significant fibrosis is 68%. It ranges from 71% for a

CAP value of 200 dB/m to 62% for a value of 350 dB/m.

3.6 | Diagnostic Flowchart using LSM andaccounting for aetiology, ALT, AST and CAP

The above results suggest that LSM be optimised for ruling out cir-

rhosis and, if possible, significant fibrosis, but imply that “ruling in”

will be of limited value. Results can be improved by using specificity-

optimised cut-offs and the resulting flowchart is presented in Fig-

ure 6. The precise performance will depend strongly on prevalence,

but for our cohort, this strategy implied 98% NPV for ruling out cir-

rhosis, but only 55% PPV for “fairly probable” cirrhosis. The chance

of cirrhosis is “quite low” (80% NPV) in a further group with LSM in

a grey zone. The chance of significant fibrosis is also “low” (NPV of

70%) for patients below the ALT and AST optimised cut-off. Neither

fibrosis nor cirrhosis can be ruled in with acceptable certainties (PPV

of 68% or 55%), despite very high, specificity-optimised cut-offs.

The particular values of NPV and PPV will depend on prevalence,

but the qualitative statements in the flowchart hold for a wide range

of scenarios. A very similar picture emerges when ruling out

advanced fibrosis, and the corresponding flowchart is presented in

Figure S6.

4 | DISCUSSION

This paper provides aetiology-specific cut-offs from a very large

cohort and confirms that LSM is a highly accurate tool for ruling out

cirrhosis. This holds independent of the CAP value. Significant fibro-

sis can be ruled out with sufficient certainty only in populations with

low prevalence and consideration of CAP can improve the negative

predictive value slightly. The accuracy of LSM for detecting signifi-

cant fibrosis, advanced fibrosis or cirrhosis is limited, due to high

false-positive rates, which is especially true for large CAP values.

These results have been summarised in diagnostic flowcharts to

facilitate interpretation of LSM in clinical practice.

There is growing pressure to develop reliable tools for screen-

ing programs in large populations aimed at detecting fibrosis.22 The

current standard for non-invasive assessment is vibration-controlled

High CAP values imply somewhat lower risk of fibrosis/cirrhosis.

Slight Relevance of CAP for LSM interpretation Negligible

Optimal cut-off:

• NAFLD 8.4 kPa

• HBV 8.0 kPa*

• HCV 6.6 kPa*

Liver stiffness measurement

ALT/AST < 5x ULN

Cirrhosis?

Optimal cut-off:• NAFLD 13.2 kPa• HBV 10.8 kPa• HCV 10.2 kPa

Ruled out

Significant fibrosis (≥F2)?

Cut-off for90% specificity:• NAFLD 14.6 kPa• HBV 17.7 kPa*• HCV 13.8 kPa*

Really cirrhosis ?

Fairly probableChances quite low

Further evaluation unavoidable.

Consider ALT/AST elevation.

Use independent non-invasive

test or biopsy.

Further evaluationif clinical doubt.

Further evaluation

recommended:

chance of cirrhosis ~55%

depending on prevalence.

Chances low Fairly probable

Rec

omm

ende

d

use

of L

SM

Fur

ther

pos

sibl

e

use

of L

SM

Pot

entia

l

use

of C

AP

Possible

F IGURE 6 Proposed diagnostic flowchart for liver stiffness measurements taking into account aetiology, transaminase values and CAP. Forthe relevant fibrosis cut-off, the sum of sensitivity and specificity (Youden index) were optimised, whereas for positive predictive values (PPV),90% specificity was required. The asterisk indicates that elevated and normal ALT/AST for HBV/HCV were considered and the lower (higher)cut-off for NPV (PPV) was chosen. See Table S1 for more details on cut-offs. ALT, alanine transaminase; AST, aspartate transaminase; HBV,hepatitis B; HCV, hepatitis C; NAFLD, non-alcoholic fatty liver disease

996 | KARLAS ET AL.

