Detecting Hepatic Steatosis Based on Transient Elastography
Detecting Hepatic Steatosis Based on Transient Elastography
Patients for this study were recruited from a prospective, multicenter trial comparing the utility of the FibroScan® M and XL probes for the staging of liver fibrosis in overweight and obese patients. Adults (≥18 years) with chronic liver disease of any aetiology and a body mass index (BMI) ≥28 kg/m who had undergone percutaneous liver biopsy within 6 months, or were scheduled to undergo biopsy within 1 month, were eligible. Patients were recruited from five Canadian hepatology centres between July 2009 and July 2010. Exclusion criteria were: (1) contraindications to LSM (e.g. pregnancy, ascites, etc.); (2) BMI <28 kg/m; (3) previous liver transplant; (4) malignancy or other terminal disease; and (5) refusal to undergo a biopsy. Health Canada and the local research ethics boards approved the protocol.
Before transient elastography (TE), demographics, liver disease aetiology and anthropometric measurements (BMI and waist circumference [in cm]) were obtained. Biochemical data including liver biochemistry, platelets, fasting glucose, cholesterol and triglycerides from within 6 months of screening were recorded. The Fatty Liver Index (FLI) and Hepatic Steatosis Index (HSI), alternative noninvasive means of assessing hepatic steatosis based on routinely collected parameters, were also calculated according to the following formulas:
Experienced operators performed all FibroScan® examinations as per the manufacturer's recommendations. With the patient lying in the dorsal decubitus position, the tip of the transducer probe was placed on the skin between the ribs over the right lobe of the liver. Assisted by a sonographic image, a portion of the liver at least 6 cm thick and free of large vascular structures was identified using a portable 10 MHz ultrasound transducer (Mindray DP-6600; Mindray, Shenzhen, China). At this site, the distance between the skin and liver capsule (skin-capsular distance) was measured and an attempt was made to collect ≥10 valid LSMs. For this report, only LSMs using the M probe were considered because the CAP algorithm is specific to this device. Examinations with no successful measurements after ≥10 attempts were deemed failures. The median liver stiffness value was considered representative of the elastic modulus of the liver. As an indicator of variability, the ratio of the interquartile range (IQR) of liver stiffness to the median (IQR/MLSM) was calculated.
The CAP measures ultrasonic attenuation in the liver at 3.5 MHz using signals acquired by the FibroScan® M probe based on vibration-controlled transient elastography (VCTE™). The principles have been described elsewhere. The CAP is measured only on validated measurements according to the same criteria used for LSM and on the same signals, ensuring that one obtains a liver ultrasonic attenuation simultaneously and in the same volume of liver parenchyma as the LSM. The final CAP value, which ranges from 100 to 400 decibels per metre (dB/m), is the median of individual measurements. As an indicator of variability, the ratio of the IQR of CAP values to the median (IQR/MCAP) was calculated.
Liver biopsies were fixed, paraffin-embedded and stained with at least hematoxylin and eosin and Masson's trichrome. Two experienced hepatopathologists analysed biopsies independently without knowledge of clinical data. A consensus was reached in cases of disagreement. Steatosis was categorized according to the NAFLD Activity Score (NAS) (S0, <5%; S1, 5–33%; S2, 34–66%; and S3, >66%) and assessed as the percentage of hepatocytes containing lipid droplets. The primary outcome was the differentiation of significant (≥10%) from insignificant steatosis (<10%). The staging of fibrosis was dependent on liver disease aetiology. In patients with NAFLD, fibrosis was staged according to the classification of Kleiner et al. (F0 = no fibrosis; F1 = perisinusoidal or portal fibrosis; F2 = perisinusoidal and portal/periportal fibrosis; F3 = septal or bridging fibrosis; and F4 = cirrhosis). In the remainder, the METAVIR classification was used (F0 = no fibrosis; F1 = portal fibrosis without septa; F2 = portal fibrosis with few septa; F3 = portal fibrosis with many septa; and F4 = cirrhosis). Inflammation was graded according to the METAVIR classification in patients with viral hepatitis [20] and the lobular component of the NAS in those with NAFLD. Finally, the length of biopsy specimens and the number of portal tracts were recorded as measures of quality. Biopsies <1.5 cm in length and/or with <6 portal tracts were deemed uninterpretable.
