Fractals:

Click Here for podcast episode details and listening links.

 

[Colin Miller] 

Hello, I'm Colin Miller, CEO at the Bracken Group, and this is Fractals, Life Science Conversations. Bracken is the professional services firm for life sciences and digital health organizations. Our intelligence ecosystem fulfils consulting, regulatory, marketing and analytics needs with an integrated and strategic approach. 

Today, we're exploring the intersective nuclear medicine, clinical innovation and strategic leadership in a rapidly evolving world of radiopharmaceuticals. I'm delighted to be joined by my colleague, friend, Dr. Deepak Behera, an accomplished physician, scientist, strategist and industry leader whose career spans the medical, academic and pharmaceutical arenas. Deepak began his career in clinical medicine before transitioning to translational research at Stanford University, where his work led to the birth of two startups, five early phase trials and multiple patents. 

Over the past two decades, he has held senior leadership roles across the industry, including senior medical director, head of medical affairs, strategic medical director and chief medical officer. His expertise bridges the full spectrum of clinical development and commercialization, navigating complex phase one to phase three study designs, guiding FDA interactions and leading the launch of two commercial products. Today, as president and CEO of Adaptive Research and senior partner at the Bracken Group, Deepak brings a uniquely global perspective to his work, having collaborated with companies across the US, Europe and Asia. 

Deepak's also authored more than 20 peer reviewed publications, delivered over a dozen invited lectures and continues to help define best practices at the intersection of science, strategy and innovation in radiopharmaceutical development. Deepak, it's great to welcome you back to this program again. Thank you for joining me today. 

 

[Deepak Behera] 

Thank you for the warm introduction and invitation, Colin. It is a pleasure to join you again on this episode. Looking forward to the conversation. 

 

[Colin Miller] 

Deepak, it's fantastic again to have you on the Fractals podcast. I think it's been about two years since you and I first chatted on this. And radiopharmaceuticals has continued to grow, but the field of radiopharmaceuticals is still a relatively young commercial category. 

And what do you see as the key trust barriers among regulators, clinicians and investors? And how can companies address them? 

 

[Deepak Behera] 

Colin, when people often say trust barriers, the way that I think, it's really about clarity. And each group, the regulators, clinicians and investors, they're looking for a different kind of clarity. And if we intend to address them appropriately, the trust barriers are no longer barriers. 

We just we gain that trust, essentially. So for regulators, as an example, in my experience, regulators generally do trust good data. And so our job is to make claims that match the evidence that we generated, that we have, and expand those claims as the data matures. 

And so they are also open to many non-traditional ways and methods. We not necessarily check the boxes when we provide the right context. So why an imaging-led or dosimetry-rich data set answers their core questions on whatever they're looking for. 

Where teams get into trouble is when either over-engineering trials by borrowing complexity from elsewhere, from traditional drug development, that is not always needed. Or sometimes even going too simple with discrete data points that imply a conclusion, but the dots haven't been connected. So they don't necessarily connect the dots. 

I think the sequence should be, what do we want to, what is the message, the story for the stakeholder, for in this case, the regulators? And what kind of evidence can we generate to satisfy that story or build that story for the regulators there? And that will gain their confidence and their trust, in my opinion. 

And when we go to the others, the clinicians, the biggest hurdle for the clinicians is the fear of the unknown. Right now, as you mentioned, the field of radiopharmaceuticals is relatively young still. And while radiopharma has moved aggressively from diagnostic into therapeutic applications, it still sits alongside, certainly not against, but alongside established treatments. 

And clinicians want to know, where does this fit in my patient's journey? What are the side effects? What should I affect? 

What is the radiobiology? The side effects are driven by radiation biology, not the target mechanism of action, which is what they are used to traditionally. And so what do rapidly decaying drugs mean for operationally in their clinics or for their patients? 

What are the licensing, handling, SOPs, logistics, etc. that they need to have in place? So I think the way to address this would be like education, involvement. 

And we do, we have seen excellent adoption when treating physicians are included early, especially with the PSMA agents. But I think the same similar models and similar efforts should be put into every new area, every new product that comes into play with involving the clinicians in decision making, understanding the field and educating them in all of these things. Cross-collaboration between nuclear medicine and the disease area specific societies will help. 

