Elevated antidrug antibodies against atezolizumab and associated clinical outcomes in advanced non-small-cell lung cancer Authors: Miran Jang, Chan Kim, So Jung Kong, Hannah Yang, Won Suk Lee, Ho Yun Lee, Hong Jae Chon, Se-Hoon Lee Published: 2025 ABSTRACT AbstractBackgroundAlthough atezolizumab has demonstrated efficacy in advanced non-small-cell lung cancer (NSCLC), antidrug antibody (ADA) may reduce its effectiveness by lowering drug exposure. We explored the association between ADA levels and clinical outcomes.MethodsWe retrospectively analyzed 86 patients with advanced NSCLC who received atezolizumab monotherapy (1200 mg every 3 weeks) between August 2018 and September 2022 at Samsung Medical Center, Seoul, Korea. Blood samples were collected prior to the first and second doses (baseline and week 3). ADA levels were measured by ELISA and correlated with plasma atezolizumab concentrations and clinical outcomes using Kaplan–Meier estimates and Cox proportional hazards models.ResultsAll 86 patients received atezolizumab as a second-line or later treatment (second-line, n = 65; third-line, n = 12; ≥ fourth-line, n = 9). Patients (median age [IQR], 67 [61–73] years; 73 [84.9%] male) showed significantly elevated ADA levels three weeks after treatment (median [IQR] 0 [0–0] vs. 530.3 [146.9–3050.5] ng/mL; p < 0.001). Strong ADA levels (≥ 1000 ng/mL) were observed in 32 (37.2%) patients and were associated with shorter PFS (HR = 1.86, 95% CI: 1.15–3.02, p = 0.010) and OS (HR = 1.92, 95% CI: 1.14–3.23, p = 0.013). Furthermore, patients with high ADA levels exhibited lower atezolizumab concentrations and reduced response rates compared to those with low ADA levels. Importantly, high ADA levels independently predicted poor prognosis in a multivariable analysis adjusted for clinical variables.ConclusionsHigh ADA levels were linked to lower atezolizumab exposure and worse outcomes. ADA monitoring may help predict prognosis and guide immunotherapy strategies.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00262-025-04245-3.Keywords: Antidrug antibody, Atezolizumab, Immunotherapy, Non-small-cell lung cancer, PD-L1 Introduction The introduction of immune checkpoint inhibitors (ICIs) has revolutionized the treatment of non-small-cell lung cancer (NSCLC) [1, 2]. Among them, atezolizumab demonstrated superior efficacy and safety in the Phase III OAK trial, showing a significant improvement in overall survival (OS) compared with docetaxel in previously treated patients (median OS, 13.8 vs. 9.6 months) [3]. The Phase II POPLAR study further confirmed these findings and highlighted programmed death-ligand 1 (PD-L1) expression as a predictive biomarker of response to atezolizumab [4]. Despite these advances, a substantial subset of patients exhibits primary resistance to immunotherapy, underscoring the need to identify additional mechanisms influencing treatment outcomes [5, 6]. The use of ICIs can elicit immune responses that trigger the formation of antidrug antibodies (ADA) [7, 8]. ADA can alter the pharmacokinetics and pharmacodynamics of therapeutic antibodies, potentially reducing efficacy or leading to adverse immune reactions [9–11]. Notably, ADA formation has been associated with decreased drug exposure and inferior clinical outcomes in hepatocellular carcinoma (HCC) patients treated with atezolizumab and bevacizumab [12]. Moreover, atezolizumab has been reported to induce ADA more frequently than other ICIs [11, 13]. Here, we evaluated ADA levels following atezolizumab monotherapy and assessed their association with clinical outcomes in patients with advanced NSCLC. Methods Patients and sample collection This study received approval from the Institutional Review Board of Samsung Medical Center (IRB No. SMC-2018–04-048) and adhered to the principles of the Declaration of Helsinki. Inclusion criteria were patients with histologically confirmed advanced NSCLC who received atezolizumab monotherapy at Samsung Medical Center. Eligible patients had available blood samples collected both prior to the first dose (baseline) and three weeks after treatment initiation for ADA and drug concentration analysis. Patients were excluded if they (1) received atezolizumab as part of combination therapy, (2) had insufficient follow-up for survival evaluation, or (3) lacked adequate clinical or laboratory data required for ADA assessment. A total of 86 patients were included, all of whom provided written informed consent before participation. Tumor response was evaluated using RECIST version 1.1, and all patients received a fixed dose of atezolizumab (1200 mg every three weeks). Blood samples were collected before the first and second infusions (approximately three