NEW YORK (GenomeWeb News) – A gene expression signature may help predict whether an individual’s liver cancer will return, according to new research.
In a paper appearing online today in the New England Journal of Medicine, an international team of researchers did genome-wide expression profiling on fixed tissue samples from more than 300 individuals with hepatocellular carcinoma. By comparing gene expression in tumors and nearby tissues, the researchers were able to uncover a genetic signature in tissues adjacent to tumors that can help predict liver cancer survival. That, in turn, may eventually help guide liver cancer treatment.
“Our findings underscore the potential of genomic signatures to help identify treatments that will be most beneficial to individual patients,” senior author Todd Golub, a Howard Hughes Medical Institute investigator who is also affiliated with the Broad Institute and the Dana-Farber Cancer Institute, said in a statement.
In developed countries liver cancers tend to be detected relatively early, since those at highest risk — for instance, individuals with cirrhosis and hepatitis — are usually screened regularly. Even so, even when liver cancer is detected and treated early, it’s difficult to predict whether it will return years down the road.
Gene expression signatures are one potential avenue for trying to gauge recurrence risk and predict patient survival. But efforts to come up with such signatures have been hindered because standard gene expression tests rely on frozen tissue samples, whereas most patient samples are fixed.
“In most hospitals and clinics, the prevailing method of storing patient tissue involves a chemical fixative, which often precludes future genome-scale analyses,” Golub said. “That means the vast majority of patient samples have effectively been off-limits to a variety of important questions.”
For this paper, the team demonstrated that they could determine gene expression in formalin-fixed, paraffin-embedded samples using a modified version of Illumina’s DASL (DNA-mediated annealing, selection, extension, and ligation) assay, a multiplex, locus-specific PCR assay.
“[W]e modified the DASL method for probe selection and analysis and performed a bioinformatics meta-analysis to identify 6,000 transcripts that captured the majority of variance in gene expression across the human transcriptome,” Golub and his colleagues wrote.
First, the team looked at the gene expression profiles in tumor tissues and adjacent, healthy tissues from 82 Japanese liver cancer patients who had undergone surgery in Tokyo between 1990 and 2001, focusing on the 6,100 most informative genes. They then determined whether certain gene expression profiles were associated with liver cancer recurrence or survival in these patients, who were followed for a median of 7.8 years after surgery.
The team purposely included a large number of patients with very early-stage hepatocellular carcinoma, noting that “these patients represent the greatest clinical challenge with respect to outcome prediction.” Otherwise, both the training and validation groups were heterogeneous.
The researchers noted that this approach successfully measured gene expression in fixed tissues about 90 percent of the time, including tissues collected dozens of years ago.
Although they did not detect a significant association between tumor gene expression and either survival or tumor recurrence, the researchers did find an association between the gene expression in the tissue adjacent to tumors and survival outcomes.
The team found a 186-gene signature that was significantly associated with survival. The expression pattern of this signature suggests that those with normal liver function in tissues adjacent to tumors have the best outcomes. Those with impaired liver function, on the other hand, are most likely to have poor outcomes. The poor-prognosis signature also contained genes linked to inflammation.
The researchers validated this 186-gene signature in 225 American and European liver cancer patients who’d had surgery between 1994 and 2005 at New York’s Mount Sinai School of Medicine, the Hospital Clinic Barcelona, or the National Cancer Institute of Milan and found that the signature could predict survival differences in these patients over a median follow-up time of just over two years.
When they focused on the 168 individuals who were followed longer after surgery (about 2.8 years), the signature was even more reliable.
The team also developed and tested a 132-gene signature that could help predict late recurrence, tumor recurrence that occurs more than two years after a patient’s initial surgery. Early recurring tumors, on the other hand, were better predicted by clinical and histopathological factors. That indicates that late recurring tumors are actually new primary tumors, the researchers argued.
In an editorial appearing in the same issue of NEJM, University of Toronto researcher Morris Sherman lauded the work, which he said “has direct implications for the prediction of survival and late recurrence after resection for hepatocellular carcinoma.”
“By demonstrating an association between survival and recurrence signatures in nontumorous liver tissue, the authors bring the possibility of individualized therapy for hepatocellular carcinoma one step closer,” Sherman wrote. He also noted that the approach could be useful for understanding other types of cancers and disease, since it demonstrates the usefulness of fixed samples for gene expression studies.
And although the so-called survival signature described in the paper needs to be verified in larger studies, the results suggest that gene expression profiles of nontumorous liver tissue could eventually inform liver cancer treatments.
“We envision the use of this test to identify the patients at highest risk for recurrence of hepatocellular carcinoma and to target intensive clinical follow-up or chemopreventive strategies in such patients,” Golub and his team concluded.