Defining clinically relevant myosteatosis in patients with liver cirrhosis

Background: Myosteatosis is increasingly recognized as an indicator of high mortality risk in cirrhosis. However, no validated criteria to define this skeletal muscle alteration are currently available for clinical practice. We aimed to validate the recently proposed University of Alberta criteria and to assess their association with mortality. Methods: Individual patient data analysis combining two cirrhosis cohorts from Italy and Canada. Muscle attenuation and skeletal muscle index were assessed on CT scans at the L3 or L3-L4 level using a standardized protocol. Interobserver agreement was evaluated using intraclass correlation coefficients. Since the performance of the University of Alberta criteria in the Italian cohort was unsatisfactory, we used ROC analysis to derive a new attenuation threshold in the Italian cohort and validated it in the Canadian cohort. The association of the new cutoff for myosteatosis and its combination with sarcopenia and 1-year mortality was assessed by a proportional hazards model for competing risks, with liver transplantation as a competing event. Results: A total of 1,296 patients were included, with substantial differences in disease severity and baseline mortality rates between the cohorts. Interobserver agreement for muscle attenuation and skeletal muscle index was excellent (ICC >0.97). ROC analysis in the Italian cohort identified a muscle attenuation threshold of 26 Hounsfield units (HU) as the most discriminant cut-off for mortality risk, independent of sex and BMI. Patients' reclassification according to this threshold showed a sHR of 1.83 (95% CI 1.03-3.24; p=0.038) in the Italian and 1.63 (95%CI 1.25-2.11; p<0.001) in the Canadian cohort. Corresponding figures for myosteatosis combined with sarcopenia were 1.35 (95% CI, 1.07-1.71; p=0.01) and 1.32 (95% CI, 1.20-1.47; p<0.001), respectively. Conclusions: A muscle attenuation cut-off of HU ≤26 allows reproducible detection of myosteatosis in cirrhosis, across populations with different baseline disease severity.