weeks apart). Whole blood samples were processed to isolate plasma, which was stored at − 80 °C until analysis. The data cutoff date was September 12, 2023. Assessment of anti-atezolizumab antibodies and atezolizumab levels The levels of anti-atezolizumab antibodies were assessed through an enzyme-linked immunosorbent assay (ELISA) (KBI2027 ver 4.0, KRISHGEN BioSystems) following the manufacturer's guidelines. In summary, samples were incubated for 120 min at 37 °C in a microwell pre-coated with atezolizumab. Following multiple washes, horseradish peroxidase–conjugated atezolizumab was introduced, and the plate was further incubated for 60 min at 37 °C. After additional washing steps, TMB (3,3′,5,5′-Tetramethylbenzidine) substrate solution was added, and the plate was incubated for 30 min at 37 °C in darkness. The absorbance was then measured at 450 nm using a microplate reader (BioTek). The plasma atezolizumab concentration was determined using a commercial ELISA assay (KBI1027, KRISHGEN BioSystems). Survival outcome and statistical analysis Progression-free survival (PFS) was determined from the initiation of ICI treatment to the date of either the first instance of tumor progression or death or the last follow-up. OS was defined as the duration from the start date of ICI treatment to the date of death from any cause or the last follow-up. Survival curves were generated using the Kaplan–Meier method, and subgroup comparisons were performed using the log-rank test. Cox proportional hazards regression was used for univariate and multivariate analyses of PFS and OS. R software (version 4.0.5; R Foundation for Statistical Computing, Vienna, Austria) was used for statistical analyses. Results Patient characteristics We enrolled 86 patients with advanced NSCLC who received atezolizumab monotherapy between August 2018 and September 2022 at Samsung Medical Center. The median age was 67 years (range, 61–73), with 73 males (84.9%) and 13 females (15.1%). The histological types included 58.1% adenocarcinoma, 30.2% squamous cell carcinoma, and 11.6% other types. Among the patients, 1 achieved a complete response (CR), 10 achieved a partial response (PR), 12 had stable disease (SD), and 63 experienced progressive disease (PD) according to RECIST guidelines. All 86 patients received atezolizumab as a second-line or later treatment (second-line, n = 65; third-line, n = 12; ≥ fourth-line, n = 9). Table 1 summarizes the baseline clinical characteristics of the study cohort. Table 1Patient characteristicsCharacteristicNo. (%)Total, No.86Age, median (IQR)67 (61–73)Male sex73 (84.9)ECOG performance status 0–184 (97.7) 22 (2.3)Smoking history Never15 (17.4) Former41 (47.7) Current30 (34.9)Histology Adenocarcinoma50 (58.1) Squamous cell carcinoma26 (30.2) Others10 (11.6)Line of treatment Second-line65 (75.6) Third-line12 (13.9) Fourth-line or later9 (10.5)irAEs Yes11 (12.8) No75 (87.2)Concomitant medications* Received any3 (3.5) None83 (96.5)ADA, median (IQR), ng/mL At baseline0 At 3 weeks530.3 (146.9–3050.5)ADA levels at 3 weeks Negative or low (< 1000 ng/mL)54 (62.8) High (≥ 1000 ng/mL)32 (37.2)Data are presented as number (%) or median (range). ECOG Eastern Cooperative Oncology Group, irAEs Immune-related adverse events, ADA antidrug antibody*Only four classes of medications—antiplatelet agents, anticoagulants, corticosteroids, and immunosuppressants—were reviewed Patient characteristics Data are presented as number (%) or median (range). ECOG Eastern Cooperative Oncology Group, irAEs Immune-related adverse events, ADA antidrug antibody *Only four classes of medications—antiplatelet agents, anticoagulants, corticosteroids, and immunosuppressants—were reviewed Atezolizumab ADA levels in plasma In comparison to the baseline levels, the ADA levels of atezolizumab were significantly higher after three weeks of atezolizumab treatment (median [IQR] 0 [0–0] vs. 530.3 [146.9–3050.5] ng/mL; P < 0.001). To classify ADA status, a cutoff of 1000 ng/mL was applied, as previously reported, which set this threshold to minimize false-positive results and to focus on post-treatment ADA responses [12]. Using this cutoff, 32 patients (37.2%) developed high ADA levels. In contrast, when the cutoff was set at the median, 43 patients (50.0%) exhibited high ADA levels (Supplementary Fig. 1). Clinical outcomes based on antidrug antibody levels We analyzed the clinical outcomes of atezolizumab treatment according to ADA status after three weeks of therapy. To identify patients with markedly elevated ADA levels, patients were categorized using a cutoff of 1000 ng/mL, and treatment outcomes were subsequently compared between the ADA-high and ADA-low groups. Patients with high ADA