Study participants were consecutively and prospectively recruited from the University Hospital of Clermont-Ferrand, France, from January 2021 to October 2021 by physicians of the hepatology department. Inclusion criteria were: Age over 18 years, patients with known or suspected NAFLD, referred for non-invasive fibrosis evaluation using impulse elastography, with willingness and ability to participate. Exclusion criteria were clinical, laboratory, or histologic evidence of a liver disease other than NAFLD (chronic hepatitis B or C, autoimmune liver disease, excessive alcohol consumption defined by WHO criteria, Wilson's disease or other), hepatocellular carcinoma and all other liver tumors, secondary causes of liver steatosis (genetic disease, steatogenic or hepatoxic medication use) and contraindications to MRI.

Demographic and anthropometric data were recorded. Biological data collected included platelets, prothrombin time (PT), international normalized ratio (INR), albumin, bilirubin, transaminases (aspartate aminotransferase, AST and alanine aminotransferase, ALT), gamma-glutamyltransferase (GGT), alkaline phosphatase (ALP), total cholesterol, high density lipoprotein (HDL) cholesterol, low density lipoprotein (LDL) cholesterol, triglycerides, fasting glucose, fasting insulinemia.

The study design was approved by local ethic committee (study M210401, Clermont-Ferrand University Hospital) and informed consent was obtained for all participants.

All participants underwent US liver examinations using the Aixplorer MACH 30^® US system (Supersonic Imagine, Aix-en-Provence, France). Participants were asked to fast for at least 3 h prior to US examinations. They were positioned in the dorsal decubitus position with the right arm at maximum abduction for intercostal stretching. A 3.5 MHz abdominal curved transducer (C6-1X probe) was used. Acquisitions were performed during neutral respiratory apnea as follows: (1) Right intercostal (between the 7th and 9th intercostal space) and subcostal view was considered for every patient; (2) US transducer was placed so that the hepatic capsule was parallel to the US transducer; and (3) Care was taken to avoid the presence of large hepatic vessels or artifacts in the image (ultrasound reflection caused by abdominal gas or by rib interposition). For HRI, the probe was placed to obtain an adequate visualization of the liver and the right kidney in a sagittal or oblique image with minimal artifacts. HRI was calculated using the region of interest (ROI) measure tool, with average brightness ratio between two ROI at least 3 mm wide placed at the same depth in hepatic parenchyma and in renal cortex (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Hepato-renal index measurement with 6 mm wide regions of interest, at a depth of 9.4 cm in liver right lobe and right kidney, University Hospital of Clermont-Ferrand.

For each patient, 5 measures of sound speed, 5 of attenuation coefficient (AC) and 3 of HRI acquisitions were performed by the same physician. The mean values were used for statistical analysis.

Simultaneously, liver stiffness and cCAP measurements were performed using FibroScan SmartExam^® (Echosens, Paris, France) by experienced operators. All patients were measured using either an M- or an XL-probe, according to the device automatically-selected probe.

A liver US was also performed by a radiologist at patient’s discretion, in any radiological center, unaware of the study, in the standard clinical care of NAFLD. The presence of a hyperechogenic liver described on the US report was noted (steatosis vs. no steatosis or not mentioned).

Within the month following US examinations, all participants underwent a chemical shift-encoded liver MRI using a 3.0 Tesla MAGNETOM Vida^® system (Siemens Medical Solutions, Erlangen, Germany). A gradient echo dedicated sequence developed by the manufacturer to measure hepatic PDFF was systematically included in MRI acquisition protocols. This sequence is characterized by a low flip angle to reduce T1 bias and six echoes to correct for T2* effect. Images were acquired during a single breath hold. In and out-phases imaging was used prior to fat quantification to assess homogeneity of fat distribution. PDFF (%) estimation was obtained by placing three large regions of interest in the liver parenchyma. Steatosis was regarded as fat fraction ≥ 5.6 %, as defined by EASL. Steatosis grading was not evaluated because correspondence between the steatosis histologic grading and PDFF value has not yet been standardized in the literature[27]. Iron content was also measured using this technique.

Patients data were recorded in e-case report form via a secure web platform (REDCap^® version 9.3.7, Vanderbilt University, United States) where these data were monitored and a database extraction was performed.

