Article pubs.acs.org/crt
Pragmatic Approaches to Determine the Exposures of Drug Metabolites in Preclinical and Clinical Subjects in the MIST Evaluation of the Clinical Development Phase Johanna Haglund,*,† Magnus M. Halldin,‡ Åsa Brunnström,† Göran Eklund,† Antti Kautiainen,† Anna Sandholm,§ and Suzanne L. Iverson∥ AstraZeneca R&D, DMPK Södertälje, SE-151 85 Södertälje, Sweden ABSTRACT: The recent stream of regulatory guidelines on the Safety Testing of Drug Metabolites by the FDA in 2008 and the ICH in 2009 and 2012 has cast light on the importance of qualifying metabolite exposure as part of the safety evaluation of new drugs and has provided a much needed framework for the drug safety researcher. Since then, numerous publications interpreting the practicalities of the guidelines have appeared in the literature focusing on strategic approaches and/or adaptation of modern analytical methodologies, e.g., NMR and AMS, for the identification and quantification of metabolites in the species used in preclinical safety assessments and in humans. Surprisingly, there are few literature accounts demonstrating how, in practice, a particular strategy or analytical method has been used to qualify drug metabolites during the safety evaluation of a drug during clinical development. At the same time as the initial FDA and ICH guideline releases, the neuroscience therapy area of AstraZeneca had a number of projects in clinical development, or approaching this phase, which gave the authors a scaffold upon which to build knowledge regarding the safety testing of drug metabolites. In this article, we present how the MIST strategy was developed to meet the guidelines. Pragmatic approaches have evolved from the experience learned in various projects in DMPK at AstraZeneca, Södertälje, Sweden. Our experience dictates that there is no single strategy for qualifying the safety of drug metabolites in humans; however, all activities should be tied to two unifying themes: first that the exposure to drug metabolites should be compared between species at repeated administration using the relative method or a similar one; and second that the internal regulatory documentation of the metabolite qualification should be agnostic to external criteria (guidelines), indication, dose given, and timing.
■
INTRODUCTION The safety of metabolites of drugs, in addition to the safety of the drug itself, must be established before a drug is considered for marketing approval by regulatory agencies. During the early development phase, and after the selection of a drug candidate, the compound is safety-tested both from an on-target (i.e., primary pharmacological target) as well as an off-target perspective in preclinical species prior to it being considered for testing in humans. During these studies, the compound is given to the animal by the intended clinical route (e.g., oral, intravenous, and inhalation), where appropriate. Furthermore, a range of doses is covered with the low end hovering just above the intended therapeutic dose and the high end inducing an adverse effect(s). The safety assessment of drugs, certainly before the first time in a human study, relies on the calculation of a safety margin, that being the fold-difference between the expected exposures to be explored in humans of the parent compound in the blood and the highest exposure of drug where no adverse effects are observed (NOAEL) in the preclinical safety studies. Furthermore, it is important to see the total exposure of the administered drug, and thus, clearance is a parameter of importance when considering and designing © 2014 American Chemical Society
clinical studies. These quantitative end points are achieved through the development and use of sensitive validated bioanalytical assays that are specific for detecting the active compound(s) in blood and/or blood plasma. One of the major mechanisms of clearance of a drug is its metabolism. After oral administration, a drug is usually converted in the GI-tract, liver, and/or extrahepatic tissues by various enzymes to a variety of metabolites. It follows then that during preclinical safety testing of drug candidates within a given animal species, it is not only the parent drug that is being tested for its on and off-target pharmacological effects but also its metabolites. Because of species-, strain-, and sometimes sexspecific isoforms of enzymes and/or capacities, each group has the potential to generate unique or disproportionate amounts of drug metabolites. Since long-term drug safety studies are not performed on human subjects per se (i.e., with toxicological end points), scientists must establish if the metabolites generated in humans have also been formed and adequately exposed in the preclinical species, and in some cases in vitro Received: December 4, 2013 Published: March 4, 2014 601
dx.doi.org/10.1021/tx400449z | Chem. Res. Toxicol. 2014, 27, 601−610
Chemical Research in Toxicology
Article
above has been suggested for compounds given to patients with life threatening diseases, e.g., late stage cancer treatment23,24 and certain types of metabolites like the chemically stable phase II varieties, e.g., some glucuronides and sulphates. Since the guidelines on MIST have been made available, there has been much discussion in the literature about how to fulfill these expectations from choice of bioanalytical methodology15,25−33 to an overall flow-chart of where these analyses fit into the scheme of drug discovery and clinical development.2,9,12−14,16,21 AstraZeneca, like other pharmaceutical companies, established an internal strategy for the safety assessment of drug metabolites that aims to satisfy both good science as well as the ICH M3 (R2) guidelines described above and which has been presented in part by Isin et al.34 In our experience, the MIST strategy is only one of many strategies taken into consideration during the planning of a drug development program and most often is not the defining strategy. While there are many published articles of MIST strategies and theoretical reasoning demonstrating the possible impact