iwCAR-T 2026 | Improving TIL therapy for solid tumors: engineering, persistence, and implementation

Benjamin Creelan
We just finished a really exciting session showing the latest and greatest for tumor-infiltrating lymphocytes in solid tumors here at the IWCART meeting 2026 in Tampa, Florida. And I think our panelists today each gave important talks introducing what this is going to look like in two or three years.

So I’d first like to introduce Dr Rodabe Amaria from MD Anderson, who presented on OBX-115, Dr Lilit Karapetyan from Moffitt Cancer Center, and Dr. Daniel Olson from University of Chicago.

So really, wow , where do you guys think this takes us from today? I know that we’ve been addressing a lot of different problems we see in tumor-infiltrating lymphocytes, right? Thirty percent response rates, twenty percent response rates in some other tumors, how do we do better? I think each of you addressed some of the problems we face: gene editing, going back to the lab to figure out how to improve on this, and then Dan, going earlier, right? Going earlier.

What approach do you think is going to win in the short term?

Rodabe Amaria
Well, I think that’s debatable, and I’m not sure if “win” is the right word. I think it’s important to have multiple different approaches all going on in parallel.

There’s been a lot of interest in modifying TIL, and there are different ways to modify TIL. You can use viral vectors to gene edit. You can do CRISPR editing. There’s a variety of different ways you can approach that. So I think those are all very important.

A lot of times when you’re using these very experimental approaches, you’re going into the heaviest patient population with the greatest need, which is treatment-refractory disease. Only once you can prove that your therapy works, is effective, and is safe, then I think you can shift where in the treatment paradigm you actually offer that therapy.

Benjamin Creelan
And so this concept of, like you said, completely IL-2-free TIL infusions. I think that holds a lot of promise for us. Is there a possibility to do dual IL-12, IL-15, IL-18 armoring? Do you see the possibility of adding even more armoring to our TIL that we use in clinic? Obviously, you’re showing a lot of promise with OBX-115.

Rodabe Amaria
Yeah, I think we learn a lot now from what they’re doing on the CAR-T side. You’re seeing these incredibly complicated CARs with all different kinds of armors and targets and things that they’re doing.

I think right now we’re not quite there in solid tumors. I think we’re trying. I think we need to pick a problem and try to figure out how to best solve it without necessarily doing too much at once.

I think the OBX model does allow for an exogenous cytokine-free environment, which I think is safer and expands the pool of eligible patients. And that’s a good starting point, but there’s certainly a lot of work left to be done.

Benjamin Creelan
And I think that exquisite control you have, being able to add acetazolamide and stop it, which activates or deactivates the product…

Rodabe Amaria
Yeah, so you have an inducible and regulatable TIL, which again was learned kind of from CAR-T having a kill switch. This isn’t exactly the same thing, but you have the ability to turn the cells on or off, which allows you as the physician to really control things on an individual patient basis.

If patients are having issues, you can back off. If patients are doing fine, they can continue on with the regimen. But again, the whole point is to enhance patient safety, which makes more patients potentially eligible.

Benjamin Creelan
I think the analogy I heard is that when we give IL-2 to a patient, we’re basically flooding the hallways with caffeine. And if you give them a gene-edited recombinant cytokine that’s membrane-bound, you’re basically giving each person their own cup of coffee. You get to choose who gets it, and when they’re activated they get to take a sip or not.

But if the hallways were flooded with caffeine, we’d all be burned out and zonked out. And that’s what happens with conventional TIL, right?

And so, Lilit, you were showing that the PD-1/ICI patients had a much lower amount of central memory T cells.

Lilit Karapetyan
Right. I think what was really important for our cohort of patients was that we had a lot of patients who received immediate ICI before tumor harvest. Some received it as part of a trial, others before trials.

What was really important to highlight was the diminished stem cell memory cells, which, as you know, although described with a different phenotype by other groups, have an important role in durable response.

They definitely decreased. Also, the impact on functional activity, like costimulatory signaling, was really interesting to observe because it was clearly diminished in those with prior ICI exposure.

I think that’s an important argument when we sequence therapies, especially nowadays when we have approved combination therapies and are using those in a neoadjuvant setting in patients with advanced melanoma. It’s an important piece of data to think about when we introduce a new line of therapy in that setting.

Benjamin Creelan
Yeah, certainly in lung cancer we’re just now getting on board with neoadjuvant therapy. But I think the profound thing was that even if we look back at the original LN-202 trial, the patients who had fewer doses of immune checkpoint therapy did better than people who were on it for more than six months. Maybe they had more of those stem cell memory cells, right?

