A new single-cell sequencing to validate new cancer cell types

Cancer is a disease that stems from the disruption of cellular state. Through genetic perturbations, tumor cells attain cellular states that give them proliferative advantage over the surrounding normal tissue. The inherent variability of this process has hampered efforts to find highly effective common therapies, thereby ushering the need for precision medicine. The scale of single-cell experiments is poised to revolutionize personalized medicine by effective characterization of the complete heterogeneity within a tumor for each individual patient.

Single cell mRNA analysis of cancer cells is one of the leading edge techniques being used to better understand cancer biology. The data generated can be used to try to disrupt cancers with drugs or work out how cancers arise in the first place. Studying the cancer transcriptomes depends on accurate identification of cancer cells. Therefore, the foundational step of tumour single cell analyses is separating cancer from non-cancer cells. The simplest approach to identifying cancer cells is to use expression of cancer specific marker genes. However, such genes do not always exist and are generally insufficiently precise, especially without corroborating readouts such as cellular morphology. Another approach is to infer the presence of tumour-defining somatic copy-number changes from shifts in average expression. The idea here is that gains or losses of genomic regions will generally increase or decrease the expression level of genes in these regions respectively. Challenges with this approach include smoothing and denoising expression changes, establishing a baseline against which to measure shifts in expression, segmenting the genome, and identifying changes in expression not due to copy-number changes. Despite these challenges, both marker genes and shifts in average expression may accurately identify cancer cells in certain circumstances. However, if there is any novelty or ambiguity in the identity of cancer cells, then these two approaches are inherently fallible as they are both based on expression and not direct evidence that a cell is cancerous, that it carries the somatic cancer genome.

A fundamental step in this process is separating cancer and non-cancer cells, but this isn’t always an easy task. As well as the many types of cancer, there will also be molecular differences between cancer cells of the same type within a single tumor. Currently, the best method of doing this is to measure the average gene expression of cells in the sample, with higher or lower expression used to distinguish cancer cells from healthy cells but this method can lead to false conclusions.

A new method of separating cancer cells from non-cancer cells has been developed by researchers at the Wellcome Sanger Institute, in a boost for those working to better understand cancer biology using single-cell mRNA sequencing. The study, published today in Communications Biology, improves on existing methods to identify which cells in a sample are cancerous and provides crucial data on the microenvironment of tumors. The software is now openly available for researchers around the world to apply to their own data, advancing the effectiveness of single-cell sequencing to understand cancer.

The research team performed whole genome sequencing and single-cell mRNA sequencing on clinical samples. By locating imbalances of alleles in these data, which indicate copy number changes in the genome, the team was able to identify cancer cells more reliably than with previous methods. This approach will primarily be useful for validating new cancer cell types and better understanding the microenvironment of tumor tissue. It is also hopeful the new technique will serve in the near future for the discovery of novel prognostic signatures and of therapeutic strategies that may enhance the implementation of precision medicine and will improve the clinical management of various cancers.