Dr. Myers is supported by a Clinical Investigator Award from the Alberta Heritage Foundation for Medical Research (now Alberta Innovates – Health Solutions) and New Investigator Award from the Canadian Institutes for Health Research. The study was sponsored by Echosens (Paris, France).
Potential conflict of interest: Dr. Myers consults for GE Healthcare. He is in the speakers' bureau of HNS Canada and received grants from Echosens. Dr. Pomier-Layrargues received grants from Echosens and Bristol-Myers Squibb.
Failure of liver stiffness measurement (LSM) by transient elastography (TE, FibroScan) and unreliable results occur in ≈5% and 15% of patients, respectively, mainly due to obesity. In this multicenter study, we evaluated the feasibility and performance of the novel FibroScan XL probe in 276 patients with chronic liver disease (42% viral hepatitis, 46% nonalcoholic fatty liver disease [NAFLD]) and a body mass index (BMI) ≥28 kg/m2. Patients underwent liver biopsy and TE with the standard M and XL probes. TE failure was defined as no valid LSMs and unreliable examinations as <10 valid LSMs or an interquartile range (IQR)/LSM >30% or success rate <60%. Probe performance for diagnosing ≥F2 fibrosis and cirrhosis (F4) versus biopsy were examined using areas under receiver operating characteristic curves (AUROC). FibroScan failure was less frequent with the XL probe than the M probe (1.1% versus 16%) and the XL probe was more often reliable (73% versus 50%; both P < 0.00005). Reliable results with the XL probe were obtained in 61% of patients in whom the M probe was unreliable. Among 178 patients with ≥10 valid LSMs using both probes, liver stiffness was highly correlated between probes (ρ = 0.86; P < 0.0005); however, median liver stiffness was lower using the XL probe (6.8 versus 7.8 kPa; P < 0.00005). The AUROC of the XL and M probes were similar for ≥F2 fibrosis (0.83 versus 0.86; P = 0.19) and cirrhosis (0.94 versus 0.91; P = 0.28). Conclusion: Compared with the M probe, the FibroScan XL probe reduces TE failure and facilitates reliable LSM in obese patients. Although the probes have comparable accuracy, lower liver stiffness cutoffs will be necessary when the XL probe is used to noninvasively assess liver fibrosis. (HEPATOLOGY 2012)
Transient elastography (TE) using the FibroScan (Echosens; Paris, France) is a noninvasive method for measuring liver stiffness as a surrogate of liver fibrosis.1, 2 First studied in patients with hepatitis C,3, 4 TE has now been validated in populations with various liver disorders and the technology has gained widespread use in many regions.5, 6 The diagnostic performance of TE is excellent for cirrhosis and moderate for significant fibrosis.5, 7, 8 Advantages of TE include its simplicity, short performance time, immediate results, patient acceptance, and ease of incorporation into an outpatient clinical setting. A disadvantage of TE is the inability to accurately assess liver stiffness in some patients, predominantly due to obesity. Because subcutaneous fat attenuates the transmission of shear waves into the liver and the ultrasonic signals used to measure their speed of propagation, FibroScan failure (i.e., no valid measurements) and unreliable results occur in ≈3%-5% and 10%-15% of patients, respectively.6, 9-13 Numerous studies have shown that obesity, defined as a body mass index (BMI) ≥30 kg/m2, is the strongest predictor of failed or unreliable liver stiffness measurement (LSM).6, 9, 12, 13 Moreover, subcutaneous adipose tissue may lead to overestimation of liver stiffness. Due to the rising prevalence of obesity and associated nonalcoholic fatty liver disease (NAFLD),14 this limitation is a potentially important barrier to the effective use of TE in clinical practice. To mitigate this limitation, a new FibroScan probe—designated the “XL” probe—has been designed specifically for use in obese patients. The XL probe differs from the standard M probe by its utilization of a lower frequency and more sensitive ultrasonic transducer, a deeper focal length, a larger vibration amplitude, and a greater depth of measurement below the skin surface. Preliminary data suggest that the XL probe improves the feasibility of LSM in obese patients; however, histological data confirming its diagnostic accuracy are limited.15, 16
The objectives of this prospective, multicenter study were to compare the feasibility and reliability of the XL and M probes for LSM in overweight and obese patients with various liver disorders. In addition, the diagnostic accuracy of the two probes was compared using liver biopsy as the reference standard.
