Chem. Res. Toxicol. 2009, 22, 243–256
243
PerspectiVe Predicting Circulating Human Metabolites: How Good Are We? Shelby Anderson, Debra Luffer-Atlas, and Mary Pat Knadler* Department of Drug Disposition, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285 ReceiVed October 31, 2008
The FDA issued a guidance on the safety testing of metabolites in February 2008, in which they stated that metabolites of concern are those that are detected at levels greater than 10% of the systemic exposure of the parent at steady state. This has presented many challenges in determining the circulating human metabolites at an early stage of development. The intention of this perspective is to address the question of how effective in vitro metabolism and early exploratory clinical data are in predicting the circulating metabolites from both a qualitative and a quantitative perspective. To this end, data were reviewed from 17 molecules in the Lilly portfolio for which there were in vitro data and a radiolabeled study in humans. Twelve example cases are presented in detail to demonstrate trends for when in vitro data adequately predicted in vivo (41%), when in vitro data underpredicted the circulating metabolites (35%), and when in vitro data overpredicted the circulating metabolites (24%). In addition, cases that present special challenges due to very low levels of the circulating parent or long half-lives of the parent and/or metabolites are presented. The trends indicate that the more complex the metabolism, the less likely the in vitro data were to predict the circulating metabolites. The in vitro data were also less predictive for N-glucuronidations and non-P450-mediated cleavage reactions. Although the in vitro data were better at predicting clearance pathways, the data set often failed to predict the quantity of metabolites, which is needed in consideration of whether or not a “disproportionate” metabolite may be circulating in human plasma. Contents 1. Introduction 2. Predicting Circulating Human Metabolites 3. In Vitro Data That Predict Circulating Human Metabolites 3.1. Case Study 1 3.2. Case Study 2 3.3. Case Study 3 4. In Vitro Data That Underpredict Circulating Human Metabolites 4.1. Case Study 4 4.2. Case Study 5 4.3. Case Study 6 5. In Vitro Data That Overpredict the Role of Oxidative Metabolism 5.1. Case Study 7 5.2. Case Study 8 5.3. Case Study 9 6. Quantification of Human Metabolites Is Often Challenging 6.1. Case Studies 10 and 11 6.2. Case Study 12 7. Conclusions
1. Introduction 243 244 245 246 247 247 248 248 249 251 251 251 252 253 253 253 254 254
In February 2008, the FDA issued the Guidance for Industry on the Safety Testing of Drug Metabolites (1). The guidance only applies to small molecule nonbiologic drug products, so the discussion is limited to such therapeutics. Prior to the issuance of this guidance and the draft version that preceded it (2), it was well-accepted in the pharmaceutical industry that the demonstration of any level of a metabolite in an animal species was considered reasonable demonstration of nonclinical metabolite coverage. Following the formal publication of the guidance, the new standard safety testing paradigm is considered to demonstrate that the human metabolic plasma profile is covered by the plasma profile in the animal species used in the toxicology studies. In other words, as long as one or more of those animal species has been exposed to approximately equal or greater concentrations of the metabolite to that found in human plasma, the metabolites are considered to have been sufficiently tested for safety. When this cannot be demonstrated, safety testing of the metabolites found circulating in humans may be needed. As stated in the guidance, human metabolites that may raise a safety concern are those formed at greater than 10% of the systemic exposure of the parent drug at steady state in humans. Several reviews and debates have been published on the issue of metabolite exposure threshold, necessitating additional nonclinical studies on the safety of metabolites and what these safety studies should be (3-10). * To whom correspondence should be addressed. Tel: 317-276-0711. Fax: 317-433-6432. E-mail:
[email protected].
