DLL3 in Neuroendocrine Malignancies: From Expression Marker to Predictive Biomarker?

Debate Introduction

Round 1

Is DLL3 a Clinically Actionable Predictive Biomarker in NEC?

Early clinical data suggest a potential relationship between delta-like ligand 3 (DLL3) expression and response to DLL3-directed therapies, but predictive utility has not yet been established. In Round 1, faculty examine what current clinical evidence does—and does not—show, while considering whether DLL3 can be measured and interpreted with sufficient reproducibility to support predictive use. 

Round 2

If DLL3 Becomes Actionable in NEC, Who Should Be Tested and When?

As DLL3-directed therapies move into earlier stages of clinical development, questions around testing strategy and timing are becoming increasingly relevant. In Round 2, faculty weigh selective versus broader, earlier DLL3 characterization, considering disease context, intended clinical or trial use, tissue availability, specimen quality, and the potential consequences of delaying testing.

Round 3

What Is Needed to Establish DLL3 as a Clinically Actionable Biomarker?

In the final round, the discussion shifts from opposing positions to the path forward. Faculty explore the clinical evidence, assay and scoring standards, specimen requirements, reporting practices, and pathology–oncology workflows that would be needed for DLL3 results to become reliably interpretable and potentially useful in treatment or clinical trial decisions. 

Transcript

Jonathan Strosberg, MD: All right, so we're going to be talking primarily about extrapulmonary neuroendocrine carcinomas. In other words, poorly differentiated neuroendocrine carcinomas that start outside of the lungs, so not really about small cell lung cancer. We know they're similar to small cell lung cancer, and small cell lung cancer has a pretty dismal prognosis. Survival rate at 5 years is only 7%. Extrapulmonary neuroendocrine cancers vary a little bit by primary site, but pretty much the same, the median survival is less than a year on average. 

There is even less evidence to guide treatment for extrapulmonary neuroendocrine carcinoma than there is for small cell lung cancer. We often borrow treatments from small cell lung cancer, for example, platinum etoposide. We've been doing that for many years. But the evidence level for extrapulmonary NEC [neuroendocrine carcinoma] is relatively poor. Is it better, for example, to treat a GI NEC with platinum etoposide compared to FOLFOX or FOLFIRINOX? We really don't know. There's some studies looking at that. We don't know whether immunotherapy, whether PD-1 inhibitors play a role in extrapulmonary NEC as they do in small cell lung cancer. We have very limited data. 

Now, DLL3 [delta-like ligand 3], as you know, is an emerging biomarker. It's been proven to be important target in small cell lung cancer with approval of tarlatamab in 2024, which is a bispecific T-cell engager targeting DLL3 and CD3. There are currently no other approved drugs for poorly differentiated neuroendocrine carcinoma, including small cell, but there are many in development, particularly looking at DLL3. Currently, treatment with tarlatamab does not require DLL3 testing. We note it's expressed in nearly all small cell lung cancers. As we'll see, it's not clear that's the case for all extrapulmonary neuroendocrine carcinomas. There are numerous other studies targeting DLL3, including antibody drug conjugates, radio pharmaceuticals, CAR-T and NK therapies, as well as other bispecific and trispecific T-cell engagers. So a lot of activity around this relatively new target. 

We know that DLL3 has emerged as a biologically relevant therapeutic target around neuroendocrine carcinomas. The question we're trying to figure out is whether it's just a target or is it a predictive biomarker? In other words, do you need to have evidence of DLL3 positivity? And if so, at what threshold in order to justify use of DLL3 targeting drugs? And is that information clinically actionable today? I want to emphasize again that extrapulmonary neuroendocrine carcinoma is not exactly the same as small cell lung cancer. It's similar, but it's not identical. 

For example, extrapulmonary neuroendocrine carcinoma often arises out of mixed histology. For example, for GI, it's often a mixture of adenocarcinoma and neuroendocrine. We think it might start out with mutations that are characteristic of adenocarcinoma before developing mutations targeting like P53 and RB1 aberrations that are more characteristic of neuroendocrine carcinoma. For extrapulmonary NEC, the large cell phenotype is actually probably more common than small cell, and often it's not specified. We'll get reports just saying poorly differentiated neuroendocrine carcinoma with small cell features or large cell features. 

