Clues from Outside

+ See all authors and affiliations

Science  05 Dec 2008:
Vol. 322, Issue 5907, pp. 1439
DOI: 10.1126/science.322.5907.1439c

Tumors that appear clinically related can respond quite differently to treatment and radically alter the outcome for the patient. Gene expression profiling on tumors is thus useful for detecting differences that can help to improve diagnosis and prognosis and predict patient response to treatment. It has already been used successfully in the clinic, particularly for patients with breast cancer. Hepatocellular carcinoma is often caused by hepatitis infection and liver cirrhosis. It is a major cause of cancer death worldwide, but disease recurrence has so far proven difficult to predict. Expression profiling has been limited by a requirement for frozen tissues, as many specimens have been, and are still being, formalin-fixed. Hoshida et al. developed a method for accurately profiling the expression of 6000 genes from formalin-fixed and paraffin-embedded samples, some of which were 24 years old. They found no association between the gene expression profiles of hepatocellular carcinoma tumors and prognosis. Instead, they found an expression signature from the surrounding non-tumoral liver tissue that could predict the late recurrence of tumors. The authors suggest that this survival signature indicates the state of the liver and how likely it is to become malignant, which may be determined by a prior event such as viral exposure. Profiling the surrounding tissue, rather than the actual tumor as is customary, may prove to be a useful tool for the treatment of other cancers. — HP*

N. Engl. J. Med. 359, 1995 (2008).

  • * Helen Pickersgill is a locum editor in Science's editorial department.

Navigate This Article