10.1021/tx8004086 CCC: $40.75 2009 American Chemical Society Published on Web 01/12/2009
244
Chem. Res. Toxicol., Vol. 22, No. 2, 2009
Identification of potentially major metabolites as well as information on metabolites circulating in humans have clearly become a critical focus in the development of drugs for many reasons. Besides the recent focus on safety testing of drug metabolites, it is well-known that coverage of significant human metabolites must be demonstrated in at least one of the rodent species (rat or mouse) that are used for 2 year carcinogenicity testing. In addition, identification of major and/or active metabolites is considered necessary for potential monitoring of the definitive clinical QT cardiac safety study. Elucidation of metabolic clearance and excretory pathways (e.g., through the various cytochrome P450 enzymes or via renal excretion) is a predecessor to the design of many drug-drug interaction, hepatic impairment, and renal impairment studies that form the basis of the clinical pharmacology package for the drug label. Related to this is the recent focus on pharmacogenomics, whereby identification of the major clearance pathways and sources of genetic variability in various populations can often be used to explain the pharmacokinetics and pharmacodynamics in diverse groups of patients. Specifically related to the issue of metabolite safety testing, it becomes advantageous to know as early as possible if any metabolites circulate in humans at or above the 10% level and whether or not the animal species have been exposed to these metabolites. When a metabolite is unique to humans or is present in human plasma at concentrations higher than that circulating in the animals, the metabolite is termed a “disproportionate” metabolite. A recent review by Luffer-Atlas (11) has illustrated a prevalence-based and relevance-based approach to these activities and the timing that was used to establish whether metabolites were either major or unique to humans. The pharmaceutical industry standard to identify and assess metabolic profiles in animals and humans is through the administration of radiolabeled drug, but there is not general agreement on the ideal timing of these studies relative to the submission of a new drug application (NDA). In the general context of drug development, a specific margin for coverage of human metabolites vs animal exposure has not been consistently communicated by regulators (12-16). However, within the context of the Guidance for Safety Testing of Drug Metabolites (1), coverage for disproportionate metabolites is considered sufficient if the metabolite circulates in at least one animal species at concentrations approaching those found in human circulation at pharmacologically relevant doses. This may sound relatively straightforward to determine, but often, important a priori decisions are made concerning which metabolites are likely to be major and whether they are likely to be covered in animals prior to the definitive studies that establish the metabolic pathways in human (typically the human radiolabeled drug ADME study).
2. Predicting Circulating Human Metabolites Early in the drug discovery and development process, in vitro studies involving incubation of test compound in liver microsomes, hepatocytes, or liver slices are usually conducted to generate metabolites and compare these metabolic profiles between animals and humans. Although these studies are valuable, the in vitro metabolic stability and profile do not always accurately predict the in vivo metabolic profile, especially from a quantitative standpoint. Similarly, many researchers examine plasma, serum, and/or urine from the initial phase 1 clinical studies in an attempt to obtain metabolic data earlier than the conduct of the radiolabeled study. This approach usually adds to the knowledge of the human metabolic profile and is
Anderson et al.
often considered the next step after in vitro metabolism but is again rarely quantitative. Both in vitro metabolic studies and exploratory metabolism in the initial phase 1 studies have their advantages and limitations. The first chance to evaluate the predictive power of the in vitro systems occurs with dosing of the first preclinical pharmacokinetic or toxicity studies. Plasma, urine, and/or bile obtained from preliminary rodent and nonrodent toxicity studies are often analyzed by LC/MS or LC/UV techniques. These samples are typically profiled to evaluate clearance pathways and to potentially identify prevalent circulating metabolites. The qualitative identification of human metabolites from these early studies is quite reliable, allowing for the prediction of clearance pathways. Quantitative profiling is, however, more challenging as LC/UV and LC/MS signals do not provide a robust measure of the relative contribution of a metabolite to the drug-related material in circulation in the absence of a standard for the relevant metabolite(s). Modifications of the chemical structure can lead to significant changes in the UV absorbance and/or MS ionization potential relative to the parent drug. Preliminary data from both in vitro studies and preclinical toxicity studies often provide an invaluable, early assessment of the potential overall metabolic scheme. However, these assessments often miss the mark in terms of predicting a priori major circulating human metabolites and estimating the levels of those metabolites in human plasma. These same limitations also apply to the first human dose of a new molecular entity. First human dose studies are typically single and multiple dose escalation studies, which are intended to evaluate the safety, tolerability, and pharmacokinetics of the clinical candidate in humans. These studies provide for the first in vivo evaluation of metabolic pathways in human matrices and, in particular, their role in human plasma. As is done for the preclinical studies, human matrices (typically plasma and urine) are profiled for metabolites using LC/UV and/or LC/MS, and as in the preclinical species, these data often provide an invaluable insight into potential metabolic pathways. However, at best, the data are semiquantitative. Metabolites with poor ionization potential and/or low UV absorption will be underpredicted, and metabolites with increased MS sensitivity can be overestimated. In addition, metabolites can be formed that are not predicted from the in vitro experiments or animal experiments; thus, these metabolites may not be detected. Despite these limitations, profiling metabolites in these studies typically provides valuable insight into the human metabolic pathways and can often even provide a means to accurately predict the likely major circulating human metabolites. Such restrictions around quantitation of metabolites must be considered when making decisions based on these preliminary profiles. In some cases, the preponderance of the evidence (in vitro and in vivo) may clearly point to one or more metabolites that may be worthy of follow-up, including the preparation of a synthetic standard to test for pharmacological activity and/or to validate a bioanalytical assay to quantify metabolite(s) in the plasma of the preclinical species and humans. In other cases, selection of major metabolites may not be possible as only numerous small peaks are noted with no clear manner to select the few metabolites of greatest concern. To overcome the quantitative limitations inherent when synthetic standards are not available, an approach has been described that estimates concentrations of circulating human metabolites prior to conduct of the human ADME study (17). This radiometric calibration technique creates a “calibration standard” by comparing the ratio of radioactive and mass
