Combining Response and Toxicity Data to Implement Project Optimus

In this JCO Article Insights episode, Subodh Selukar interviews author Dr. Robert Maki on "Combining Response and Toxicity Data to Implement Project Optimus" by Maki, et al published in the Journal of Clinical Oncology September 11, 2024.

TRANSCRIPT

Subodh Selukar: Welcome to this episode of JCO Article Insights. This is Subodh Selukar, JCO's editorial fellow. Today, I am interviewing Dr. Robert Maki on his recent editorial, “Combining Response and Toxicity Data to Implement Project Optimus.”

At the time of this recording, our guest has disclosures that are available in the manuscript and will be linked in the transcript.

Dr. Maki, welcome to our podcast.

Dr. Robert Maki: Hi, Subodh. It's a pleasure to be able to take part.

Subodh Selukar: Yeah, thank you.

So, to start us off, would you give an overview of your article?

Dr. Robert Maki: Yes. Well, it's not my article, but it's just an editorial which is a commentary on an article by authors Cheng and Associates. It's called, “Exposure-Response-Based Multiattribute Clinical Utility Score Framework to Facilitate Optimal Dose Selection for Oncology Drugs.” That's a very technical title and so forth, and yet it's a JCO article because we think that it makes an important point that in oncological trials, we talk a lot about primary endpoints, oftentimes of overall survival or progression free survival, sometimes even just response rates, but most of the time, we don't take into account the toxicity of an agent. So, you can imagine that if a drug is relatively nontoxic, then what you see is what you get. Progression free survival could be associated with what is called some sort of so-called clinical benefit. However, if a drug is really toxic and you're just laid up on the couch all day or bed bound, or need transfusions three days a week, where is that really beneficial for the patient? But, by the same token, there's no quality of life without life itself. You have to have some sort of evidence that someone is going to be around for a longer period of time as an indication of benefit. So, these are ideas that have been played out to some degree for the better part of a quarter of a century.

There's a biostatistician at MD Anderson named Peter Thall, who's one of the first people to think about this idea of combining toxicity data and response data as some sort of a combination primary endpoint for a trial. And where this comes into play for Project Optimus, this FDA initiative to come up with not just necessarily one dose or one dose and schedule, but rather a range or multiple doses and schedules for a drug based on the toxicity that's seen, is that this new paper by Dr. Cheng and colleagues provides one mechanism for doing this, for combining not just traditional clinical outcomes data, but also toxicity data.

Subodh Selukar: So, you mentioned Project Optimus is an important component of all of this. So, can you tell a little bit about what Project Optimus is and maybe a little bit potentially about how Project Optimus has affected you so far?

Dr. Robert Maki: I'd say it's having an effect mostly in the earlier phases of drug development. I'm not certain, but I think it was an outgrowth of some of the toxicity that was seen in some of the studies that were done over the course of the last 10 to 15 years with kinase-targeted drugs. The overall goal from the FDA Project Optimus was to work with companies, with academia, groups like ASCO and regulatory authorities, as well as patients to try and come up with dosing for everyone basically based on patient characteristics that they're focusing not just on those outcomes, such as progression, pre survival, overall survival, but also looking for quality of life and adding that into the mix in terms of how you choose a dose. So that's an effort that's been going on for the last several years now. There's been some nice articles on that from FDA on that and perhaps we could provide some links to those as well for people who are interested in some of the more introductory core information about Project Optimus.

Subodh Selukar: Yeah, for sure. And so, I mean you're on the editorial board at JCO and you've written this editorial, but has Project Optimus affected your clinical research yet?

