WEBVTT
00:00:01.120 --> 00:00:14.640
Welcome to the Mind the Methods Podcast, brought to you by the International Society for CNS Clinical Trials and Methodology, your gateway to the latest insights, innovations, and conversations at the forefront of central nervous system clinical research.
00:00:14.880 --> 00:00:22.640
This series dives into the challenges, progress, and people shaping the future of CNS therapeutics development and trial methodology.
00:00:22.879 --> 00:00:31.839
In this first season, we'll explore innovations in CNS trial design, from digital endpoints and adaptive protocols to patient-centric models and generative AI.
00:00:32.079 --> 00:00:41.200
Here we bring together researchers, clinicians, regulators, and industry leaders to discuss the challenges and breakthroughs shaping the future of CNS therapeutics development.
00:00:41.359 --> 00:00:50.159
From advancing trial methodology to improving outcomes for patients, our conversations dive deep into the science and strategy behind CNS research.
00:00:50.399 --> 00:00:51.439
Thanks for joining us.
00:00:51.520 --> 00:00:52.399
Let's get started.
00:00:52.719 --> 00:00:59.359
Mind the Methods, brought to you by the International Society for CNS Trials and Methodology.
00:01:00.320 --> 00:01:06.719
Hi, my name is Camilleolo Gemel, and I am a scientific committee member at the ISCTM.
00:01:07.120 --> 00:01:11.040
I'm very pleased today to be speaking to Mo Haydren.
00:01:11.280 --> 00:01:11.599
Dr.
00:01:11.680 --> 00:01:19.519
Haydran brings a rare trifactor of being a scientist, a regulator turn strategist, and an industry operator.
00:01:20.000 --> 00:01:33.599
Mo spent several years at the FDA's division of cellular and gene therapies, 15 years in biotech planning research, product, and IP development, and nine years early in his career at the NIH.
00:01:33.920 --> 00:01:43.359
He's an author of more than 50 original research articles and a holder of 25 issued patents with 54 additional pending patents.
00:01:43.599 --> 00:01:57.680
He's a founder of a nonprofit organization dedicated to promoting education and translational research to better understand root causes of the diseases of the brain, including physical conditions that affect our mental health.
00:01:58.000 --> 00:02:09.520
Mo is currently serving as chief regulatory scientist at Cell X Incorporated Consulting, making him a particularly compelling voice for our audience here at ISCTM.
00:02:09.680 --> 00:02:14.719
And today's topics precision and safety in genetic medicines for CNS.
00:02:14.960 --> 00:02:23.280
This session is born out of a continuation of our February meeting where we tackled regulatory flexibility to accelerate CNS drug development.
00:02:23.439 --> 00:02:39.919
And today we'll cover basic topics such as taxonomy of gene editing platforms, CNS delivery systems, preclinical safety, trial design methodology, and lastly, where AI is beginning to shape how we think about editing all together.
00:02:40.240 --> 00:02:42.240
Mo, welcome to Mind the Methods.
00:02:42.319 --> 00:02:43.439
We're so glad you're here.
00:02:43.840 --> 00:02:45.120
Thank you so much, Camille.
00:02:45.680 --> 00:02:52.479
I'm glad to be here and I'm hoping that I can contribute to some of the topics that you outlined.
00:02:52.879 --> 00:03:03.759
So just to begin, Mo, you've sat on three distinct sides of our industry: the bench at NIH review table at FDA Seber and now industry strategy.
00:03:04.000 --> 00:03:15.280
And when you look at where genetic medicines, especially for the CNS, are today, does the progress feel faster or slower than you would have predicted from your SIBR days?
00:03:15.759 --> 00:03:16.639
Great question.
00:03:16.879 --> 00:03:25.439
I think before I get into your question, I do I want to come up with some definition and framework for our discussion.
00:03:25.599 --> 00:03:37.840
Obviously, genetic medicine is a broad term, and it includes a number of different modalities for replacing, correcting genes or silencing genes.
00:03:38.319 --> 00:03:42.879
Replacing correcting genes or silencing genes.
00:03:43.120 --> 00:03:57.120
And this topic of interest was fueled in 2010 with the discovery of some of these genome editing modalities and products that have really fueled the technology.
00:03:57.360 --> 00:04:15.520
So the underlying technology, for example, for genome editing is uh is very recent as compared to the use of AV or antiviruses to deliver genes for gene replacement strategy, which has been around for a few decades.
00:04:16.079 --> 00:04:20.639
So it really depends on the context of our discussion.
00:04:20.879 --> 00:04:33.360
But uh in reality, we have done an incredible job translating technologies that were used for research a few decades ago, and now they're used to treat patients.
00:04:33.600 --> 00:04:37.759
So from my perspective, that's that is an incredible progress.
00:04:38.160 --> 00:04:47.360
So when you think of the term genetic medicine, are you thinking of the broader framework that you just described, or in terms of DNA editing or gene editing?
00:04:47.839 --> 00:05:14.079
Genetic medicine is a broader term, but I think the conversation that we can have today with focus on genome editing, I consider that as a subset of genetic medicine, a subset, and it's fueled by recent progress that has been made, and everybody's aware of developing personalized genetic medicine for patients.
00:05:14.240 --> 00:05:28.639
And that is based on recent progress with Baby KJ's uh story of treating a single patient using gene editing modalities that can be done within six months.
00:05:28.800 --> 00:05:33.920
And that has created a lot of enthusiasm about personalized medicine.
00:05:34.160 --> 00:05:51.199
Personalized medicine has also a broader definition, but in the context of genome editing, really talks about treating and a single patient based on their genetic mutations as detected by companion diagnostics.
00:05:51.600 --> 00:06:12.319
So, in reality, I can have a broader conversation, but I realize that we want to focus due to the time constraint on the modalities that are used to change um genetic abnormalities that are found in rare condition, rare monogenic diseases as we go forward.
00:06:12.720 --> 00:06:13.839
And that makes a lot of sense.
00:06:13.920 --> 00:06:16.079
And yeah, thank you for bringing up baby KJ.
00:06:16.319 --> 00:06:21.360
I think we're all still pretty in awe about um the progress that he's made so far.
00:06:21.600 --> 00:06:34.079
So, Mo, just thinking about uh taxonomy and CRISPR-Cas9 versus base editing versus prime editing from a CNS development standpoint, how do you stratify these tools?
00:06:34.240 --> 00:06:38.399
And is the selection driven by the biology of the target?
00:06:38.720 --> 00:06:46.480
You know, say loss of function versus gain of function, or are delivery and safety constraints the dominant filter here?
00:06:46.879 --> 00:06:47.120
Yes.
00:06:47.279 --> 00:06:49.759
Um couple of things before I get into it.
00:06:49.839 --> 00:06:58.560
Uh, for genome editing, obviously, that's an area that has been going on maybe for more than a decade in the sense of clinical studies.
00:06:58.720 --> 00:07:02.800
And we're really dealing with different uh modalities.