PerspectiVe
spectrometric (MS) signals obtained from a biologically derived metabolite (e.g., circulating in animal plasma and/or generated in in vitro incubations). Subsequently, the concentration of a presumably identical nonlabeled metabolite in human plasma can be estimated by comparing its MS response to the metabolite standard “calibrant” signal ratio. While some important assumptions must be made concerning the radiopurity of the peak relating to the metabolite standard, this technique does offer the ability to get a ballpark estimate of circulating human metabolite levels in early phase clinical trials before synthetic standards are made available. On a more qualitative front, high resolution or fractional mass filtering LC/MS techniques (18, 19) can be useful in analyzing plasma from early phase clinical trials. This approach takes advantage of the fact that accurate mass measurement of the parent drug provides a mass defect component that can be used as a “signature” for drug-related ions. Fractional mass filtering does not provide quantitative information; however, it will effectively filter for both anticipated and unanticipated mass changes. Levels of the corresponding metabolites can then be estimated using the radiometric calibration technique described above, all prior to the human ADME study with radiolabeled drug. Alternatively, the technique of accelerator mass spectrometry (AMS) has been applied early in drug development to obtain information about pharmacokinetics and circulating drug-related metabolites well before the traditional human radiolabel ADME study can be conducted (20). The main advantage of AMS as compared to conventional radiometric methods is that nCi doses are administered to humans. From a safety perspective, this dose is not considered to be “radioactive”, and conventional animal dosimetry studies are not required to support the study. The primary disadvantage is that time-consuming fraction collection is required to provide a radiometric time course and conventional LC/MS techniques must be conducted in parallel to enable metabolite identification. In recognition of the growing popularity of this approach, the FDA has issued a guidance concerning application of microdose techniques in the so-called “exploratory IND” format (21). Ultimately, definitive metabolic data for both preclinical species and humans will be obtained via conduct of a radiolabel study. The radiotracer allows metabolites to be profiled in a more or less quantitative manner, allowing direct comparison of the relative exposure to the total circulating radioactivity, circulating levels of parent drug, and between species. Radiolabel studies are typically conducted in the preclinical species approximately in parallel with phase 1 clinical work. Timing of the radiolabel human study varies significantly across the industry and can occur in parallel to phase 1 (relatively early in drug development), in phase 2 (prior to the end of phase 2 meeting with the FDA), or in phase 3 (late in development). As described in depth in a previous review (11), the Lilly strategy has placed the conduct of the human radiolabel ADME studies as early in phase 2 as practical, understanding that in some cases this study is being conducted at risk before proof of concept has been achieved in the clinic. Although these data are not a mandatory part of the end of phase 2 submission, Lilly has affirmatively acknowledged that they are critical to understand metabolite coverage and thus safety assessments. The definitive human ADME profile is also necessary to drive study design for subsequent in vitro metabolic assessments of the enzymes and transporters involved in clearance pathways as well as numerous so-called “clinical pharmacology” studies (including drug interactions, renal and hepatic impairment, thorough
Chem. Res. Toxicol., Vol. 22, No. 2, 2009 245
QT studies, etc.), which assess both the intrinsic and the extrinsic factors that influence the pharmacokinetics of the drug. These studies are typically considered an integral part of the package, and the results are included in the product “label”. For the following analysis, 17 ongoing and past drug development programs across the Lilly portfolio were reviewed. All of these programs completed human ADME studies with a 14Clabeled drug (between 1993 and 2008). Twelve of these programs are presented in some detail as part of this perspective. The clinical development candidates from these 12 programs have all been blinded to protect the confidentiality of the assets, but key metabolic profiling data have been shared when relevant to the case. The predictability of the in vitro and exploratory human data for each of these programs has been considered to determine the predictive nature of these data in a retrospective manner. Programs were categorized according to how successfully their human in vitro data predicted major circulating human metabolites that were ultimately identified in the human ADME study with radiolabeled drug. This approach was selected to specifically address safety coverage concerns consistent with the FDA Guidance for Industry on the Safety Testing of Drug Metabolites (1). Because the FDA Guidance is primarily concerned with circulating human metabolites, this perspective will focus mainly on the correlation between the in vitro metabolic profiles and the metabolites ultimately identified in human plasma following administration of a radiolabeled drug in a human metabolism and disposition study. In this analysis, Lilly molecules were categorized as having (1) in vitro data that were essentially predictive of major circulating human metabolites, (2) in vitro data that underpredicted the extent of metabolites in human circulation, or (3) in vitro data that overpredicted the role of oxidative metabolites in human circulation. In addition, some molecules have been categorized separately as those that do not really conform to the FDA recommendations to demonstrate preclinical coverage for human metabolites >10% of circulating parent (at steady state) because of special considerations (e.g., prodrugs and those with long-lived metabolites in circulation). Individual case studies are presented and discussed in the context of the final FDA guidance. Because it was challenging to make a definite classification in each case as to whether the in vitro data were predictive or not, applied were the following criteria: A data set was assumed to be essentially predictive of human metabolites if the in vitro metabolites were observed to any degree in human circulation. Alternatively, if major circulating human metabolites were observed that had never been identified in any in vitro incubation, this was defined as an underpredictive data set. Finally, if many in vitro metabolites were observed that did not appear in human plasma at any level, then this was considered to be an overpredictive case. This means that even if the in vitro metabolites were found to be excreted into urine and/or feces but were not detected in human plasma at any level, the in vitro data set was still viewed as being overpredictive.
