Evaluating feasibility and accuracy of non-invasive tests for nonalcoholic fatty liver disease in severe and morbid obesity.

Ooi, Geraldine J; Earnest, Arul; Kemp, William W; Burton, Paul R; Laurie, Cheryl; Majeed, Ammar; Johnson, Nathan; McLean, Catriona; Roberts, Stuart K; Brown, Wendy A

INTRODUCTION In obese individuals, nonalcoholic fatty liver disease (NAFLD) is common but often goes undiagnosed, and therefore untreated. The presence of significant fibrosis is a key determinant of NAFLD progression, and liver steatosis has substantial cardiovascular implications. We aimed to determine the diagnostic accuracy of common noninvasive diagnostic tests for steatosis and fibrosis in the obese. METHODS We recruited 182 severely and morbidly obese individuals undergoing bariatric surgery (age 44 ± 12 years, body mass index 45.1 ± 8.3 kg/m2). Medical history, blood tests and liver biopsy were taken on the day of surgery. Serum steatosis and fibrosis scores were calculated. In a subgroup of patients, transient elastography with controlled attenuation parameter (TE/CAP) (n = 82) and proton magnetic resonance spectroscopy (1H-MRS) (n = 49) were performed. RESULTS 1H-MRS had excellent diagnostic accuracy for steatosis, with strong correlation to steatosis (r = 0.647, p < 0.001), good AUROC (0.852, p = 0.001), sensitivity (81.3%) and specificity (87.5%). However, due to low feasibility in this cohort (65.3% success), this was substantially decreased with intention-to-diagnose analysis (sensitivity 50.0%, specificity 60.9%). CAP had good feasibility (80.5%), and performed better in intention-to-diagnose analysis (AUROC 0.688, sensitivity 84.8%, specificity 47.2%). Serum steatosis scores performed poorly, with comparable accuracy to ALT. For significant fibrosis, TE had the best accuracy (AUROC 0.903, p = 0.007), which remained reasonable after intention-to-diagnose analysis (sensitivity 100%, specificity 59.0%). A combination approach using CAP with ALT for steatosis and TE with Forn index for fibrosis yielded reasonable overall accuracy. CONCLUSIONS 1H-MRS and TE/CAP had greatest accuracy for NAFLD-related steatosis and fibrosis. Failure rates in obesity significantly diminished diagnostic ability. Use of a combination of serum and imaging tests improved overall feasibility of assessment and diagnostic accuracy in obese individuals.