In the study, we established and validated a noninvasive model named AAPL which were composed of easily available laboratory and clinical indexes. The model can be used to identify the absence or presence of significant liver injury in CHB patients with normal ALT level, and to evaluate whether this part of patients need antiviral therapy immediately.

CHB patients have an increased risk of liver cirrhosis and HCC [17]. Early control of HBV activity can prevent the occurrence of end-stage liver diseases [9, 18]. Studies have shown that antiviral therapy can promote HBeAg and HBsAg seroconversion. It is reported that HBeAg is important in the history of chronic HBV infection. During the course of HBV infection, HBeAg seroconversion often coincides with normalization of biochemical test and clinical remission [19,20,21]. There is a lower incidence of cirrhosis and HCC among patients who undergo spontaneous or treatment-induced HBeAg seroconversion [21, 22]. HBeAg seroconversion is also associated with a higher probability of HBsAg loss and seroconversion, which is considered to be associated with clinical cure of HBV infection [21, 23]. All CHB patients meeting the antiviral therapy indication should get treated in order to prevent the progression of liver injury [10].

Based on the current guidelines, HBeAg positive patients with normal ALT level and high HBV viral load, as well as HBeAg negative with normal ALT level and low HBV viral load are required regular follow-up [9, 10]. There is a proportion of CHB patients with normal ALT level having significant liver injury, who should get antiviral therapy immediately. Traditionally, such patients are identified after liver biopsy. However, liver biopsy, as an invasive examination with a series of potential complications is often not acceptable to patients [11, 24]. Thus, it is challenging to identify CHB patients with normal ALT level who need antiviral therapy. So, we establish a noninvasive prediction model to solve this problem.

It has been showed in previous studies that ALT, Age, PLT and LS were independent variables for severity of liver injury [25,26,27]. In the study, we also found that there are positive correlations between these indexes and liver injury, in agreement with the previous reports. Furthermore, we developed the prediction model based on these 4 indexes. Our noninvasive model proved to have satisfactory diagnostic value, which helps to identify those who need initiating antiviral therapy immediately among CHB patients with normal ALT level.

Most noninvasive models only focus on predicting the stage of liver fibrosis or the grade of the liver inflammation. However, according to clinical guidelines on HBV infection nowadays, we should take both fibrosis stage and inflammation grade into account. A noninvasive model composed of PT, AST, GGT, and anti-hepatitis B core antibody levels is valuable in predicting significant liver inflammation in CHB patients [28], however, it ignores the fibrosis stage. Conversely, another noninvasive model called CPHBV could accurately predict liver fibrosis in CHB patients with normal ALT level, regardless of liver inflammation [29]. AGH model composed of AST, hepatitis B core antigen and GGT is a reliable index to predict moderate to severe inflammation or significant fibrosis, which helps to determine the initiation of antiviral therapy [30]. But the predictive value of our model is better than AGH model.

Based on the logistic analysis, we establish a noninvasive model named AAPL, which combined ALT, LS, Age and PLT. According to the ROC curve, the AUCs in the training cohort and the validation cohort were 0.85 and 0.87, both satisfactory. Furthermore, for CHB patient with normal ALT level, we recommend antiviral therapy when AAPL index is above the high cut-off value, and regular monitoring when it is below the low cut-off value. Or rather, patients whose AAPL index is beyond the 2 cut-off values could avoid liver biopsy.

The treatment of CHB was mainly based on serum HBV DNA level, serum ALT level and severity of hepatic inflammation and fibrosis according to the guidelines [31, 32]. Previous studies suggested that LS, Age and PLT were risk factors for liver injury in patients with chronic HBV infection [25, 26]. In our study, logistics analysis showed that ALT level, LS, Age and PLT were independent factors for liver injury. Our findings were consistent with clinical guidelines and previous studies, which indicated the reliability of the model.

Our study has several advantages. It included a relatively large cohort of treatment-naïve CHB patients with normal ALT level. The research is a multicenter study, and the noninvasive model has been validated in a cohort from 2 centers. The newly established model consists of clinical and laboratory variables readily available during regular follow-up of CHB in clinical practice. It is easy for patients to use, and has good performance in the identification of antiviral therapy. Treatment-naïve CHB patients could benefit from the model to assess their hepatic histology during their follow-up, and to start their antiviral therapy on time without liver biopsy.

In conclusion, a proportion of CHB patients with normal ALT level, who proves to have significant liver injury referring to liver biopsy, should receive antiviral therapy. ALT level, LS, Age and PLT were independent variables to identify significant liver injury. The noninvasive model consisting of the 4 indexes, can be used to identify histological indication for antiviral therapy, which helps to reduce the need of liver biopsy.