Impact of Target Interactions on Small-Molecule Drug Disposition

CORRESPONDENCE LINK TO ORIGINAL ARTICLE

Typically, the impact of a target on the pharmacokinetics of a drug depends on: the Impact of target interactions on affinity of the drug for the target; the total binding capacity of the target within the body; small-molecule drug disposition: the affinity and binding capacity of other competing body constituents; the location an overlooked area of the target; and the dose and corresponding concentration of the drug. Some of these factors are interdependent, but they can be Robert A. B. van Waterschoot1, Neil J. Parrott1, Andrés Olivares-Morales1, categorized into broad effects. 1 2 3 Thierry Lavé , Malcolm Rowland and Dennis A. Smith • If the target is tissue-based and is a major binding constituent within the body, it The selection of an appropriate drug target, as target-mediated drug disposition (TMDD)3 will influence local tissue distribution and well as a solid understanding of drug pharma- on small-molecule drug discovery, as well as V globally, but will have no effect on CL. cokinetics and pharmacodynamics, are cru- approaches to address this issue. Typically, V will decrease with increasing cial for successful drug discovery1. Although it dose as a result of the saturation of target is generally accepted that the free (unbound) Impact of TMDD binding sites, and the half-life of the drug concentration of a drug is the driving force Impact on in vitro studies. Nowadays, a large will decrease proportionally. for its pharmacological effects (as only free proportion of early discovery assays overex- • If the target is plasma-based, saturation of drug molecules are available to interact with press the pharmacological target to create the target with increasing dose will result the therapeutic target), there are several mis- a viable in vitro assay. The general assump- in lower plasma binding, which will be conceptions regarding the impact of drug tion is that the concentrations of target pro- particularly marked when the target is the binding on pharmacokinetics2. It is important teins are considerably lower than the affinity major binding protein. This will typically to understand that the binding of a drug to (equilibrium dissociation constant (Kd)) of be associated with an increase in both V constituents of the blood or tissues does not the compounds being tested. With nanomolar and CL with a change in half-life, unless alter the overall unbound drug concentra- and even picomolar affinities being routinely V and CL change to the same extent. For tion following chronic drug administration, achieved in drug discovery, however, this plasma-based (or blood-based) targets, as this is primarily determined by intrinsic assumption may be incorrect. As such, with there is a high probability that some influ- clearance2. However, drug binding can highly a fixed high concentration of receptor in the ence of target binding will be observed influence the pharmacokinetic profile, owing assay, the apparent affinity of compounds in vivo. The influence of the target may be to the effects on the volume of distribution will reach a lower limit approximating to the missed in in vitro studies unless plasma (V) and total drug clearance (CL), which are concentration of receptor. Thus a series of protein binding-type assays are conducted based on measurements of the total plasma compounds with a range of potencies would over a concentration range commencing concentration. appear equipotent if their affinity values were at a value below the Kd of the drug for its The major nonspecific drug-binding con- screened in an assay with a receptor concen- pharmacological target. stituents are albumin (which is found in the tration greater than those values. Clearly, this • When the target is the predominant bind- blood and interstitial fluids) and acidic phos- may mean underestimation of compound ing constituent in the body (or in the pholipids (which are predominantly found affinity and selectivity, as well as overestima- blood) and the therapeutic unbound drug in the tissues). Although drugs have only tion of the predicted clinical dose. Similarly, it concentration approximates to the target moderate affinity (potency) for these con- could result in futile lead optimization cycles concentration, the pharmacokinetics of the stituents, the impact of these constituents on with the same or different chemical series. drug will show pronounced nonlinearity drug pharmacokinetics is high because they over relatively small dose increments due to are present at high concentrations. However, Impact on in vivo studies. In vivo, different saturation of target binding. Multiple dose the impact of the drug binding to the target considerations are relevant for highly expressed studies at doses giving rise to similar con- itself is often not discussed, although the targets. The binding of the compound to its centrations as the target may demonstrate affinity of drugs for their targets will often target can have a major influence on both the this aspect of nonlinearity as an unantici- be much higher than the affinity of drugs for distribution and the CL, and can make a large pated accumulation profile of the drug. nonspecific blood and/or tissue constituents. contribution to nonlinear pharmacokinetics. • Given that relatively high doses are often Consequently, many small-molecule drug Although such TMDD is typically considered used in preclinical pharmacokinetic and discovery projects are mounted with a firm for large-molecule drugs such as monoclonal toxicology studies, it is likely that satura- biological rationale for the target, yet have antibodies, it is often overlooked in small- tion of the target will prevail at this stage little knowledge of the actual concentration molecule drug research. In many cases, this and obscure TMDD, meaning that TMDD of this protein, whether in the initial in vitro is not an issue, as there is indeed little influ- may first be encountered during clinical screens, in the subsequent in vivo animal ence of the target on the overall distribution of development. studies, or in healthy subjects or patients. small molecules, because the mass of the tar- Of course, other processes such as satura- The potential consequences of this lack of gets comprises only a tiny fraction of the mass tion of metabolism or transporter-dependent knowledge can range from unreliable con- of the body constituents that bind the drug. clearance may contribute to nonlinearity in clusions about compound potency to unusual However, for some small-molecule drugs, the addition to target binding, which will add and even highly variable pharmacokinetics target represents a higher proportion of these complexity to interpretation. Several exam- in humans. In this article, we briefly high- body constituents and thus the effect on the ples of small molecules that exhibit substan- light the underappreciated impact of pharmacokinetics can be dramatic. tial TMDD have been reported, including