AUROC, area under the receiver operating characteristic curve; CI, confidence interval; IQR, interquartile range; IQR/M, IQR over the median; LSM, liver stiffness measurement; NAFLD, nonalcoholic fatty liver disease; NAS, NAFLD Activity Score; OR, odds ratio; TE, transient elastography.
Patients and Methods
In this prospective study, adults (≥18 years) with chronic liver disease of any etiology and a BMI ≥28 kg/m2 who had undergone percutaneous liver biopsy within 6 months, or were scheduled to undergo biopsy within 1 month, were eligible. Patients were recruited from four Canadian academic centers (the University of Calgary, the University of Western Ontario, the Toronto Western Hospital Liver Centre, and Hôpital St. Luc in Montréal, Quebec) and a community-based hepatology clinic (the Liver Centre in Toronto, Ontario) between July 2009 and July 2010. Patients meeting any of the following criteria were ineligible: (1) contraindications to LSM (e.g., pregnancy, ascites, implantable cardiac devices, etc.); (2) BMI <28 kg/m2; (3) previous liver transplant; (4) known malignancy or other terminal disease; and (5) refusal to undergo a liver biopsy. Health Canada and the research ethics boards of the participating institutions approved the study protocol (clinicaltrials.gov, NCT 00926224). The study sponsor (Echosens; Paris, France) oversaw data collection and monitoring, but had no role in data analysis, drafting of the article, or in the decision to submit the article for publication.
Before TE examination, demographic details, etiology of liver disease, and anthropometric measurements (weight, height, BMI, waist circumference, and thoracic perimeter measured at the xiphoid process) were obtained. Biochemical data including liver biochemistry, platelets, albumin, and fasting glucose, cholesterol, and triglycerides from within 6 months of screening were recorded. Presence of the metabolic syndrome was defined according to guidelines of the American Heart Association and National Heart, Lung, and Blood Institute.17
Liver Stiffness Measurement.
Nine experienced operators at the five study sites performed all FibroScan examinations as per the manufacturer's recommendations. All operators had completed at least 50 prior exams (four had performed >500 exams; one had >200; three had >100; and one had >50). Briefly, 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 at least 10 valid measurements with each of the M and XL probes. Specific differences between the M and XL probes include their central ultrasound frequency (3.5 versus 2.5 MHz), vibration amplitude (2 versus 3 mm), diameter of their tips (9 versus 12 mm), and measurement depth from the skin surface (25-65 versus 35-75 mm).15 The manufacturer recommends that the XL probe be used in patients with a skin-capsular distance ≥25 mm. Examinations with no successful measurements after at least 10 attempts were deemed failures. The median liver stiffness value (in kPa) 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 value (IQR/M) was calculated. Examinations with fewer than 10 valid measurements or an IQR/M >30% or a success rate <60% were considered potentially unreliable.
Liver biopsy tissue, obtained under ultrasound guidance, was fixed, paraffin-embedded, and stained with at least hematoxylin and eosin and Masson's trichrome. Two experienced hepatopathologists analyzed biopsy specimens independently without knowledge of liver stiffness results or other clinical data. In cases of disagreement, a consensus was reached by joint discussion. In patients with NAFLD, fibrosis was staged according to the classification of Kleiner et al.18 (F0 = no fibrosis; F1 = perisinusoidal or portal fibrosis; F2 = perisinusoidal and portal/periportal fibrosis; F3 = septal or bridging fibrosis; and F4 = cirrhosis). In patients with all other liver conditions, including chronic viral hepatitis (with or without coexistent steatosis), fibrosis was staged according to the METAVIR classification (F0 = no fibrosis; F1 = portal fibrosis without septa; F2 = portal fibrosis with few septa; F3 = portal fibrosis with many septa; and F4 = cirrhosis). Steatosis was graded according to the NAFLD Activity Score (NAS: S0, <5% of hepatocytes affected; S1, 5%-33%; S2, 34%-66%; and S3, >66%).18 The severity of hepatic inflammation was graded according to the METAVIR classification in patients with viral hepatitis19 and the lobular component of the NAS in those with NAFLD.18 Finally, the length of biopsy specimens and the number of portal tracts sampled were recorded as measures of biopsy quality. Biopsies less than 15 mm in length and/or with fewer than six portal triads were deemed uninterpretable.
