Recently, we showed that a linear two-compartment model (Schiavon et al., DTT 2015) is able to describe plasma-to-interstitial fluid (ISF) glucose kinetics both in steady as well as non-steady state conditions (Schiavon et al., ATTD 2019) using multi-tracer plasma and microdialysis data. The model allows estimation of plasma-to-ISF equilibration time (τ). However, on average, slower kinetics and greater variability was shown in non-steady than steady state conditions. Here we aim to test in silico the role that experiment design variables may have on τ estimation.
The 100 virtual adult population of the UVA/Padova T1D simulator (Visentin at al., JDST 2018) was used to simulate plasma-to-ISF glucose kinetics in fasting (steady state) and postprandial (non-steady state) conditions. In addition, a primed-constant i.v. infusion of glucose tracer was simulated. Measurements of glucose concentrations and tracer enrichments were simulated in both plasma and ISF. Different experimental settings were simulated while τ estimation was performed by fitting the model to ISF glucose tracer data using plasma measurements as forcing functions.
The model is able to describe the data in the various experimental settings. An effect of sampling schedule and data pooling in both steady state and non-steady state conditions have been observed. The role of measuring glucose concentration in ISF has also been assessed.
Experiment design is critical to accurately assess plasma-to-ISF glucose kinetics and should be taken into account in evaluating the plasma-to-ISF equilibration time.