00:07:03.040 --> 00:07:36.160
Initially, we were talking about the use of endonucleases which were larger, and then in 2012, about CAS series of endonucleases in the context of CRISPR's uh were developed, and essentially CRISPR really dominated the old uh legacy tradition of genome editing product, which are still under investigation by some companies, but CRISPR has become the dominant player.
00:07:36.399 --> 00:07:43.759
After CRISPR development, we see uh uh an emerging technology of developing better CAS.
00:07:44.240 --> 00:08:01.839
CAS-9 was the first one, and then there was a CAS-12A that was a variant of CAS molecule, and then we have base editors um focusing on adenine and cytosine base editors, and then prime editing.
00:08:02.000 --> 00:08:18.480
Each one of these modalities have really different profile, different characteristics, different safety characteristics in view of the presence of on-target versus off-target effects and other modalities that has to be factored in.
00:08:18.720 --> 00:08:34.639
So when we go from one platform to another, I believe we are dealing with uh multiple types of products, and we have to be cognizant that each product has a very unique characteristic that needs to be factored in.
00:08:34.960 --> 00:08:41.360
And just thinking further about the broad category of genetic medicines versus small molecules.
00:08:41.759 --> 00:08:57.360
For discussion, which I think is very important uh to sort of focus on these personalized medicines, which are in neurology's a definite target, and that is monogenic conditions that have mutation within a single gene.
00:08:57.600 --> 00:09:17.200
In the context of personalized medicine, I think one thing I want to elaborate on in the first question, make sure that we understand, and depending on what type of correction that needs to be made, there's going to be a sort of optimization of what platform would be most suitable.
00:09:17.360 --> 00:09:21.279
And this is something I missed in our previous discussion.
00:09:21.519 --> 00:09:38.559
And I want to kind of emphasize that although we have different modalities, we also have to optimize editors based on the target sequence, and that target sequence really impacts uh what's the most optimal way of developing drugs.
00:09:38.799 --> 00:09:49.279
To answer your question about developing drug product in genetic medicine, it's really driven by mutations that are found in the patient.
00:09:49.519 --> 00:10:00.159
For example, case of baby KJ, which had deficiency in CSP1, we know there are thousands of mutations within a single gene.
00:10:00.320 --> 00:10:05.279
And each patient has a very unique mutation that needs to be addressed.
00:10:05.519 --> 00:10:11.840
Not all mutations can be addressed using the same modality, and that has to be factored in.
00:10:12.000 --> 00:10:34.879
But because the mechanism of action is very well defined, this accelerates the drug development, because we start with having knowledge of what is the target sequence and ability to design appropriate product to correct that mutation in vitro, which is called ex vivo, as well as in vivo.
00:10:35.279 --> 00:10:45.919
From my perspective, because the mechanism of action is well defined, the target is identified and the modalities are available.
00:10:46.080 --> 00:11:01.279
The process of developing a drug should be accelerated tremendously, as evidenced by products that we have in market that have been developed within five or six years that are essentially close to approval or approved.
00:11:01.519 --> 00:11:09.440
So I think well-defined mechanism of action is very unique with respect to the broader category of genetic medicine.
00:11:09.840 --> 00:11:18.639
Yeah, and I think underpinning all that, um, a well-characterized biology and understanding that is the key to having a well-defined mechanism of action.
00:11:18.960 --> 00:11:21.120
So I appreciate your perspectives there, Mo.
00:11:21.600 --> 00:11:39.200
So thinking now about your role, both when you were at FDA and now as a consultant advising companies, what when planning I and D-enabling uh preclinical studies, are there any studies that you see sponsors underweighting chronically?
00:11:39.360 --> 00:11:42.159
For example, biodistribution, genotox.
00:11:42.240 --> 00:11:44.799
You you mentioned off-target effects previously.
00:11:45.120 --> 00:11:48.720
You know, what are you seeing in your conversations right now with sponsors?
00:11:49.200 --> 00:11:54.799
The conversation is really uh centers around the complexity of these products.
00:11:55.120 --> 00:12:03.120
These products are multi-components, meaning that they are not really essentially consist of one component.
00:12:03.440 --> 00:12:20.399
For example, a genome editing product, as we know, consists of the endonuclease, which is guided through guide RNA to the target sequence and delivered using, as let's say, a virus or non-viral mechanism using lipid nanoparticle.
00:12:20.720 --> 00:12:32.320
Because of the complexity of the products, each component poses an additional burden of quality requirement that is going to impact the final drug product.
00:12:32.480 --> 00:12:42.720
So, in a sense, having these multi-component products that are very complex really impacts discussion that we will have going forward.
00:12:42.879 --> 00:12:47.759
And it's really specific to the product of choice and going forward.
00:12:48.240 --> 00:12:54.799
Here you're saying is that product development in the space is increasingly complex due to multiple components.
00:12:55.360 --> 00:13:04.159
From safety perspective, is uh sometimes is the quality of material and manufacturing and manufacturing consistency.
00:13:04.559 --> 00:13:16.320
Because we, you know, 10 years ago, some of these reagents were research reagents, and now we have the requirement from a regulatory perspective to use high-quality materials.
00:13:16.559 --> 00:13:26.559
Sometimes manufacturing of these quality materials are not necessarily something that investigators are focusing on initially.
00:13:26.720 --> 00:13:33.519
For example, in Baby KJ, essentially in six months, a product was developed and the cost was very low.
00:13:33.759 --> 00:13:55.360
But when you have subsequent discussion with FDA, FDA has been flexible in many ways, but they're still really focusing on quality and the requirement for release tests for different components of product that needs to be factored in, that really create pressure on the costs, etc.
00:13:55.759 --> 00:14:23.600
The other aspect is about identifying on-target versus off-target efficiency of the genome editing product and really looking at what you describe as genotoxicity studies early on that needs to be factored in, as well as uh looking at possibility of germline transmission, et cetera, which is not a major focus during earlier stages.
00:14:23.919 --> 00:14:44.960
But uh in reality, after one identified the gene of interest and the target sequence, from that point on you build that manufacturing platform to produce these very specialized, highly complicated genome editing products that need to be delivered appropriately.
00:14:45.200 --> 00:15:11.039
So ultimately, there are major criteria or obstacles that one has to face that relates to quality of material, cost of material, reproducibility of obtaining those materials, manufacturing the genome editing products, and being able to evaluate those in vitro for safety as well as in preclinical studies.
00:15:11.200 --> 00:15:25.919
And the preclinical studies has been really an area that things have evolved in the last few years with respect to FDA's interest to really reduce the number of animal studies.
00:15:26.240 --> 00:15:47.200
And I think there has been some flexibility from FDA and reviewers to accept cell-based assay in lieu of some of the preclinical studies to assess off-target effects using cell lines, etc., that harbor different mutations that exist in the patient.
00:15:47.440 --> 00:15:51.600
So there are a number of things that essentially needs to be focused on.