Patient characteristics and clinical data were descriptively summarized and are reported as medians (IQR) and proportions. Between-group comparisons were made using Fisher's exact and chi-square tests for categorical variables, and Mann-Whitney and Wilcoxon matched pairs sign-rank tests for continuous variables. Correlations between variables, including liver stiffness measured using the M and XL probes, were described using Spearman correlation coefficients (ρ). Further analysis of the agreement between probes was assessed using Bland-Altman plots of the intraindividual differences in LSM with each probe versus the mean measurement.20 To identify independent predictors of reliable FibroScan examinations, multivariate logistic regression models including age, gender, liver disease etiology, BMI (categorized as <30, 30 to <35, 35 to <40, and ≥40 kg/m2), skin-capsular distance (categorized as <25 and ≥25 mm), diabetes mellitus, and moderate to severe hepatic steatosis were analyzed. Separate disease-specific models determined the influence of moderate to severe hepatic inflammation on FibroScan reliability.
The diagnostic performances of the M and XL probes compared with liver histology as the reference standard were determined using areas under receiver operating characteristic (AUROC) curves.21 The diagnostic utilities for significant fibrosis (≥F2), severe fibrosis (≥F3), and cirrhosis (F4) were examined, both overall and according to the main etiologies (viral hepatitis and NAFLD). Statistical comparisons of probe performance were limited to patients with ≥10 valid measurements with both probes; AUROCs were compared using the method of DeLong et al.22 We also calculated the sensitivity, specificity, and positive (PPV) and negative predictive values (NPV) of the FibroScan with each probe. For these analyses, optimal liver stiffness cutoffs that maximized the sum of sensitivity and specificity were determined overall and within specific disease categories. All statistical analyses were performed using SAS v. 9.2 (SAS Institute, Cary, NC) and Stata v. 11.0 (StataCorp, College Station, TX). Two-sided P-values less than 0.05 were considered statistically significant.
Between July 2009 and July 2010, 306 patients were screened for the study at the five participating centers. Thirty patients were excluded due to withdrawal of consent (n = 2) and refusal to undergo liver biopsy following LSM (n = 28). The characteristics of the remaining 276 patients are outlined in Table 1. The majority (63%) was male and the median age was 50 years (interquartile range [IQR] 43-57). Forty-two percent of patients had chronic hepatitis B and/or C (32% with coexistent steatosis) and 46% had NAFLD. The prevalence of diabetes mellitus was 24% (viral 16%, NAFLD 33%, other 21%) and 33% had moderate to severe (>33%) steatosis. The median BMI was 30 kg/m2 (IQR 29-33; range 28-53); 15% of patients were extremely obese (BMI ≥40 kg/m2). The median skin-capsular distance was 22 mm. The skin-capsular distance was <25 mm in 68% of patients, 25 to 34 mm in 27%, and ≥35 mm in 5%. BMI was moderately correlated with the skin-capsular distance (ρ = 0.51) and thoracic perimeter (ρ = 0.53), as were the latter variables together (ρ = 0.47; all P < 0.0005).
Table 1. Characteristics of the Study Cohort (n=276)
Median (IQR) or Proportion (%, n)
Complete data for determination of presence of the metabolic syndrome available in 235 patients.
Skin-capsular distance missing in five patients.
Numbers in parentheses indicate the number of patients with complete laboratory data.
Feasibility and Reliability of LSM with the FibroScan M and XL Probes.
