Insulin disposition in the lung following Oral inhalation in humans a meta-analysis of its pharmacokinetics

Oral inhalation of insulin potentially offers non-invasive treatment and better glycaemic control in diabetes by virtue of its apparently faster absorption into the systemic circulation compared with subcutaneous injection. Nevertheless, the lung kinetics of inhaled insulin in humans have yet to be fully clarified because of the complexity of insulin-glucose (patho)physiology and the difficulty in approximating the inhaled dose. As a result, there remains considerable debate on the mechanisms of absorption and metabolism of insulin in the lung. To develop and apply a physiologically realistic insulin-glucose kinetic model to a meta-analysis of insulin-glucose profiles from well-controlled clinical studies of inhaled insulin published in the literature, and thereby, to derive the kinetic descriptors of insulin in the lung following inhalation through curve fitting. The model assumed first-order absorption (k(a,L)) and parallel non-absorptive loss (k(mm,L)), the latter primarily occurring via metabolism and mucociliary clearance in the lung, alongside two systemic compartments. Where necessary, glucose-dependent endogenous pancreatic insulin secretion was also taken into account by using blood glucose data as the second independent variable. Despite the model's simplicity and the use of mean data, 16 insulin-glucose profiles from ten clinical studies were successfully fitted to the model, yielding values for the rate constants k(a,L) and k(mm,L). Whole serum insulin profiles were rate-determined by k(a,L) and k(mm,L) combined, representing 'flip-flop' pharmacokinetics. The best estimate for k(a,L) was found to be 0.020-0.032 h(-1), effectively unchanged across doses (0.3-1.8 IU/kg), formulations (powder vs liquid) and subjects (healthy vs diabetic), suggesting passive diffusive absorption of insulin from the lung. In contrast, the values for k(mm,L) were much larger (0.5-1.6 h(-1)) and decreased with increasing inhaled dose. Therefore, it is likely that dose-dependent saturable lung metabolism controls the value of k(mm,L), alongside mucociliary clearance. As a result, the absolute bioavailability ranged from 1.5% to 4.8%. The modelling analysis also enabled derivation of increased values for both k(a,L) and k(mm,L) as a possible cause of faster absorption for deep inspiratory manoeuvres and increased absorption in smokers, and faster and increased absorption for insulin lispro. Although some of these results need to be substantiated experimentally, it appea