Approved Research

Analyzing common cancers and immune diseases using multi-trait polygenic risk score subtyping and imputed expression dimensions for the discovery of novel etiological factors.

Lay summary

Cancer is an umbrella term. While all cancers are characterized by an excessive proliferation of cells, cancers are usually considered by body site. However, while each cancer-type has similarities, even within each type there are many differences. These differences, or heterogeneities, are due to the many different causes that exist within and across cancer/s. The presence of multiple causes limits our power to detect these causes.

If we could divide patients with a certain cancer-type into subgroups that reflect different underlying etiologies we would increase our understanding of cancer development and potential treatment. Also, if we could add together subgroups of different cancer-types that have shared causes, we would increase power for discovery. These appear somewhat paradoxical, but there is evidence for both. Evidence for subgroups within cancer-types includes observations of differing tumor markers, survival, and responses to therapies. Evidence for shared causes across cancer-types comes from observations of multiple cancers in families and some cancers responding to similar drugs. Immune response has also been offered as a major player in cancer risk. We will define natural clusters based only on genetics across all people in the UKBiobank.

Once these genetic clusters are defined, we will characterize them by the cancers and immune diseases they contain, as well as other health-related, lifestyle, and environmental factors. We have also developed a way to predict the expression of genes for a person based on their genetics. We will use this method to predict gene expression for all people in the UKBiobank, and compare these predicted measures across our genetic clusters. Last, we will look for additional inherited genetic variants that are different across clusters. Through an improved understanding of these clusters, we hope to increase knowledge about cancer-type subgroups (which are split across clusters) and which different subgroups of cancers were clustered together (shared risk) as well as some of the biological relevance of these using gene expression predictions and novel genetic risk factors.