transient elastography, which has already been adopted by interna-

tional guidelines.1,2,7,23 Accurate measurement and interpretation

rely on consideration of aetiology, inflammatory activity and other

covariates,3 whereas the relevance of steatosis has been a matter

of debate.10,13-15,24,25 In a large, biopsy controlled NAFLD cohort,

Petta et al concluded that CAP should be used to gauge the risk

of false-positive LSM results and propose a diagnostic flowchart.13

We show that steatosis does indeed affect mean LSM values

slightly and confirm that CAP can be used as a surrogate in inter-

preting LSM measurements, but qualify the recommendations from

Petta et al in important ways. For one thing, CAP does not suffi-

ciently help increase the probability of correct fibrosis classification

for clinical use. Moreover, when focusing on ruling out cirrhosis,

consideration of CAP is unnecessary, but does indeed lead to

slightly higher negative predictive values for ruling out significant

fibrosis. More specifically, aetiology, prevalence and CAP should

be taken into account, particularly above a threshold of about

250 dB/m, which also arose as an optimal cut-off for detecting

steatosis in this cohort.16

We choose to focus on positive and negative predictive values

(PPV/NPV), since these are what the clinician needs to act upon,

and note the prevalence dependence inherent in them. Other mea-

sures of diagnostic precision such as likelihood ratios and areas

under the curve are also precisely interpretable, but do not have

such immediate bearing on the clinical decision to be made: the PPV

(NPV) is precisely the probability that the patient is sick (healthy),

given the test result. In the context of LSM, false-positive results

lead to reduced PPV, which indeed falls with increasing CAP, but is

never high enough to warrant a definite diagnosis. Thus, CAP will

not greatly influence the consequences of a positive/increased LSM

result, since a positive fibrosis or cirrhosis diagnosis cannot be made

on the basis of LSM alone.

The low PPV we observed despite having excluded advanced

obesity and patients with severe hepatic inflammation may seem

surprising at first. However, a comprehensive meta-analysis26

demonstrated that diagnostic performance for detection of cirrhosis

depends significantly on aetiology and country of origin and found

an AUC of 0.94 and optimal cut-off, sensitivity and specificity of

13 kPa, 0.92 and 0.89 respectively, implying a PPV of 51% at our

observed prevalence of 11%. This meta-analysis contained primarily

HCV patients and its results concur quite well with our correspond-

ing AUC of 0.94 and optimal sensitivity and specificity of 0.89 and

0.87. The cut-off chosen depends greatly on choice of optimisation

and we recommended 10.2 kPa for ruling out cirrhosis in HCV

patients, but 13.8 kPa for “ruling it in”. A non-negligible proportion

of false-positive findings seems to be inherent to vibration-con-

trolled transient elastography (Figure S5), which becomes especially

relevant at low prevalence, eg, in the screening setting. This impor-

tant observation also holds for F0-F2 vs F3-F4 in a recently pub-

lished diagnostic flowchart that notes a “low risk of F3-F4 false

positive results” compared to higher risks at higher CAP values, but

also provides a PPV of only 76% even for the lowest CAP tertile.13

A recent study looking at various combinations of non-invasive

methods found comparable PPV for detecting F3-F4.27 The low PPV

indicates the need to develop and use a combination of methods to

detect fibrosis and (even more so) cirrhosis that rely on complemen-

tary and distinct principles.27-30

NPV could be increased for ruling out fibrosis by choosing sensi-

tivity optimised cut-offs. This, however, would result in extremely

low cut-offs well into the normal range, would only include a small

fraction of the population of interest and would still not provide

extremely high NPVs. Hence, we chose a less extreme optimisation

strategy when choosing the respective cut-offs. No universally

accepted cut-offs are available3 and given the size of our cohort, it

was considered appropriate to use those we determined despite

slight overestimations regarding clinical performance. For the basic

approach of the flowchart proposed, the precise choice of cut-off

will have only minor consequences.