The other thing is it's not strongly associated with smoking. There's a very weak smoking association. This is in contrast to small cell lung cancer, which is typically homogeneous. It's not mixed with adenocarcinoma, or squamous cell carcinoma, or transitional. It's virtually always associated with either mutations or dysregulation of P53 and RB1. That's true for the majority of extrapulmonary NEC, but certainly not for all. Of course, there's an almost universal association with smoking. These are similar diseases, but not identical. 

So I'd like to have Dr Roden come up and talk a little bit more about neuroendocrine diagnosis. 

Anja Roden, MD: All right, good morning everyone. It's a great pleasure to be here. I'm actually a pathologist, so we will talk a little bit about pathology and DLL3 testing. Neuroendocrine neoplasms are divided in neuroendocrine carcinomas and neuroendocrine tumors based on morphology. Neuroendocrine carcinomas are poorly differentiated. They have a worse outcome. Neuroendocrine tumors are well-differentiated and they have a much better outcome. 

Amongst the neuroendocrine carcinomas in the lung, we have small cell carcinomas and large cell neuroendocrine carcinomas. Then as you already heard, there are also the extrapulmonary neuroendocrine carcinomas, which are mainly found in the gastrointestinal tract, the pancreatobiliary tract, GU and GYN, but they can be also seen elsewhere. We also have what we call the MiNENs, which are the mixed neuroendocrine/non-neuroendocrine tumors, or neuroendocrine neoplasms. 

How do we establish the diagnosis? Neuroendocrine neoplasms have what we call the neuroendocrine morphology. It's an organic morphology. Usually we see a nested cause pattern of the cells. Sometimes we see a trabecular or pseudo-setting cause pattern. Then we have to prove that it's actually of neuroendocrine differentiation, so we use neuroendocrine markers such as synaptophysin, chromogranin, and INSM1. You can see actually a very nice example over here. This is INSM1 expression. 

The tumors we are talking about today express keratin, as you can appreciate over here. There are some neuroendocrine neoplasms that do not express keratin, but the neoplasms we talk about today are carotene-expressing. Ki-67, as a pathologist, helps us to differentiate between a neuroendocrine tumor and a neuroendocrine carcinoma, specifically on small biopsies when we have a lot of crush artifact. However, in neuroendocrine tumors, we are also often asked to actually perform a Ki-67 and then this is more for prognostic value. 

All right, so DLL3 is an inhibitory notch ligand that has been implicated in the neuroendocrine neoplasm tumor oregenesis. Notch activation is associated with suppressed neuroendocrine differentiation and the DLL3 over-expression causes reduced surface localization of the notch responses. So that means notch silencing, and thereby promotes neuroendocrine phenotype switching.  

DLL3 is highly expressed in small cell carcinomas. We basically see expression of DLL3 in almost all of the small cell carcinomas, but it's really variably expressed in extrapulmonary neuroendocrine carcinomas. One of the reasons may be that they are different from small cell carcinomas, as you just heard. 

Here is an example that you can see that DLL3 is differently expressed on, for instance, the same instance, which is the uterine cervical neuroendocrine carcinoma. You see four different neuroendocrine carcinomas of the cervix. On the upper left-hand side, this tumor was negative for DLL3. On the upper right-hand side, here is weak expression, what we would call a 1+ staining intensity. In this tumor, there is moderate staining intensity, and this one has a strong or 3+ intensity. 

Now, many of the studies on DLL3 actually report the percent of tumor cells that are positive for DLL3, but other studies report an age score. What is an age score? The age score is actually a combination of the staining intensity and the tumor cell expression. How it works is if you have, for instance, a tumor that diffusely expresses DLL3 at 1+ intensity, so 100% of the tumor cells express DLL3 1+, you would multiply 1 by 100%, which is an H score of 100. 

If you have a tumor that strongly diffusely expresses DLL3, so 100% of tumor cells are 3+ positive, you multiply the 3 with 100, you have an H score of 300. Now I will turn it back to Joe and see where we go with the DLL3. We can report DLL3, but do we really know that it is a predictive biomarker?