Data were expressed as descriptive statistics (mean ± SD, median with IQR or number with corresponding percentage as appropriate). The normality assumption was assessed using the Shapiro-Wilk's test. Chi-Square and Student t-test were used to compare categorical and continuous variables. If normality was not ascertained, the Mann-Whitney test was used for continuous variables. Receiver operating curve (ROC) analysis was performed to evaluate the diagnostic performance of US tools compared to MRI PDFF. Optimal cutoff values were identified by maximizing the Youden index, and corresponding sensitivities, specificities, likelihood ratios (LR) were derived. Comparison of areas under receiver operating characteristic curves was made according to Hanley and McNeil. Pearson and Spearman linear correlation coefficients were used to evaluate the relationship between continuous variables, respectively when normality was assessed or not. A non-linear regression analysis was also conducted to assess the relationship between SSE AC, cCAP, HRI and PDFF. All tests were considered statistically significant in the case of P < 0.05. Statistical analysis were performed and graphs were designed using GraphPad Prism^® (v8, GraphPad Software, La Jolla, California, United States). Multivariable linear regression was used to evaluate the influence of anthropometric parameters and fibrosis severity on US measures (R Software, R Foundation for Statistical Computing, Vienna, Austria).

One hundred and five patients were included in this study. Main characteristics are presented in Table 1. Median BMI was 31 kg/m². Most patients were overweight or obese (90%) and had metabolic syndrome (70%). One third suffered from diabetes. Steatosis was identified in 85 patients (81%) according to PDFF. Twenty-five patients (20%) were suspected to have advanced liver disease according to shear-wave elastography.

Mean AST and ALT were respectively 35 (± 28) U/L and 65 (± 49) U/L. Mean GGT was 120 (± 125) U/mL. Only ALT levels correlated significantly but moderately with PDFF (Spearman's ρ = 0.4, P < 0.001). One patient failed to receive Fibroscan^® and HRI measurement due to poor US signal.

Patients characteristics according to the presence of steatosis are shown in Table 2. Steatotic patients had significantly higher BMI and waist circumference. As expected, metabolic syndrome was more prevalent in patients with steatosis. Mean SSE, AC and cCAP values significantly differed between non-steatotic and steatotic patients, but not liver stiffness.

SSE, AC, HRI and cCAP correlated with PDFF, with respective Spearman correlation coefficient of -0.39, 0.42, 0.59 and 0.54 (P < 0.01) (Figure 2). SSE, AC, cCAP and HRI specifically displayed a significant non-linear relationship with PDFF as previously demonstrated (Supplementary Figure 1)[28].

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Scatterplots. A-D: Scatterplots showing linear relationship between sound speed estimation (A), attenuation coefficient (B), hepato-renal index (C) and continuous controlled attenuation parameter (D) with magnetic resonance imaging-proton density fat fraction using 6-echo gradient (P < 0.001). PDFF: Proton density fat fraction; cCAP: Continuous controlled attenuation parameter; SSE: Sound speed estimation; AC: Attenuation coefficient; HRI: Hepato-renal index.

Intercostal SSE performed moderately for the diagnosis of steatosis with an area under curve (AUC) of 0.73 [0.62-0.84] (P = 0.001) (Figure 3A). Using the manufacturer cutoff of 1537 m/s resulted in a sensitivity and specificity of 80% and 45%. The best cutoff identified in our work was 1518 m/s but performances remained modest with sensitivity and specificity of 60% and 80% respectively. Of note, using 5 measures of SSE compared to 3 measures did not improve performances (data not shown).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Receiver operating curve. A-D: Receiver operating curve for sound speed estimation (A), attenuation coefficient (B), continuous controlled attenuation parameter (C) and hepato-renal index (D). PDFF: Proton density fat fraction; cCAP: Continuous controlled attenuation parameter; SSE: Sound speed estimation; AC: Attenuation coefficient; HRI: Hepato-renal index.

Intercostal AC performed well for the diagnosis of steatosis with an AUC of 0.82 (0.70-0.93) (P = 0.001) (Figure 3B). The best cutoff identified in our work was 0.42 dB/cm/MHz with sensitivity and specificity of 84 % and 75 % respectively.

cCAP was accurate for the diagnosis of steatosis with AUC of 0.79 (0.66-0.92) but did not perform better than AC. The best cutoff identified in our work was 275 dB/m but lacked sensitivity and specificity (respectively 72% and 80%) (Figure 3C).

