Abstract
Drug-drug interactions (DDIs) between therapeutic proteins (TPs) and small-molecule drugs have recently drawn the attention of regulatory agencies, the pharmaceutical industry, and academia. TP-DDIs are mainly caused by proinflammatory cytokine or cytokine modulator–mediated effects on the expression of cytochrome P450 enzymes. To build consensus among industry and regulatory agencies on expectations and challenges in this area, a working group was initiated to review the preclinical state of the art. This white paper represents the observations and recommendations of the working group on the value of in vitro human hepatocyte studies for the prediction of clinical TP-DDI. The white paper was developed following a “Workshop on Recent Advances in the Investigation of Therapeutic Protein Drug-Drug Interactions: Preclinical and Clinical Approaches” held at the Food and Drug Administration White Oak Conference Center on June 4 and 5, 2012. Results of a workshop poll, cross-laboratory data comparisons, and the overall recommendations of the in vitro working group are presented herein. The working group observed that evaluation of TP-DDI for anticytokine monoclonal antibodies is currently best accomplished with a clinical study in patients with inflammatory disease. Treatment-induced changes in appropriate biomarkers in phase 2 and 3 studies may indicate the potential for a clinically measurable treatment effect on cytochrome P450 enzymes. Cytokine-mediated DDIs observed with anti-inflammatory TPs cannot currently be predicted using in vitro data. Future success in predicting clinical TP-DDIs will require an understanding of disease biology, physiologically relevant in vitro systems, and more examples of well conducted clinical TP-DDI trials.
Introduction
The increased clinical use of therapeutic proteins (TPs) has raised awareness of drug-drug interactions (DDIs) between TPs and small-molecule drugs (SMDs). Several recent workshops have addressed TP-DDIs, including the 2010 American Association of Pharmaceutical Scientists workshop on “Strategies to Address Therapeutic Protein–Drug Interactions during Clinical Development” (Girish et al., 2011), and related roundtables at the National Biotechnology Conference in 2012 (Kenny et al., 2013). Recent publications from academia, industry, and regulatory agencies have summarized the mechanisms involving TP-DDIs, evaluations of TP-DDIs included in Biologics License Application (BLA) submissions, and current strategies for the evaluation of TP-DDI potential during drug development (Morgan, 2009; Huang et al., 2010; Lee et al., 2010; Girish et al., 2011; Kraynov et al., 2011; Zhou and Mascelli, 2011; Lloyd et al., 2012; Dallas et al., 2013b; Slatter et al., 2013). As a result of discussions at several of these meetings, a collaboration among academia, industry, and regulatory agencies was initiated under the leadership of the TP-DDI Steering Committee (see Appendix) to address knowledge gaps in the mechanisms of TP-DDI, the relevance of in vitro systems, current industry practices for assessing these interactions, and to develop a general framework of risk-based approaches for TP-DDI assessment during drug development. The efforts were focused on pharmacokinetic (PK) and metabolism-based DDIs of monoclonal antibodies (mAbs), fusion proteins, cytokines, and cytokine modulators. As such, TPs as perpetrators of effects on drugs metabolized by cytochrome P450 enzymes and as victims of other SMDs or TPs were in scope. The Steering Committee subsequently formed two working subgroups as follows: the In Vitro TP-DDI Working Group (see Appendix) focused on hepatocyte-based methodologies to assess TP-DDI and in vitro–in vivo correlation, and the Population Pharmacokinetics (Pop PK) Working Group (see Appendix) focused on development of appropriate model-based TP-DDI evaluations, such as Population PK–based analyses. Both groups met by teleconference for more than 1 year, and organized a 2-day workshop at the Food and Drug Administration (FDA) White Oak Conference Center on June 4 and 5, 2012. The workshop was cosponsored by the FDA Office of Clinical Pharmacology and the Drug Metabolism and Clinical Pharmacology Leadership Groups of the International Consortium for Innovation and Quality in Pharmaceutical Development (The IQ Consortium; see Appendix). The workshop was attended by more than 70 delegates, including scientists from two regulatory agencies, invited academic speakers, and pharmaceutical industry scientists.
Following the workshop, both working groups agreed to publish independent white papers. The white paper by the Pop PK group will appear elsewhere (Chow et al., manuscript in preparation). In this white paper, we present cross-laboratory comparisons of cytokine effects on cytochrome P450 enzymes in hepatocyte culture, the results of a current practices poll, and the observations and recommendations of the In Vitro Working Group regarding the use of in vitro models to study cytochrome P450–related TP-DDIs.
Overview of Cytokine-Mediated Clinical Drug Interactions Involving TPs
TPs are often coadministered with SMDs. Initially, pharmacokinetic interactions between TPs and SMDs were viewed as unlikely, as mechanisms involved in the disposition of both modalities are fundamentally different. In the case of SMDs, various drug-metabolizing enzymes and transporters are involved in absorption, distribution, metabolism, and excretion, whereas clearance of TPs is typically mediated by nonspecific catabolism or target-mediated elimination (Lobo et al., 2004; Deng et al., 2012). However, as more TPs progressed to the clinic, isolated examples of clinically measurable effects of cytokines or cytokine modulators on SMD PK have been described (Table 1). Clinical administration of proinflammatory cytokines such as interferon-α (IFN-α) and interleukin-6 (IL-6) can reduce the activity of various cytochrome P450 enzymes, but in general these effects have been relatively small and variable (Lee et al., 2010). This class of TPs alters the expression of a number of drug-metabolizing enzymes and transporters in vitro (Aitken et al., 2006; Fardel and Le Vee, 2009; Huang et al., 2010; Lee et al., 2010). Several clinical studies have shown that SMD clearance can be impaired in individuals with inflammatory disease or cytokine-induced inflammation (Mahmood and Green, 2007; Schmitt et al., 2011, 2012). Accordingly, there are now clinical studies showing that some proinflammatory cytokines (particularly IL-6) and cytokine modulators (particularly the anti–IL-6R mAb tocilizumab) can affect the pharmacokinetics of coadministered SMDs (Schmitt et al., 2011, 2012).