Table 2 compares the feasibility of LSM between the M and XL probes. The XL probe nearly eliminated FibroScan failure (i.e., no valid measurements: 1.1% versus 16% with the M probe; P < 0.00005). Success with the XL probe was consistent across BMI categories (P = 0.17; Fig. 1) and skin-capsular distance (<25 versus ≥25 mm: failure in 0.5% versus 2.3%; P = 0.24). On the contrary, failure of the M probe increased markedly as the BMI increased (P < 0.0005) and in patients with skin capsular distance ≥25 mm (versus <25 mm: 33% versus 9%; P < 0.0005). As illustrated in Fig. 2, among the 44 patients (16%) in whom the M probe failed, the XL probe successfully measured liver stiffness in 42 (95%). The XL probe failed in only one patient (0.4%) in whom the M probe was successful (skin-capsular distance 23 mm).
Table 2. Feasibility and Performance of the FibroScan M and XL Probes
For 178 patients with ≥10 valid measurements with both probes.
The XL probe was also significantly more likely than the M probe to obtain ≥10 valid LSMs (93% versus 65%; both P < 0.00005). The proportions of patients with ≥10 valid measurements according to BMI category and probe is illustrated in Fig. 3, along with the proportion with a skin-capsular distance <25 mm. The two probes obtained ≥10 valid measurements in a similar proportion of nonobese patients (BMI<30 kg/m2); the difference between probes was most evident in higher BMI categories. The ability to obtain ≥10 valid measurements using the M probe paralleled the prevalence of a skin-capsular distance <25 mm, which decreased in frequency at higher BMI categories. On the contrary, success with the XL probe was largely independent of BMI, except in the extremely obese (BMI ≥40 kg/m2), in whom 10 valid measurements were obtained in 71% of patients (versus 95%-100% with BMI <40 kg/m2).
Variability between LSMs, as assessed by the ratio of IQR/M, was not significantly different between the M and XL probes (P = 0.65; Table 2). However, the XL probe was more likely to provide a reliable assessment of liver stiffness, as defined by ≥10 valid measurements, an IQR/M ≤30%, and a success rate ≥60% (73% versus 50%; P < 0.00005). As illustrated in Fig. 4, among the 138 patients (50%) in whom the M probe was unreliable, the XL probe obtained reliable results in 84 (61%).
Table 3 includes the results of multivariate analyses evaluating factors associated with reliable LSMs using the M and XL probes. Age, sex, liver disease etiology, and moderate to severe hepatic steatosis (>33%) were not significant predictors with either probe. For the M probe, reliable LSMs were less likely with a skin-capsular distance ≥25 mm and BMI >35 kg/m2. For the XL probe, reliable measurements were less likely in patients with a BMI ≥40 kg/m2 and those with diabetes mellitus. In supplementary analyses that included the presence of the metabolic syndrome instead of diabetes mellitus, the metabolic syndrome was not associated with reliable LSM using either the M (odds ratio [OR] 0.83; 95% confidence interval [CI] 0.46-1.48) or XL probes (OR 0.69; 95% CI 0.37-1.29). In disease-specific analyses, moderate to severe necroinflammation (METAVIR grades 2 to 3) was not associated with reliability using either the M or XL probes among patients with viral hepatitis (data not shown). However, among patients with NAFLD the presence of at least moderate lobular inflammation (NAS grade 2) was associated with a lower likelihood of achieving reliable results using both the M (OR 0.22; 95% CI 0.05-0.96; P = 0.04) and XL probes (OR 0.23; 95% CI 0.06-0.89; P = 0.03).
Table 3. Independent Predictors of Reliable Liver Stiffness Measurement* Using the FibroScan M and XL Probes
Odds Ratio (95% CI)
Odds Ratio (95% CI)
Valid shots ≥10, IQR/M ≤30%, and success rate ≥60%.
Age, per year
≥ 40 kg/m2
Steatosis >33% (S2-S3)
Skin-capsular distance ≥25 mm
Correlation Between Liver Stiffness Measured Using the M and XL Probes.