3. In Vitro Data That Predict Circulating Human Metabolites In some cases, in vitro and exploratory human data can be quite predictive of human pathways, based on the identification of major circulating metabolites. This section describes in some detail three programs where the in vitro assessment would have provided teams with the ability to reasonably predict the major circulating human metabolites prior to conducting the human metabolic study. Of the 17 programs evaluated as part of this
246
Chem. Res. Toxicol., Vol. 22, No. 2, 2009 Scheme 1. Proposed Metabolic Pathways for LY01
perspective according to the strict criteria outlined above, seven of the programs fit into this category. 3.1. Case Study 1. LY01 was a phase 2 clinical candidate. A clinical trial was conducted with 14C-LY01 using a single radiocarbon dose of 100 µCi per subject in six healthy male subjects to evaluate the human metabolism and disposition of this investigational drug. LY01 was labeled with 14C in two positions based on metabolic cleavage routes that were identified in the preclinical species; one of the label sites was on the cyano moiety. This molecule demonstrated three “major” circulating drug-related peaks: parent (accounting for ∼24% of total circulating radioactivity), thiocyanate (SCN, ∼31% of total circulating radioactivity), and a P450-mediated oxidative product, a benzylic alcohol (M9, ∼35% of total circulating radioactivity) (Scheme 1). According to the 2005 FDA Draft Guidance on Safety Testing of Drug Metabolites (2), which was in effect at the time that this molecule was being clinically evaluated, only the SCN and M9 metabolites were greater than 10% of the total circulating drug-related material as measured by the area under the curve (AUC) of metabolite relative to the AUC of total radioactivity in plasma. In accordance with the guidance, these metabolic pathways were evaluated to determine whether exposure to these “major” human metabolites was similar or higher based on circulating levels in the preclinical species used for the safety assessment of LY01. Coverage for SCN (resulting from a cyano moiety within the LY01 molecule) and M9 was assessed in the preclinical species (mouse, rat, and monkey). AUC exposures in the animals were estimated for SCN and M9 based on data from previous preclinical single dose radiolabel studies with LY01 (doses ranging from 5 to 20 mg/kg). SCN levels in all of the preclinical toxicity species significantly exceeded those reported in the human 14C trial, even at doses much lower than those administered in the chronic toxicity studies. In addition, because of questions from the FDA concerning the formation of SCN, levels of SCN were quantitated in a series of human clinical pharmacology studies after both single and multiple doses. SCN exposure is common across the human population due to dietary and other environmental sources of cyanide (22). LY01 administration was determined not to increase the background levels of SCN in the evaluated trials. For these reasons, SCN production resulting from the administration of LY01 was considered safe. All species (mouse, rat, and monkey) were also found to produce significant levels of the M9 metabolite; however, total exposure in humans exceeded levels reported in at least the rat and monkey single dose radiolabel studies (dosed at 5 mg/kg). A recommendation was made to evaluate M9 exposure across
Anderson et al.
