Hugo Perazzo1, Isabela Benseñor2, José Geraldo Mill3, Antônio G Pacheco4, Maria de Jesus Mendes da Fonseca5, Rosane Härter Griep6, Paulo Lotufo2, Dora Chor5. 1. Laboratory of Clinical Research on STD/AIDS, Evandro Chagas National Institute of Infectious Disease (INI). 2. Center for Clinical and Epidemiologic Research of the University of São Paulo, São Paulo. 3. Department of Physiological Sciences, Federal University of Espírito Santo, Vitória, Brazil. 4. Scientific Computing Program (PROCC). 5. National School of Public Health. 6. Laboratory of Health and Environment Education, Oswaldo Cruz Institute, Oswaldo Cruz Foundation (FIOCRUZ), Rio de Janeiro.
Abstract
GOALS: To develop a noninvasive algorithm for diagnosis of liver steatosis and to compare its diagnostic value with available predictive models. BACKGROUND: Liver steatosis represents the most frequent liver disease worldwide. STUDY: This cross-sectional study analyzed data from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Patients were randomly divided into training (n=6571) and validation (n=3286) cohort. Abdominal ultrasound (US), used to grade steatosis, and overnight fasting blood tests were performed at the same day. Fatty Liver Index (FLI), Hepatic Steatosis Index, and Nonalcoholic Fatty Liver Disease-Liver Fat Score were calculated. A backward stepwise multivariate logistic regression analysis was used to develop the new predictive model, Steato-ELSA. RESULTS: In total, 9857 subjects [58% female, age=51 (interquartile range, 45 to 58) years, body mass index=26.4 (23.9 to 29.6) Kg/m] were included. Body mass index, waist circumference, homeostasis model of assessment of insulin resistance, transaminases, and triglycerides were independently associated with steatosis in the multivariate model (Hosmer-Lemeshow P=0.279). In the validation cohort, the area under the receiver-operator characteristics (95% confidence interval) for prediction of mild and moderate steatosis were: (i) 0.768 (0.751-0.784) and 0.829 (0.810-0.848) for Steato-ELSA; (ii) 0.762 (0.745-0.779) and 0.819 (0.799-0.838) for Fatty Liver Index; (iii) 0.743 (0.727-0.761) and 0.800 (0.779-0.822) for Hepatic Steatosis Index; and (iv) 0.719 (0.701-0.737) and 0.769 (0.747-0.791) for Nonalcoholic Fatty Liver Disease-Liver Fat Score. Steato-ELSA performed significantly better than other models and yielded sensitivity (Se)/specificity (Sp) (95% confidence interval): (i) for mild steatosis (score ≥0.386): Se=65.6% (63.0-68.3) and Sp=73.7% (71.8-75.6); (ii) for moderate steatosis (score ≥0.403): Se=83.5% (80.0-86.9) and Sp=68.7% (67.0-70.4). CONCLUSIONS: Steato-ELSA is an accurate and inexpensive tool that uses simple parameters to identify individuals at high risk of liver steatosis.
GOALS: To develop a noninvasive algorithm for diagnosis of liver steatosis and to compare its diagnostic value with available predictive models. BACKGROUND:Liver steatosis represents the most frequent liver disease worldwide. STUDY: This cross-sectional study analyzed data from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Patients were randomly divided into training (n=6571) and validation (n=3286) cohort. Abdominal ultrasound (US), used to grade steatosis, and overnight fasting blood tests were performed at the same day. Fatty Liver Index (FLI), Hepatic Steatosis Index, and Nonalcoholic Fatty Liver Disease-Liver Fat Score were calculated. A backward stepwise multivariate logistic regression analysis was used to develop the new predictive model, Steato-ELSA. RESULTS: In total, 9857 subjects [58% female, age=51 (interquartile range, 45 to 58) years, body mass index=26.4 (23.9 to 29.6) Kg/m] were included. Body mass index, waist circumference, homeostasis model of assessment of insulin resistance, transaminases, and triglycerides were independently associated with steatosis in the multivariate model (Hosmer-Lemeshow P=0.279). In the validation cohort, the area under the receiver-operator characteristics (95% confidence interval) for prediction of mild and moderate steatosis were: (i) 0.768 (0.751-0.784) and 0.829 (0.810-0.848) for Steato-ELSA; (ii) 0.762 (0.745-0.779) and 0.819 (0.799-0.838) for Fatty Liver Index; (iii) 0.743 (0.727-0.761) and 0.800 (0.779-0.822) for Hepatic Steatosis Index; and (iv) 0.719 (0.701-0.737) and 0.769 (0.747-0.791) for Nonalcoholic Fatty Liver Disease-Liver Fat Score. Steato-ELSA performed significantly better than other models and yielded sensitivity (Se)/specificity (Sp) (95% confidence interval): (i) for mild steatosis (score ≥0.386): Se=65.6% (63.0-68.3) and Sp=73.7% (71.8-75.6); (ii) for moderate steatosis (score ≥0.403): Se=83.5% (80.0-86.9) and Sp=68.7% (67.0-70.4). CONCLUSIONS: Steato-ELSA is an accurate and inexpensive tool that uses simple parameters to identify individuals at high risk of liver steatosis.
Authors: Daniella Braz Parente; Hugo Perazzo; Fernando Fernandes Paiva; Carlos Frederico Ferreira Campos; Carlos José Saboya; Silvia Elaine Pereira; Felipe d'Almeida E Silva; Rosana Souza Rodrigues; Renata de Mello Perez Journal: Sci Rep Date: 2020-09-14 Impact factor: 4.379