For the anti-IL-6R mAb tocilizumab (8 mg/kg) in rheumatoid arthritis patients, a weak reduction (28%) was observed in the plasma exposure of omeprazole (a CYP2C19 substrate), and no effect was observed for dextromethorphan (a CYP2D6 substrate) or methotrexate (a substrate for Organic Anion Transporters (OATs) and Organic Anion Transporting Polypeptides (OATPs)) (Table 1; Hoffmann-La Roche, 2008; http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4371b1-01-FDA.pdf; Zhang et al., 2009; Schmitt et al., 2012). Treatment of rheumatoid arthritis patients with a single dose of 10 mg/kg tocilizumab resulted in a 43% reduction in plasma exposure to the CYP3A4 substrate simvastatin (Schmitt et al., 2011). This reduction indicates reversal of IL-6–mediated CYP3A4 suppression in rheumatoid arthritis patients with tocilizumab exposure (Schmitt et al., 2011). In addition to the IL-6 mAb example, other cytokine modulators have been shown to affect drug-metabolizing enzymes clinically. For the anti-IL2R mAb basiliximab and anti-CD3 mAb muronomab, increased plasma levels of cyclosporine A have been observed, but these subsided over time (Vasquez and Pollak, 1997; Strehlau et a., 2000; Sifontis et al., 2002). These DDIs may occur as a result of changes in circulating proinflammatory cytokines that are induced by anticytokine mAb-mediated inflammatory disease mitigation, or can be caused by proinflammatory therapeutic cytokines used in infectious disease or cancer therapy (Morgan, 2009).
Overall, TP effects on SMD exposure observed so far have been modest (<2- to 3-fold), and therefore may only be relevant for drugs such as cyclosporine or warfarin that have a narrow therapeutic index and routinely undergo therapeutic monitoring. Furthermore, these effects are disease state– and/or drug target–dependent, and are therefore different from typical pharmacokinetic DDIs between SMDs. As such, for inflammatory disease treatment, this is a drug-disease interaction, as opposed to a classic DDI.
Preclinical Approaches to Cytokine-Mediated TP-DDI Evaluation
Cellular Mechanisms of Cytochrome P450 Suppression by Cytokines and Inflammatory Mediators.
No single molecular mechanism can account for inflammation-mediated effects on various drug-metabolizing enzymes, and transcriptional effects may be coordinated by multiple different nuclear receptors and transcription factors (Zhou et al., 2009). Nuclear factor-κB–mediated cross-talk with the pregnane X receptor (PXR; NR1I2) has been implicated (reviewed by Zordoky and El-Kadi, 2009), and Gu et al. (2006) demonstrated a direct interaction of nuclear factor-κB (p65) with the DNA-binding domain of retinoid X receptor (RXR) α. Ghose et al. (2004) proposed that inflammation-induced effects on cytochrome P450 could be related to the subcellular localization of RXRα, which is a partner in the heterodimeric nuclear receptors PXR, constitutive androstane receptor, liver X receptor, farnesoid X receptor, RXR and peroxisome proliferator–activated receptor α. Mechanisms such as these may account for some of the general suppression effects that are observed across select enzymes involved in steroid, bile acid, and lipid metabolism (Slatter et al., 2013).
In addition to effects on RXR, studies have implicated post-transcriptional mechanisms such as nitric oxide–dependent ubiquitination and subsequent proteasomal degradation in the cytokine-mediated downregulation of cytochrome P450 enzymes (Ferrari et al., 2001; Lee et al., 2009). In rats, Lee et al. (2009) have shown that Cyp3a1 downregulation by IL-1β occurs via two distinct modes: a nitric oxide- and proteasome-dependent mechanism at early time points (<9 hours) and a nitric oxide- and proteasome-independent mechanism at later time points (24 hours), with no decrease in Cyp3a1 mRNA at either time point. This study highlights the complexity of cytochrome P450 regulation by cytokines, and suggests that inflammation-mediated suppression of metabolic enzymes can be a dynamic process and may involve several different measurable effects. For more in-depth reviews of cellular mechanisms of cytochrome P450 suppression, the reader is referred to a number of review articles (Morgan, 2001; Morgan et al., 2002; Aitken et al., 2006; Jover et al., 2009; Zordoky and El-Kadi, 2009).
In Vitro Models to Study Cytochrome P450 Suppression by Cytokines.
Unlike SMD metabolism and DDI, for which multiple in vitro methods are used routinely, there are only a few in vitro models that have been used for predictive assessment of TP-DDIs (Abdel-Razzak et al., 1993; Aitken and Morgan, 2007; Dickmann et al., 2011; Dallas et al., 2013b). Primary human hepatocytes are the current in vitro model of choice for studies on the effects of individual proinflammatory cytokines on hepatic cytochrome P450 expression and activity. Reports on the effect of cytokines on cytochrome P450 enzymes in primary hepatocyte culture date back to the early 1990s. Rat and human primary hepatocytes in sandwich culture were used to assess acute-phase protein responses to cytokine stimulation (Bader et al., 1992). Abdel-Razzak et al. (1993) and Donato et al. (1993) showed that proinflammatory cytokines such as IL-6, tumor necrosis factor-α (TNF-α), and IFN-γ suppressed the constitutive expression of cytochrome P450 enzymes in human hepatocyte culture. The magnitude and direction of effects on cytochrome P450 activity and expression depended on the particular cytokine and cytochrome P450 isoenzyme measured (Nguyen et al., 2013). To maximize in vitro response, cytokine concentrations were based on or exceeded the maximum levels observed during the acute phase of inflammation (Muntane-Relat et al., 1995; Chen et al., 2011). Physiologically relevant concentration thresholds for the in vitro effects of cytokines on hepatocytes are difficult to discern. Recent work in industry laboratories has attempted to benchmark the cytokine-mediated suppression of cytochrome P450 enzymes to serum cytokine concentrations by measuring dose-response IC50 values at cytokine concentrations relevant to chronic inflammatory diseases such as rheumatoid arthritis (Dickmann et al., 2011, 2012a).
