Michael Hofman: Around 10 years ago, we started the very first randomized trial of Lutetium PSMA, the therapy trial, comparing it to cabazitaxel. And we made this video actually commissioned by the Prostate Cancer Foundation of Australia. And it was played at the launch of the trial at the ANZUP cooperative trials group meeting. And the medical oncologists in the audience were outraged. They said, "We'll never randomize patients if we show this video because all the patients will want the Lutetium arm and nobody will want the taxane chemotherapy." And indeed, this became a problem in the phase-three VISION trial with patients dropping out of the control arm. I don't think any of my disclosures are relevant for this talk.

Prior to the therapy study, off the back of compassionate access data from Germany, we did the very first phase-two trial of PSMA-617 at Peter Mac. And we produced this image, which won the SNMMI Image of the year in 2018. Perhaps hype because Lancet Oncology made us take this image out of the publication. They said, "We're just showing the eight best responders. It may be overstating the results." I said, "I'll move the figure to the supplementary appendix." And they said, "No, remove it completely." So I did, and it was published in Journal of Nuclear Medicine, and they liked it so much they awarded it the Image of the Year, also highlighting the differences in the way we think as nuclear medicine physicians compared to our medical oncology colleagues.

But indeed, Lancet Oncology was right because seven of eight of these patients died of prostate cancer. They all progressed, and some of them happened in a relatively short period of time. When we ran the therapy trial, it was a small phase 2 trial, only 200 patients, 100 randomized to Lutetium, 100 randomized to Lutetium PSMA. Because it was a small trial, we selected patients quite carefully. So everyone had both a PSMA and an FDG. We excluded around 30% of patients that either had low PSMA or FDG-positive discordant sites of disease. And we randomized 200 patients. And the early results were super promising. When you look at the PSA, 50% response rate, 66 compared to 37%. This is a massive difference between the two arms.

If we look at CT RECIST, 49% compared to 24% for cabazitaxel, again, a huge difference between the two arms. And these results were evident early on because these are sort of early readout. And this made me very excited because I thought with these large differences, we're going to see a big difference in overall survival despite being a small trial.

So in my brain, I envisaged the Kaplan-Meier survival curve was going to look like this. I thought Lutetium was going to blow cabazitaxel out of the water, but the reality was different. Actually, with longer follow-up, we showed no difference in overall survival.

We then hypothesized that this was due to crossover and when patients had cabazitaxel, they had Lutetium afterwards and vice versa. This is usually what we do when we don't get the result we like. We try to dig around the data, find an explanation, but we did get our biostatisticians onto it and they reassured us that this was not a crossover effect, that even when we tried to undo the crossover with advanced statistics, this was real. And they actually compared the overall survival between the therapy trial and the vision trial. And in fact, the survival in the Lutetium arm sort of matches. And it seems that in the vision trial, the survival advantage is potentially due to the suboptimal comparator arm as opposed to cabazitaxel and effective therapy that we chose in our therapy trial.

So it's very clear that Lutetium PSMA as a tumor-targeted agent is very effective. We see tumor shrinkage. We see that in the therapy trial. 50% of patients having RECIST-measurable responses. We see PSA going markedly down as a measure of a tumor-targeted therapy, but maybe Lutetium PSMA is less effective at stopping new metastases forming. And this is something I think it's real and something that I've come to realize over time. And if we conceptualize this, when we see PSA resist response in the Lutetium arm, this is a strength of Lutetium PSMA. But when we see the small increment in radiographic PFS and the lack of OS, maybe this is because we're not altering the natural biology of the disease as much as we would like, stopping these new metastases forming and maybe taxane chemotherapy is more effective at that.

Now, when we look at the ultimate outcomes, the only two things that matter rather than these surrogate outcomes, it's patient-reported outcomes living better or overall survival, living longer. And this is a mix of both the tumor-targeting and the new metastasis formation. And no doubt we actually do really good with Lutetium PSMA because it's an extremely well-tolerated treatment. We've treated some patients with up to 20 cycles of Lutetium PSMA over five year period. This is a patient rapid doubling time, 3 months, had failed taxane, had 20 cycles over 5 years with a cumulative renal dose of 56 gray, but actually after 20 cycles, his renal function was completely normal, EGFR over 90, and he actually had cabazitaxel after Lutetium PSMA. So this is not hype, this is a real patient, but this is very unusual. Not many patients get to this. When we look at the reality, this patient falls into this top group that has a complete response. And when we retreat them, they do well and we can repeat that multiple times.

But many patients are like patient number two. They have a progressive response to therapy, but by the time we finish six cycles, they still have residual disease. And some patients are like patient three. They have stable disease through cycles one and two, and then they progress. And some patients are like number four. Even with high PSMA expression, they just have primary progression even when we get effective targeting.

