The Future of DLL3 in epNEC: Diagnostic Integration and Emerging Clinical Directions

Dr David Klimstra explores how delta-like ligand 3 (DLL3) and other emerging biomarkers may be incorporated into extrapulmonary neuroendocrine carcinoma (epNEC) diagnostic and multidisciplinary workflows. This video addresses unmet needs in epNEC, timing of biomarker assessment, tissue stewardship, clinical trial identification, pathology-oncology communication, and the pathologist’s evolving role in supporting biomarker-directed care.

Transcript

David Klimstra, MD: I'm Dr David Klimstra. I'm a pathologist, professor of pathology in the Department of Pathology at Yale University School of Medicine.

There are many opportunities from both a diagnostic perspective and from a clinical treatment perspective to advance biomarker-directed care in extrapulmonary neuroendocrine carcinomas, or epNECs. The treatment of epNECs is really modeled on small cell lung cancer. And really, there are very few data to suggest that the treatment of other types of neuroendocrine carcinomas should be exactly the same. And so we've extrapolated from small cell lung cancer to employ platinum-based chemotherapy for all neuroendocrine carcinomas. And the problem is that we really don't have a clear biomarker to show which patients will respond well to that treatment and which will not. And so the opportunity to target treatment based on a specific biomarker is huge in neuroendocrine carcinomas because so far we don't have that.

The argument has always been that small cell neuroendocrine carcinoma, whether it's arising in the lung or arising elsewhere, is a homogeneous disease because it shares genomic features, namely the frequent mutation of the retinoblastoma gene. But outside of small cell carcinoma, extrapulmonary neuroendocrine carcinomas are a mixed bag. There really are no biomarkers to adequately distinguish them. They've been classified in a variety of different ways. And so having the opportunity to provide treatment based on the expression of a protein, for instance, DLL3, that can be assessed by pathologists is an enormous step forward. Having the opportunity to direct treatment at that protein is an enormous step forward.

Another important consideration, especially when the tissue sample is a biopsy, is to ensure that it is preserved for biomarker testing. Sometimes pathologists undertake a range of different diagnostic studies, which are important, no doubt, but there's a potential to deplete critical tissue that will no longer be available for the test that will truly determine the treatment the patient will receive. So, for instance, if immunohistochemistry is being performed to establish the diagnosis, it may be reasonable to reserve some sections for DLL3 immunohistochemistry to avoid the need to go back to the paraffin block and potentially deplete it.

Another important point is the multidisciplinary approach. Establish a diagnosis, but when you're talking about biomarkers, this is something that should be decided in a multidisciplinary fashion. The oncologists know what data they need to make treatment decisions, what trials a patient may be suitable for, and they can ensure that the appropriate studies are done by working together with the pathologists who are in the position to conduct those studies. So back-and-forth discussion about how to manage these types of cancers is really critical.

So there are a number of places where DLL3 assessment can be useful in evaluating the diagnostic features of epNECs and in determining whether treatment based on DLL3 expression is appropriate. Importantly, DLL3 is not a neuroendocrine marker. It's important to first classify these carcinomas, to distinguish them from well-differentiated neuroendocrine tumors, which can be high-grade like neuroendocrine carcinomas, but have very different genetic and clinical features. It's important to establish the neuroendocrine differentiation using well-accepted neuroendocrine markers. Currently, the WHO suggests that we employ at least 2 markers among the three that are commonly used, being chromogranin A, synaptophysin, and INSM1.

And so that must be done first. The tumor should be graded based on the Ki-67 proliferation index of the mitotic rate. And an evaluation of the origin, if possible, should be undertaken based both on pathology findings and also on clinical and radiographic features. Once it's established that you are dealing with a neuroendocrine carcinoma, then assessment of DLL3 is appropriate. And this can be done either at the time of first diagnosis from a biopsy. It can be done at the time of tumor progression. It can be done on a resection specimen. Since the method of detection is immunohistochemistry, the technique can be applied to even small samples with relatively minimal neoplastic cellularity.

