Research led by the University of Texas Southwestern Medical Center reveals a protein signature that can reliably predict whether patients with nonalcoholic fatty liver disease (NAFLD) are likely to develop hepatocellular carcinoma, the most common form of liver cancer.

The result of many years of development work, the team has created a test that measures the levels of a panel of proteins in the blood and could evaluate whether a person with NAFLD is at high or low risk for cancer without the need for regular ultrasound examinations.

As outlined in the June 22, 2022, issue of Science Translational Medicine, the researchers also found the risk signatures were altered in patients undergoing bariatric surgery, statin treatment and cancer immunotherapy suggesting they could play a role in new treatment development and monitoring of disease progression and treatment.

Rates of liver cancer are rapidly increasing. There has been a more than three-fold increase in the U.S. since the early 1980s, with hepatocellular carcinoma accounting for around 75% of cases. In 2020, there were more than 900,000 cases and over 800,000 deaths around the world.

Factors such as hepatitis infection, alcohol use and other liver conditions all impact risk for hepatocellular carcinoma, with obesity-associated NAFLD being one such factor. Although the risk for cancer in these patients is small (less than 1% per year), the high numbers of people with NAFLD make it an important group to monitor.

"Some epidemiological studies suggest, shockingly, that one-fourth of the obese adults are supposed to have fatty liver disease. So, this is a huge population we need to watch," lead study researcher and University of Texas associate professor Yujin Hoshida told BioWorld Science.

"The vast majority of fatty liver disease is benign..... but a very small fraction of the patients develop cancer. Because the denominator is huge, even if the fractional proportion of the patients with liver cancer is small, the actual number is very large," he explained.

Standard practice currently recommends 6-monthly ultrasound scans for people with NAFLD, but in practice this is hard to apply. In reality, less than 25% of potentially at-risk patients undergo such screening. This is a problem, particularly as survival rates from hepatocellular carcinoma are much higher if it is detected early.

"Practically, it's impossible to watch one-fourth of adults with obesity. So, we really need to have a test to tell us who should be in the small, small subset of fatty liver patients who we really need to take a close look at because of the elevated cancer risk," said Hoshida.

Applying molecular techniques

Hoshida and colleagues have been working to improve cancer risk prediction in liver disease patients for many years. This study assessed the protein expression of a panel of 133 genes expressed in the liver and serum, 80 of which are linked to increased risk and 53 to low risk for developing hepatocellular carcinoma.

To define the risk signature, the team first assessed gene expression in 48 NAFLD patients who previously experienced hepatocellular carcinoma and created a score to predict high or low risk for developing this type of liver cancer.

"When you see an increase of high-risk proteins and a reduction of low-risk proteins, we make a prediction for high risk and if you see low high-risk proteins and the high low-risk proteins, it's a pattern of good prognosis or low risk," noted Hoshida.

The risk signature was then validated in a group of 106 patients with NAFLD who had never had hepatocellular carcinoma and a separate group of 59 individuals with NAFLD who had undergone treatment for hepatocellular carcinoma in the past.

In the group who had never previously had cancer, the incidence rates of hepatocellular carcinoma after 15 years of observation were 22.7% in the high-risk group and 0% in the low-risk group. Similarly, after 5 years of follow-up, 71.8% and 42.9% of high- and low-risk designated patients had a recurrence of their cancer in the group with previous hepatocellular carcinoma.

"We've been working on this biomarker project over the past decade. We initially [started] from a more experimental approach, looking at liver tissue by taking biopsies... This is an invasive method, but you can get more reliable molecular information, because we are getting the information directly from the liver," explained Hoshida.

Taking liver biopsies from the number of patients impacted by NAFLD would be problematic due to the invasive nature of the approach and the medical expertise and time needed. "Based on that, we spent years translating the information to measure less invasively by using blood samples," said Hoshida.

The researchers used the previously developed bioinformatics tool TexSEC to convert their larger protein expression panel into a blood-based four-protein secretome panel. This blood test was validated in a group of 59 individuals with NAFLD but no prior liver cancer. In this group, incidence rates at 15 years were 37.6% and 0% in the high- and low-risk groups, respectively.

Overall, individuals with a high-risk score had high levels of IDO1, an enzyme known to suppress antitumor immunity. "IDO1 may be a therapeutic target to modulate high-risk prognostic liver signature-NAFLD and reduce hepatocellular carcinoma incidence in NAFLD," the authors wrote. Other factors that distinguished high risk individuals included the presence of dysfunctional CD8+ T cells and impaired metabolic regulators. In contrast, those at low risk had higher levels of bone marrow-derived monocytes.

Notably, the risk signatures were altered in patients undergoing bariatric surgery, lipophilic statin treatment and immunotherapy (IDO1 inhibitors) suggesting they could also help monitor treatment efficacy in clinical trials, as well as potentially guiding new hepatocellular cancer treatments.

The research team behind this study has previously created another protein-based risk score, which can be used to predict cancer risk across different liver conditions including alcoholic liver disease, hepatitis and NAFLD. When combined with the NAFLD specific score, they found that hepatocellular carcinoma risk prediction was further improved.

Hoshida explained that next steps for the NAFLD risk score are to validate it further using an NIH liver disease biobank, including 1,500 patients, as well as using a similar Texas-based resource – the Texas Hepatocellular Carcinoma Consortium – which has 2,000 samples.

To accelerate its journey to the clinic, the researchers are also keen to work with industrial partners to help develop the predictive score into an FDA-approved test that can be rolled out more widely.