species using a validated assay; however, clinical development of LY01 was halted before this assessment was completed. In accordance with the final FDA Guidance on Safety Testing of Drug Metabolites (1), which recommends assessing safety coverage for human metabolites that represent g10% of the plasma AUC exposure of the parent drug at steady state, two additional metabolites of LY01 may have qualified as “major”, N-dealkylated LY01 (M5) and the glucuronide conjugate of M9 (M6). This evaluation was, however, roughly estimated based on the height of the corresponding radiolabel peaks in the radiocount vs time chromatograms (collected using TopCount technology) over the time course of 24 h. Because of the relatively low counts of radioactivity in plasma, a rigorous quantitation of the AUC exposure of these metabolites relative to parent is not feasible given the available data (Figure 1). Whether a rigorous assessment would be feasible would depend on whether or not the plasma could be significantly concentrated for additional metabolite profiling. For some molecules, a 10-20-fold concentration factor is feasible with good extraction efficiency (>85-90%). Here, LY01 accounts for approximately 24% of the circulating radioactivity. To meet the 2008 Guidance criteria, it would be necessary to quantitate metabolites accounting for approximately 90% of the dose). In vivo, the predominant metabolites of both LY10 and LY11 were glucuronide conjugates on phenolic hydroxyl groups. None
254
Chem. Res. Toxicol., Vol. 22, No. 2, 2009
of the conjugates were acyl glucuronides, but in both cases, there was evidence that the conjugate was hydrolyzed back to the parent molecule. In addition to conjugates, there was an oxidative metabolite in plasma for LY11 formed by Odemethylation. This metabolite also formed ether glucuronide conjugates. Radiodetection was not sensitive enough to adequately characterize the parent or the oxidative metabolite, so percentages were based on LC/MS/MS methods to quantitate both the parent and the O-desmethyl metabolite. The glucuronide conjugates in plasma of both LY10 and LY11 were characterized after enzymatic hydrolysis. Analysis of fecal samples proved to be misleading, as the majority of the fecal radioactivity was due to the presence of LY10 or LY11, but this was shown to be due to hydrolysis of their respective conjugates and not due to poor absorption. In vitro studies with LY10 utilized human liver microsomes. When UDPGA was added to the incubations at 14C-LY10 concentrations of 100 µM, two large peaks were found in the radiochromatogram that corresponded to two separate glucuronide conjugates. These corresponded to the conjugates found in vivo, except that the ratio of which conjugate predominated was different. In vitro studies with 14C-LY11 were conducted utilizing human liver slices. These experiments suggested limited metabolism of LY11, as 78-96% of the radioactivity remained as parent. The two metabolites that were detected were a glucuronide conjugate and a sulfate conjugate. The slices underpredicted the extent of metabolism and suggested that sulfation was much more prevalent than it actually was in vivo. As the major metabolites of both LY10 and LY11 were ether glucuronide conjugates, there should not be a concern for safety according to the 2008 FDA guidance, and the conjugates were found in plasma from the animal species. These two examples illustrate, however, that it may not be feasible to detect metabolites that circulate at levels of 10% of parent when the parent circulates at very low levels and accounts for only a small fraction of the circulating radioactivity. 6.2. Case Study 12. LY12 was an investigational compound in phase 2, which possesses a chiral center (S-enantiomer). A human liver slice incubation with LY12 (50 µM) demonstrated one principal metabolic pathway, featuring para-hydroxylation of a terminal phenyl ether group (M2), followed by sulfate conjugation of the phenol (M1). In addition, there were trace levels of two other metabolites, M7 (dehydrogenation) and M9 (dehydrogenation plus oxidation plus sulfation). Exploratory human metabolism work was conducted as part of the initial human safety assessment study, and three metabolites were identified in human plasma: M1, M2, and M3, the last of which was an O-glucuronide metabolite of LY12. A clinical trial was subsequently conducted with 14C-LY12 in six healthy male subjects, which demonstrated poor recovery, with only 67% of the dose recovered over 1296 h (54 days). LY12 accounted for the majority of the radioactivity (∼85%) within the first 24 h and persisted as the major circulating component up to 336 h postdose. In addition, three metabolites were identified in plasma, two of which were the in vitro metabolites previously identified (M1 and M2). Interestingly, the most abundant metabolite observed in plasma was the result of an enzymatic chiral inversion of LY12 to the R-enantiomer, which had not been observed in the liver slice incubation. Even more surprising was the fact that M3, which had been observed in plasma of subjects administered unlabeled drug and analyzed using LC/MS, was not observed as a circulating metabolite in any of the subjects in the human ADME study with 14C-LY12.
Anderson et al.
Becasue of the long observed half-life of the parent compound and its circulating human metabolites, this made it particularly challenging to determine whether these metabolites would be considered “disproportionate” according to the definition in the recent FDA Guidance (1). Specifically, it was noted that while parent drug concentrations declined very gradually in plasma over time, the concentrations of M1 did not decline as quickly; thus, this metabolite might be considered as a major metabolite (relative to the AUC of parent drug at steady state). For example, at 336 h postdose, LY12 accounted for 39% of the total radioactivity, whereas M1 still represented as much as 23% of the total. Thus, this case study demonstrates that a compound with a long half-life in conjunction with even longer-lived circulating metabolites can provide inconsistent decisions on which metabolites, if any, may exceed the FDA-recommended 10% action limit for a disproportionate metabolite.
