2230. Association of Anthropometric Measurements and Fat Distribution with Predictors of Liver Disease in Patients with NAFLD
Authors * Denotes Presenting Author
  1. Juliana Sitta *; University of Mississippi Medical Center
  2. Edward Florez; University of Mississippi Medical Center
  3. Baylor Obert; University of Mississippi Medical Center
  4. Jeremiah Reese; University of Mississippi Medical Center
  5. Jon-Michael Stork; University of Mississippi Medical Center
  6. Elliot Varney; University of Mississippi Medical Center
  7. Candace Howard; University of Mississippi Medical Center
Non-alcoholic fatty liver disease (NAFLD) is the most common liver disease in the world, and it closely relates to obesity, diabetes, and dyslipidemia. Currently, there is not a well-established non-invasive, low-cost biomarker to predict NAFLD progression to fibrosis and stage of chronic liver disease. However, computed tomography (CT), magnetic resonance elastography (MRE), and ultrasound elastography have been increasingly accepted as noninvasive alternatives to the gold standard liver biopsy. CT is a promising technique due to its widespread use, its cost (compared to MRE), and its application in obese patients (compared to ultrasound). This study assessed the association of anthropometric measurements and specific distributions of adiposity with liver surface nodularity (LSN) and NAFLD clinical index in a diverse population diagnosed with NAFLD.

Materials and Methods:
For this HIPPA-compliant, IRB-approved, retrospective study, overweight or obese adult patients with various degrees of NAFLD and non-enhanced CT images of the abdomen and pelvis were selected (N=681). Patients whose CT did not meet the minimum necessary parameters were excluded (N=116), totaling a final sample of 565 subjects. From the final cohort, a sub-group of 287 patients with available clinical data to calculate the NAFLD index and another sample of 454 patients with BMI were analyzed. Three readers performed the segmentation of fat and muscle depots on 24 abdominal CT slices centered at the L4-L5 intervertebral space, using a validated segmentation software. Two readers obtained the LSN score using previously validated software. Sagittal abdominal diameter (SAD) and waist circumference (WC) were measured using a Digital Imaging and Communications in Medicine viewer. The coefficient of Pearson correlation was used to associate BMI, anthropometric measurements, and body composition with LSN scores and NAFLD index. Intra-class correlation coefficient (ICC) with 95% confidence intervals (CI) was used to assess inter-observer agreement.

LSN and BMI showed a direct correlation with SAD (r=0.58 and 0.77 respectively, p<0.001) and WC (r=0.52 and 0.60 respectively, p<0.001). Correlations between NAFLD index and anthropometric measurements showed minimal clinical significance (r=0.29 for WC and 0.21 for SAD, p<0.005). Total muscle attenuation (HU) showed an inverse association with WC (r=-0.28) and SAD (r=-0.18). Intra- and inter-observer agreement was excellent for the LSN scores (>0.88, 95%CI 0.82-0.92) as well as for the segmentation (>0.93, 95%CI 0.88-0.96) in a random sub-cohort (n=60). A multivariate regression model including LSN score, anthropometric indices, BMI, age, and body composition for predicting the NAFLD index was performed: NAFLD = -8.0886 + 0.028*WC - 0.160*SAD + 0.076*BMI + 0.063*Age + 4.073E-5*TotalFAT + 1.005*LSN.

In patients diagnosed with NAFLD, anthropomorphic measurements were linked to increased LSN scores, which are directly proportionate to clinical stage of chronic liver disease and have previously been shown to be independent markers of liver decompensation and death.