levels exhibited a markedly poorer response, with 90.6% experiencing PD and none achieving PR or CR. In contrast, among those with low ADA levels, 63.0% had PD, while 20.4% achieved PR or CR (Fig. 1a). Response to atezolizumab treatment and survival outcomes based on ADA status. a Response for atezolizumab treatment according to ADA status. ADA, antidrug antibody; PD, progressive disease; SD, stable disease; PR, partial response; CR, complete response. b Progression-free survival and overall survival based on ADA levels. The analysis was conducted using an ADA cutoff value of 1000 ng/mL Survival analysis further supported these findings. Patients with ADA-high had significantly shorter PFS (hazard ratio [HR] = 1.86, 95% CI: 1.15–3.02, P = 0.010) and OS (HR = 1.92, 95% CI: 1.14–3.23, P = 0.013) compared with ADA-low patients (Fig. 1b). To evaluate the robustness of these results, the same analysis was performed using the median ADA value as an additional cutoff, which yielded similar outcomes (Supplementary Fig. 2a–b). Correlation between ADA status and atezolizumab concentrations To better understand the impact of elevated ADA levels on the immunotherapeutic effectiveness of atezolizumab, we examined atezolizumab concentrations based on ADA status. Following three weeks of atezolizumab treatment, the concentration of atezolizumab exhibited a negative correlation with ADA levels, and there was a notable decrease in atezolizumab concentration in patients with ADA values of 1000 ng/mL or higher (Fig. 2a, b). Even when the cutoff was set to the median, the similar results were obtained (Supplementary Fig. 3). Correlation between ADA and atezolizumab concentration. a Atezolizumab concentrations in relation to ADA status. The analysis was conducted using an ADA cutoff value of 1000 ng/mL. b Changes in atezolizumab concentration according to ADA level Impact of ADA levels on atezolizumab efficacy based on PD-L1 expression Because baseline PD-L1 expression is one of the most important factors in predicting response to atezolizumab, we examined differences in response according to ADA level stratified by PD-L1 status. The objective response rate (ORR) decreased if the patients have ADA higher than 1000 ng/mL (Fig. 3) or median (Supplementary Fig. 4), irrespective of PD-L1 status. The absolute amount of ORR decrement is more prominent in PD-L1 positive patients than PD-L1 negative ones (26.7% vs 12.5% when using 1000 ng/mL and 36.4% vs 15.4% when using median as a cutoff, respectively) (Fig. 3 and Supplementary Fig. 4). Influence of antidrug antibodies on atezolizumab efficacy according to PD-L1 expression. The analysis was conducted using an ADA cutoff value of 1000 ng/mL Multivariate analysis of PFS and OS based on ADA status In the multivariable analysis, variables with a p-value < 0.1 in the univariate analysis were included. High ADA levels remained independently associated with shorter PFS and OS even after adjusting for age, sex, smoking status, PD-L1 expression, albumin, and derived neutrophil-to-lymphocyte ratio (dNLR). The ADA-high group with 1000 ng/mL or higher showed a significantly increased risk of disease progression (adjusted HR = 1.6, 95% CI: 1.0–2.7, P = 0.053) and death (adjusted HR = 1.9, 95% CI: 1.1–3.3, P = 0.015) (Table 2). Consistent results were observed when using the median ADA value as the cutoff (Supplementary Table 1). Table 2Multivariate analysis of PFS and OS based on ADA levels using 1000 ng/mL as cutoffVariablesHR (95% CI)p-valueProgression-free survival Univariate analysisAge < 65 yr versus ≥ 65 yr0.7 (0.4–1.0)0.077SexFemale versus Male0.9 (0.5–1.7)0.722SmokingNever versus Ever1.4 (0.9–2.3)0.123PD-L1 < 1% versus ≥ 1%1.0 (0.6–1.7)0.944Albumin < median versus ≥ median0.7 (0.4–1.0)0.067dNLR < median versus ≥ median1.1 (0.7–1.8)0.592ADA < 1000 ng/mL versus ≥ 1000 ng/mL1.9 (1.2–3.0)0.011 Multivariate analysisAge < 65 yr versus ≥ 65 yr0.7 (0.4–1.1)0.128Albumin < median versus ≥ median0.7 (0.4–1.1)0.088ADA < 1000 ng/mL versus ≥ 1000 ng/mL1.6 (1.0–2.7)0.053VariablesHR (95% CI)p-valueOverall survival Univariate analysisAge < 65 yr versus ≥ 65 yr0.9 (0.5–1.6)0.779SexFemale versus Male2.3 (0.9–5.8)0.072SmokingNever versus Ever1.3 (0.8–2.1)0.340PD-L1 < 1% versus ≥ 1%1.0 (0.5–1.8)0.960Albumin < median versus ≥ median0.5 (0.3–0.9)0.013dNLR < median versus ≥ median1.5 (0.9–2.4)0.151ADA < 1000 ng/mL versus ≥ 1000 ng/mL1.9 (1.1–3.2)0.014 Multivariate analysisSexFemale versus Male2.4 (1.0–6.0)0.064Albumin < median versus ≥ median0.5 (0.3–0.9)0.021ADA < 1000 ng/mL versus ≥ 1000 ng/mL1.9 (1.1–3.3)0.015dNLR, derived neutrophil-to-lymphocyte ratio Multivariate analysis of PFS and OS based on ADA levels using 1000 ng/mL as cutoff