7. Conclusions Given the data presented above, along with the data for programs that were not detailed herein, the in vitro data were predictive of the circulating human metabolites 41% of the time, were underpredictive 35% of the time, and were overpredictive 24% of the time. Table 1 presents the previously discussed data in a tabular format. On the basis of our criteria of predicting circulating metabolites in humans, the success rate is 41%, which is smaller than some other researchers have concluded based upon prediction of metabolism or clearance pathways in general. In addition, these success rates were achieved by comparing single dose human metabolite profiles with in vitro metabolism data. Because the final version of the FDA guidance mandates coverage to be established under steady-state conditions (1), it is conceivable that the proportion of the metabolites observed at steady state could be somewhat greater than after a single dose, and thus might result in a further reduced success rate for in vitro-in vivo metabolite correlations. In this discussion, we have focused on circulating metabolites, as these are the criteria in the FDA guidance on safety testing of metabolites. Our data demonstrate how difficult it is to predict both qualitatively and quantitatively circulating metabolites from in vitro data, as there are many factors that determine the extent to which a metabolite is formed and whether a metabolite circulates once formed. In vitro systems, whether microsomes, hepatocytes, or slices, are isolated systems that do not have all of the complexities of the multiorgan systems working in concert as in humans. In vitro metabolism assessments are typically conducted under standard conditions, including substrate and protein concentrations, and, as such, are not optimized for a particular conversion. The data derived from these in vitro systems can be influenced by how much effort went into optimizing the system for a particular compound. Concentration of protein, substrate, cofactors present, and time of incubation can all impact the outcome. In the data presented, the in vitro systems were “standardized” and were not optimized for each individual compound; thus, in some cases, there was little turnover, which may have adversely influenced the predictions. With more customized assays and the ever-more sensitive mass detection technologies, in vitro systems may become more accurate at predicting clearance pathways, but they are unlikely to accurately predict major human circulating metabolites due to all of the complexities that contribute to whether a metabolite circulates or is rapidly cleared once formed. Our data also demonstrate that in general, the more complex the metabolic pathways, the more difficult the ultimate metabo-
PerspectiVe
Chem. Res. Toxicol., Vol. 22, No. 2, 2009 255
Table 1. Summary of Major Circulating Human Metabolites and Primary Clearance Pathways from Human 14C Studies for Drugs across the Lilly Drug Development Portfolio drug LY01 base
“major” circulating metabolites
M9 M2 M3 M4
LY03 base
M8 M6
N-oxide keto reduction and N-oxide N-glucuronide 4-OH + glucuronide
LY10 base
4′-O-glucuronide
ether-glucuronide
LY11 base
6-O-glucuronide O-glucuronide
ether-glucuronide ether-glucuronide
LY12 acid
R-enantiomer
chiral inversion
M1 M2
+ O + sulfate +O
LY04 base
M6
M7 M9 M10 LY05 base
M3 M7
LY06 neutral
M15 M26 M2 M3
LY07 base
M11
LY08 base
N-glucuronide
LY09 acid
M5
in vitro matrix
in vitro data essentially predictive of major human circulating metabolites SCN metabolism: oxidation and hepatocytes glucuronidation N-dealkyl benzylic + O + glucuronide benzylic + O keto reduction metabolism: oxidation, microsome reduction
SCN M5 M6
LY02 base
primary clearance pathway(s)
metabolite descriptor
metabolism: oxidation and glucuronidation metabolism: glucuronidation
M9 (N-glucuronide with keto reduction); M2, M3, M4, and M8
M2 4-OH, M6
microsome with UDPGA
4′-O-glucuronide
metabolism: glucuronidation
slices
6-O-glucuronide O-glucuronide
metabolism: oxidation, reduction, conjugation (glucuronidation and sulfation)
slices
O-sulfate + O + sulfate
M12 + O; M14 + O
M23 (NdesCH3) M6, M7, M8 + O M22 + O trace metabolites only
low turnover; +O major route
metabolism: hydrolysis and oxidation
in vitro data that overpredicted the role of oxidative metabolites on human circulation N-glucuronide renal/biliary parent microsomes slices (>60% dose); metabolism: glucuronidation
acyl-glucuronide
M5, M6, M9, and N-glucuronide
microsome slices
in vitro data that underpredicted the role of metabolism on human circulation 4-OH + glucuronide metabolism: oxidation and microsomes slices conjugation (glucuronidation, sulfation, methylation) CH3 cathecol + sulfate cathecol + glucuronide CH3 cathecol + glucuronide +O renal parent (∼20% dose) microsomes hepatocytes slices + O-2H metabolism: oxidation and glucuronidation OdesCH3 + glucuronide N-carbamate glucuronide amide hydrolysis renal parent (>40% dose) microsomes slices benzylic + O
significant in vitro metabolites
metabolism: glucuronidation and oxidation renal/biliary parent (>80% dose)
