Functional genomics identifies AMPD2 as a new prognostic marker for undifferentiated pleomorphic sarcoma
Soft-tissue sarcomas are rare, heterogeneous and often aggressive mesenchymal cancers. Many of them are associated with poor outcome, in part because biomarkers that can reliably identify high-risk patients are lacking. Studies on sarcomas often are limited by small sample sizes rendering the identification of novel biomarkers difficult when focusing only on individual cohorts. However, the increasing number of publicly available omics data opens inroads to overcome this obstacle.\n\nHere, we combine high-throughput transcriptome analyses, immunohistochemistry, and functional assays to show that high adenosine monophosphate deaminase 2 (AMPD2) is a robust prognostic biomarker for worse patient outcome in undifferentiated pleomorphic sarcoma (UPS). Publicly available gene expression and survival data for UPS from two independent studies, The Cancer Genome Atlas (TCGA) and the CINSARC reference dataset, were subjected to survival association testing. Genes, whose high expression was significantly correlated with worse outcome in both cohorts (overall and metastasis-free survival), were considered as prognostic marker candidates. The best candidate, AMPD2, was validated on protein level in an independent tissue microarray. Analysis of DNA copy-number and matched gene expression data indicated that high AMPD2 expression is significantly correlated with copy-number gains at the AMPD2 locus. Gene-set enrichment analyses of AMPD2 co-expressed genes in both UPS gene expression datasets suggested that highly AMPD2 expressing tumors are enriched in gene signatures involved in tumorigenesis. Consistent with this prediction in primary tumors, knockdown of AMPD2 by RNA interference with pooled siRNAs or a doxycycline-inducible shRNA construct in the UPS cell line FPS-1 markedly inhibited proliferation in vitro and tumorigenicity in vivo.\n\nCollectively, these results provide evidence that AMPD2 may serve as a novel biomarker for outcome prediction in UPS. Our study exemplifies how the integration of available omics data, immunohistochemical analyses, and functional experiments can identify novel biomarkers even in a rare sarcoma, which may serve as a blueprint for biomarker identification for other rare cancers.