Jung Il Lee, Hyun Woong Lee, Kwan Sik Lee

Abstract

Key words: Fibroscan; Controlled attenuation parameter; Liver stiffness; Nonalcoholic steatohepatitis; Liver fibrosis

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is the most common cause of chronic liver disease around worldwide. The spectrum of NAFLD ranges from simple steatosis without evidence of liver injury to nonalcoholic steatohepatitis (NASH) with or without liver fibrosis[1]. Although the natural history of NAFLD requires further investigation, studies demonstrate that the severity of liver fibrosis is the most important determinant of mortality and morbidity in patients with NAFLD[2-4].Patients with significant liver fibrosis [Kleiner classification fibrosis stage (F) ≥ 2]showed decreased survival compared to those with no or minimal fibrosis (F0-1)[4].NAFLD may progress from simple steatosis to NASH with fibrosis, and estimation of severity of liver fibrosis is critical not only for the initial workup but also for followup[5].

Although liver biopsy is considered the gold standard for assessing the severity of fibrosis[6], it is an invasive procedure that might not be practical to perform sequentially. Instead, the liver stiffness measurement (LSM), obtained by transient elastograpy (TE) is a useful noninvasive means of assessing liver fibrosis. LSM values are well correlated with the biopsy determined severity of fibrosis[7-9]. However, the diagnostic performance of LSM is known to be affected by obesity and the severity of steatosis, which are closely associated with NAFLD, resulting in overestimation of the LSM in patients with NAFLD[10-12]. Recently, FibroScan, a type of TE, has been equipped with controlled attenuation parameter (CAP), software to enable noninvasive quantification of hepatic steatosis. The CAP value was strongly correlated with the histologically assessed percentage of liver fat in patients with NAFLD, but is susceptible to interference by liver fibrosis[13]. However, CAP may enhance the accuracy of TE measured LSM in patients with NAFLD[14]. We evaluated the role of the CAP value in predicting the liver fibrosis stage based on LSM in patients with biopsy proven NASH.

MATERIALS AND METHODS

Study design

This retrospective study involved patients with biopsy proven NASH evaluated at Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea. Liver biopsy was performed to confirm the diagnosis of NASH in patients with ultrasound findings of fatty liver and persistent (＞ 6 mo) elevation of the alanine aminotransferase (ALT) or aspartate aminotransferase (AST) level without excessive alcohol consumption (30 g/day in men and 20 g/day in women). Patients in whom liver stiffness was evaluated within 1 mo before the liver biopsy were included in the analysis. The exclusion criteria were as follows: (1) Liver disease of other or mixed etiology (such as hepatitis B infection, hepatitis C infection, alcohol abuse,autoimmune liver disease, Wilson's disease, or drug-induced liver disease); (2) LSM evaluated while the AST or ALT level was more than fivefold the upper limit of normal (ULN); (3) Hepatocellular carcinoma; (4) Advanced liver cirrhosis (Child-Turcotte-Pugh B and C); (5) Previous treatment with steatosis-inducing drugs such as tamoxifen, aromatase inhibitor, valproic acid, amiodarone or corticosteroid; (6)Human immunodeficiency virus infection; (7) Active intravenous drug addiction or use of cannabis; and (8) Insufficient clinical data.

This study was performed in accordance with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Institutional Review Board (IRB) of Gangnam Severance Hospital (permit no: 3-2019-0010). The requirement for written informed consent was exempted by the IRB since the database was accessed only for analysis purposes and the patients' personal information was anonymized by coding.

Clinical assessment

Demographic, clinical and anthropometric data were collected at the time of liver biopsy. Hypertension was defined as use of antihypertensive medication and type 2 diabetes mellitus was considered present if the fasting glucose level was ≥ 126 mg/dL or antidiabetic agents were being used. Body mass index (BMI) was calculated as body weight in kilogram divided by the square of height in meters, and a BMI ≥ 25 kg/m2was considered to indicate obesity based on the criteria used in the Asian-Pacific region[15].

Liver stiffness measurement

TE was performed using a FibroScan (Echosens, Paris, France), medical device, with a standard probe. Only LS values with at least 10 valid measurements, a success rate of at least 60%, and an interquartile range-to-median ratio of ＜ 30% were considered reliable, as suggested by previous studies[16,17]. In addition, patients in whom LS was measured while the AST or ALT ＞ 5 × ULN was present, were excluded from the analysis due to possible exaggerated LSM values as previous studies demonstrated[18,19]. The baseline LSM was obtained within 1 month before liver biopsy.The follow-up LSM was performed after 12 mo of NASH treatment with daily dose of 15 mg pioglitazone, peroxisome proliferator-activated receptor (PPAR)-γ agonist, or 300 mg/day ursodeoxycholic acid (UDCA) following liver biopsy.