lite is to predict. For example, when a metabolite is oxidized and then conjugated, microsomal incubations may not have contained all of the necessary cofactors and hepatocytes may not have been as efficient in the conjugation of the oxidized products. The more steps needed to reach the ultimate circulating metabolite, the more difficult it was to predict. Metabolism pathways such as direct N-glucuronidation were also more difficult to predict, as were pathways resulting in a non-P450mediated cleavage reactions (such as ester or amide hydrolyses) or pathways catalyzed by extrahepatic enzymes. It was some of the more difficult to predict metabolites that were often found to be “disproportionate” metabolites. In general, these analyses found that the in vitro data were much better at predicting clearance pathways than “disproportionate” metabolites. In vitro data can be invaluable in predicting clearance pathways for selection of the nonrodent toxicology
microsomes slices
low turnover: M2 and M4 + O; M5 (NdesCH3) M1 + O + sulf; M11 (N-glucuronide) M5 (NdesCH3) N-desCH3; +O; N-oxide low turnover: M5 (acyl-glucuronide); M9 (CH3 to COOH); M3 + O
species and for developing a pharmacogenomic strategy. In each of these cases, these data play a critical role when deciding to bring a molecule forward in drug development. In both cases, however, the in vivo radiolabeled studies are required to confirm the decisions made using the in vitro data. Until the radiolabeled studies are conducted, the early decisions are made at risk. The in vitro data are also an important precursor to the exploratory metabolite work conducted using plasma and urine samples from the first human dose studies. The in vitro work allows identification of possible metabolites so that the in vivo samples can be more effectively profiled for the predicted metabolites. It cannot be overemphasized, however, that these types of analyses are qualitative, since there is no radiotracer in the samples and the UV and MS spectra may be different for the parent and metabolites. Expecting the in vitro experiments to predict the quantities of metabolites in the circulation
256
Chem. Res. Toxicol., Vol. 22, No. 2, 2009
is misguided and will frequently result in false expectations. The FDA guidance on safety testing of metabolites requires both qualitative and quantitative assessments, so the final answer will not be known until the human radiolabeled studies are conducted. To answer the initial question “How good are we in predicting the human circulating metabolites?”, it appears that although progress has been made with in vitro and exploratory studies, it is not possible to predict both qualitatively and quantitatively which metabolites will circulate in humans at >10% of the parent. In vitro and exploratory data provide the starting points, providing project teams with an early assessment of whether metabolism is likely to play a large or small role in the drug’s overall clearance, as well as a preliminary read on what may be major circulating metabolites. There is clear value in both in vitro experiments and in vivo exploratory analyses, but the definitive answer remains in the radiolabeled studies. Finally, the recommendation of the 2008 FDA guidance to provide safety coverage for metabolites representing greater than 10% of the systemic exposure of parent at steady state in humans is a significant challenge for the pharmaceutical industry. Even for drugs with significant circulating parent drug, operational challenges can be significant. Moreover, these challenges are even more problematic for certain classes of drugs, including issues associated with prodrugs, cases where the parent circulates at very small quantities, and cases where the metabolites have much longer half-lives than the parent. If parent circulates at very low levels, there is theoretically an unlimited number of metabolites that would meet the guidance criteria of greater than 10% of parent and would, thus, theoretically require follow-up. Although in vitro systems play an incredibly valuable role in predicting human metabolites, clearance pathways, and crossspecies profiles, the human plasma profiles of drug candidates are not routinely predicted by in vitro systems. Even with newer approaches and technologies such as the use of radiometric calibrants to estimate metabolite exposure in nonradiolabeled studies and high resolution mass filtering to identify unanticipated metabolites, reliable quantitative data concerning important circulating metabolites typically result from the human radiolabel ADME study. As suggested by the FDA guidances on both the safety testing of metabolites (1) and the drug-drug interactions (23), it is concluded that radiolabeled studies in humans, and thus, actual circulating metabolite data, are best obtained as early as practical in the drug development program to allow appropriate time to respond to these data prior to large scale testing of a drug. Ultimately, no in vitro or exploratory data can substitute for the definitive radiolabel assessment, and as such, any decisions made prior to the conduct of a human radiolabel ADME study are made at risk and may need to be reevaluated once the definitive data are available. Acknowledgment. We thank all members of the Drug Disposition Department at Eli Lilly and Company, both present and past, whose work has been described or was included in our survey.
References (1) Food and Drug Administration (2008) Guidance for Industry: Safety Testing of Drug Metabolites, FDA, Rockville, MD. (2) Food and Drug Administration (2005) Guidance for Industry (Draft): Safety Testing of Drug Metabolites, FDA, Rockville, MD.