Measurement of IC50 values for proinflammatory cytokine-mediated decreases in cytochrome P450 activity requires that adequate basal cytochrome P450 metabolism is available to quantify probe substrate metabolism both before and after suppression. Early hepatocyte studies suffered from a lack of highly sensitive and specific analytical assays, and in the early 1990s, were probably confounded by low basal cytochrome P450 expression, most likely explained by nonoptimal culture conditions. To overcome these limitations, hepatocytes were treated with cytochrome P450 inducers, such as β-naphthoflavone (an Aryl Hydrocarbon Receptor (AhR) activator and CYP1A2 inducer) and rifampicin (a PXR activator and CYP3A4 inducer), in the presence and absence of different cytokines (Muntane-Relat et al., 1995). In these studies, TNF-α and IL-6 suppressed the induction of CYP1A2 and CYP3A4, respectively. These early studies showed that decreases in cytochrome P450 expression correlated with cytochrome P450 enzyme activities due to effects of the cytokines on transcriptional and post-transcriptional events, rather than on the rate of translation of cytochrome P450 mRNA or the rate of cytochrome P450 protein degradation (Muntane-Relat et al., 1995). Dickmann et al. (2011) recently measured the IC50 for the IL-6–mediated suppression of CYP3A4 in human hepatocytes in the presence and absence of the inducer rifampicin and showed that the IC50 value increased in induced hepatocytes. The degree to which induction ablates proinflammatory cytokine–mediated cytochrome P450 suppression in vivo may depend on multiple factors, such as the isoform involved, the inducer and dose, the cytokine involved, and the physiologic concentrations of the cytokines evaluated.
Recent studies have used cryopreserved human hepatocytes (Dickmann et al., 2011; Dallas et al., 2012), which offer the advantage of experiment scheduling, preselection and characterization of cells, and improved reproducibility in studies using the same donor. Dallas et al. (2012) conducted a comprehensive analysis of the effects of IL-2, IL-6, and TNF-α on the gene expression and activities of multiple cytochrome P450 enzymes. This study reflects the findings of earlier studies using fresh human hepatocytes, and supports the use of cryopreserved hepatocytes as a model for cytokine-mediated suppression of cytochrome P450 enzymes.
Although primary hepatocyte culture offers a useful model system to investigate the effect of a single cytokine on cytochrome P450 expression, the model does not adequately reflect the complex interactions of multiple cytokines and cell types that are in play in vivo during inflammatory disease. For this reason, in vitro studies with a single cytokine and hepatocytes only may not accurately predict the magnitude of an effect in vivo.
Studies using the human hepatocyte-Kupffer cell coculture model are currently rare due to the limited (commercial) availability of human Kupffer cells. In human hepatocytes cocultured with hepatic Kupffer cells, the effects of three individual cytokines on CYP3A4 activities were measured by Sunman et al. (2004). The presence of Kupffer cells did not significantly perturb the suppression of CYP3A4 by IL-1 and IL-6. In contrast, IL-2 caused only a transient decrease in CYP3A when incubated with human hepatocytes alone but, in coculture with Kupffer cells, caused a concentration-dependent and sustained suppression in CYP3A activity after 72 hours.
In studies using lipopolysaccharide (LPS), the importance of physical contact between Kupffer cells and hepatocytes in vitro was demonstrated for various inflammatory cytokines (Hoebe et al., 2001). In these studies, LPS (1 and 10 µg/ml) was incubated with a single cell type or in coculture with or without direct physical contact, as provided by a semipermeable membrane insert. Cultures with physical contact between cell types had 10-fold (TNF-α) to 500-fold higher release of cytokines (IL-6). Phase 1 and 2 enzyme activities in the hepatocytes (e.g., CYP3A and UDP glucuronosyltransferase) were decreased to a greater extent relative to hepatocyte-only or noncontact cocultures. These in vitro data suggest that Kupffer cells have a significant role in priming hepatocytes to the suppressive effects of certain cytokines such as IL-6 (Freudenberg et al., 1982). Hepatocyte coculture with Kupffer cells, as currently used, may not be able to predict TP-DDI quantitatively due to the complexity of the immune system. However, coculture or other systems that recapitulate aspects of in vivo liver physiology may be worthy of further exploration for mechanistic studies on proinflammatory cytokine-mediated cytochrome P450 suppression (Kenny et al., 2013).
Studies on cytochrome P450 suppression by multiple cytokines in combination are rare, and historically, LPS was most commonly used to elicit a generalized acute-phase response. A recent study extended single-cytokine studies on IL-6 alone to examine the effect of combinations of IL-1β, IL-6, and TNF-α on cytochrome P450 in human hepatocytes (Dickmann et al., 2012a). When hepatocytes were treated with IL-1β and IL-6 in combination at concentrations ranging from 1 to 100 pg/ml, IL-6 was the main determinant of increases in acute-phase response marker mRNA and of decreases in CYP3A4 mRNA (Dickmann et al., 2012a). There was no synergy between IL-1β and IL-6 in the regulation of cytochrome P450 mRNA, although the effects of the two cytokines in combination were additive in certain instances. The preliminary study examined only a single donor, and therefore awaits more detailed experiments that may shed light on intersubject variability.
