Model predicts progression to acute liver failure in chronic HBV
A new prediction model revealed that hepatitis B DNA, international normalized ratio and older age correlated with an increased risk for the development of acute-on-chronic liver failure in patients with acute exacerbation of chronic HBV.
“During the natural course of chronic HBV infection, most patients will experience hepatitis flares with various degrees of liver injury,” researchers from the department of infectious disease, Affiliated Hospital of Zunyi Medical Center, Guizhou, China, wrote. “Patients with [acute exacerbation] of chronic HBV infection are at risk of further deterioration and progression to hepatic decompensation and acute-on-chronic liver failure (ACLF).”
The study comprised 474 patients with acute exacerbation of chronic HBV. Thirty-six patients progressed to ACLF after hospital admission, 15 of whom died during 3 months of follow-up.
Investigators randomly assigned patients into a derivation cohort (n = 280) and a validation cohort (n = 194) for the final model. Patients in the derivation cohort had higher levels of albumin and prealbumin and were more likely to have cirrhosis compared with the validation cohort. Otherwise, clinical characteristics were comparable between the two groups.
In the derivation cohort, patients who developed ACLF were more likely to be older (43.7 vs. 37.4 years; P = .012), have higher levels of HBV DNA (7.5 vs. 6.1 log copies/mL; P = .001), total bilirubin (211.2 vs. 133.3 mol/L; P = .032) and international normalized ratio (1.59 vs. 1.23; P < .0005), and have lower levels of prealbumin (47.5 vs. 70 mg/L; P = .009), sodium (135.6 vs. 137.2 mmol/L; P = .012) and prothrombin activity (46.5% vs. 70.3%; P < .0005) compared with those without ACLF.
Univariate analysis showed that age, HBV DNA, total bilirubin, INR, prealbumin, prothrombin activity and sodium levels correlated significantly with progression to ACLF. However, after multivariate analysis only age (OR = 1.054; 95% CI, 1.004-1.107), HBV DNA (OR = 1.705; 95% CI, 1.105-2.631) and INR (OR = 37.74; 95% CI, 6.7-211.2) remained significant factors.
The data indicate this model better predicted ACLF progression (AUROC = 0.867; 95% CI, 0.781-0.9954) compared with MELD score (AUROC = 0.706; 95% CI, 0.577-0.834) and MELD-Na score (AUROC = 0.7; 95% CI, 0.575-0.824). With a cut-off value of –2.43, the model also achieved a higher sensitivity (85%), specificity (80%), positive predictive value (24.6%) and negative predictive value (99.6%) compared with MELD and MELD-Na score.
Results from the validation cohort showed the model continued to produce better results (AUROC = 0.827; 95% CI, 0.766-0.877) than MELD score (AUROC = 0.678; 95% CI, 0.607-0.743) or MELD-Na score (AUROC = 0.667; 95% CI, 0.596-0.733). With the cut-off value of –2.43, the model achieved a higher sensitivity (87.5%), specificity (73.6%), positive predictive value (23%) and negative predictive value (98.5%) than MELD or MELD-Na score.
“Although all patients in our study had started antiviral therapy after admission, baseline HBV DNA level was found to be one of only three independent risk factors,” the researchers wrote. “Our results provide further evidence to support the hypothesis that baseline HBV DNA level plays an important role in determining progression in [acute exacerbation] patients despite early antiviral treatment.” – by Talitha Bennett
Disclosure: The authors report no relevant financial disclosures.