00:15:51.759 --> 00:16:15.600
But if there is one topic that I can stress here more as we go forward, is emphasis on being able to manufacture these very complicated products using CDMOs or other entities that you depend on and be having control over that process to be able to have reproducible products.
00:16:15.759 --> 00:16:30.000
And the secondary to that is really focusing on genotoxicity up front and accessing on-target and off-target in view of any given product by having a robust package of in vitro and vivo data.
00:16:30.399 --> 00:16:39.200
Definitely don't cut corners on quality and manufacturing and genotoxicity assessments in the preclinical package.
00:16:39.759 --> 00:16:40.320
Exactly.
00:16:40.480 --> 00:16:45.679
And there are in this field, we have uh really different modalities.
00:16:45.759 --> 00:16:52.799
And sometimes base editing is more suitable for certain targets, sometimes prime editing is better.
00:16:53.120 --> 00:17:32.319
And there are always uh claims made by different innovators that each modality has different characteristics and so forth, and it is very important from the safety perspective on off-target effects, on on-target efficiency, and off-tissue on-target edit with respect to in vivo editing, that investigators understand that although FDA is very flexible, that the claim that are made regarding safety has to be supported with robust package of data when different modalities are used for genome editing application.
00:17:32.720 --> 00:17:37.680
So I love the topic of safety, and I have two follow-up questions with respect to safety.
00:17:38.000 --> 00:17:40.799
The first is immunogenicity.
00:17:41.039 --> 00:17:44.640
How should we think about incorporating that into the development?
00:17:44.880 --> 00:17:53.200
And then the second question I wanted to ask you was about follow-up recommendations for gene therapy programs in terms of uh long-term monitoring.
00:17:53.599 --> 00:18:03.039
Yes, immunogenicity is uh is an important aspect that uh is hard to handle using animal models per se.
00:18:03.440 --> 00:18:13.119
But I just wanted to remind everybody that uh these endonucleases are bacterial endonucleases in the context of genome editing, for example.
00:18:13.359 --> 00:18:28.480
And if we talk about broader sense uh use of the viruses or AV or lentiviruses, these are uh essentially uh foreign entities uh to any given human being as a recipient.
00:18:28.799 --> 00:18:38.640
So there is always going to be innate immunity to viruses and foreign agents, and then there's gonna be adaptive immunity as well.
00:18:38.960 --> 00:18:48.640
So to some extent, the dosing regimen, whether it's done uh repeatedly or not, that's gonna impact your immunogenicity studies.
00:18:49.119 --> 00:19:07.200
But in a larger sense, most of what we know with respect to immunogenicity of the products is discovered during clinical studies and when adverse events are detected that could be attributed to the immunological responses.
00:19:07.440 --> 00:19:28.559
So I believe immunogenicity is something that should be addressed in the context of or accumulating knowledge of what these products induce as far as anti-drug immunity and in the context of whether they're used single-time or used repeatedly.
00:19:28.799 --> 00:19:41.200
And depending on the moiety that is the major constituent, there would be different types of immunity that needs to be handled and thought about in the development cycle.
00:19:41.519 --> 00:20:04.559
But in essence, most of the immunogenicity that we detect in patients, sometimes those issues result from residual impurities in the drug product due to the quality, and sometimes they are really driven specifically by the type of antigens that are introduced.
00:20:04.799 --> 00:20:13.119
So it is a complex issue driven specifically by the type of antigens that are introduced.
00:20:13.359 --> 00:20:18.240
So it is a complex issue and needs a lot of attention going forward.
00:20:18.559 --> 00:20:19.200
Thank you for that.
00:20:19.279 --> 00:20:34.079
And I and personally, with my experience, I found that when you have repeat dosing, it's always best to plan to be able to conduct ADA testing, even if you're banking those samples, but you want to be prepared with your assays as soon as possible.
00:20:34.319 --> 00:20:40.319
What are your views, Mo, on the adequacy of a 15-year follow-up period for gene therapy?
00:20:40.799 --> 00:20:43.599
Well, uh that requirement is evolving.
00:20:43.839 --> 00:20:46.160
Essentially, 15 years came about.
00:20:46.400 --> 00:20:51.359
It was an arbitrary number of years that FDA decided on.
00:20:51.519 --> 00:21:02.960
And it came about really the concern that was there with respect to long-term effects that you could have using these uh type of genetic medicine.
00:21:03.119 --> 00:21:28.000
And FDA has revised their follow-up guidance, and I encourage everybody to take a look, but really the main risk is when essentially you have possibility of random gene integration that results in oncogenic activation and cancer formation, or off-target effects that have safety concerns.
00:21:28.240 --> 00:21:39.599
And we as a community, selling gene therapy industry, we are accumulating more knowledge about different modalities and looking at long-term effects.
00:21:39.839 --> 00:21:57.759
But with products, for example, if there are plenty virus products that result in preferred but random integration, that concern of oncogenic activation is potentially there and needs to be monitored from my perspective for long term.
00:21:57.920 --> 00:21:59.599
I'm not sure if 15 years is.
00:22:00.319 --> 00:22:04.480
Really, the mandatory time that needs to be monitoring.
00:22:04.640 --> 00:22:12.720
But I think we need to really exercise long-term monitoring to some extent for some of these modalities more than others.
00:22:12.960 --> 00:22:20.960
If you, for example, have a product that is RNA-based and the likelihood of integration is much lower.
00:22:21.119 --> 00:22:24.400
Obviously, the risk is going to be much lower.
00:22:24.559 --> 00:22:28.880
So, based on that, the follow-up would be impacted.
00:22:29.039 --> 00:22:45.119
Again, as FDA and reviewers gain more knowledge about the safety profile of these products, for example, AAV, the requirement for follow-up may change to some extent based on the existing safety profile.
00:22:45.359 --> 00:23:00.559
But I think long-term follow-up is designed to actually make these products sustainable in order to identify any sort of obstacles or safety issues that needs to be addressed going forward.
00:23:00.720 --> 00:23:07.440
And is the way of identifying those safety issues so we can develop better drugs in this field.
00:23:07.759 --> 00:23:09.119
No, thanks for your perspective.
00:23:09.359 --> 00:23:10.160
Very important.
00:23:10.559 --> 00:23:20.799
So moving on to biomarkers and stratification, you know, genetic medicines are inherently precision interventions as you described previously.
00:23:21.039 --> 00:23:31.839
But yet many of the indications we work on in neuroscience, especially in rare and orphan indications, enroll genetically and phenotypically heterogeneous populations.
00:23:32.160 --> 00:23:40.000
I'm just interested in your perspective on how we should be integrating genomic stratification into eligibility criteria and response assessment.
00:23:40.400 --> 00:23:48.480
For any uh monogenic diseases, we have 7,000, uh, depending on who you talk to, maybe 10,000 monogenic diseases.
