Authors
J. MERTENS (1), J. WEYLER (1), E. DIRINCK (2), L. VONGHIA (1), L. MORTELMANS (1), C. DE BLOCK (2), S. FRANCQUE (1) / [1] Antwerp University Hospital, Edegem, Belgium, Gastroenterology and Hepatology, [2] Antwerp University Hospital, Edegem, Belgium, Endocrinology, Diabetology and Metabolism
Introduction
Data are limited concerning the value of scores based on clinical and biochemical parameters to diagnose NAFLD in type 1 diabetes (T1D).
Aim
We aimed to evaluate the effectiveness of controlled attenuation parameter (CAP), Hepatic Steatosis Index (HSI) and Fatty Liver Index (FLI) compared to conventional ultrasound (US) as reference method in a study of 407 subjects with T1D.
Methods
Adult T1D subjects were screened for NAFLD using simultaneously US, CAP (criterion for NAFLD diagnosis: ≥215 dB/m on M probe or ≥250 dB/m on XL probe according to skin-liver capsule distance), FLI (criterion for NAFLD diagnosis ≥60) AND HSI (criterion for NAFLD diagnosis ≥36). We subtracted 2 points from each HSI score to correct for diabetes, since it gives weight to the presence of diabetes.
Results
407 adult subjects were included (male sex proportion: 56.8%, age 46 (IQR: 29) years, HbA1c 7.5 ± 1.0 %, BMI 25.5 ± 4.0 kg/m2, diabetes duration 26.3 ± 14.2 years, metabolic syndrome proportion: 31.7%). The prevalence of NAFLD was 20.4% based on US, 52.1% based on CAP, 43.8% based on HSI and 18.3% based on FLI. There was strong correlation between HSI and FLI (r:0.724, p<0.001), moderate correlation between FLI and CAP (r:0.578, p<0.005) and between HSI and CAP (r:0.400, p<0.001) and weak correlation between US and CAP (r: 0290, p<0.001). FLI and HSI (k: 0.395, p<0.001), HSI and CAP (k: 0.246, p<0.001) and US and CAP (k: 0.228, p<0.001) showed fair agreement, while FLI and CAP showed slight agreement (kappa: 0.177, p<0.001). Sensitivity of CAP versus US was 81%, specificity: 55%, PPV: 32%, NPV: 92%. AUROC for CAP yielded 0.77 [0.71-0.82], p<0.001. Sensitivity of FLI versus US was 52%, specificity: 90%, PPV: 58%, NPV: 88%. AUROC for FLI yielded 0.79 [0.73-0.85], p<0.001. Sensitivity of HSI versus US was 76%, specificity: 64%, PPV: 35%, NPV: 92%. AUROC for HSI yielded 0.74 [0.68-0.80], p<0.001. All tests correlated moderately with the metabolic syndrome (HSI r:0.433, p<0.001, CAP r:0.135, p<0.001), except the FLI, which correlated more strongly (FLI r:0.568, p<0.001). In logistic regression analysis, adjusting for age, gender, systolic and diastolic blood pressure or antihypertensive drug use, HDL-c and triglycerides levels, FLI (OR:1.04 [1.03-10.6];p<0.001), CAP (OR:1.02 [0.01-1.02], p<0.001) and HSI (OR:1.17 [1.11-1.24], p<0.001) were all associated with US-determined NAFLD.
Conclusions
Based on US, NAFLD seems prevalent in T1D. Clinicobiochemical scores show moderate diagnostic accuracy compared to US to determine NAFLD in this pilot study. Mutual agreement between diagnostic tools is moderate, stressing the need for more specific diagnostic tools in subjects withT1D.
