(P 389) OSTEOBLASTIC-GENE SIGNATURE TO PREDICT SURVIVAL IN PEDIATRIC OSTEOSARCOMA (OS) AND TO FOLLOW PATIENTS TOWARDS LIQUID BIOPSY: A COLLABORATIVE WORK IN OS2006 PROTOCOL.

Objective: The survival of pediatric osteosarcomas (OS) is stable worldwide for two decades. The management of OS is lacking new approaches to classify closely the patients and adapt thereafter the chemo-treatments. The objectives of our molecular study, using allelotyping, quantitative PCR (qPCR) and droplet digital PCR (ddPCR), were to determine the survival impact of specific genes previously involved in the bone de-differentiation osteosarcomagenesis (e.g., APC, TWIST1 and CMET). This panel was already identified in the large tumor cohort of the French national protocol OS94, where this 3-gene deregulation was significantly associated to a worst prognosis. By the past, looking in mouse model and combining those gene defects, we were able to reproduce osteosarcomas. The objectives are now to go further and understand the 3-gene signature significance throughout the recent French national OS2006 protocol.

Methods: Using tumors (e.g., initial tumor biopsies, the paired-tumor resection after neoadjuvant chemotherapy and on circulating DNA at different treatment steps), we applied retrospectively in this OS2006 protocol cohort (144 patients) the same molecular assessments comprising the targeting of those 3 genes with 3 techniques evaluating rearrangements. Targeting the 3 genes, we performed in parallel allelotyping, copy number variations assessment by real-time qPCR and ddPCR. A preliminary cohort of 28 patients with liquid biopsies at diagnosis, before tumor surgery and at the end of chemotherapy was screened by ddPCR with this 3-gene panel.

Results: In terms of results, we were able to determine frequent rearrangements in the regions containing these 3 genes and a statistically significant correlation was made between the presence of a gene rearrangement and a worst outcome. 71% of the entire OS studied population rearranged on at least one of these 3 genes, 34.4% presenting 2-gene rearrangements and 14.6% presented abnormalities in the 3 genes. All rearrangement in the 3-gene panel was correlated with a worst EFS and OS. Using this 3-gene signature in QPCR analyses at a prognostic point of view, the sensitivity was of 96% and specificity of 85%. The predictivity of worst outcome when presenting a rearrangement in 2 or 3 genes is of 92%. No correlation with chemoresponse for any biomarker was observed. Furthermore, on the tumor resection, the persistence or appearance of new clones was significantly correlated to a higher risk of relapse. This 3-gene signature was also usable for residual disease assessment on plasmatic DNA, and we were able to follow closely osteosarcoma patients and their evolution.

Conclusion: In conclusion, this molecular signature seems to be clearly associated with patient outcome and was able to provide evidence that this approach is a powerful tool to be used in pediatric OS follow up.