00:23:48.799 --> 00:23:56.400
A subset of these monogenic diseases really include multiple mutations that vary from patient to patient.
00:23:56.640 --> 00:24:09.119
And because of that, we need to really have more precision in our upfront diagnostic of the patient and to detect the target sequence.
00:24:09.359 --> 00:24:22.480
And as a result, we are using whole genome sequencing, NGS and others to identify target sequences in the patient that is going to be the driver of developing drugs.
00:24:22.720 --> 00:24:49.599
I want in the context of genome editing, I want to sort of a little bit elaborate and talk a little bit about conditions where you have monogenic conditions and it's a single mutation and is very homogeneous in the population, and that would be very different than, as you alluded to, to a condition where the mutation is uh varies from patient to patient and there is a high degree of heterogeneity.
00:24:50.079 --> 00:25:02.559
So when you have a high degree of heterogeneity, there needs to be precision in upfront diagnostic in order to validate the target accurately in that patient.
00:25:02.720 --> 00:25:05.119
And that becomes an important factor.
00:25:05.440 --> 00:25:20.480
So upfront diagnostic, genetic base, using the sequencing technique, etc., is really highly encouraged and required for some of these therapies to become a reality.
00:25:20.720 --> 00:25:28.960
And that precision of sequencing is needs to be another consideration that companies need to think about.
00:25:29.200 --> 00:25:43.119
There was a actually, I want to just put a plug-in, another guidance document that was just published, draft guidance, using NGS sequencing for assessing safety of genetic medicine products.
00:25:43.279 --> 00:25:52.799
And they talk about specific requirements for using NGS to detect target sequences, to assess off-target effects, et cetera.
00:25:52.880 --> 00:25:56.160
And I found that document to be very helpful and detailed.
00:25:56.319 --> 00:26:10.400
And it also deals with not only looking at off-target effects, but it talks about detecting variation in off-target effects in different patients that have different HLA or characteristics, etc.
00:26:11.039 --> 00:26:16.960
And obviously talks about variation in the elite frequency, et cetera.
00:26:17.200 --> 00:26:40.400
So they there is definite uh prescriptive advice that is given in this draft guidance, best thinking of the reviewers within FDA that is really helpful to read, that deals with the use of NGS not only for upfront patient stratification, but also to assess off-target effects, et cetera.
00:26:40.960 --> 00:26:44.079
No discussion would be complete without mentioning AI.
00:26:44.640 --> 00:27:00.000
And from someone like you who's both a scientist and someone who's been a regulator, how are you thinking about AI in terms of being able to predict off-target effects or its applications in other ways to help us to validate?
00:27:00.160 --> 00:27:00.400
Yeah.
00:27:01.119 --> 00:27:01.359
Yeah.
00:27:01.759 --> 00:27:02.480
Great question.
00:27:02.640 --> 00:27:15.200
You know, we talk about AI a lot recently, and I've been in a few conferences where AI was brought up in the context of genetic medicine, other drug discovery, et cetera.
00:27:15.440 --> 00:27:34.720
FDA has uh, you know, I want to um take this uh advantage of this uh podcast to convey the message that what FDA says at the level of reviewers that AI is a tool, can be summarized as a very sophisticated software that requires a very sophisticated hardware.
00:27:34.880 --> 00:27:42.559
And that tool can be used in many different ways to help us with some of the challenges that we have.
00:27:42.880 --> 00:27:45.759
AI has an application in sequencing.
00:27:46.000 --> 00:27:59.359
AI has an application and utility in identifying off-target effects theoretically using bioinformatics based on a design sequence of guide RNA, et cetera.
00:27:59.680 --> 00:28:10.880
AI has applicability in really helping us facilitate manufacturing of high-quality products, et cetera, in the context of manufacturing.
00:28:11.039 --> 00:28:30.400
So there are many applications for AIs, but ultimately AI will be, to some extent, topic of discussion at FDA, and AI is designed to accelerate, but it is important to emphasize that AI is not a decision maker.
00:28:30.799 --> 00:28:39.440
And a decision maker always has to be an individual and have affiliation with quality control, etc.
00:28:39.920 --> 00:29:15.359
Um, so I want to sort of uh put a plug-in for AI as a tool to accelerate a lot of challenging things that we do, but at the same time say that AI has some limitations and in the constraint of compliance and CGMPs, AI does not uh represent an entity that can make a decision, but can support decision made by others that are highly qualified and are responsible for I and D execution.
00:29:15.920 --> 00:29:17.359
I definitely agree with you there.
00:29:17.519 --> 00:29:27.519
AI as a tool, I don't think will supplant uh the human mind, but definitely has helped us in in many ways to increase the efficiency of drug development.
00:29:27.759 --> 00:29:33.440
We could talk for a long time, but I have one more question for you, and I'm so grateful for your your time today.
00:29:33.599 --> 00:29:37.599
Just thinking about your your career and industry.
00:29:37.759 --> 00:29:43.839
What's one important thing that you've changed your mind about um with respect to genetic medicines in the past decade?
00:29:44.160 --> 00:29:52.240
Initially, I was skeptical about the application of genome editing for treatment of diseases in vivo.
00:29:52.640 --> 00:30:23.440
Um ex vivo, there has been ample evidence of being able to mutate a gene of interest or target uh in vitro, but in vivo application is quite complicated that involves not only having a product, but a product that can target the right cell type and making efficiently a change, uh single nucleotide change without changing dramatically other sequences that are being targeted.
00:30:23.680 --> 00:31:03.759
So initially I was skeptical about the utility of genome editing in vivo, going beyond ex vivo approaches, but recent data that is available is clearly suggests that in vivo targeting of genome editing can be done, and there would be additional room for improvement to target these genome editors to the right cell type, uh right tissue inside the body, and to ensure that they're safe and are not biodistributed to off-tissue sites and have off-target effects that are undesirable.
00:31:04.000 --> 00:31:18.720
But the surprise that I had is based on the pace of discovery that is really fueling development in this area, transitioning from ex-vivo approaches to in vivo approaches quite rapidly.
00:31:18.880 --> 00:31:31.519
And to see some results that are highly promising, we will see our first probably in vivo genome editing product approved next year from Intellio.
00:31:31.759 --> 00:31:34.160
Hopefully, if things go well.
00:31:34.400 --> 00:31:43.359
And that pace from my perspective has been very surprisingly quick and has been very efficient in the industry.
00:31:43.680 --> 00:31:44.720
Well said, Mel.
00:31:44.799 --> 00:31:51.039
Thank you so much for your perspectives and for joining us on today's Mind the Methods podcast.
00:31:51.359 --> 00:31:52.400
Thank you so much, Kami.
00:31:52.720 --> 00:32:03.039
I appreciate the time to talk to you and spend a few minutes about genome editing, which is a huge area of interest for the whole industry going forward.
00:32:04.480 --> 00:32:10.640
Mind the Methods brought to you by the International Society for CNS Trials and Methodology.
