(P 03) EVOLUTION OF CLINICALLY ACTIONABLE VARIANTS IN ADVANCED PEDIATRIC AND AYA BONE SARCOMAS

Objective: Few pediatric and AYA sarcoma studies have investigated how gene alterations detected by targeted cancer panels change in the same patient across multiple tumor samples over time. Our study leverages targeted somatic DNA next-generation sequencing (NGS) data from a large cohort of young patients with osteosarcoma (OS) or Ewing sarcoma (EWS) to identify changes in oncogenic variants and variants potentially predicting response to matched molecularly targeted therapy (MTT) with the aim of addressing the clinical utility of biopsy and resequencing pediatric and AYA bone sarcomas at recurrence.

Methods: The iCat2/GAIN (Genomic Assessment Informs Novel Therapy) Consortium Study is a multicenter cohort study involving 13 US cancer institutions evaluating outcomes after receipt of MTT in young patients with advanced extra-cranial solid tumors. Tumors are sequenced using OncoPanel, a targeted DNA NGS test performed in a CLIA/CAP certified laboratory assessing single nucleotide variants (SNV), copy number variants (CNV) and structural variants (SV) in up to 447 cancer genes and rearrangements in up to 60 cancer genes. In 1 case, a similar targeted NGS tests was performed. Multiple tumor samples from the same patient at different timepoints (initial diagnosis or relapse) and tumor sites (primary or metastatic) are sequenced when available.

We selected OS or EWS patients with at least 2 successful targeted NGS panels. We assessed changes in pathogenic variants in genes that are either: 1) known to be biologically important in OS or EWS (oncogenic variants), or 2) gene alterations which may predict response to matched MTT (actionable variants). We categorized patterns of change as emergence, disappearance, or persistence. Pattern of change was variant-based rather than patient-based, so an individual patient could have more than one pattern present.

Results: Of 784 iCat2/GAIN patients there were 155 patients with OS and 107 with EWS. The analytic cohort consisted of 27 OS patients and 19 EWS patients with at least 2 tumor sequenced samples. The median number of sequenced samples per patient was 2 (range 2 to 6).

In EWS, 10 (53%) patients had an oncogenic or actionable variant in at least one sample. The predominant pattern of change in EWS was emergence of oncogenic and actionable variants occurring in 9/10 patients. Of the 19 EWS patients with >2 samples, 26% had emergence of a new pathogenic SNV in STAG2 or TP53 and 21% had emergence of new actionable variants in other genes (in MAP2K, FLCN, FGFR1, MYC, and CDKN2A). Persistence of oncogenic and actionable variants was also seen in 26% of patients. Disappearance of oncogenic or actionable variants was only seen in one case (in CCND1).

In OS, all 27 patients had at least one oncogenic or actionable variant in at least one sample. The predominant pattern of change in OS was persistence of oncogenic or actionable variants occurring in 22/27 patients, most commonly in TP53 (n=17), CDK4 (n=4), RB1 (n=4), CCND3 (n=3), and MCL1 (n=3). Emergence of new oncogenic or actionable variants occurred in 14/27 (52%) patients, with emergence of new TP53 mutations being the most common (n=8). There was also a high rate of disappearance of variants [12/27 (44%) patients], with disappearance of MYC amplification (n=6) and TP53 mutations (n=6) being the most common.

Conclusion: In EWS we observed a predominant pattern of emergence of new biological and/or actionable variants over time with emergence of new STAG2 or TP53 mutations occurring in 26% of patients, confirming the importance of these genes in resistant disease. In OS, we observed mixed patterns of change including persistence, emergence and disappearance of variants, suggesting the previously described unstable genome in OS impacts key oncogenes and potentially targetable genes. Given the observed patterns of genomic change in OS and EWS, there may be utility to repeat sequencing of relapsed tumor biopsies.