Patient characteristics and clinical data were descriptively summarized and are reported as medians (IQR) and proportions. Between groups comparisons were made using Fisher's exact, chi-square and Mann–Whitney tests. Correlations between variables were described using Spearman correlation coefficients (ρ). To study interactions between the steatosis grade, fibrosis stage and severe inflammation with the CAP, multivariate analyses of variance (anova) were performed. The diagnostic performance of the CAP for grading steatosis compared with liver histology was determined using areas under receiver operating characteristic (AUROC) curves. Sensitivity analyses were conducted according to age, gender, liver disease, obesity, diabetes, skin-capsular distance (≤25 vs >25 mm; the measurement depth of the M probe), interval between liver biopsy and CAP measurement (< vs. ≥35 days [the median]), and the presence of moderate to severe fibrosis (F0–F1 vs. F2–F4) and inflammation (A0–A1 vs. A2–A3). Based on literature describing an increased risk of discordance between fibrosis staged histologically and as estimated by LSM in patients with highly variable LSM (i.e. high IQR/MLSM), sensitivity analyses examined the influence of IQR/MCAP on CAP performance. Finally, the performance of CAP compared with the FLI and HSI were compared using the method of DeLong et al.
To determine the ability of the CAP to differentiate between individual steatosis grades, we also calculated AUROCs between two steatosis grades only (e.g. S0 vs. S1, excluding patients with S2–S3). In addition, the Obuchowski measure was assessed taking into account the distribution of steatosis grades in the cohort. The Obuchowski measure is a multinomial version of the AUROC designed for situations characterized by a non-binary reference standard. It can be interpreted as the probability that the CAP will correctly rank two randomly sampled patients with different steatosis grades. For this analysis, weighting was based on the relative distribution of steatosis grades in the cohort. Moreover, a penalty function was applied to adjust for the 'distance' between steatosis grades under comparison, as described by Lambert et al.
Finally, we calculated the sensitivity, specificity, positive (PPV) and negative predictive values (NPV), and accuracy of the CAP at optimal thresholds defined by the maximal sum of sensitivity and specificity. All statistical analyses were performed using Stata v11.0 (StataCorp; College Station, TX, USA) and R software (R Development Core Team 2008). Two-sided P-values <0.05 were considered significant.
Patients and Methods
Study Population
Patients for this study were recruited from a prospective, multicenter trial comparing the utility of the FibroScan® M and XL probes for the staging of liver fibrosis in overweight and obese patients. Adults (≥18 years) with chronic liver disease of any aetiology and a body mass index (BMI) ≥28 kg/m who had undergone percutaneous liver biopsy within 6 months, or were scheduled to undergo biopsy within 1 month, were eligible. Patients were recruited from five Canadian hepatology centres between July 2009 and July 2010. Exclusion criteria were: (1) contraindications to LSM (e.g. pregnancy, ascites, etc.); (2) BMI <28 kg/m; (3) previous liver transplant; (4) malignancy or other terminal disease; and (5) refusal to undergo a biopsy. Health Canada and the local research ethics boards approved the protocol.
Clinical Data
Before transient elastography (TE), demographics, liver disease aetiology and anthropometric measurements (BMI and waist circumference [in cm]) were obtained. Biochemical data including liver biochemistry, platelets, fasting glucose, cholesterol and triglycerides from within 6 months of screening were recorded. The Fatty Liver Index (FLI) and Hepatic Steatosis Index (HSI), alternative noninvasive means of assessing hepatic steatosis based on routinely collected parameters, were also calculated according to the following formulas:
Liver Stiffness Measurement
Experienced operators performed all FibroScan® examinations as per the manufacturer's recommendations. With the patient lying in the dorsal decubitus position, the tip of the transducer probe was placed on the skin between the ribs over the right lobe of the liver. Assisted by a sonographic image, a portion of the liver at least 6 cm thick and free of large vascular structures was identified using a portable 10 MHz ultrasound transducer (Mindray DP-6600; Mindray, Shenzhen, China). At this site, the distance between the skin and liver capsule (skin-capsular distance) was measured and an attempt was made to collect ≥10 valid LSMs. For this report, only LSMs using the M probe were considered because the CAP algorithm is specific to this device. Examinations with no successful measurements after ≥10 attempts were deemed failures. The median liver stiffness value was considered representative of the elastic modulus of the liver. As an indicator of variability, the ratio of the interquartile range (IQR) of liver stiffness to the median (IQR/MLSM) was calculated.
