(P 226) ASSESSING PREDICTIVE VALUE OF RNA-SEQ OF B4GALNT1 AND ST8SIA1 AS BIOMARKERS OF GD2 EXPRESSION IN PEDIATRIC SARCOMAS

Objective: GD2 is a tumor-associated ganglioside that is highly expressed in neuroblastoma and is targeted by two FDA-approved agents, dinutuximab and naxitamab. ST8SIA1 and B4GALNT1 are two genes whose protein products are critical to GD2 ganglioside biosynthesis. Recent in vitro work in neuroblastoma and osteosarcoma cell lines has shown correlation of RNA overexpression of ST8SIA1 and B4GALNT1 with cell surface GD2 protein expression via IHC. To date, clinical samples have not been assessed to evaluate ST8SIA1 or B4GALNT1 RNA expression levels nor to correlate it to either GD2 expression or response to dinutuximab or naxitimab. As upfront sequencing of tumors becomes a more wide-spread practice, identifying gene expression patterns that reliably predict self-surface targets is imperative thus the goal of this project is to validate if either of these biomarkers can predict GD2 expression and or response in pediatric cancers.

Methods: RNA sequence data was obtained from specimens obtained during initial biopsy at diagnosis. Tumor specimens from 69 pediatric patients with either neuroblastoma (n=11) or a variety of sarcomas (n=58) underwent whole exome sequencing and RNA-seq between April 2016 to April 2022. Tumor sequencing was provided from a variety of CLIA-approved labs including Tempus, Nantomics, Ashion, or Caris. Given the potential for variability, RNA normalization across sequencing providers was performed using housekeeping genes, GAPDH and ACTB. The average transcripts per million (TPM) for GAPDH and ACTB was determined for each patient in this cohort. B4GALNT1 TPM was then multiplied by the fractional ACTB or GAPDH value. ACTB/GAPDH paired normalization was then performed by adding paired ACTB and GAPDH fractional values and dividing by two. These data were then compared utilizing linear regression to assess correlation of these two RNA biomarkers.

Results: RNA expression of both B4GALNT1 and ST8SIA1 on average was highest in the patients with neuroblastoma as expected. Clustering of neuroblastoma controls was most effective evaluating B4GALNT1 when GAPDH/ACTB were used in combination with TPM. Of the 11 patients with neuroblastoma, 9 received dinutuximab or naxitimab with reduction of Curie score with antibody therapy in 8 of these patients. One patient with neuroblastoma had very low expression of B4GALNT1, but had the highest expression of ST8SIA1 of the cohort. When evaluating the sarcoma samples, malignant peripheral nerve sheath tumors (MPNST, 10.3 +/- 2.6 TPM) had the highest expression levels closely followed by osteosarcoma (5.3 +/- 1.1 TPM) when evaluating TPM from the average normalized B4GALNT1 data. The expression of B4GALNT1in rhabdomyosarcoma varied on histology as alveolar rhabdomyosarcoma had far higher expression (4.9 +/- 1.5 TPM) than embryonal rhabdomyosarcoma (1.4 +/- 0.2 TPM). Ewing sarcoma had the lowest expression of B4GALNT1 all of the sarcoma subtypes (0.9 +/- 0.2 TPM).

Conclusion: RNA expression of B4GALNT1 seemed to correlate and cluster better with known levels of GD2 expression than ST8SIA1 in neuroblastoma patients who historically have very high levels of GD2 expression via IHC. In sarcomas, MPNSTs had expression levels higher than that of even osteosarcoma which has been the center of clinical trials in sarcomas for GD2-targeted therapy. Given the response of bone disease with dinutuximab and naxitinib in pediatric patients with neuroblastoma, it suggests potential utility of treatment in sarcomas outside of neuroblastoma. Correlation with GD2 IHC expression levels in these patient tumor samples and expansion of this sarcoma data cohort with our adult colleagues is currently underway.