Transcriptome Profiling in Multiple Tumour Types in Adult AML

The following case study features published data from a poster presented at AACR.

Study Aim 

The aim of this study was to exploit the publicly available datasets of The Cancer Genome Atlas (TGCA), PanCanAtlas, and Cancer Cell Line Encyclopaedia (CCLE) in order to extend transcriptomic and prognostic profiling of specific genes of interest in 33 tumour types across over 10,000 cancer patient samples and cell lines. 

Acute myeloid leukaemia (AML) can affect both adults and children and impairs the normal functioning of bone marrow, impacting upon myeloblasts, red blood cells and platelets. E-selectin is a cell adhesion glycoprotein which is expressed on endothelial cells and promotes environmental–mediated drug resistance, metastasis and confers poor clinical outcome in multiple cancers. 

Recent clinical data has demonstrated a correlation between AML cell surface E–selectin ligands and response to uproleselan, a specific E-selectin inhibitor, which can be used in combination with chemotherapy to treat patients with AML. As an E-selectin antagonist, uproleselan increases susceptibility to cytotoxic therapies. 

Paediatric AML tumours express two genes involved in the glycan synthesis of E–selectin ligands at high levels: FUT7 and ST3GAL4. Both genes are associated with poor outcome and cell surface E–selectin ligand expression. This study therefore sought to map expression of these genes across multiple adult cancer tumour types, including AML, as well as to determine their prognostic potential in a subset of AML patients with particularly poor survival.

Analysis and Results

The analysis was completed by Fios using the TCGA PanCanAtlas dataset encompassing 10,258 patient samples across 33 different tumour types, and in the CCLE dataset encompassing 1,457 cell line samples. Expression levels of two genes, ST3GAL4 and FUT7, were evaluated. ST3GAL4 and FUT7 were consistently expressed in the majority of cancers in the PanCanAtlas dataset, while augmented expression of FUT7 and ST3GAL4 was observed in an analysis of 39 AML cell lines in the CCLE dataset.  

ST3GAL4 expression was greatest in melanoma, kidney chromophobe, adrenocortical carcinoma and bladder urothelial carcinoma, while FUT7 expression was greatest in AML, diffuse large B-cell lymphoma, thymoma testicular germ cell tumours and head and neck squamous cell carcinoma. Adult AML had the highest expression of FUT7 (mean log2 gene expression = 8.1) and high expression of ST3GAL4 (mean log2 gene expression = 9.4).   

Prognostic significance of ST3GAL4 and FUT7 in adult AML was further assessed using the TCGA LAML RNAseq dataset. The TCGA LAML RNAseq dataset contained 151 RNAseq profiles of bone marrow samples from adult AML patients and the status of the FMS-like tyrosine kinase 3 (FLT3) proto-oncogene was considered. Mutations in FLT3 are associated with higher risk of relapse and shorter overall survival. The expression of both ST3GAL4 and FUT7 was upregulated in mutated FLT3 samples (fold change in expression of 1.73 and 1.40, respectively). In samples with FLT3 internal tandem duplication, FUT7 expression was significantly associated with poor prognosis and decreased overall survival (HR=0.223, p=0.015).  

Conclusions  

Interrogation of publicly available datasets can help to further evaluate therapeutic targets. ST3GAL4 and FUT7 may both serve as predictive biomarkers in adult AML, while there is evidence to suggest that E-Selectin inhibition with uproleselan should be investigated as a potential treatment across multiple tumour types.  

View the full poster on the AACR website

View more case studies and publications.