Most drug–drug interaction (DDI) studies take place in the laboratory, where scientists introduce an experimental drug into cell culture, checking for its ability to induce or inhibit drug-metabolizing cytochrome P450 (CYP) enzymes. These in vitro studies investigate how a drug is metabolized, how it may inhibit the metabolism of other drugs, and how it may impact enzymes and other cellular processes. The goal of these in vitro studies is to predict whether a drug is likely to be a “perpetrator” or a “victim” of DDI effects.
Although in vitro studies that evaluate drug metabolism are well established, they are also showing their limitations. For example, they may struggle to accommodate increasingly diverse drug candidates, which now include biologic drugs.
To overcome their limitations, DDI studies need to keep up with technological improvements and regulatory requirements. In 1997, when the U.S. Food and Drug Administration (FDA) issued its first guidance document on DDIs. Basically, the FDA described the steps pharmaceutical companies and an emerging DDI industry could take to predict adverse DDIs. Since then, the European Medicines Agency (EMA) has issued its own guidance documents. Moreover, both institutions have repeatedly updated their recommendations.
Partly in response to regulatory activity, and partly in response to technological innovations, such as sophisticated computer modeling, the DDI industry has gradually changed over the past decade. For example, the DDI industry now includes a growing number of specialist laboratories, that is, laboratories that cultivate niche areas of expertise. In fact, some companies, like Nuventra Pharma Services, don’t run wet labs at all. Instead, they offer consulting services to help clients navigate the myriad possibilities that are emerging for DDI studies.
Among the companies that have laboratories, almost all offer the mainstay of DDI testing: in vitro drug studies that explore a new drug’s potential to inhibit or induce drug-metabolizing enzymes. For example, BioIVT, a global provider of laboratory services, has expertise in using culture hepatocytes to perform induction studies.
Meanwhile, the “field has evolved over the past decade” to “understand the importance of transporter science,” says Phil Butler, PhD, head of ADME, Cyprotex. In contrast to drug-metabolizing enzymes, which are typically found in specific organs, transporters are found in many tissues throughout the body and regulate the access of substances to different sites. Many laboratories, including Cyprotex and Charles River Laboratories, tout their ability to conduct studies that evaluate a new drug’s potential to induce or inhibit drug transporters.
Other companies are focusing on innovating in silico models. For example, drug development consultancy Certara is using computer-based modeling and simulation to predict DDIs and simulate different scenarios that inform future in vitro and in vivo DDI studies.
These and other specialist activities are contributing to an overarching trend in DDI studies. Essentially, many laboratories are moving away from the routine box-checking stipulated by regulatory guidance documents and toward a more thoughtful consultancy-style approach to helping pharma clients design new types of studies to predict DDIs.
The FDA’s catch-22
The FDA issued a guidance document in 2017 that introduced a shift in the agency’s approach to in vitro DDI studies, says Ronald Laethem, PhD, director, ADME services operations, BioIVT. A similar point is made by Cyprotex’s Butler: “One of the main things to come out of the FDA guidance is a strong suggestion that the [in vitro] studies should be done earlier in the value chain.”
Traditionally, the concentration of drug used during in vitro DDI studies has been informed by tests in healthy volunteers. The results of these tests guide researcher’s understanding of Cmax, or the peak concentration of a drug inside the human body.
“The  guidelines recommend that you use concentrations that are human-relevant, but they also say that before you go in vivo, you need to have in vitro data,” says Jelle Reinen, PhD, group leader, in vitro drug metabolism and pharmacokinetics, Charles River Laboratories. “It’s a bit of a catch-22,” Reinen remarks.
If the FDA wants in vitro studies before trials in humans, then the in vitro models might not benefit from an understanding of Cmax, explains Laethem, who added that, moving forward, in vitro studies will need to be even more carefully designed to ensure laboratory experiments are physiologically relevant.
Still, there are distinct benefits to this approach, Laethem continues. Consider Phase I studies, which test relatively unknown candidates. These studies typically exclude wide segments of the population, such as women who may be pregnant, people who already take at least one additional medication or people with renal deficiencies. By conducting in vitro DDI studies first, Laethem says, “you can include more people in those [Phase I studies] and get better results.”
