Non-invasive tests for liver fibrosis assessment in patients with chronic liver diseases: a prospective study

There is an urgent need of non-invasive tests (NITs) for monitoring treatment response and disease progression in chronic liver disease. Liver stiffness (LS) evaluated by transient elastography (TE), shear wave elastography (SWE), and magnetic resonance elastography (MRE) and serum markers e.g. APRI and FIB-4 scores were assessed at baseline and the 1-year follow-up. In all, 89 chronic hepatitis C virus (HCV) patients with sustained virological response and 93 non-alcoholic fatty liver disease (NAFLD) patients were included. There was a significantly strong correlation among imaging techniques. Using MRE as the reference standard, the area under the receiver operating characteristics curves for TE, SWE, APRI, and FIB-4 in detecting stage1–4 fibrosis were 0.88–0.95, 0.87–0.96, 0.83–0.89, and 0.79–0.92, respectively. In chronic HCV patients, the values of TE, SWE, MRE, APRI and FIB-4 significantly decreased from baseline to the 1-year follow-up. Liver steatosis did not significantly change over time. In NAFLD, compared to obese patients, non-obese patients had less LS and steatosis at baseline, and these values did not show significant changes at the 1-year follow-up. Our study suggests that the current NITs have a good correlation and accuracy in monitoring the treatment outcomes in patients with chronic liver diseases.

Assessment of liver fibrosis has been a cornerstone of evaluating patients with any chronic liver disease (CLD) for therapeutic decision-making, prognostication, and evaluation of treatment response. Although liver biopsy is currently the reference standard for the identification and staging of liver fibrosis, its role has significantly declined because of several limitations such as invasiveness of the procedure, sampling errors, and intra-and interobserver variability 1,2 . The advent of non-invasive tests (NITs) has now replaced the role of liver biopsy. These approaches can overcome the limitations of liver biopsy and have become more widely used in routine clinical practice. Imaging-based elastography and serum biomarkers are currently the two main NITs for liver fibrosis staging. Liver stiffness (LS) measurement by imaging techniques such as transient elastography (TE), shear wave ultrasound elastography (SWE), and magnetic resonance elastography (MRE) have been investigated for their diagnostic performance in fibrosis staging and prediction of clinical outcomes in various CLDs [3][4][5] . Simple biomarker blood tests for fibrosis, including the aspartate aminotransferase-platelet ratio index (APRI) and fibrosis-4 index (FIB-4) score have been extensively validated and used given their easy accessibility 6 . In terms of non-invasive quantification of liver steatosis, TE-based controlled attenuation parameter (CAP) and magnetic resonance imaging-estimated proton density fat fraction (MRI-PDFF) have been investigated in patients with several CLDs, particularly non-alcoholic fatty liver disease (NAFLD) 3,7,8 . The accuracy of CAP and MRI-PDFF for diagnosing all grades of steatosis in patients with NAFLD was 0.85 and 0.99, respectively 3 . MRI-PDFF has been reported to perform better than CAP for detecting liver steatosis 3,7 .
Several issues need to be addressed regarding the aetiology-specific consideration of NITs for staging of liver fibrosis. In chronic hepatitis C virus (HCV) infection, the introduction of direct-acting antiviral (DAA) therapies has led to a paradigm shift in the treatment eradication results in high sustained virological response (SVR) achievement. Previous studies have shown that serum markers and TE values decline following SVR in chronic Assessment of liver stiffness and steatosis. All studied participants underwent LS measurement using three imaging elastography techniques, namely TE, SWE, and MRE, before DAA therapy and at 1-year after treatment initiation. All elastography tests were done on the same day.