I think engaging patient advocacy groups or patient education groups could be catalytic because informed patients ask about a new option or are better positioned and prepared for radiation related precautions and reduce the burden on clinicians to explain and educate patients in turn as well. Industry and investors, I think the issue here is context. Without radiation training or radiopharma training, the risk benefit of each data point is less well understood. 

It's hard to capture and you add real concerns around CMC supply chain scalability, etc. And market sizing, and that's another part that in a precision world, sizing the market becomes difficult as well. So it's driven by the precision diagnostic there and which subsegment of patients truly benefits. 

I think that's one of the key goals. The good news is with the diagnostics, the diagnostic twin can give go, no-go signals very early on. I can test hypothesis quickly, we can see the targeting and the pharmacology in humans and course correct before expensive commitments are taken. 

So with this kind of a playbook, I think investors and industry overall can have a greater level of trust with each step as it goes along. Certainly expert interpretation of the data and again, putting it in the context that the investors can understand could help gain their trust as well. 

 

[Colin Miller] 

Wow, Deepak, that was a fantastic answer, multiple answers to a lot of the other questions. And I'd like to pick up on several of those items that you touched in and drill into them a little bit more. The first one actually was on the diagnostic side. 

And I think we hear it a little bit on the consulting side and have a struggle with it. And I wonder what your take is where people have come from pharma, particularly in the preclinical or early translational stage go, hey, we can just use a Lutetium product and we just do imaging, we don't need a diagnostic. And what's your thought on that? 

 

[Deepak Behera] 

Very well said, Colin. I think that is a part which many traditional drug developers find it a little bit difficult to grasp. In fact, I think they grasp it conceptually, but it is difficult to translate into endpoints and actual actions if they're not used to it. 

And the dearth of specialists in this area does not help. The idea, however, is that imaging shows you pictures of what's going on in the body once a drug is delivered right away and can help you draw conclusions in terms of the proof of concept, the mechanism of action, the distribution, excretion, pharmacokinetics, the potential side effects based on organ residence times or the potential treatment effects based on the tumor to background ratios and blood ratios and all as well. So I think all this information can be captured as at least a no-go decision if you don't see favorable imaging characteristics. And that helps without having to wait for a very long time for blood biomarkers or resist measurements or other ways of looking at responses or efficacy and safety. 

You can have a very early signal of whether we ought to drop it at the list or continue investigating this further. And that's a very, I think, powerful advantage that imaging has in theranostics, especially because the concept of theranostics is that imaging is a very close counterpart of the therapeutic and therefore can be extrapolated much better in that way. 

 

[Colin Miller] 

Yeah, no, I couldn't agree more with you, but that was a fascinating insight. So just in general, as we head out of 2025 and look at 2026 and into the latter half of the decade, with putting a radiopharmaceutical hat on, what surprised you perhaps in the last year? And anything in the future that you see, wow, that I wouldn't have predicted, but or you can predict where it's going? 

 

[Deepak Behera] 

Because the field is still maturing, right? We will still see surprises, right? And I think we'll see the surprises. 

We'll see more and more surprises and creative solutions in a good way. And I think one of the nice, I shouldn't say surprise, but it was a good thing to see, unexpected outcome was what we just discussed, right? So we used an imaging-only study, the first segment of a study, to understand the biology before treating and essentially asking the question, let's see what the target is really doing in humans and where it is distributing and what's going on. 

The surprise was how decisive the pictures were. What the images showed was for uptake in one particular tumor type much higher than many others, whereas, and it was not entirely discordant, but it was unexpected from the distribution that you'd see in histopathology. But this was very helpful because now we were able to narrow down the indication to the high-yield tumor type, saving significant resources by not chasing the lower probability settings. 

And again, the other part of this is the imaging was so strong that the company that was pursuing primarily the therapeutic, with the diagnostics to support the therapeutic, added a parallel diagnostic-only path, right? And they were keen on now understanding or started treating the diagnostic as an independent product in its own right. And that's the beauty of theranostics. 