dNLR, derived neutrophil-to-lymphocyte ratio Discussion In this study, patients with advanced NSCLC who exhibited a significant increase in ADA levels three weeks after starting atezolizumab administration, compared to baseline levels, showed poor clinical outcomes. The concentration of atezolizumab displayed an inverse correlation with ADA levels, notably decreasing in patients with ADA values exceeding 1000 ng/mL three weeks after atezolizumab administration. These findings suggest that the early development of elevated ADA levels may reduce the efficacy of atezolizumab immunotherapy. The use of immune checkpoint inhibitors plays a significant role in the contemporary treatment of cancer patients [14–17]. However, single-agent treatment often fails to produce sustained responses, and resistance eventually develops [18]. ADA can be considered one of the mechanisms of resistance to ICIs. One study examined the incidences of ADA developing with different immune checkpoint inhibitors [13]. For atezolizumab, pembrolizumab, durvalumab, and avelumab, the highest reported rates of ADA positivity were 54.1%, 2.1%, 2.9%, and 5.9%, respectively, presenting the highest incidence of ADA with atezolizumab [13]. In a previous study of patients with advanced HCC treated with atezolizumab plus bevacizumab, it was reported that most patients did not develop atezolizumab ADAs or showed very low ADA levels three weeks after treatment, whereas a subset exhibited a rapid and robust ADA response [12]. Hence, additional studies focusing specifically on atezolizumab are essential to clarify the impact of ADA formation and guide strategies to mitigate its effects in cancer therapy. The development of ADA seems to occur early, often during the initial drug infusion, and in some cases, as early as the first administration [19]. The early development of ADA during atezolizumab treatment in patients with advanced NSCLC is significant just three weeks post-administration. When we use a stringent cutoff of 1000 ng/mL, 37.2% exhibited high ADA levels, which correlated negatively with PFS and OS. This suggests that high ADA levels may reduce therapeutic efficacy, possibly by neutralizing the drug or accelerating its clearance. These findings underscore the need for early and frequent monitoring of ADA levels, particularly in the initial treatment stages, to manage patients at risk of rapid ADA development. However, in this study, ADA levels were assessed at only two time points (baseline and three weeks after treatment initiation). This design does not allow differentiation between transient ADA positivity and persistent high-titer ADA. Longitudinal monitoring across multiple timepoints would therefore be essential to elucidate ADA kinetics, determine whether early ADA responses are sustained or transient, and better understand their clinical relevance. In addition, although this study focused on patients receiving atezolizumab monotherapy after prior treatments, this population represents a setting that is becoming less common in current clinical practice, as ICIs are increasingly used in the first-line setting and in combination with chemotherapy. Future investigations evaluating ADA development and its clinical relevance in these contemporary treatment contexts would be valuable, as chemotherapy may potentially mitigate the formation of ADA and modify their clinical impact. Collectively, these limitations highlight the importance of future studies aimed at elucidating the mechanisms driving ADA formation and developing strategies to minimize its impact, ultimately improving the therapeutic effectiveness of immune checkpoint inhibitors in clinical practice. It is also important to evaluate the molecular characteristics of patients in whom a negative correlation between ADA levels and response to atezolizumab was not observed. Patients showing no anti-tumor response despite having low ADA levels may originally be due to the inherently low responsiveness to ICI itself; however, it is also possible that persistent exposure to low ADA titers, even if the levels are low, could partially affect therapeutic efficacy. Conversely, patients who exhibit high ADA levels but demonstrate a favorable prognosis may have a higher proportion of low-affinity and non-neutralizing immunoglobulin M-type ADA. Therefore, the duration of ADA production and the varying isotypes of ADA might contribute to the heterogeneous relationship between ADA levels and clinical outcomes. Although this study relied solely on total ADA quantification using an ELISA-based assay, future analyses including neutralizing versus non-neutralizing activities, IgG/IgM subclass distribution, and antibody affinity will be