Early detection of hepatocellular carcinoma (HCC), a highly deadly cancer, is essential for improving patient outcomes. Researchers have developed a serum fusion-gene machine-learning model that could greatly improve the five-year survival rate for HCC patients, increasing it from 20% to 90%. This is due to the model’s enhanced accuracy in detecting HCC in its early stages and its ability to monitor the effectiveness of treatment.The study reports on the creation of a serum fusion-gene machine-learning model, which is a crucial screening tool that could potentially increase the five-year survival rate of patients with HCC from 20% to 90%. This is due to its improved accuracy in the early diagnosis of HCC and its ability to monitor the impact of treatment. The findings were published in The American Journal of Pathology by Elsevier.
HCC, which accounts for around 90% of liver cancer cases, is the most common form of the disease. The current screening test for the HCC biomarker, serum alpha-fetal protein, is not always accurate, and it is estimated that up to 60% of liver cancers are only Diagnosed in later stages, resulting in a survival rate of only about 20%.
Lead investigator Jian-Hua Luo, MD, PhD, Department of Pathology, High Throughput Genome Center, and Pittsburgh Liver Research Center, University of Pittsburgh School of Medicine, stated: “Detecting liver cancer early saves lives. However, most liver cancers develop gradually and without many symptoms. This makes early detection difficult. What we need is an affordable, precise, and convenient test to screen for early-stage liver cancer in human populations. We wanted to see if a machine-learning approach could improve the accuracy.The study focused on screening for HCC (hepatocellular carcinoma) by examining fusion genes. Researchers looked at nine fusion transcripts in serum samples from 61 HCC patients and 75 patients with non-HCC conditions using real-time quantitative reverse transcription PCR (RT-PCR). Seven of the nine fusions were commonly found in HCC patients. The researchers used machine-learning models based on serum fusion-gene levels to predict HCC in the training cohort, using the leave-one-out cross-validation approach. A logistic model based on four fusion genes was able to predict HCC accurately.The logistic regression model was able to predict the occurrence of HCC with an accuracy of 83% to 91%. When combined with serum alpha-fetal protein, the fusion gene plus alpha-fetal protein logistic regression model achieved 95% accuracy for all cohorts. Additionally, the quantification of fusion gene transcripts in serum samples effectively assessed the impact of treatment and could monitor cancer recurrence.
Dr. Luo stated, “The fusion gene machine-learning model greatly enhances the early detection of HCC compared to serum alpha-fetal protein alone. It could be a valuable tool for screening for HCC.”
When treating liver cancer, it is important to monitor its impact. This test can identify patients who are likely to have liver cancer.
According to Dr. Luo, early treatment of liver cancer leads to a 90% five-year survival rate, whereas late treatment only results in a 20% survival rate. The alternative to this test is to conduct imaging analysis every six months on individuals at risk of liver cancer, which is costly and ineffective. Furthermore, when imaging results are inconclusive, this test can help distinguish between malignant and benign lesions.