Early, the visual goes to the signals that you can get. And this was an example of a pleasant surprise. Unpleasant in the sense that certain tumor types were excluded now, but pleasant in the sense that you didn't pursue them to failure, right? 

There are different aspects of it. 

 

[Colin Miller] 

Yeah, it was a very quick go-no-go decision on certain tumor types to your point of earlier on, of how a diagnostic can really help you. And I think you've just characterized the example there. And the diagnostic shouldn't be underthought of, if you will, or undervalued. 

It's in this whole space. And just because you come from a therapeutic space and you don't know diagnostics, don't ignore them. They are there. 

They're a really key biomarker to help you. Yeah, and moving through from clinical trial to commercialization, you've helped launch two industry-leading products. What were the biggest lessons learned in translating this complexity of radiopharmaceutical science into real-world patient benefit? 

 

[Deepak Behera] 

The biggest lessons from commercial launches weren't necessarily about the science, but about how medicine actually works on, say, Monday morning. So getting approval... 

 

[Colin Miller] 

Changes on the weekend. Don't take your medicines on a Friday. They're not going to work. 

It's going to be Monday morning. Okay. Sorry. 

 

[Deepak Behera] 

I had to... Getting approval is necessary. Understanding the science is necessary. 

Of course, those are the building blocks to getting a product established in terms of safety and efficacy. But adoption happens when a lot of other factors, evidence, the operations, education, economics, everything all together, come together, line up in a proper way. The trials are optimized for regulatory approval, but they don't necessarily... 

One of those learnings is that they don't necessarily answer certain practical questions that clinicians have. And these launches have fortunately educated me to plan a post-approval evidence layer and bring that into a phase three trial, for instance, or even earlier on if necessary. Some pragmatic studies, registries, sub-analysis, and all those things, which speak directly to tumor boards. 

So we can build answer-dense materials, treating, holding rules, supportive care guides and all, which allow the context of practicing clinician and under-treatment patient stay in picture in addition to the evidence that we generate for regulatory approval. The payers also see a real-world impact differently. They're not just looking at response rates. 

They're not just looking at diagnostic efficacy. They're asking who exactly benefits for how long and at what total cost of care, bringing some of those insights, even if it is incomplete, but at least teasing, beginning to tease those questions early on with early health economic modeling or line of therapy specific outcomes, or maybe considering resource use data. So for example, how many imaging visits, how many hospital days, how many infusion chair time, how much infusion chair time comes into picture. 

So it gives a clearer story on sub-segments of the market. And again, because theranostics, like I said earlier, theranostics are precise by design. They do not necessarily apply to the entire patient population. 

So because of that precision, it is important for stakeholders, including payers, to understand what portion of patients are benefiting out of that particular therapy. And the other part to this, I believe, is operational. We, in trials, a sponsor de-risks any operational logistic. 

It is on the sponsor to deliver the drug. It is on the sponsor to figure out whether the site can get the product in time and those kind of things. But in practice, the clinics must handle the rapidly decaying drugs. 

They must handle the radiation licensing, handling SOPs and dose scheduling and coordination across nuclear medicine, oncology, pharmacy, all those kind of things. The site readiness to adopt a new product is something that is useful in consideration when we design trials themselves so that the eventual label that comes on the trial is conducive to an operational simplification as much as possible, right, in addition to everything else. I think that is also helpful to bring in early on in drug development. 

 

[Colin Miller] 

Because you've seen the back end to the front end, that key aspect. That's one of the reasons whenever I know we're working together on a project, you always come in from the orthogonal direction because you've seen it all the way across the spectrum. And I find it fascinating that, as you've just described multiple times in different sort of contexts, how you do that and think about it and how that changes. 

We've talked about the sort of amazing opportunities in the future and the way this whole process is going. So as Radiopharma grows, as is being talked about a lot in the industry, there's demand for skilled professionals, which is currently outpacing supply. So what advice do you give for early career stage physicians, scientists who want to enter this field and make an impact? 

What do you do to encourage them to come here? 

 

[Deepak Behera] 

The encouragement is this is a growing field, a very exciting field and very intellectually stimulating field. So you can probably find all sorts of roles and positions and interests satisfied. So please come on, join Radiopharma. 

