Prediction of Sustained Response After Nucleo(s)tide Analogue Cessation Using HBsAg and HBcrAg Levels: A Multicenter Study (CREATE)

Abstract

BACKGROUND & AIMS: Predictors of successful nucleo(s)tide analogue (NA) therapy withdrawal remain elusive. We studied the relationship between end-of-treatment levels of hepatitis B core-related antigen (HBcrAg) and hepatitis B surface antigen (HBsAg) and outcome after therapy cessation. METHODS: Patients who discontinued NA therapy in centers in Asia and Europe were enrolled. HBcrAg and HBsAg were measured at treatment cessation, and associations with off-treatment outcomes were explored. The SCALE-B (Surface antigen, Core-related antigen, Age, ALT, and tenofovir for HBV) score was calculated as previously reported. End points included sustained virologic response (VR; hepatitis B virus DNA level <2000 IU/mL), HBsAg loss, and alanine aminotransferase (ALT) flares (>3 X upper limit of normal). Re-treated patients were considered nonresponders. RESULTS: We analyzed 572 patients, 457 (80%) were Asian and 95 (17%) were hepatitis B e antigen positive at the start of NA therapy. The median treatment duration was 295 weeks. VR was observed in 267 (47%), HBsAg loss was observed in 24 (4.2%), and ALT flare was observed in 92 (16%). VR (67% vs 42%) and HBsAg loss (15% vs 1.5%) was observed more frequently in non-Asian patients when compared to Asian patients (P < .001). Lower HBcrAg levels were associated with higher rates of VR (odds ratio [OR], 0.701; P < .001) and HBsAg loss (OR, 0.476; P < .001), and lower rates of ALT flares (OR, 1.288; P = .005). Similar results were observed with HBsAg (VR: OR, 0.812; P = .011; HBsAg loss: OR, 0.380; P < .001; and ALT flare: OR, 1.833; P < .001). Lower SCALE-B scores were associated with higher rates of VR, HBsAg loss, and lower rates of ALT flares in both Asian and non-Asian patients (P < .001). CONCLUSIONS: In this multicenter study, off-treatment outcomes after NA cessation varied with ethnicity. Lower levels of HBcrAg and HBsAg were associated with favorable outcomes. A risk score comprising both factors can be used for risk stratification.