Cancer's Fingerprint

An analysis of gene expression patterns has revealed a telltale signature for cancer stem cells consistent across many kinds of cancer. The findings, reported in August by HMS researchers, suggest new avenues for cancer diagnostics and therapies. The paper, titled "A gene expression profile of stem cell pluripotentiality and differentiation is conserved across diverse solid and hematopoietic cancers," was published Aug. 21 in the journal Genome Biology. Tumors result from the unchecked growth of cells that, to an extent, resemble the tissue of origin. For example, the cells of a colon tumor are close relatives of normal colon tissue, and the cells of a brain tumor closely resemble normal brain cells, and so on. Just as normal cells are derived from a lineage of stem cells and precursor cells, a major focus of cancer research is identifying “cancer stem cells,” or those cells with the ability to initiate and sustain the growth of a tumor. The therapeutic potential is tremendous: If researchers can identify those cells that cause the tumor to proliferate, clinicians will have a specific target for treatment. No more cancer stem cells, the thinking goes, no more tumor growth. Previous studies have identified signatures specific to a handful of individual tumor types (e.g., melanoma stem cells and breast cancer stem cells). Now, researchers at HMS have identified a set of genes whose expression in normal stem cells is tightly controlled, and shown how this same set of genes is similarly expressed in a wide variety of cancers (blood, brain and many epithelial tumors). "In other words, from the perspective of which genes are being expressed, the cancers always looked like their normal tissue counterparts, plus a fraction of the stem cell 'signature' that we identified," said senior author Isaac Kohane, the Lawrence J. Henderson Professor of Pediatrics at HMS and Boston Children's Hospital, and co-director of the Center for Biomedical Informatics at the Countway Library of Medicine. "This signature may therefore serve as a starting point for therapeutic design." Further, Kohane's team was also able to show that how intensely this stem cell "signature" was expressed in the cancers correlated strongly to clinical grading (i.e., the more "stem-like" the tumor, the higher the stage of the cancer). This latter observation may pave the way to cheap, in-silico assistance for oncologic pathologists. "Specifically, we may be able to use this stemness measure to identify tumors that present as low- or mid-grade under a microscope but are 'under the hood' about to become highly aggressive," said first author Nathan Palmer, a research fellow in Kohane's lab at the Center.