Ganglioside targeting: Exerting sweet revenge on neuroblastoma

Neuroblastoma (NBL) is the most common extracranial solid tumor diagnosed in children and is responsible for 11% of pediatric cancer deaths. Dinutuximab is an FDA approved monoclonal antibody (mAb) against GD2, used in the treatment of high-risk NBL. Despite integration of dinutuximab into upfront treatment protocols, 40-50% of patients still relapse and eventually die of their disease. Resistance to anti-GD2 therapy is poorly understood. Previously, we have shown GD2 loss in xenografts in response to anti-GD2 immunotherapy. This resistance mechanism is driven by downregulation of one enzyme, ST8SIA1 while other enzymes in the GD2 pathway remain unchanged. We hypothesized that loss of ST8SIA1 leads to reduced GD2 and a compensatory increase in GM2. In the present work, we validate our hypothesis, vetting GM2 as an immunotherapy target for which we engineered GM2-specific chimeric antigen receptors (CAR) and mAbs and tested their efficacy and safety in murine models.First, we tested anti-GM2 mAbs as a monotherapy in 3 metastatic NBL xenograft models. Regardless of initial efficacy, monotherapy fail to provide long-lasting responses and eventually all mice relapsed. However, because each antibody showed partial activity in the GD2-high model, we added a combination-therapy arm in the next experiment. The combination proved to be superior, as all mice receiving it saw tumor eradication and long-term survival.We then tested anti-GM2 CAR T cells, based on the same binder used for mAbs, and saw robust and durable responses. Dual targeting GD2 and GM2, with a tandem GM2-GD2 CAR, was able to eliminate the tumors completely and provided long-term survival.Ganglioside distribution in normal tissue is poorly understood. However, it is known that GD2 is present in peripheral nerves, where it causes neuropathic pain, the main dose-limiting toxicity. To anticipate potential toxicities the combination could present, we optimized ganglioside detection by IHC on frozen tissue and successfully screened healthy tissue microarrays (TMAs). GM2 showed limited tissue distribution, being restricted to the kidneys and thyroid. More importantly, no overlap between GD2 and GM2 was observed.We also screened murine tissues from in search of a safety model. The kidneys of C57Bl/6 mice showed GM2 levels comparable to human kidneys, so were used for toxicity studies. GM2 mAbs were well tolerated with no evidence of off-target toxicity, no weight loss or noticeable pathological features.Finally, courtesy of NANT, we procured viably frozen bone marrow samples from relapsed NBL patients, previously treated with anti-GD2. These samples showed heterogeneous GD2, yet high GM2.Together these data shows, GD2 loss driven by immunotherapy and a concomitant increase in GM2. More importantly, it validates GM2 as an alternative target for GD2 low NBL and because GM2 has limited normal-tissue expression, it can be safely targeted alongside GD2 to prevent antigen escape