Statistical Topics in Gene Therapy

Statistical Topics in Gene Therapy

Alexander: [00:00:00] Welcome to another episode of the effective set session and I can’t believe that it’s close to 400 episodes and I have never ever talked about gene therapy. Maybe that’s because I personally never really got involved in that space. Wasn’t working in at any of the biotech companies that are heavily involved in it and wasn’t cooperating on and of course Most of my career, I worked in the very late stages to kind of post approval.

And well, that also doesn’t help if you work on these kind of very, very new and innovative therapies. And for that, I have two awesome guests here on the show. And you will learn throughout this podcast, why these are great guests to have on the show for this topic. So Avery and Alex. So maybe you can start introducing yourself, [00:01:00] Avery, do you want to go first?

Avery: Sure. Thank you so much, Alexander. So I’m Avery McIntosh. I’m a statistician based in Collegeville, Pennsylvania. I work for Pfizer and I currently work in our internal medicine division where we have a number of compounds. In hematology and metabolic disorders, cardiovascular I met Alex actually working at Novartis.

When I, I first came out of graduate school, I was hired into Novartis. And while at Novartis we discovered the really interesting world of, of gene therapies and were able to work together on a number of projects. And some of those projects moved forward and some did not, but one of the ones that did move forward that I’m sure we’ll talk about is this book that we recently developed on on gene therapy drug development.

So happy to speak about that and happy to be with you here today. Cool. Alex, what about you?

Alex: Yeah. Thank you for a nice [00:02:00] intro. I’m Alex Sverdlov. I’m a statistician at Novartis. So I’ve been with Novartis for eight years now during the company in 2016. And I’m working in the field of neuroscience and supporting the early stage of development, which involves the preclinical studies, the first in human studies and proof of concept studies.

And as Avery said, yeah, I was lucky to meet with him back in 2017. And then we worked together on a couple of projects. And it wasn’t until 2021 when we Decided to develop a edited volume, which is the scope of today’s presentation. And yes, happy to talk more about it.

Alexander: Awesome. Very good. So before we [00:03:00] go a little bit more into the book let’s first talk about what actually is gene therapy.

Avery: Sure. I can start there. And Alex, if I miss anything, please feel free to chime in. So gene therapy, you mentioned earlier, Alexander, that it’s innovative and new, and I would say innovative. Yes, but new, actually not so much. So these Drugs have really been around and the idea behind them has been around for a long time, really stretching back into the 20th century when it was realized that some diseases are heritable and genetic in nature, but the pharmacologic science and the basic biology that it took to develop a drug and all of the elements that new drugs have to have to be safe and tolerable and to go through all of the regulatory exercises really didn’t take shape until the 1990s.

very much. But from that time forward there was a number of years where there was no approved therapies. And then in 2017 in 2018, there were a number of [00:04:00] regulatory approvals. The first was for Kymriah, which is a cell therapy for relapsed refractory leukemia, acute lymphocytic leukemia. And then zolgensma for spinal muscular atrophy and luxterna, which is for a form of inherited retinal blindness.

And since that time, they’ve now been I think as of today, maybe 33 FDA approvals and most of those are approved in EMA as well. So it’s really accelerating quickly and just a note on the mechanism of action. These drugs are the term gene therapy is really important. Thank you. A broad term to encompass any sort of drug that affects the expression of genetic information.

So all of our cells have DNA nucleic acids, and these are transcribed into RNA and then translated into proteins, which really powers all of the activities of our bodies and ourselves. So gene therapy drugs, correct, or address some form of genetic disease. Some people may be born with a [00:05:00] genetic disorder that prevents the appropriate expression of cell proteins and results in some sort of deficiency and downstream disease.

And some examples that your, your listeners may have heard of are Huntington’s disease, as I mentioned, spinal muscular atrophy, cystic fibrosis, and

Alexander: So the target of action is from gene to DNA to RNA and from RNA to the proteins. Is it also directly targeting the gene itself? So changing the chromosomes?

Avery: Well, that sort of depends on the mechanism of action. So there’s this broad label of gene therapies, but some of them transfer a therapeutic gene into the cell and that gene exists in an epizomal form so like a circular DNA form that’s then translated and transcribed every [00:06:00] time that the DNA is read.

in the cell, but there are other modalities, gene editing modalities. Many people have heard of CRISPR. That’s been in the news a lot. And and those generally do impact the chromosomal DNA structure itself theoretically in a permanent manner. Yeah. So it really depends on the type. And as I mentioned, there’s also those ex vivo cell therapies that some people call gene therapy.

