Anthony, Katy, and Don got together in the last week of September, 2021 to discuss both short- and long-term solutions to the concerns regarding AAV safety and efficacy: everything from the need for spec guidelines to possible AdComm patterns to possible technical solutions. We’ll kick it off with one of @ADaviesDHCG’s many tweets from his virtual meeting attendance:
Right, so that tweet was from the morning of Sept 2nd and by the end of the AAV AdComm meeting we still hadn’t received any news that was even guideline-adjacent. I’ve been beating this drum for a while now and, while we certainly have received many a guidance doc in the last several years (including these two just from earlier this week: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/interpreting-sameness-gene-therapy-products-under-orphan-drug-regulations and https://www.fda.gov/regulatory-information/search-fda-guidance-documents/studying-multiple-versions-cellular-or-gene-therapy-product-early-phase-clinical-trial), we haven’t seen specification guidelines, and that’s something that we really need. To explain why, let’s first step sidewise for clarity.
The bad things that AAV will do to you are largely capsid mediated: pro-inflammatory immunostimulatory organ-overloading. But the purpose of that capsid is of course to deliver a payload, which takes us to a discussion of things that can go wrong inside the capsid.
If the capsid is empty, it’s just useless. If it’s got an incomplete payload, it is best-case useless and worst-case problematic. If the capsid is misformed and looks ok but cannot deliver its payload, then it’s also useless. If we want to maximize efficacy while minimizing those pro-inflammatory effects, it is critical to maintain control of the percentage of what I’ll call “viable viruses,” those capable of delivering their payload.
At this point in this field, we should have a sense for what is feasible here; I’d encourage all of us to be more willing to go ahead and say a number.
OK, so let’s pick a number. What would it be?
Look, I’m intrigued by the similarities between dead cells and empty capsids. Potency? Zero on both counts. Potential to cause inflammation? High on both counts. Right? It took us from 1991 to 2008 for specific guidance on a cell viability specification to be published, and that number is 70%. I’d be willing to receive a copy of that same document with the words “cell” or “cellular” simply replaced with the words “viral vector.” What are the chances that the viral vector viability minimum acceptable number will at some point be placed at 70%? I think that’s a good solid guess. Just you wait… #DHCpredicts
And this can be done even in the absence of a universal reference standard because, in the course of validating an assay, you have to generate your own reference standard.
Impurities have specs: let’s start identifying and narrowing down those we know we should be concerned about. That would serve a dual purpose: providing a guardrail for developers and also identifying avenues for potential new technologies that might identify or correct for impurities.
In fact, there’s already an established mechanism for anyone, anyone at all, to write specific guidance and submit it to the FDA. We need some experts in the field to go ahead and do that. Give the FDA a place to start – a first draft, if you will. And as Anthony mentions above, if you had nothing else to add you could literally start with the latest cell therapy guidance as an example and suggest 70% for the full:empty capsid ratio.
I know some developers may not want to see a specific number put forward, but in reality, as long as the number proposed is realistic, I don’t think they should be. As I often point out, guidance (even updated “final” FDA guidance) is not the same as an unbreakable rule, so why are we so loath to put that stake in the ground?
Guidance acts as guardrails, an effective way to set expectations, so that developers know approximately what ranges to strive for as well as where they might be expected to make the case for why something is a reasonable exception to that norm.
Developers just need those guardrails in place so that they can meet expectations. We’ve all seen what can happen when sponsors get misaligned with FDA expectations—just look at the 2020/2021 spate of delays and rejections—so let’s not make meeting them a guessing game. Time is of the essence for these therapies; lives are on the line.
That urgency was definitely felt in the meeting, which is why I had higher hopes for a request of viral vector spec guidelines to come out of the committee. I think we all have an opinion on why that didn’t happen.
I’ll jump in with a little perspective, first. It’s important to understand that this recent AAV AdComm was different from the type of meeting that is typically conducted. It reflects an interest in using the cell and gene therapy advisory committee differently so that instead of being convened only when product approvals are on the line, they can be used to have a public airing that spotlights specific issues of concern. So, the meeting was intentional and it reflects justifiable concern in the industry related to AAV toxicity.
