Optimal Evaluation of Programmed Death Ligand-1 on Tumor Cells Versus Immune Cells Requires Different Detection Methods

Optimal Evaluation of Programmed Death Ligand-1 on Tumor Cells Versus Immune Cells Requires Different Detection Methods Kelly A. Schats, PharmD; Emily A. Van Vré, PhD; Carolien Boeckx, PhD; Martine De Bie, BSc; Dorien M. Schrijvers, PhD; Bart Neyns, MD, PhD; Ingrid De Meester, PharmD, PhD; Mark M. Kockx, MD, PhD Context. The benefit of programmed death ligand-1 (PD-L1) immunohistochemistry (IHC) as a method to select patients who may benefit from programmed death receptor-1 (PD-1)/PD-L1 immunotherapies remains uncertain in many tumor indications. Objectives. To compare the commercially available, approved PD-L1 IHC assays (22C3, 28-8, SP142, SP263), specifically identifying the changes in staining output created by altering the detection method. Design. This pilot study investigates the respective PD- L1 kit assay staining patterns and related scoring of tumor cells and immune cells on lung carcinoma and melanoma. Furthermore, the influence of the detection method (platform and related reagents) on PD-L1 antibody performance is studied. Results. The SP142 kit reveals more immune cell staining but less tumor cell staining than the other PD-L1 kits. Alternatively, the 22C3 and 28-8 kits show good tumor cell sensitivity, but less pronounced immune cell staining, even in tonsil. Tumor cell staining by the SP263 kit is comparable to that of 22C3 and 28-8 kits, while immune cell staining is better. Strikingly, the selection of the detection method has a major impact on the sensitivity of the assay for PD-L1 detection per cell type. Switching the detection method of the kits could largely circumvent the observed staining differences. Conclusions. The diverse sensitivities caused by the choice of the detection method should be taken into consideration when selecting PD-L1 kits or developing PD- L1 IHC laboratory-developed tests. When using alternative kits or laboratory-developed tests, it is strongly recommended to reestablish their clinical utility per therapeutic agent or compare them with the original kit. (Arch Pathol Lab Med. 2018;142:982 991; doi: 10.5858/ arpa.2017-0159-oa) Many promising results were obtained by using anti programmed death receptor-1 (PD-1) or anti programmed death ligand-1 (PD-L1) therapy in different tumor types, and several PD-1/PD-L1 blocking agents are already approved by the US Food and Drug Administration (FDA). Some of these are in combination with a companion or complementary PD-L1 immunohistochemistry (IHC) assay, which was developed in parallel with the drug (overview in Table 1). 1,2 However, not all studies showed association between PD-L1 expression and better clinical response. Additionally, the predictive value of PD-L1 cannot easily be compared between different clinical trials, as different PD- L1 IHC assays were used. Therefore, there is an urgent need to perform comparison studies by using the commercially available kits with the prominent PD-L1 antibody clones 22C3, 28-8, SP142, and SP263. The aim of this pilot study is to compare the staining pattern of 4 PD-L1 IHC kits. The assay performance, particularly the sensitivity of PD-L1 antigen detection, was studied on tonsil, a selection of non small cell lung cancer, and metastatic melanoma samples. Accepted for publication October 24, 2017. Published as an Early Online Release April 2, 2018. Supplemental digital content is available for this article at www. archivesofpathology.org in the August 2018 table of contents. From the Departments of Immunohistochemistry (Drs Schats, Van Vré, Boeckx, and Schrijvers and Ms De Bie) and Molecular Pathology (Dr Kockx), HistoGeneX, Antwerp, Belgium; the Department of Medical Biochemistry, University of Antwerp, Antwerp, Belgium (Drs Schats and De Meester); and the Department of Medical Oncology, Universitair Ziekenhuis Brussel, Belgium (Dr Neyns). MATERIALS AND METHODS KAS, EAVV, CB, MDB, and DMS are employees of HistoGeneX NV. MMK is CEO of HistoGeneX NV, which performs immunohistochemical and molecular testing for pharmaceutical companies as Control and Tumor Tissue part of (preclinical) clinical studies that evaluate new anticancer Formalin-fixed paraffin-embedded tonsil, placenta, and 7 drugs. The authors have no other relevant affiliations or financial melanoma metastasis samples were obtained in accordance involvements with any organization or entity with a financial interest with the Helsinki Declaration of 1975, following patient in or financial conflict with the subject matter or materials discussed privacy procedures and approval by the hospital ethics in the manuscript apart from those disclosed. Corresponding author: Kelly A. Schats, PharmD, HistoGeneX NV, committee (EC/PC/avl/2016.003 and CME 2010/266). Two Prinses Josephine Charlottelaan 10, 2600 Antwerp, Belgium (email: additional non small cell lung cancer (1 adenocarcinoma kelly.schats@histogenex.com). and 1 squamous cell carcinoma) and 2 melanoma resection 982 Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al