Some people call cell therapies. They’re kind of, you know, A fuzzy delineation there where cells are transfused out of the body and then genetically engineered to express certain antigens that trigger the immune system to fight cancer and then are reinfused into the body. So those cells are engineered, but whether they’re lifelong or, or last for a long duration, the way that so called in vivo cell therapies might work is. I think not clear to

Alex: that also that so [00:07:00] just from the name, the gene therapy, so it affects the fundamental biology of a human body. And so we can think about these gene therapies as more like like organ transplants instead of pharmacological modalities. That’s where we. either insert the missing gene or deactivate the deficient gene or do some other sort of editing of genetic information.

Alexander: So if we actually change the gene itself then and that is permanent. So then it’s basically potentially a cure to the disease if we know exactly what we’re doing.

Avery: Yeah, that’s been a hot topic in the field is what is the durability of these products? And you know, as statisticians, this is something that falls on us [00:08:00] ultimately, to adjudicate the durability of these products and all gene therapies. In major regulatory environments have to the patients who were administered a gene therapy have to be followed for 5 to 15 years after their initial dosing.

And it’s really an open question for different modalities. Which ones of these are permanent and which one of And once of these are long lasting, you know, the initial targets for gene therapies were neurological indications because the neuron cells are generally non dividing. So then you could introduce an epizomal DNA structure and theoretically it wouldn’t be diluted when, when cells divide.

But since then there’s been a lot of work in this area and now there’s approved gene therapies for hemophilia that target the liver, right? Which is rapidly divided. So that it really does depend on the disease and the nature of the gene therapy, the vector and a lot of features as to whether these are curative [00:09:00] or not.

Alexander: Okay, okay. But it’s still a pretty interesting hope to have to cure some of these really disabilitating diseases. So as we talk about gene therapy, are, are these all kind of rare diseases, often indications?

Alex: So I can start and Eric can pick up on that. So yes, the gene therapies are usually developed for rare diseases.

And of course, there’s no universal definition of the rare disease in the United States. It’s defined as a disease or condition that affects fewer than 200, 000 people. And from this, one may imagine that the development of gene therapies would be more challenging compared to the more common diseases because the patient populations are limited and [00:10:00] then all sorts of restrictions apply.

But this is not the only therapeutic area. So not, not limited to rare diseases. So we may have more common diseases. That have to do with heart failure or for example, Alzheimer’s disease, which is not as rare. Pick up on that.

Alexander: Yeah. Okay. Okay. That’s super interesting. So no. What are the specific challenges beyond, let’s say, that it’s a rare disease, that make it harder for gene therapies to get approved?

Or why is it actually different to, let’s say, a typical, typical drug?

Avery: Avery? Sure, yeah. Well and I think it’s more difficult, certainly partly because of that rare disease component that Alex was speaking of there’s also very high manufacturing standards and purity and chemistry manufacturing and control standards. And, you [00:11:00] know, if you look into the news, there’s been many clinical holds.

And and, and pauses and even complete response letters from regulators wanting more information about the manufacturing and the purity and the potency of these products. So that’s been a major bottleneck in terms of scaling up because creating research grade product has been generally feasible, but then creating enough products meeting the same standards to Treat an entire population has, has been a real burden.

And there are a number of statisticians involved in the CMC space who, who work in this area, very brilliant statisticians. That’s not something that myself or Alex have done, although we’ve maybe moonlighted a little bit. The, the small sample sizes is a challenge. And I’ll, I’ll say another one that I’m Currently working on quite a bit is how you perform dose finding in, in these types of populations because these patients due to the immunogenetic nature of, of the therapies really can only be treated one time.[00:12:00]

So it’s a single administration, potentially lifelong impactful treatment. And if a patient were to receive a sub therapeutic dose or a dose that was too high. One of two things could happen. So if it’s too low, they may be precluded from receiving a gene therapy again in the future that that has a similar vector because their immune systems been triggered.

And then if that gene therapies administered again or at a higher dose or a similar gene therapy, their immune system could could target that gene therapy and recognize it or create an overactive immune response. And if the dose is too high, you know, we’re usually trying to get these proteins expression to a some sort of therapeutic window similar to non gene therapy drugs, then that gene therapy is not going to wash out and not going to be cleared by the liver and the kidneys the same way that a small molecule or a biologic can.

Might. And so those that [00:13:00] upregulation quote unquote overdose could be potentially permanent. And so we do have a lot of statistical tools for dose finding in non gene therapy settings. And these have really been developed mostly in the context of oncology. There’s been just brilliant work done over the past 10, 15 years in Bayesian models.

But applying that in the gene therapy space is a little bit tricky because most of these compounds usually only have two maximum three dose levels under study. And so, for models like Bayesian logistic regression, which is very commonly used in oncology trials, it’s a Bayesian model, and so you have a prior probability distribution, but if you only have two doses, And you have a small number of patients before you expand and, and choose your, your target dose to enroll more patients, then that type of model will really be much more dependent on the prior probability.

