Determination of PD-L1 Expression in Circulating Tumor Cells of NSCLC Patients and Correlation with Response to PD-1/PD-L1 Inhibitors

cancers Article Determination of PD-L1 Expression in Circulating Tumor Cells of NSCLC Patients and Correlation with Response to PD-1/PD-L1 Inhibitors Melanie Janning 1,2,† , Franca Kobus 1,2,† , Anna Babayan 2 , Harriet Wikman 2 , Janna-Lisa Velthaus 1,2 , Sonja Bergmann 2 , Stefanie Schatz 3 , Markus Falk 3 , Lars-Arne Berger 1 , Lisa-Marie Böttcher 1,2 , Sarina Päsler 1,2 , Tobias M. Gorges 2,‡ , Linda O’Flaherty 2,4 , Claudia Hille 2 , Simon A. Joosse 2 , Ronald Simon 5 , Markus Tiemann 3 , Carsten Bokemeyer 1 , Martin Reck 6 , Sabine Riethdorf 2 , Klaus Pantel 2 and Sonja Loges 1,2, * 1 2 3 4 5 6 * † ‡ Department of Oncology, Hematology and Bone Marrow Transplantation with section Pneumology, Hubertus Wald University Comprehensive Cancer Center Hamburg, University Medical Center Hamburg-Eppendorf, 20246 Hamburg, Germany; (M.J.); (F.K.); (J.-L.V.); (L.-A.B.); (L.-M.B.); (S.P.); (C.B.) Department of Tumor Biology, Center of Experimental Medicine, University Medical Center Hamburg-Eppendorf, 20246 Hamburg, Germany; (A.B.); (H.W.); (S.B.); (T.M.G.); (L.O.); (C.H.); (S.A.J.); (S.R.); (K.P.) Institute of Hematopathology Hamburg HpH, 22547 Hamburg, Germany; (S.S.); (M.F.); (M.T.) Boehringer Ingelheim Pharma GmbH & Co. KG, 88397 Biberach, Germany Institute of Pathology, University Medical Center Hamburg-Eppendorf, 20246 Hamburg, Germany; Lung Clinic Grosshansdorf, Airway Research Center North, German Center of Lung Research, 22927 Grosshansdorf, Germany; Correspondence: ; Tel.: +49-40-7410-51962 These authors contributed equally to this work. Current address: AstraZeneca GmbH, 22880 Wedel, Germany. Received: 27 May 2019; Accepted: 12 June 2019; Published: 17 June 2019



 Abstract: Circulating tumor cells (CTCs) hold great potential to answer key questions of how non-small cell lung cancer (NSCLC) evolves and develops resistance upon anti-PD-1/PD-L1 treatment. Currently, their clinical utility in NSCLC is compromised by a low detection rate with the established, Food and Drug Administration (FDA)-approved, EpCAM-based CellSearch® System. We tested an epitope-independent method (ParsortixTM system) and utilized it to assess PD-L1 expression of CTCs from NSCLC patients. We prospectively collected 127 samples, 97 of which were analyzed with the epitope-independent system in comparison to the CellSearch system. CTCs were determined by immunocytochemistry as intact, nucleated, CD45− , pankeratins (K)+ cells. PD-L1 status of CTCs was evaluated from 89 samples. With the epitope-independent system, ≥1 CTC per blood sample was detected in 59 samples (61%) compared to 31 samples (32%) with the EpCAM-based system. Upon PD-L1 staining, 47% of patients harbored only PD-L1+ CTCs, 47% had PD-L1+ and PD-L1− CTCs, and only 7% displayed exclusively PD-L1− CTCs. The percentage of PD-L1+ CTCs did not correlate with the percentage of PD-L1+ in biopsies determined by immunohistochemistry (p = 0.179). Upon disease progression, all patients showed an increase in PD-L1+ CTCs, while no change or a decrease in PD-L1+ CTCs was observed in responding patients (n = 11; p = 0.001). Our data show a considerable heterogeneity in the PD-L1 status of CTCs from NSCLC patients. An increase of PD-L1+ CTCs holds potential to predict resistance to PD-1/PD-L1 inhibitors. Cancers 2019, 11, 835; doi:10.3390/cancers11060835 www.mdpi.com/journal/cancers Cancers 2019, 11, 835 2 of 16 Keywords: NSCLC; circulating tumor cells; PD-1/PD-L1 inhibition; resistance 1. Introduction Blockade of the interaction of programmed death ligand 1 (PD-L1) and its receptor PD-1 has revolutionized treatment of patients with non-small cell lung cancer (NSCLC) [1]. Therapy with different anti-PD-1/PD-L1 antibodies was approved in NSCLC in the first- and second-line setting. A correlation between PD-L1 expression by tumor or immune cells and efficacy of anti-PD-1/PD-L1 antibodies has been confirmed in various trials. However, the predictive valus of PD-L1 immunohistochemistry (IHC) is still controversial since activity has also been observed in patients with PD-L1-negative tumors [2,3]. One possible explanation for these findings could be that a single tissue biopsy does not adequately reflect the heterogeneity of PD-L1 expression in stage IV NSCLC patients with multiple tumor sites. Furthermore, dynamic changes of PD-L1 expression in tumor cells might occur before or under therapy with PD-1/PD-L1 inhibitors, leading to different sensitivity to PD-1/PD-L1 blockade which would similarly be missed by one single biopsy. Circulating tumor cells (CTCs) hold promise to better reflect the tumor heterogeneity compared to tissue biopsies because they originate from different tumor sites. In addition, they could lead to important insights on how tumor cells become resistant to immune therapy because they can be analyzed longitudinally as liquid biopsies [4]. One major challenge remains the specific detection of CTCs in cancer patients. The semiautomated, EpCAM/pankeratins(K)-dependent CellSearch® system is the only CTC detection method approved by the Food and Drug administration (FDA). While this system proved clinical utility in breast, prostate, and colorectal cancer, its usefulness in NSCLC remains challenging [5]. Few reports suggest that in NSCLC, one or more CTCs can be detected by CellSearch in only 30% of patients compared to 50% and 57% of patients with more than five CTCs in breast and prostate cancer, respectively [6,7]. Thus, alternative methods for CTC detection are warranted, especially in NSCLC, in order to allow clinically meaningful applications in a representative number of patients. Especially based on recent reports on CTCs with lower EpCAM expression and increased CTC heterogeneity, one strategy is to use label-independent, microfluidic devices [8,9]. In this prospective study, we aim to evaluated the label-independent, microfluidic ParsortixTM system (ANGLE plc., Guildford, United Kingdom), which selects CTCs based on size and rigidity in a cohort of 127 samples form NSCLC patients [10]. To the best of our knowledge, this is the first study using the label-independent Parsortix system for CTC detection in a large cohort of NSCLC patients. We also compared CTC detection directly to the CellSearch system, which is considered the gold-standard, using matched samples. In a second step, we established a workflow to detect PD-L1+ CTCs in Parsortix-enriched samples in order to shed light on the largely unanswered question about heterogeneity of PD-L1 expression by CTCs in NSCLC. Furthermore, we assessed the PD-L1 expression by CTCs under therapy with anti-PD-L1/PD-1 antibodies. 2. Results 2.1. CTC Enumeration With the label-independent Parsortix system, we detected ≥1 CD45− /K+ cell in 59 and ≥3 CD45− /K+ cells in 26 out of 97 samples (60.8% and 26.8%, respectively, range: 1–54 CTCs, median = 2, Table 1, matched cohort and Figure 1A). Figure 2 shows representative images of CD45− /K+ (...truncated)