Optimal threshold of alpha-fetoprotein response in patients with unresectable hepatocellular carcinoma treated with atezolizumab and bevacizumab

Summary Alpha-fetoprotein (AFP) response (relative decline in AFP) is associated with imaging response evaluated by response evaluation criteria in solid tumors ver1.1 (RECIST) and survival in treatment for hepatocellular carcinoma (HCC). However, the optimal threshold of AFP response is still unknown, especially in atezolizumab and bevacizumab (Atez/Bev) treatment. In this prospective multicenter study, we aimed to investigate an optimal threshold of AFP response in Atez/Bev treatment. Out of 284 patients with unresectable HCC who were treated with Atez/Bev, 91 patients with AFP ≥ 10 ng/ml were enrolled in the multicenter study. We investigated the relationship between various AFP response thresholds (relative decline ≥ 20%, ≥ 50%, and ≥ 75%) and treatment response and progression-free survival (PFS). An AFP relative decrease of ≥ 50% was associated with an overall response rate (ORR) with an odds ratio (95% confidence interval [CI]) of 5.7 (1.9–17). Disease control rate (DCR) was associated with an AFP relative decrease of ≥ 20%, with a 100% positive predictive value and a 52.0% sensitivity. AFP relative decreases of ≥ 50% and ≥ 20% were significantly associated with PFS with a hazard ratio (HR) of 5.60 (95% CI: 1.6–19, p = 0.006) and a HR of 4.44 (95% CI: 1.9–10, p < 0.001), respectively. AFP response of ≥ 50% and ≥ 20% were related to ORR and DCR, respectively, and both of these responses were also associated with PFS. AFP can be used as a real-time monitor during Atez/Bev treatment and is helpful for treatment optimization.