(P 427) MULTI-OMIC AND MULTI-REGION PROFILING OF DESMOID TUMORS REVEALS INTRA- AND INTER-TUMOR HETEROGENEITY

Objective: Desmoid tumors are bland fibroblastic tumors with little histologic variation between different regions of the tumor. While desmoid tumors do not metastasize, they have a high rate of local recurrence and no reliable predictors of clinical behavior exist. Previously published studies proposed two different transcriptomic signatures to predict relapse (PMIDs: 25878329, 33444924). However, these two transcriptomic signatures had no overlapping genes or represented pathways, and neither of them have been introduced into routine clinical practice. In this study, we sought to develop a new prognostic signature for desmoid tumors and validate the previously published signatures. This led us to a comprehensive multi-omic investigation of the possible molecular intra- and inter-tumor heterogeneity of desmoid tumors.

Methods: We performed transcriptomic profiling of 31 specimens from 20 primary desmoid tumors to identify genes associated with disease recurrence. Among these 20 patients, 13 patients developed local recurrence after surgical resection of the primary tumor. Next, we performed an in-depth multi-omic analysis including DNA methylation (EPIC microarrays, Illumina), DNA copy number alterations (OncoScan FFPE assay, Affymetrix), point mutations (whole exome sequencing, Roche) and gene expression (whole transcriptome RNA-seq, Illumina) on 24 specimens from different regions of primary and recurrent desmoid tumors obtained from 3 patients (7-9 specimens per patient).

Results: Through differential expression analysis, we identified 14 genes associated with disease recurrence (log2 fold change >2 or <-2, adjusted p < 0.05). However, there was no overlap between our 14-gene signature and the two previously reported transcriptomic prognostic signatures, in terms of individual genes or enriched pathways. Furthermore, significant differences in expression levels of individual genes were identified between different areas from the same surgical specimen. Gene expression signatures associated with favorable and unfavorable outcome were detected in different regions within the same tumor. Further multi-omic studies showed a remarkable intra- and inter-tumor heterogeneity of genomic, epigenomic and transcriptomic patterns. Through reconstruction of phylogenetic trees based on single nucleotide variants, we identified novel clonal and subclonal somatic mutations in desmoid tumors. In one of the patients, we detected two co-existing mutations in the CTNNB1 gene (T41A and S45P) in different regions of the same recurrent tumor. Among the four types of data, the transcriptomic profiles showed the highest degree of variability within tumors and between primary and recurrent tumors from the same patient. We also observed a significant enrichment of different molecular pathways between distinct regions of the same tumor. DNA methylation profiles showed substantial differences between the primary and recurrent tumors, but little intra-tumor heterogeneity within individual tumors.

Conclusion: This study shows a remarkable and unexpected degree of intra- and inter-tumor heterogeneity in desmoid tumors. The presence of molecular heterogeneity has been well established in cancer and high-grade sarcomas, and here we report it for the first time also within histologically bland lesions such as desmoid tumors. Our analysis indicates that molecular analysis of a single tumor biopsy may not accurately reflect the extent and variety of molecular alterations in these tumors. Our study also shows that recurrent desmoid tumors may acquire multiple molecular alterations that are not present in the patient-matched primary tumor. Thus, molecular heterogeneity is an important consideration in drug development and validation of prognostic and predictive biomarkers for desmoid tumors. Our findings may indicate that clonal selection and evolution may contribute to the variable clinical behavior of observed in these lesions.