Low GD2 expression as a mechanism of resistance to chemoimmunotherapy with dinutuximab
Introduction
Chemoimmunotherapy with anti-GD2 antibodies is used for treating progressive disease (PD) neuroblastoma and is being evaluated with induction chemotherapy. While most neuroblastomas express GD2, low or negative GD2-expressing neuroblastomas occur and may not benefit from anti-GD2 therapy. We sought to determine the relationship between GD2 expression and response to chemoimmunotherapy in patient-derived xenografts (PDXs) and to determine the frequency of low GD2 expression in neuroblastoma patients.
Methods
We quantified dinutuximab binding to neuroblastoma cells in marrow by multi-color flow cytometry, gating out CD45+ cells and gating on cells positive for the non-GD2 neuroblastoma antibody HSAN (binds to 100% of neuroblastomas) to define % positive and median fluorescence intensity (MFI). We assessed 93 bone marrow (68 pretherapy, 25 at PD), 9 tumor, and 12 blood samples from neuroblastoma patients on Children’s Oncology Group protocol ANBL00B1. Event-free survival (EFS) was assessed for subcutaneous PDXs in nu/nu mice treated with 15 mg/kg temozolomide (TMZ) + irinotecan 7.5 mg/kg (IRN) (days 1-5 and 21-25) +/- 15 mg/kg dinutuximab (days 2, 4, 22, 24).
Results
Based on % positive and intensity of dinutuximab binding we defined four groups of patient samples: high, intermediate, low, and very low dinutuximab binding. GD2 expression was very low in 3% pretherapy and 11% PD patient samples, and low in 20% pretherapy and 39% PD samples. We assessed activity of TMZ/IRN +/- dinutuximab in neuroblastoma PDXs (3 PD, 1 established pretherapy). In COG-N-480x (pre-therapy PDX, high GD2 expression) dinutuximab increased 300-day EFS to 100% vs 20% with TMZ/IRN (p=0.04). In COG-N-452x (intermediate GD2 expression) dinutuximab increased 300-day EFS to 50% vs. 0% for TMZ/IRN (p=0.01). Dinutuximab did not enhance TMZ/IRN activity in COG-N-519x (very low GD2) or Felix PDX (low GD2).
Conclusion
Low GD2 expression occurs in neuroblastoma. Neuroblastoma PDXs in nu/nu mice provide a preclinical model to assess antibody activity when combined with chemotherapy. In > 10% of patient marrow samples analyzed, dinutuximab binding was comparable to levels seen in PDX’s that did not benefit from dinutuximab. Quantifying GD2 expression by flow cytometry is a potential biomarker of activity for dinutuximab in patients treated with anti-GD2 based chemoimmunotherapy.