The other point is that DLL3 assessment can be done locally or it can be done in a centralized laboratory. For patients who are being treated locally on clinical protocols, if there is a significant interest in the oncology group, it may be worthwhile for pathologists to obtain the antibody, characterize it, and undertake the staining locally. If instead it's an unusual occurrence in the practice, then these studies can be sent to a reference laboratory. And for patients enrolled in a clinical trial, it may become necessary for a central lab to perform the studies.

So there are a number of things that pathologists and oncologists can be doing now to get ready for a more biomarker-informed approach to epNEC. For one thing, it's very important for pathologists to standardize how these carcinomas are classified. They need to become well-informed about the diagnostic criteria, how to distinguish them from their mimics, the importance of recognizing mixed differentiation when it occurs, the importance of employing the standard immunohistochemical assessment.

And so they must become very familiar with making the diagnosis because we don't employ biomarkers until the diagnosis has been established. It's also important for pathologists to gain familiarity with interpreting DLL3 immunohistochemistry. Trials and different studies have used different thresholds for scoring DLL3 expression.

And it's important for pathologists to have the experience of examining these preparations, understanding the extent of immunolabeling that can occur, and become comfortable with the different scoring systems. The clinical studies that are being conducted now have defined the parameters for DLL3 immunopositivity, but it's also important to record exactly how many cells, what percentage of cells, and how intensely the antibody is expressed when it's positive so that we can, in the future, reanalyze the data in any manner that may be important.

For oncologists, of course, it's important to understand the different DLL3-directed therapeutic strategies. These are evolving, and data are emerging specific to non-small cell carcinomas and extrapulmonary primaries. And I think this is a very important thing to understand as we encounter patients with these uncommon types of carcinoma.

So there's several important takeaways for pathologists to consider when implementing DLL3 biomarker testing. Neoplasm you're studying is indeed a high-grade extrapulmonary neuroendocrine carcinoma. To exclude the mimics and to make that diagnosis first, DLL3 immunohistochemistry is not a neuroendocrine marker. It's only employed after the diagnosis of neuroendocrine carcinoma has been established by conventional methods.

The second important message is to consider biomarker testing, either in response to a clinical request on a specific patient or potentially even as a reflex test, since this therapeutic option has the potential to offer patients a novel and more effective way of treating what is indeed a very aggressive carcinoma.

The third important takeaway is a more general message, which is to ensure you have fluid communication with the oncologist to understand what is their therapeutic approach and how do your diagnoses impact on the treatment they will choose. Pathologists have an enormous role to play in guiding the treatment, whether it's from simply making the diagnosis or from assessing a range of different biomarkers. And only by maintaining open communication with the clinical team can you be sure that you're up to date in all of the testing that would be necessary.

David Klimstra, MD, is the Louis J. M. Zinterhofer Professor of Pathology at Yale University School of Medicine, the strategic director of Digital Pathology and AI, and chief of the Expert Consultation Service. He attended Carleton College in Minnesota and received his MD cum laude from Yale University, where he also trained as a resident and administrative chief resident in anatomic pathology. He received fellowship training in oncologic surgical pathology with Dr Juan Rosai at Memorial Sloan Kettering Cancer Center (MSK), and in 1992 he joined the faculty as an assistant attending pathologist. He held a range of leadership positions at MSK, including fellowship training program director, chief of surgical pathology service, acting chairman of the Pathology Department, and chairman of the Pathology Department and the James Ewing Alumni Chair in Pathology. He was also a professor of pathology and laboratory medicine at Weill Cornell Medicine. 

Dr Klimstra’s research has focused on the pathologic characterization of tumors of the gastrointestinal tract, liver, and pancreas. His studies have correlated specific pathologic features of tumors with their clinical biology and molecular genetics. He has a particular interest in precursors of pancreatic cancer and has characterized the pathologic features of several such entities. He has also studied the classification, grading and pathogenesis of neuroendocrine tumors of the gut and pancreas and introduced novel concepts about neoplastic progression in neuroendocrine tumors that have become a new standard of diagnosis. Recent work has focused on the development and practical utility of artificial intelligence algorithms based on digital scans of pathology slides, which can enable enhanced diagnostic sensitivity, efficiency, and detection of novel digital biomarkers to predict prognosis, genomic alterations, and response to therapy. Dr Klimstra has authored over 475 peer-reviewed articles on these topics, along with 125 review articles and book chapters.