RESUMEN
BACKGROUND & AIMS: Transient elastometry is a noninvasive procedure used to measure fibrosis when patients are diagnosed with liver disease; it might be used to monitor changes over time. We investigated whether there are short-term variations in stiffness measurements that are not attributable to changes in fibrosis by studying patients with stable liver disease. METHODS: We performed a retrospective analysis of 531 paired liver stiffness measurements made by Fibroscan when the study began (LSM1) and at follow-up (LSM2), more than 1 day and less than 1 year apart, from 432 stable (for body mass index, waist circumference, and alcohol consumption), untreated, immunocompetent patients with chronic liver disease (from January 2006 through March 2009). Variations between the first and follow-up measurements were expressed as absolute (LSM2-LSM1, kPa) or relative ([LSM2-LSM1]/LSM1*100) or as changes in fibrosis stage. RESULTS: There was >20% variation in 49.7%, >30% in 34.3%, and >50% in 12.2% of paired measurements; this variation was constant across the spectrum of LSM1 values. The variations produced a 1-fibrosis stage difference in 31.5% of pairs and a ≥ 2-stage difference in 9.8% of pairs. Patients with LSM1 >7 kPa had increased probability of having a different stage of fibrosis at LSM2, compared with patients with LSM1 <7 kPa. Factors associated with variation included measurements made by 2 different operators or at least 1 non-senior operator, ratios of interquartile range:median values, significant fibrosis (≥ 7 kPa) at LSM1, baseline body mass index, or a 2-fold difference in level of alanine aminotransferase between measurements. When the analyses were restricted to measurements made by the same operator, the variation was slightly reduced; fibrosis stage differed between measurements for only 34.3% of cases. CONCLUSIONS: Operator-related and patient-related factors produce significant variations in liver stiffness measurements made by transient elastometry, limiting its use in monitoring patients. These variations are unrelated to disease progression. The lowest levels of variation occur in measurements made in patients with no or early-stage fibrosis or by a single experienced operator.