of the new guidelines, there are to our knowledge few real life, published reports of prospective applications of the various companies’ strategies and the lessons learned from them.9,25−32,35 Through the presentation of cases, the present article describes how the AstraZeneca Neuroscience ADME group deepened its understanding of the measurement of metabolite exposure and thereby developed its MIST strategy postguidelines. The actions described in the cases reflect an evolving MIST strategy, in that we were focused on trying to fulfill the regulatory guideline of the day (FDA or ICH) in an ongoing development project with the biological samples at hand. It is in this way that we view the case approaches as pragmatic, but the cases also demonstrate that we do not believe categorically in an arbitrary cutoff point for further in vivo testing of a disproportionate metabolite. Just as metabolites cannot be treated the same in a risk assessment, neither can the drug project, which we also demonstrate in the cases presented.
screens (e.g., Ames), that are used in the toxicological and safety pharmacological studies throughout drug development. The DMPK scientist has long understood the importance of comparing metabolite profiles across species in particular because of the on- and off-target toxicities that can arise due to active metabolites as well as the formation of reactive metabolites. Active metabolites against the primary pharmacological target generally follow similar regulatory requirements as parent compounds. In contrast, reactive metabolites are difficult to quantitate (because they are chemically reactive and difficult to identify), which in turn renders this class of metabolite difficult to risk assess and regulate. Metabolites are characterized as active or reactive only if they have been identified as such. Other metabolites generated that are not given those labels also need to be considered in a safety evaluation. When it comes to metabolite profiles in general, the regulatory agencies have recommended simply that metabolites are to be considered during a safety evaluation. Further complicating a safety evaluation is the issue of plasma protein binding. The unbound fraction of a parent drug is routinely measured and considered in the risk assessment of compounds. Since the accurate measurement of the free fraction of drug requires a synthetic standard and a sensitive bioanalytical assay, it is not feasible to measure the unbound fractions of potentially many metabolites, and as such, total concentrations of metabolites are most often considered. Historically, there has been a self-professed inconsistency across the pharmaceutical industry on how to conduct not only these measurements but also the resulting safety assessments of drug metabolites.1−16 This has been impeded by the length of time over which these evaluations need to occur in that it is not a one-time study; rather, studies over the course of years need to be compared, including the human studies that come after the preclinical studies. Furthermore, it is practically unreasonable to generate sensitive validated bioanalytical methods for each metabolite in order to generate the quantitative data, including plasma protein binding, needed to accurately assess the circulating exposure to metabolites. Finally, even if a metabolite could be quantified there remains the question of how much metabolite is needed to trigger other safety evaluations. To aid in the toxicological assessment of the metabolites of new chemical entities, the US Food and Drug Administration (FDA) and the International Conference on Harmonization (ICH) have offered a much-needed framework in the form of guidelines, which will be referred to as Metabolites in Safety Testing (MIST) guidelines.17−19 The overall aim is to identify and assess exposure, preferably at steady-state of the parent compound and of not only the parent compound but also circulating human metabolites and to compare these to the exposure of the same metabolites in preclinical safety species. The FDA guidance, which preceded the ICH viewpoint, encouraged the sponsor to conduct further preclinical safety studies on any human metabolite representing >10% of the parent drug exposure at steady-state and with significantly lower exposure in the preclinical species. There were strong reactions to these criteria among industry scientists, which were expressed in a stream of position articles.8−16,20−22 The ICH guideline followed, and in principle has superseded the FDA guideline, which recommends further safety-testing of a human metabolite that is observed at exposures >10% of total drug-related exposure and at significantly greater levels in humans than the maximum exposure seen in the safety studies.18,19 An exception to the
■
MIST STRATEGY The goal of assessing metabolite coverage throughout a clinical program is, or at least ought to be, to decrease the risk of a safety liability due to a disproportionately exposed metabolite in a human population. In our opinion, a MIST assessment should start with the results from in vitro metabolite profiling and identification experiments and therefore prior to the first dose in humans. While the early species comparison studies give a first peek, our strategy to date is similar to that of other published strategies where the key study in a MIST assessment is the first multiple ascending dose (MAD) study in humans. In this clinical study, exposures to metabolites at steady-state of parent are compared to their counterparts in preclinical safety species also at steady-state of parent (samples usually taken from toxicokinetic (TK) satellite animals of a pivotal preclinical toxicity study). Furthermore, in the MAD study, higher doses are normally studied compared to the remainder of clinical development. That these particular studies are central to a MIST risk assessment imposes an analytical challenge since these studies are performed without radiolabel tracers. Logistical challenges have also come to light and are highlighted in the strategy-building cases below.
