Advancing Oncology Trials: Innovative Designs and Breakthroughs in Cancer Research

In this engaging talk, the speaker Prof. Richard Sullivan from King’s College London shares insights on the evolution of early-phase clinical trials in oncology. Highlighting the shift from traditional designs to modern, complex approaches like basket and umbrella trials, they emphasize the importance of pharmacodynamic endpoints, innovative drug development, and the critical role of statistical design.

Richard Sullivan is a Professor of Cancer and Global Health at King’s College London, where he directs the King’s Institute of Cancer Policy and co-directs the Conflict and Health Research Group. His research spans from global cancer studies to conflict and health, with a focus on capacity building in conflict zones, humanitarian medicine, women’s health, and digital innovation in surgery.

Ladies and gentlemen, first of all, wonderful to be here. Thank you so much, and particularly to Gavel for hosting myself and Ajay. It’s really wonderful.

We’re obviously from Guy’s and St. Thomas’ Hospital in London, and also King’s College London, which you must all come and visit at some point. I won’t say we know His Majesty the King very well, but we’ll try and get an audience with him. And can I say, we literally arrived this morning, so if I start falling asleep a little bit, just throw things at me and wake me up.

But it’s really wonderful to be here. We’re going to spend a few days with you, and we’re really looking forward to it. And I guess I’m doing a little bit of scene setting, and Gavel’s asked me to say a little bit about early phase clinical trials.

And I’m going to say, I’m not going to bore you to death with agonizing detail about how to run these. I thought what I’d do is talk a little bit about my personal history working on early phase clinical trials in the UK, and also talk something about the contemporaneous nature of the way we run early phase trials now. And it’s more to illuminate the challenges, and also to illuminate perhaps the sort of individuals you need to build in an institution to run contemporaneous early phase clinical trials.

So I hope this is a nice story. I think it should be approachable to all of you in the room, but it also will have a little bit of technical detail in. So a little bit of history.

Once upon a time, I was the clinical director of an organization called Cancer Research UK, which is a very large funder of research in the UK. And very unusual, and I’m going back to the 1990s here, and prior to that, we ran our own drug development organization where we developed an awful lot of very, very interesting molecules, some of which some of you will probably use today, and I’ll go through those in a minute. This is a fantastic paper.

If you want to read about the history of public drug development from a combinatorial chemistry point of view, this is one of the best papers. And really, the history of early phase clinical trials starts with small molecules. Here’s a question, a little bit of a quiz.

Does anybody know what drug this is? Anyone? Who wants to have a second?

No. No, good. No, that’s good.

I like guesses. Nope. No.

It’s unusual. I’m going to give you a clue. It’s for a rare cancer.

No. No, I can’t believe this. No, no one.

I’ll tell you what, cash prize for anybody who gets this. They’re all going to be looking at it on the phone now. This is really interesting.

Second, which one? No, but you’re kind of in the right direction. This is temozolomide.

Yeah, so temozolomide was really one of the first big success stories for combinatorial chemistry. And the reason I put this up is not because it was one of our drugs from Cancer Research UK, and we outsourced it in the end to a company called Shearing Plough, which no longer exists now. I’ve forgotten who Shearing Plough got bought up by.

You might be able to remember. Anyway, it disappeared. But they developed it.

Why was this drug so important? Because this was the beginning of the modern era for early phase trials in cancer. Prior to this, we were really doing trials of a 3 plus 3 design, bit of pharmacokinetics.

We didn’t really think about what we call pharmacodynamic endpoints. And this was the first drug for which a lot of work was done on pharmacodynamic endpoints. And the reason we were able to do it is because of technology.

At the time, a lady called Pat Price in the UK worked out how to put carbon-11 into this molecule. And that allowed us to actually image without having to stick needles into people where the drug was being distributed. And so we were able to prove that the drug was super concentrating in gliomas.

So this was using MRS. So the point about this is, first of all, there was clever chemistry from combinatorial libraries. And secondly, it was the beginning of the pharmacodynamic endpoint era. And at that point in the 1990s, it became unacceptable, really, to do early phase trials without having really good PK, pharmacokinetics, and really good pharmacodynamics.

Now, what is interesting is it wasn’t just temozolomide. And again, anyone have a guess what these two molecules are? Come on.

Ah, very good. Which is cis-bacteria. Yeah, OK, very good.

Cis is on the load. Yeah, very good. And what’s the other one?

No, but I’ll go. Yeah, that’s near. Carboplatin, yeah.