Histologic assessment

A single liver-dedicated expert pathologist, blinded to the patients' identity,performed the histologic analysis. A ≥ 15-mm-long biopsy specimen or the presence of at least 10 complete portal tracts was considered adequate for the analysis[20]. NASH was diagnosed according to the NASH Clinical Research Network System, and was defined as the presence of ≥ 5% hepatic steatosis and inflammation with hepatocyte injury such as ballooning with or without fibrosis[21].

Statistical analysis

All statistical tests were performed using IBM SPSS Statistics 22.0 (IBM, Armonk, NY,United States). Continuous variables are expressed as means ± standard deviation(SD) or medians (range). The area under the receiver-operating characteristic (ROC)curve was calculated to reflect the overall accuracy of LSM for diagnosing significant fibrosis (F2-4). Categorical variables were compared by using two-sided χ2-test (or Fisher's exact test, or McNemar test, as appropriate) and continuous variables by independent or Mann-Whitney test as appropriate. A paired t test was performed to evaluate changes in LSM. A two-sided P value of ＜ 0.05 was considered indicative of statistical significance.

RESULTS

Patients

From January 2010 to December 2017, 325 patients underwent liver biopsy and LSM assessed due to suspicion of NASH. Among them, 184 patients met the inclusion and exclusion criteria and were thus included in the analysis. The baseline characteristics of the 184 patients with NASH are listed in Table 1.

Assessment of steatosis using CAP

Among the 184 patients with biopsy proven NASH, the distribution of the histologically assessed steatosis grade (S) was as follows: S1, n = 44 (2.9%); S2, n = 81(44.0%); S3, n = 59 (32.1%) (Figure 1). The CAP scores were significantly different between S1 and S2-S3 (P ＜ 0.001). However, no significant difference was detectedbetween S2 and S3 (P = 0.075). The ROC curve showed the optimum CAP cutoff for ≥S2 was 313.5 dB/m [area under the curve (AUC), 0.736; sensitivity, 72.9%; specificity,63.6%]. On the other hands, the accuracy dropped to an AUC of 0.656 for S3, with a cutoff of 323.5 dB/m (sensitivity 64.4%; specificity 55.2%).

Table 1 Demographic, clinical, biochemical and histological characteristics of 184 patients with nonalcoholic steatohepatitis

Factors associated with LSM

LSM significantly increased with the histologically detected fibrosis stage (F0, 7.5 ±2.1; F1, 9.8 ± 2.7; F2, 11.8 ± 4.9; F3, 15.4 ± 6.9; and F4, 20.3 ± 8.8 kPa) (P ＜ 0.001). The ROC curve showed that the optimum LSM cutoff for ≥ F2 was 8.95 kPa (AUC, 0.730;sensitivity, 72.5%; specificity, 65.4%). According to univariate and multivariate analyses, the CAP values and pathologically detected fibrosis stage was significantly associated with LSM (Table 2).

Figure 1 Distribution of controlled attenuation parameter scores according to histologically assessed steatosis grades (S). CAP: Controlled attenuation parameter.

Although there were significant differences in CAP scores between S1 and S2-S3, no cutoff could differentiate S3 from S2. S2 and S3 accounted for the majority of the patients (76.1%, 140/184). Furthermore, in a multivariate analysis, the CAP value, but not the pathologically detected steatosis grade, was associated with LSM. Therefore,the variations in the LSM value for each stage of liver fibrosis was evaluated according to arbitrary CAP tertiles (lower, 223-310; middle, 311-339; high 340-400 dB/m) (Figure 2). For F0-1, LSM values significantly increased at high CAP tertile (P =0.001) (Figure 2A). For F2, LSM values were higher for the middle and high CAP tertiles (P = 0.027) (Figure 2B). However, the LSM values did not differ significantly according to CAP tertile in patients with NASH and advanced fibrosis (F3-4) (P =0.752) (Figure 2C).

