The severity of steatosis influences liver stiffness measurement in patients with nonalcoholic fatty liver disease

Patients

The study assessed consecutive patients with biopsy-proven NAFLD, recruited at the Gastrointestinal Liver Unit of the University Hospital in Palermo, Italy, and fulfilled all inclusion and exclusion criteria detailed below. The diagnosis of NAFLD was based on chronically elevated alanine aminotransferase for at least 6 months, alcohol consumption <20 g/day in the last year, and steatosis (≥5% of hepatocytes) at histology with or without necroinflammation and/or fibrosis. Exclusion criteria were as follows: (1) advanced cirrhosis (Child-Turcotte-Pugh B and C); (2) hepatocellular carcinoma; (3) other causes of liver disease or mixed etiologies (alcohol abuse, hepatitis C, hepatitis B, autoimmune liver disease, Wilson's disease, hemochromatosis, or α1-antitrypsin deficiency); (4) human immunodeficiency viral infection; (5) previous treatment with immunosuppressive drugs and/or regular use of steatosis-inducing drugs, evaluated by a questionnaire (for example, corticosteroid, valproic acid, tamoxifen, amiodarone); or (6) active intravenous drug addiction or use of cannabis.

The study was performed in accordance with the principles of the Declaration of Helsinki and its appendices and with local and national laws. Approval was obtained from the hospital's Internal Review Board and Ethics Committee, and written informed consent was obtained from all patients.

Clinical and Laboratory Assessment

Demographic, clinical, and anthropometric data were collected at the time of liver biopsy. The diagnosis of arterial hypertension was based on a systolic blood pressure ≥135 mm Hg and/or a diastolic blood pressure ≥85 mm Hg (measured three times within 30 minutes using a brachial sphygmomanometer) or on use of blood pressure–lowering agents. The diagnosis of type 2 diabetes was based on the revised criteria of the American Diabetes Association, using a fasting blood glucose value of ≥126 mg/dL on at least two occasions.27 A current therapy with insulin or oral hypoglycemic agents was documented in patients with a previous diagnosis of type 2 diabetes.

A 12-hour overnight fasting blood sample was drawn at the time of biopsy to determine serum levels of alanine aminotransferase, platelet count, total cholesterol, high-density lipoprotein cholesterol, triglycerides, blood glucose, and insulin. Insulin resistance was assessed with the homeostasis model assessment method using the following equation28: Insulin resistance = [Fasting insulin (μU/mL) × Fasting glucose (mmol/L)]/22.5.

Liver Stiffness Measurement

Transient elastography was performed with the FibroScan (Echosens, Paris, France) medical device, using the M probe (also named as the standard probe). On the same day of liver biopsy, LSM was assessed, before the procedure and after an overnight fast, by trained operators who had previously performed at least 300 determinations in patients with chronic liver disease. As recently reported in the literature,29 we classified all LSM examinations into three reliability categories: “very reliable” (interquartile range/median [IQR/M] ≤0.10), “reliable” (0.10 < IQR/M ≤ 0.30 or IQR/M > 0.30 with LSM median <7.1 kPa), and “poorly reliable” (IQR/M >0.30 with LSM median ≥7.1 kPa). “Poorly reliable” results were excluded from the analysis.

US Assessment

The US assessment was performed in the morning, on the same day of liver biopsy, by one operator trained for US techniques and particularly dedicated to liver examination, using a real-time Hitachi H21 apparatus with a 2-5 MHz, convex, multifrequency probe. Presence of hepatic steatosis was defined through detection of bright liver echo pattern (BLEP), that is, fine, packed, and high-amplitude echoes, with consequent brightness of liver, increase in liver-kidney contrast, and possible evidence of vascular blurring and deep attenuation signs.30 Mild steatosis was thus recognized by a slight increase in liver echogenicity, a slight exaggeration of liver and kidney echo discrepancy, and relative preservation of echoes from the walls of the portal vein. Moderate steatosis was accompanied by loss of echoes from the walls of the portal veins, particularly from the peripheral branches, resulting in a featureless appearance of the liver. In addition, greater posterior beam attenuation was found as well as a greater discrepancy between hepatic and renal echoes. Severe steatosis was recognized by a greater reduction in beam penetration, loss of echoes from most of the portal vein wall including the main branches, and a large discrepancy between hepatic and renal echoes.31

Histological Assessment

Histological slides were coded and read by one expert pathologist, who was unaware of the patient's identity and history. A minimum length of 15 mm of the biopsy specimen or the presence of at least 10 complete portal tracts was required.32 Steatosis was assessed as the percentage of hepatocytes containing fat droplets (minimum 5%) and evaluated as a categorical variable. The Kleiner classification33 was used to compute steatosis and lobular inflammation and to stage fibrosis from 0 to 4.

Statistics

The results are reported as frequencies and percentage for categorical variables, mean ± standard deviation for normally distributed continuous variables, and median and IQR for nonnormally distributed continuous variables. The Student t test, the Mann-Whitney, and the χ² test were used to compare continuous or categorical variables. Multivariate analysis, including all of the significant baseline variables (P < 0.05), was also performed using linear regression to identify independent variables associated with LSM as continuous variables.

Receiver operating characteristic curves were applied to find the best cutoff values and to identify the area under the receiver operating characteristic curve of LSM that is able to discriminate the different stages of fibrosis. Furthermore, descriptions of the operating characteristics (sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio) of US for the detection of steatosis >66%, with the assumption that the gold standard for the diagnosis of steatosis was histologic examination. All analyses were performed using SPSS v. 20.0 statistical package for MacIntosh (SPSS, Inc., Chicago, IL).