Bizarrely, this was actually invented in the UK. This was Eve Wiltshire at the Institute for Cancer Research in 1972. And it was cisplatin that was discovered first.

But originally, it was horrendous as a drug in early phase trials. And if you look at the history, what you’ll discover is, of course, vast numbers of patients got hideous nausea and vomiting, and of course, renal toxicity. And this is the era before we had 5-HT inhibitors, and before we’d sussed out that you needed to diurese and hydrate people really well.

So again, what it taught us an awful lot about is managing toxicities in early phase clinical trials. Now, of course, out of that, a lot of other spin-outs. We did a lot more work.

And a gentleman called Hilary Calvert and others developed this other drug, which was called carboplatin. Much, much better renal toxicity profile. But the problem was it had terrible hematological toxicity.

Hence, the formula on the bottom. Does anyone know what that formula is? Calvert 4, yeah, bravo.

So this is Hilary Calvert. And Hilary Calvert, again, but what it says is, Hilary Calvert was interesting chap because he was a clinician and an early phase clinical trialist, but also had a really deep interest in chemistry and pharmacokinetics. And I guess what I’m kind of speaking to a little bit here is that you need to kind of think, when you’re thinking about early phase clinical, it’s just not the act of doing the trial.

It’s actually thinking around the whole biology. It’s thinking about the team. And I’ve put up a few kind of lessons that we learned, really.

This is really true today, even. There’s lots of elegant working hypothesis of how things are successful. But the way they act is often nothing to do with the working hypothesis.

So you can have a great working hypothesis for a new drug in early phase clinical trials. And actually, the way it really acts in reality has got nothing to do with it. And there’s no direct correlation between the elegance of the working hypothesis and the success in the clinic.

We had lots of fantastic drugs, flavonacetic acid, that cured animal cancers. So if you’ve ever got a dog or a mouse that gets cancer, a class of compounds called FAA will cure the cancers. Cancer will vanish.

It does nothing in human beings. Beautiful, beautiful idea. We had a lot of cytoreductase inhibitors, for example, that required very high levels of DT-diophorese inside cells.

Great inside animal models, but does absolutely nothing inside humans. And this is the other thing we learned, of course, is, and I guess in the immunodynamic endpoint, this is even more important now, is just because something works in cell lines doesn’t mean it’ll work in vivo. Just because it works in vivo doesn’t necessarily mean it’s going to work in humans.

And the mantra now is that it’s human beings as the model system. You can do all the work you like in the animal models, and, you know, you have to, but the reality is until they go first into human, you really don’t know what they’re going to do. And the last point I want to make here is everyone focuses enormously, and I’ll make a more point about this in a moment, to technology, technology, technology, technology, particularly to do laboratory studies associated with early phase clinical trials.

But it doesn’t substitute for original thinking. And that takes a long time to develop, all of you out there. And it doesn’t matter whether it’s the nursing, the PK experts in the lab, the clinicians.

So evolution of early phase clinical trials, I’m going to rock it through this, literally starting with a rocket. This is how we used to do it. It was very formulaic.

It went on phase one, phase two, phase three, et cetera. The world has changed a lot since then. And really, I’m not going to go into any detail with this.

And it’s more to make the point here that there’s a chap we hang out with and we go to a particular course in India called Krida, called Xavier. He’s French, obviously, with a name like Xavier. And he’s quite remarkable.

He’s a brilliant statistician, particularly in early phase clinical trials. Now, once upon a time, again, back in the 1980s and the dark old era, you know, individuals like me could do a standard three plus three design. We could do Fibonacci dose escalation.

It wasn’t too difficult because the rules were really set. The rules these days from a statistical perspective in terms of design are super complicated. And paediatrics have their own ideas.

Adult solids have their own ideas. There are different ideas with those who run immuno-oncology from small molecule inhibitors. Haemato-oncology is completely different.

And the point I want to make here is you need good statistics. And you need really excellent people who dedicate themselves to trial design for any particular combat, because it will make or break your early phase clinical trial. This is becoming an increasingly interesting and popular area, but it’s not without huge controversy.

What we’re seeing more and more now is early phase clinical trials with what we call synthetic arms, historical controls. And I’m not going to have time to go into the nuances of this now. There is a lot of argument, both for and against using synthetic arms.

And I’ve just sort of illuminated a few things from there. Right now at the moment, there is a big difference between how the FDA thinks of synthetic arm trials, early phase trials, and how the EMA, the EMA, European Medicine Association, is much more conservative. And they’re much more concerned.