In Vivo Models to Study Suppression of Cytochrome P450 Enzymes by Cytokines.
Yang and Lee (2008) have reviewed studies on LPS-endotoxin effects on cytochrome P450 expression in rodent models. In general, LPS suppresses cytochrome P450 probe substrate clearances by about 1- to 2-fold. Few studies have looked at disease treatment effects on cytochrome P450 suppression. Piquette-Miller and Jamali (1995) examined the effect of ketoprofen on propranolol pharmacokinetics in a rodent model of severe adjuvant arthritis (AA). Plasma concentrations of propranolol enantiomers in AA were increased in the disease model and were related to the degree of inflammation and α1-acid glycoprotein levels. Ketoprofen treatment decreased the disease-induced suppression of propranolol elimination. Ling and Jamali (2009) studied the effects of the anti–TNF-α monoclonal antibody infliximab on cytochrome P450 enzymes in the resolution of early AA in rats. Hepatic Cyp1a and Cyp3a protein expression were decreased in rats with AA, and infliximab treatment increased Cyp1a and Cyp3a protein expression. Ashino et al. (2007) used a murine anti–IL-6 monoclonal antibody to treat arthritis in a Human T-Cell Lymphotropic Virus type 1 (HTLV-1) transgenic mouse arthritis model. IL-6 was significantly increased in the arthritic mice, and Cyp3a protein was 60% of nonarthritic transgenic controls. The anti–IL-6 antibody normalized the Cyp3a response. More recently, the murine collagen antibody induced arthritis model was used to study cytochrome P450 suppression (Dickmann et al., 2012b). Most hepatic cytochrome P450 mRNA levels were downregulated, with the exception of Cyp2d9 and 3a13, which increased 2.3- and 4.4-fold, respectively, relative to control mice. Cyp1a2, 2c29, 2b9, and 3a11 mRNA were decreased by 2- to 3-fold compared with controls, and the level of Cyp2e1 was unchanged. Intrinsic clearance data obtained with cytochrome P450 isoform–selective probe substrates and hepatic S9 indicated the largest suppression of cytochrome P450 activity was ∼2-fold. Strain- and isoform-dependent cytochrome P450 suppression has also been described in other in vivo models (reviewed in Dickmann et al., 2012b). Preclinical in vivo models related to LPS-mediated acute-phase responses or animal models of chronic disease are labor-intensive, time-sensitive, and strain/species-dependent, and therefore do not appear to be robust enough for prediction of the probability or magnitude of a clinical TP-DDI.
Hepatic Suppression of Drug Transporters.
Although the focus of TP-DDI studies has been around cytochrome P450 enzymes, in vitro and preclinical in vivo studies have revealed evidence of proinflammatory cytokine-mediated effects on drug transporters. However, clinically relevant drug interactions involving inflammation-mediated suppression of transporter activity have not been reported. P-glycoprotein (P-gp) expression in the liver, intestine, and blood brain barrier is decreased in rodent models of inflammation (LPS and turpentine) and increased in the kidney (Kalitsky-Szirtes et al., 2004; Hartmann et al., 2005; Wang et al., 2005; Hartz et al., 2006; Matsumoto et al., 2012). In a mouse model of LPS-induced inflammation, biliary clearance of the P-gp substrate doxorubicin was decreased by 50%, and a 3-fold increase in renal clearance was observed. These changes were associated with reduced P-gp protein levels in the liver and increased levels in the kidney (Hartmann et al., 2005). Using radiolabeled P-gp substrate, 99mTc-sestamibi, Wang et al. (2005) showed that LPS-induced systemic inflammation increased retention of 99mTc-sestamibi in the brain, heart, liver, and fetal tissue, and this correlated with a reduction in multidrug resistance protein Mdr1a mRNA levels in these organs. In vitro effects of proinflammatory cytokines on the expression of various transporters have also been reported (Petrovic et al., 2007). Three proinflammatory cytokines, IL-6, IL-1, and TNF-α, downregulate hepatic transporter expression in hepatocyte sandwich cultures (Diao et al., 2010). IFN-γ and oncostatin M (a member of the IL-6 family) also influenced transporter levels in cultured hepatocytes (Donato et al., 1993; Guillen et al., 1998; Le Vee et al., 2009; Chen et al., 2011). IL-1β, TNF-α, and IL-6 alter the expression profile of human hepatic transporters in vitro (Aitken et al., 2006; Fardel and Le Vee, 2009). The bile salt export pump as well as sinusoidal solute carrier uptake transporters are generally suppressed, whereas ATP-binding cassette drug efflux pumps are either unchanged or are up or downregulated depending on the transporter isoform. These changes are observed at mRNA levels for some transporters, and at protein and activity levels for others (Diao et al., 2010). Currently, the clinical relevance of these changes in transporter levels is unclear.
Modeling, Simulation, and Clinical Prediction
In vitro and preclinical in vivo models of cytokine effects on drug-metabolizing enzymes may afford interesting data regarding mechanism of action, but unfortunately cannot currently provide much guidance on the likelihood of clinically relevant TP-DDI (Morgan, 2009; Dickmann et al., 2011). To enable TP-DDI clinical prediction and guide clinical development teams on the need for a clinical study (the current gold standard), many factors need to be considered. These minimally include TP pharmacology and clearance pathways, a thorough understanding of relevant cytokines and their impact on cytochrome P450 enzymes in the disease setting, the patient population, and comedications. Physiology-based PK/pharmacodynamics (PD) modeling has been used in retrospective attempts to relate in vitro IL-6 effects on CYP3A4 to the clinical tocilizumab:simvastatin interaction and to SMD clearance in organ transplant and surgical trauma (Machavaram et al., 2013). Unfortunately, IL-6 is currently the only cytokine-mediated TP-DDI for which both in vitro and in vivo data are available. Advances in in vitro–in vivo extrapolation have been limited by the subtle magnitude of the interaction, the challenges in generating physiologically relevant quantitative in vitro data, and the paucity of positive clinical TP-DDI examples (Kenny et al., 2013). Disease-centric approaches, rather than the current cytokine- and target-centric approaches, have recently been proposed that may enable clinical risk assessment based primarily on the disease indication and treatment effectiveness (reviewed by Kenny et al., 2013 and Slatter et al., 2013).