■
BIOANALYTICAL METHODS
When we set out to interpret the 2008 FDA MIST guidance through the lens of the then bioanalytical paradigm in the pharmaceutical 602
dx.doi.org/10.1021/tx400449z | Chem. Res. Toxicol. 2014, 27, 601−610
Chemical Research in Toxicology
Article
Table 1. Comparison between Various Bioanalytical Techniques Used in a MIST Evaluation bioanalytical method • synthesize the metabolites and develop a bioanalytical method • ultimate but time-consuming and challenging
response factor methods
relative method (MIST screening)
• response factors are established between LC-MS and other analytical techniques, e.g., radioactivity detection • feasible but also time-consuming
• NO absolute quantification • direct exposure comparison • NO radioactivity required
Figure 1. MS chromatogram from an analysis of metabolites in an AUC pool from plasma collected within 24 h at a steady-state of the parent after repeated oral administration. undergoing extensive metabolism in vivo; thus, the anticipation was that even performing a MIST analysis using response factors would be too laborious for the time scales of early development. We indeed proved this to be true during our first postguideline MIST analysis (the first case described below) and begged the question as to whether this large effort is needed when most of the time we end up identifying covered metabolites. This led to our adoption of the relative method described by Gao et al.,25 where direct exposure comparison is performed between plasma samples collected during the human MAD and preclinical safety studies without absolute quantification. In the relative method, plasma samples from each subject within each dose group and species from the last day of administration in the MAD and safety studies, respectively, are initially pooled using the same aliquot from each subject within same sampling time. The pools
industry, it was interpreted as if any circulating metabolite above 10% of the parent in human plasma had to be properly quantified in repeated dose studies from humans and preclinical species using synthesized standards for all or the most relevant metabolites (referred to here as the Bioanalytical Method) (Table 1). This did not seem ideal from a speed and cost perspective. At that time in our company, radiolabeled ADME studies were often performed during early drug development, so we knew we could use response factors for metabolites identified in radioactivity studies (e.g., rat ADME) to semiquantify the metabolites found in phase 1 human studies (unlabeled compound). However, the availability of ADME studies did not always coincide with phase 1, and without radioactivity, we would have to, at least in that paradigm, fall back on traditional bioanalysis. Furthermore, we were accustomed to our compounds 603
dx.doi.org/10.1021/tx400449z | Chem. Res. Toxicol. 2014, 27, 601−610
Chemical Research in Toxicology
Article
of individuals are then time-proportionately pooled according to Hamilton et al.36 to obtain one sample per species and dose group. The samples are matrix-matched by the addition of blank animal plasma to the human plasma sample and vice versa. This is done to avoid differences in matrix effects on the ionization in the electrospray mass spectrometer. The assessment of response linearity is done in parallel via serial dilution of blood/plasma samples. All samples are prepared by precipitation of proteins through the addition of ice-cold acetonitrile and centrifuged. The supernatants are transferred to LC vials and consecutively analyzed on the same occasion. The mass defect filtered chromatograms (or narrow Da window) are used to determine the peak areas, which correspond to the area under the concentration−time curve (AUC) for each metabolite over a dose interval at steady-state for the parent compound (AUCss). Using the relative method, the exposure (AUC) of any metabolite can easily be evaluated regardless of its abundance. The exposure data are then used to establish exposure ratios between the human and preclinical safety species, which serve as a basis for the consideration of adequate exposure in safety species. The wording adequate exposure (or significantly exposed) relates to the guidance’s