essential to more precisely delineate the biological relevance of ADA. Importantly, multiple mechanisms of resistance to immunotherapy, such as intrinsic tumor factors, host immune contexture, and pharmacokinetic variability, may also contribute to these observations and should be considered in future investigations. In conclusion, our study demonstrated a potential correlation between significantly elevated ADA levels at three weeks and poor clinical outcomes in patients with advanced NSCLC receiving atezolizumab treatment. When assessing the worldwide applicability of ADA incidence and predicting prognosis in atezolizumab administration, our findings highlight the potential to refine patient monitoring and develop tailored clinical interventions, ultimately improving therapeutic outcomes. Supplementary Information Below is the link to the electronic supplementary material. Supplementary file1 (PDF 499 KB) Acknowledgements The authors acknowledge the Korea Research Environment Open Network (KREONET) service and the usage of the Global Science Experimental Data Hub Center (GSDC) provided by Korea Institute of Science and Technology Information (KISTI). This author also thanks Korea Institute of Science and Technology Information (KISTI) for the usage of the Global Science Experimental Data Hub Center (GSDC) and Korea Research Environment Open Network (KREONET/GLORIAD) resources, and for the technological supports. Author contributions MJ was responsible for conceptualization, methodology development, validation, formal analysis, investigation, data curation, visualization, and drafting of the original manuscript. CK contributed to conceptualization, validation, formal analysis, investigation, data curation, manuscript review and editing, and funding acquisition. SK, HY, and WL contributed to methodology development, validation, formal analysis, investigation, and data curation. HL contributed to investigation, provided resources, and supported funding acquisition. HC and SL contributed to conceptualization, provided resources, and was responsible for manuscript review and editing, supervision, project administration, and funding acquisition. Funding This research was supported by the Bio&Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2023–00225255). This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Grant Number: RS-2025–00519956). This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (Grant Number: RS-2020-KH088685). This study was supported by Future Medicine 2030 Project of the Samsung Medical Center [#SMX1250111]. This study was supported by Future Medicine 2030 Project of the Samsung Medical Center [#SMO1250061]. This research was supported by a grant of Korean ARPA-H Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (Grant Number: RS-2024–00512539). This research was supported by a grant of the Korea-US Collaborative Research Fund(KUCRF), funded by the Ministry of Science and ICT and Ministry of Health & Welfare, Republic of Korea (Grant Number: RS-2024–00466776). This work was supported by the National Research Foundation of Korea (NRF) grants funded by the Korean government (MSIT) (NRF-2023R1A2C2004339). This work was supported by the National Research Foundation of Korea [NRF] grants funded by the Korean government [MSIT] (NRF-2023R1A2C2006375). Data availability A data sharing plan was not included in the research protocol and hence data sharing will be conditional upon receiving the approval of the Institutional Review Board of the Samsung Medical Center. A data access agreement will be required for access. Declarations Conflict of interest Hong Jae Chon holds consulting or advisory roles with Eisai, Roche, Bayer, ONO, MSD, BMS, BeOne, Boehringer Ingelheim, Sanofi, Servier, AstraZeneca, and IMBDx, and has received research grants from Roche, BeOne, Guardant Health, Servier, HANDOK Inc., Dong-A ST, and Boryung Pharmaceuticals. The other authors declare no competing interests. Ethical approval This study received approval from the institutional review board of Samsung Medical Center and adhered to the principles of the Helsinki Declaration. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Miran Jang and Chan Kim have contributed equally to this work. Contributor Information Hong Jae Chon, Email: minidoctor@cha.ac.kr. Se-Hoon Lee, Email: shlee119@skku.edu. References Associated Data Supplementary Materials Supplementary file1 (PDF 499 KB) Data Availability Statement A data sharing plan was not included in the research protocol and hence data sharing will be conditional upon receiving the approval of the Institutional Review Board of the Samsung Medical Center. A data access agreement will be required for access.