Dr. Robert Maki: It's just beginning to. So, in phase 1 and 2 clinical trials, especially in phase 1, the goal is not necessarily to look for activity, but just to come up with a recommended phase 2 dose and schedule of a drug. Well, Project Optimus says, “Okay. Well, maybe there's more than one dose and schedule that should arise.” And as I was alluding to earlier, this may have arisen out of what was seen previously where a number of the multi targeted tyrosine kinase inhibitors were developed. But when you got to the phase 3 trial, it was necessary to have dose reductions in 30%, 40%, 50%, 60%, even 70% of patients in some situations. So that to me represents a drug or a development pathway for that drug that was in essence incorrect. Yes, we talk about in traditional chemotherapy of trying to get the maximum dose we can, but is that always the best thing for the patient? And we recognize that there really is a plateau usually for systemic therapies we give, that there is a limit to dose escalation even within an individual patient to try and achieve that same benefit. At some point you're just going to add toxicity. The idea is to bring some element of toxicity into the decision making for a recommended phase 2 dose and schedule or schedules in that case.

Subodh Selukar: And so, building on that, so I think one advantage of these different approaches is that they might identify a single optimal dose, or maybe they'll recommend this range of doses that maximize some maybe clinical utility score combining these different aspects. In the current paradigm, it seems like probably response and toxicity are just these separate concepts that aren't typically linked together. But we typically do have a single recommended dose. But like you said, they might in subsequent trials have a lot of dose reductions and stuff like that.

So how do you think about the process now where this is a single recommended dose of, but there are deviations from that recommended dose in the research process. Like you said, in subsequent trials or within a trial, maybe patients are needing their own dose reductions as well. And then separately once a product is approved, what do you think about deviating from the recommended dose for your standard clinical practice?

Dr. Robert Maki: Oftentimes a work in progress. So even after phase 1, maybe having only treated 30 to 50 patients, they may be relatively homogeneous and that they have to be healthier to qualify for phase 1 trial. Once the drug is released to the whole wide world, then it becomes a different scenario, and you may have patients with poor performance status to start with. Can they still get the same benefit as the patients who got the medication in the context of a clinical trial? And it may not be the case.

And I think this is where Project Optimus and the idea of giving more than one dose or schedule may be useful and say, “Okay. Well, you can give 20% less,” and what's the trade off? Maybe the drug doesn't work as well, but it is less toxic. On average, do you really lose a whole lot as a matter of a few weeks of median progression free survival? Or does the response rate really drop off as you decrease the dose intensity of your drug? One concern about having more than one dose and schedule is could you potentially be underdosing patients by the same token? Since we usually have some amount of time, at least a few weeks, to work out what's tolerable for our patient, at least the parameters of having more than one dose and schedule to choose from can be useful.

Subodh Selukar: So then thinking about potentially maybe we would have a range of doses to recommend, what do you think are going to be challenges once that starts to be incorporated into clinical practice? What kind of complications do you think might happen explaining this to a patient?

Dr. Robert Maki: That's a really, really good question and something that we- I think, just have a difficult time with just the regular consent form. It used to be that maybe you had a couple of information sheets on a standard drug, or if it's a clinical trial, then you'll have a relatively modest consent form that's supposed to be at, whatever, 7th, 8th, 9th grade reading level. But now you start adding this form with complex text to a consent form for a clinical trial. What are people really signing up for? They get a 40-page document, and I don't think they really understand that.

So, the idea that you're trying to relate to them, pushing as hard as you can, but by the same token watching out for that toxicity, I think really does speak to those endpoints of the program, that it really can be a patient-friendly idea. Are we going to necessarily get it right every time? No. As I was mentioning previously there, if you're only treating 30 to 50 patients, you may only have partial information and you come up with some sense of dose and schedule to give. And then you move that into phase 2 and phase 3, and you may have to, you see that maybe one dose and schedule is a lot more effective as you get into a randomized portion of a phase 2 trial before you move to phase 3, for example, or you see that the toxicity is much greater with no better evidence of progression free survival. So those two scenarios could certainly rise. You can't predict them in the early phases of development of a drug, but you have to be able to react or be able to react with a solid clinical trial design that allows you to have that flexibility to make those decisions later. This is where discussion with the regulators, obviously is very important to make sure that what you're doing really still fits these guardrails, as it were, of traditional clinical trial design, or these ideas of adding in the toxicity-based information from Project Optimus.

