Landscape of functional DLL3 expression in gastroenteropancreatic neuroendocrine neoplasms (GEP NENs).

4146 Background: Delta-like ligand 3 (DLL3) is an emerging target in multiple neuroendocrine cancers including small cell lung cancer but remains underexplored in GEP NENs. With the ongoing development of multiple classes of therapeutics against DLL3, there is a need to understand the landscape of functional DLL3 expression in GEP NENs. Methods: DLL3 immunohistochemistry (IHC) was completed on available tumor samples from patients (pts) with GEP poorly differentiated neuroendocrine carcinomas (GEP NECs) and grade 3 well differentiated pancreatic NETs (G3 WD PanNETs) treated between 2018-2024 and a tissue microarray of resected G1-G2 WD PanNETs (185 samples total). DLL3 positivity (+) was defined as ≥ 5% weak (1+) IHC staining. H-scores were calculated by combining % of + tumor cells and degree of staining (1, 2, 3+), ranging from 0-300. Correlations between DLL3 status and clinicopathologic features and outcomes were analyzed. Among selected pts with DLL3 IHC+ GEP NENs, DLL3 immunoPET imaging using the diagnostic tracer [ 89 Zr]Zr-DFO-SC16.56 was completed to evaluate functional expression. Results: Among GEP NECs overall, 50/69 were DLL3+ (72%; median H-score 50, interquartile range [IQR] 0-160, range 0-300), including 13/16 esophagogastric (median 45), 11/13 pancreatic (median 60), 7/11 hepatobiliary (median 120), 16/26 colorectal (median 32.5), and 3/3 NECs of other/unknown origin (median 60), with DLL3 expression higher in small cell vs large cell histology (median 120 vs 15, P = 0.011). Among GEP NECs, there was no association between DLL3+ and individual genomic alterations, PFS to 1L platinum-based therapy (median 4.6 mo vs 4.7 mo in DLL3-negative [-], P = 0.435), or OS from diagnosis of advanced disease (median 15.5 vs 12.2 mo in DLL3-, P = 0.629). Among WD PanNETs, DLL3 expression was detected in 3/46 (7%) G1, 1/23 (4%) G2, and 19/47 (40%) G3 tumors, with median Ki67 higher among DLL3+ vs DLL3- tumors overall (42% vs 6%, P < 0.001) and within G3 WD PanNETs alone (48% vs 30%, P = 0.009). Among pts with advanced G3 WD PanNETs, DLL3+ was associated with shorter OS from diagnosis of advanced G3 disease (median OS 23.1 mo vs 43.9 mo in DLL3-, P = 0.012). [ 89 Zr]Zr-DFO-SC16.56 DLL3 PET imaging was completed on 5 pts with DLL3 IHC+ GEP NENs at progression on standard systemic therapy. Notably, a pt with pancreatic NEC and liver metastases (DLL3 IHC H-score 60) demonstrated high tumoral tracer uptake (SUV max 36.7) with 95% of tumor lesions demonstrating DLL3 PET avidity. Among 4 pts with DLL3 IHC+ G3 WD PanNETs, DLL3 PET was positive in 3/4, with SUV max ranging from 14.4-27.5 and % of DLL3 PET+ tumor lesions ranging from 50-100%. Conclusions: DLL3 is expressed on a majority of GEP NECs and on a minority of well differentiated PanNETs marked by high grade disease and poor clinical outcomes. Functional DLL3 PET imaging highly suggests DLL3 as a promising therapeutic target in both GEP NECs and high grade WD PanNETs.