But as we often do for these patients with more high-risk de novo metastatic disease with high Gleason scores, we get conventional imaging throughout their treatment course, we don't just rely on their PSA. And this patient in fact, did develop a hypodense lesion in the posterior right hepatic lobe, as you can see there. So we biopsy this and just six months into his treatment course, he now had treatment-emergent neuroendocrine prostate cancer. High-grade KI67 was high, indicating a high proliferation index, was negative for our canonical adenocarcinoma markers, was positive for INSM1, synaptophysin. And so we did next-gen sequencing on this tumor biopsy. This patient had dual loss of RV1, TP53, and actually did get testing for DLL3, which showed 80% tumor positivity.
This patient had platinum etoposide for six cycles, and as we see in other high-grade neuroendocrine cancers, had an initial response, but then subsequently progressed. And we'll come back to his case a little bit later in my talk. When we think about DLL3, we can't talk about that with first talking about lineage plasticity and treatment-emergent neuroendocrine prostate cancer. And this is a slide I put together a few years ago, but it really just highlights that this is really a spectrum. It's really hard to bend patients in discrete categories when we know there's just significant heterogeneity within a patient across patients.
We know that epigenetic reprogramming is really at the forefront of how this disease drives from an androgen-dependent to an androgen-independent state. And along the way, you can have different disease features that can get captured by different clinical tests that we perform, pathology, IHC, imaging, next-gen sequencing by genomics. We really think about DLL3 as kind of at the right end of that spectrum. These are tumors that have terminally differentiated in a neuroendocrine prostate cancer. And DLL3, we think is important along with ASCL1, which drives the transcription of DLL3, is important in maintaining that neuroendocrine state. And as I'll talk about in my talk today, there are different imaging modalities and biopsies and things like that, that we can use to identify these patients.
Just a quick word about neuroendocrine prostate cancer, we know this is rare at diagnosis, but in the treatment-emergent setting, is actually not uncommon. And so as we think about developing new drugs in prostate cancer in the androgen-resistant setting, we have to think and keep in mind, neuroendocrine prostate cancer. In our prospective series from our West Coast dream team, when we biopsied patients with metastatic CRPC and we profile their tumors, not looking for neuroendocrine, but just anyone with an accessible lesion, we found small-cell neuroendocrine histology in 17% of all of our biopsies that we perform. So this is not a tiny subset of androgen-independent prostate cancer, but in fact, is a pretty large subset that we need to sort of figure out what are the right treatment modalities for.
When we talk about DLL3 expression, we know that it's quite heterogeneous. The expression receptor density is lower and more heterogeneous than we think about in other surface antigens that we think about for targeting with ADCs, T-cell engagers, radioligand therapy. We think that in small-cell neuroendocrine prostate cancer, about 60% to 80% of tumors express DLL3, but the expression is quite heterogeneous. If you look at the figure on the bottom right, these are circulating tumor cells from patients with neuroendocrine prostate cancer. And on the left you see in red, really the wide range of expression of DLL3, even within a given patient, just speaking to the heterogeneity that we see. And on the bottom left, we see from the rapid autopsy series from the Fred Hutch, just the expression of DLL3 in individual tumors and across patients.
So shown in the yellow bar is the DLL3 expression, which is primarily focused in the neuroendocrine-positive, AR-negative subset, but you see a wide range of expression within that subset. This is a heterogeneous target, but fairly commonly expressed in the treatment-emergent neuroendocrine setting. So how do we target DLL3? We think about our ADCs, and classically we think of ADCs as requiring a high surface receptor density, greater than 10,000 copies per cell. And of course, this treatment effect is mediated via receptor, mediated endocytosis, and the payloads that we've seen tested clinically had varying levels of bystander effect. We think of relatively low bystander effect with a microtubular targeting agents and higher bystander effect with a TOPA-1 inhibitors, for example.
T-cell engagers, of course, internalization is not required with these, but immunologic synapse is required and may be effective even in situations where we have lower expressing or heterogeneously expressing situations like DLL3. And of course, targeted radioligand therapy, here we leverage the crossfire effect and properties to really enhance the binding when you have a lower expressing antigen, driving up the binding affinity and staying on the target and internalizing within the tumor are really important when we think about targeting something like DLL3. Let's first focus on the antibody-drug conjugates. Our first experience in DLL3 targeted ADCs comes from Rova-T, which I think was a challenge for a lot of us that treat patients in the clinic.