An important strength of the current analysis is the well-charac-

terised biopsy controlled LSM cohort, which is one of the largest

whatsoever. Data come from many countries around the world com-

prising three major aetiologies of chronic liver disease: NAFLD, HBV

and HCV. Patients at risk of unreliable results were excluded, as

were invalid measurements. Care was taken to ensure that published

evidence was not artificially “enhanced” by basing conclusions on

those same patients. The cut-offs derived from this large cohort of

individual data with rigorous exclusion criteria confirm results from

many smaller studies.

One weaknesses of the analysis is that it relies on a database

collected for assessing CAP alone and not LSM. As such, it is not a

meta-analysis, since studies were not selected for the purposes of

this analysis, but for the evaluation of CAP. On the other hand, this

suggests that the data are more representative of clinical routine

than is to be expected from a dedicated “LSM study”. However, the

sparse data available for aetiologies such as alcoholic and auto-

immune/cholestatic liver disease mean that a proportion of patients

are underrepresented. This is a void in research that should be filled,

though care must be taken to account for specific LSM characteris-

tics of differing aetiologies. Optimising cut-offs and applying them to

one and the same cohort is known to lead to overoptimistic esti-

mates. Here, we applied the strategy of adopting a wide “grey zone”,

so as to have good PPV and NPV, and do not rely on precise optimi-

sation. An inherent fact when collecting individual patient data

worldwide is that histological evaluation and biopsy quality may dif-

fer between sites as was pointed out in the primary paper, which

may be even more relevant in fibrosis staging and requires conver-

sion between scoring systems.16 This and further limitations can and

have been treated as centre effects in statistical models, but suffer

from uncertainties regarding what can be attributed to differences

between centres and what between aetiologies. Spectrum bias may

also imply that the analyses here are appropriate given current medi-

cal practice, but may have to be reconsidered in a screening setting

or in situations in which the case-mix differs notably from what was

available in our data. An LSM cut-off is not strictly speaking neces-

sary, and much information can be gained by taking into account the

precise value of a measurement, see eg, Thiele et al31 Such analyses

KARLAS ET AL. | 997

go beyond the scope of this paper, however, which orients itself to

clinical routine. CAP may be considered an imperfect surrogate for

steatosis modulated by further metabolic factors16 with the advan-

tage of being a by-product of the LSM process. However, it can also

be regarded as a measure on its own, attaining value through its pre-

dictive power as opposed to its biological interpretation. Recently,

magnetic resonance (MR) elastography and spectroscopy have been

shown to be accurate alternatives for non-invasive fibrosis staging

and steatosis grading.32-34 We were unable to take this technology

into account because of lack of data at the time our database was

created in 2015. Beyond the technical precision of MR methods,

questions regarding clinical relevance, including the effects of preva-

lence and the efficient use of such resources still need to be

answered.35,36 The current analysis only included data from vibra-

tion-controlled transient elastography as opposed to further ultra-

sound-based elastography methods, but these lack an established

means of quantifying steatosis.3 Our database only includes data

from the M-probe, which thus also restricts the BMI range that can

be considered. Moreover, data were not always available on skin to

liver capsule distance, which could be a relevant independent covari-

ate. Although the XL-probe enables examination of obese patients

for LSM,37 there are not yet sufficient data available nor is there

consensus regarding cut-offs to enable an IPDMA of comparable

scope at this point in time. In addition, XL-probe CAP was intro-

duced only recently38 and thus lacks sufficient evaluation. However,

the first comparisons indicate that the probes hardly differ regarding

CAP.39

Liver stiffness is a single number that cannot possibly replace a

multi-dimensional clinical examination. However, after taking into

account covariates, LSM provides an important guide for taking clini-

cal decisions, but the low PPV that prevails demonstrates that “false-

positives” findings are always an issue, regardless of how high the

LSM value may be. Furthermore, liver stiffness has been shown to

provide prognostic information beyond fibrosis staging alone1 imply-

ing that it is not always sound to refer to the histological staging

and grading as the gold standard.40

In conclusion, liver stiffness measurement values below aetiol-

ogy-specific cut-offs are very useful for ruling out cirrhosis and

advanced fibrosis and can help rule out significant fibrosis. High val-

ues alone are never sufficiently predictive to diagnose cirrhosis or

even significant fibrosis. Controlled Attenuation Parameter values

can be used to improve the interpretation of liver stiffness measure-

ment results only slightly. Our data indicate that CAP-guided LSM

will not be of major clinical relevance in routine practice.

