Aim
We sought to create a prediction model for intrahepatic covalently closed circular DNA (IH‐cccDNA) level in chronic hepatitis B (CHB) patients and to validate the model's predictive accuracy.
Methods
Patients who did not receive previous nucleoside analogue (NA) therapy were assigned to the training cohort (n = 57), and those who received previous NA therapy were assigned to the validation cohort (n = 69). Factors linked to IH‐cccDNA levels in the training cohort were analyzed and a formula to predict IH‐cccDNA levels was constructed. Next, the reproducibility of that formula was assessed.
Results
In the multivariate analysis for the prediction of IH‐cccDNA level in the training cohort, fasting blood sugar (FBS) (P = 0.0227), hepatitis B e antigen (HBeAg) (P = 0.0067) and log10(HB surface antigen [HBsAg]) (P = 0.0497) were significant, whereas HB core‐related antigen (HBcrAg) tended to be significant (P = 0.0562). The formula was constructed and named the FBS‐cres score based on the variables used (FBS, HBcrAg, HBeAg, and HBsAg). The FBS‐cres score was calculated as: 3.1686 − (0.0148 × FBS) + (0.1982 × HBcrAg) + (0.0008168 × HBeAg) + (0.1761 × log10(HBsAg)). In the training cohort, a significant correlation was noted between HBcrAg and IH‐cccDNA levels (P < 0.0001, r = 0.67), whereas the FBS‐cres score was more closely correlated to IH‐cccDNA level (P < 0.0001, r = 0.81). In the validation cohort, significant correlation was found between HBcrAg and IH‐cccDNA levels (P = 0.0012, r = 0.38), whereas the FBS‐cres score was more closely linked to IH‐cccDNA levels (P < 0.0001, r = 0.51). Similar tendencies were observed in all subgroup analyses.
Conclusion
Our proposed model for the prediction of IH‐cccDNA level could be helpful in CHB patients.