Prediction of Insulin Resistance with Anthropometric and Clinical Laboratory Measures in Nondiabetic Teenagers

Adolescent obesity is an important risk factor for cardiovascular disease in part due to its relationship to type 2 diabetes. Therefore, screening for type 2 diabetes is recommended. However, since insulin resistance (IR) is an early precursor of diabetes, it would be beneficial if IR predictors from routine assessments could identify those at greatest metabolic risk before diabetes has developed.

Adolescents aged 14-18 years with normal fasting plasma glucose (n = 252) were selected from the National Health and Nutrition Examination Survey 2013-2014. IR was estimated with homeostatic model assessment (HOMA). Multivariable linear/logistic regressions with demographic covariates assessed HOMA Index (HI)-determined IR associations with common clinical measurements.

The constellation of body mass index (BMI) percentile (P < 0.0001), diastolic blood pressure (DBP, P = 0.0056), and serum alanine aminotransferase (ALT, P = 0.0295) predicted log-transformed HI; cholesterol, triglycerides, and high- and low-density lipoprotein cholesterol were not predictive. Secondary gender analyses indicated significant associations in males (BMI percentile, P = 0.0008; DBP, P = 0.0003; ALT, P = 0.0436), but only BMI percentile in females (P = 0.0001). After stratifying predictors at clinical thresholds, logistic regressions had increased specificity at HI's 85th percentile compared with its 75th [85th: area under curve (AUC) = 0.87, 75th: AUC = 0.80]. At the 85th HI percentile, IR was associated with these components [BMI percentile: adjusted odds ratio (aOR) = 34.31, 95% confidence interval (CI): 8.5-139.3; ALT: aOR = 11.36, 95% CI: 3.0-43.7; DBP: aOR = 7.73, 95% CI: 2.4-24.6]. In males, IR was associated with elevated BMI percentile (aOR 24.0) and ALT (aOR 23.9) and in females with elevated BMI percentile (aOR 45.8) and DBP (aOR 17.9).

Three routine clinical assessments, BMI percentile, DBP, and ALT, predicted IR at HI's 75th and 85th percentiles in nondiabetic adolescents. Prospective validation may yield simple gender-specific screening identifying nondiabetic adolescents at greatest need of treatment intervention.