ACKNOWLEDGEMENTS

Declaration of personal interests: TK, DP, VK and JW received an

unrestricted grant from Echosens, Paris France. MS is an Echosens

scientific employee in the R&D department, but contributed in her

role as scientist and author of papers in this meta-analysis. In partic-

ular, MS did not disclose scientific contents of this work to her

employer. GF and CF were provided a Fibroscan device for research

purposes. VW, GW and SM received lecture fees from Echosens. No

other authors have conflicts of interest.

Declaration of funding interests: This project is supported in part

by the German Federal Ministry of Education and Research (BMBF,

FKZ 01EO1001). The company Echosens provided funding, but does

not have access to the data, did not influence analyses and did not

receive the manuscript before submission.

AUTHORSHIP

Guarantor of the article: David Petroff is acting as the submission’s

guarantor, taking responsibility for the integrity of the work as a

whole, from inception to published article.

Author contributions: TK, DP, VK, JW, study concept and

design, collected data from participating centres, data review and

analysis, interpretation of data, drafting of the manuscript, statisti-

cal analysis, obtained funding; MS, study concept and design,

drafting of protocol, coded and provided data, interpretation of

data, statistical advice, critical revision of the manuscript; PB,

drafting of protocol, interpretation of data, critical revision of the

manuscript; All remaining authors, drafting of protocol, coded and

provided data, interpretation of data, critical revision of the manu-

script. All authors have approved the final version of the article,

including the authorship list.

ORCID

T. Karlas http://orcid.org/0000-0002-8109-8526

D. Petroff http://orcid.org/0000-0002-6916-1465

V. de L�edinghen http://orcid.org/0000-0001-6414-1951

G. Ferraioli http://orcid.org/0000-0002-6344-697X

W.-K. Chan http://orcid.org/0000-0002-9105-5837

V. W.-S. Wong http://orcid.org/0000-0003-2215-9410

S. K. Sarin http://orcid.org/0000-0002-0544-5610

REFERENCES

1. European Association for Study of Liver. Asociacion Latinoamericana

para el Estudio del Higado.. EASL-ALEH Clinical Practice Guidelines:

non-invasive tests for evaluation of liver disease severity and prog-

nosis. J Hepatology. 2015;63:237-264. https://doi.org/10.1016/j.jhe

p.2015.04.006

2. Singh S, Muir AJ, Dieterich DT, Falck-Ytter YT. American gastroen-

terological association institute technical review on the role of elas-

tography in chronic liver diseases. Gastroenterology. 2017;152:1544-

1577. https://doi.org/10.1053/j.gastro.2017.03.016

3. Dietrich CF, Bamber J, Berzigotti A, et al. EFSUMB guidelines and

recommendations on the clinical use of liver ultrasound elastography,

Update 2017 (Long Version). Ultraschall Med. 2017;38:e16-e47.

https://doi.org/10.1055/s-0043-103952

4. Arena U, Lupsor Platon M, Stasi C, et al. Liver stiffness is influenced

by a standardized meal in patients with chronic hepatitis C virus at

different stages of fibrotic evolution. Hepatology. 2013;58:65-72.

https://doi.org/10.1002/hep.26343

5. Chalasani N, Younossi Z, Lavine JE, et al. The diagnosis and manage-

ment of non-alcoholic fatty liver disease: practice guideline by the

998 | KARLAS ET AL.

