OBJECTIVE. To identify clinical, socio-demographic, dietary and biological markers to be used in a non-invasive and cost-effective clinical tool for screening for non-alcoholic fatty liver disease (NAFLD) in obese adolescents.
METHODS. We conducted a cross-sectional analysis using baseline data from 77 obese adolescents enrolled in a drug trial for the Glaser Pediatric Research Network, between October 2003 and August 2007. NAFLD was defined as the presence of fatty liver infiltration assessed by computed tomography. Receiver operation characteristic (ROC) analyses were performed to identify variables with the highest area under the curve (AUC) for NAFLD. Serum biomarkers were dichotomized using sensitivity analyses to identify the best cutoff point for NAFLD. Multiple logistic regression models were created to predict prevalent NAFLD.
RESULTS. Serum triglycerides was identified as the best biomarker for NAFLD (AUC 0.790; pseudo R2 0.235). Additional adjustment for sex, age and Tanner stage improved the AUC to 0.846 and the pseudo R2 to 0.290. We then explored adding a simple biochemical marker for predicting NAFLD (HOMA-B, ALT or glutamate) and found that HOMA-B led to greater improvement in AUC, ALT to a greater improvement in sensitivity and glutamate to a greater improvement in the pseudo R2. Thus, all three factors individually improved overall model performance to some degree and inclusion of all three led to an AUC=0.907 and pseudo R2=0.433. Our second objective was to develop a more complex exploratory model starting with the inclusion of important clinical predictors (triglycerides, sex, age, Tanner stage, SBP, BMI, waist circumference); this yielded an AUC of 0.871 and pseudo R2 of 0.342. Further adjustment for HOMA-B, ALT and glutamate gave an AUC=0.913 and pseudo R2=0.497.
CONCLUSION. Simple clinical and biochemical factors may be used to screen for prevalent NAFLD. Our simplest clinically relevant model using triglycerides, age, sex and Tanner stage provided a reasonable screening tool for NAFLD in obese adolescents. A second more complex model that warrants further testing includes triglycerides, sex, age, Tanner stage, SPB, BMI, waist circumference, HOMA-B, ALT and glutamate. In this study, this model was more accurate for detecting undiagnosed cases of NAFLD in this pediatric population.
Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/20802 |
Date | 09 March 2017 |
Creators | Lara-Castor, Laura |
Source Sets | Boston University |
Language | en_US |
Detected Language | English |
Type | Thesis/Dissertation |
Page generated in 0.0018 seconds