Andrea Miyahira: Hi everyone. I'm Andrea Miyahira, and I'm the Senior Director of Global Research and Scientific Communications at PCF. Today, I'm joined by Andrew Armstrong, a professor at Duke University. He and his colleagues recently published the paper, PSMA Positive Circulating Tumor Cell Detection and Outcomes with Abiraterone or Enzalutamide Treatment in Men with Metastatic Castrate Resistant Prostate Cancer in Clinical Cancer Research. Dr. Armstrong, thank you for joining me to discuss this today.

Andrew Armstrong: Andrea, it's a real privilege to be here with you. I'm going to share my slides and go through this as like a mini journal club. So as the audience knows really well, PSMA-targeting therapies are one of the hottest things in our clinic right now. Patients are living longer, they're living better, and the treatment with PSMA-targeting radioligand therapy is evolving very quickly into earlier and earlier disease settings than we have right now, which is for patients who have failed a AR inhibitor and a taxane, it's rapidly moving into the pre-chemotherapy and castration-resistant setting as well as the hormone sensitive setting. So there's a great unmet need for biomarkers that can optimally identify patients who would benefit or not benefit from PSMA-targeting therapies. We all know that PSMA PET scans are a good biomarker for that, but we also know that many patients who test positive using the conventional PSMA PET scan may not benefit as much as others. So we sought to develop a liquid biopsy using circulating tumor cells that we could associate with outcomes in men with mCRPC and ultimately apply that to a PSMA-targeting context.

So the purpose of our study and manuscript that we're talking about is to develop that assay measuring PSMA protein expression on CTCs, capture that heterogeneity, both positive and negative of PSMAs in a liquid biopsy, and look at that longitudinally in the context of AR-targeting therapies, correlate that with genotype as well as other important phenotypes of patients such as neuroendocrine disease, AR-v7 and genotype. So this is the paper that's now out in CCR, and so this was funded through a PCF Challenge Grant, PCF Movember funded this in 2014, and we've published many papers related to the PROPHECY study where we initially validated AR-v7 seven across multiple assays, but also looked at a number of key liquid biopsies like neuroendocrine phenotype, chromosomal instability phenotype, enumeration, genomics. We've presented on this at retreats in the past. One of the great things about this PCF Challenge award is that we bank samples and we're able to collect and annotate those samples and ask new questions. So the gift that keeps on giving.

For this assay, we used the Epic Sciences platform, which is like a blood smear and it uses immunofluorescence in multiple colors as well as a facial recognition software to identify CTCs in the blood. We've shown that the number of these CTCs correlates with overall survival. We've also shown that using a nuclear AR-v7 protein assay that we can associate AR-v7 with overall survival in an AR therapy context. This led to the first Medicare reimbursed and validated liquid biopsy assay that can inform treatment decision -making between a taxane and an AR therapy. So it can be useful, and you can see here from the original PCF Challenge award in the multi-center blinded independent prospective study that we showed that with Abi and Enza, AR-v7 detection associated with worse outcomes, but not with taxanes. So we wondered what this would be like for PSMA. I do also want to point out that we developed and validated a neuroendocrine phenotype of CTCs also showing in PROPHECY as well as in a validation cohort that when you have small cells, neuroendocrine cells floating around in the bloodstream, that's associated with very poor outcomes.  So this will come back in our PSMA study.

We showed that there's lots of CTC phenotypes in our patients. Some are positive for AR-v7. Some have a more neuroendocrine signature, some have both. Some have chromosomal instability, such as BRCA mutations and P53RV loss that can lead to genomic instability. When you have these biomarkers, it's associated with very poor outcomes as opposed to not having any of these biomarkers where many patients do very well. So with this new project, we developed the PSMA assay. Everyone knows PSMA is expressed on some normal tissues like salivary glands and small bowel and kidneys as well as lacrimal glands, but it's fairly ubiquitous in adenocarcinoma, the prostate. It tends to be lost in neuroendocrine. So we know that some tumors are very heterogeneous. We also know that with PET scans and radioligand therapies, that patients can live much longer, that the PSMA PET scans can be essentially turned off at least for a little while from these radioligand therapies that you see PSA declines and better patient reported outcomes.

