Evaluating GD2-targeting antibody-drug conjugates for DIPG using human organoid-based preclinical platforms

Diffuse intrinsic pontine glioma (DIPG) is a universally fatal pediatric brain tumor with a median survival of less than 12 months. Although the recent FDA accelerated approval of ONC201 offers a long-awaited sense of progress for the DIPG/DMG community, the central therapeutic problem remains far from solved. The frequent presence of H3K27M mutations in DIPG is associated with elevated expression of GD2, a clinically actionable surface antigen. GD2-directed CAR-T therapies have shown encouraging early activity in DIPG. However, challenges such as antigen heterogeneity, inherent manufacturing lead time, and risks of neurotoxicity remain. Antibody-drug conjugates (ADCs) offer a more controllable, off-the-shelf therapeutic approach that could overcome these limitations, yet their potential in DIPG remains unexplored. To evaluate the efficacy, resistance, and safety of GD2-targeting ADCs, we employed two complementary human-derived organoid platforms. First, DIPG models were generated using NanoGlio, a high-throughput nanoliter-volume droplet 3D organoid system. Second, we established a co-culture model with DIPG NanoGlio and iPSC-derived brain organoids to assess neurotoxicity. We successfully generated NanoGlio derived from 9 DIPG models reflecting the key genetic alterations typically identified in DIPG, including mutations in H3.3 (H3F3A) or H3.1 (HIST1H3B). We assessed the expression of GD2 by flow cytometry and 3D-Fluorescence Antibody Cell Tracking (3D-FACT) and observed widely variable GD2 expression across DIPG models. We tested M3554, a first-in-class clinical-stage anti-GD2 ADC conjugated to exatecan, a topoisomerase I inhibitor, in these models. M3554 exhibited potent cytotoxicity across DIPG NanoGlios. Resistance profiling using its unconjugated payload, Exatecan, revealed that reduced ADC response was primarily driven by intrinsic payload resistance rather than low GD2 expression. DIPG models with very low GD2 expression, including SF7761 and SU-DIPG-IV, also responded strongly to M3554, consistent with a bystander killing effect. iPSC-derived brain organoids (GD2-negative) showed minimal toxicity with M3554, whereas equivalent doses of Exatecan induced extensive cell death. Future work will systematically evaluate GD2 ADC bystander effect via co-encapsulation of GD2+ and GD2- DIPG cells, and assess the efficacy and neurotoxicity of M3554 using our DIPG-brain co-culture model. Successful completion of this study will provide critical preclinical evidence supporting the assessment and use of GD2-targeting ADCs in DIPG treatment.