Fasting indices of glucose-insulin-metabolism across life span and prediction of glycemic deterioration in children with obesity from new diagnostic cut-offs

Abstract

Background: Fasting indices of glucose-insulin-metabolism are an easy and affordable tool to assess insulin resistance. We aimed to establish reference ranges for fasting insulin indices that reflect age-dependent variation over the entire life span and subsequently test their clinical application regarding the prediction of glycemic deterioration in children.

Methods: We calculated age- and puberty-dependent reference values for HOMA-IR, HOMA2-IR, HOMA-β, McAuley index, fasting insulin, and fasting glucose from 6994 observations of 5512 non-obese healthy subjects aged 5-80 years. Applying those references, we determined the prevalence of insulin resistance among 2538 subjects with obesity. Furthermore, we investigated the intraindividual stability and the predictive values for future dysglycemia of these fasting indices in 516 children and adolescents with obesity up to 19 years of follow-up. We validated the results in three independent cohorts.

Findings: There was a strong age-dependent variation of all indices throughout the life span, including prolonged recovery of pubertal insulin resistance and a subsequent continuous increase throughout adulthood. Already from age 5 years onwards, >40% of children with obesity presented with elevated parameters of insulin resistance. Applying newly developed reference ranges, insulin resistance among children with obesity doubled the risk for future glycemic deterioration (HOMA-IR HR 1.88 (95% CI 1.1-3.21)), fasting insulin HR 1.89 (95% CI 1.11-3.23). In contrast, fasting glucose alone was not predictive for emerging dysglycemia in children with obesity (HR 1.03 (95% CI 0.62-1.71)). The new insulin-based thresholds were superior to fasting glucose and HbA1c in detecting children eventually manifesting with dysglycemia in prospective analyses.

Interpretation: The variation of fasting glucose-insulin-metabolism across the life span necessitates age-specific reference ranges. The improved prediction of future glycemic deterioration by indices based on fasting insulin beyond simple glucose measures alone could help to stratify risk characteristics of children with obesity in order to guide patient-tailored prevention and intervention approaches.

Funding: German Research Foundation (DFG)-through SFB 1052, project number 209933838, subproject C5; Federal Ministry of Education and Research, Germany; European Union-European Regional Development Fund; Free State of Saxony. The German Diabetes Association, the CarbHealth consortium (01EA1908B). EU-IMI2-Consortium SOPHIA (grant agreement No 875534), German Center for Diabetes Research (DZD), grant number 82DZD14E03.

Keywords: Childhood obesity; HOMA-IR; Insulin resistance; Insulin sensitivity; McAuley index; Obesity; Prediabetes; Reference values; Type 2 diabetes.

Fig. 1

Distribution of fasting indices over age (5–80 years). Observations from the non-obese reference cohort are depicted as solid line, whereas the obesity cohort is depicted as dashed line. Curved lines and ribbons represent local polynomial regression fitting with 95% confidence intervals. ∗cut-off for the diagnosis of prediabetes according to the American diabetes association; HOMA-IR, Homeostasis model assessment of insulin resistance index.

Fig. 2

Age-dependent distribution of fasting indices among subjects without and with obesity. The dashed lines represent the age-specific 97.5th percentile (2.5th percentile respectively for McAuley index) as a cut-off for pathological values. The prevalence of pathological findings among subjects with obesity is depicted next to the violin plots. Black symbols represent the median of subjects without obesity and with obesity.

Fig. 3

Prediction of future dysglycemia. Participants with obesity but without dysglycemia at baseline (n = 516) were stratified according to their fasting index value. Hereby, the 90th (P90) percentile (or P10 respectively for McAuley index) serves as the cut-off to separate a risk group (blue) from a non-risk group (gray). Differences of event-free survival (time without occurrence of dysglycemia) were compared between risk groups by log-rank test. Ribbons around the survival curves represent the 95% confidence interval and ticks represent right-censored data.

Fig. 4

Graphicalabstract.