The BioSafe Survey
The BioSafe Pharmacokinetics and Disposition Expert Working Group conducted a survey in January to March 2010 on TP-DDIs in drug development (Lloyd et al., 2012). The survey assessed practices of 21 BioSafe industry trade group member companies. The survey indicated that 1) most of the DDI studies for TPs have been undertaken in the oncology and immunology therapeutic areas, 2) most of the DDI studies in the oncology therapeutic area considered the TP as a potential victim, and 3) most of the DDI studies in the immunology therapeutic area considered the TP as a perpetrator. The survey was a snapshot-in-time that showed industry strategy in the early days of this issue. The survey prompted subsequent industry efforts to understand underlying mechanisms and clinical implications of TP-DDIs.
Change in the Regulatory Landscape
The number of TPs under development has steadily increased in recent years. For example, 46 BLAs were approved between 2000 and 2009, as compared with 25 BLAs in 1990–1999 and less than 10 BLAs in 1980–1989 (http://www.fda.gov/downloads/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/UCM333444.pdf). Historically, drug interaction studies involving TPs were not routine, mainly because there were few examples of TP-DDI–related and clinically relevant adverse events. The general lack of understanding of potential interaction mechanisms and of predictive tools became evident when clinically measurable TP-DDIs started to appear. Reviews published in 2007 highlighted examples of drug interactions involving TPs and illustrated possible interaction mechanisms (Mahmood and Green, 2007; Seitz and Zhou, 2007). TP-DDI-related information (either PK- or PD-based) started to appear in product labeling, as summarized in a labeling overview published by the FDA in 2010 and more recently in 2013 (Huang et al., 2010; Lee et al., 2010; Zhao et al., 2013). In addition to clinical drug interaction studies, in vitro evaluation of TP-DDI started to influence drug labels and regulatory requests. For example, product labeling of tocilizumab and ustekinumab included results from in vitro studies of cytokine-mediated effects on cytochrome P450 enzymes (Tocilizumab Product Label, 2010; http://www.accessdata.fda.gov/drugsatfda_docs/label/2010/125276lbl.pdf; Ustekinumab Product Label, 2012; http://www.accessdata.fda.gov/drugsatfda_docs/label/2012/125261s0059lbl.pdf).
Regulatory Guidelines.
The 2007 European Medicines Agency guideline entitled “Guideline on the Clinical Investigation of the Pharmacokinetics of Therapeutic Proteins” described DDI concerns about immunomodulators such as cytokines that have potential for inhibition or induction of cytochrome P450 enzymes (European Medicines Agency, 2007; http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500003029.pdf). The guideline suggested in vitro and/or in vivo studies should be considered on a case-by-case basis. The 2012 FDA draft guidance on drug interactions expanded recommendations on TP-drug interaction assessment relative to the previous 2006 draft (FDA Guidance for Industry, 2012; http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm064982.htm). The new draft guidance contained a decision tree which reflected the FDA’s current view of TP-DDI as follows:
If an investigational TP is a cytokine or cytokine modulator that has known effects on cytochrome P450 enzymes or transporters, studies should be conducted to determine the magnitude of the TP’s effects on drugs that are substrates of the affected cytochrome P450 enzymes or transporters.
In vitro or animal studies have limited value in the qualitative and quantitative projection of clinical interactions because translation of in vitro to in vivo and animal to human results to date has been inconsistent, necessitating clinical drug interaction studies. The in vivo evaluations of TPs in targeted patient populations can be conducted with individual substrates for specific cytochrome P450 enzymes and transporters, or studies can be conducted using a “cocktail approach.”
If a TP will be used with other drug products (small molecule or TP) as a combination therapy, studies should evaluate the effect of each product on the other (PK or PD when appropriate). This evaluation is particularly important when the drug used in combination has a narrow therapeutic index (e.g., chemotherapeutic agents).
If there are known mechanisms for interactions or prior experience suggests potential PK or PD interactions, appropriate evaluations for possible interactions should be conducted. Some interactions between drugs and TPs are based on mechanisms other than cytochrome P450 enzyme or transporter modulation.
In Vitro TP-DDI Working Group Cross-Company Evaluation of Hepatocyte Studies
The Introduction of this review summarized the body of literature that describes the effects of cytokines on cytochrome P450 expression in isolated human hepatocytes. Overall, high interlaboratory variability in cytochrome P450 suppression by cytokines has been noted (Abdel-Razzak et al., 1993; Donato et al., 1997; Pascussi et al., 2000; Aitken and Morgan, 2007; Nguyen et al., 2013). Many factors could contribute to variability, including experimental design (e.g., plate format, seeding densities, and time points), intrinsic donor attributes (age, disease state, and medical history), and sources of reagents (cytokines and culture media) (Kenny et al., 2013; Nguyen et al., 2013). Interlaboratory variability is a concern from both a pharmaceutical industry and regulatory perspective, as it complicates the assessment of the validity of data and interlaboratory data comparison.