So in the PROPHECY study, we had a cohort of men that were PSMA CTC positive, and this is the baseline table that compares to patients who are positive versus those that are negative. You can basically just see the patients that have lots of CTCs tend to have worse prognostic characteristics. They have tend to have more metastatic disease, the diagnosis, faster PSA doubling times, more radiographic progression, more CTCs, and tend to be a little bit more prevalent for AR-V7. We developed an assay using control cells, those that lacked PSMA, like PC-3, and those that had fairly high intense PSMA expression like LNCAPS. This was performed in the Epic Sciences laboratory on these spike samples, optimizing that this assay for signal-to-noise detection. In PROPHECY, we started with 120 patients and we have 97 patients that had CTCs evaluable for this question and another 57 that had progression samples after progression on Abi or Enza for this analysis. This is just a picture of what these PSMA CTCs look like in the fourth channel.

You can see examples of high PSMA expression, and at the bottom, basically absent PSMA expression where you really have to zoom in and overexpose just to see a tiny bit of PSMA expression. So you can see that heterogeneity just graphically right in front of you. When you graph this for all the patients, according to patient ID, you can see that a good number of patients have no CTCs. So they're obviously not evaluable for PSMA expression. About half of patients have PSMA detection. That means half lacked PSMA. Of those that were PSMA positive, about 10 to 15% were PSMA ubiquitous. That means all their CTCs had PSMA and that would be shown here in the blue bars as compared to the red bars. I think what you can grasp from this plot is there's great heterogeneity between patients and even within patients and over time that the vast majority of the CTCs that we see in the circulation are actually lacking PSMA, even when a patient is positive. This heterogeneity is probably what's driving a lot of the patient outcomes.

You see that again at progression, a little bit more PSMA expression, but the vast majority of patients, even when they're positive, they often have great heterogeneity for that PSMA expression where most patients are not ubiquitously PSMA positive.

We did see some interesting correlations with chromosomal instability. We found that most patients with that chromosomal instability have very intense PSMA expression. Inversely, when patients have small cell neuroendocrine disease, they very rarely have PSMA positivity. Most of these CTCs are PSMA negative, in fact. AR-v7, you see great heterogeneity of the PSMA CTCs and a wide range of heterogeneity. So it's a little bit inversely associated that AR-v7 we know is associated with CTC heterogeneity, but also with PSMA heterogeneity. So those were some interesting correlations. Importantly for this paper, we show that using a cutoff of two PSMA positive CTCs, that Susan Hallabi identified as a statistically optimal threshold for prognosis, we showed that that's associated with both overall survival and progression-free survival, even after adjusting for the Hallabi clinical factors and the number of CTCs.

So a hazard ratio of about three for death and almost three for progression-free survival on Abi or Enza, the overall survival multi-variable hazard ratio is nearly three. So suggesting that this can provide an independent prognostic signal for patients in the context of AR-targeting therapies. Interestingly, CTC homogeneity is associated with a better survival. So it's really what's driving that poor survival. It's not necessarily the PSMA, but what that is measuring is a greater disease burden and a greater heterogeneity. PSMA is probably what's driving that poor outcome. As you can see in the blue curve, the patients who meet that threshold, but have many CTCs that are also negative for PSMA is what's driving that core survival while patients, again, about 12% of patients, had ubiquitous PSMA expression. These patients are on the red line. They have one of the best survivals. So our conclusion is that we have a liquid biopsy now that can quantify protein expression. This would be missed in conventional liquid biopsy assays such as CT DNA, where this is not encoded necessarily in the DNA. It would not be predictable by conventional assays.

