A microfluidic in vitro method predicting the fate of peptide drugs after subcutaneous administration

For many biopharmaceuticals, subcutaneous (sc) administration is the only viable route. However, there is no in vitro method available accurately predicting the absorption profiles of subcutaneously injected pharmaceuticals. In this work, we show that a recently developed microfluidics method for interaction studies (MIS) has the potential to be useful in this respect. The method utilises the responsiveness of polyelectrolyte microgel networks to oppositely charged molecules as a means to monitor the interaction between peptides and hyaluronic acid (HA), a major constituent of the subcutaneous extracellular matrix. We use the method to determine parameters describing the strength of interaction between peptide and HA as well as the peptide's aggregation tendency and transport properties in HA networks. The results from MIS studies of the peptide drugs exenatide, pramlintide, vancomycin, polymyxin B, lanreotide, MEDI7219 and AZD2820 are compared with results from measurements with the commercially available SCISSOR system and in vivo absorption and bioavailability data from the literature. We show that both MIS and SCISSOR reveal differences in the peptides' diffusivity and tendency to aggregate in the presence of HA. We show that MIS is particularly good at discriminating between peptides forming aggregates stabilised by non-electrostatic forces in the presence of HA, and peptides forming complexes stabilised by electrostatic interactions with HA. The method provides two parameters that can be used to quantify the peptides' aggregation tendency, the one describing the peptide packing density in complexes with HA and the other the apparent diffusivity upon release in a medium of physiological ionic strength and pH. The order of the peptides when ranked by increasing binding strength at pH 7.4 determined with MIS is shown to be in agreement with the order when ranked by the apparent 1st order absorption rate constant (k_a) after sc administration in humans: lanreotide (Autogel) < exenatide (IRF) < AZD2820 < pramlintide < lanreotide (IRF) (IRF: Immediate release formulation). A correlation is found between the 1st order release rate constant determined with SCISSOR and k_a for lanreotide (Autogel), exenatide and AZD2820. A mechanism relating the magnitude of k_a to the peptides' charge is proposed.