Figure 1 illustrates variability in IL-6 suppression of CYP3A4 mRNA and activity in laboratories of several In Vitro TP-DDI Working Group members. Figure 1A shows how incubation time can influence potency assessments. Cryopreserved human hepatocytes from four donors studied by company A revealed that IC50 values for CYP3A4 mRNA suppression by IL-6 were generally higher following longer incubation times (72 versus 48 hours). This may be due to IL-6–mediated suppression of IL-6 receptor expression over time (Dickmann et al., 2011). Intrinsic donor characteristics may also have affected experimental results as shown by IL-6–mediated downregulation of CYP3A4 mRNA (Fig. 1A) and CYP3A4 activity (Fig. 1B) in different donors. These data indicate that experimental design and intrinsic donor attributes should be taken into account when comparing the effects of different cytokines on cytochrome P450 enzymes. To better understand interlaboratory experimental variability, six laboratories from the TP-DDI Working Group performed a cross-company in vitro study that assessed the effect of IL-6 on CYP3A4 in a single hepatocyte donor.
Experimental Design.
A single preselected human hepatocyte donor was used by all six laboratories (Table 2). The donor was preselected following initial studies to confirm that cytochrome P450 enzymes were induced by prototypical inducers (β-naphthoflavone, phenobarbital, and rifampicin) and suppressed by IL-6. During the cross-company evaluation, cryopreserved cells were incubated for 48 hours with recombinant human IL-6. CYP3A enzyme activity was measured as testosterone metabolism to 6-β-hydroxytestosterone, and CYP3A4 mRNA expression was evaluated with reverse-transcription polymerase chain reaction. C-reactive protein (CRP) mRNA expression was measured as an indicator of the responsiveness of the experimental system to IL-6 (Dickmann et al., 2011). Rifampin (10 μM) was included as a control to determine that the cells were responsive to a prototypical inducer. The same hepatocyte donor, IL-6 source, IL-6 concentrations, and incubation time were used, and remaining experimental conditions were flexible to accommodate individual laboratory protocols. These included culture media (hepatocyte recovery, plating, and maintenance media), culture time prior to cytokine treatment, maintenance media supplements (with or without dexamethasone, fetal bovine serum, and other common tissue culture supplements), absence or presence of a matrigel overlay, and plate layout (24- versus 96-well format). Table 2 summarizes experimental conditions across individual laboratories.
Cross-Company Hepatocyte Study Results.
Table 3 summarizes the cross-company assessment of IL-6 effects on CYP3A4 mRNA and activity. In general, suppression of CYP3A4 by IL-6 was consistently observed. Figure 2 depicts the concentration-effect profile of IL-6–mediated inhibition of CYP3A4 mRNA expression (Fig. 2A) and activity (Fig. 2B) from company D.
The influence of glucocorticoids such as dexamethasone on CYP3A4 expression is well known (Hewitt et al., 2007). In general, there is a decline in cytochrome P450 activity levels as hepatocytes de-differentiate in culture. As a consequence, dexamethasone is often used in hepatocyte culture protocols to decrease this effect on cytochrome P450. To evaluate the effect of dexamethasone on IL-6–mediated suppression of CYP3A4, several laboratories performed the cross-company study in both the presence and absence of dexamethasone. Table 3 shows that maximum suppression of CYP3A4 activity by IL-6 across laboratories ranged between 20 and 38% and between 68 and 91% in the absence or presence of 100 nM dexamethasone, respectively. For IL-6–mediated suppression of CYP3A4 mRNA expression, maximum suppression was between 55 and 95% and between 93 and 99.6% in the absence or presence of dexamethasone, respectively (Table 3). The level of CYP3A4 induction by rifampin was highly variable and did not correlate with either the expression of CRP or suppression of CYP3A4 mRNA by IL-6.
Hepatocytes from the single donor had low to moderate basal levels of 6β-testosterone hydroxylase activity, depending on the culture conditions used [Table 3; 2.7–13 pmol/min/106 cells (without dexamethasone) and 27–220 pmol/min/106 cells (with dexamethasone)], and this may have contributed to variability in study results. Despite variation in experimental conditions and low basal cytochrome P450 levels, three of six independent laboratories that reported IC50 values for IL-6–mediated suppression of CYP3A4 mRNA expression provided results within 2.2-fold, and four of six within 11.4-fold of each other (Table 3). However, a wide range of IC50 values for suppression of CYP3A4 activity (24–1750 pg/ml) and mRNA expression (8.3–1600 pg/ml) were observed. These single-donor data showed interlaboratory variability in the measurement of IC50 values for CYP3A4 suppression by IL-6 and do not support quantitative interlaboratory comparisons of cytochrome P450 inhibition.
IL-6 resulted in dose-dependent upregulation of CRP mRNA (Fig. 3), although the EC50 values were variable across the laboratories [Table 3; 0.114–346 ng/ml (without dexamethasone) and 0.95–47.5 ng/ml (with dexamethasone)]. The maximum fold increase in CRP expression over untreated controls ranged between 208- and 1000-fold, and 572- and 1342-fold in the absence or presence of dexamethasone, respectively (Table 3). These data indicated that the hepatocyte experimental system was responsive to IL-6.
Technical Considerations.
Fetal bovine serum (FBS) is known to aid hepatocyte attachment and provide higher-quality hepatocyte monolayers (Hewitt et al., 2007). In the cross-company comparison, the impact of FBS on cytochrome P450 suppression was minimal, since laboratories that incorporated or omitted FBS in recovery and plating media all reported good cell attachment at the beginning of the study period. Three of six companies using 24-well collagen-coated plates generally reported good cell monolayers throughout the entire study period. Three of six companies used a 96-well format, and well attached cells were generally reported for up to 48 hours. Culture time was also an important contributor to cell health, with better hepatocyte monolayers observed in the first 48 hours after cytokine dosing. Preselection and validation of donor performance in a given plate format is recommended prior to embarking on studies.
