Arrowhead Achieves Robust Alpha-1 Antitrypsin Lowering As Grounds for Trial Termination Still Shrouded in Mystery

Last Friday, Arrowhead Pharmaceuticals presented first data from its phase I study of ARO-AAT being developed for addressing liver disease in people with certain mutations in the alpha-1 antitrypsin gene.  Based on the selective data release, Arrowhead has assumed the lead in this indication as Alnylam struggles to regain its footing following apparent off-target-related liver toxicity almost 2 years ago.

The press release can be found here, the actual presentation here. 

With a mean maximal target gene knockdown of -87% between 6 to 8 weeks following a single subcutaneous injection of 100mg dose of the GalNAc-enabled ARO-AAT, Arrowhead has shattered the previous AAT knockdown record by Alnylam’s ALN-AAT of -80% at a ~4x dose of ARO-AAT.  The dynamics of the knockdown also suggests that infrequent dosing should be feasible with ARO-AAT, although without seeing the multi-dose data from this study, it is difficult to predict whether we are talking about quarterly or semi-annual dosing here.

It will also be interesting to find out whether the variation of knockdown in the 4 patients- ~-70% knockdown at 4 weeks for the 2 weaker responders and ~-90% for the 2 better responders- had to do the polymorphism issues similar to those encountered by Alnylam before.

Safety mystery remains

When Arrowhead announced on June 18 that it would terminate the healthy volunteer trial prematurely given that it had escalated above doses at which maximal knockdown can be observed, I was hesitant in taking it at face value or whether this decision also had something to do with the safety profile of ARO-AAT.

The Alpha-1 National Education Conference update only added to the impression.  Firstly, because Arrowhead must already know the knockdown from the 200mg and perhaps also the 300mg open-label cohorts since at least June 18, why didn’t the company simply show the data?   

Regarding safety, the company chose a cut-off date of June 11, that is a week before the June 18 decision.  By June 11, there were only 2 drug-related injection site reactions among the 32 subjects that had received at least 1 dose of either ARO-AAT (n=20) or placebo (n=12) per slide 17 of the presentation.  Those happened to be in the 2 of 4 100mg open-label subjects, the only subjects for which the knockdown had been reported.  

Confusingly, the company also reported that 44 subjects had received at least 1 dose (at least as of the trial termination decision date of 18/6), meaning that 8 subjects had further received 300mg of ARO-AAT (4 open-label, 4 blinded) and 4 placebo (all blinded).  Adding to the confusion, a company representative emailed me in a response to a tweet of mine on the ISR frequency that the safety update referred to those 40 subjects that had received at least 1 injection as of 11/6, but- as I said- contrary to this statement the table on slide 17 only included 32 subjects.  For all those others, the safety data were missing entirely.

The question now is whether the omission was intended as a teaser for the Liver Meeting presentation in November, a minor math issue, or whether there was something more nefarious to it.  The speed with which the company will move into patients (which already have or are at risk for liver disease) will be an important indication whether liver toxicity or the like has been observed subsequent to June 11.

Alnylam GalNAc Improvements Incremental, but Likely Enough to Beat PCSK9 Antibodies

Over the last week, Alnylam presented pre-clinical data for their new development candidates for the treatment of hypercholesterolemia (ALN-PCSsc) and liver disease related to forms of alpha-1 antitrypsin deficiency (ALN-AAT).  These candidates are based on second-generation GalNAc-conjugate chemistry that the company is now dubbing ESC (Enhanced Stabilization Chemistry).  They involve the increased use of chemical nucleic acid modifications for stability with attendant improvements in knockdown potency and duration over first-generation GalNAc conjugates such as ALN-TTRsc.  ALN-TTRsc is the lead GalNAc candidate and currently in phase II development.

Borderline first-generation GalNAcs

In a phase I study of ALN-TTRsc, potent knockdowns were achieved with about:

-60% knockdown at 2.5mg/kg;

-80% knockdown at 5.0mg/kg (ED80), and

-90% knockdown at 10.0mg/kg.

At the risk of insulting medical geneticists for oversimplifying, assume that an RNAi technology that can safely achieve an 80% target gene knockdown provides for a solid platform.  In the case where a subcutaneous route of administration is desired and/or necessary, this should ideally also fit into a 1ml injection volume which in the case of GalNAc conjugates would correspond to a 2.5mg/kg dose.

