Skip to main content

Clinical Trials

Revusiran’s Failure Revisited

I wrote about Alnylam’s recent clinical disaster here, where they had to stop a Phase III siRNA therapy trial against a rare amyloidosis with cardiac complications. A reader sends along the link to this 2008 paper that suggests a possible reason for the excess deaths seen in the trial. The authors studied a range of different siRNA constructs in mice, and found that (as you’d expect and/or hope) that sequences targeting genes for VEGF inhibited vascularization. That’s what that growth factor is for (among other things), so that’s fine.

The problem, though, is that they found that pretty much everything else they tried (siRNAs for both pro- and antiangiogenic genes, for totally unrelated genes, for things that don’t even match a target at all) also showed these effects. This was one of the results from this period that alerted people to the effects of some siRNA agents on TLR3 signaling, which is one of the sensors of foreign RNA (and thus of infection). Human endothelial cells most definitely express TLR3, so (as the paper put it), you could say that siRNAs, even untargeted siRNAs, might be a clinically useful antiangiogenesis agent, or (alternatively), if you’re trying to target something else, you might be asking for vascularization side effects that you don’t want.

These concerns complicated the development of bevasiranib and AGN-745/Sirna-027 (both siRNAs directed at VEGF), but it does have to be noted that neither of them was associated with any severe side effects (rather, the bigger problem seemed to be efficacy). That said, these were injected directly into the eye compartment, so the possibility of systemic effects was probably minimal in any case. Overall, though, many suspect that what efficacy the first wave of siRNA clinical agents had may well have been due to this nonspecific TLR activation, and this was something that the field had to overcome.

VEGF effects would certainly seem to be something you’d want to look out for in patients that are already showing cardiac damage, but what I don’t know is the degree to which Alnylam’s GalNAc conjugate siRNAs (like revusiran) activate the TLRs. This seems like an issue that would have been publicly addressed, but I’m having trouble putting my finger on anything that directly does so. Perhaps it was dealt with right up front, and I’m missing it, but if anyone who follows the RNA therapeutics world more closely than I do can shed some light on this, I’d be glad to put up some links. . .


4 comments on “Revusiran’s Failure Revisited”

  1. KTP says:

    TLR activation and off-target silencing have long been a focus of study in the siRNA field from what I have seen. Off-target silencing is a more recent discovery, where the siRNA acts as a miRNA (reading only positions 1-8 of a target as opposed to a full sequence pairing) which allows for more indiscriminate silencing. If you look into the chemistry of RNA modifications, there have been a number of studies (Pete Beal @ UCDavis stands out) that have investigated what modifications (simple, or designed) can help control these effects. People have established how adding modifications to the major/minor grooves of RNA duplexes they can control TLR effects, and Beal’s more recent work with modified guide strands has established that off-target silencing can be controlled as well. As far as Alnylam’s clinical endeavors, I imagine it depends on what Alnylam did as far as modifying their RNAs, but if there was any nonspecific silencing, who knows what they were hitting besides their target.

  2. Charles U. Farley says:

    Alnylam (and all other modern RNAi companies) use *highly* chemically modified RNAs for their drugs. The (multiple, and multiple types of) modifications do two things, as mentioned above:
    1) abrogate immune stimulation
    2) greatly reduce off-target cleavage.
    Combined with modern algorithms for off-target screening, the concerns raised in the old literature (pre-2010) have been about as well taken care of as possible.
    Now, whether and how other class effects, such as:
    – protein binding by the siRNA
    -“typical” tox baggage due to retention in non-target tissues
    – on-target toxic phenotypes further down in the target gene’s pathway(s)
    etc. contribute to the tox they’ve seen … that remains to be determined.

    The issue in the revusiran trial isn’t tox per se as much as lack of efficacy and a concomitant poor risk/benefit ratio.

  3. steve says:

    The TLR effects are well-known and it’s unlikely to be involved in the observed toxicity. There are a number of chemical modifications of siRNAs that have been described that avoid TLR activation while retaining gene suppression activity (e.g., 2′-O-methyl modifications). TLR activation is also unlikely to explain the neuropathy observed in the trial. I think we need to look elsewhere to explain the observed tox.

  4. jtsuhmd says:

    The investigators will perform an analysis to look at causality. Off-target side-effects significant enough to cause premature death are one potential interpretation. Another plausible explanation is the study design. Perhaps the study population was too sick or the intervention was too late in the course of the disease. With a small study group in a heart failure population there is always a risk for unexplainable statistical abberations.

Comments are closed.