Patients

From January 2008 to October 2013, we included 306 consecutive patients with NAFLD who underwent TE, US, and liver biopsy. Fifty-three (17%) failed to obtain 10 valid LSM acquisitions due to obesity or to unreliable results according to the manufacturer's recommendations (see above). Hence, 253 patients with valid LSM acquisitions could be included in the analysis. Table 1 shows the comparison between NAFLD patients with reliable and those with unreliable LSM. As expected, the latter were more obese compared to their counterparts, even if the overall severity of the liver damage was not different between the two groups. About the baseline features of the 253 included patients, the mean age was 45.2 years, with a prevalence of males (70%). The greater proportion of patients was in the overweight or obese range, one-quarter was hypertensive, and diabetes was present in about 20% of subjects. Mean values for total, high-density lipoprotein cholesterol, and triglycerides were within the normal range, whereas mean homeostasis model assessment values were elevated (4.0 ± 3.3).

At liver biopsy, approximately one patient in two had fibrosis ≥2 by Kleiner score, whereas severe liver fibrosis (≥3) prevalence was 20%. About 28% of the cases had steatosis of severe grade (≥66%).

Performance of US on the Assessment of Severe Steatosis

All patients had BLEP at US examination, while a severe BLEP was observed in 65 patients (25.7%). Sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio of US for the detection of steatosis ≥66% were 48.6%, 83.1%, 53.8%, 80%, 2.88, and 0.62, respectively.

Factors Associated With LSM and Accuracy of TE for Fibrosis Assessment

Median LSM was 6.8 (IQR 5.4-10.0). Liver stiffness values significantly increased according to fibrosis stage: in F0 5.7 (IQR 4.8-7.4), in F1 6.1 (IQR 5.2-7.7), in F2 7.2 (IQR 6.0-10.0), in F3 10.7 (IQR 8.3-13.3), and in F4 18.4 (IQR 12.1-25.0, P < 0.001).

Table 2 shows the univariate and multivariate analyses of parameters associated with LSM as continuous variables. Notably, steatosis grade was an independent predictor of higher LSM (P = 0.03) even after correction for clinicometabolic and histological variables, as well as for quality criteria such as biopsy length and IQR/M value of LSM.

Receiver operating characteristic curves identified the best cutoffs of LSM that are able to maximize the accuracy of TE at 6.9 kPa for the diagnosis of significant fibrosis (area under the curve = 0.784, sensitivity 70.8%, specificity 68.6%) and at 8.4 kPa for severe fibrosis (area under the curve = 0.866, sensitivity 77.4%, specificity 76%). At the LSM cutoff of 6.9 kPa for significant fibrosis, the false-positive rate was 31.4% (44/140) and the false-negative rate was 29.2% (33/113). Concerning the performance of LSM >8.4 kPa for severe fibrosis, the false-positive rate was 24% (48/200) and the false-negative rate was 22.6% (12/53). The accuracy of LSM in diagnosing significant and severe fibrosis in the entire NAFLD cohort is summarized (Fig. 1).

Impact of Severe Steatosis Detected by Histology or by BLEP on LSM Values and on Accuracy of TE

Due to the independent association of liver stiffness with steatosis, we assessed the variations of LSM values and of TE accuracy within the same stages of liver fibrosis and according to the presence or absence of severe steatosis (≥66%) on histological examination or on severe BLEP detection on US, a widely used, if not optimal, surrogate of steatosis.

Supporting Figs. 1 and 2 show that when patients were considered at each fibrosis stage, for lower grades of fibrosis only, LSM values were significantly higher in patients with severe steatosis by histology or by US compared to those without. Along this line, similar results were observed when discriminating patients according to the presence or absence of significant (F2-F4) or severe (F3-F4) fibrosis. Specifically, Fig. 2 shows, in subgroups of patients with and without severe steatosis, median LSM values among patients with F0-F1 fibrosis stages (6.9 versus 5.8, P = 0.04; Fig. 2A) and F2-F4 fibrosis stages (9.0 versus 8.8, P = 0.70; Fig. 2B) and among patients with F0-F2 fibrosis stages (7.4 versus 6.0, P = 0.001; Fig. 2c) and F3-F4 fibrosis stages (11.9 versus 12.0, P = 0.70; Fig. 2d). As expected, similar results were obtained when the entire cohort was split according to the presence or absence of severe BLEP on US (Fig. 3).

To further strengthen the independent effect of steatosis on LSM, Table 3 depicts the LSM value distribution among patients with F0-F1 fibrosis stages and F0-F2 fibrosis stages and the same inflammation grade and/or hepatocyte ballooning presence, according to the presence or absence of severe steatosis and of severe BLEP. Overall, the presence of severe steatosis and its US detection remained significantly associated with higher LSM values in almost all these subgroups.

When considering the impact of steatosis on TE accuracy, a higher rate of false-positive LSM results for the diagnosis of significant fibrosis (F2-F4) was found among patients with steatosis ≥66% (17/72, 23.6%) compared to those without (27/181, 14.9%) (Fig. 4A) and in patients with detection of severe BLEP (16/72, 22.2%) compared to those without (28/181, 15.4%) (Fig. 4B). Similar differences in false-positive LSM results for the diagnosis of severe (F3-F4) fibrosis were observed (Fig. 4C,D). When comparing the false-positive rates among patients with F0-F1 or F0-F2 fibrosis, discriminated according to the presence or absence of steatosis ≥66%/severe BLEP, we obtained again higher false-positive rates in patients with severe steatosis/BLEP (Supporting Fig. 3).

Conversely, the impact of severe steatosis and of severe BLEP on false-negative LSM results was negligible (data not shown).

Table 4 summarizes the best cutoffs of LSM and their operating characteristics at different grades of histological or US steatosis.