Lilit Karapetyan
Right. In general, even with the real-world data that we looked at here, I think lines of therapy definitely matter. And duration.

The only caveat is that sometimes I think about those patients who have melanoma and have gone through many lines of therapy and are still durable — that patient probably also has good disease biology. That’s one small caveat.

Benjamin Creelan
But lower duration of ICI had more response. That was the surprising thing, I thought.

Rodabe Amaria
I think the sequencing issue is a huge deal in melanoma especially. Because we do have frontline ICI that can potentially durably benefit maybe 40% of people.

So I think it’s a little hard, in my opinion, to say we need to be doing TIL earlier and earlier. I agree we shouldn’t be doing TIL as sixth-line therapy. Maybe not even fourth line. Maybe the sweet spot is second line.

It’s a little hard in my mind to move it completely frontline just because we do have effective ICI options.

If you’re looking at it from a socioeconomic standpoint, though, you could say one-time TIL therapy might be cheaper and have better quality of life than two years of anti-PD-1-based therapy. So I think that’s a viable approach. But I’m reluctant to say it should be frontline for everybody.

Daniel Olson
Yeah, I think the debate is really in that second- and third-line setting.

We have a lot of evidence suggesting it’s really hard to get patients to TIL therapy beyond that. And certainly the way the label is written right now, requiring BRAF exposure, that’s a challenge because many people are using PD-1 antibodies and BRAF inhibitors first to maximize response and then trying to send the patient to a TIL center afterward.

More than sixty percent of patients with melanoma are not treated by academic physicians. They’re being referred at a very different stage.

So I think there are practical things we can do from a messaging standpoint to say earlier is better. If you’re starting BRAF therapy, you should hopefully already be starting the TIL referral process.

And we can think about ways to get patients procured or treated before they’re fully resistant, because we don’t want to leave anything on the table.

At the same time, we’re hoping to develop better therapies. Gene-edited TILs and phenotypic selection of TILs are all really interesting.

Ben, maybe that’s a question for you. We talked about the challenges of neoantigen-selected TIL, and companies that tried to do that aren’t around anymore. I think they showed how difficult it is.

So now we’re using surrogate markers for phenotypic selection. Can you talk about the logistical complexity of that compared to neoantigen selection?

Benjamin Creelan
Yeah. I think the Rube Goldberg idea of synthesizing neoantigens, predicting them genomically, which sometimes takes four or five months, just isn’t feasible in real practice.

But these surrogate markers enrich maybe two-fold, five-fold, ten-fold for antigen-reactive cells in that initial culture. I think that’s a very viable method.

One of the promising approaches presented from the Netherlands showed tumor clusters using cell sorting for CD146, which allows you to select T cells physically associated with tumor. If that can be put into a trial, I’d be very excited to see how it performs.

That being said, sometimes simpler is better. Over the last five years, we’ve surprisingly stuck with bulk TIL without a lot of elaborate methods.

Going back to referral patterns, let’s say the phase III trial is positive for LN-202 in melanoma. Are community physicians going to be referring newly diagnosed melanoma patients to University of Chicago? Will they start pembrolizumab beforehand? Are you going to be trying to resect a rapidly shrinking tumor?

Daniel Olson
Yeah, it’s a good question. I don’t think we’ll know until it happens.

My suspicion is that the routine patient will not get referred for TIL early. That’s just not how incentives align in the world.

I think it will be for the highly motivated patient who’s trying to hit a home run on the first swing.

That makes sense for some people. The “one and done” concept, not wanting long-term exposure to multiple therapies, wanting to take the best shot first, that may be right for certain patients.

But realistically, those are going to be patients who are plugged in or treated at academic centers. Sometimes where you are and what your options are determines what you get. That’s just the reality of the landscape.

Benjamin Creelan
An interesting point Dr Miklos mentioned was that in TIL we don’t put in a genetic marker so we can track TIL over time. But as several of us have said, we can track the TCR barcode of infused T cells over time.

There’s this recurring theme that persistence is important. Having the ability to use acetazolamide as an inducible way of reactivating TIL seems really attractive.

The idea of vaccination against neoantigens, which hasn’t really been fully fleshed out with TIL in melanoma yet, is another strategy.

What are your thoughts on solving the persistence problem?

Rodabe Amaria
Yeah, I think it’s tough to know for sure.