American Association for the study of liver diseases, American Col-

lege of Gastroenterology, and the American Gastroenterological

Association. Hepatology. 2012;55:2005-2023. https://doi.org/10.

1002/hep.25762

6. Everhart JE, Lok AS, Kim H, et al. Weight-related effects on disease

progression in the hepatitis C antiviral long-term treatment against

cirrhosis trial. Gastroenterology. 2009;137:549-557. https://doi.org/

10.1053/j.gastro.2009.05.007

7. European Association for the Study of Diabetes (EASD); European

Association for the Study of Obesity (EASO). EASL–EASD–EASO.

EASL–EASD–EASO Clinical Practice Guidelines for the management

of non-alcoholic fatty liver disease. J Hepatol. 2016;64:1388-1402.

https://doi.org/10.1016/j.jhep.2015.11.004.

8. Karlas T, Wiegand J, Berg T. Gastrointestinal complications of obe-

sity: non-alcoholic fatty liver disease (NAFLD) and its sequelae. Best

Pract Res Clin Endocrinol Metab. 2013;27:195-208. https://doi.org/

10.1016/j.beem.2013.02.002

9. Boursier J, de Ledinghen V, Sturm N, et al. Precise evaluation of liver

histology by computerized morphometry shows that steatosis influ-

ences liver stiffness measured by transient elastography in chronic

hepatitis C. J Gastroenterol. 2014;49:527-537. https://doi.org/10.

1007/s00535-013-0819-9

10. Petta S, Maida M, Macaluso FS, et al. The severity of steatosis influ-

ences liver stiffness measurement in patients with nonalcoholic fatty

liver disease. Hepatology. 2015;62:1101-1110. https://doi.org/10.

1002/hep.27844

11. Macaluso FS, Maida M, Camm�a C, et al. Steatosis affects the perfor-

mance of liver stiffness measurement for fibrosis assessment in

patients with genotype 1 chronic hepatitis C. J Hepatol.

2014;61:523-529. https://doi.org/10.1016/j.jhep.2014.04.045

12. Berzigotti A. Getting closer to a point-of-care diagnostic assessment

in patients with chronic liver disease: controlled attenuation parame-

ter for steatosis. J Hepatol. 2014;60:910-912. https://doi.org/10.

1016/j.jhep.2014.01.017

13. Petta S, Wai-Sun Wong V, Camma C, et al. Improved noninvasive

prediction of liver fibrosis by liver stiffness measurement in patients

with nonalcoholic fatty liver disease accounting for controlled atten-

uation parameter values. Hepatology (Baltimore, MD). 2017;65:1145-

1155. https://doi.org/10.1002/hep.28843

14. Eddowes P, Sasso M, Fournier C, Vuppalanchi R, Newsome P.

Steatosis and liver stiffness measurements using transient elastogra-

phy. Hepatology. 2016;64:700. https://doi.org/10.1002/hep.28515

15. Karlas T, Beer S, Babel J, et al. Do we need controlled attenuation

parameter adjustment for fibrosis estimation in nonalcoholic fatty

liver disease patients? Hepatology. 2017;65:2126-2128. https://doi.

org/10.1002/hep.29067

16. Karlas T, Petroff D, Sasso M, et al. Individual patient data meta-ana-

lysis of controlled attenuation parameter (CAP) technology for

assessing steatosis. J Hepatol. 2017;66:1022-1030. https://doi.org/

10.1016/j.jhep.2016.12.022

17. Bedossa P. The French Metavir Cooperative Study Group. Intraob-

server and interobserver variations in liver biopsy interpretation in

patients with chronic hepatitis C. Hepatology. 1994;20:15-20.

https://doi.org/10.1002/hep.1840200104

18. Kleiner DE, Brunt EM, vanNatta M, et al. Design and validation of a

histological scoring system for nonalcoholic fatty liver disease. Hepa-

tology (Baltimore, MD). 2005;41:1313-1321. https://doi.org/10.

