By Doug Macron
BETHESDA, Md. — As improvements in chemical modifications and delivery technologies boost the potency of oligonucleotide-based therapeutics, developers of such drugs must remain vigilant about the potential for hybridization-dependent off-target effects, according to a group of researchers from the industrial, academic, and regulatory fields.
In regards to oligonucleotide-based drugs, "many of the toxicities that we have described over the years … are related to the chemistry of the oligonucleotide," Santaris Pharma Chief Development Officer Art Levin, a member of the Oligonucleotide Safety Working Group, said this week at the Drug Information Association's annual Oligonucleotide-Based Therapeutics Conference.
These hybridization-independent toxicities "may be related to sequence, but they are not necessarily related to the fact that the oligonucleotide binds to a target because of Watson-and-Crick base pairing," he explained.
And while these "have been the most prominent toxicities that we've seen in 15 years of oligonucleotide-development experience ... we're now improving the potency of our ligands, we're adding modifications to our oligonucleotide therapeutics that are improving their affinity, [and] we're creating improved delivery systems for siRNAs," Levin said. As a result, "we should continue to be vigilant with respect to the potential for [hybridization-dependent] off-target events."
These events can be defined as "hybridization-dependent alterations in the expression of unintended targets." They potentially include effects caused by inadvertently "tying up" microRNA machinery — a toxicity described by Stanford University researcher Mark Kay and colleagues in 2006 (see RNAi News, 5/25/2006) and further examined by Harvard Medical School's Debora Marks and colleagues last year — as well as by directly impacting miRNAs with antagonists and agonists, he said.
To address the possibility of these kinds of off-target events, Levin said the OSWG drafted "clear recommendations," including that "candidate drug sequences, either antisense or siRNAs, should be selected to minimize the number of meaningful off-target effects.
"In practice, there are many small companies that fall in love with a sequence and then figure out that maybe they ought to look for off-target events," he noted. The OSWG "strongly felt that we … should make it clear that the best place to eliminate off-target events is by understanding what you have [early] in the selection process."
The best way to do that, he said, "is to search gene databases … for the presence of homologies" with candidate oligo sequences.
Levin also noted that while avoiding sequences with homology may not always be possible, "it may not always be necessary."
For example, a "critical factor in understanding the relevance of any particular off-target event is … the IC50 between knocking down your intended target and knocking down your off-target," he said. "If you have one or two mismatches in your off-target gene, is it going to necessarily be meaningful when you're getting a ten picomolar inhibition of your target gene and you need a 200 picomolar inhibition of your off-target gene?
"We all need to keep that in mind before we go running off getting worried about a potential off-target event," he noted. "If we're not delivering toxicologic amounts of drug and we're delivering pharmacologic amounts of drug, these kinds of relationships become very important. It may be that off-target events are going to be almost non-existent in a real-world situation."
One of the best ways of evaluating the significance of an off-target event is "by looking at human cell lines," Levin said. "Once you've identified something in an in silico screen, will you get a real hit when you use human cell lines?"
[ pagebreak ]
And while "the ultimate test of hits from in silico screening are in vivo studies," differences exist between species, Levin pointed out. "Obviously, not all mRNAs from all species are going to be the same, and we're always going to be left with questions. That is why [the OSWG places] so much emphasis on looking in human cell lines."
But in the end, "understanding what happens in real animals under real conditions of potential use that mimic the clinical situation as best as possible" remains the best way to identify the unintended effects of candidate oligo drugs, he said.
The OSWG also recommended that drug developers use transcriptional profiling to explain toxicities or unexpected effects, although the "value of them in routine testing is unknown," according to an overview of the group's findings.
Additionally, perturbations are to be expected with oligo therapeutics, as with any other kind of drug, "and their clinical significance will be difficult to interpret."
"In the small-molecule world, we've become very familiar with the fact that there are going to be off-target events … [but] don't have any way of really defining what off-target events are," Levin said. By comparison, "in the oligonucleotide world, we come up with a list of the most likely suspects simply on the basis of bioinformatics information.
"Here we have to be careful," he noted. "Just because we know more, does that mean we have to be treated differently? Just because we understand the mechanism for a potential off-target event, does that obligate us to always look for off-target events?"
The OSWG recommended that researchers characterize off-target effects "on a case-by-case basis as to when it is needed and what role it should play in the drug-development and regulatory processes."
The OSWG includes researchers from Topigen, Alnylam Pharmaceuticals, Roche, Merck, Pfizer, Santaris Pharma, the US Food and Drug Administration, Quark Pharmaceuticals, RXi Pharmaceuticals, GlaxoSmithKline, and the University of Washington, and made its recommendations based on discussions held over the past few months, Levin said.
He stressed during his presentation that the findings and recommendations of the committee do not necessarily reflect the positions of the institutions and companies to which the committee members belong.