When cutoff of 8.95 kPa was used to diagnose significant fibrosis (F ≥ 2), positive predictive values in lower, middle and high CAP tertiles was 21/31 (67.7%), 19/32(59.4%) and 23/49 (46.9%), respectively. False positive rates increased with increasing CAP tertile (Figure 3A). However, when different cutoffs were used for each CAP tertile, differences in false positive rates among different CAP tertile were reduced and the false positive rate in high CAP tertile decreased (Figure 3B). ROC curves showed that the optimum LSM cutoff for diagnosis of F ≥ 2 was 8.05 kPa (AUC, 0.682;sensitivity, 73.5%; specificity 51.7%) for the lower CAP tertile, LSM of 9.35 kPa (AUC,0.843; sensitivity, 90.5%; specificity 71.8%) for the middle CAP tertile, and LSM of 10.55 kPa (AUC 0.682; sensitivity, 76.0%; specificity 52.8%) for high CAP tertile.

Effect of PPAR-γ agonist on follow up LSM

Among the 184 patients with biopsy proven NASH, 130 patients had LS measured 1 year after liver biopsy and were treated with PPAR-γ agonist (pioglitazone) 15 mg/day (n = 80) or UDCA 300 mg/day (n = 50). Regarding the baseline characteristics, there was no significant difference in BMI, LS value or CAP score between the two groups (Table 3). However, the pioglitazone group had higher rates of DM and hypertension (P = 0.048, P = 0.049, respectively).

The patients treated with pioglitazone demonstrated a decreased LSM value after 1- year of treatment (P ＜ 0.001), when that in UDCA-treated patients did not change significantly (P = 0.068) (Figure 4). The CAP values did not show significant changes after treatment with pioglitazone (318.2 ± 37.9 vs 313.2 ± 41.5 dB/m, P = 0.197) or UDCA (319.3 ± 37.1 vs 309.2 ± 38.2 dB/m, P = 0.057). Changes in the proportion of patients with LSM suggesting significant fibrosis (F ≥ 2) were assessed separately in the pioglitazone and UDCA groups according to the LSM cutoffs for each CAP tertile,that were obtained from the analysis with 184 patients. Among the patient with lower CAP tertile (223-309 dB/m), those treated with pioglitazone showed a decreased proportion of high LSM values (F ≥ 2) but the proportion of high LSM values did not change significantly among the patients treated with UDCA (Figure 4A). Similar results were noted in patients with middle (310-332 dB/m) and high CAP tertile (333-400 dB/m) (Figure 4B and C).

DISCUSSION

This study of 184 patients with biopsy proven NASH demonstrated that high CAP scores are associated with increased LSM values at the same fibrosis stage, resulting in overestimation of liver fibrosis. Lower positive predictive values were noted in patients in the high CAP tertile, particularly those with F0-2. Therefore, higher LSM cutoffs might be useful for identifying significant fibrosis in patients with NAFLD andhigh CAP values. Although a “high CAP cutoff value” had yet to be defined, the cutoff for the high CAP tertile in this study was 330-340 dB/m.

Table 2 Univariate and multivariate analysis of factors associated with liver stiffness measurements as continuous variable in 184 patients with nonalcoholic steatohepatitis by linear regression analysis

Two prospective cohort studies on the natural history of NAFLD proposed that the severity of liver fibrosis is the most important predictor of liver-related complications as well as survival in patients with NAFLD[2,3]. In addition, recent studies suggested that only the severity of fibrosis is an important prognostic factor for NAFLD, and is independent of NASH and the severity of inflammation[4,22]. These studies investigated the prognostic value of the baseline liver fibrosis stage, and one also assessed the progression of liver fibrosis in patients with NAFLD[5]. Therefore, when sequential liver biopsy is not practical, accurate prediction of fibrosis stage using noninvasive methods is important. To reduce the effect of hepatic steatosis, as indicated by CAP scores, on the prediction of fibrosis based on LSM, we calculated the cutoff values for significant fibrosis in according to CAP tertile. We applied arbitrary CAP tertiles since the CAP score did not accurately differentiate S2 from S3 which accounted for the majority of the patients. Also, CAP reportedly cannot differentiate adjacent grades of steatosis with high precision[23]. As a noninvasive means of steatosis measurement, magnetic resonance imaging (MRI)-proton density fat traction (PDFF) is reported to be more accurate for predicting hepatic steatosis compared to CAP[24]. Magnetic resonance elastography (MRE) has the highest diagnostic accuracy for staging fibrosis in patients with NAFLD. However, both MRIPDFF and MRE are MRI-based tools that require more space and more costly than FibroScan with CAP[25].