This drug is an ADC with a PBD payload that causes DNA damage to exert its antitumor effect and had great preclinical data, a very strong rationale, but this is a great example of where the payload really matters. And in the clinic, this drug was just very toxic related to the PBD warhead. A lot of pleural pericardial effusions, pneumonitis, transaminitis. And so subsequently, development was halted after the futility was reached for their Phase-3 trials in small-cell lung cancer. We did a basket study in neuroendocrine prostate cancer along with other high-grade neuroendocrine cancers, and there were some patients that had benefit and there was a suggestion that the responses were related to the degree of DLL3 expression.
Of course, we want to think about how can we broaden the therapeutic index of ADCs. We think about payloads with a higher bystander effect and more predictable toxicity profile. And there are DLL3 targeting ADCs in clinical trials that include patients with neuroendocrine prostate cancer. A couple of examples are listed there. We think about the improved technology here in terms of the linker technology, optimizing the DAR, site-specific conjugation with leads to a more predictable DAR, higher binding affinity, more efficient internalization. All these things are really important. I want to highlight and piggyback on what Tanya said.
I think bi-specifics are really important here, both given the heterogeneity of the disease, as well as this concept that when you target two different antigens, perhaps you get more efficient internalization within the cancer cell. So one plus one may be actually greater than two in this scenario. Then of course, moving beyond the full antibodies to smaller molecules and small peptides, shorter systemic circulation may lead to less toxicity in this regard. Let's talk quickly about the T-cell engagers. This is work that was led by John Chao, one of our physician scientists at UCSF, evaluating Tarlatamab, which is a DLL-3 TCE in neuroendocrine prostate cancer tumor models. These are all PDX models. And the point being is that you see really effective activity here that is DLL3 expression dependent.
So you have three different examples of PDXs with varying levels of DLL3 expression, and there was a nice correlation between the degree of response and the depth and the degree of DLL3 expression shown here. Interestingly, and I think we need more data, what happens to DLL3 over time. In this preclinical experiment, those tumors that had late relapse actually seemed to have lower DLL3 expression. So temporally analyzing what happens with DLL3 or expression over time is going to be important. We subsequently led a clinical trial with Tarlatamab and neuroendocrine prostate cancer. This was a very broad trial that enrolled patients across the spectrum of aggressive variant prostate cancer, did not prospectively select based on DLL3.
But in fact, when you look at the subset of patients that had DLL3 expression, we see that there's a correlation. Those tumors that had higher DLL3 expression were more likely to respond to Tarlatamab. Just quickly going back to our case, I'm running out of time here, this patient was enrolled on a Tarlatamab clinical trial and actually had a really nice durable response lasting over a year with platinum-resistant prostate cancer that had neuroendocrine differentiation. Finally, just talking about targeted radioligand therapy, I don't need to tell this group too much about the radiochemistry here. This is really work led out of MSKCC, looking at different constructs to achieve therapeutic benefit in neuroendocrine prostate cancer.
This was the same antibody backbone that was used as part of the Rova-T, and you can see really nice preclinical data with tumor regressions. This has been translated into the clinic and this was data published in Lancet Oncology. An imaging agent based on Zirconium-89 that showed a really nice response and imaging uptake of several patients with neuroendocrine prostate cancer. And so I'll just finish by saying that they're merging new targeted radioligand therapies, next-generation Lutetium-177 led to 12 actinium compounds that are showing really nice preclinical data and are in early-phase clinical testing.
So where do we go with DLL3 testing? We think about combination therapy, given the heterogeneity of the disease, targeting different surface antigens, different payloads. Understanding the mechanisms, what drives surface expression of DLL3, and then optimizing patient selection. Going beyond just looking at neuroendocrine prostate cancer, but really focusing on the target, I want to highlight a trial of Tarlatamab and DLL3-positive prostate cancer led by Karen Audio at MSKCC, really getting away from histology and focusing on DLL3, understanding if Tarlatamab has benefit in that regard. Thank you very much.