Potency and plasma protein binding of drugs in vitro-a potentially misleading pair for predicting in vivo efficacious concentrations in humans

In drug discovery or preclinical stages of development, potency parameters such as $IC_{50}$, $K_i$, or $K_d$ in vitro have been routinely used to predict the parameters of efficacious exposure (AUC, $C_{min}$, etc.) in humans. However, to our knowledge, the fundamental assumption that the potency in vitro is correlated with the efficacious concentration in vivo in humans has not been investigated extensively. Thus, the present review examined this assumption by comparing a wide range of published pharmacokinetic (PK) and potency data. If the drug potency in vitro and its in vivo effectiveness in humans are well correlated, the steady-state average unbound concentrations in humans [$C_{u_-ss.avg}=f_u{\cdot}F{\cdot}Dose/(CL{\cdot}{\tau})=f_u{\cdot}AUCss/{\tau}$] after treatment with approved dosage regimens should be higher than, or at least comparable to, the potency parameters assessed in vitro. We reviewed the ratios of $C_{u_-ss.avg}$/potency in vitro for a total of 54 drug entities (13 major therapeutic classes) using the dosage, PK, and in vitro potency reported in the published literature. For 54 drugs, the $C_{u_-ss.avg}$/in vitro potency ratios were < 1 for 38 (69%) and < 0.1 for 22 (34%) drugs. When the ratios were plotted against $f_u$ (unbound fraction), "ratio < 1" was predominant for drugs with high protein binding (90% of drugs with $f_u{\leq}5%$; i.e., 28 of 31 drugs). Thus, predicting the in vivo efficacious unbound concentrations in humans using only in vitro potency data and $f_u$ should be avoided, especially for molecules with high protein binding.