A lot of synthetic arms don’t really reflect reality. But this is the way we are going, because the object is, we were just talking about accelerated approval here, the object is fast to market now for any form of molecule. And that has its trade-offs, and we’ll discuss that, I think, a bit later.

You’ll hear from Ajay and from myself a bit what this means with accelerated approval. This is, again, also a very popular way of doing early phase trials, is you have expansion cohorts now. Because you’re trying to make the very best of the work you’re doing at an early phase.

And again, there are trade-offs with what works and what doesn’t work. I guess you’ll have heard of things like basket trials here, and how we do them. This is an interesting one.

This is, of course, extremely popular at the moment, because this is about the determination of a target here, where the subtype, basically the subtrial, enrolls multiple cancer types with one common genetic mutation. But again, the disadvantages are, I’ve put at the bottom here, that, of course, the molecular variant may not be a driver of the tumor, and the contextual complexity of various histologies may exist. And so just because you’re targeting that specific mutation in these different cancer types, doesn’t necessarily mean the readout’s the same.

And so again, basket trials are very popular, but you have to read those with caution. Umbrella trials as well, you’ll hear that phrase, where you have a single histology there. And again, umbrella studies really do have difficulty enrolling these rare genetic mutations of a single tumor.

And also the introduction of new standard of cares during a trial, I think really does change the environment. So again, when you’re looking at umbrella trials, the design’s really important, but there are lots of trade-offs. And just to give you a practical example, and there are so many of these, I mean, these are being run all over the world now, it’s kind of almost new.

You get these strange hybrids, what we call umbrella basket trials. So you get one big umbrella, which allows lots of molecules to get in. And then with it, you get lots of different baskets where you’ve got different mutations with different drugs.

Each of these have their own unique set of rules. And as you can imagine, I want to go back to the beginning, the statistics is everything. And so these are designed with very, very nuanced statistics, particularly the new types of Bayesian studies.

And again, this is well outside my comfort zone and my knowledge. What you need is people who really understand the statistical way of doing it. But this is how things are going.

It’s to get more and more out of the system. But I kind of want to put out what the problems are here. The first is you have to have serious access to drugs here.

So the determined umbrella basket trial that’s going on in the UK at the moment, I mean, it’s got relationships with Roche and with other companies because it needs access to all these different sorts of inhibitors. And each of those inhibitors, of course, and targets have their own nuances. And again, it requires a really deep level of expertise in these different signalling pathways to put these two together.

And the point I want to make at the bottom with this whole genome sequence is not, it’s just the amount of technology you need to run these. It is staggering now when you look at these basket umbrella trials, umbrella basket trials, just how much tech you have to have to run all the PD or immunodynamic endpoints. Now, I’m not really saying much more about that in terms of trial design, because again, it’s getting into the weeds a little bit and you just want to stay out at that level.

That’s kind of where we are at the moment with early phase clinical trials. The idea is to get a lot more out of your early patients. The idea is to accelerate your drug development pathway.

And your idea is to be clever in terms of what you’re trying to target. But with increasing complexity in terms of trial design and statistics comes increasing numbers of bear traps. You can make major mistakes.

Either you can kill a drug, which is really good, or you can allow a drug through that really isn’t going to deliver anything from a clinically meaningful benefit. Quickly going through the immuno-oncology, I mean, wow, we couldn’t believe, you know, when immuno-oncology came along, I worked on something called Cetuximab, which was fascinating because it’s an EGFR inhibitor. And we were kind of killing ourselves over what PD endpoints were until we realized, you know, the butterfly rash you get, perfect.

It was the perfect PD endpoint. If they got a rash, guess what? They got a response.

So suddenly out of nowhere, all these times it was like, great, you know, you don’t have to, you don’t have to use rhesus criteria here, do you? You can just look and say it’s melted away. So it slightly made life easier.

But then, of course, things got super complicated. The more you look into it, and this is the truth. I mean, I came from an era of cell signaling.

The more molecules discovered, the more immuno-molecule stuff we work on, the harder the environment is. And again, it talks to this entering an era, not just of immuno-oncology, but of combinations in early phase trials. And in that sense, I want to just talk very quickly, because there’s only one example I want to get.

Because this comes from an era before we get into this deeper understanding of combination early phase clinical trials. This was published in 2016. The trial itself was done, I think, in 2013 or 14.

