Figure_2_UKRI.jpg

Big picture

The first genetic steps towards cancer occur in healthy tissue decades before cancer onset. This raises the possibility that these early events could be used as a bellwether for predicting who is most at risk of developing cancer. However, at present, we are unable to reliably identify which mutant clones will progress to lethal cancers.

Serial blood samples collected annually from hundreds of thousands of healthy people give us a superpower to better predict which clones are the bad actors. By “zooming in” on the people who develop cancer, we can “rewind” time by analysing blood samples collected years before the cancer was diagnosed. By comparing these signals to those measured in healthy individuals, we aim to understand how many years before traditional diagnosis cancer be predicted.

To achieve this objective, we combine lineage-tracing experiments, longitudinal sequencing data and mathematical modelling approaches in projects outlined below.


Clonal hematopoiesis and Cancer risk Tracked over 20 years

edited.jpg

We make ~5,000,000 new blood cells each second. To achieve this, haematopoietic stem- and progenitor-cells (HSPCs) must constantly divide to replenish the cells we lose. Throughout our lifetimes, therefore, our populations of HSPCs evolve: accumulating genetic alterations, some of which cause clonal expansions. Certain expansing clones are known to increase your risk of developing a subsequent blood cancer, and hence are thought to be the earliest events in this process. We are studying this by performing ultra-deep sequencing on longitudinal blood samples collected over decade-long timescales to quantitatively understand the earliest stages of cancer evolution.

People: Caroline Watson & Jonathon Cordova

(collaborators: Doug Easton, Usha Menon, Daniel Fisher, Todd Druley)


Mathematical Modelling of clonal dynamics in tissues

The fate of a new mutation depends on the evolutionary forces of mutation, genetic drift and selection. In both normal and cancerous tissue evolution, the relative roles of each of these forces remains controversial. How much of the somatic evolution we observe is due to chance or effect? This project aims to address these questions by making quantitative predictions of mutation frequencies under different evolutionary scenarios and comparing these predictions to available data. These models will provide a rational basis for detecting abnormal clones.

People: Gladys Poon


in-situ DNA barcoding of tissues

Portable Network Graphics image-1F03337387E5-1.png

DNA barcodes are short (~30bp) sequences of DNA which are stably integrated into a cell’s genome and which serve as “name-tags”, allowing one to track millions of cell lineages through time by counting each tag using next-generation sequencing. In this project we aim to develop an in-situ barcoding technology for mammalian cells based on an existing genetic technology developed in yeast. Because DNA-based lineage tracking is both quantitative and high-throughput, the development of this tool in mammalian cells has the potential to shed light on many questions in cancer and stem cell biology. This project is a collaboration with the lab of Sasha Levy.  

People: Jinqi Fu, Adriana Fonseca


TCR repertoire profiling in Ovarian cancer

Immune cells (including T- and B-cells) do not share the same DNA sequence as the rest of our cells. They shuffle a specific part of their genome to generate a diverse set of sequences that are used to recognise foreign “antigens”.

An individual’s T-cell receptor (TCR) repertoire (i.e. the set of all their T-cell sequences) therefore records information about what antigens an individual has been exposed to throughout their life – including infectious diseases and cancers. As tumours develop, neo-antigens on the surface of the tumour cells are recognised by T-cells, causing changes in the TCR repertoire. This cancer-specific signal may be detectable years before traditional diagnosis. We are analysing TCR repertoires in 100s of Ovarian cancer-cases and 100s of controls to determine whether statistical differences exist that can distinguish between cases and controls and how early these differences arise.

People: Jinqi Fu, Gladys Poon

Collaborators: Doug Easton, Harlan Robins