So we have a CTC biomarker that can capture the expression of PSMA and its heterogeneity that can be tracked longitudinally. We show that it's associated with outcomes, even adjusting for clinical features that it's inversely associated, as you might expect, with disease lineage plasticity that's associated with loss of PSMA expression and heterogeneity. What we're planning to do with this is to launch a prospective study that will start this summer. We haven't come up with the acronym for that. So we're definitely open to suggestions where we will take patients in the context of PSMA lutetium targeting therapy and measure this liquid biopsy more as a predictor of efficacy of PSMA therapies and clinical outcomes now. We were not able to do this until this year when we have now, finally, patients receiving this therapy in an FDA approved context.

We will look at now what are the biomarkers associated with better outcomes with this specific PSMA-targeting therapy, their genotypes, phenotypes, and how does that change in the context of PSMA-targeting, and over time, what cells emerge that we can maybe better target with PCF-funded novel targeting therapies. So we speculate that patients that have no CTCs will do the best, patients with great heterogeneity or lack of PSMA will do the worst, and that patients with more homogeneity will do very well with PSMA-targeting therapies. I'd like to thank the PCF Movember funding agencies for the Challenge Award, as well as the big team at Memorial Sloan Kettering, Hopkins, Chicago, and Cornell, who contributed that to this paper as well as Epic Sciences who provided much of the assay work behind this.

Andrea Miyahira: Thank you for sharing that, Andy. Some questions that I had were, why do you think the presence of PSMA positive CTCs were associated with worse outcomes? Was there more baseline disease burden was higher? Were there a higher number of CTCs in those patients?

Andrew Armstrong: Yes. So as we showed in the baseline table, patients who are PSMA positive have to have more CTCs. So when we did the adjusted analysis, we had to adjust for that. Despite the adjustment for the total numbers, we still saw a worse prognostic association. I think diving into the heterogeneity, I think that's probably driving this, when you have a huge disease burden such as visceral metastases or widespread bone metastases, these are the patients that have more CTCs in their blood. They will have more heterogeneity as well. So their disease is being driven not just by the PSMA positive CTCs, but also by PSMA negative CTCs. They can benefit from PSMA-targeting therapies, we think, but we tend to see resistance to drugs like PSMA lutetium emerge after just six to 12 months probably because it's not targeting the whole cancer, it's targeting just the PSMA positive portion of the cancer. So we need to ideally develop better therapies that will target that PSMA invisible disease.

Andrea Miyahira: Thank you. What do you think this means in terms of CTCs as biomarkers? One thing that did intrigue me was that your PSMA negative group had better outcomes, but how heterogeneous were the patients in that group also?

Andrew Armstrong: Yeah, the patients who do the best have no CTCs. So it can be a useful biomarker to identify a good prognosis group, but it's not as useful there because there's no disease that's detectable in the blood that's good for the patient, but it's frustrating for those developing CTC biomarkers. That's true of plasma DNA as well. One of the best patient groups is those who have no circulating tumor DNA, so you don't get an informative assay for genotype as well. So that's where it may be helpful to have the PET scan as another biomarker that can complement a liquid biopsy. We don't think this PSMA CTC assay will replace PET scans, but rather compliment the PET scans. Even patients who test positive by a whole body PET scan may have heterogeneity in their blood. During that disseminating process, that may help you identify those who might be best suited for PSMA lutetium and those who maybe needed to be followed more closely, and those who may not benefit at all where they might benefit from some alternative treatment like docetaxel, cabazitaxel, radium, or an investigational approach.

Andrea Miyahira: Well, thank you. I think it's clear from the PSMA lutetium studies that it's a very promising treatment, but we do need better biomarkers. So I look forward to what your studies are going to teach us in the future.

Andrew Armstrong: Thanks so much.

Andrea Miyahira: Yeah. Thank you for joining me today.

Andrew Armstrong: Thanks to the PCF. We really appreciate the support over the last 10 years.