1002/hep.20701

19. Ishak K, Baptista A, Bianchi L, et al. Histological grading and staging

of chronic hepatitis. J Hepatol. 1995;22:696-699. PMID:7560864.

20. R Core Team. R: A language and environment for statistical computing.

Vienna, Austria: R Foundation for Statistical Computing; 2017.

21. Robin X, Turck N, Hainard A, et al. pROC: an open-source package

for R and S+ to analyze and compare ROC curves. BMC

Bioinformatics. 2011;12:77. PMID:21414208; https://doi.org/10.

1186/1471-2105-12-77

22. Gin�es P, Graupera I, Lammert F, et al. Screening for liver fibrosis in

the general population: a call for action. The Lancet Gastroenterology

& Hepatology. 2016;1:256-260. https://doi.org/10.1016/S2468-

1253(16)30081-4

23. Shiha G, Ibrahim A, Helmy A, et al. Asian-Pacific Association for the

Study of the Liver (APASL) consensus guidelines on invasive and

non-invasive assessment of hepatic fibrosis: a 2016 update. Hepatol

Int. 2017;11:1-30. PMID:27714681; https://doi.org/10.1007/

s12072-016-9760-3

24. Wong GL-H, Wong VW-S, Chan HL-Y. Is transient elastography

inaccurate in chronic hepatitis B and non-alcoholic fatty liver dis-

ease? J Hepatology. 2011;55:497; author reply 498;

PMID:21251939; https://doi.org/10.1016/j.jhep.2010.12.035

25. Gaia S, Carenzi S, Barilli AL, et al. Reliability of transient elastogra-

phy for the detection of fibrosis in non-alcoholic fatty liver disease

and chronic viral hepatitis. J Hepatol. 2011;54:64-71. https://doi.

org/10.1016/j.jhep.2010.06.022

26. Friedrich-Rust M, Ong MF, Martens S, et al. Performance of tran-

sient elastography for the staging of liver fibrosis: a meta-analysis.

Gastroenterology. 2008;134:960-974. PMID:18395077; https://doi.

org/10.1053/j.gastro.2008.01.034.

27. Petta S, Wong VW-S, Camm�a C, et al. Serial combination of non-

invasive tools improves the diagnostic accuracy of severe liver fibro-

sis in patients with NAFLD. Aliment Pharmacol Ther. 2017;46:617-

627. PMID:28752524; https://doi.org/10.1111/apt.14219.

28. Cales P, Boursier J, Lebigot J, et al. Liver fibrosis diagnosis by blood

test and elastography in chronic hepatitis C: agreement or combina-

tion? Aliment Pharmacol Ther. 2017;45:991-1003; PMID:28164327;

https://doi.org/10.1111/apt.13954.

29. Blond E, Disse E, Cuerq C, et al. EASL-EASD-EASO clinical practice

guidelines for the management of non-alcoholic fatty liver disease in

severely obese people: do they lead to over-referral? Diabetologia.

2017; PMID:28352941; https://doi.org/10.1007/s00125-017-4264-

9.

30. Tilg H. How to approach a patient with nonalcoholic fatty liver dis-

ease. Gastroenterology. 2017;153:345-349. PMID:28647352;

https://doi.org/10.1053/j.gastro.2017.06.016.

31. Thiele M, Detlefsen S, Sevelsted Moller L, et al. Transient and 2-

dimensional shear-wave elastography provide comparable assess-

ment of alcoholic liver fibrosis and cirrhosis. Gastroenterology.

2016;150:123-133. PMID:26435270; https://doi.org/10.1053/j.gas

tro.2015.09.040.

32. Raptis DA, Fischer MA, Graf R, et al. MRI: the new reference stan-

dard in quantifying hepatic steatosis? Gut. 2011;61:117-127.

https://doi.org/10.1136/gutjnl-2011-300155

33. Imajo K, Kessoku T, Honda Y, et al. Magnetic resonance imaging

more accurately classifies steatosis and fibrosis in patients with non-

alcoholic fatty liver disease than transient elastography. Gastroen-

terology. 2016;150:626-637.e7; PMID:26677985; https://doi.org/10.