CORRESPONDENCE

the endothelin receptor antagonist bosentan, a Day 1 (dose normalized) b Steady state (dose normalized) the angiotensin-converting enzyme inhibi- 100 100 tor enalaprilat, the aldose reductase inhibitor Kd =5nM Kd =5nM imirestat, the dipeptidyl peptidase 4 inhibitor linagliptin, the monoamine oxidase B inhibitor selegiline and the vitamin K epoxide 10 10 reductase inhibitor warfarin4. A recent example is the potent 11β-hydroxysteroid dehy- drogenase type 1 inhibitor ABT-384, which showed marked nonlinear pharmacokinetics 1 1 in humans that was attributed to TMDD4. At low doses (1, 2 or 4 mg) ABT-384 appeared (ng/mL/mg) Concentration (ng/mL/mg) Concentration to have a higher V than at higher doses (8, 30 or 100 mg). Of note, after repeated 0.1 0.1 0 6 12 18 24 0 6 12 18 24 dosing, disproportionately higher accumula- Time (hours) Time (hours) tion of ABT-384 occurred at the lower doses than predicted based on half-life and dosing c Day 1 (dose normalized) d Steady state (dose normalized) frequency, but not at the higher doses. These 100 100 results are consistent with the notion of a Kd =1,000nM Kd =1,000nM strong influence of the target on the pharmacokinetics, which becomes particularly obvi- ous at low doses. Once the target is saturated 10 10 by administration of a high dose or multiple low doses, the pharmacokinetics are again linear. However, depending on the extent of target saturation, the influence on the 1 1 pharmacokinetics can be highly complex5.

A simulation illustrating the effects of TMDD (ng/mL/mg) Concentration (ng/mL/mg) Concentration for a hypothetical small molecule is shown in FIG. 1. 0.1 0.1 0 6 12 18 24 0 6 12 18 24 Impact on microdosing studies. TMDD Time (hours) Time (hours) also has implications when considering Dose of drug human microdosing approaches, which are 3 mg 6 mg 12 mg 24 mg 50 mg increasingly being used in cases where the extrapolation of pharmacokinetics from ani- Figure 1 | A simulated example of target-mediated drug disposition. The top panels of the figure Nature Reviews | Drug Discovery mals and in vitro measurements to humans illustrate the pharmacokinetic consequences following single-dose (part a) and repeated-dose (part b) administration of a hypothetical small-molecule drug that has a high affinity (dissociation constant is problematic. Such studies can yield valu- (Kd) = 5 nM) for a tissue-based target that results in target-mediated drug disposition (TMDD). Dose- able information regarding human pharma- normalized plasma concentration–time profiles are shown. Increasing the dose of the drug from 3 to cokinetic properties at an early stage in drug 50 mg causes saturation of the target, resulting in a marked reduction in the volume of distribution (V), development, but success depends on phar- and a corresponding decrease in half-life (t1/2); as clearance is unaffected, this translates into clear macokinetic linearity, which is violated when dose-nonlinearity on day 1 (part a). After multiple administrations, the plasma concentration at steady TMDD occurs. For example, warfarin is very state, being dependent on clearance, shows no evidence of nonlinearity (part b). The lower panels of highly bound to albumin but even more the figure illustrate the same protocol for a less potent compound (with a 200-fold higher Kd), for which highly bound to its high-affinity, low-capacity TMDD and changes in V are not observed (parts c,d). therapeutic target vitamin K epoxide reductase. Accordingly, due to TMDD, warfarin has a much larger volume of distribution and • Knockout (and knock-in) mouse studies. highly expressed, the tissue/plasma ratio is longer half-life following a microdose com- These are an elegant way of demonstrat- expected to decrease with increasing dose9. pared with a therapeutic dose6. This clearly ing whether the target can have a profound • Comparison of single with multiple indicates TMDD can confound microdos- effect on the pharmacokinetics7. ascending doses. This provides further ing studies, potentially leading to incorrect • Co-administration of a selective inhibitor insight into TMDD as well as ruling out expectations for the pharmacokinetics at for the pharmacological target. This will other sources of nonlinear pharmaco therapeutic doses. typically result in a decrease in V with no kinetics, such as saturation of clearance effect on CL with increasing dose of the mechanisms. Typically, the nonlinear Understanding the potential for TMDD inhibitor when the target is expressed in tis- pharmacokinetic effects will be dimin- In order to gain a better understanding of sues, and an increase in both V and CL when ished after repeated dosing (FIG. 1). the potential for target-mediated effects on the target is primarily expressed in blood8. • Radioactive tracer studies. For exam- small-molecule drug pharmacokinetics, sev- • Determining tissue to plasma partition ple, the half-life of a tracer dose of eral preclinical and clinical activities can be after a dose-ranging study in preclinical 14C-warfarin was substantially shortened considered. species. For tissues in which targets are when co-administered with a high dose of