That's the encouraging part of it. If I'm probably not, let's say, old enough to advise, but if I were to do this, I'd say build your skill set as a, let's say, T. So definitely be deep in one area. 

But also broaden your area into so many other skills, because this is a very multidisciplinary therapeutic area or field, if you will. It is so multidisciplinary. You have to think about physics, biology, chemistry, logistics and supply, your interactions between various treatment groups, clinical groups, of course, regulatory and science, and in addition to all of the others as well. 

Focusing and staying in one group or one skill set is pretty good, definitely. But I would advise people to take a look at other skill sets around, interact outside of your own zone and learn as much as you can, because all the strings connect to each other. And it is very helpful. 

At least I found it very helpful having the breadth of experience that I have been fortunate to have. 

 

[Colin Miller] 

It's not a linear path in radiopharma, is it? It's, I guess it's the inverse of an exponential decay. What's that curve called? 

Exponential rise? It's a geometric progression. It's not a geometric, it's an arithmetic progression. 

So I think it's the opposite. But yes, to your point there. But yeah, no, it's a tough one. 

But I guess the other part to this is that AI, while it will play a part, is still so far off being able to help us in this space. I haven't seen it making a major impact at this stage. Have you? 

 

[Deepak Behera] 

No. And I think the primary reason is exactly this. So if we don't have sufficient skill sets spanning across these multiple areas, including AI, the breadth that is necessary for AI to influence this entire field probably will have to be learned over time. 

And we're still in very early stages of people understanding AI. Those who understand radiopharma may not understand AI. Those who understand AI may not understand radiopharma sufficiently to have that marriage happen at this point. 

And again, I'm saying radiopharma imaging is probably the easiest way to go. There are data, and AI can look at data very quickly. But so many other aspects, which AI needs context to operate. 

And building that context is key for us as a field to provide to AI. 

 

[Colin Miller] 

And to your point, it's so nascent in the knowledge, I think, and the nuances are so subtle. And I'll just give you a funny case example that happened in the last 24, 48 hours. Someone within our team was pulling together a whole string of information, both statistical and operational, and trying to simulate it. 

So put it into an AI chat and ask AI to simulate all of this. And AI got it right about 75% of the time. But it got it wrong, maybe 20%, 25%. 

But the 25% was so intertwined, it made everything else not quite right. And it was really, you actually read it and realized that you had to back out of the whole thing and really almost start again. And it's been fascinating to watch and understand how AI works, because it's been given the world's data as it existed up to about two or three years ago. 

And the incremental lift that it's getting is relatively small. And in an area like radiopharmaceuticals, we're learning it all at the front end. So AI is always going to be behind us. 

And the knowledge that we have within the agency is residing with us. It's not in the available world yet. So I put all that together, and I'm beginning to realize that humans still are needed in some places. 

But I don't know, you were going to say something. I don't know if you have a thought. 

 

[Deepak Behera] 

Yeah, absolutely. I totally agree. I think humans are needed in the leading edge. 

We're going to lead and build the context necessary for AI to then take on. So humans will be needed on the leading edge, pretty much. Having said that, what I was going to say is, I remember a phrase that was used when computers and software became prevalent early on, garbage in, garbage out. 

AI currently is very much similar, except that no one understands what's garbage here. So AI has the garbage in, garbage out, with AI as well, which essentially means that if we provide proper prompts and proper directions, it's like proper codes and commands to the software, you'll get a proper output from the software. In this case, it's going to be proper prompts, which is similar to proper commands for a software. 

And those proper prompts are the part where humans become very important and the context becomes really important. Most of us ask AI a question, I still do, as if I was Googling something. I'll get the response based on, again, AI's observation of similar questions being asked by all over the place or something else that is all over the world. 

If the more specific I can make my question, the more specific is the response that I will receive. And that engineering, the prompt engineering, is the context engineering that we need to have. And right now, I think, at least in radiopharma, we don't have many people that understand enough all around to create that context for AI to respond appropriately, in my opinion. 