The AdComm was extremely informative in terms of highlighting potential safety issues and in that sense it was useful, but if anyone’s goal was to hear a “we back you FDA on these positions and your efforts to address” from the committee or, as Anthony mentioned, a call for specification guidelines, then it wasn’t successful.
And to Anthony’s point about why that happened, I agree that it looks pretty clear from where we sit. Clinical experts are extraordinarily valuable and were obviously needed in a meeting like this in which deaths during clinical trials were the impetus for convening the meeting, but especially in C>, the answer to these clinical issues is likely to run through improvements in CMC.
First, a quick note on those clinical issues you mention, and it has to do with patient populations. In these high-dose AAV studies, the human deaths have been in the Astellas/Audentes trial.
That is relevant because we need to be facing the cruel truth that the life expectancy for participants in the Astellas/Audentes trial is one of the worst in a high-dose AAV study, with an average life expectancy of 29 months. Not only that, but the diseases being targeted here align with a history of liver issues. So, we have the heartbreaking reality of families with kids facing nearly certain death at an impossibly early age, who have prior liver issues, and who are taking a chance on a therapy that cannot (at this stage, at least) avoid affecting the liver. A trial wouldn’t typically accept participants who are facing odds like this but it’s the only way to attempt to give these kids even a chance. That doesn’t make these deaths even the least bit less tragic, but it does provide critical context for why we need to maintain a sense of urgency in looking for those CMC solutions.
As we love to do when the question of CMC in C> comes up, let’s quote Scott Gottlieb, from his remarks to ARM’s annual board meeting in May of 2018:
“In contrast to traditional drug review, where 80 percent of the review is focused on the clinical portion of that process, and maybe 20 percent is focused on the product issues, I’d say that this general principal is almost completely inverted when it comes to cell and gene therapy.”
From the clinical side of things we’re identifying the problem and providing clarity on any patterns, any further questions that need to be answered, etc., but then in order to solve that problem we’d need to turn to the CMC folks. Without a CMC-featured component to the meeting, there wouldn’t be any way to even identify which product specs were needed. And of course clinicians wouldn’t be expected to have answers to that.
Right. I mentioned in last quarter’s Straight from the Horse’s Mouth that the AdComm that was convened for Mesoblast’s remestemcel-L (RyoncilTM) two summers ago wasn’t the right match for the questions being raised and this AdComm saw a redux that we didn’t need or want. We’ve seen two examples of this disconnect thus far and while two may not yet demonstrate a pattern, it’s worrisomely close to doing so.
Getting the right experts for the job is a critical short-term need. FDA is in the habit of convening clinical ad comms because they’re in the habit of the issues at hand being clinical issues but they haven’t yet taken that logical next step of putting CMC folks on the AdComms for cell and gene.
I’ll say it louder for those in the back: either C> AdComms need to be split between clinician/CMC experts or we need a new separate follow-up AdComm populated with CMC experts.
Agreed; in fact, I’m on record already as saying that the wrong people were invited. If we try to take a look at who should have been invited, of course, we run into complications with ethics screening, which takes time, as well as conflicts of interest, which all of the most informed individuals are going to have. So, finding the overlap between people who should be there and people who could be there is the challenge.
That said, we should be looking at people like Zenobia Taraporewala from the Office of Tissue and Advanced Therapies at CBER. She’s publicly said (and I’m paraphrasing,) “anything more than a 50% impurity is a non-starter”—why wasn’t she there? Or Nicole Paulk or Andrew May, two of the authors on last year’s Methods Matter paper. Or Frasier Wright, formerly of Spark Therapeutics, now at Stanford.
There are people like these who have expertise on these CMC topics. If we avoid convening the right experts for both sides of the equation, then actionable answers are going to continue to evade us.
So, if specs and a CMC AdComm are two shorter-term necessary steps, what’s a longer-term solution that we can get started on so that we’re working the problem from all angles?
