Controlled attenuation parameter for the assessment of liver steatosis in comparison with liver histology : a single-centre real life experience

BACKGROUND AND AIMS: Accurate diagnosis and staging of non-alcoholic fatty liver disease are essential for the management of this disorder. Controlled attenuation parameter (CAP) has been suggested as a new noninvasive measurement made during transient elastography to assess liver steatosis. The aim of this study was to evaluate CAP as a diagnostic tool for identifying the presence and degree of hepatic steatosis in consecutive patients in an outpatient liver unit of a tertiary centre.


Introduction
Non-alcoholic fatty liver disease (NAFLD) has become the most common liver disease in industrialised countries with a prevalence of up to 35% worldwide and 25% in Europe [1][2][3][4].NAFLD includes steatosis (isolated fat accumulation in hepatocytes) and steatohepatitis (steatosis with liver-cell injury/death and accumulation of inflammatory cells).Patients with NAFLD have a significantly increased risk of developing hepatic fibrosis or cirrhosis and therefore increased morbidity and mortality due to complications of liver cirrhosis or hepatocellular carcinoma.In addition, NAFLD has a strong association with metabolic syndrome and is also associated with an increased risk of cardiovascular disease [5].Both cardiovascular morbidity and complications of cirrhosis are associated with a severe impairment of quality of life and represent an increasing clinical and economic burden [6][7][8].Furthermore, NAFLD is also observed with increasing frequency concomitant to other liver diseases such as hepatitis B and C, autoimmune hepatitis, hemochromatosis, primary biliary cholangitis or alpha1-antitrypsin deficiency [9].As a concomitant disease, NAFLD can accelerate the progression of the underlying disease and promote the development of hepatocellular carcinoma [10][11][12].Therefore, early detection and treatment of NAFLD, both as separate disease and as cofactor with other liver diseases, is crucial.
Liver biopsy is the gold standard for both the diagnosis of NAFLD and the differentiation between isolated steatosis and non-alcoholic steatohepatitis (NASH).However, liver biopsy is an invasive procedure, involves a certain risk of complications, is costly and bears the risk of sampling errors [13,14].Therefore, a non-invasive diagnostic approach is preferred.The current guidelines from the European Association for Study of the Liver (EASL) mention ultrasonography, serum biomarkers and magnetic resonance spectroscopy ( 1 H-MRS) as non-invasive tests for NAFLD [15].However, ultrasonography and serum biomarker tests are relatively reliable in predicting steatosis, but not in its classification.In comparison, 1 H-MRS is the optimal examination for grading the steatosis and avoids sampling errors [16,17].However, 1 H-MRS is expensive and only available at selected centres.
Measurement of controlled attenuation parameter (CAP) is a novel tool for the non-invasive quantitative and qualitative assessment of liver steatosis.CAP is integrated in the elastography tool FibroScan ® , which is widely used to assess liver fibrosis in chronic viral hepatitis [18,19].CAP measures ultrasonic attenuation (in dB/m) at a frequency of 3.5 MHz (on a go-and-return path).Values range from 100 to 400 dB/m [20 -22].To date, CAP has been studied in various patient cohorts, mainly in Europe and Asia and in patients with metabolic syndrome and hepatitis B [23][24][25][26][27].
The aim of this study was to evaluate CAP as a diagnostic tool for identifying the presence and the degree of hepatic steatosis in consecutive patients in an outpatient liver unit of a tertiary centre.

Study population
This retrospective study was conducted at the Cantonal Hospital St Gallen, a tertiary referral centre in Switzerland.Consecutive patients who underwent liver biopsy between March 2015 and August 2016 were evaluated for inclusion in the study.Every patient considered for liver biopsy without a focal lesion and aged over 18 years was eligible for inclusion.The following clinical parameters were recorded: age, gender, body mass index (BMI) and liver diseases.Data analysis was in accordance with the ethical guidelines of the Helsinki Declaration 1975 and the study was approved by the Ethics Committee Ostschweiz (EKOS), St Gallen (Project ID 2018-02092).

Transient elastography and CAP measurement
Transient elastography and CAP measurements were performed with the FibroScan ® Touch 502 device using M or XL probes (Echosens, Paris, France).Details on CAP measurements are described elsewhere [28,29].The result is the median of ≥10 valid measurements.The measurements were taken at a depth of between 25 and 65 mm, measured on the go-and-return path at 3.5 MHz.The measurements (and the biopsy) were made on the right lateral side (intercostal space) in the liver segment VI or VII.If CAP measurement and biopsy were performed on the same day, CAP measurement was always performed before biopsy.If CAP measurement and biopsy were not performed on the same day, the maximum allowed interval was 7 days.Only valid measurements were included, based on the device's internal algorithm.Reliability was defined as (1) ≥10 valid measurements, (2) interquartile range / median (IQR/med) <30% in elastography and (3) success rate >60% [30].The physicians performing the elastography (and biopsy) had knowledge of the available clinical information.