So how do we understand this? And in nuclear medicine, we like to use dosimetry, and we've done very fancy dosimetry in our phase-two trial. And what we could see is that Lutetium's not magical. It's a radiation delivery device. We give an intravenous injection, the radiation finds its way to the tumors. We know that the normal organ doses are actually very similar between patients. So these are the 50 patients in the trial, the parotid kidney dose, very similar between the patient, but the tumor dose is actually varied, tenfold between patients. And this maybe explains our heterogeneous response. And indeed, when we look at a dose-response relationship between a surrogate outcome, PSA response and dose, there's a very strong correlation.

But this actually falls apart when we try to compare dose to overall survival, maybe akin to the earlier hypothesis that we are affecting tumor killing, but not necessarily affecting the natural history as much as we want.

So in theranostics, is it a drug or is it a type of radiation? And I think it's a little bit of a both. And as we design our clinical trials, it's actually very good to have nuclear medicine, radiation oncology, medical oncology working together to come up with a little bit of a hybrid model.

Now, do we need dosimetry or can we just rely on our PET imaging? Because the SUV, the intensity of the uptake on the PET scan is a surrogate for the amount of tumor-targeting we are going to get when we deliver our isotope. And I developed this schema a couple of years ago, and in blue we have the non-game-changing isotopes. These are the ones where the uptake is really low. Germo referred to these yesterday as the theranostic zombies, and he articulated that there's quite a lot of these theranostic zombies out in the commercialization world at the moment. They've got low uptake on PET. You get low doses to tumor. They're really very unlikely to ever help a patient. You can see it on patient number one because you really have to wind up the PET scan to see any uptake.

PSMA and DOTATATE are both game changers. They're approved, and now there's a move to use them both earlier, combined and personalized. But we actually still can't beat good old radioactive iodine for treating thyroid cancer. First used this year will actually be 85th anniversary, and we cure people with a single dose of radioactive iodine who have bone or lung metastases, and that's because you get SUVs in the thousands. You get doses above 200, above 400 gray to tumor, and you really eradicate tumor.

Another challenge we have is tumor heterogeneity that we simply can't overcome with theranostics, but we can visualize it. And some patients just have a large burden of disease that we cannot target. And I think in precision medicine, actually, the patient selection is the key. And this doesn't matter whether you're using an ADC, theranostics or a T-cell engager. I think this is going to be a fundamental limitation, but we don't fully understand it, but we can use our PET imaging to at least try to biopsy discordant sites and try to understand this better.

Now, some people think we can't do PET scanning because of the costs and accessibility, but I want to disagree with this. This year will actually be the 50th anniversary of the first PET scan in oncology done in University of Pennsylvania. This took two hours to acquire this blurry image. Recently at Peter Mac, we acquired this PET scan just a few weeks ago. Tom, to put you on the spot, how long did this PET scan take to acquire?

This is a five-second PET acquisition, and this is on our total-body PET. So we've reconstructed this PET. You can see on the standard-of-care PET scanner, it took 15 minutes, and we can now do a PET scan in 5 seconds, 15 seconds. Actually, the 30-second scan is equivalent to our 15-minute scan on the old device. And if we scan for 1 minute, we get significantly superior image quality to the old 15-minute scan. So there's no reason that we shouldn't be doing more PET CT imaging to really interrogate what targets are expressed to fully evaluate tumor heterogeneity and use multiple traces to really get a full picture of what is happening.

The other promise is that we've got lots of different isotopes. We've got alpha, beta, OJ. And we've got lutetium, which is really the only approved radial ligand apart from radium for prostate cancer. Is it the perfect drug? Can we do better? We have the whole periodic table to play with, but we heard from Matthias Ober yesterday that actinium's been around for a decade and despite all the promise and hype, there's still very little prospective data showing that it's better than good old-fashioned Lutetium.

We've got a whole bunch of isotopes we can play with and maybe some of these will come to fruition. We've got my favorite Terbium-161 as a dual beta emitter. Can we eradicate those very last micrometastases if we use a ultra short emitter like Terbium-161, or can we do better with a alpha emitter, maybe Lead-212 with a half-life that's better matched to the peptides that we are using in clinical practice?

Another way to move forward is with combination therapies. We heard Shahneen Sandhu talk about this yesterday. We've had a big interest in this in our center. We've tried combining Lutetium PSMA with hormone therapy, chemotherapy, immunotherapy, PARP inhibitors, multiple radioligands, maybe in combination with surgery or external beam radiation. As some of these trials have read out, the rest will continue to read out. These are all small phase-one, -two trials, and we really need a little bit more industry investment in some of these combinations to work out, which of them are really going to make it in the real world.

So I'll leave you with this slide of the theranostic zombies. They are chasing the next Pluvicto. No doubt somewhere in the zombies is hiding the next Pluvicto.

And if you really want to find maybe whichever is going to be the next killer theranostics, then you might need to travel to Melbourne to our prostate theranostics meeting, which is going to be September 3 to 5 this year. Thank you.