In the 1990s, Carlo Croce, then director of the Kimmel Cancer Center in Philadelphia, Pennsylvania, was hunting for genes involved in chronic lymphocytic leukaemia. The disease was consistently associated with a lesion in chromosome 13, and so, back before the human genome was sequenced, his lab determined the identity of the nucleotides in an 800-kilobase stretch from the deleted region and began searching for protein-coding genes. “We failed, for six years,” recalls Croce, now director of the Human Cancer Genetics Program at the Ohio State University in Columbus. His lead graduate student left science to go to business school.

Rosetta Genomics is using microarrays and PCR to identify microRNA biomarkers and develop tests. Credit: ROSETTA GENOMICS

After another false start and a lucky conversation, Croce obtained cells from a patient who had leukaemia involving a very small translocation, only about 30 kilobases. Croce examined this region thoroughly enough to convince himself that it contained no genes. Then he read about microRNAs, which had just been discovered in mice. Further experiments quickly revealed that the region encoded two microRNAs, which, in 2002, were the first to be implicated in disease4.

A few years later, researchers led by Todd Golub at the Broad Institute in Cambridge, Massachusetts, examined RNA molecules in tumours and reported that using just a small number of microRNAs, about 200, provided a better classification of tumours by type and source than using 16,000 messenger RNAs5. There's great potential, says Croce. “There is no doubt in my mind that microRNA can be used for diagnostic and prognostic purposes.”

Indeed, although insurance companies will not yet pay for them, at least two companies offer services that involve testing for microRNAs in biopsies from patients with cancer: Rosetta Genomics in Rehovot, Israel, and Asuragen in Austin, Texas.

Such tests are possible because microRNAs are surprisingly stable both in the body and in paraffin blocks, says Muneesh Tewari, a researcher at the Fred Hutchinson Cancer Research Center in Seattle, Washington, and lead author on one of the first papers showing that microRNA can be extracted from plasma and serum6. “The hard part of this,” he says, “is working with small quantities of starting RNAs and applying this to the technologies” that can identify them. Once extracted, known microRNAs can be identified by using microarrays and quantitative PCR; both known and unknown microRNAs can be identified by sequencing.

But finding microRNAs in samples from patients is only the first step to identifying which microRNAs carry information about disease, Tewari says. “Very little is known about the variation of microRNAs.” The results of experiments can vary for reasons besides disease, says Jun Lu, a genetics researcher at Yale University in New Haven, Connecticut, and first author on the paper with Golub5. But, he says, “if it's a strong discriminator, then you should get the same answer no matter what platform you're using”.

M.B.