Subodh Selukar: One of the challenges in early phase trials is, like you said, we might have 30 to 50 patients at the end of the study. I think in the editorial, you mentioned that some of these newer metrics might require more and more patients. Maybe we need 30 to 50 patients on a single dose in order to have reliable understanding of these clinical utility scores. Whereas right now a sample size at a single dose might be six patients, it might even be fewer. What are your thoughts on that aspect of it?

Dr. Robert Maki: That’s an important point, too. When you're doing, let's say, a quick and dirty, as you might say, 3+3 design, which has very large error bars in terms of the confidence intervals around a dose and schedule compared to some of the newer Bayesian-based designs, yes, you can get a phase 1 trial quote done, especially if it's a ‘me too’ sort of drug, so say, another checkpoint inhibitor, you kind of know the characteristics of those over another inhibitor of a specific kinase, you know the toxicities to expect when you block, let's say, EGF receptor. So, if you have some idea, and therefore you're able to more rapidly get to that recommended phase 2 dose from a phase 1 trial, if it ends up being a new drug, then maybe 30 to 50 patients isn't enough. And you really do need to continue that assessment of both response and toxicity as the trials move forward into phase 2 and phase 3.

So, it's kind of one of those ideas of continuous process improvement that if we are going to do this, we really do need to include it, not just in early phase trials, but especially for agents that are acting through a new mechanism of action, that we look at that holistically across the drug development spectrum. And now that trials are kind of being smashed together, phase 1 and 2, now phase 2 and 3, that really increases our need to also add in the assessment of toxicity, and maybe not just on the basis of our own evaluations or lab evaluations of toxicity, but patient reported outcomes, which is something that wasn't addressed in the Cheng article and really hasn't been well addressed in clinical trials in general, I would offer.

There are precious few trials that incorporate patient reported outcome data as a means to determine what's too toxic for a patient, for example. So how do we do that? As you know, we do have patient reported CTCAE clinical toxicity criteria that are based on patient reported outcomes. And wouldn't it be interesting, at the very least, as an academic project, but even more importantly, later on, to use those as the key means to determine whether a dose is too toxic or not in the development of the drug. That, to me, would be really, really interesting and kind of turns the idea of some of the data that we collect on its head. I guess, yes, we do need to collect things like liver function tests and so forth. It is one metric of toxicity of a drug. But patients have a lot of fatigue, we really do a poor job of documenting that as clinicians, and not to mention the elements that go into what that fatigue is. To be able to capture that through PROs would be another noble effort that I think has been underutilized and underappreciated in oncology clinical trials overall.

Subodh Selukar: And so, what do you think are barriers to doing it now?

Dr. Robert Maki: We tend to, for lack of a better term, cut and paste from what we've done before, to develop new, let's say, by patient reported outcome score or metric or worksheet for a given diagnosis. That can be hard, that takes a lot in and of itself, and perhaps has been one of the barriers that we don't have enough disease specific PROs, at least for some diagnoses. For others we do. And the fact that we do have PRO-scored CTCAE sorts of score tables, now, certainly makes it easier to validate and use these tools in clinical trials. So, I would love to see more of that, even if it ends up being secondary tertiary endpoints on phase 1, 2, and 3 trials. It's a pretty easy thing to add, even if you're doing that for the first time. Get some experience with it, and it can only help patients get through a trial or even just assessing it as part of a standard of care that will help our patients in the longer run.

Subodh Selukar: Yeah. And so, thinking about other metrics of success, you mentioned a couple in your article. These aren't necessarily patient reported outcome ones, but like RECIST and RANO. I was curious. I think the Cheng article, maybe I would think about it as a general framework for combining response and toxicity together, whereas some of these other metrics are a lot more disease specific, potentially, or agent and disease specific, maybe even. Do you think that clinical research will end up settling on these metrics that are kind of increasingly specific, or do you think that there's a possibility for general frameworks?

