Editors' ChoiceCancer

Lumpers and Splitters: A New Molecular Taxonomy for Cancer

See allHide authors and affiliations

Science Translational Medicine  13 Aug 2014:
Vol. 6, Issue 249, pp. 249ec138
DOI: 10.1126/scitranslmed.3010118

Different cancers are usually grouped together on the basis of their tissue of origin. However, it is becoming increasingly clear that cancers originating from different tissues and organs exhibit molecular similarities. With the advent of tools to generate high-throughput “omic” data, a new molecular cancer taxonomy is now possible. Now, Hoadley et al. integrate data from several “omic” platforms—including whole-exome DNA sequences, DNA copy number variation, DNA methylation, and mRNA and protein expression—from 12 different tumor types. They identified 13 different clusters of tumors according to these features. Both the molecular clusters and the tissue of origin both independently predicted patient survival, demonstrating that this new molecular taxonomy may be clinically useful.

Some tumors, such as glioblastoma multiforme and ovarian cancer, showed molecular clusters that were clearly matched to the tissue of origin, whereas other tumors had more complex patterns. In some cases, tissues from different histological cancer types, such as colon cancer and rectal cancer, converged into a single molecular cluster. Four distinct tumor types—including lung squamous carcinoma, head and neck squamous carcinoma, and some bladder cancers—all clustered into a squamous-like molecular subtype despite differences in the tissue of origin. There were substantial molecular similarities between the C2-squamous–like basal breast cancer and ovarian cancer subtypes, including a high frequency of TP53 mutations, enrichment of amplifications of 3q26 and 8q24/cMYC, and copy number losses in chromosomes 4q, 5q, 8p, and 18. On the other hand, one histological cancer type was split into multiple molecular clusters: Bladder cancer showed a very divergent pattern, with samples mapping to three different molecular subtype groups. Interestingly, these molecular subtypes predicted patient survival, with samples in the C2-squamous–like and C1-lung adenocarcinoma-enriched groups correlating with worse overall survival of patients as compared with samples in the C8-bladder cancer group.

This new molecular classification confirms close relationships among the molecular signatures of tumors that originate from different tissues and confirms that tissue of origin continues to be an important factor in predicting patient outcomes.

K. A. Hoadley et al., Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin. Cell 10.1016/j.cell.2014.06.049 (2104). [Full Text]

Stay Connected to Science Translational Medicine

Navigate This Article