Abstract 4919: Required numbers of biopsy samples and tumor cells for the assessment of PD-L1 expression to predict a response to nivolumab treatment in patients with NSCLC

Abstract

Abstract Background: PD-L1 expression is a predictive biomarker of response to nivolumab in non-squamous non-small cell lung carcinoma (Non-Sq NSCLC). PD-L1 expression in NSCLC exhibits significant intratumoral heterogeneity. Although multiple biopsy samples are usually obtained during a biopsy procedure, the evaluation of PD-L1 expression is typically performed using only one small sample. The aim of this study was to investigate the required numbers of samples and tumor cells (TCs) to assess PD-L1 expression using biopsy samples to predict the response to nivolumab. Methods: A total of 225 biopsy samples obtained from 80 Non-Sq NSCLC patients treated with nivolumab between January 2016 and August 2017 were collected. In each patient, the samples were obtained from the same lesion at the same time before nivolumab treatment. The number of TCs and the PD-L1 score after staining with anti-PD-L1 antibody (Dako 28-8) were evaluated. We defined the sample containing the largest number of TCs as Max-TCs, the sample containing the smallest number of TCs as Min-TCs and the sum of all the samples from each patient as Multiple-TCs. The concordances of the PD-L1 scores among the Max-TCs, Min-TCs and Multiple-TCs were assessed. And then, the impacts of the numbers of samples and TCs on the prediction of the efficacy of nivolumab using the evaluation of PD-L1 expression were evaluated. Results: The median number of samples was 2 (range: 2-6). The median number of TCs was 148 (13-2976) for the Mini-TCs, 570 (29-4826) for the Max-TCs, and 1022 (55-7062) for the Multiple-TCs.The correlation of the PD-L1 scores between the Max-TCs and Multiple-TCs was significantly high (R2 = 0.98). A mismatch of PD-L1 <1%/≥1% was observed between the Max-TC and the Multiple-TC in one patient (1%). The correlation of the PD-L1 scores between the Max-TCs and the Mini-TCs (R2 = 0.83) was weaker than that between the Max-TCs and the Multiple-TCs. A mismatch of PD-L1 <1%/≥1% was observed between the Max-TCs and the Mini-TCs in six patients (8%). The ROC analysis indicated that the required number of TCs to match PD-L1 <1%/≥1% between the Max-TCs and the Mini-TCs was 80 (AUC: 0.62). PD-L1 ≥1% in Mini-TCs containing the ≥80 TCs was associated with a longer progression-free survival (PFS) (median 8.7 vs 1.8 months, HR 0.39, P < 0.01) and a prolonged overall survival (OS) (median 20.9 vs 9.2 months, HR 0.41, P = 0.01) compared with PD-L1 <1%. On the other hand, there were no difference in the PFS (median 3.0 vs 2.4 months, HR 0.77, P = 0.59) and the OS (median 7.7 vs 5.6 months, HR 0.89, P = 0.81) between PD-L1 ≥1% and PD-L1 <1% in the Mini-TCs containing <80 TCs. Conclusion: Evaluating the PD-L1 score using multiple biopsy samples did not provide any added benefit. One biopsy sample containing ≥80 TCs is required to evaluate PD-L1 expression to predict the response to nivolumab. Citation Format: Jun Sato, Tomoyuki Naito, Hibiki Udagawa, Hidehito Horinouchi, Shuji Murakami, Yasushi Goto, Shintaro Kanda, Yutaka Fujiwara, Noboru Yamamoto, Yuichiro Ohe, Yoshitaka Zenke, Keisuke Kirita, Shingo Matsumoto, Kiyotaka Yoh, Seiji Niho, Noriko Motoi, Genichiro Ishii, Koichi Goto. Required numbers of biopsy samples and tumor cells for the assessment of PD-L1 expression to predict a response to nivolumab treatment in patients with NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4919.