00:00:01.120 --> 00:00:14.640
Welcome to the Mind the Methods Podcast, brought to you by the International Society for CNS Clinical Trials and Methodology, your gateway to the latest insights, innovations, and conversations at the forefront of central nervous system clinical research.
00:00:14.880 --> 00:00:22.640
This series dives into the challenges, progress, and people shaping the future of CNS therapeutics development and trial methodology.
00:00:22.879 --> 00:00:31.839
In this first season, we'll explore innovations in CNS trial design, from digital endpoints and adaptive protocols to patient-centric models and generative AI.
00:00:32.079 --> 00:00:41.200
Here we bring together researchers, clinicians, regulators, and industry leaders to discuss the challenges and breakthroughs shaping the future of CNS therapeutics development.
00:00:41.359 --> 00:00:50.159
From advancing trial methodology to improving outcomes for patients, our conversations dive deep into the science and strategy behind CNS research.
00:00:50.399 --> 00:00:51.439
Thanks for joining us.
00:00:51.520 --> 00:00:52.399
Let's get started.
00:00:52.719 --> 00:00:59.359
Mind the Methods, brought to you by the International Society for CNS Trials and Methodology.
00:01:00.320 --> 00:01:06.719
Hi, my name is Camilleolo Gemel, and I am a scientific committee member at the ISCTM.
00:01:07.120 --> 00:01:11.040
I'm very pleased today to be speaking to Mo Haydren.
00:01:11.280 --> 00:01:11.599
Dr.
00:01:11.680 --> 00:01:19.519
Haydran brings a rare trifactor of being a scientist, a regulator turn strategist, and an industry operator.
00:01:20.000 --> 00:01:33.599
Mo spent several years at the FDA's division of cellular and gene therapies, 15 years in biotech planning research, product, and IP development, and nine years early in his career at the NIH.
00:01:33.920 --> 00:01:43.359
He's an author of more than 50 original research articles and a holder of 25 issued patents with 54 additional pending patents.
00:01:43.599 --> 00:01:57.680
He's a founder of a nonprofit organization dedicated to promoting education and translational research to better understand root causes of the diseases of the brain, including physical conditions that affect our mental health.
00:01:58.000 --> 00:02:09.520
Mo is currently serving as chief regulatory scientist at Cell X Incorporated Consulting, making him a particularly compelling voice for our audience here at ISCTM.
00:02:09.680 --> 00:02:14.719
And today's topics precision and safety in genetic medicines for CNS.
00:02:14.960 --> 00:02:23.280
This session is born out of a continuation of our February meeting where we tackled regulatory flexibility to accelerate CNS drug development.
00:02:23.439 --> 00:02:39.919
And today we'll cover basic topics such as taxonomy of gene editing platforms, CNS delivery systems, preclinical safety, trial design methodology, and lastly, where AI is beginning to shape how we think about editing all together.
00:02:40.240 --> 00:02:42.240
Mo, welcome to Mind the Methods.
00:02:42.319 --> 00:02:43.439
We're so glad you're here.
00:02:43.840 --> 00:02:45.120
Thank you so much, Camille.
00:02:45.680 --> 00:02:52.479
I'm glad to be here and I'm hoping that I can contribute to some of the topics that you outlined.
00:02:52.879 --> 00:03:03.759
So just to begin, Mo, you've sat on three distinct sides of our industry: the bench at NIH review table at FDA Seber and now industry strategy.
00:03:04.000 --> 00:03:15.280
And when you look at where genetic medicines, especially for the CNS, are today, does the progress feel faster or slower than you would have predicted from your SIBR days?
00:03:15.759 --> 00:03:16.639
Great question.
00:03:16.879 --> 00:03:25.439
I think before I get into your question, I do I want to come up with some definition and framework for our discussion.
00:03:25.599 --> 00:03:37.840
Obviously, genetic medicine is a broad term, and it includes a number of different modalities for replacing, correcting genes or silencing genes.
00:03:38.319 --> 00:03:42.879
Replacing correcting genes or silencing genes.
00:03:43.120 --> 00:03:57.120
And this topic of interest was fueled in 2010 with the discovery of some of these genome editing modalities and products that have really fueled the technology.
00:03:57.360 --> 00:04:15.520
So the underlying technology, for example, for genome editing is uh is very recent as compared to the use of AV or antiviruses to deliver genes for gene replacement strategy, which has been around for a few decades.
00:04:16.079 --> 00:04:20.639
So it really depends on the context of our discussion.
00:04:20.879 --> 00:04:33.360
But uh in reality, we have done an incredible job translating technologies that were used for research a few decades ago, and now they're used to treat patients.
00:04:33.600 --> 00:04:37.759
So from my perspective, that's that is an incredible progress.
00:04:38.160 --> 00:04:47.360
So when you think of the term genetic medicine, are you thinking of the broader framework that you just described, or in terms of DNA editing or gene editing?
00:04:47.839 --> 00:05:14.079
Genetic medicine is a broader term, but I think the conversation that we can have today with focus on genome editing, I consider that as a subset of genetic medicine, a subset, and it's fueled by recent progress that has been made, and everybody's aware of developing personalized genetic medicine for patients.
00:05:14.240 --> 00:05:28.639
And that is based on recent progress with Baby KJ's uh story of treating a single patient using gene editing modalities that can be done within six months.
00:05:28.800 --> 00:05:33.920
And that has created a lot of enthusiasm about personalized medicine.
00:05:34.160 --> 00:05:51.199
Personalized medicine has also a broader definition, but in the context of genome editing, really talks about treating and a single patient based on their genetic mutations as detected by companion diagnostics.
00:05:51.600 --> 00:06:12.319
So, in reality, I can have a broader conversation, but I realize that we want to focus due to the time constraint on the modalities that are used to change um genetic abnormalities that are found in rare condition, rare monogenic diseases as we go forward.
00:06:12.720 --> 00:06:13.839
And that makes a lot of sense.
00:06:13.920 --> 00:06:16.079
And yeah, thank you for bringing up baby KJ.
00:06:16.319 --> 00:06:21.360
I think we're all still pretty in awe about um the progress that he's made so far.
00:06:21.600 --> 00:06:34.079
So, Mo, just thinking about uh taxonomy and CRISPR-Cas9 versus base editing versus prime editing from a CNS development standpoint, how do you stratify these tools?
00:06:34.240 --> 00:06:38.399
And is the selection driven by the biology of the target?
00:06:38.720 --> 00:06:46.480
You know, say loss of function versus gain of function, or are delivery and safety constraints the dominant filter here?
00:06:46.879 --> 00:06:47.120
Yes.
00:06:47.279 --> 00:06:49.759
Um couple of things before I get into it.
00:06:49.839 --> 00:06:58.560
Uh, for genome editing, obviously, that's an area that has been going on maybe for more than a decade in the sense of clinical studies.
00:06:58.720 --> 00:07:02.800
And we're really dealing with different uh modalities.