Liver biopsy and histology
The liver biopsy was always performed by two physicians together (SB, IB or DS).The biopsies were always taken under sonographic control from the right lateral side (segment VI or VII) at the same place as the elastography (if performed before).All biopsies were made with BioPince ® Full Core Biopsy needles with a diameter of 18G and a cylinder length of 33 mm.Liver biopsy specimens were formalin-fixed and paraffin-embedded.Serial sections were stained with haematoxylin and eosin, and chromotrope aniline blue using standard methods.Sections were evaluated by two experienced liver pathologists (WJ and JN), who were blinded to CAP und elastography results.Liver steatosis was categorised as S0, S1, S2 or S3 based on the percentage of hepatocytes with large or medium-sized fat droplets, using the SAF (steatosis, activity, fibrosis) scoring system: S0 <5%; S1 5-33%, mild; S2 34-66%, moderate; S3 >66%, marked [31][32].In addition, liver sections were evaluated for hepatocyte ballooning, lobular inflammation and fibrosis using the SAF scoring system.The degree of fibrosis was also scored using the METAVIR system [33].

Statistical analysis
Descriptive statistics are reported for all included patients, for patients stratified according to their steatosis category and for the subgroup with a diagnosis of NAFLD.For measurements on a continuous scale the mean and standard deviation and for nominal characteristics the number of observations and percentages were reported.Continuous measures were assumed to follow a normal distribution, except for the liver stiffness and steatosis grade (in %).Parametric test statistics were used to compare the null hypothesis of no differences between patient groups.For comparisons between two patient groups, a p-value based on a two-sided t-test was reported.In the case of more than two groups, the reported p-values were based on an F-test.The liver stiffness and steatosis grade (in %) were compared by a nonparametric Mann-Whitney U-test and a Kruskal-Wallis test.Characteristics on a nominal scale were compared by Fisher's exact test.
The performance of the CAP parameter to predict the steatosis category was examined by the area under the curve (AUC) obtained from the receiver operating characteristics (ROC).A 95% confidence interval (CI) for the AUC of each patient group was provided and for the cutoff point, which is related to the maximum Youden index (defined as sensitivity + specificity − 1), the specificity, sensitivity, negative predictive value (NPV), positive predictive value (PPV), the accuracy and the positive as well as the negative likelihood ratio (LR).For all these measures a 95% CI was reported, if available.The ROC was carried out for groups with different steatosis categories: first the S0 patients were compared against all other and then higher steatosis grade categories were reassigned con-tinually to the S0 category.In a second analysis all steatosis grade categories were compared group wise in a separate ROC.All analyses were made with R version 3.2.3(2015-12-10) and for the ROC analyses the R-package pROC version 1.8 was used [34,35].
The influence of possible confounders, which could have an impact on the association between CAP and steatosis, was established by a multiple linear regression model, which includes the CAP measurements as outcome.The included, possible confounding factors, were the patient's age, sex, BMI, liver stiffness and the steatoses grade in percent and if a patient suffered from NASH or NAFLD.The reference category refers to female patients with no NASH.

Patient characteristics
From March 2015 to August 2016 a total 290 patients were screened and 224 were included in the analysis (fig.1).Thirty-nine patients were not included because of unreliable CAP measurements.In 7 patients fewer than 10 measurements were valid, in 11 the success rate was ≤60% and in 21 LSM IQR/med was ≥30%.In 27 cases the interval from CAP measurement to liver biopsy was more than 7 days.The majority (65%) of the 224 patients were male.Median age was 49 years (IQR 39-60).
The M or XL probe was chosen according to the device's internal algorithm.The XL probe was used on 20% of the patients.For patients with NAFLD, the XL probe was used in 69%, for non-NAFLD patients in 14%; for BMI ≥25 kg/ m 2 the XL probe was used in 32% and for BMI <25 kg/m 2 in 7%.

Association between CAP and fibrosis
A potential association between CAP and the degree of fibrosis was analysed.A comparison of the median CAP values in the fibrosis groups F0 to F4 showed no significant difference (fig.5).
A similar analysis for each pair of individual steatosis categories is presented in table S1 and figure S1 (appendix 1).

Correlation of CAP with other clinical and histological parameters
Table 4 shows the coefficients of the factors included in the regression model together with a 95% CIs and a p-values.Multivariate analysis showed correlation with CAP value for BMI, sex, liver stiffness (kPa), presence of NAFLD and steatosis in percent.No association was found for the factor age.
The comparison of liver stiffness measurements between the steatosis groups is summarised in table S2 and figure S2 (appendix 1).