The Obsidian approach is interesting because we think we’re reactivating the OBX cells by pulsing with acetazolamide. It’s a little hard to prove.

It would be helpful to have serial biopsies to show these cells are actually going into the tumor and mediating killing.

We can track these engineered cells in the blood pretty easily, but what becomes challenging is maintaining those numbers over time.

We do see evidence the cells persist, but it’s still a very small subset of the actual bulk infusion product.

And so it ends up being a terribly inefficient system with weeks and weeks and huge expense to expand these cells. But absent better technology, that’s where we are right now.

And the beauty and challenge of the science is that whatever one or two clones or ten clones are working in patient A are completely different from what’s happening in patient B. That becomes very difficult to understand and target.

Daniel Olson
I think it’s really interesting how advanced the science is right now and how many questions we’re asking.

We’re looking at different phenotypes and what associates with durable responses, and we’re still just characterizing all of this.

At the same time, there’s so much we can do. We can manipulate TIL ex vivo. There are so many engineered cell therapy approaches and gene editing techniques.

But we also have to think about what’s practical.

Sometimes perfect is the enemy of good. There’s so much we could do, but when that happens, manufacturing gets extended so long that it’s no longer practical.

So maybe there are intermediate states where phenotypic markers that can be assessed relatively quickly give us something better and represent a good compromise.

It really feels like the Wild West. There’s so much you can do, but how much should we do? What’s realistic? Those are the important questions.

Benjamin Creelan
I face this question all the time. I have a lymph node, a liver lesion, a lung lesion, often I only have one area to operate on.

Are we selecting certain areas over others for TIL resection at this point? Aside from operational reasons, would I prefer to reach for a lung tumor or a lymph node?

Rodabe Amaria
I think Iovance has the dogma that lymph nodes may be better starting material and maybe liver is not as good.

But I really think that’s center-dependent.

At my institution we harvest a lot from liver because we have excellent minimally invasive surgeons. We’ve had very few out-of-specification products.

So I really think it varies center by center and surgeon by surgeon.

Besides lymph nodes, which we all think are probably the best, I’m not sure there’s a clear second-best site.

Daniel Olson
Yeah, I think lymph nodes are good for generating a product that meets quality specifications. But we don’t know whether an in-spec product from liver is different from an in-spec product from lymph node in terms of outcomes.

Those are questions we’ll hopefully answer through real-world data.

Benjamin Creelan
And with that real-world data, is there maybe a threshold around 30 or 40 billion cells? Or is it more of a black box involving the percentage of antigen-reactive cells present?

Lilit Karapetyan
Yeah, I think in terms of cell count, specifically for TIL persistence and durable response, I do believe threshold matters.

I’ve seen that even in real-world lifileucel patients.

But it’s not necessarily linear, it’s not always that higher is better,  but there definitely is a threshold we need to meet.

Daniel Olson
If I understood your data correctly, it was really the first quartile where outcomes weren’t as good, but beyond that there wasn’t as much distinction?

Lilit Karapetyan
Correct. And even with the Iovance real-world data, there wasn’t much difference because their numbers were very similar, around twenty to forty billion cells, and of course we’re talking only about in-spec products.

Benjamin Creelan
Right, exactly.

Certainly in our lung cancer experience, we had some patients who received relatively few cells, maybe five million, and they didn’t respond.

Whereas your initial patient on OBX-115 was amazing. Multiple disease sites. You know it when you see it, right? The patient’s having CRS, edema in the brain, almost like a TENS-type syndrome, and the lymphocyte count shoots up to 30,000.

Have you seen that too, that patients with higher lymphocyte counts are basically having antigen-driven expansion of cells in the body?

Rodabe Amaria
Yeah, I had never seen it before this.

And I still think there are some patients with that product who are exceptional responders and really do have those same kinetics, maybe not as high and maybe not 30,000 absolute lymphocytes within five days, but there’s definitely a subset that expands amazingly well.

The problem is we can’t predict who those patients are at all. It seems completely random.

But those patients with that kind of exponential initial engraftment seem to do the best long-term and have durable responses without additional therapy.

It’s crazy because we can’t figure out who they are.

Benjamin Creelan
Well, thanks for this quick chat, this little huddle after our presentations.

Excited to see what the rest of this conference will hold on Solid Tumor Day today with CARs for solid tumors, TCRs, and TIL. We’ll see bispecifics later tomorrow.

And thanks for presenting on this really exciting future of cell therapy.