00:07:03.040 --> 00:07:36.160
Initially, we were talking about the use of endonucleases which were larger, and then in 2012, about CAS series of endonucleases in the context of CRISPR's uh were developed, and essentially CRISPR really dominated the old uh legacy tradition of genome editing product, which are still under investigation by some companies, but CRISPR has become the dominant player.
00:07:36.399 --> 00:07:43.759
After CRISPR development, we see uh uh an emerging technology of developing better CAS.
00:07:44.240 --> 00:08:01.839
CAS-9 was the first one, and then there was a CAS-12A that was a variant of CAS molecule, and then we have base editors um focusing on adenine and cytosine base editors, and then prime editing.
00:08:02.000 --> 00:08:18.480
Each one of these modalities have really different profile, different characteristics, different safety characteristics in view of the presence of on-target versus off-target effects and other modalities that has to be factored in.
00:08:18.720 --> 00:08:34.639
So when we go from one platform to another, I believe we are dealing with uh multiple types of products, and we have to be cognizant that each product has a very unique characteristic that needs to be factored in.
00:08:34.960 --> 00:08:41.360
And just thinking further about the broad category of genetic medicines versus small molecules.
00:08:41.759 --> 00:08:57.360
For discussion, which I think is very important uh to sort of focus on these personalized medicines, which are in neurology's a definite target, and that is monogenic conditions that have mutation within a single gene.
00:08:57.600 --> 00:09:17.200
In the context of personalized medicine, I think one thing I want to elaborate on in the first question, make sure that we understand, and depending on what type of correction that needs to be made, there's going to be a sort of optimization of what platform would be most suitable.
00:09:17.360 --> 00:09:21.279
And this is something I missed in our previous discussion.
00:09:21.519 --> 00:09:38.559
And I want to kind of emphasize that although we have different modalities, we also have to optimize editors based on the target sequence, and that target sequence really impacts uh what's the most optimal way of developing drugs.
00:09:38.799 --> 00:09:49.279
To answer your question about developing drug product in genetic medicine, it's really driven by mutations that are found in the patient.
00:09:49.519 --> 00:10:00.159
For example, case of baby KJ, which had deficiency in CSP1, we know there are thousands of mutations within a single gene.
00:10:00.320 --> 00:10:05.279
And each patient has a very unique mutation that needs to be addressed.
00:10:05.519 --> 00:10:11.840
Not all mutations can be addressed using the same modality, and that has to be factored in.
00:10:12.000 --> 00:10:34.879
But because the mechanism of action is very well defined, this accelerates the drug development, because we start with having knowledge of what is the target sequence and ability to design appropriate product to correct that mutation in vitro, which is called ex vivo, as well as in vivo.
00:10:35.279 --> 00:10:45.919
From my perspective, because the mechanism of action is well defined, the target is identified and the modalities are available.
00:10:46.080 --> 00:11:01.279
The process of developing a drug should be accelerated tremendously, as evidenced by products that we have in market that have been developed within five or six years that are essentially close to approval or approved.
00:11:01.519 --> 00:11:09.440
So I think well-defined mechanism of action is very unique with respect to the broader category of genetic medicine.
00:11:09.840 --> 00:11:18.639
Yeah, and I think underpinning all that, um, a well-characterized biology and understanding that is the key to having a well-defined mechanism of action.
00:11:18.960 --> 00:11:21.120
So I appreciate your perspectives there, Mo.
00:11:21.600 --> 00:11:39.200
So thinking now about your role, both when you were at FDA and now as a consultant advising companies, what when planning I and D-enabling uh preclinical studies, are there any studies that you see sponsors underweighting chronically?
00:11:39.360 --> 00:11:42.159
For example, biodistribution, genotox.
00:11:42.240 --> 00:11:44.799
You you mentioned off-target effects previously.
00:11:45.120 --> 00:11:48.720
You know, what are you seeing in your conversations right now with sponsors?
00:11:49.200 --> 00:11:54.799
The conversation is really uh centers around the complexity of these products.
00:11:55.120 --> 00:12:03.120
These products are multi-components, meaning that they are not really essentially consist of one component.
00:12:03.440 --> 00:12:20.399
For example, a genome editing product, as we know, consists of the endonuclease, which is guided through guide RNA to the target sequence and delivered using, as let's say, a virus or non-viral mechanism using lipid nanoparticle.
00:12:20.720 --> 00:12:32.320
Because of the complexity of the products, each component poses an additional burden of quality requirement that is going to impact the final drug product.
00:12:32.480 --> 00:12:42.720
So, in a sense, having these multi-component products that are very complex really impacts discussion that we will have going forward.
00:12:42.879 --> 00:12:47.759
And it's really specific to the product of choice and going forward.
00:12:48.240 --> 00:12:54.799
Here you're saying is that product development in the space is increasingly complex due to multiple components.
00:12:55.360 --> 00:13:04.159
From safety perspective, is uh sometimes is the quality of material and manufacturing and manufacturing consistency.
00:13:04.559 --> 00:13:16.320
Because we, you know, 10 years ago, some of these reagents were research reagents, and now we have the requirement from a regulatory perspective to use high-quality materials.
00:13:16.559 --> 00:13:26.559
Sometimes manufacturing of these quality materials are not necessarily something that investigators are focusing on initially.
00:13:26.720 --> 00:13:33.519
For example, in Baby KJ, essentially in six months, a product was developed and the cost was very low.
00:13:33.759 --> 00:13:55.360
But when you have subsequent discussion with FDA, FDA has been flexible in many ways, but they're still really focusing on quality and the requirement for release tests for different components of product that needs to be factored in, that really create pressure on the costs, etc.
00:13:55.759 --> 00:14:23.600
The other aspect is about identifying on-target versus off-target efficiency of the genome editing product and really looking at what you describe as genotoxicity studies early on that needs to be factored in, as well as uh looking at possibility of germline transmission, et cetera, which is not a major focus during earlier stages.
00:14:23.919 --> 00:14:44.960
But uh in reality, after one identified the gene of interest and the target sequence, from that point on you build that manufacturing platform to produce these very specialized, highly complicated genome editing products that need to be delivered appropriately.
00:14:45.200 --> 00:15:11.039
So ultimately, there are major criteria or obstacles that one has to face that relates to quality of material, cost of material, reproducibility of obtaining those materials, manufacturing the genome editing products, and being able to evaluate those in vitro for safety as well as in preclinical studies.
00:15:11.200 --> 00:15:25.919
And the preclinical studies has been really an area that things have evolved in the last few years with respect to FDA's interest to really reduce the number of animal studies.
00:15:26.240 --> 00:15:47.200
And I think there has been some flexibility from FDA and reviewers to accept cell-based assay in lieu of some of the preclinical studies to assess off-target effects using cell lines, etc., that harbor different mutations that exist in the patient.
00:15:47.440 --> 00:15:51.600
So there are a number of things that essentially needs to be focused on.
