Genetic ‘signature’ in healthy liver cells predicts risk of liver cancer recurrence

Researchers have found a gene ‘signature’ – a specific set of genes – in the healthy, noncancerous cells in the liver of people who have had liver cancer that is associated with better chances of survival and a second gene signature associated with late cancer recurrence, in an international study that has shown for the first time that tissue samples preserved by traditional methods can be used for genetic analysis.

The researchers analysed the genetic profiles of liver samples from 307 people with liver cancer who had taken part in clinical studies in four countries – Japan, USA, Italy and Spain.

The liver samples had been preserved using traditional methods of fixing tissue with formalin and embedding in paraffin, rather than being frozen. Some had been stored for more than 24 years. Current methods of genomewide expression profiling require frozen tissue for analysis. This has – until now - limited efforts to study the genetic profile of liver cancer cells, because tissue banks from most clinical trials have collected formalin-fixed, paraffin-embedded specimens.

The researchers used a new method for gene profiling that was able to reconstruct the thousands of genes that are cut into tiny pieces when tissue samples are treated with a chemical fixative and stored in wax. They used DNA microarrays (DNA chips) to analyse the gene expression simultaneously of a 6000 genes across the human genome, and found it was successfully in 90% of the specimens.

When they used the gene profiling technique to look at the tissue samples from people with liver cancer they were surprised to find that the genetic profile of the tumour tissue was not predictive of outcome, survival or late recurrence (a further tumour occurring more than two years after the first cancer). However, when they analysed apparently normal liver tissue samples surrounding the patient’s tumours they found a gene signature of 186 genes that was highly correlated with survival (p=0.04).

This ‘good prognosis’ signature contained genes associated with normal liver function, including genes coding for plasma proteins and for several drug-metabolising enzymes. In contrast, the ‘poor prognosis’ signature – found in patients at higher risk of liver cancer recurrence – contained gene sets associated with inflammation, including those related to interferon signalling, activation of nuclear factor-κB, and signalling by tumour necrosis factor α.

The findings fit with the ‘field defect’ theory, in which researchers have previously suggested that factors in the entire liver may predispose a person to liver cancer. There may be genetic abnormalities in liver tissue that appears normal that could give rise to new tumours after an initial tumour has been removed.

The lead author of the study, Yujin Hoshida, from the Broad Institute of Massachusetts Institute of Technology and Harvard University, Cambridge, USA, said: “These findings indicate that we might be able to identify patients at high risk of recurrence and target those patients to prevent it.” Although the treatment of hepatocellular carcinoma is evolving, hepatic resection remains the treatment of choice for most patients. Resection is associated with a five-year survival rate of 50%, but the recurrence rate remains 70%. For most patients, resection is not a cure.

Dr Hoshida added: “The fact that the predictive information comes not from the tumour but from surrounding tissue could offer important insights into the mechanism of liver cancer.”

In an editorial accompanying the report of the results, Morris Sherman, from the University of Toronto, Canada, agreed that the new findings added to the understanding of the development of liver cancer. “The research has opened the door to identifying the relevant gene expression in the pathogenesis of hepatocellular carcinoma as it evolves from non-tumourous liver, as well as initiating research into a molecular method for determining more precisely who is at risk for the development of hepatocellular carcinoma.” He added that these findings “bring the possibility of individualised therapy for hepatocellular carcinoma one step closer.”

The researchers suggested that, in future research, the new gene profiling technique could potentially be used in any type of cancer. “We don’t know whether there will be a recurrence signature in the non-tumour tissue of breast cancer. But it is now possible to explore that possibility.”