At least 10 valid LSMs with both probes were obtained in 178 patients (89%). In these individuals, liver stiffness as assessed by the M and XL probes was highly correlated (ρ = 0.86; P < 0.0005). The correlation between LSMs was strongest at lower values (Fig. 5A). This relationship was confirmed in a Bland-Altman plot (Fig. 5B), which demonstrated a greater difference in LSMs between probes at higher mean values (Pitman's test of difference in variance: r = 0.429; P < 0.0005). In general, liver stiffness was lower with the XL probe than the M probe (median 6.8 kPa [IQR 5.0-10.5] versus 7.8 kPa [IQR 6.1-13.9]; P < 0.0005). The mean and median differences between measurements were 2.3 kPa (95% CI 1.6-3.0) and 1.4 kPa (IQR 0.2-3.3), respectively. Boxplots illustrating the relationships between liver stiffness measured using both probes and fibrosis stage in patients with viral hepatitis and NAFLD are illustrated in Fig. 6A,B, respectively.
A post-hoc exploratory analysis examining the influence of measurement depth on the difference between liver stiffness measurements of the M and XL probes is illustrated in Supporting Fig. 1. When the FibroScan data were reprocessed to a measurement depth of 35 to 65 mm from the skin, the mean difference between the M and XL probes was 0 kPa (95% CI −0.5 to 0.5). At a common measurement depth of 35 to 75 mm from the skin (standard with the XL probe), the mean difference was −0.1 kPa (95% CI −1.0 to 0.7).
Diagnostic Performance of the M and XL Probes for the Assessment of Liver Fibrosis.
Among 178 patients with ≥10 valid LSMs using both probes, 159 (94%) had interpretable liver biopsies. The median interval between LSM and liver biopsy was 34 days (IQR 15-64); median biopsy length was 28 mm (IQR 23-33; range 15-53 mm); and the median number of portal tracts was 13 (IQR 10-15; range 7-39). Forty-nine percent of patients had significant (≥F2) fibrosis, 27% had severe fibrosis (≥F3), and 12% were cirrhotic (F4). Table 4 includes AUROCs for these outcomes, both overall and according to disease etiology. The only significant difference between the M and XL probes was in the differentiation of F2-4 from F0-1 fibrosis among patients with viral hepatitis (n = 69). In these patients, the AUROCs (95% CI) for the M and XL probes were 0.90 (0.83-0.98) and 0.82 (0.72-0.92), respectively (P = 0.02). Similar findings were observed in analyses restricted to patients with reliable LSM (Table 4), ≥5 valid measurements with both probes, and ≥10 valid measurements with either probe (Supporting Table 1).
Table 4. Diagnostic Performance of the M and XL Probes for the Assessment of Liver Fibrosis
Optimal Liver Stiffness Cutoffs with the M and XL Probes.
The optimal stiffness cutoffs using the M and XL probes for the diagnosis of significant fibrosis and cirrhosis both overall and according to disease etiology are outlined in Table 5. In general, the cutoffs and their operating characteristics are within the range of previous reports. Notable is that the optimal cutoffs for the XL probe were lower than those of the M probe (with the exception of significant fibrosis in patients with viral hepatitis). For example, for the diagnosis of ≥F2 fibrosis in patients with NAFLD, the optimal cutoffs for the M and XL probes were 7.8 and 6.4 kPa, respectively.
Table 5. Optimal Liver Stiffness Cutoffs Using the M and XL Probes According to Liver Disease Etiology*
Sensitivity % (95% CI)
Specificity % (95% CI)
PPV % (95% CI)
NPV % (95% CI)
Sensitivity % (95% CI)
Specificity % (95% CI)
PPV % (95% CI)
NPV % (95% CI)
Subgroup analyses restricted to patients with ≥10 valid measurements with the respective probes and interpretable liver biopsies. See Supplementary Table 1 for the numbers of patients within each analysis. Optimal liver stiffness cutoffs defined by the maximal sum of sensitivity and specificity.