00:15:51.759 --> 00:16:15.600
But if there is one topic that I can stress here more as we go forward, is emphasis on being able to manufacture these very complicated products using CDMOs or other entities that you depend on and be having control over that process to be able to have reproducible products.
00:16:15.759 --> 00:16:30.000
And the secondary to that is really focusing on genotoxicity up front and accessing on-target and off-target in view of any given product by having a robust package of in vitro and vivo data.
00:16:30.399 --> 00:16:39.200
Definitely don't cut corners on quality and manufacturing and genotoxicity assessments in the preclinical package.
00:16:39.759 --> 00:16:40.320
Exactly.
00:16:40.480 --> 00:16:45.679
And there are in this field, we have uh really different modalities.
00:16:45.759 --> 00:16:52.799
And sometimes base editing is more suitable for certain targets, sometimes prime editing is better.
00:16:53.120 --> 00:17:32.319
And there are always uh claims made by different innovators that each modality has different characteristics and so forth, and it is very important from the safety perspective on off-target effects, on on-target efficiency, and off-tissue on-target edit with respect to in vivo editing, that investigators understand that although FDA is very flexible, that the claim that are made regarding safety has to be supported with robust package of data when different modalities are used for genome editing application.
00:17:32.720 --> 00:17:37.680
So I love the topic of safety, and I have two follow-up questions with respect to safety.
00:17:38.000 --> 00:17:40.799
The first is immunogenicity.
00:17:41.039 --> 00:17:44.640
How should we think about incorporating that into the development?
00:17:44.880 --> 00:17:53.200
And then the second question I wanted to ask you was about follow-up recommendations for gene therapy programs in terms of uh long-term monitoring.
00:17:53.599 --> 00:18:03.039
Yes, immunogenicity is uh is an important aspect that uh is hard to handle using animal models per se.
00:18:03.440 --> 00:18:13.119
But I just wanted to remind everybody that uh these endonucleases are bacterial endonucleases in the context of genome editing, for example.
00:18:13.359 --> 00:18:28.480
And if we talk about broader sense uh use of the viruses or AV or lentiviruses, these are uh essentially uh foreign entities uh to any given human being as a recipient.
00:18:28.799 --> 00:18:38.640
So there is always going to be innate immunity to viruses and foreign agents, and then there's gonna be adaptive immunity as well.
00:18:38.960 --> 00:18:48.640
So to some extent, the dosing regimen, whether it's done uh repeatedly or not, that's gonna impact your immunogenicity studies.
00:18:49.119 --> 00:19:07.200
But in a larger sense, most of what we know with respect to immunogenicity of the products is discovered during clinical studies and when adverse events are detected that could be attributed to the immunological responses.
00:19:07.440 --> 00:19:28.559
So I believe immunogenicity is something that should be addressed in the context of or accumulating knowledge of what these products induce as far as anti-drug immunity and in the context of whether they're used single-time or used repeatedly.
00:19:28.799 --> 00:19:41.200
And depending on the moiety that is the major constituent, there would be different types of immunity that needs to be handled and thought about in the development cycle.
00:19:41.519 --> 00:20:04.559
But in essence, most of the immunogenicity that we detect in patients, sometimes those issues result from residual impurities in the drug product due to the quality, and sometimes they are really driven specifically by the type of antigens that are introduced.
00:20:04.799 --> 00:20:13.119
So it is a complex issue driven specifically by the type of antigens that are introduced.
00:20:13.359 --> 00:20:18.240
So it is a complex issue and needs a lot of attention going forward.
00:20:18.559 --> 00:20:19.200
Thank you for that.
00:20:19.279 --> 00:20:34.079
And I and personally, with my experience, I found that when you have repeat dosing, it's always best to plan to be able to conduct ADA testing, even if you're banking those samples, but you want to be prepared with your assays as soon as possible.
00:20:34.319 --> 00:20:40.319
What are your views, Mo, on the adequacy of a 15-year follow-up period for gene therapy?
00:20:40.799 --> 00:20:43.599
Well, uh that requirement is evolving.
00:20:43.839 --> 00:20:46.160
Essentially, 15 years came about.
00:20:46.400 --> 00:20:51.359
It was an arbitrary number of years that FDA decided on.
00:20:51.519 --> 00:21:02.960
And it came about really the concern that was there with respect to long-term effects that you could have using these uh type of genetic medicine.
00:21:03.119 --> 00:21:28.000
And FDA has revised their follow-up guidance, and I encourage everybody to take a look, but really the main risk is when essentially you have possibility of random gene integration that results in oncogenic activation and cancer formation, or off-target effects that have safety concerns.
00:21:28.240 --> 00:21:39.599
And we as a community, selling gene therapy industry, we are accumulating more knowledge about different modalities and looking at long-term effects.
00:21:39.839 --> 00:21:57.759
But with products, for example, if there are plenty virus products that result in preferred but random integration, that concern of oncogenic activation is potentially there and needs to be monitored from my perspective for long term.
00:21:57.920 --> 00:21:59.599
I'm not sure if 15 years is.
00:22:00.319 --> 00:22:04.480
Really, the mandatory time that needs to be monitoring.
00:22:04.640 --> 00:22:12.720
But I think we need to really exercise long-term monitoring to some extent for some of these modalities more than others.
00:22:12.960 --> 00:22:20.960
If you, for example, have a product that is RNA-based and the likelihood of integration is much lower.
00:22:21.119 --> 00:22:24.400
Obviously, the risk is going to be much lower.
00:22:24.559 --> 00:22:28.880
So, based on that, the follow-up would be impacted.
00:22:29.039 --> 00:22:45.119
Again, as FDA and reviewers gain more knowledge about the safety profile of these products, for example, AAV, the requirement for follow-up may change to some extent based on the existing safety profile.
00:22:45.359 --> 00:23:00.559
But I think long-term follow-up is designed to actually make these products sustainable in order to identify any sort of obstacles or safety issues that needs to be addressed going forward.
00:23:00.720 --> 00:23:07.440
And is the way of identifying those safety issues so we can develop better drugs in this field.
00:23:07.759 --> 00:23:09.119
No, thanks for your perspective.
00:23:09.359 --> 00:23:10.160
Very important.
00:23:10.559 --> 00:23:20.799
So moving on to biomarkers and stratification, you know, genetic medicines are inherently precision interventions as you described previously.
00:23:21.039 --> 00:23:31.839
But yet many of the indications we work on in neuroscience, especially in rare and orphan indications, enroll genetically and phenotypically heterogeneous populations.
00:23:32.160 --> 00:23:40.000
I'm just interested in your perspective on how we should be integrating genomic stratification into eligibility criteria and response assessment.
00:23:40.400 --> 00:23:48.480
For any uh monogenic diseases, we have 7,000, uh, depending on who you talk to, maybe 10,000 monogenic diseases.