Discussion
This study investigated the performance of CAP to identify liver steatosis in a "real life" scenario in consecutive patients at an outpatient liver unit of a tertiary centre.The study included more than 10 different liver diseases.However, the subgroup with hepatitis C was disproportionately large (63%), because liver biopsy was mandatory for the treatment with direct-acting antiviral drugs during trial recruitment.This study shows that CAP performs well not  only in patients with metabolic syndrome or hepatitis B (as shown in several studies before), but also in patients with a broad spectrum of underlying liver diseases [23][24][25][26][27].This is important for the future use of CAP as a screening tool for liver steatosis not only as an independent disease, but also as a concomitant condition with other liver diseases.
The AUC values of 0.78 (for the diagnosis of steatosis ≥5%), 0.83 (≥33%) and 0.82 (≥66%) are comparable to data from previous studies such as those of de Lédinghen et al. [26] (in patients with known NAFLD) or Chen et al. [25] and Ferraioli et al. [20] (in chronic viral hepatitis).Despite several existing studies with different populations, there is still no consensus on the cut-off values for CAP measurements.Compared to previous studies, the values proposed here are in the middle range, except for the value for S0 vs S1-3.This value, at 258.5 dB/m, was relatively high compared with other findings [20, 23, 25-27, 36, 37].This may be due to the more heterogeneous patient population.However, even in this population of patients with a variety of liver diseases, the cut-off values were clearly separated from each other.Additional studies will have to establish whether theses cut-off values can be applied in patients with a broad spectrum of liver diseases.
From the comparisons between the BMI groups (≥25 kg/ m 2 vs <25 kg/m 2 ) and the groups with confirmed NAFLD/ NASH versus all other liver diseases, it can be assumed that patients with a high BMI and patients with a diagnosis of NAFLD/NASH present with higher CAP values.This is confirmed by the following results: 280 dB/m and 229 dB/ m for the BMI ≥/< 25 kg/m 2 subgroups; 324.5 dB/m and 239.5 dB/m for NAFLD/NASH versus other liver diseases groups.However, it is interesting to note that the CAP values in the groups for the comparison of NAFLD/NASH with others are significantly higher than the values for the comparison of BMI ≥/< 25 kg/m 2 , which is in line with the data presented by Karlas et al. [38].Similarly, the presence of T2DM also results in increased CAP values (288 vs 253 dB/m) in comparison with non-T2DM patients.This finding could be useful for screening for NAFLD/NASH in populations at risk, such as patients with metabolic syndrome and obesity.
This study found no significant difference between CAP values of patients with steatohepatitis and those without an inflammatory component.The presence of steatohepatitis was quantified with the SAF score.Based on our findings, we assume that the inflammatory cells and ballooning have no relevant physical influence on the attenuation of the sonic wave.In this context it would be interesting to compare CAP values in a larger group of patients with pure steatosis (without ballooning, inflammation and fibrosis) with CAP values in NASH patients.However, it is unlikely that CAP will have the same prognostic impact as the components fibrosis and inflammation in NAFLD patients [39,40].Nevertheless, steatosis is a relevant co-factor in liver diseases such as chronic hepatitis C and even after sustained virological response steatosis remains a relevant risk factor for fibrosis progression [41][42][43][44][45][46][47].Therefore the use of CAP has a role as an additional prognostic tool.
This study has several limitations.Because of incomplete data sets due to too long intervals between liver biopsy and laboratory tests, we decided not to evaluate the laboratory data.A correlation with certain laboratory values and certain indices (e.g., fatty liver index) would be interesting, since according to previous studies there is still some disagreement about the association of CAP with gammaglutamyltransferase, alkaline phosphatase, aspartate or alanine transaminase, or bilirubin [24,25,27,48].
A recurring problem is the use of the M or XL probe [49][50][51][52].For this study we decided to use the algorithm suggested by FibroScan ® Touch 502.However, this algorithm refers primarily to the elastography.It is therefore unclear whether it can be used in the same way for steatosis measurement or whether the XL probe should be used according to a separate algorithm such as BMI or thickness of the subcutaneous fat tissue.
Another methodological limitation is the introduction of a potential bias by the exclusion of patients who did not fulfil the reliability criteria.For this study, we chose the rather restrictive criteria of Castera et al. (≥10 valid measurements, IQR/med <30% in elastography and success rate >60%) [30].This resulted in a total of 39 exclusions, in which NAFLD patients (n = 21, 54%) were overrepresented.This is a consequence of the more difficult and therefore more error-prone examination in obese patients.
In this study we could not use 1 H-MRS because it is not available in our centre.It would be desirable to involve the 1 H-MRS in future studies.Because the 1 H-MRS seems to be an ideal (but expensive) method for the quantification of liver steatosis and in particular takes account of the partly inhomogeneous fat distribution in the liver [16].
In conclusion, this study shows, that CAP offers a quantitative measurement of hepatic steatosis with excellent accuracy when compared with histology.This is an important advantage as it qualifies CAP not only as a noninvasive screening tool, but also for assessing the course and severity of steatosis.In addition, CAP is readily available with the well-established elastography and cheaper than biopsy or MR spectroscopy [16,53].This makes CAP measurement to a potential standard tool for steatosis screening.
This study was presented orally in abstract form at: Jahreskongress Schweizerische Gesellschaft für Gastroenterologie 2017, Interlaken; and United European Gastroenterology (UEG) Week 2017, Barcelona.

Disclosure statement
No financial support and no other potential conflict of interest relevant to this article was reported.

Figure 2 :
Figure 2: Controlled attenuation parameter (CAP) values according to the degree of steatosis S0 to S3.

Table 2 :
Characteristics of patients with non-alcoholic fatty liver disease.

Table 3 :
Performance of CAP on the basis of the area under receiver operating characteristics curves (AUROC).