CORRESPONDENCE

unlabelled drug that saturated the target, TMDD for small molecules to be more com- 3. Levy, G. Pharmacologic target–mediated drug 10 disposition. Clin. Pharmacol. Ther. 56, 248–252 owing to a substantial decrease in V . mon than is appreciated and it might in fact (1994). • Intravenous studies. These are most suited often be overlooked. For drugs where it is 4. An, G. Small-molecule compounds exhibiting target– mediated drug disposition (TMDD): a minireview. to assess TMDD, as after oral administra- indeed important, pharmacokinetic and phar- J. Clin. Pharmacol. 57, 131–136 (2017). tion, certain absorption and first-pass macodynamic modelling and simulation can 5. Peletier, L. A. & Gabrielsson, J. Pharmacokinetic steady-states highlight interesting target-mediated clearance processes could mask TMDD. provide useful insights as to the appropriate disposition properties. AAPS J. 19, 772–786 • In vitro binding assays. For plasma-based dosage regimen. (2017). 6. Lappin, G. et al. Use of microdosing to predict targets, in vitro binding assays for the drug pharmacokinetics at the therapeutic dose: experience conducted over the therapeutic concentra- Robert A. B. van Waterschoot, Neil J. Parrott, with five drugs. Clin. Pharmacol. Ther. 80, 203–217 tion range can indicate whether any non- Andrés Olivares-Morales and Thierry Lavé are at (2006). Roche Pharmaceutical Research and Early 7. Fuchs, H., Tillement, J. P., Urien, S., Greischel, A. & Roth, W. Concentration-dependent plasma protein linear effects are occurring. Development, Pharmaceutical Sciences, Roche binding of the novel dipeptidyl peptidase 4 inhibitor BI Innovation Center Basel, Switzerland. 1356 due to saturable binding to its target in plasma Outlook Malcolm Rowland is at the Centre for Applied of mice, rats and humans. J. Pharm. Pharmacol. 61, 55–62 (2009). Pharmacokinetic Research, Manchester Pharmacy TMDD for small molecules has not received 8. Chien, J. Y., Banfield, C. R., Brazzell, R. K., Mayer, P. R. much attention and often only becomes School, University of Manchester, Manchester, UK. & Slattery, J. T. Saturable tissue binding and imirestat pharmacokinetics in rats. Pharm. Res. 9, 469–473 Dennis A. Smith is at the Department of Chemistry, apparent during clinical trials, with the (1992). potential for data misinterpretation. TMDD University of Capetown, Capetown, South Africa. 9. Cheung, W. K. & Levy, G. Comparative pharmacokinetics of coumarin anticoagulants. XLIX: can have a major influence on drug distribu- Correspondence to R.A.B.vW. Nonlinear tissue distribution of S-warfarin in rats. tion and sometimes clearance, resulting in [email protected] J. Pharm. Sci. 78, 541–546 (1989). 10. Takada, K. & Levy, G. Comparative pharmacokinetics unpredictable, nonlinear pharmacokinetics. doi:10.1038/nrd.2018.26 of coumarin anticoagulants XLIV: dose-dependent As such, consideration of TMDD is crucial Published online 23 Feb 2018 pharmacokinetics of warfarin in rats. J. Pharm. Sci. 69, 9–14 (1980). for meaningful interpretation of clinical data 1. Morgan, P. et al. Impact of a five-dimensional and could have an impact on the probability framework on R&D productivity at AstraZeneca. Competing interests Nat. Rev. Drug Discov. https://doi.org/10.1038/ The authors declare competing interests: see Web version of success in drug discovery and develop- nrd.2017.244 (2018). for details. ment, particularly with the drive for increas- 2. Smith, D. A., Di, L. & Kerns, E. H. The effect of plasma Publisher’s note 4 protein binding on in vivo efficacy: misconceptions in ing potency of drugs. Based on recent reports drug discovery. Nat. Rev. Drug Discov. 9, 929–939 Springer Nature remains neutral with regard to jurisdictional as well as observations at Roche, we consider (2010). claims in published maps and institutional affiliations.