TE was performed using a Fibroscan® 520 Touch (Echosens, Paris, France) on the right liver lobe by a 3.5 MHz M or 2.5 MHz XL probe, according to the manufacturer's instructions. The M probe was used initially, followed by the XL probe if the M probe failed. Patients were investigated under at least 2 hours of fasting. The  www.nature.com/scientificreports/ measurement depth was between 25 and 65 mm. The TE result was considered valid if the following criteria were met; (1) ≥ 10 successful measurements; (2) a success rate ≥ 60%; and (3) an interquartile range (IQR)/ median ratio < 30% 15 . Meanwhile, liver steatosis was evaluated using the CAP value provided by the machine and expressed in decibel/milliwatt (dB/m). The mean and IQR value of successful LS and CAP measurements was calculated by the device. One official certified (PT) operator was responsible for LS and CAP measurement. Two-dimensional SWE was performed using a LOGIQ E9 (GE Healthcare, Milwaukee, WI, USA) and a convex probe (1-6 MHz) by one of two abdominal radiologists (BC and NT), both of whom had evaluated the SWE in more than 100 patients. The operators were blinded to the results obtained by other techniques during the measurement. The right liver lobe was visualized through an intercostal space, and a region of interest (ROI) was placed at less than 5 cm below the liver surface and avoided major vessels. Measurements were taken 10-15 times while the patients held their breath for 5-10 seconds. The result was considered reliable when 10 successful shots and an IQR/median ratio < 30% was required. Finally, the LS value was automatically calculated as kilopascals (kPa) using shear wave velocity in the liver tissue by the machine.
MRE was performed using the Discovery™ 750w 3.0 T MRI system (GE Healthcare, Milwaukee, WI, USA) according to the protocol described previously 16 . In brief, a passive driver was placed on a supine patient over the right upper quadrant of the abdominal wall. The driver generated low-amplitude 60-Hz vibrations into the liver and the signal can be measured in the two-dimensional gradient echo sequences (2D GRE). MRE pulse sequence was used to capture the propagation of shear wave. LS values were obtained by drawing ROIs around the whole liver on each axial slice, avoiding the edges of the liver and large blood vessels. The ROI was placed by one of two abdominal radiologists (BC and NT) with more than 5 years' clinical experience. Mean LS (kPa) was calculated from three individual slices. Patients were instructed to hold their breath while obtaining each slice. The LS was considered reliable if the coefficient of variation (CV) was < 30%. Liver steatosis was evaluated using MRI-PDFF with a validated multi-echo gradient-echo pulse sequence 17  In chronic HCV patients, HCV RNA quantification was performed before and at SVR12 using a real-time reverse transcription-polymerase chain reaction (PCR)-based method (COBAS Taqman HCV assay, Roche Diagnostics, Basel, Switzerland). HCV genotype was assessed via the INNO-LiPA HCV II assay (Innogenetics, Ghent, Belgium). SVR was confirmed as undetectable HCV RNA in the plasma 12 weeks after the completion of treatment (SVR12). Liver cirrhosis was diagnosed by radiological imaging.
Statistical analysis. Continuous variables were expressed as median and standard deviation. The Mann-Whitney U test was used to compare continuous variables, and Pearson's chi-square or Fisher's exact test was used for categorical variables. Correlations between the value of serum fibrosis score, LS, and liver steatosis values were assessed using Spearman's correlation coefficient. For all analyses, P < 0.05 was considered to indicate statistical significance. The diagnostic accuracy of the studied NITs was evaluated by calculation of the area under the receiver operating characteristic (AUROC) curves using MRE as the reference. Statistical analysis was performed using SPSS software (version 22, IBM Corporation, Armonk, NY, USA).

Patient baseline characteristics.
A total of 182 patients were finally enrolled during the study period.
Of these, there were 89 patients with chronic HCV infection and 93 with NAFLD. The patients consisted of 97 (53.3%) females with a mean age (standard deviation: SD) of 53.5 (12.4) years. Forty-five patients (24.7%) had liver cirrhosis. Baseline characteristics, demographic data, laboratory parameters, LS, and steatosis values are shown in Table 1.

Correlations of liver stiffness and steatosis among imaging and serum biomarker tests.
The correlation between all NITs in study participants at baseline was investigated. MRE showed a strong positive correlation with TE (r = 0.82, p < 0.001); SWE (r = 0.83, p < 0.001); and APRI score (r = 0.71, p < 0.001) and moderate positive correlation with FIB-4 score (r = 0.69, p < 0.001). SWE was strongly positively correlated with TE (r = 0.80, p < 0.001); APRI score (r = 0.72, p < 0.001); and FIB-4 score (r = 0.71, p < 0.001). TE showed a strong positive correlation with APRI score (r = 0.74, p < 0.001) and moderate positive correlation with FIB-4 index With respect to NITs for liver steatosis measurement, MRI-PDFF was strongly positively correlated with TE-based CAP (r = 0.70, p < 0.001). However, in the subgroup analysis, the strong correlation was found only in patients with non-cirrhosis (r = 0.73, p < 0.001), while there was low correlation in patients with cirrhosis (r = 0.35, p = 0.02). Figure 2 shows the correlation between the studied NITs for liver fibrosis and steatosis.