In a sense, the prior almost dictates your decisions. Yeah. Right. [00:14:00] So that’s, that’s some of the other challenges.

Alex: And I would also add to that, there are also some ethical challenges that stem from that we have patients who are very ill, right? So they have genetic disorders and potentially these are pediatric patients.

So pediatric patients, their central nervous system is still forming, and when introducing the gene therapy, will have a profound impact on their body and the entire system organ. So that means that we have to be very mindful about not just statistical aspects, how we do those findings. sample size justification, but also keeping in mind these restrictions related to the ethical considerations.

And in later phase clinical trials, for example, there may be ethical challenge, the use [00:15:00] of placebo group. So it would be really unethical to do some sham injection and deprive the patient from the opportunity to receive gene therapy, especially if it’s a very severe disease. So all sorts of things arise.

Alexander: Okay, in terms of that these later challenges in terms of randomized studies and things like that. So well, the

not to use placebo is not a new kind of argument in lots of, lots of areas, especially, you know, oncology or other life threatening diseases. People say, well, you can’t use placebo, however, then of course, the kind of it’s very hard on the safety profile to understand what’s really going on.

And of course. You don’t really know whether the new treatment will actually help. Yeah, so it’s [00:16:00] a little bit of a conundrum. Yeah, you need to run the study to know that it was unethical to use placebo. So at the time you make the decision to run a confirmatory study, you actually don’t know. So, so it’s kind of, you know, a little bit like you know the famous cat in the box where you don’t know whether it’s dead or alive.

Alex: That’s almost like a catch 22. Yes, because we, we are facing a problem of clinical research versus clinical practice, because that ideally we would certainly like to have participants who. are exposed to these new treatments, they would like to provide them with most benefit so that they derive the opportunity and not miss the treatment that’s efficacious in the trial.

However, when we run the trial, we don’t really know [00:17:00] which one is efficacious and which one may be unsafe. So it’s a leap of faith. to a large extent for these patients who volunteer to participate. And so there are ethical challenges also related to the informed consent. How Patients actually consent to take part in such experiments, especially when it comes to pediatric patients.

So there are obviously obligations for their parents to basically sign the informed consent. So that really speaks to the importance of all these medical ethics questions that have to be cleared. And also the health authorities would have to really authorize them and give the permission to run this experiment.

Alexander: Is there anything else that makes it different to let’s say it’s a typical pharmacological interventions that we study and get approved?

Avery: Well, [00:18:00] just back to the placebo topic again, beyond being, you know, potentially unethical, it may actually be infeasible because you used, Alexander, the word curative earlier.

And a lot of people do have that, that connotation around gene therapies. They, they think that it’s almost like a miracle cure, right? And of course investigational gene therapies that fail. Not just like every other kind of drug but there is some information that we might have with gene therapies that we might not have with other types of diseases or drugs that we would work in, and that’s a very well defined target.

So unlike any sort of complex disease heart failure, Alzheimer’s where genes may be implicated in, in a person’s risk, you know, it’s, it’s not like Mendelian where if you have the gene, the variants you, you will be effective. So there’s some very good animal models. That have been developed and used for many types of gene therapies that are studied in these [00:19:00] animals.

And then if there’s adequate rescue in the animal species, then they would have a lot of faith. That these would be advantageous and efficacious and usually safe for human populations. Now, there’s always an asterisk. Around that, because, you know, the translatability of these is quite tricky. But historically, it’s been the case that FDA and EMA’s not very let’s say a little bit more skeptical of the use of historical data or synthetic data for placebo arms, and that’s not done unless they’re, the disease under study is very severe.

Well, it so happens that for many gene therapies, the disease is extremely severe. And so there’s probably more use of historical data in the gene therapy space than in other spaces, but there’s certainly, as you mentioned, plenty of cases where regulators and sponsors will say, we need a placebo control trial.

So just finding that balance is a bit [00:20:00] tricky. And then exploring all of the statistical methods that we might have, it’s a very, a vast space, actually, of, of how you can run single arm trials or trials with historical controls or, you know with Bayesian augmented control designs and all sorts of new techniques for that.

So that in and of itself could certainly be a a topic for your podcast.

Alexander: Yeah, well these topics are ongoing on my podcast. Historical controls is definitely a hot topic for, for quite some time. Single noun trials or trials that have don’t have a one to one randomization, but maybe I don’t know, one to five or whatever randomization is, is definitely a really, really important one.

Now. If we think about gene therapies and one, is there the potential that you would have such huge effects that kind of you can actually get around a placebo just because of that? [00:21:00]