Cutoff values according to CAP tertile were applied to estimate the effect of pioglitazone and UDCA on the LSM values at the 1-year follow up. Treatment with pioglitazone reduced LSM-estimated proportion of significant fibrosis. However, the CAP score did not change significantly after pioglitazone or UDCA administration. In recent NAFLD practice guidance, pioglitazone and 800 IU/day vitamin E are recommended to improve liver histology in patients with NASH[26]. The 184 patients from which the LSM cutoff values for significant fibrosis were estimated included 130 patients treated with pioglitazone or UDCA. Although pioglitazone improved the LSM value, the utility of this result is limited by several factors. Ironically, the accuracy of LSM cutoff determined based on the CAP score cannot be validated without follow-up biopsy after pioglitazone or UDCA treatment in patients with NASH. Therefore, studies involving paired liver biopsies are needed, until more reliable LSM standards for NAFLD are established. Secondly, being a retrospective,observational study, the baseline demographic parameters of the pioglitazone and UDCA-treated groups were not matched. Patients treated with pioglitazone were more likely to be diabetic and hypertensive which are important elements of metabolic syndrome. Among NAFLD patients with fibrosis progression, 80% were diabetic, suggesting that diabetes promotes the progression of NASH[5]. Although it involved a larger number of patients with diabetes, our study showed that pioglitazone resulted in a reduced proportion of patients with high LSM values compared to UDCA. However, the sample size was too small to reach a definite conclusion. Finally, unlike previous investigations of the effect of pioglitazone on NASH, the patients in this study received low dose of pioglitazone (15 mg/day). Two randomized studies of the effect of pioglitazone on NASH prescribed pioglitazone 30 mg or 45 mg daily to patients with NASH[27,28]. The mean BMIs of the patients in these previous studies were about 33-35 kg/m2, compare to 29.2 ± 4.5 kg/m2in this study,and the lower dose of pioglitazone may have been effective due to the lower BMI of our patients. Further studies investigating smaller doses of pioglitazone on patients with NASH are needed to verify our results.

Figure 2 Distribution of liver stiffness values within the same Kleiner fibrosis stages (F0-4) in different controlled attenuation parameter tertiles. A: F0-1; B:F2; C: F3-4. CAP: Controlled attenuation parameter.

In conclusion, LSM in patients with NASH may overestimate the liver fibrosis stage, particularly in those with high CAP values. Interpretation of LSM results taking into consideration the simultaneously measured CAP scores may prevent the performance of unnecessary liver biopsy in patients with NAFLD.

Table 3 Demographic, clinical, biochemical and histological characteristics of patients with nonalcoholic steatohepatitis under either pioglitazone or ursodeoxycholic acid

Figure 3 False-positive rates when evaluating liver fibrosis stages by liver stiffness measurements. A: The cutoff of 8.95 kPa for detecting Kleiner fibrosis stage (F) ≥ 2 was applied in all controlled attenuation parameter (CAP) tertiles; B: Different cutoffs were applied according to CAP tertiles. CAP: Controlled attenuation parameter; LSM: Liver stiffness measurements.

Figure 4 Change in the proportion of liver stiffness measurements predicting significant liver fibrosis, Kleiner fibrosis stage F ≥ 2, after 1-year of pioglitazone or ursodeoxycholic acid treatment. A: In the lower controlled attenuation parameter (CAP) tertile; B: In the middle CAP tertile; C: In the high CAP tertile. The liver stiffness measurements (LSM) cutoff values for F ≥ 2 are derived from the analysis 184 patients with biopsy proven nonalcoholic steatohepatitis after considering different CAP scores in tertiles. The changes in LSM values are compared by paired t-test. CAP: Controlled attenuation parameter; LSM: Liver stiffness measurements.

ARTICLE HIGHLIGHTS

In conclusion, LSM in NASH may overestimate the liver fibrosis stage particularly in patients with high CAP values. Interpretation of LSM considering simultaneously measured CAP scores may provide more helpful information preventing unnecessary liver biopsy in patients with NAFLD.

Research perspectives

In patients with NAFLD with high CAP scores, LSM cutoff that leads to liver biopsy may need to be set higher than in those with other chronic liver diseases. Validation studies for more precise LSM cutoffs should be performed incorporating larger number of patients with biopsy proven NAFLD. With more reliable LSM cutoffs for noninvasive diagnosis of liver fibrosis in NAFLD,clinical studies evaluating efficacies of treatment would be more widely preformed in NAFLD.