1053/j.gastro.2015.11.048.

34. Park CC, Nguyen P, Hernandez C, et al. Magnetic resonance elastog-

raphy vs transient elastography in detection of fibrosis and noninva-

sive measurement of steatosis in patients with biopsy-proven

nonalcoholic fatty liver disease. Gastroenterology. 2017;152:598-

607.e2. https://doi.org/10.1053/j.gastro.2016.10.026.

35. Karlas T, Petroff D. Is magnetic resonance imaging really more accu-

rate for classifying steatosis than controlled attenuation parameter?

Gastroenterology. 2016;in press.

36. Karlas T, Petroff D, Wiegand J. Collaboration, not competition: the

role of magnetic resonance, transient elastography, and liver biopsy

in the diagnosis of nonalcoholic fatty liver disease. Gastroenterology.

2017;152:479-481. https://doi.org/10.1053/j.gastro.2016.12.013

KARLAS ET AL. | 999

37. Kumagai E, Korenaga K, Korenaga M, et al. Appropriate use of vir-

tual touch quantification and FibroScan M and XL probes according

to the skin capsular distance. J Gastroenterol. 2016;51:496-505.

PMID:26463734; https://doi.org/10.1007/s00535-015-1127-3.

38. Sasso M, Audiere S, Kemgang A, et al. Liver steatosis assessed by

controlled attenuation parameter (CAP) measured with the XL probe

of the fibroscan: a pilot study assessing diagnostic accuracy. Ultra-

sound Med Biol. 2016;42:92-103. PMID:26386476; https://doi.org/

10.1016/j.ultrasmedbio.2015.08.008.

39. de L�edinghen V, Hiriart J-B, Vergniol J, Merrouche W, Bedossa P,

Paradis V. Controlled Attenuation Parameter (CAP) with the XL

Probe of the Fibroscan: a Comparative Study with the M Probe and

Liver Biopsy. Dig Dis Sci. 2017;62:2569-2577.

40. Bedossa P, Carrat F. Liver biopsy: the best, not the gold standard. J

Hepatol. 2009;50:1-3. PMID:19017551; https://doi.org/10.1016/j.

jhep.2008.10.014.

SUPPORTING INFORMATION

Additional Supporting Information will be found online in the

supporting information tab for this article.

How to cite this article: Karlas T, Petroff D, Sasso M, et al.

Impact of controlled attenuation parameter on detecting

fibrosis using liver stiffness measurement. Aliment Pharmacol

Ther. 2018;47:989–1000. https://doi.org/10.1111/apt.14529

APPENDIX 1

THE NAMES AND FULL ADDRESSES OF THEINTERNATIONAL CAP-IPDMA GROUP

Thomas Karlas, Division of Gastroenterology and Rheumatology,

University Hospital Leipzig, Leipzig, Germany; David Petroff, Clinical

Trial Centre, University of Leipzig, Leipzig, Germany and IFB Adipos-

ityDiseases, University of Leipzig, Leipzig, Germany; Magali Sasso, R &

D Department, Echosens, Paris, France; Jian-Gao Fan, Center for Fatty

Liver, Department of Gastroenterology, XinHua Hospital, Shanghai

Jiao Tong University School of Medicine, Shanghai, China; Yu-Qiang

Mi, Research Institute of Liver Diseases, Tianjin Second People’s

Hospital, Tianjin, China; Victor de L�edinghen, Centre d’Investigation de

la Fibrose h�epatique, Hopital Haut-L�eveque, Centre Hospitalo-Univer-

sitaire de Bordeaux, Pessac, France; Manoj Kumar, Department of

Hepatology, Institute of Liver and Biliary Sciences, New Delhi, India;

