Su1006 Longer Duration of Transient Elastography Predicts Unreliable Measurements and Measurement Failure

Abstract

Introduction: Transient elastography (TE) is a noninvasive, validated method to assess liver fibrosis by obtaining liver stiffness measurements (LSM). Measurement failure (MF) and unreliable measurement (UM) occur in up to 3% and 15% of patients, respectively. The relationship between time taken to perform TE (duration) and UM has not been studied. Aim: To determine whether the examination duration correlates with unreliable measurement and measurement failure of transient elastography. Materials and Methods: This is a prospective study of the frequency and predictors of UM based on LSM obtained by FibroscanTM from September 2008 to June 2013. UM was defined as any of the following: 30% of the median LSM. MF was defined as zero valid measurements. All operators had performed >100 prior examinations. Data collected included age, sex, BMI (kg/m2), probe used (M or XL), stiffness (LSM in kPa), duration of TE (in minutes:seconds) and aetiology of liver disease. Patients were categorised as cirrhotic according to validated disease-specific median liver stiffness cut-offs. Multivariate binary logistic regression was performed to identify factors independently associated with UM. Results: Liver stiffness measurements of 2835 patients were analysed (55% male, median age 51 years [IQR 4160]). The aetiologies of liver disease were hepatitis B (42.7%), hepatitis C (23.7%), nonalcoholic fatty liver disease (7.8%), alcohol (3.4%) and other (22.4%). MF and UM occurred in 3.1% and 19.0%, respectively. Duration, LSM and obesity were independent predictors of UM. The area under the receiver operating curve (AUROC) of duration for diagnosing UM was 0.716 (0.69-0.74) with an optimal cut-off of 3mins 47s (sensitivity 70%, specificity 65%, odds ratio (OR) 4.2[95% CI 2.7-6.6], P 30kg/m2) was an independent predictor of UM (OR 1.9[1.2-3.0], P=0.009). Patients with median LSM falling in the cirrhosis range were more likely to have UM (OR 2.3[1.4-3.9], P=0.002). The majority of these UM (72.7%) were attributed to IQR >30%, indicating higher variability at the upper ranges of LSM. Conclusions: In experienced hands, duration is a strong predictor of UM in patients undergoing TE. Examinations longer than 4 minutes are more likely to be unreliable. Examinations longer than 8 minutes are unlikely to yield a valid result and should be considered a futility endpoint. Obesity and elevated LSM in the cirrhosis range are independent, albeit weaker, predictors of UM. Table1. Duration cut-offs for UM at 90, 95 and 99% specificity