00:23:48.799 --> 00:23:56.400
A subset of these monogenic diseases really include multiple mutations that vary from patient to patient.
00:23:56.640 --> 00:24:09.119
And because of that, we need to really have more precision in our upfront diagnostic of the patient and to detect the target sequence.
00:24:09.359 --> 00:24:22.480
And as a result, we are using whole genome sequencing, NGS and others to identify target sequences in the patient that is going to be the driver of developing drugs.
00:24:22.720 --> 00:24:49.599
I want in the context of genome editing, I want to sort of a little bit elaborate and talk a little bit about conditions where you have monogenic conditions and it's a single mutation and is very homogeneous in the population, and that would be very different than, as you alluded to, to a condition where the mutation is uh varies from patient to patient and there is a high degree of heterogeneity.
00:24:50.079 --> 00:25:02.559
So when you have a high degree of heterogeneity, there needs to be precision in upfront diagnostic in order to validate the target accurately in that patient.
00:25:02.720 --> 00:25:05.119
And that becomes an important factor.
00:25:05.440 --> 00:25:20.480
So upfront diagnostic, genetic base, using the sequencing technique, etc., is really highly encouraged and required for some of these therapies to become a reality.
00:25:20.720 --> 00:25:28.960
And that precision of sequencing is needs to be another consideration that companies need to think about.
00:25:29.200 --> 00:25:43.119
There was a actually, I want to just put a plug-in, another guidance document that was just published, draft guidance, using NGS sequencing for assessing safety of genetic medicine products.
00:25:43.279 --> 00:25:52.799
And they talk about specific requirements for using NGS to detect target sequences, to assess off-target effects, et cetera.
00:25:52.880 --> 00:25:56.160
And I found that document to be very helpful and detailed.
00:25:56.319 --> 00:26:10.400
And it also deals with not only looking at off-target effects, but it talks about detecting variation in off-target effects in different patients that have different HLA or characteristics, etc.
00:26:11.039 --> 00:26:16.960
And obviously talks about variation in the elite frequency, et cetera.
00:26:17.200 --> 00:26:40.400
So they there is definite uh prescriptive advice that is given in this draft guidance, best thinking of the reviewers within FDA that is really helpful to read, that deals with the use of NGS not only for upfront patient stratification, but also to assess off-target effects, et cetera.
00:26:40.960 --> 00:26:44.079
No discussion would be complete without mentioning AI.
00:26:44.640 --> 00:27:00.000
And from someone like you who's both a scientist and someone who's been a regulator, how are you thinking about AI in terms of being able to predict off-target effects or its applications in other ways to help us to validate?
00:27:00.160 --> 00:27:00.400
Yeah.
00:27:01.119 --> 00:27:01.359
Yeah.
00:27:01.759 --> 00:27:02.480
Great question.
00:27:02.640 --> 00:27:15.200
You know, we talk about AI a lot recently, and I've been in a few conferences where AI was brought up in the context of genetic medicine, other drug discovery, et cetera.
00:27:15.440 --> 00:27:34.720
FDA has uh, you know, I want to um take this uh advantage of this uh podcast to convey the message that what FDA says at the level of reviewers that AI is a tool, can be summarized as a very sophisticated software that requires a very sophisticated hardware.
00:27:34.880 --> 00:27:42.559
And that tool can be used in many different ways to help us with some of the challenges that we have.
00:27:42.880 --> 00:27:45.759
AI has an application in sequencing.
00:27:46.000 --> 00:27:59.359
AI has an application and utility in identifying off-target effects theoretically using bioinformatics based on a design sequence of guide RNA, et cetera.
00:27:59.680 --> 00:28:10.880
AI has applicability in really helping us facilitate manufacturing of high-quality products, et cetera, in the context of manufacturing.
00:28:11.039 --> 00:28:30.400
So there are many applications for AIs, but ultimately AI will be, to some extent, topic of discussion at FDA, and AI is designed to accelerate, but it is important to emphasize that AI is not a decision maker.
00:28:30.799 --> 00:28:39.440
And a decision maker always has to be an individual and have affiliation with quality control, etc.
00:28:39.920 --> 00:29:15.359
Um, so I want to sort of uh put a plug-in for AI as a tool to accelerate a lot of challenging things that we do, but at the same time say that AI has some limitations and in the constraint of compliance and CGMPs, AI does not uh represent an entity that can make a decision, but can support decision made by others that are highly qualified and are responsible for I and D execution.
00:29:15.920 --> 00:29:17.359
I definitely agree with you there.
00:29:17.519 --> 00:29:27.519
AI as a tool, I don't think will supplant uh the human mind, but definitely has helped us in in many ways to increase the efficiency of drug development.
00:29:27.759 --> 00:29:33.440
We could talk for a long time, but I have one more question for you, and I'm so grateful for your your time today.
00:29:33.599 --> 00:29:37.599
Just thinking about your your career and industry.
00:29:37.759 --> 00:29:43.839
What's one important thing that you've changed your mind about um with respect to genetic medicines in the past decade?
00:29:44.160 --> 00:29:52.240
Initially, I was skeptical about the application of genome editing for treatment of diseases in vivo.
00:29:52.640 --> 00:30:23.440
Um ex vivo, there has been ample evidence of being able to mutate a gene of interest or target uh in vitro, but in vivo application is quite complicated that involves not only having a product, but a product that can target the right cell type and making efficiently a change, uh single nucleotide change without changing dramatically other sequences that are being targeted.
00:30:23.680 --> 00:31:03.759
So initially I was skeptical about the utility of genome editing in vivo, going beyond ex vivo approaches, but recent data that is available is clearly suggests that in vivo targeting of genome editing can be done, and there would be additional room for improvement to target these genome editors to the right cell type, uh right tissue inside the body, and to ensure that they're safe and are not biodistributed to off-tissue sites and have off-target effects that are undesirable.
00:31:04.000 --> 00:31:18.720
But the surprise that I had is based on the pace of discovery that is really fueling development in this area, transitioning from ex-vivo approaches to in vivo approaches quite rapidly.
00:31:18.880 --> 00:31:31.519
And to see some results that are highly promising, we will see our first probably in vivo genome editing product approved next year from Intellio.
00:31:31.759 --> 00:31:34.160
Hopefully, if things go well.
00:31:34.400 --> 00:31:43.359
And that pace from my perspective has been very surprisingly quick and has been very efficient in the industry.
00:31:43.680 --> 00:31:44.720
Well said, Mel.
00:31:44.799 --> 00:31:51.039
Thank you so much for your perspectives and for joining us on today's Mind the Methods podcast.
00:31:51.359 --> 00:31:52.400
Thank you so much, Kami.
00:31:52.720 --> 00:32:03.039
I appreciate the time to talk to you and spend a few minutes about genome editing, which is a huge area of interest for the whole industry going forward.
00:32:04.480 --> 00:32:10.640
Mind the Methods brought to you by the International Society for CNS Trials and Methodology.
