Another Second-Generation GalNAc-RNAi Rocked by Off-Target Concerns

With the approval of ONPATTRO and the current overall clinical picture, in particular the highly encouraging safety profile seen with PCSK9-targeting Inclisiran in well over a 1000 subjects (à The Medicines Company and Alnylam), the RNAi mechanism can be considered a clinically validated  new drug modality. 

Liver enzyme elevations, however, have been seen across a few clinical programs.  This includes ALN-AAT for alpha-1 antitrypsin-related liver disease, ALN-AT3 for hemophilia, and now also Givosiran (ALN-AS1).  The most likelyreason for these observations is that the minimization of off-targeting had been neglected for these ‘2^nd-generation GalNAc-RNAi triggers’, thereby raising the risk that the relatively high off-target noise (dozens of off-targets) can be detrimental to target cell health. 

To wit, the spectrum of off-target genes varies tremendously from RNAi trigger sequence to RNAi trigger sequence and affects biologically unrelated genes.  This makes it very hard to predict whether the off-target noise will have adverse effects or will be just irrelevant noise to the cell.  Not helping is the fact that lab animals due to their genome sequence differences cannot model the safety around off-targeting.  The hope is that perhaps organ explants or mice with humanized organs (e.g. humanized mouse livers) will be better predictors since the target cell is almost the same as in the clinical setting.

Marked liver enzyme elevation in Givosiran pivotal trial

This week, Alnylam provided an interim read-out from the ENVISION trial, the registrational study with GalNAc-siRNA Givosiran for acute hepatic porphyria (AHP).  AHP is a rare disease caused by defects in heme synthesis.  This results in the accumulation of toxic intermediates such as PBG and ALA which can cause a range of symptoms, including severe pain attacks.

The study envisioned an interim biomarker read-out in a first cohort of patients that had been treated for at least 3 months with study drug.  It had been agreed before with the FDA that if lowering of ALA could be found in this read-out, it might form the basis for an accelerated approval.

Unsurprisingly, urinary ALA levels were reported to be statistically significantly lower in patients treated with Givosiran compared to placebo- in-line with the phase I/II experience (p<0 .001="" span="" style="mso-spacerun: yes;"> 

Unfortunately, there were numerically more than twice as many Givosiran-treated sujects (5/23) that experienced a serious adverse event (SAE) than in the placebo group (2/20).  One of the Givosiran-treated subjects had to discontinue treatment due to liver enzyme levels above the pre-specified threshold (>8x upper limit of normal).  While no liver enzyme elevations had been reported in a similar number of patients in the earlier-stage studies, I wished the company could have talked more about the presence of other liver enzyme increases of less than 8x ULN to get a better sense of the magnitude of the problem and whether such incidence would be compatible with giving the drugs in a chronic fashion.

Having said that, with 5 SAEs in 23 subjects treated a little more than 3 months on average (on top of a case of drug-related anaphylaxis in the phase I/II open-label extension study), SAEs that might be drug-related, would already seem to invalidate Givosiran being used in those with the same genetic condition, but not necessarily with recurrent pain attacks.  This would have tremendously extended the market opportunity of Givosiran.   

In the same vein, with the present safety questions, an accelerated approval that would shave off a few months before the full dataset will be available in early 2019, can now be ruled out with high certainty.

To add insult to injury, Akcea reported earlier this week that a GalNAc-version of the competitive RNaseH antisense technology did not suffer from thrombocytopenia as unconjugated ASOs typically do, thus putting ASOs back in the race for liver targets.  Still, in the long-term and in particular with the more specific new GalNAc-RNAi trigger chemistries, RNAi should win for most liver targets, but for some targets where either RNAi has inherent difficulties achieving high potencies or in cases where off-targeting toxicity still strikes (albeit at a much lower rate), RNaseH ASOs may emerge victorious.

Disclosure: long ALNY, but have taken some off the table following these data and questions around cardiac issues with ONPATTRO; it is important now for Inclisiran to re-instill confidence in the safety of RNAi Therapeutics; short AKCA as I do not believe that their current TTR and triglyceride-lowering drugs TEGSEDI and WAYLIVRA, respectively, can be commercially successful drugs and its artificially bloated market cap is a result of the low float and short interest in this stock and not a reflection of the value that the market sees in Akcea.