Anderson et al. (3) Baillie, T. A., Cayen, M. N., Fouda, H., Gerson, R. J., Green, J. D., Grossman, S. J., Klunk, L. J., LeBlanc, B., Perkins, D. G., and Shipley, L. A. (2002) Drug metabolites in safety testing. Toxicol. Appl. Pharmacol. 182, 188–96. (4) Smith, D. A., and Obach, R. S. (2005) Seeing through the MIST: Abundance versus percentage. Commentary of metabolites in safety testing. Drug Metab. Dispos. Fast Forward, DOI: 10.1124/ dmd.105.005041. (5) Davis-Bruno, K. L., and Atrakchi, A. (2006) A regulatory perspective on issues and approaches in characterizing human metabolites. Chem. Res. Toxicol. 19, 1561–1563. (6) Guengerich, F. P. (2006) Safety assessment of stable drug metabolites. Chem. Res. Toxicol. 19, 1559–1560. (7) Humphreys, W. G., and Unger, S. E. (2006) Safety assessment of drug metabolites: characterization of chemically stable metabolites. Chem. Res. Toxicol. 19, 1564–1569. (8) Prueksaritanont, T., Lin, J. H., and Baillie, T. A. (2006) Complicating factors in safety testing of drug metabolites: Kinetic differences between generated and preformed metabolites. Toxicol. Appl. Pharmacol. 217, 143–152. (9) Smith, D. A., and Obach, R. S. (2006) Metabolites and safety: What are the concerns, and how should we address them? Chem. Res. Toxicol. 19, 1570–1579. (10) Naito, S., Furuta, S., Yoshida, T., Kitada, O., Unno, T., Ohno, Y., Kawamura, N., Kurokawa, M., Sagami, F., Shinoda, K., Nakazawa, T., and Yamazaki, T. (2007) Current opinion: Safety evaluation of drug metabolites in development of pharmaceuticals. J. Toxicol. Sci. 32 (4), 329–341. (11) Luffer-Atlas, D. (2008) Unique/major human metabolites: Why, how, and when to test for safety in animals. Drug Metab. ReV. 40, 447– 463. (12) International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (1997) Guideline S1b: Testing for Carcinogenicity of Pharmaceuticals, ICH Steering Committee. (13) International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (2005) Guideline S7B: The Nonclinical Evaluation of the Potential for Delayed Ventricular Repolarization (QT Interval Prolongation) by Human Pharmaceuticals, ICH Steering Committee. (14) International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (2006a) Guideline S1C(R2) [Revision of ICH S1C and S1C(R1)] Dose Selection for Carcinogenicity Studies of Pharmaceuticals & Limit Dose, ICH Steering Committee. (15) International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (2006b) Guideline M3(R2): Revision of ICH M3(R1): Maintenance of the ICH Guideline on Nonclinical Safety Studies for the Conduct of Human Clinical Trials for Pharmaceuticals, ICH Steering Committee. (16) Food and Drug Administration (2002) Guidance for Industry: Carcinogenicity Study Protocol Submissions, FDA, Rockville, MD. (17) Yu, C. P., Chen, C. L., Gorycki, F. L., and Neiss, T. G. (2007) A rapid method for quantitatively estimating metabolites in human plasma in the absence of synthetic standards using a combination of liquid chromatography/mass spectrometry and radiometric detection. Rapid Commun. Mass Spectrom. 21, 497–502. (18) Tiller, P. R., Yu, S., Bateman, K. P., Castro-Perez, J., McIntosh, I. S., Kuo, Y., and Baillie, T. A. (2008) Fractional mass filtering as a means to assess circulating metabolites in early human clinical studies. Rapid Commun. Mass Spectrom. 22, 3510–3516. (19) Zhu, M., Ma, L., Zhang, D., Ray, K., Zhao, W., Humphreys, W. G., Skiles, G., Sanders, M., and Zhang, H. (2006) Detection and characterization of metabolites in biological matrices using mass defect filtering of liquid chromatography/high resolution mass spectrometry data. Drug Metab. Dispos. 34, 1722–1733. (20) Lappin, G., and Stevens, L. (2008) Biomedical acceleratory mass spectrometry: recent applications in metabolism and pharmacokinetics. Expert Opin. Drug Metab. Toxicol. 4 (8), 1021–1033. (21) Food and Drug Administration (2006) Guidance for Industry, InVestigators, and ReViewers: Exploratory IND Studies, FDA, Rockville, MD. (22) Tsuge, K., Kataoka, M., and Seto, Y. (2000) Cyanide and thiocyanate levels in blood and saliva of healthy adult volunteers. J. Health Sci. 46, 343–350. (23) Food and Drug Administration (2006) Guidance for Industry (Draft): Drug Interaction StudiessStudy Design, Data Analysis, and Implications for Dosing and Labeling, FDA, Rockville, MD.
TX8004086
