Author information
1Department of Gastroenterology, All India Institute of Medical Sciences, New Delhi, India.
2Department of Pathology, All India Institute of Medical Sciences, New Delhi, India.
3Department of Gastrointestinal Surgery and Liver Transplant, All India Institute of Medical Sciences, New Delhi, India.
4Department of Surgical Disciplines, All India Institute of Medical Sciences, New Delhi, India.
5Department of Radiodiagnosis, All India Institute of Medical Sciences, New Delhi, India.
6Department of Gastroenterology, All India Institute of Medical Sciences, Patna, India.
7Department of Gastroenterology, All India Institute of Medical Sciences, New Delhi, India. drshalimar@yahoo.com.
Abstract
Background: Advanced fibrosis and cirrhosis (compensated advanced chronic liver disease [cACLD]) are clinically indistinguishable and increase risk of developing clinically significant portal hypertension. Baveno VII recommends using elastography to rule out and diagnose cACLD with liver stiffness measurement (LSM) cut-offs of 10/15 kPa.
Methods: In a retrospective analysis of 330 nonalcoholic fatty liver disease (NAFLD) patients, performance of the Baveno VII cut-offs for diagnosing cACLD was compared with newly suggested lower cut-offs (8/12 kPa). A model for detecting cACLD among those with LSM between 8 and 12 kPa was developed and compared with recently published models.
Results: Seventy (21.2%) of the 330 NAFLD patients had biopsy-proven cACLD. The Baveno VII cut-offs (10/15 kPa) had a lower sensitivity of 72.8% (60.9-82.8%) and a specificity of 93.4% (89.7-96.1%). Sensitivity and specificity of lower cut-offs (8/12 kPa) were 91.4% (82.3-96.8%) and 88.5% (83.9-92.1%), respectively. Modeling based on the presence of diabetes (odds ratio [OR] 3.625[1.161-11.320], p = 0.027) and serum aspartate aminotransferase (AST) levels (OR 1.636[1.098-2.436], p = 0.015) correctly identified 75.7% of patients with LSM between 8 and 12 kPa. Our model performed best with an area under receiver operator curve (AUROC) of 0.725 (95%CI 0.609-0.822), compared to Papatheodoridi (AUROC 0.626, CI 0.506-.736) and Zhou (AUROC 0.523, CI 0.403-0.640) models. A two-step strategy comprising application of lower LSM cut-offs followed by the predictive model correctly identified the presence of cACLD in 83% of the patients as compared to 75% by the Baveno VII cut-offs.
Conclusion: A two-step strategy employing lower LSM cut-offs and modeling based on diabetes and AST levels outperforms Baveno VII cut-offs for identifying cACLD in NAFLD patients.