Tumor cell surface targeting of high-grade neuroendocrine carcinomas

High-grade neuroendocrine carcinomas (hgNECs) are aggressive malignancies that arise, most commonly, from the aerodigestive tracts, though, more rarely from other sites. Frustratingly, hgNECs are, initially, exquisitely sensitive to chemotherapy and/or radiation, but these responses are short-lived and inevitable relapses occur. Recent therapeutic developments have focused on hgNEC-associated cell surface antigens which, when targeted with an antibody, can serve as beacons for delivery of cytotoxic or immunologic payloads. T-cell engagers (TCEs) and/or antibody-drug conjugates (ADCs) against hgNEC antigens, including DLL3 and SEZ6, have demonstrated unprecedented responses in relapsed hgNECs. In some cases, such as small cell lung cancer (SCLC), the expression of these antigens is nearly ubiquitous and sensitivity to surface-targeting therapies is contingent primarily on payload. However, in other hgNECs, neuroendocrine (NE) features, including DLL3 and SEZ6 expression, are bimodal, which necessitates strategies for patient selection and alternative antigens for NE-low hgNECs. DLL3 levels have been evaluated by CLIA-validated immunohistochemistry in >340 patients at MDACC, including those diagnosed with rare extrapulmonary hgNECs (i.e., Merkel cell carcinoma, small cell of the breast, thyroid, etc.) for off-label use of tarlatamab (DLL3-targeting TCE). We utilized expression data from public hgNEC cohorts, as well as our own patient and preclinical model cohorts to characterize the relationship between known cell surface target expression (i.e., DLL3, etc.), various established biomarkers, including NE status and SLFN11, and sensitivity to surface targeting therapies. NE status defined two major subsets of hgNECs. NE-high specimens expressed high DLL3 and SEZ6 and were responsive to therapies targeting these antigens. In contrast, the NE-low subsets expressed high TROP2 and HER2. Surprisingly, target expression was not the dominant predictive biomarker for ADCs targeting these antigens as, instead, biomarkers of payload sensitivity (e.g. SLFN11 for topoisomerase I inhibitor payloads, P=0.01) offered superior insight into efficacy. Surface-targeting therapies represent a new paradigm for hgNEC therapeutics and while DLL3- and SEZ6-targeting therapies are showing promise across all hgNECs, their impact will be limited outside SCLC without patient selection. Utilizing biomarkers, including those for target expression, inflammatory potential, and payload sensitivity, will be critical to optimize drug selection for patients with these malignancies.