Three of six laboratories that used dexamethasone supplementation reported generally higher basal levels of cytochrome P450 activity. Without dexamethasone, incorporation of a matrigel overlay can sometimes mitigate the decline of cytochrome P450 enzymes (Hewitt et al., 2007). However, the cross-company in vitro experiment did not conclusively demonstrate a benefit of a matrigel overlay.
Three of six laboratories used the same commercially available recovery media and similar culture media (William’s E Media) supplemented with dexamethasone. Company F used a different commercial media for recovery, plating, and maintenance. The IC50 of IL-6–mediated suppression of CYP3A4 mRNA reported by company F was significantly higher than the other companies. Therefore, culture media may partially explain differences in the observed IC50 values. The cross-company studies highlight that preselection and validation of human donor lots should be performed prior to initiating TP-DDI studies in cryopreserved human hepatocytes. This ensures optimal cell attachment, growth, and response to positive control repressors such as IL-6 under experimental conditions that will ultimately be used by each laboratory.
In summary, the working group concludes, based on experience with primary hepatocyte culture studies on the well-studied CYP3A4 suppressor IL-6, that significant variability in IC50 values across laboratories can be expected (Fig. 1; Table 3). Experimental conditions that should be considered for mechanistic TP-DDI investigations are listed in Table 4. Future research should focus on achieving consensus on 1) optimal assay conditions, 2) appropriate positive controls for various cytochrome P450 enzymes, 3) the clinical relevance of effects on cytochrome P450 enzymes other than CYP3A4 and on non–cytochrome P450 absorption, distribution, metabolism, excretion enzymes and transporters, 4) the effect of mixtures of cytokines on gene expression, 5) temporal changes in cytokine and cytokine receptor production and their ability to regulate each other in vitro, and 6) mechanistic understanding of cytochrome P450 suppression via multiple (additive or synergistic) pathways.
The TP-DDI Workshop
The objective of the workshop (for meeting agenda, see Supplemental Fig. 1) was to facilitate better understanding of the current science, investigative approaches, and knowledge gaps in the study of TP-DDI. The audience was made up of industrial scientists (84%, from 20 companies), with invited academic speakers and regulatory scientists representing 16% of attendees. Following lectures and panel discussions that covered the most recent state-of-the-art data, the meeting culminated with presentation of the In Vitro Working Group recommendations as follows:
Evaluation of TP-DDI requires scientific diligence and is currently best accomplished with a clinical study. The need for a dedicated clinical TP-DDI study can be discerned with a four-step assessment (Kenny et al., 2013). First consider the disease indication and disease biology, then the TP class, and follow this with an analysis of risk related to concomitant medications, clearance mechanisms, and patient factors (Fig. 5).
A thorough understanding of relevant cytokines in the disease indication and their impact on cytochrome P450 enzymes, along with evidence that a novel TP will influence proinflammatory cytokines, will help determine whether a clinical study is necessary. When there is insufficient information for risk assessment, potential TP effects on important concomitant medications could be explored in phase 2 studies using Population PK approaches.
Treatment-induced changes in appropriate biomarkers in phase 2 and 3 studies, possibly CRP or CYP3A markers, may indicate the potential for a clinically measurable treatment effect on cytochrome P450 enzymes.
Preclinical quantitative assessment of TP-DDI to predict a clinical interaction is not recommended at this time.
In vitro studies on the mechanism of proinflammatory cytokine or cytokine modulator effects on cytochrome P450 may help further elucidate the limitations and potential utility of current and future technologies.
Poll Results: In Vitro Evaluation of TP-DDI Potential.
To develop recommendations that would reflect a cumulative opinion/recommendation of the workshop participants, a live polling session was conducted after presentations and panel discussions. The polling session covered in vitro studies, clinical observations, Pop PK approaches, and regulatory interactions in the field of TP-DDIs. The poll comprised 18 scientific questions and 2 questions on audience demographics. All responses were blind with respect to the individual and company/institution. The poll was performed prior to presentation of the recommendations of the In Vitro Working Group, to gauge audience opinion without influence from the working group recommendations. All poll data are available as supplementary material (Supplemental Fig. 2), and a summary of three key in vitro questions is provided in Fig. 4.
The entire audience (industry and academic/regulatory) was polled, with a 98% response rate. The majority of those polled were not clear about regulatory requirements on TP-DDI evaluation for cytokines and/or cytokine modulators. Forty-five percent of respondents indicated they did not have full understanding, compared with 16% who were clear on regulatory requirements. A larger proportion of those polled were clear on clinical expectations (25%) compared with preclinical expectations (14%). Industrial attendees were asked if regulatory agencies had requested in vitro and/or clinical data on TP-DDI, and two-thirds of the industry audience indicated they had received requests. Only one-third of the industry members polled indicated their company had an internal guidance or strategy for addressing TP-DDI.
Approximately half of the industry audience (49%) had performed in vitro work in human hepatocytes to investigate cytokine and/or cytokine modulators as perpetrators of TP-DDI (Fig. 4). Where those studies were performed for a specific drug discovery program, there was an equal distribution of opinion on whether these studies were useful (21%) or not useful (19%) for decision making and/or project progression. Twenty-three percent of the industry audience indicated their company had included in vitro data in regulatory submissions as justification that clinical studies were not needed. There was an equal distribution of opinion on whether in vitro data were accepted (13%) or were not accepted (13%) by regulatory agencies as justification that clinical studies were not needed.