In silico studies
In 2018, the FDA and the EMA released respective guidance documents on in silico modeling, also called physiologically based pharmacokinetic (PBPK) modeling. It may be performed with the aid of proprietary PBPK software systems, such as Certara’s Simcyp and Simulations Plus’ GastroPlus.
PBPK models can predict drug disposition throughout the body and help evaluate potential DDIs. Despite rapid advancements in PBPK models over the past decade, uncertainty remains in the field about the exact role PBPK models should play in drug development.
“From my perspective…there isn’t any uniform way that companies are doing this,” says Amin Rostami, PhD, senior vice president of research and development and chief scientific officer, Certara. Rostami, who is also a professor of systems pharmacology at the University of Manchester, adds that the latest FDA and EMA guidance documents fully support the use of PBPK modeling to bolster existing DDI studies.
The advantage of the regulatory guidance is also emphasized by Mark Bush, PhD, associate vice president, Nuventra. “One thing that was really nice about the guidance,” he points out, “is there was this standardized algorithm to help us consistently look at interaction risks.”
Today, models can be used to simulate situations that would be too cumbersome or expensive to capture using in vitro or in vivo DDI studies. Models can guide dosing and predict outcomes in pediatric patients, polypharmacy patients, renally impaired patients, and patients in other niche groups.
“These models are not actually replacing a clinical trial that was going to happen,” says Rostami. Rather, they are “replacing a study that was never going to happen.” Going forward, PBPK models may be trusted with core tasks. Already, the FDA’s DDI guidance says that the agency will consider PBPK models “in lieu of some prospective DDI studies.”
Regulators are more accepting of in silico models, Rostami notes, because companies like Certara have been transparent about how the models work. “It’s not like artificial intelligence, where nobody knows what’s happening,” he adds. Biologists and computer scientists have created computer simulations of DDI experiment components.
For more than 40 FDA-approved drugs, approval processes included PBPK modeling. According to Rostami, the most common application of PBPK modeling has been to predict DDIs.
Much ado about biologics
The drug development pipeline includes many biologics and therapeutic proteins. But when it comes to DDI studies, “anything that’s not a small molecule has the potential to be challenging,” warns Butler. Adds Laethem, “the models we have are not really adequate for addressing questions” raised by non-small molecule drugs.
The FDA’s 2012 guidance briefly mentions therapeutic proteins, but the 2017 guidance clearly addresses only small molecule DDI studies, leaving industry unsure of how to proceed. “It’s not clear what needs to be done,” admits Reinen. “Most often, the proteins are really well soluble, so it’s not an issue. We use the same approach we use for pharmaceuticals.”
Generally, experts agree that small molecules have the highest DDI risk, so it hasn’t been quite as urgent to map the DDI risk of biologics, observes Bush. “Most of these proteins and peptides don’t interact with cytochrome P450 in the same way,” he explains. Nonetheless, he recommends a degree of caution: “Just because you’re working with a protein or peptide doesn’t mean you’re 100% off the hook with drug interactions.”
There could be downstream synergistic effects that create adverse events. For example, says Bush, the glucagon-like peptide-1 class of diabetes drugs can cause delayed gastric emptying, which might, in turn, affect the absorption of other oral medications. “Our standard approach with small molecules,” Butler remarks, “doesn’t necessarily fit outside of that space.”
Although companies are trying to keep up by developing new systems, the DDI studies for non-small molecule drug candidates are, in Laethem’s estimation, more experimental. “While some companies are willing to try these things out, I think one of the biggest drivers is going to be a blessing from the FDA,” he emphasizes. That blessing may soon come, with the FDA announcing1 last year that it would be taking comments and suggestions on the best way to evaluate DDI studies for therapeutic proteins.
Today, more than 20% of adults take three or more prescription drugs, according to the Centers for Disease Control and Prevention.2 As the number of people taking more than one drug continues to increase, so too will the risk of drug interactions. Soon, many DDI scenarios could be mapped using computer simulations, but for now, laboratories across the globe are working to keep up with ever-shifting regulations for DDI laboratory studies.