Controlled Attenuation Parameter (CAP) Measurement
The CAP measures ultrasonic attenuation in the liver at 3.5 MHz using signals acquired by the FibroScan® M probe based on vibration-controlled transient elastography (VCTE™). The principles have been described elsewhere. The CAP is measured only on validated measurements according to the same criteria used for LSM and on the same signals, ensuring that one obtains a liver ultrasonic attenuation simultaneously and in the same volume of liver parenchyma as the LSM. The final CAP value, which ranges from 100 to 400 decibels per metre (dB/m), is the median of individual measurements. As an indicator of variability, the ratio of the IQR of CAP values to the median (IQR/MCAP) was calculated.
Liver Biopsy
Liver biopsies were fixed, paraffin-embedded and stained with at least hematoxylin and eosin and Masson's trichrome. Two experienced hepatopathologists analysed biopsies independently without knowledge of clinical data. A consensus was reached in cases of disagreement. Steatosis was categorized according to the NAFLD Activity Score (NAS) (S0, <5%; S1, 5–33%; S2, 34–66%; and S3, >66%) and assessed as the percentage of hepatocytes containing lipid droplets. The primary outcome was the differentiation of significant (≥10%) from insignificant steatosis (<10%). The staging of fibrosis was dependent on liver disease aetiology. In patients with NAFLD, fibrosis was staged according to the classification of Kleiner et al. (F0 = no fibrosis; F1 = perisinusoidal or portal fibrosis; F2 = perisinusoidal and portal/periportal fibrosis; F3 = septal or bridging fibrosis; and F4 = cirrhosis). In the remainder, the METAVIR classification was used (F0 = no fibrosis; F1 = portal fibrosis without septa; F2 = portal fibrosis with few septa; F3 = portal fibrosis with many septa; and F4 = cirrhosis). Inflammation was graded according to the METAVIR classification in patients with viral hepatitis [20] and the lobular component of the NAS in those with NAFLD. Finally, the length of biopsy specimens and the number of portal tracts were recorded as measures of quality. Biopsies <1.5 cm in length and/or with <6 portal tracts were deemed uninterpretable.
Statistical Analyses
Patient characteristics and clinical data were descriptively summarized and are reported as medians (IQR) and proportions. Between groups comparisons were made using Fisher's exact, chi-square and Mann–Whitney tests. Correlations between variables were described using Spearman correlation coefficients (ρ). To study interactions between the steatosis grade, fibrosis stage and severe inflammation with the CAP, multivariate analyses of variance (anova) were performed. The diagnostic performance of the CAP for grading steatosis compared with liver histology was determined using areas under receiver operating characteristic (AUROC) curves. Sensitivity analyses were conducted according to age, gender, liver disease, obesity, diabetes, skin-capsular distance (≤25 vs >25 mm; the measurement depth of the M probe), interval between liver biopsy and CAP measurement (< vs. ≥35 days [the median]), and the presence of moderate to severe fibrosis (F0–F1 vs. F2–F4) and inflammation (A0–A1 vs. A2–A3). Based on literature describing an increased risk of discordance between fibrosis staged histologically and as estimated by LSM in patients with highly variable LSM (i.e. high IQR/MLSM), sensitivity analyses examined the influence of IQR/MCAP on CAP performance. Finally, the performance of CAP compared with the FLI and HSI were compared using the method of DeLong et al.
To determine the ability of the CAP to differentiate between individual steatosis grades, we also calculated AUROCs between two steatosis grades only (e.g. S0 vs. S1, excluding patients with S2–S3). In addition, the Obuchowski measure was assessed taking into account the distribution of steatosis grades in the cohort. The Obuchowski measure is a multinomial version of the AUROC designed for situations characterized by a non-binary reference standard. It can be interpreted as the probability that the CAP will correctly rank two randomly sampled patients with different steatosis grades. For this analysis, weighting was based on the relative distribution of steatosis grades in the cohort. Moreover, a penalty function was applied to adjust for the 'distance' between steatosis grades under comparison, as described by Lambert et al.
Finally, we calculated the sensitivity, specificity, positive (PPV) and negative predictive values (NPV), and accuracy of the CAP at optimal thresholds defined by the maximal sum of sensitivity and specificity. All statistical analyses were performed using Stata v11.0 (StataCorp; College Station, TX, USA) and R software (R Development Core Team 2008). Two-sided P-values <0.05 were considered significant.
Source...