As expected, using cCAP resulted in low coefficient variation with 96% of patients having a coefficient of less than 10%. The diagnostic accuracy of cCAP was significantly improved when SD of cCAP was < 15 dB/m with an AUC of 0.91 (95% confidence interval: 0.83-0.98, P < 0.01) resulting in a sensitivity and specificity of 74% and 92%. In this work, one third of patients had an SD > 15 dB/m (Figure 4).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Continuous controlled attenuation parameter area under the receiver operating characteristic curve for the diagnosis of steatosis according to SD. PDFF: Proton density fat fraction; AUROC: Area under the receiver operating characteristic curve.

HRI performed the best with an AUC of 0.91 (0.83-0.99) (P < 0.0001) (Figure 3D). Using a cutoff set at 1.3 resulted in a sensitivity of 82% and specificity of 98%. In other words, only 2 patients had an increased hepatorenal gradient without having liver steatosis. The first patient had borderline steatosis (4% on MRI) and severe fibrosis on elastography (19 KPa) which could explain the increased HRI. For the other patient, moderate iron overload may have increased hepatorenal gradient. Among the 15 false negative patients, two third had mild steatosis on MRI (defined as PDFF < 10%).

Hyperechogenic liver was described in 73/85 steatotic patients (Figure 5). It showed a sensitivity of 86% and specificity of 70%. Among the 6 patients presenting a hyperechogenic liver on US but no steatosis on MRI, none had severe fibrosis (assessed using Fibroscan) or iron overload. Among the 12 patients with no hyperechogenic liver but steatosis on MRI, 7/12 (58%) had mild steatosis based on PDFF. No other steatosis sign was described by radiologists.

Open in New Tab   Full Size Figure   Download Figure

Figure 5 Ultrasound finding according to liver steatosis based on proton density fat fraction.

SSE and AC measurements were made for every patient using intercostal and subcostal view. However, if intercostal access was successful in every patient (success rate: 100% for SSE and AC), subcostal access showed a lower success rate (94% for SSE, 92% for AC) and lower performances (data not shown).

Using multivariable linear regression, we demonstrated that both HRI, cCAP, AC and SSE were associated with the presence of steatosis on MRI, independently from BMI, waist circumference elastography, iron overload and the presence of diabetes or metabolic syndrome.

Non-alcoholic fatty liver disease (NAFLD) is the most common liver disease worldwide, ranging from simple steatosis to aggressive hepatitis leading to liver fibrosis, cirrhosis and hepatocellular carcinoma.

Diagnose fatty liver disease and assess its severity during follow-up and after treatment is a key in clinical practice. Liver biopsy can deliver this information, but it is an invasive procedure with potentially severe. Therefore, non-invasive techniques were developed to stage steatosis. Ultrasound is the primary imaging modality in the assessment of patients with confirmed or suspected NAFLD.

We wanted to evaluate new ultrasonographic tools to detect and measure hepatic steatosis.

One hundred and five patients underwent ultrasonographic measurement of liver sound speed estimation (SSE) and attenuation coefficient (AC) using Aixplorer MACH 30 (Supersonic Imagine, France), continuous Controlled Attenuation Parameter (cCAP) using Fibroscan (Echosens, France) and standard liver ultrasound with hepato-renal index (HRI) calculation. Hepatic steatosis was then classified according to MRI proton density fat fraction (PDFF) as gold standard.

SSE, AC, cCAP and HRI correlated with PDFF, with respective Spearman correlation coefficient of -0.39, 0.42, 0.54 and 0.59 (P < 0.01). Area under the receiver operating characteristic curve (AUROC) for detection of steatosis with HRI was 0.91 (0.83-0.99), with the best cut-off value being 1.3 (Se = 83%, Sp = 98%). The optimal cCAP threshold of 275 dB/m, corresponding to the recent EASL-suggested threshold, had a sensitivity of 72% and a specificity of 80%. Corresponding AUROC was 0.79 (0.66-0.92). The diagnostic accuracy of cCAP was more reliable when standard deviation was < 15 dB/m with an AUC of 0.91 (0.83-0.98). An AC threshold of 0.42 dB/cm/MHz had an AUROC was 0.82 (0.70-0.93). SSE performed moderately with an AUROC of 0.73 (0.62-0.84).

HRI had the best performance. It is also the simplest and most available method as most US scans are equipped with this module.

Measurement quality criteria need to be defined and validated for a wider use of theses techniques. Their improvement could open the way to efficient and easily accessible non-invasive steatosis grading.