Mapping cancer origins

Abstract

Cancer comprises a bewildering assortment of diseases that kill 7.5 million people each year. Poor understanding of cancer's diversity currently thwarts our goal of a cure for every patient, but recent integration of genomic and stem cell technologies promises a route through this impasse.

Figure 1. Cross-species genomics matches cells in developing and adult mouse tissues with human tumor subgroups

Matching of developing mouse tissues with childhood cancers (left). Different cell stages in an embryonic mouse tissue hierarchy are colored according to differentiation stage. Expression profiling segregates these cells according to transcriptome (top-left). Histologically similar but clinically distinct tumors from the corresponding childhood tissue express different transcriptomes driven by different mutations (mut^1–3) (bottom-left). Cross-species genomics matches human tumors with their candidate cells-of-origin in the corresponding developing mouse tissue. As development proceeds (right), the spectrum of normal cell types (top right) and cancers (bottom right) changes. The same approach shown for childhood cancers can match tumors with candidate cells-of-origin in adult tissues.

Figure 2. Modeling cancer heterogeneity

Subgroups of the childhood cancer shown in Figure 1 can be modeled accurately when transcriptome matched normal cell types are challenged with mutations found in the corresponding cancer. These models should prove extremely useful for developing new therapies for specific cancer subgroups.