Nonlinear mixed effects modeling of glucagon kinetics assessed using [13C15N]-glucagon in healthy and type 1 diabetic subjects

To date, few studies have quantified glucagon kinetics in humans with and without diabetes, with results often varying due to differences in experimental designs and glucagon assays methodologies. These variations have limited the ability to study glucagon secretion in vivo using methods such as deconvolution, which are commonly employed for hormones like insulin and C-peptide. To overcome these limitations, a novel stable glucagon tracer, [13C15N]-glucagon, along with high-resolution mass spectrometry, has been used to precisely measure glucagon kinetics and turnover. In this work, we present a nonlinear mixed effect modeling approach to describe glucagon kinetics in healthy and type 1 diabetic subjects exploiting [13C15N]-glucagon data collected from 9 healthy controls and 12 individuals with T1D following an intravenous bolus injection under basal steady-state, post-absorptive conditions. Within the Monolix framework, models of increasing complexity were developed and applied to the tracer data, with model selection guided by criteria such as data fit quality, precision and physiological plausibility of parameter estimates and parsimony. A one-compartment model and a two-compartment model were tested, excluding and including the earliest time points (t = 1 min), respectively. Both models effectively described typical population glucagon kinetics (TPK) along with the between-subject variability (BSV) and inter-occasion variability (IOV), linking individual differences to easily accessible subject characteristics, such as body weight. The two-compartment model, in particular, successfully captured the early decline in glucagon concentration, effectively describing the rapid initial phase post-administration. These findings are crucial for assessing glucagon turnover in the post-absorptive state in humans. This work represents a significant step toward the quantification of glucagon secretion in humans and lays the foundation for future studies involving both healthy individuals and those with type 1 diabetes. Once validated, these models could be integrated into diabetes simulation platforms and, in combination with models of glucagon action and subcutaneous absorption, used to evaluate the efficacy of dual-hormone artificial pancreas systems.