F2-4 vs. F0-1 fibrosis
F4 vs. F0-3 fibrosis
In this prospective multicenter study, we confirmed the feasibility and performance of LSM using the FibroScan XL probe in overweight and obese patients with a variety of liver disorders. The major advantage of this new probe designed specifically for use in obese patients is that it facilitates LSM in more patients than is feasible with the standard M probe. For example, failure of LSM occurred in only 1% of patients with the XL probe compared with 16% with the M probe. Corresponding failure rates in patients with extreme obesity (BMI ≥40 kg/m2) were 5% and 59%, respectively. Similarly, at least 10 valid LSMs were obtained in 93% of patients with the XL probe compared with only 65% with the M probe; 81% of patients with fewer than 10 LSMs with the M probe were successfully assessed using the XL probe. These encouraging results confirm data from two small pilot studies evaluating the XL probe.15, 16 Among 17 patients with a BMI ≥30 kg/m2, Friedrich-Rust et al.16 obtained 10 valid measurements in 94% of patients with the XL probe versus only 65% with the M probe. Similarly, de Ledinghen et al.15 reported that 59% of obese patients for whom it was impossible to obtain 10 valid LSMs with the M probe were successfully measured using an XL probe prototype (which differed in several respects from the probe currently under study). The lower rates of success in the French study likely reflect the higher mean BMI of their cohort (41.5 versus 34.3 kg/m2 in our study), who were hospitalized specifically for obesity management. Considering the rising prevalence of obesity and NAFLD, these findings represent an important advance in the management of patients with chronic liver disease. With recent estimates suggesting that over 100 million Americans are obese, and that the majority of obese individuals have NAFLD,14 noninvasive and widely applicable screening tools for the assessment of liver fibrosis are desperately needed.
Failure of TE measurement in obese patients is predominantly related to hindrance of propagation of the FibroScan's shear wave and ultrasonic measurement due to subcutaneous adipose tissue. Previous studies have shown that the skin-capsular distance15, 16 and correlated measures (e.g., thoracic fold thickness and waist circumference) are important determinants of FibroScan failure.6, 12, 23 Indeed, the skin-capsular distance was ≥25 mm in 57% of patients in whom the M probe obtained fewer than 10 valid measurements. Moreover, the skin-capsular distance was an independent predictor of M probe (but not XL probe) reliability in our multivariate analysis (Table 3). Because the XL probe measures liver stiffness at a greater depth below the cutaneous surface (35-75 mm versus 25-65 mm with the M probe), the M probe is recommended for use in patients with a skin-capsular distance less than 25 mm; in the remainder, the XL probe is advised.15 Data from our study support these recommendations. Specifically, reliable TE assessments (≥10 valid LSMs, IQR/M ≤30%, and success rate ≥60%) were obtained using the M probe in 61% of patients with a skin-capsular distance <25 mm, 30% between 25 and 34 mm, and 8% with a distance ≥35 mm. Corresponding rates using the XL probe were 80%, 64%, and 31%, respectively. These data also highlight the fact that the FibroScan's quality control software that identifies LSMs as invalid cannot be relied upon in isolation to gauge the validity of TE results. According to the aforementioned criteria, LSMs were deemed reliable by both probes in 30%-40% of patients with a skin-capsular distance greater than the measurement depth of the probe.
Our data also shed light on a related issue, namely, the choice of probe in settings where an ultrasound is not available for measurement of the skin-capsular distance. As illustrated in Fig. 3, the proportion of patients with a skin-capsular distance >25 mm (i.e., greater than the depth of M probe measurement) mirrors the BMI. In nonobese patients (BMI 28 to <30 kg/m2), the skin-capsular distance exceeded 25 mm in only 8% of patients; in this group, the XL probe did not offer an advantage over the M probe. However, among higher BMI categories, in which the skin-capsular distance was >25 mm in 20%-74% of patients, the XL probe was more successful. Therefore, although ideally one would base the selection of FibroScan probe on the skin-capsular distance, use of the XL probe in obese patients and the M probe in those with a BMI <30 kg/m2 is a reasonable approach where an ultrasound is not available to measure this parameter.