There is some controversy around acceptable injection volumes and as often is the case, increased standards are applied to RNAi Therapeutics.  I say this because drugs and drug candidates such as expected mega-blockbuster PCSK9 antibody from Amgen, AMG-145, has been administered at 2ml volumes.  In fact, to achieve once-every-4-week dosing, 6ml (3x2ml) have been administered (see Giugliano et al. 2012).

In light of this, ALN-TTRsc has failed the 1ml test, but 80% are certainly possible with this first-generation GalNAc chemistry.  There is therefore room, and in some cases a competitive need (àcompetition with more potent delivery technologies such as Tekmira’s SNALP LNPs and Arrowhead’s single molecule DPCs) for improvements in GalNAc conjugation technology.    

Second-generation data could indicate progress

In agreement with this, Alnylam is now advertising the ESC second-generation GalNAc and has presented critical non-human primate data for their new development candidates ALN-PCSsc and ALN-AAT.  Non-human primate data are ‘critical’ because the RNAi knockdown observed in monkeys typically closely predicts what will be seen in humans based on the experiences with ALN-PCS (SNALP), ALN-TTR (SNALP), and ALN-TTRsc. 

In interpreting the newly presented data, it should be noted that in the case of ALN-TTRsc, an 80% knockdown was already observed at 2.5mg/kg (and 5mg/kg) in non-human primates, but that this shifted to 5mg/kg in humans.  Numerically a relatively small difference, in practical terms an important one.   

ALN-PCSsc achieved 80% PCSK9 target gene knockdowns at 2mg/kg in a weekly multi-dose study in non-human primates.  This resulted in a highly competitive 60% LDLc lowering in the absence of statins.  Similarly impressive in light of the monoclonal antibody competition was that 80% PCSK9 knockdown and 50-60% LDLc reductions were achieved and sustained for over 3 months (!) when a single dose of 10mg/kg (2x2ml) was given.  Compare this to 57% LDLc lowering with AMG145 in a once-every-4 week regimen in the MENDEL-2 phase III study of AMG145:

-ALN-PCSsc (RNAi): 50-60% LDLc reduction, 2x2ml subcutaneous, once-every-3-months

-AMG145 (monoclonal antibody): 57% LDLc reduction, 3x2ml subcutaneous, once-every-4-weeks (phase III MENDEL-2 monotherapy study)

(Yes, I do keep an eye on cash-rich, sub-$2B market cap The Medicines Company, Alnylam’s licensee for ALN-PCSsc for this reason).

Before I get carried away with all the advantages of the RNAi platform over monoclonal antibodies for even extracellular targets such as PCSK9, in terms of GalNAc improvements, these knockdown results are very much in line with what was seen for TTR in non-human primates. 

The same applies for ALN-AAT where a single dose of 3mg/kg translated into a 60% knockdown which, based on TTR and PCSK9, will likely translate into a weekly multi-dose ED80 of 2.0-2.5mg/kg in non-human primates.

However, with the caveat of different half-lives for different target genes, the durability of the PCSK9 knockdown is quite impressive and more than what is typically seen with e.g. SNALP LNPs which have historically utilized minimally modified RNAi triggers.  Such triggers may be turned over more rapidly in liver cells.

Similarly, the RNAi trigger sequence in the TTR development candidate is an extraordinarily potent one with low single-digit picomolar EC50 potency in tissue culture.  It is therefore possible that equivalent animal potencies have now been achieved with less potent sequences meaning that the underlying delivery technology has improved.

It’s been fascinating to watch the various delivery platforms, and indeed RNaseH antisense, compete over the years and pushing each other to new heights.  In this context, I am anxiously waiting for Arrowhead’s presentation at next week’s TIDES meeting with the intriguing title:  “Next Generation Dynamic Polyconjugates for siRNA Delivery in vivo,”

Market commentary: I remain mostly on the sidelines.  Years ago, these scientific data would have made me jump head-first into the market with me buying all I could in the opening minutes.  These days, however, I view them as ensuring the long-term value of RNAi stocks, but fail to see how they will support share prices in the current, growth-to-value rotating trading environment for more than a trading week or so.  I am therefore speculating on a capitulation event before some of the important clinical data read-outs roll in.