Statistical agreement between ultrasound (US) and computerized tomography (CT) for non-alcoholic liver disease (NAFLD) diagnosis

Purpose of the Study

To assess K statistic of non-invasive imaging evaluations to diagnose NAFLD by means of US fatty liver indicator (FLI) score and liver to spleen attenuation ratio by CT.

Methods

Cross-sectional observational study that included all consecutive HIV-infected patients seen at a metabolic clinic who were screened for NAFLD with US and CT in the same day. HCV, HBV and heavy alcohol users (>20 g/day) were excluded. With CT, NAFLD diagnosis was defined with an L/S<1.1. US scanning was performed by a single operator. Presence and severity of steatosis was evaluated by a semiquantitative echographic score (FLI) graduated from 2 to 8. "Sine qua non" condition for the diagnosis of steatosis was the presence of liver/kidney contrast, graduated as mild/moderate (score 2) and severe (score 3). Additional criteria (score 1 each) included posterior attenuation of ultrasound beam, vessel blurring, difficult visualization of gallbladder wall or diaphragm and areas of focal sparing. US NAFLD was diagnosed by the score ≥2.

Summary of Results

46 patients were included. NAFLD prevalence with US was 19.57% and with CT 50.00%. (Tables 1 and 2).

K of Cohen coefficent between L/S ≤1.1 and US score ≥2 was 0.21 (agreement = 60.87%), US score ≥3 was 0.45 (agreement = 80.43%), US score ≥4 was 0.29 (agreement = 80.43%).

Table 1

Table 2

Conclusion

Non-invasive imaging agreement between US and CT to diagnose NAFLD is less than ideal. Best concordance is found for US ≥3 and L/S ≤1.1. Liver biopsy studies are needed to validate cut-off of US evaluation and L/S in people living with HIV.