Injection Site Reactions and Liver Toxicity Emerge as Major Issues for GalNAc-siRNA Technology

Alnylam this morning reported top-line results from a phase II study of their first clinical candidate based on the GalNAc-siRNA conjugate delivery technology, ALN-TTRsc for TTR amyloidosis.  Accordingly, average knockdowns of ~85% were seen in the 5mg/kg dose group, thus confirming the robust potency at around the dose that the company plans to take forward in later-stage studies.

Such potency, however, came at the apparent expense of relatively frequent injection site reactions (23% of patients), with an additional skin reaction seen outside the area of injection.  Moreover, there was a trend towards elevated liver enzymes, a marker of liver toxicity, including one that was adjudicated a serious adverse event (SAE; ~4x ULN deemed mild severity).

The efficacy data do not come as a surprise given that they were largely in line with that seen in the phase I study a year ago, which included the same dose group (5mg/kg) at the same dosing schedule (first 5x daily, then weekly for 5 weeks).  In both cases average TTR reductions  of ~85% were seen, with the difference being that this study involved 23 subjects at this dose (plus 3 subjects at 7.5mg/kg which was not further pursued for undisclosed reasons) while the phase I study involved only 3 comparable subjects.

The efficacy is thus in line with the intravenous ALN-TTR02 which utilizes Tekmira’s liposomal delivery technology and which so far has been very well tolerated.  Critics (aka LNP haters), however, are keen to point out the use of (transient) immune suppression.  The efficacy of ALN-TTRsc is superior to the antisense compound by ISIS and GSK which has shown 70% target gene knockdown in a short 4-week phase I study.  Assuming maximal knockdown efficacy has not been reached at this time-point, ISIS-TTRRx is likely to max out between 75 and 80%.

Similar to potency, the injection site reactions were not really a surprise given that in the phase I study this was the most common side effect.  What is new is that there was a skin reaction that occurred outside the area of injection possibly indicating systemic immune activation.

The liver enzyme elevations, however, were certainly new.  This could be related to histologic observations of granulations in the cytoplasm of hepatocytes, likely reflecting storage sites of the fairly stable modified RNAi triggers.

My hope and expectation is that particularly the injection site reactions, but also liver enzyme elevations will lessen with the lower doses enabled by the more effective second generation ESC GalNAc chemistry.  Still, for ALN-TTRsc the safety profile looks adequate for an indication like FAC (familial amyloidotic cardiomyopathy), but 23 is still a small number to be sure.  

Lastly, if I had to choose between ALN-TTRsc, ISIS-TTRRx, and ALN-TTR02, I would go with ALN-TTR02 with the best apparent risk:reward, regardless of whether it's an intravenous procedure or not.   

Disclosure: Long ISIS Pharmaceuticals, no positions in Tekmira and Alnylam.

ARC520 for Chronic HepB: The Immune System…It’s Reactivating!

Usually, immune stimulation which manifest itself as flu-like symptoms and the like is something that you do not want to see with an RNAi Therapeutic.  Having said that, there are settings in which you want to see immune stimulations.  This is also the case for the treatment of chronic HBV with a HBsAg knockdown approach. 

ARC520 is the lead candidate in this effort and yesterday the sponsor, Arrowhead Research, presented very exciting data in a chronically infected chimpanzee that had been treated with the RNAi agent.  The data are not only exciting because they presumably represent the fastest reduction of the HBsAg antigen ever seen in a real infection (note: only the chimpanzee is thought to reflect human HBV infection), no, what is even more exciting is that the data are consistent with a HBsAg-specific T-cell immune response.  

As the success of an HBsAg knockdown approach for the treatment of chronic HBV hinges on the hypothesis that by knocking down HBsAg it might be possible for the body to re-activate an adaptive immune response against the virus, the data in my mind are the biggest de-risking event in the development of ARC520.