 

[Colin Miller] 

That's a fascinating insight. I hadn't put it into that context. And you're right, to your point about the prompts, but also understanding how to prompt the right question, putting the context into it, and also contextualizing the output, how it needs to be thought through as well. 

And AI will work those pieces out, but unless they're done, it takes you down a totally wrong path, I believe. Indeed. Yeah. 

So with that, I always like to ask a bit more of a personal question and thought process. And I know you've got two teenage children. And the question is, what advice are you giving them for careers with AI? 

And how are you talking to them about how they might use it in the long term? 

 

[Deepak Behera] 

Oh, boy. That is a very difficult question, actually. I'm still struggling with the concept of how to prepare our kids for the next what is coming, right? 

Are the traditional skills of calculations and differentials and knowledge and writing and all those skillsets still relevant in the near future? Are there other skillsets that become more important? Because now, AI can do all of the mundane tasks, right? 

So I think where, and this is still all speculation, right? Decision making, executive thinking, or relationship building, and those kind of things. Are they a team approach and all these soft skills, which AI will not give you, and which still are important in building something, creating something, taking things to the next level. 

Do they become more important than academic, let's say, competence, is the question that I ask myself. I can't stay away from it because right now, the system still values academic competence, or I guess academic competence would be redefined as AI takes on a larger and larger role. And where does that sit? 

We're probably on the cusp of something changing, and I just don't know what my kids will have to go through, and what shape or form their education or their work life will take in the future. There are certainly thoughts around those, but I have no idea. 

 

[Colin Miller] 

Thank you for that. It's a tough position as a parent. I can appreciate that, where you go. 

Yeah, now with 2020 hindsight, as you look back and we now see AI in there, is there anything different that you would have done or told yourself that you should have done to prep for AI? 

 

[Deepak Behera] 

Not if AI is coming where it is coming right now. I think I'm in a place, and many places, many people in my generation, and above, are in a place where I think we have the soft skills and the thought processes in place to take advantage of AI's powers. If we think it through, and if we learn how to do it, spend enough time, we still are drawn into our day-to-day activities and work and everything else, meaning that we are learning slower than AI is progressing. 

But if we were to take advantage, we could probably be the makers, and we are in that stage of our lives. But we are learning it as we go. There's no one to teach us. 

Will I ask, will I have, if I were to meet my younger self today, or if I time travel five years back or something, or 20 years back, and have to tell my younger self something, I think it wouldn't be very different from what I'd say prepare them from AI. It wouldn't be very different from real-life lessons. Think about what is the end goal, and structure a logical pathway for reaching that end goal. 

Verbalize those questions or steps as steps between now and the end goal. And those steps essentially become your AI prompts, if you're able to verbalize them and articulate them properly. So that thought process is prompt engineering in many ways. 

And I think that is probably the best advice I can give to myself to prepare myself for AI. Again, we are still in the very early stages. I have no idea what this is going to look like later on. 

Two years later, I might be saying something else. 

 

[Colin Miller] 

So there's the invitation, Deepak. This is the two-year stepping point. You've just set yourself up for two years' time. 

We're going to come back and go, was that the right answer? Was it the right answer? 

 

[Deepak Behera] 

Yeah, absolutely. The other part, I guess, is pay attention to the soft skills, pay attention to relationships, pay attention to teamwork, leadership, and those kinds of things, which AI will not help you with. 

 

[Colin Miller] 

I think that is the important part. Yeah, that's quite foundational and quite insightful. Deepak, thank you for such a fascinating and insightful journey on this podcast. 

We've touched so many different areas. And even as we moved into the last part of AI, the concepts, I think, can carry forward for a lot of us. I really appreciate your thoughtfulness there. 

So thank you for joining us on today's episode of Frank Talks. 

 

[Deepak Behera] 

Thank you, Colin. It was a pleasure. All the pleasure is completely mine. 

And I enjoyed the episode and our discussions. 

 

[Colin Miller] 

Fractals is brought to you by Bracken, the professional services firm for life science and digital health organizations. Subscribe to Fractals via your preferred podcast platform. Visit us at thebrackengroup.com or reach out directly on LinkedIn. We'll be delighted to speak with you. I'm Colin Miller, wishing you sound business and good health. Thanks for listening.