The use of unbound drug concentrations is crucial for the prediction of efficacious doses. Hence, dose predictions for antibiotics targeting respiratory pathogens should be based on free, rather than the currently used, total drug concentrations in epithelial lining fluid (ELF). In this work, we describe an assay to estimate the percent unbound of drugs in ELF using simulated epithelial lining fluid (sELF) containing the most abundant components of ELF in healthy humans. A diverse set of 85 compounds showed a broad range of unbound values ranging from <0.01 to 100%. Binding in sELF was influenced by ionization, with basic compounds typically resulting in a stronger binding than neutral and acidic compounds (median percent unbound values 17, 50, and 62%, respectively). A permanent positive charge further increased binding (median percent unbound 11%), while zwitterions showed a lower binding (median percent unbound 69%). In lipid-free sELF, the binding of basic compounds was less pronounced, while compounds of other ionization classes were less impacted, indicating that lipids are involved in the binding of bases. A reasonable correlation was found between binding in sELF and human plasma (R2 = 0.75); however, plasma binding poorly predicted sELF binding for basic compounds (R2 = 0.50). Bases are an important compound class for antibacterial drug development since positive charges affect permeability into Gram-negative bacteria, which are important in terms of bacterial pneumonia. To evaluate in vivo activity, we selected two bases, for which strong sELF binding was observed (percent unbound <1 and 7%) and conducted an analysis of antibacterial efficacy in the neutropenic murine lung efficacy model and total vs free ELF drug concentrations. In both cases, the total ELF resulted in an overprediction of expected efficacy, while the corrected free ELF explained the observed in vivo efficacy. This supports that free, and not total, ELF concentrations should be used for the efficacious dose prediction for pneumonia and highlights the importance of determining binding in this matrix.
Keywords: antibiotics; efficacious dose prediction; epithelial lining fluid; percent unbound; pneumonia; unbound drug concentration.