NEW YORK (GenomeWeb) – The perfect drug molecule, according to Pfizer's Jean-Claude Marshall, is one that is right on target, has no off-target effects, and is highly bioavailable. Importantly, it is metabolized by a range of enzymes but not dependent on those really buggy ones, like CYP2D6, that can be variable.
"No drug is going to be that. We know that right up front," Marshall, director of Pfizer's Clinical Pharmacogenomics Lab, told GenomeWeb last week. "It's just our goal to attain that."
Pharma's quest for the perfect drug is supported by an ever improving understanding of pharmacogenetics. Pfizer began looking at the impact of cytochrome P450 genotypes on drug response in the 1990s, but it wasn't until 2009 that the drugmaker formed a dedicated PGx unit. Biomarker tests developed by this division are a critical part of the development process that enables the drug giant to decide whether a therapy needs a companion test, whether its label should be updated, or whether to move forward with a molecule at all.
"The testing menu that we offer has continued to evolve over the past several decades and certainly is much more in depth now than it ever was," Marshall said. "Before, we probably [tested for] a small handful of genotypes. Now, we're doing whole array-based approaches where we look at multiple genotypes for multiple CYP450 and drug transporter genes."
Marshall recently discussed his group's work at the Festival of Genomics in Boston, and noted that the majority of Pfizer's drugs coming on the market are metabolized primarily through the CYP3A5 pathway. Some published retrospective analyses using small cohorts have suggested that carriers of non functional alleles, such as CYP3A5 *3, *6, or *7, may have varying response to certain drugs.
For molecules metabolized by this pathway, researchers try to assess early on the precise contribution of the enzyme in breaking down and converting the drug into its active form. "Our general recommendation to clinical teams is if you have greater than 30 percent signal for CYP3A5, then we would recommend full genotyping be conducted at time of enrollment for at least Phase I/IIa and b studies," Marshall said at the Festival of Genomics.
Keeping the ideal drug molecule in mind, Pfizer teams try to "push away" from drugs metabolized primarily by polymorphic enzymes, such as CYP2D6. The enzyme is involved in metabolizing or activating 25 percent of prescribed drugs, but because it sits next to a pseudo-gene, it is challenging to assay.
If a drug is metabolized mostly by CYP2D6, "it's not a deal breaker" and Pfizer may still decide to pursue development, Marshall said. "But it could raise red flags." Given the choice between two molecules that are similar in all other characteristics, Marshall noted that the drugmaker would likely go with the treatment that was less dependent on highly variable metabolism enzymes.
Pharmacogenomics is particularly indispensible now that drug trials are global, and it is well known that specific drug metabolism enzymes are more prevalent in certain ethnic groups. For example, between 38 percent and 79 percent of Pacific Islanders, and between 40 percent and 50 percent of East Asians harbor CYP2C19 alterations that make them poor metabolizers of the anti-platelet drug Plavix (clopidogrel). "More and more regulators around the world are requiring these types of analyses up front before approval," Marshall said. "Ten years ago that wasn't the case."
Regulatory authorities around the world already ask drugmakers to conduct studies to demonstrate that a treatment works in the native population. Traditionally, pharmacogenetic assessments tend to be first performed in Caucasians, but Pfizer is working on enhancing genomic understanding of drug response in different populations.
For example, the firm recently launched a precision medicine center of excellence in Chile and provided a grant to help cancer patients receive genomic testing free of charge in Canada. Marshall noted that Pfizer is also starting to address these knowledge gaps by conducting separate Phase I and Phase II drug trials in different ethnic populations.
Pfizer's growing focus on genomic analysis is in line with mounting interest in the drug industry as a whole. A report this year from the Tufts Center for the Study of Drug Development (CSDD) found that biomarker data informs the development of 73 percent of cancer compounds and 42 percent of all compounds in pharmaceutical pipelines. Five years ago, Tufts had reported that between 12 percent and 50 percent of drugmakers' pipelines involved personalized therapies.
Pfizer's investment in the field has already yielded successes. Its non-small cell lung cancer drug Xalkori (crizotinib) is an example of the positive impact advancing genetics knowledge can have on drug development. It takes an average of 15 years for a compound to make it to market from the time of discovery. In comparison, Xalkori came to market in less than half that time.
Pfizer discovered molecule PF-02341066 in 2005; Japanese researchers published on the role of the EML4-ALK fusion gene in NSCLC in 2007; and the FDA approved the drug alongside a companion ALK FISH test in 2011. Since then, Pfizer has continued to explore different test platforms for predicting who will respond to treatment, identified new subpopulations of patients that might benefit from Xalkori, and worked with diagnostic shops to bring additional FDA-approved companion tests to market.