Monica Lupsor-Platon, Department of Medical Imaging, Iuliu Hatie-

ganu University of Medicine and Pharmacy, Regional Institute of

Gastroenterology and Hepatology “Prof. Dr. Octavian Fodor”,

Cluj-Napoca, Romania; Kwang-Hyub Han, Department of Internal

Medicine, Yonsei University College of Medicine, Seoul, Korea; Ana C

Cardoso, HUCFF - Universidade Federal do Rio de Janeiro and Depart-

ment of Hepatology and INSERM U773-CRB3, Hopital Beaujon,

APHP, University of Paris 7, Clichy, France; Giovanna Ferraioli,

Department of Infectious Diseases, Fondazione IRCCS Policlinico San

Matteo, Medical School University of Pavia, Pavia, Italy; Wah-Kheong

Chan, Gastroenterology and Hepatology Unit, Department of Medi-

cine, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malay-

sia; Vincent Wai-Sun Wong, Department of Medicine and

Therapeutics, The Chinese University of Hong Kong, Hong Kong;

Robert P Myers, Liver Unit, Division of Gastroenterology and Hepatol-

ogy, Department of Medicine, University of Calgary, Calgary, Alberta,

Canada; Kazuaki Chayama, Departments of Gastroenterology and

Metabolism, Hiroshima University Hospital, Hiroshima, Japan; Mireen

Friedrich-Rust, Department of Internal Medicine, J.W. Goethe-Univer-

sity Hospital, Frankfurt, Germany; Michel Beaugrand, Department of

Hepatology, Hopital Jean Verdier, Bondy, France; Feng Shen, Center

for Fatty Liver, Department of Gastroenterology, XinHua Hospital,

Shanghai Jiao Tong University School of Medicine, Shanghai, China;

Jean-Baptiste Hiriart, Centre d’Investigation de la Fibrose h�epatique,

Hopital Haut-L�eveque, Centre Hospitalo-Universitaire de Bordeaux,

Pessac, France; Shiv K Sarin, Department of Hepatology, Institute of

Liver and Biliary Sciences, New Delhi, India; Radu Badea, Department

of Medical Imaging, Iuliu Hatieganu University of Medicine and Phar-

macy, Regional Institute of Gastroenterology and Hepatology “Prof.

Dr. Octavian Fodor”, Cluj-Napoca, Romania; Hye Won Lee, Depart-

ment of Internal Medicine, Yonsei University College of Medicine,

Seoul, Korea; Patrick Marcellin, Department of Hepatology and

INSERM U773-CRB3, Hopital Beaujon, APHP, University of Paris 7,

Clichy, France; Carlo Filice, Department of Infectious Diseases, Fon-

dazione IRCCS Policlinico San Matteo, Medical School University of

Pavia, Pavia, Italy; Sanjiv Mahadeva, Gastroenterology and Hepatology

Unit, Department of Medicine, Faculty of Medicine, University of

Malaya, Kuala Lumpur, Malaysia; Grace Lai-Hung Wong, Department

of Medicine and Therapeutics, The Chinese University of Hong Kong,

Hong Kong; Pam Crotty, Liver Unit, Division of Gastroenterology and

Hepatology, Department of Medicine, University of Calgary, Calgary,

Alberta, Canada; Keiichi Masaki, Departments of Gastroenterology

and Metabolism, Hiroshima University Hospital, Hiroshima, Japan;

Joerg Bojunga, Department of Internal Medicine, J.W. Goethe-Univer-

sity Hospital, Frankfurt, Germany; Pierre Bedossa, Department of

Hepatology, Hopital Jean Verdier, Bondy, France and Department of

Pathology, Physiology and Imaging, University Paris Diderot, Paris,

France; Volker Keim, Division of Gastroenterology and Rheumatology,

University Hospital Leipzig, Leipzig, Germany; Johannes Wiegand,

Division of Gastroenterology and Rheumatology, University Hospital

Leipzig, Leipzig, Germany.

1000 | KARLAS ET AL.