HCC risk prediction model for patients with chronic hepatitis B - The Hope of Precision Medicine- A review of high scores.

In the past 10 years, more than 12 risk scoring models have been developed to predict the risk of hepatocellular carcinoma (HCC) in patients with chronic hepatitis B (CHB).these scoring models divide the risk of HCC occurrence in CHB patients into three levels: low, medium and high, so as to evaluate whether the patients need HCC monitoring and monitoring strategies.generally speaking, a risk score model with high reliability can be selected to evaluate low-risk patients. For example, the model of negative predictive value (NPV) & gt; 99% can be used to evaluate low-risk patients. If the cumulative 5-year incidence of HCC & lt; 1% in patients can be considered as low-risk HCC, such patients do not need regular HCC monitoring.on the other hand, it is recommended that patients with moderate and high risk should continue to receive HCC monitoring.in this review, the HCC risk scores of untreated and Na treated CHB patients were analyzed. The similarities and differences of these scores were summarized to understand how na treatment affects the risk of HCC and the relationship between HCC risk factors.HCC risk score for untreated patients, the researchers included a total of 10 HCC risk assessment models for predicting untreated patients, all of whom were from East Asia (Table 1).Table 1 Summary of HCC risk score models note: * derived cohort treated with ETV or tenofovir (TDF) der = derived cohort; alid = validation cohort; ALT = alanine aminotransferase; AFP = alpha fetoprotein; PLT = platelets; LSM = liver hardness determination; as expected, age was the most frequently included variable in these evaluation models Nine models were included in the scoring system, followed by HBV DNA (n = 8), gender (n = 7), liver cirrhosis, liver stiffness and HBeAg (n = 3 each).other less common indicators included family history of HCC, alt, albumin, bilirubin, AFP and HBsAg titers, HBV genotype, platelet and spleen diameter.as shown in Table 2, the accuracy of these models is represented by auroc.in terms of assessing low-risk population, reach-b, lsm-hcc and d2as have higher NPV (> 99%) and can predict low-risk patients more accurately.Table 2 accuracy comparison of prediction models note: except for special identification, this table is the prediction of 5-year cumulative risk of HCC * 3-year cumulative risk of HCC occurrence; ා 10-year cumulative risk prediction of HCC occurrence; auroc = area under the work characteristic curve of subjects; der = derivation queue; alid = validation cohort.HCC risk scores for Na treated patients. To date, seven HCC risk scores have been used to predict HCC risk in Na treated CHB patients (Table 1).six of the models included age and gender variables.cirrhosis, platelets and serum albumin were included in each of the three scores.less selected indicators included alpha fetoprotein and diabetes.Table 2 shows that the accuracy of these prediction models is high.independent validation of the HCC risk score, the researchers validated the performance of several HCC risk models in an independent cohort of at least 16 treated or untreated patients, or were more representative of actual follow-up CHB patients (Figure 1). in Figure 1, several HCC risk models were independently verified to evaluate their accuracy in predicting the cumulative risk of HCC in 5 years. A: the ordinate is the auroc value; B: the ordinate is the NPV value. In Figure 1a, the aurocs of each model are mainly between 0.6 and 0.8. although it is difficult to directly compare these models, reach-b seems to have the lowest auroc, possibly because liver fibrosis or cirrhosis is not considered. figure 1b compares the NPV values of each model, and the Page-B and mpage-b models seem to have the highest NPV values. the variables in Table 1 can be divided into three categories, including: (1) degree of liver fibrosis and cirrhosis; (2) degree of hepatitis activity; (3) host factors, including comorbidity. Liver cirrhosis / hepatic fibrosis cirrhosis is the strongest risk factor for HCC. With the progression of the disease, liver fibrosis develops into cirrhosis, and the compensatory stage progresses to decompensated stage, and the risk of HCC increases gradually. among the predictors of HCC risk over 5 years, the important variables were age at year 5, lower platelet count and liver stiffness & gt; 12 kPa. these data suggest that reversion of cirrhosis and resolution of fibrosis significantly reduce the risk of HCC. the applicability of the current score after liver cirrhosis reversion remains to be studied. The risk of HCC is related to hepatitis activity, which can be determined by serum HBV DNA and ALT levels. although chronic inflammation promotes fiber formation and thus increases the risk of HCC, inflammation itself has always been associated with various human cancers. inflammatory necrosis and repeated circulation of hepatocyte regeneration lead to hepatocyte variation and abnormal proliferation. other factors that promote the occurrence of HCC include oxidative stress, up regulation of cytokines and growth factors, and DNA methylation. a randomized controlled trial by Liaw showed that inhibition of viral replication reduces the risk of HCC. In host factors, age, gender, severity of fibrosis or liver dysfunction were included in most models, while family history, drinking or smoking were excluded. part of the reason may be that host factors (such as family history) have less influence than stronger predictors (such as cirrhosis). summary in the past 10 years, more than 12 risk scoring models have been developed to predict the risk of hepatocellular carcinoma (HCC) in patients with chronic hepatitis B (CHB). many of these models are based on CHB patients who have not received antiviral treatment, while others are based on patients receiving antiviral treatment. the variables included in these scoring models can be divided into three groups: (1) degree of liver fibrosis and cirrhosis; (2) degree of hepatitis activity; (3) host factors, including comorbidity. HCC risk scores for untreated patients tend to include all three types of variables, while models for treated patients rarely include variables representing hepatitis activity. some variables recognized as major risk factors for HCC, such as family history, were not included in most of the scores. however, more work needs to be done to comprehensively consider all relevant variables and provide individualized prediction, so as to be applied to a wider range of CHB population. yimaitong was translated from: George v. papathodoridis, thodoris Voulgaris, Margarita papathodoridi, et al. Risk scores for hepatocellular carcinoma in chronic hepatitis B: a promise for precision medicine. Hepatology. Doi: 10.1002/hep.31440?