In addition to comparing the feasibility of the M and XL probes, we confirmed the strong correlation between liver stiffness measured using both devices (ρ = 0.86; P < 0.0005).15, 16 Importantly, in patients successfully measured using both probes, liver stiffness tended to be lower using the XL probe. The mean and median differences between measurements were 2.3 kPa and 1.4 kPa, respectively. These differences were greatest at higher values of liver stiffness (Fig. 4) and independent of liver disease etiology (Fig. 5). These findings presumably reflect the presence of adipose tissue in the region of interest explored by the M probe in obese patients, leading to overestimation of liver stiffness. In addition, heterogeneity in hepatic fibrosis (e.g., greater fibrous tissue deposition in the subcapsular region) and the differences in measurement depth between probes likely play a role in these findings. For example, when tested on phantoms with homogeneous stiffness distribution, the M and XL probes give nearly identical stiffness measurements (V. Miette, Echosens; unpubl. data). In addition, our post-hoc analyses of data in which the M and XL probe were recalibrated to measure the same region of interest (both 35-65 mm and 35-75 mm from the skin) show elimination of this bias between probes (Supporting Fig. 1). In light of these findings, the optimal thresholds for interpreting liver stiffness using the XL probe are, in general, 1 to 2 kPa lower than those for the M probe (Table 5). However, the optimal cutoffs defined in our cohort require confirmation in light of the small sample sizes of some of these subgroups, particularly the disease-specific analyses. For example, the optimal M probe threshold for cirrhosis in the overall cohort (21.5 kPa)—which was derived based on data from only 19 cirrhotic patients—is significantly lower than typically reported (≈13 kPa).7, 8 Finally, as prospective studies of the FibroScan emerge, the relatively reduced range of LSMs observed with the XL probe compared with the M probe (Fig. 6) will require consideration when interpreting the significance of temporal changes in liver stiffness with this probe.
Our data confirm the diagnostic performance of the XL probe across various liver disorders. In the two previously published pilot studies of this probe,15, 16 histological confirmation was available in only one, which was restricted to 50 patients with NAFLD.16 In our entire cohort the AUROCs for the XL probe were 0.83 (95% CI 0.77-0.90) for significant fibrosis (≥F2) and 0.94 (95% CI 0.90-0.98) for cirrhosis. These figures are comparable with those observed for the M probe in our study (Table 4) and in numerous prior reports.7, 8 For example, in a meta-analysis of 50 studies of the M probe,7 Friedrich-Rust et al. reported summary AUROCs of 0.84 for significant fibrosis and 0.94 for cirrhosis. The diagnostic performance of the M and XL probes was similar across diseases except for significant viral hepatitis-related fibrosis, in which the M probe appeared more accurate (AUROCs, 0.90 versus 0.82 for the XL probe; P = 0.02). However, because there is no clear rationale for this discrepancy, these results must be interpreted cautiously considering the small number of patients in this analysis (n = 69).
Our study has several limitations. As in all studies that utilize liver biopsy to evaluate the performance of noninvasive tools for fibrosis assessment, sampling error and interobserver agreement in staging must be considered.24, 25 In order to mitigate these limitations, two experienced pathologists staged liver fibrosis and reached a consensus in cases of disagreement. Moreover, all biopsies were ≥15 mm in length and included ≥6 portal triads. Second, our cohort included patients with numerous liver diseases that may be staged using different scoring systems. However, the majority of our patients (88%) had viral hepatitis or NAFLD; fibrosis in these patients was staged according to standard classification systems18, 19 and appropriate subgroup analyses of probe performance were reported. In addition, disease-specific differences in liver stiffness have been described,26 as supported by the variability in optimal stiffness thresholds observed across conditions (Table 5). Potential explanations for these findings include differences in fibrosis staging systems and the quantity and character of fibrosis deposition (e.g., perisinusoidal/perivenular in NAFLD; periportal in viral hepatitis), and the influence of nonfibrotic histological features (e.g., steatosis and inflammation) on liver stiffness that may differ between disorders. These data suggest that different liver stiffness cutoff values may be necessary for patients with viral hepatitis and NAFLD, although the different fibrosis stage distributions between disorders may have influenced our results due to spectrum bias.27, 28 Nevertheless, additional studies including a larger number of patients will be necessary to validate these findings.
In conclusion, TE with the FibroScan XL probe facilitates LSM in a significantly greater number of obese patients than the standard M probe, while maintaining comparable accuracy. Our data suggest that lower liver stiffness cutoffs will be necessary when using the XL probe; however, additional large studies that adjust for important patient-related variables (e.g., liver disease etiology and BMI) will be necessary to validate the optimal cutoffs identified in our study.