At the 1-year follow-up, there was no significant change from baseline with respect to body weight, APRI, FIB-4 index, LS values as evaluated using imaging tests, liver steatosis values assessed by TE-based CAP, and MRI-PDFF in both obese and non-obese patients with NAFLD. Besides, the proportions of patients who had liver fibrosis progression, defined as an increase by at least one stage of fibrosis evaluated by MRE from baseline to the 1-year follow-up, was not different between patients with non-obese and obese NAFLD (6.7% vs. 9.5%, p = 0.45). Table 3 shows a comparison of the mean changes of simple serum fibrotic score, LS, and steatosis values at baseline and the 1-year follow-up in obese and non-obese patients with NAFLD.    Non-invasive tests for assessment of liver fibrosis are becoming more desirable and widely used to improve the diagnosis and prognostication of CLD of various aetiologies. TE, SWE, and MRE exhibited excellent diagnostic accuracy for detecting liver fibrosis and have emerged into clinical practice and research purposes 22,23 . However, data on the correlations among three elastography techniques in patients with CLD patients are scarce. Our study showed a good correlation among imaging elastography tests and between imaging tests and simple serum biomarkers, consistent with a previous study in which MRE and TE showed moderate correlation 24 . MRE can be considered the most accurate NIT for detecting liver fibrosis, especially in advanced fibrosis and cirrhosis 25 . In a recent study, MRE outperformed TE and SWE in detecting stage 4 liver fibrosis in patients with NAFLD 22 . Furthermore, MRE has an advantage over TE or SWE in that it visualizes the whole liver rather than just small hepatic areas. Hence, MRE was selected as the reference method for the staging of liver fibrosis in the current study. Overall, TE, SWE, APRI, and FIB-4 scores had a high diagnostic performance for detecting stage 1-4 liver fibrosis. The accuracy of all tests was comparable. Thus, although MRE is currently accepted as the most precise NIT for evaluating liver fibrosis, our study showed that TE, SWE, APRI, and FIB-4 had a good correlation and can be used particularly in centres with limited availability of MRE. In the current study, imaging Table 3. Baseline characteristics and change of serum simple fibrotic marker, elastography and steatosis value in Lean and obese NAFLD. AST aspartate aminotransferase, ALT alanine aminotransferase, APRI AST to Platelet Ratio, BW body weight, BL baseline, CAP controlled attenuation parameter, FBS fasting blood sugar, FIB-4 Fibrosis-4, HbA1c Glycated haemoglobin, HDL high density lipoprotein, LDL low density lipoprotein, MRE magnetic resonance elastography, MRI magnetic resonance imaging, PDFF proton density fat fraction, SWE shear wave elastography, TE transient elastography. www.nature.com/scientificreports/ findings revealed that 45 out of 182 patients had cirrhosis. Of these, 29 patients (64.4%) had MRE-based LS cutoff ≥ 4.6 kPa (F4). The discrepancy between MRE and radiological results might be explained by interobserver variability among radiologists. Liver necroinflammation influences the increase in LS as evaluated by imaging-based elastography [26][27][28] . Therefore, the presence of advanced fibrosis or cirrhosis assessed by LS measurement in patients with chronic HCV prior to DAA treatment might be overestimated. Previous studies used TE 29,30 or MRE 31 to show that the LS values reduced from baseline to SVR24 in patients with chronic HCV treated with DAA. Our study demonstrated that the LS value based on three elastography tests, APRI, and FIB-4 index in chronic HCV patients with SVR significantly decreased from baseline to 1 year following DAA treatment. Viral eradication resulted in a reduced proportion of patients with advanced fibrosis by 13%, 28%, and 11% based on TE, SWE, and MRE, respectively. Elevated ALT levels, which indicate the presence of necroinflammatory activity, decreased between baseline and SVR12 and between SVR12 and the 1-year follow-up. Thus, the reduction of LS values and serum fibrosis scores after viral eradication were partly caused by the decrease in liver inflammation rather than liver fibrosis regression. Another previous study showed that the improvement of liver fibrosis was not evident in 20 paired biopsies in the short-term of chronic HCV patients who achieved SVR after DAA treatment 32 . Reassessing LS after achieving SVR might be more accurate for non-invasive diagnosis of advanced fibrosis and be helpful information for clinical decision-making on patient follow-up. Moreover, liver steatosis values based on Fibroscan-based CAP and MRI-PDFF were not different between the baseline and 1-year follow-up after initiation of DAA therapy in patients with chronic HCV. This was inconsistent with Tada et al. study, which showed that viral eradication improved MRI-PDFF values at SVR24 31 . The possible reason for the discordant results between both studies may be the difference of HCV genotype (47.4% genotype 1b/52.6% genotype 2 vs. 46.0% genotype 1/44.9% genotype 3) and time of repeated liver steatosis measurement (24 weeks after end-oftreatment vs. 24-36 weeks after end-of-treatment) between Tada et al. study and our study 31 . HCV genotype is strongly associated with the prevalence of liver steatosis. Genotype 3 infection is now commonly recognized as an independent aetiological factor of steatosis 33,34 .