Dr. Robert Maki: Yeah, that's a tough question. I'm just trying to think of some of those patients reported outcomes. They've got kind of the general assessment ones, and then you do have ones that are more disease specific, just like we do have response criteria that are different for, let's say, lymphoma versus brain tumors versus colorectal cancer. We do have different ways of measuring those outcomes, and we all complain that those are imperfect measures. You can always find circumstances where that patient was responding, but it was called progression or vice versa. So even from these more objective tools like RECIST and the like, it’s a challenging field, that's for sure.

We keep going around and trying to find ways of improving those sorts of systems. But let's say, for example, you used - this is part of the reason we moved from two dimensional measurements in WHO criteria versus one dimensional RECIST - if you have two dimensions, well, you have that much more variability in the measurements of the lesion. So, it turned out that we just didn't gain anything by having those bidimensional measurements. Now, since we have the ability to measure tumors better in three dimensions, should we be using volumetric assessments? Part of it depends on the size of the tumor. If you're dealing with a tumor that's 1 cm versus 8 cm, well, then the volumetric changes, you have a lot more variability, the small ones, than the big ones. Not to mention the fact that you have shapes that are not just an ovoid mass in a lot of cancers. There's just so many pitfalls in these sorts of data. What really matters at the end of the day, one thing that's underappreciated, and again is underscored by Project Optimus, is getting back to the patient.

Subodh Selukar: Your editorial made me have this one thought, and so bear with me, it's like a multi-part question. One of the reasons that we're becoming more and more interested in these alternative approaches, these clinical utility scores and everything, is that these new agents are being proposed, where there's a hypothesis that there's more complicated relationships between dose, response and toxicity. And so, 50 years ago, researchers probably didn't hypothesize that these complicated relationships were happening. They probably thought that they were more straightforward. What do you think would have happened if we had had these conversations that we're having today if we'd had them 50 years ago, what do you think would be different? Do you think that maybe we would have different therapies that kind of ended up becoming standard today? Maybe would we interpret or run studies differently today?

Dr. Robert Maki: I like that question as well. Now, if we go back to the Charles Moertel studies back from the 1970s, the whole reason that we have tumor measurements as a criterion are really based on his work, where he got a series of clinicians together and he put these masses underneath a piece of rubber sheeting, and they tried to determine how well they could determine the difference between a mass that they could palpate. And this is when we came up with the idea that a partial response was a 50% decrease in the cross-sectional area of a mass. That came from that very crude but important work from about 50 years ago. And of course, that was also a time when there really wasn't any imaging. Maybe the best you would have would be x-ray tomography to look at a lung nodule or something like that. It was a little bit of a different era. We didn't know how our drugs worked very well. We had at least some biochemical reason to use chemotherapy, and we tried to leverage that. But it was always the idea of more is better, finally disproved later on, in let's say the era of breast cancer, looking at the AC combination or doxorubicin as part of a treatment for breast cancer, that there was a ceiling to the benefit of doxorubicin in the adjuvant setting. Even then, it was clear that we needed to think about dose and schedule. We also didn't have the variety of drugs that we have now, or the different metrics that we have, circulating tumor DNA or something along those lines. Those sorts of things just never existed then either. So, we need metrics that are appropriate for their time, and we have more tools to work with.

I suspect that we'll have more specialization in oncology along disease lines, or even molecularly characterized subsets of diagnoses as well. All the detailed classification that we now need for a lymphoma, for example, or different flavors of triple negative breast cancer, all of those things are impacting how we even put a person on a trial. Similarly, since these patients are also going to get different classes of drugs that are relatively unique to them, there are a lot of drugs now that are available that really are only approved for one diagnosis. Then you really have to drill down pretty deeply in order to be able to focus on that clinical scenario. But I think we have the means to do so. Nonetheless, the general idea of these frameworks, again, the idea of combining response and toxicity data that can apply across essentially any cancer or neoplasm that we want to study.

Subodh Selukar: Okay. So, I want to move a little bit to aspirational, like where we want to move forward now. And so I think you've talked a little bit about this so far already, but would you tell me a little bit about when you're seeing a patient, interpreting results that have been given in clinical trials, are there results, metrics, summaries of trials that you wish you could communicate to them, metrics that actually already exist but don't really get implemented? You already mentioned quality of life is something that doesn't seem to be there but are there other things that maybe quality of life might not just be collected enough yet. But are there metrics on data that we have and we just don't really report them at all?

