High Levels of Antidrug Antibodies Against Atezolizumab as a Predictive Marker for Clinical Outcomes in Patients With Hepatocellular Carcinoma

Abstract

To the Editor We read with great interest the cohort study by Kim and colleagues 1 and appreciate the authors' efforts to determine the adverse prognostic association of high-level serum antidrug antibody (ADA) against atezolizumab in patients with advanced hepatocellular carcinoma (HCC) treated with atezolizumab and bevacizumab. We have 3 concerns regarding the interpretation of the findings and hope that the authors would consider them.The first concern is the authors' speculation that "high ADA levels may reduce atezolizumab exposure and attenuate the anticancer efficacy of the drug." 1 Using data from various clinical trials, other studies have found that serum atezolizumab concentrations exceeded the target saturation level of 6 μg/mL in more than 90% of patients regardless of ADA status, 2,3 suggesting that ADA development has no clinically meaningful association with atezolizumab exposure. In fact, as Figure 2B of the article shows, 1 more than 85% of patients with high ADA levels (≥1000 ng/mL) achieved the atezolizumab target exposure, indicating that ADA does not appear to be associated with atezolizumab outcomes by itself. We would be grateful if the authors would elaborate on how ADA is associated with atezolizumab outcomes.The second concern is the effect of confounding between baseline disease characteristics and ADA on atezolizumab pharmacokinetics and clinical outcomes. 1 For example, the frequency of ADA development is high in patients with increased tumor burden and elevated systemic inflammation (eg, highvolume tumor, high neutrophil-to-lymphocyte ratio, and highlevel α-fetoprotein [AFP]) at baseline, but these baseline characteristics are also associated with increased atezolizumab clearance and poor prognosis. 2,3 As eFigure 5A of the article shows, 1 the adverse prognostic indicator neutrophil-lymphocyte ratio did not affect survival when analyzed in conjunction with ADA, suggesting the confounding association. It would be helpful if the authors could provide the baseline characteristics of ADA-high and ADA-low patients, allowing for the assessment of nonnegligible confounding effects.The last concern pertains to the utility of ADA as a predictor for clinical outcomes compared with existing or emerging markers. Comprehensive molecular research has identified AFP expression in the HCC microenvironment as 1 predictor of atezolizumab and bevacizumab inefficacy. 4 In this article, 1 baseline serum AFP, similar to ADA, was independently associated with worse overall survival. On-treatment serum AFP has been recently identified as a promising biomarker for response to atezolizumab and bevacizumab in HCC. 5 Therefore, we would appreciate it if the authors would discuss whether ADA or on-treatment AFP is more accurate for predicting therapeutic response. Candidate biomarkers are emerging and must be ranked for performance.