definitions, which are not a fixed number. However, differences of ≥2 fold in (mean) AUC are generally considered meaningful in toxicokinetic evaluations.19 Thus, a metabolite would generally be considered toxicologically qualified when animal exposure is at least 50% of the exposure seen in humans. In this way, with little bioanalytical effort, a set of data from repeated dose studies can be generated, including all relevant species and dose cohorts, where the conclusion about “coverage” can be evaluated regardless of changes in criteria for coverage and even therapeutic dose. Evolution of the Departmental MIST Strategy. We had the opportunity to develop our MIST strategy by using various approaches in several of our ongoing projects in different stages of drug development. The development phase is unique for every compound, and therefore we are not suggesting a generic approach. In this article, using some live cases we will demonstrate that each project needs to be dealt with on a case-by-case basis, even in the application of bioanalytical methods. There is a growing toolbox that can be utilized in various ways and orders depending on the project/compound. It is worth emphasizing that the relative method plays a major role and creates the basis for the MIST work described here. Case Studies. Case 1: When There Is Extensive Metabolism, Validation of the Relative Method Using the Response Factor Method. Compound A was approaching phase II in parallel with the 2008 release of the FDA guidance on MIST. The human ADME (hADME) study indicated that this compound was extensively metabolized in human resulting in ca. 15 significant circulating metabolites (ca. 75 in total in all studies), all with the potential of being >10% of parent (Figure 1). The majority of metabolites were glucuronides and, although analyzed and evaluated against the framework of the FDA guidance, the work in the end focused only on the phase I metabolites (M7 and M8, Table 2). We started off assessing metabolite exposure using response factors, but the high number of metabolites inspired the use of the faster and more efficient relative method. By performing both analysis methods for the same samples, we were able to validate the relative method.
The response factor method was applied to the 10 most abundant metabolites in plasma samples from pooled subjects within each time point for each species and dose group. Each sample was analyzed, and the areas under the peaks (i.e., concentration) were quantified and transformed into molar equivalents using the established response factors. The concentration in plasma for each metabolite at all time points was used for calculations of AUC. The calculated AUCs were used to conclude which metabolites fell under the FDA criteria of being >10% of parent. Of the 10 metabolites quantified, two phase I metabolites M7 and M8 were >10% of parent, but when compared with the AUC in rats and dogs, determined as described above for human samples, the exposures were significantly higher in at least one of the safety species (Table 2). Using the same human samples, time proportional pools of plasma were generated from repeated dose studies of compound A in humans and mice. All metabolites were evaluated, and the data was presented as exposure ratios (Table 3).
Table 3. Exposure Ratios between Human and Mouse Metabolites for All Metabolites found in Mana
human (50 mg) human (90 mg) rat (NOAEL) dog (NOAEL)
compound A 8.97 14.3 106 96
M7 AUC (μM·h) 8.16 11.6 124 52
AUCHuman/AUCMouse
M7 M8 M11 M21 M38 M39 M41 M42 M43 M46 M49 M50 M65 M73
0.1 0.06 0.1 3.3 0.01 0.03 1.7 0.35 2.3 0.23 0.01 0.28 6.3 20
comment
not covered but 10% of total drug related material. In this case, even though we were not intending to assess metabolite exposure in terms of the percentage of the parent (i.e., the redundant FDA guideline), both metabolites happened to be less than 10% of the parent, and hence, we could conclude that they were also less than 10% of total drug related material. Had the metabolites been >10% of the parent, we would not have been able to say if they were 2.5
0.13 0.08 1 1
0.08 0.0 0.14 0.14
0.06 0.03 0.25 0.33
M
F (NOAEL)
dog, 50 mg/kg M
F
Human Dose Cohort (12.5 mg/day) 0.02 0.02 0.25 0.25