Dr. Robert Maki: That may be the case, or maybe the data end up in a secondary and tertiary publication, so they don't really become part of the lingua franca of the oncologist. I think it really speaks to just having the experience as an oncologist that you try the FDA-approved dose for medication for somebody and you run into trouble if they're, let's say, in their 80s, whereas the study population was in their 40s and their 50s with better bone marrows or better renal function on average, and things like that. So, another untested waters are geriatric oncology. What are the maximum tolerated doses when they're 80 versus when they’re 40 or 50? It's a real challenge. Probably they had the most experience of that with things like prostate cancer, where we do treat largely an older population of men compared to other diagnoses, potentially.

I suspect we're going to see just more specialization, just like we do with the medications. We do need more specialized assessments for those adverse events and or quality of life that will be diagnosis specific. If you have GI cancer, abdominal pain is going to be a bigger issue or obstruction sorts of questions. And the symptoms that you may have from having a tumor within the abdomen versus, let's say, another diagnosis, which may tend to give you more, let's say, lung metastases. So those little subtleties can't come out. And the toxicities of the drugs that we use in those diagnoses are also going to differ as well. So those should be kept in mind as we come up with, let's say, disease specific toxicity metrics that we want to combine with those outcome data. So, I think we're going to see more and more specialization of that over time.

You have to create the tool and you've got to validate it. So, all these things will take some time. But again, people have been interested in this for a long, long time. There are any number of careers that are built around quality of life and cancer, or for example, long term survivorship in pediatric cancer patients. And all of these things can be very useful and just require our attention, both as clinical investigators as well as clinicians, when we face our patient’s day to day.

Subodh Selukar: And so just one last question before we close. Is there anything that we haven't had a chance to talk about that you like to share with our listeners?

Dr. Robert Maki: If it's anything it’s that I'm really heartened as I get older with this very large influx of new clinicians and new investigators. Oncology continues to get more interesting and more sophisticated. We need more people- we still don't have enough oncologists, even for our population here in the United States. We'll have plenty to do for a very, very long time. So, I'm excited to see a new generation of young oncologists such as yourself and the trainees that I see here, the new fellows, junior faculty who are all beginning to answer these questions, thinking about them. And as me and some of my more senior friends can help promote this kind of idea and help together to answer some of these questions. We're still trying to figure it out and there are just so many variables and clinical scenarios that we need to chase down in terms of clinical research. It is going to be an ongoing discussion and hopefully this article is just one example towards the goal again of finding the right dose for our given patient.

Subodh Selukar: Thank you so much for sharing and yeah, I'm very excited to be a part of this as well.

This has been Subodh Selukar interviewing Dr. Robert Maki on his recent editorial, “Combining Response and Toxicity Data to Implement Project Optimus.” Thank you for listening and stay tuned for the next episode of JCO Article Insights.

The purpose of this podcast is to educate and to inform. This is not a substitute for professional medical care and is not intended for use in the diagnosis or treatment of individual conditions.

Guests on this podcast express their own opinions, experience and conclusions. Guest statements on the podcast do not express the opinions of ASCO. The mention of any product, service, organization, activity or therapy should not be construed as an ASCO endorsement.

 Dr. Robert Maki Disclosures:

Consulting or Advisory Role: Deciphera, PEEL Therapeutics, Eisai, GlaxoSmithKline, Medtronic, Boehringer Ingelheim Speakers' Bureau: MJH Life Sciences  Research Funding: Amgen, Astex Pharmaceuticals, Boehringer Ingelheim, BioAtla, C4 Therapeutics, InhibRx, Regeneron, SARC: Sarcoma Alliance for Research though Collaboration, TRACON Pharma Patents, Royalties, Other Intellectual Property, Uptodate Travel, Accommodations, Expenses Company name: Stand up to Cancer, Fondazione Enrico Pallazzo