When asked about the qualitative usefulness of in vitro hepatocyte studies for prediction of the clinical situation, nearly two-thirds of the audience thought in vitro data could be useful, with 12% answering “yes” and 53% answering “maybe,” compared with 33% who answered “no” (Fig. 4). In contrast, when asked about quantitative assessment, the majority of the audience (90%) thought in vitro hepatocyte data may not reliably predict the likelihood and magnitude of a clinical DDI. Of those polled on the need for alternative in vitro methods (e.g., hepatocytes in coculture with Kupffer cells or other immune-modulating cells), 56% indicated that alternative in vitro approaches should be investigated, 30% indicated “on a case by case basis,” 5% responded “no,” and 9% responded “don’t know.”
Application of the Working Group Recommendations to Address a TP-DDI.
Prior to initiation of preclinical evaluations, the disease impact on PK of a coadministered drug should be understood. The limited data currently available indicate that the absolute magnitude of chronic inflammatory disease effects on cytochrome P450 enzymes is modest, and that only low therapeutic index cytochrome P450 substrates present a risk of clinically significant TP-DDI. Additional factors include modality of the TP, clearance mechanisms of coadministered drugs, probability that a coadministered drug will alter clearance of the TP, and the ability of the TP to interact with endogenous proteins other than the target (Fig. 5).
The need for a TP-DDI study may require a disease indication–centered strategy in addition to “no effect” in vitro human hepatocyte data. “No effect” in vitro human hepatocyte data have recently been used to preclude the need for a clinical TP-DDI study (S. Dallas, personal communication). The anti–IL-12/IL-23 p40 antibody ustekinumab is approved for the treatment of moderate to severe plaque psoriasis. In this case, a sufficient understanding of the cytokines in question (i.e., the absence of IL-12 and IL-23 receptors in hepatocytes) was available. However, the implications of these factors with respect to TP-DDI risk had not been generally established, therefore in vitro studies were conducted and supported the hypothesis that IL-12 or IL-23 would not directly suppress cytochrome P450 enzymes (Dallas et al., 2013a). Similar studies on diverse cytokines (both pro- and anti-inflammatory) and disease indications will help to elucidate molecular pathways that are affected and determine which cytochrome P450 isoforms may be suppressed. The clinical relevance of all these in vitro results remains to be fully elucidated.
Conclusions and Future Directions
Cytokine-mediated DDIs observed with anti-inflammatory TPs cannot currently be predicted using in vitro data, and any preclinical systems used to predict such interactions need to be capable of more closely mimicking the disease state. In vitro experimental conditions used in the cross-company evaluation described in the In Vitro TP-DDI Working Group Cross-Company Evaluation of Hepatocyte Studies section and Table 4 identified a number of important parameters for TP-DDI evaluations in hepatocytes. Future success in predicting clinical TP-DDIs will require an understanding of disease biology, physiologically relevant in vitro systems, and additional clinically measurable effects to facilitate in vitro–in vivo extrapolations and mathematical modeling.
Acknowledgments
The authors thank the members of the TP-DDI Steering Committee for their encouragement and support. The authors also acknowledge the following colleagues for technical support in generating the in vitro cross-company hepatocyte data: Souvik Chattopadhyay and Carlo Sensenhauser (Janssen), Suzanne Tay and Susan Wong (Genentech), Sonal Patel (Amgen), and Xing Yang (Pfizer).
Appendix
Members of the BIO TP-DDI Steering Committee
Honghui Zhou (Janssen; Cochair), Joseph Balthazar (State University of New York–Buffalo), Raymond Evers (Merck), Shiew Mei Huang (FDA, Cochair), Lei Zhang (FDA), Amita Joshi (Genentech), Andrew Chow (Amgen), Lewis Klunk (Biogen Idec).
Members of the In Vitro TP-DDI Working Group
Raymond Evers (Merck; Chair), Shannon Dallas (Janssen), Odette A. Fahmi (Pfizer), Jane R. Kenny (Genentech), Eugenia Kraynov (Pfizer), Aarti H. Patel (GlaxoSmithKline), Theresa Nguyen (Merck), J. Greg Slatter (Amgen), Lei Zhang (FDA).
Members of the Pop PK TP-DDI Task Force
Andrew Chow (Amgen; Chair), Justin Earp (FDA), Manish Gupta (Bristol-Myers Squib), William Hanley (Merck), Chuanpu Hu (Janssen), Diane Wang (Pfizer), Min Zhu (Amgen).
The International Consortium for Innovation and Quality in Pharmaceutical Development
The IQ Consortium is an international association of pharmaceutical and biotechnology companies that aims to advance innovation and quality in the development of pharmaceutical products through scientifically driven best practices and standards. Their ultimate goal is to improve safety and efficacy of medical products for patient benefit (http://iqconsortium.com/).
Authorship Contributions
Wrote or contributed to the writing of the manuscript: Evers, Dallas, Dickmann, Fahmi, Kenny, Kraynov, Nguyen, Patel, Slatter, Zhang.
Footnotes
- Received April 1, 2013.
- Accepted June 21, 2013.
↵This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- AA
- adjuvant arthritis
- BLA
- Biologics License Application
- CRP
- C-reactive protein
- DDI
- drug-drug interaction
- FBS
- fetal bovine serum
- FDA
- Food and Drug Administration
- IFN
- interferon
- IL
- interleukin
- LPS
- lipopolysaccharide
- mAb
- monoclonal antibody
- PD
- pharmacodynamics
- P-gp
- P-glycoprotein
- PK
- pharmacokinetics
- Pop PK
- Population Pharmacokinetics Working Group
- PXR
- pregnane X receptor
- RXR
- retinoid X receptor
- SMD
- small-molecule drug
- TNF-α
- tumor necrosis factor-α
- TP
- therapeutic protein
- U.S. Government work not protected by U.S. copyright