Genome Scale Prediction of Functional Target Networks Uncovers Candidate Vulnerabilities in Sky-92 High Risk Multiple Myeloma

Introduction

High risk multiple myeloma (HR-MM) continues to result in unacceptable overall survival rates for MM patients, despite unprecedented advances in therapy. The definition of HR-MM is evolving in an effort to focus on patients with the most aggressive forms of MM and trial design is pivoting toward HR-MM risk-adapted randomisation in an effort to personalise treatments. SKY-92 is a 92-gene expression classifier that enables rapid powerful prognostication and stratification of risk in MM. Mapping gene regulatory networks enables a deeper understanding of the molecular mechanisms underpinning HR-MM and can uncover specific co-regulated modules that enable novel target identification. NetNC¹ is an algorithm for genome-scale prediction of functional transcription factor target genes. We investigated key pathways and potential targets in SKY92 HR-MM patients from the MMRF COMPASS dataset² using this network biology approach (NetNC¹).

Methods

763 patients from the MMRF CoMMpass dataset² were stratified into SKY-92-HR and SKY-92-SR risk groups based on gene expression profiling. NetNC¹, an algorithm that can predict groups of genes driving disease biology, was applied to SKY92-HR and SR groups to derive networks from genes with high or low relative expression according to ranked expression values. With NetNC², we identified functionally coherent gene networks for 500 High-Risk and 500 Standard-Risk genes, using the HumanNet gene network as a scaffold. Networks were visualised using Cytoscape and annotated with the BiNGO plugin.

Results

Out of 47,084 genes with nonzero total read count, 10720 were upregulated in SKY92-HR-MM patients (23% of all genes) and 7617 (16%) genes were downregulated (adjusted p-value < 0.05). NetNC discovered a total of eleven active networks for SKY-92-HR disease biology, with clearly distinct biological processes in-play in HR-MM vs SR-MM patients. Functional gene networks that are specific to HR-MM include regulation of cyclase activity (e.g. SSTR1), mitotic regulation (e.g. HUNK), myosin morphogenesis (MYH family members), MAGE T-cell regulators, GAGE family members, extracellular matrix proteins, GABA receptor family regulators, cellular junction proteins, cadherin proteins, ion channel regulators and DNA/RNA regulatory networks (e.g.FOXG1). Deep module analysis identifies candidate Achilles’ heel target genes that act as high dependency nodes for MM cell survival, proliferation and drug resistance.

Conclusion

Genome scale prediction of altered gene expression and co-regulated networks defines distinct processes of biological significance in HR-MM. This enables identification of candidate Achilles’ heel targets for drug development specific to this challenging patient group. This analysis identifies in particular critical cellular junction proteins, chromosomal stability regulators and extracellular proteins that are distinctly regulated in HR-MM and point to differential cellular functions that enable HR-MM therapy resistance and escape. Further work is needed to identify targeted therapies directed to these vulnerable cell processes.

References

1. Overton et al. Functional Transcription Factor Target Networks Illuminate Control of Epithelial Remodelling Cancers (2020) 12(10): 2823
2. Keats et al. Interim Analysis Of The MMRF CoMMpass Trial, a Longitudinal Study In Multiple Myeloma Relating Clinical Outcomes To Genomic and Immunophenotypic Profiles. Blood(2013) 122 (21): 532.