The prevalence of lean NAFLD varies from 5-26%, although it ranges from 5-45% among Asians and 5-20% in European populations 35,36 . It is unclear whether the clinical and histological characteristics and prognosis differ between obese and non-obese patients with NAFLD. A recent study by Younes et al. showed that despite the fact that non-obese patients with NAFLD patients had less severe histological disease, advanced fibrosis, and lower prevalence of diabetes, they still experienced similar long-term prognosis and survival compared to obese patients with NAFLD 37 . A prospective Chinese cohort of biopsy-proven NAFLD patients showed that non-obese NAFLD patients had less metabolic syndrome, NAFLD activity score, and liver fibrosis and likely a better prognosis than obese patients 38 . In contrast, a European study found that even though non-obese patients with NAFLD had fewer metabolic syndrome components, they had severe histological features similar to obese patients and had more progressive disease than obese NAFLD patients 39 . Data on liver stiffness and steatosis evaluated by imaging methods in non-obese compared to obese NAFLD patients are limited. The current study demonstrated that non-obese patients with NAFLD had less LS and liver steatosis, while they had a similar metabolic profile with similar dyslipidaemia and glycaemia when compared to patient with obese NAFLD. Changes in BMI, serum fibrosis score, LS, and steatosis values were not different between obese and non-obese patients with NAFLD at the 1-year follow-up. The reason for different observations among these studies may be explained by a major confounding factor of the younger age of non-obese patients, which has been mentioned in previous studies 37,40 . However, it is possible that this is not the true reason given that the mean age of non-obese and obese NAFLD patients was not significantly different in the Chinese and European studies. Additionally, non-obese patients in our study were older than obese patients. Further studies are needed for the long-term progression of LS and steatosis evaluated by NITs and to compare both groups of NAFLD patients.
This study has some limitations. First, the number of patients and the period of follow-up were both relatively small. Further large-scale studies with longer follow-up are needed to confirm these findings. Second, the study lacked liver biopsy results. Despite several limitations of liver biopsies, the validation process using biopsy is still mandatory to evaluate the performance of a new test. However, the performance of several tests can be assessed without the need for biopsy. Furthermore, TE, SWE, and MRE have been extensively validated in a large number of studies. Third, all TE, SWE, and MRE evaluations were performed on the same equipment for each technique. Therefore, extrapolating our findings to the clinical setting with various MRI or ultrasound models might not be possible. Fourth, the number of patients in each fibrosis stage was unequal, with 70, 42, 26, 7, and 37 patients having F0, F1, F2, F3, and F4, respectively. However, this might not have a significant impact on the results since the primary objective of the study was to assess the diagnostic performance of current NITs value in patients with chronic liver diseases. Fifth, despite the fact that hepatic inflammation affects the accuracy of NITs for liver fibrosis staging, patients with elevated aminotransferase were not excluded from studying the change of NITs value after HCV eradication. Moreover, only 8/182 (4.4%) patients had baseline serum ALT levels greater than five times the upper limit of normal.
In conclusion, TE, SWE, MRE, APRI, and the FIB-4 index have a good correlation. Using MRE as the reference standard, all studied NITs have high diagnostic accuracies for detecting liver fibrosis. Viral eradication after DAA therapy in chronic HCV patients results in the improvement of APRI, FIB-4 score, and LS as assessed by TE, SWE, and MRE. Non-obese NAFLD patients have lesser LS and steatosis than obese NAFLD patients, but changing of LS and steatosis are not different after short-term follow-up.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.