The data

At the 2013 Liver Meeting of the AASLD in Washington DC, Arrowhead Research presented more detailed data on the chimpanzee that had been treated with ARC520.  Before that we had known that two doses of ARC520 (one of 2.0mg/kg and one of 3.0mg/kg) spaced 14 days apart was able to knock down HBsAg by ~80%, HBeAg by over 90%, and serum HBV DNA by 1-2logs.  The important knockdown here is that of HBsAg, an otherwise ‘undruggable’ target.  Reverse transcriptase inhibitors such as entecavir are not able to meaningfully reduce HBsAg.

The discrepancy of the degrees of HBsAg and HBeAg/HBV DNA knockdowns is likely explained by the prolonged half-life of HBsAg.  Repeat dosing for an extended period of time should lead to even further reductions in HBsAg.  Moreover, the treated chimpanzee was an unusually tough challenge, because old (HBV for over 35 years alone), heavy and with extremely high viremia (>10exp10 genomes per ml). 

Importantly, as predicted by the HBsAg knockdown-immune reactivation hypothesis, there was an apparent T cell-mediated anti-HBsAg immune response shortly after peak HBsAg knockdown levels were reached.  The peak HBsAG knockdown occurred somewhere between ~days 25 and 38 and there was a flare-up in liver enzymes around day 43.   

 

A liver-enzyme flare-up would be expected when anti-HBsAg cytotoxic T-cells start to attack HBV-infected hepatocytes.  Consistent with it being due to a T-cell response, markers of T cell activation, most notably interferon gamma, were also up-regulated around the time of the flare-up.  Interestingly, liver enzyme levels did not fully revert back to normal, but remained somewhat elevated compared to base-line suggesting that the RNAi knockdown kicked into gear a more persistent immune response.

Such delayed dynamics are consistent with what is seen following successful treatment with interferons.  In the ~5-10% of cases where interferons are able to achieve the gold standard HBsAg elimination (in the presence or absence of HBsAg seroconversion), the elimination usually occurs after the ~52 week course of interferon, in many cases years afterwards.

Of course, the present chimpanzee data do not show an elimination of HBsAg.  While I do not exclude the possibility that Arrowhead Research will come back in another 3 months or so to report that the immune system has finally overcome HBsAg, I prefer to keep expectations for such an event low and instead consider the present data as a nice starting point that indeed ARC520 is able rekindle the desired immune response after only two doses over 14 days.

Why it is unlikely to be a non-specific immune response

It is very important here to emphasize that what we are seeing here was not due to a non-specific innate immune response triggered by an immunostimulatory RNAi agent.  Similarly, some RNAi delivery strategies run the risk of causing direct damage to hepatocytes which would also manifest itself by increases in liver enzymes.

The most convincing argument to me is in the timing of the flare-up and cytokine elevations.  While non-specific responses usually occur in the hours and days immediately following RNAi administration, in this case, they occurred 3-5 weeks after the second dose.

Consistent with a ‘clean’ safety profile of ARC520, no such liver enzyme and cytokine elevations were seen in the phase I volunteer study for which Arrowhead reported initial safety data a month ago (2mg/kg highest dose in that study).  Nevertheless, as those data only focused on the safety in the first few days following drug administration for the above reasons and the 30-day follow-up still remains to be reported, one formally cannot exclude the possibility that the unique DPC chemistry is associated with liver damage and the like only weeks after drug administration.  I consider this quite unlikely though.

Going along with this theme, I expect the phase IIa study which will involve infected patients in Hong Kong and is scheduled to initiate enrollment in early 2014 to be ARC520 on top of an RT inhibitor such as entecavir.  As RT inhibitors stabilize the liver of HepB patients and in light of the phase I data, any flare-ups that would be seen in that single-dose study would presumably be the result of HBsAg-specific immune reactivation.  

Lots of exciting catalysts ahead over the next 6 months and who knows, due to intrapatient variability, maybe there will be a cure or two in the phase IIa study already.  

PS: For more background on RNAi knockdown for HBV and ARC520, please visit my cureHBV companion blog.