The switch of cancer

For many years, scientists have struggled to understand and cure cancer. The study of the genome of multiple tumors has been fundamental to detect recurrent alterations in several types of cancer, and has facilitated their classification and the development of new therapeutic strategies. In particular, high-throughput technologies have been applied in the context of multiple international projects to detect actionable alterations, i.e. genetic changes in the genome of cancer cells that can be used to develop new targeted therapies. These studies have highlighted the heterogeneity of genetic alterations in patients suffering from the same type of cancer, motivating the development of individualized treatments. However, known actionable alterations tend to occur at low frequency, and often a tumor sample has fewer mutations than those seemingly necessary to explain the tumoral process. Thus, there is a need to expand the catalogue of cancer signatures to integrate other molecular alterations for the characterization of individual tumors.

Most of the strategies used in cancer genome projects are based on searching for genetic alterations or changes in the expression of genes. On the other hand, there is more and more evidence that alterations in the splicing regulatory program play an important role in tumor transformation. Splicing is a process by which the long RNA molecule transcribed from the gene in the genome is processed to remove segments called introns, giving rise to an RNA transcript. Alternative splicing provides a mechanism to generate multiple RNA transcripts from the same gene by eliminating introns in different ways. This process is tightly regulated, and is known to give rise to proteins with cell-type specific or opposing functions, or even provide a way to activate or deactivate gene function. This dramatic change between conditions lead by splicing is generally called a splicing switch.

switch_of_cancer_v1

Splicing switches that are not originally present and regulated in cells can induce altered cellular states, leading to disease. Accordingly, the determination of alterations in alternative splicing in tumors can be fundamental for the development of tumor specific molecular targets for prognosis and therapy. However, this analysis is generally hindered by the heterogeneity of tumors of the same origin from different individuals, as well as by the normal variability between individuals. In our group we have developed a new computational method, robust to biological and technical variability, which identifies significant splicing switches across a large number of tumor samples and shows high accuracy on held-out datasets. Moreover, the method is capable of identifying complex alternative splicing changes that cannot be described using standard methodologies. Additionally, the method is independent of parameterizations, which is especially relevant for the analysis of RNA sequencing data from samples from multiple laboratories and technological platforms.

We have applied this method to data from the Cancer Genome Atlas (TCGA) project, which is the NIH-funded branch of the ICGC project. This is the first published large-scale analysis describing the splicing alterations in 9 cancer types using RNA sequencing data from more than 4000 samples. This is possibly the first systematic study of alternative splicing alterations in 9 difference cancer types using so many patient samples. In this work, we have discovered that there exist many splicing switches in patients with the same cancer type that can separate with high accuracy tumor and normal samples, and different types of cancer from each other, providing a predictive signature. In particular, these signatures provide simple rules based on the expression of a few RNA molecules that could allow determining the cancer type from an RNA sample of a new patient. Additionally, we found such set of rules for the triple-negative breast cancer subtype, which is one of the most aggressive subtypes of breast cancer. This new computational method reveals novel signatures of cancer in terms of RNA transcript isoforms specifically expressed in tumors, providing potential novel molecular targets for prognosis and therapy.

Further reading:

Endre Sebestyén, Michał Zawisza, Eduardo Eyras. Detection of recurrent alternative splicing switches in tumor samples reveals novel signatures of cancer. Nucleic Acids Research 2015; doi: 10.1093/nar/gku1392

News at the UPF web site: “Interruptors del càncer” (in Catalan)