Technical solutions fit that bill, which is why I’m passionate about them. We need to focus on them now if we want to have ways to protect all these future developing therapies from clinical disasters. But even within that framework, some of these solutions are nearer-term than others.
So first off in this category, let’s talk about capsid engineering to improve viral tropism. As we all know, the trouble with AAV is that it’s largely a non-tropic virus, and for that reason I believe altering tropism is ultimately one of the highest potential solutions to the problem of high dose AAV safety. When we look at the highest of the high dose AAV therapies, those dosing in the range of 1014 vg/kg, these are all products for muscle diseases: Zolgensma, Astellas’ XLMTM program, Sarepta’s LGMD programs, and DMD programs from Sarepta, Solid, and Pfizer.
This is happening only for muscle diseases for what I think are two reasons:
1) The skeletal muscle is among the largest organs in the body, so even if the delivery mechanism had good muscle tropism, you’d still need a lot of virus just to transduce enough cells.
2) But, on top of that, because you’re fighting against that liver filtration effect, you’re essentially racing to get that dose to the muscle before you run up against liver tox.
So, if you could engineer a capsid that would be only one order of magnitude more effective than current capsid designs in targeting skeletal muscle over liver – and that doesn’t seem like a terribly high bar – then you could reduce your dose ten-fold. That’s huge. That widens the therapeutic window and gives all patient populations a better shot. And the more at-risk the populations are, the more of an advantage such a change will give them.
And, as promising evidence that such an order of magnitude improvement may be possible in the relatively near term, I’ll point you towards the article that came out in Cell this month about the recent identification of a muscle-tropic AAV capsid variant. In various in vitro and in vivo screens (with the obvious caveats about predictiveness of model systems), it identified improvements in muscle targeting in the range of 3 to 50-fold, with corresponding decreases in liver targeting.
Yes, but having familiarity with capsid engineering companies I can tell you that the answer to “good news, I have something way more tropic” is going to be less “ah, excellent, let’s lower that dosage,” and more Bruce Willis from Die Hard, aka “Yippee-ki-yay, MF’ers! let’s crank more virus in there!”
Good point – and I need to acknowledge that capsid engineering is exceedingly difficult. It’s going to take time and it’ll put us back at the starting line from a development and regulatory perspective, so it’s more a second-generation approach than a fix for programs already in the clinic. In that case, though, it’s better that we start exploring this solution now rather than later. Fundamentally, I do believe it’s going to be one of the eventual answers.
Also on the tech solutions front, there are also nearer term options to consider, such as a way to effectively separate full capsids from empties…
For which there’s already a credible method: ion exchange using monolith chromatography.
We could also take a look at better promoters. Due primarily to space limitations that hinder use of more selective options, current promoters basically just drive massive amounts of gene expression in a non-selective way, which is also potentially quite risky from a safety perspective.
Yes—to continue with the movie quotes, if we think of a promoter as basically the volume knob on gene expression, the options to date have all been some version of Spinal Tap’s “turn it up to eleven!”
We should also consider the possibilities of a non-viral option as a solution. mRNA-based gene editing approaches might be the ticket to a non-viral delivery and as Katy noted on LinkedIn as one of her takeaways from the ASGCT’s recent Policy Summit*, the mass scale of COVID vaccines has opened up a whole new non-viral avenue by generating an impressive proof of concept concerning feasibility, scalability, and cost -of these technologies. I’m sure we’re not the only field looking much more closely at this possibility than we might have before COVID.
So, to summarize here, we’re all in agreement that the importance of CMC issues in Cell and Gene Therapy product development argue for increased representation of CMC expertise on CGT AdComms. Yes, we need clinicians, but we also need CMC experts.
And we agree that someone—be that some guy off the street who submits brilliant guidance or be that FDA experts or anyone in between—needs to begin laying down guidance for important, potentially safety-related characteristics such as impurities in gene therapy products. We need those guiderails for the good of all in this field.
And since newer technologies may well be the solutions we will need in future in order to achieve those specs, we’d better start focusing on them sooner rather than later.
* Katy’s summations from the Policy Summit can be found here: Day 1, Day 2, and Day 3.