Pan-cancer association of a centrosome amplification gene expression signature with genomic alterations and clinical outcome

Centrosome amplification (CA) is a common feature of human tumours and a promising target for cancer therapy. However, CA’s pan-cancer prevalence, molecular role in tumourigenesis and therapeutic value in the clinical setting are still largely unexplored. Here, we used a transcriptomic signature (CA20) to characterise the landscape of CA-associated gene expression in 9,721 tumours from The Cancer Genome Atlas (TCGA). CA20 is upregulated in cancer and associated with distinct clinical and molecular features of breast cancer, consistently with our experimental CA quantification in patient samples. Moreover, we show that CA20 upregulation is positively associated with genomic instability, alteration of specific chromosomal arms and C>T mutations, and we propose novel molecular players associated with CA in cancer. Finally, high CA20 is associated with poor prognosis and, by integrating drug sensitivity with drug perturbation profiles in cell lines, we identify candidate compounds for selectively targeting cancer cells exhibiting transcriptomic evidence for CA.Author summary: Centrosome amplification, i.e. an increased number of centrosomes—structures that exist inside cells, is a hallmark of cancer cells and therefore an Achilles' heel for the development of innovative therapies that specifically target tumour cells, sparing healthy ones. To exploit centrosome amplification’s clinical potential, it is crucial to understand its role in cancer development and to identify compounds for its selective targeting. These are challenging tasks due to the technical difficulty of profiling centrosome amplification in cells. In this study, we circumvent those challenges by computationally analysing the expression of 20 genes known to promote centrosome amplification across nearly 10,000 tumours of over 30 cancer types, thereby estimating their relative centrosome amplification levels. We found that those genes are indeed highly active in tumours and associated with prognosis in different cancer types. We also show that those genes’ expression is associated with instability in the structure of cancer cells’ chromosomes and identify candidate drugs for selectively targeting those cells. Our work therefore demonstrates the potential of computational analyses of large volumes of cancer molecular and clinical data to elucidate cellular and molecular mechanisms of tumour development and propose novel therapeutic options in oncology.