With an eye toward developing more precision drugs, Pfizer is collaborating with external groups to bolster its understanding of the genomic underpinnings of diseases such as lupus, and to develop advanced diagnostic tools for identifying best responders to cancer treatments. Based on published literature, the latest knowhow, and internal data, Marshall's lab is responsible for guiding drug development teams on when a treatment needs to be studied in specific biomarker-defined patient subsets.
The small investment that it does take to support my group … you can think of it as an insurance policy.
Researchers responsible for studying the pharmacokinetics and metabolism of molecules will likely know through in vitro efforts the major and minor metabolizing enzymes for a compound before it reaches Phase I human trials. Based on this information, Marshall's lab gets involved in planning first-in-human studies well ahead of Phase II trials.
Following guidance from regulators, Pfizer will generally study a drug with genotyping if it is known at the time of development that 50 percent or more of the compound is being metabolized by a CYP450 enzyme. Pfizer will also retrospectively study a population of patients if data in the literature implicates new genotypes in patients' response to a marketed drug.
In 2014, Pfizer's PGx group validated around 30 new lab-developed tests (LDTs) for use as part of drug studies. A number of these LDTs gauged genes, like those in the CYP450 family, involved in drug absorption, distribution, metabolism, and excretion, but Marshall's group also develops biomarker tests for Pfizer's entire drug development pipeline. "Anything ranging from oncology to neuroscience can call on us to develop a new LDT for early-phase clinical trials, with the idea that if necessary, those LDTs could be transitioned to something like a companion diagnostic by the time we get to Phase III trials," Marshall said.
Although validating LDTs and conducting genomic analysis is a substantial amount of work, what the PGx group does usurps such a tiny fraction of the overall development budget that the investment is well worth it, in Marshall's view. "We don't even compare to what it costs to run a multi-center trial," he said.
PGx as insurance policy
Especially as personalized or precision medicine becomes part of the popular lexicon, drugmakers can no longer afford to not act on emerging genomics knowledge about their drugs. For example, in a deceptive marketing lawsuit last year, the state of Hawaii accused Bristol-Myers Squibb and Sanofi-Aventis of knowing since 1998 that their antiplatelet drug Plavix didn't work as well for approximately 30 percent of the state's population due to genetic traits or the influence of other drugs, but failed to sufficiently alert consumers.
"The small investment that it does take to support my group … you can think of it as an insurance policy," Marshall said. "In those cases where we do find something up front, you're saving yourself something like a clopidogrel series of lawsuits [by] knowing that a drug won't do well in a certain type of population.
"Just from that point of view, that is a significant cost savings," he continued. "There is no doubt that we have to do these kinds of analyses."
For example, Pfizer retrospectively analyzed samples to try to further investigate genotypes that might impact the pharmacokinetic variability for its kidney cancer drug Inlyta (axitinib). The label for Inlyta, which is primarily metabolized by CYP3A4/5, states that doctors may need to adjust the dose of the drug in line with a patient's ability to tolerate it, but there is no genotype-specific dosing recommendation.
Although Pfizer researchers looked for such genotypes, they could not find a statistically significant association between any studied markers and the PK variability of the drug. "If we had, then we would look at what additional studies would be needed to build the level of evidence to go back to the FDA for modification of the label," Marshall said. "We didn't get there, of course. But that's what the path forward might have looked like." Pfizer will present the details of this work at the American Society of Human Genetics annual meeting this fall.
In another example, Pfizer revisited the PGx analysis for its HIV drug Selzentry (maraviroc) when an external research group published new data suggesting that patients with certain CYP3A5 genotypes may not benefit from typical doses. CYP3A5 is involved in as many as 60 percent of prescribed drugs and its activity varies highly by ethnicity. Even though Pfizer's internal data didn't support the external group's findings, researchers at the firm went back to the 800-subject Phase III trial for Selzentry and retrospectively analyzed data for the same genotypes.
"We saw what we thought we would see," Marshall said. "We didn't find any supporting evidence for what the external publication had shown." Still, to make doubly sure, Pfizer is now conducting a prospective study to assess whether these genotypes impact patients' ability to respond to Selzentry.
This type of analysis is becoming more and more common for drugmakers, according to Marshall. For this reason, whenever possible Pfizer tries to take a research sample from every study subject so investigators, based on country-specific rules, can conduct retrospective analysis if new data emerge.
But even with sufficient samples, some types of analysis are easier than others. "We are blessed to deal with genotype data where the answer is always 'yes' or 'no,'" Marshall said at the Festival of Genomics. "When we start talking about expression data that all goes through the window."
Developing LDTs to assess gene expression can be more challenging in terms of standardizing sampling and testing methods across multiple study sites. "The site-to-site variation is something that keeps us awake at night," he said.
Still, Pfizer is certainly looking beyond genotyping technologies, and even considering information from targeted sequencing approaches to inform drug development. The PGx group hasn't yet begun incorporating whole-exome data due to cost and uncertainty about how to call all the variants that might be identified by the approach. But "this is certainly on the horizon," Marshall noted.