And it’s basically, you’ll all recognize pembrolizumab and this other drug, epicatestat, which was basically an IDO inhibitor, indolamine 2,3-deoxygenase. And basically, it’s an immunomodulatory enzyme. If it’s overexpressed, if IDO is overexpressed in prostate or pancreatic or cancers like that, you have very poor prognosis.

And what it does for cancers, basically, is it metabolizes something called tryptophan, removes that from the microenvironment, and also, as a byproduct, creates something called calinurin, calinurins, which inhibit T cells and natural killer cells. And that’s the reason why it allows the proliferation of the cancer. So it’s a survival mechanism, the overexpression.

So the view was you could put pembro together with this IDO inhibitor. And of course, you get two for the price of one. But just have a look at these individual response rates.

The overall response rate, of course, for melanoma, you know, this is the poster child of immuno-oncology, early phase clinical trials. But the red and the blue and the gray are quite interesting. That’s PD-L1 positive.

That’s PD-L1 negative is the red. And the gray is we just don’t know. So bizarrely, this was an era where we really weren’t doing our immunodynamic endpoints very well, because for the majority of those, we just don’t know.

But also what’s remarkable is you’ve got at least two patients that responded pretty well there who are supposed to be PD-L1 negative. And you’ll see again, non-small cell and renal cell cancer. And what it told us at the time was, we have a problem with thinking about what the endpoints should be and how we interpret waterfall plots.

And of course, this has just got harder and harder and harder, because more and more early phase trials now are putting together inhibitors from these different classes. We’ve just been talking about anti-CTL4. You know, we’re talking about people putting bevacuzumab together with these compounds.

So people are really starting to get complicated about the individual agents they’re putting together, which means the PD or immunodynamic readout is getting harder and harder to do. And it’s getting harder to interpret. And just again, it’s just more illumination.

If you just look at something like, this is urethelial cancer trials, and you look at all the detection antibodies, the immunohistochemical platforms, and also the cutoffs that all these different immuno-oncology drugs have. You can see where the complexity comes in, and you can see why you need individuals who are scientifically supremely literate. Are there pathologists here?

Any pathologists? We love pathologists. One pathologist.

Yeah, but they are super important. This is molecular pathology. This is a massive problem.

You have great clinicians, you have fantastic PK, amazing nurses. You don’t have decent pathology to understand where to do the cutoffs or to understand what antibodies you’re using in your IHC platform. And so, when we look back at the last five or 10 years and look at some of the trials that have really flunked, part of the problem is because they haven’t really thought through this element.

And I mean, this is really deep work in PDM and immunodynamic endpoints. So again, it’s a warning that be very nice to your pathology colleagues. Pay them heavily.

Yeah, so I’m not going to say any more now. So basically, the IOs seriously alter the biocomplexity. And I think I’m not going to talk about next generation and talk about later, but I think the next generation in biomarker work around this is going to be even more complicated.

And I think part of this is combinatorial work we’re doing. It’s harder and harder now to find niches and individual areas. And also the science becomes deeper and more difficult.

And I think areas we are very comfortable with reading across these areas. It’s just it’s just tough. So I think this is going to be an interesting area going forward.

And just to sort of finish to say that. I mean, this is the stuff that’s coming. This is unbelievable.

Cytotoxics in blue, targeted drugs in yellow and targeted biologicals in red. And that is what’s coming. That is these are drugs which have got authorization.

This is stunning. And by the way, if you’re moving to the future here, these are the drugs in pipeline. And those dark lines, that’s every therapeutic class by every major company and the dark line there, there’s dark boxes, there’s heat maps, are cancer.

So everybody is betting on cancer drugs. OK, so there are going to be stunning numbers coming down the pipeline in the future. I guess for all of you, the other thing I kind of make the point is early phase clinical trials are not without a kind of what I call a consequence and understanding what the social consequence of a drug is and the economic consequences.

The earlier you can think about that and build that into your studies, the better it’s going to be for society. And I’ll pick that up a bit later. And I guess the regulatory landscape as well as my last slide is important to understand.

You’ve got experts like Diego here who really understand and engage with the FDA. The EMA is a different beast. If you come through the MHRA in the UK, that’s a different beast.

And the regulatory nuances of what are acceptable in early phase clinical trials are very different. And you do have to think about that now, even with public sector organizations or mixed hybrid organizations. So look, there we are.

That’s what I have to say on early phase. Bit of a mixed picture. Good.

Thank you very much.

For more information visit oncodaily.com