Liam Davenport

July 03, 2015

BARCELONA, Spain — A novel risk gene score can differentiate between hepatocellular carcinoma patients at high risk for early or late recurrence after curative therapy, Belgian investigators reported here at the 17th World Congress on Gastrointestinal Cancer.

The score, which combines seven overlapping genes associated with treatment-resistant hepatocellular carcinoma, divides patients into four categories that predict recurrence risk independent of Barcelona Clinic Liver Cancer (BCLC) staging.

However, more work is required to identify the most suitable patients for the score before it can be used in clinical practice, said an expert approached for comment.

"I believe that the key issue is really to define the population that deserves this extra test on top of the very-low-risk standard clinical pathologic procedures," said Rodrigo Dienstmann, MD, from Sage Bionetworks at the Fred Hutchinson Cancer Research Center, Seattle, and the Medical Oncology Department at Vall d'Hebron University Hospital in Barcelona, Spain. He cochaired the session during which the new score was reported.

Speaking to Medscape Medical News, Dr Dienstmann explained that it may be possible to a priori "define a population based on liver profiling, for example, that will go very poorly, irrespective of the surgery and everything that you do."

He continued: "You may even avoid surgery in this situation or avoid transplantation. But I would go for the identification of the really-low-risk population that deserves transplantation, for example."

Dr Rodrigo Dienstmann

For Dr Dienstmann, liver profiling before surgery may indeed be the best use of the risk score, because "if you identify a low-risk population for late relapse, and if you have liver-only disease, that is definitely the subgroup that deserves liver transplantation."

"It's not a clear patient population, so there's still a lot of room for understanding this population and better selecting these patients," he concluded.

New Score

Dr Jeroen Dekervel

The new score was developed by a team headed by Jeroen Dekervel, MD, who is currently working on his PhD in the Department of Hepatology at University Hospitals Leuven in Belgium.

Dr Dekervel explained that patients with hepatocellular carcinoma have a "dismal prognosis." Curative treatments, such as ablation, resection, and transplantation, can only be successfully applied to patients with very-early-stage and early-stage disease, but even then there is a high risk for recurrence.

Previous research has shown that recurrence falls into two distinct phases. Early recurrence occurs less than 2 years after resection and involves intrahepatic metastasis. Risk factors include microvascular invasion and poor differentiation. For late recurrence, which occurs more than 2 years after resection and involves de novo tumor formation, the risk factors are hepatitis activity, hepatitis B viral load, and serum bilirubin.

To develop a risk gene score, the team used HepG2 liver cancer cells made resistant to sorafenib over several months to model hepatocyte dedifferentiation and aggressive tumor biology. An Affymetrix microarray platform was then used to determine the genes differentially expressed between this cell line and its nonresistant parental line.

This revealed 3201 differentially expressed genes. After comparison with three datasets, containing 640 samples that represent different forms of hepatocellular carcinoma, the team was able to narrow down the number to seven overlapping genes.

The seven genes— COL4A2, OXCT1, LRRC16A, F11, GCKR, ATP11C, and PCSK6 — were then combined into a Global Risk Score (GRS). This was validated in a dataset of 138 tumor samples and a dataset of 239 tumor samples and paired liver tissue.

In the first dataset, low and high GRS could differentiate patients at low and high risk for recurrence (P = .03) and those at low and high risk for reduced cumulative survival at 3 years (P < .001).

In the second dataset, similar patterns were seen for low and high GRS in the tumor and in the liver, with both the rate of recurrence and likelihood of survival significantly predicted.

When the researchers restricted the analysis to only patients who had recurrence after 19 months, tumor GRS was no longer predictive; however, liver GRS was able to predict both recurrence and survival.

This led to the establishment of four GRS risk categories:

  • Category I, low GRS in the tumor and the liver

  • Category  II, low GRS in the tumor and high GRS in the liver

  • Category  III, high GRS in the tumor and low GRS in the liver

  • Category  IV, high GRS in the tumor and the liver.

On multivariable Cox regression analysis, GRS category was a significant predictor of hepatocellular carcinoma recurrence, independent of BCLC staging. Compared with category I, the hazard ratio for category II was 2.458, for category III was 1.810, and for category IV was 2.372 (P = .007).

Dr Dekervel, spoke to Medscape Medical News about his presentation. Explaining the need for a novel risk score, he said: "I think it's important to integrate information coming from the liver and the tumor, to put it together to get a global estimate of the recurrence risk."

"Until now, we felt that this was not really completely done yet," he added.

A crucial step in developing the risk score was to be able to limit the number of genes included in the score. Dr Dekervel noted: "I did not emphasize this [in my presentation], but the gene signatures that are now available for the liver, for example, contain over 200 genes, so it's difficult to find a platform to validate and use this in clinical practice."

"Now we have a gene score with seven genes. This is something that might be cost-effective to do," he said, with techniques such as real-time quantitative polymerase chain reaction (qPCR).

Dr Dekervel believes that although the risk score and the use of qPCR needs validation in further studies, it could easily be translated to clinical practice, once the most appropriate cutoff values have been established.

"Of course, it's important to keep in mind that the underlying liver disease might differ," he added. "Some patients do not have cirrhosis, some patients just have fibrosis or even no liver disease at all, although they are rare, and this study should be validated in all these subcohorts."

The authors have disclosed no relevant financial relationships.

17th World Congress on Gastrointestinal Cancer (WCGC): Abstract O-013. Presented July 3, 2015.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.