Table 1. Human Programmed Cell Death Ligand-1 (PD-L1) Immunohistochemistry (IHC) Assays and Applications PD-L1 Clone 22C3 28-8 SP142 SP263 Antibody host Mouse Rabbit Rabbit Rabbit Immunogen (Source: data sheet) Used in combination with therapeutic antibody (target) Commercial name therapeutic agent (Company) FDA-approved diagnostic IHC assay FDA-approved indications and, if applicable, cutoff for treatment (% PD-L1 þ cells) Human extracellular domain of PD-L1 (Phe19-Thr239) fused to a human IgG1 fragment Pembrolizumab (PD-1) Keytruda (MSD Inc, Kenilworth, New Jersey) Purified recombinant containing the extracellular domain (Phe19-Thr239) of human PD-L1 Synthetic peptide derived from the C-terminus of protein Nivolumab (PD-1) Atezolizumab (PD-L1) Durvalumab (PD-L1) Opdivo (BMS, New York, New York) Tecentriq (Genentech, South San Francisco, California) Synthetic peptide derived from the C-terminus of protein Imfinzi (AZ, Cambridge, United Kingdom) Nivolumab (PD-1) Opdivo (BMS, New York, New York) Pembrolizumab (PD-1) Keytruda (MSD Inc, Kenilworth, New Jersey) Companion Complementary Complementary Complementary...... Metastatic NSCLC NSCLC Metastatic NSCLC... NSCLC Metastatic NSCLC 50% TCs 1% TCs 50% TCs or 10% ICs Metastatic melanoma Melanoma............ 1% TCs Classical Hodgkin lymphoma Classical Hodgkin lymphoma............ MSI-H solid tumor............... HNSCC HNSCC............ 1% TCs Advanced urothelial bladder cancer Urothelial cancer Metastatic urothelial cancer Locally advanced or metastatic urothelial cancer...... 1% TCs 5% ICs 25% TCs or ICs... RCC............ Advanced gastric cancer HCC previously treated with sorafenib Abbreviations: AZ, AstraZeneca; BMS, Bristol Myers Squibb; FDA, US Food and Drug Administration; HCC, hepatocellular cancer; HNSCC, head and neck squamous cell cancer; ICs, immune cells; IgG1, immunoglobulin G1; MSD, Merck Sharp & Dohme; MSI-H, microsatellite instable high; NSCLC, non small cell lung cancer; PD-1, programmed cell death-1; Phe, phenylalanine; RCC, renal cell cancer; TCs, tumor cells; Thr, threonine. Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al 983

Table 2. Human Programmed Cell Death Ligand-1 (PD-L1) Immunohistochemistry (IHC) Assay Staining Protocols as Described in the Kit Insert PD-L1 Clone 22C3 28-8 SP142 SP263 Ab host Mouse Rabbit Rabbit Rabbit Immunogen (Source: data sheet) Human extracellular domain of PD-L1 (Phe19-Thr239) fused to a human IgG1 fragment Purified recombinant containing the extracellular domain (Phe19-Thr239) of Synthetic peptide derived from the C-terminus of protein Synthetic peptide derived from the C-terminus of protein Immunohistochemistry protocol kit insert Vendor kit Dako North America Inc, Ventana Medical Systems Inc, Tucson, Arizona Carpinteria, California IHC platform Autostainer Link 48 Ventana Benchmark Ultra HIER TRS low ph (ph 6.1) 48 min CC1 (ph 8.5) 64 min CC1 (ph 8.5) Blocking Peroxidase block Peroxidase block Ab concentration Prediluted in kit (ready to use) Ab incubation condition Room temperature, 30 min 368C, 16 min Amplification 30 min mouse linker 30 min rabbit linker 8 min amplification No Detection system 30 min dextran coupled with peroxidase OptiView molecules and GtARb and GtAMs immunoglobulins DAB enhancer 5 min... Abbreviations: Ab, antibody; CC, cell condition; DAB, 3,3 0 -diaminobenzidine; GtAMs, goat anti-mouse; GtARb, goat anti-rabbit; HIER, heatinduced epitope retrieval; IgG1, immunoglobulin G1; Phe, phenylalanine; Thr, threonine; TRS, target retrieval solution. samples were acquired from Bioreclamation IVT (Chestertown, Maryland). The comparison was performed on serial 4-lm slides to minimize the impact of biological variability. Immunohistochemistry All 4 PD-L1 PharmDx kits 22C3, 28-8 (both from Dako, Carpinteria, California); and SP142, SP263 (both from Ventana Medical Systems Inc, Tucson, Arizona) underwent an implementation validation on their assigned kit platform. Each staining run was qualified according to the guidelines described in the kit insert. Technical details of the PD-L1 kits are summarized in Table 2. Additionally, kit protocols were adapted by switching the detection methods (platform and IHC reagents; Table 3). In-house optimized and validated CD8 (C8/144B clone, Dako), CD163 (MRQ26 clone, Ventana Medical Systems), CD4 (1F6 clone, Leica Biosystems Richmond Inc, Richmond, Illinois), and CD68 (KP-1 clone, Ventana Medical Systems) IHC assays were performed on Ventana Benchmark XT platforms. The CD11c (5D11 clone, Leica Biosystems Richmond) IHC assay was Table 3. Human Programmed Cell Death Ligand-1 (PD-L1) Immunohistochemistry (IHC) Assays and Staining Protocols Using Another Detection System PD-L1 Clone 22C3 28-8 SP142 SP263 Ab host Mouse Rabbit Rabbit Rabbit Immunogen (Source: data sheet) Vendor primary antibody Human extracellular domain of PD-L1 (Phe19-Thr239) fused to a human IgG1 fragment Dako North America Inc, Carpinteria, California Purified recombinant containing the extracellular domain (Phe19-Thr239) of Immunohistochemistry adapted protocol Abcam, Cambridge, United Kingdom Synthetic peptide derived from the C-terminus of protein Spring Bioscience, Pleasanton, California Synthetic peptide derived from the C-terminus of protein IHC platform Ventana Benchmark XT Lab Vision Autostainer 480S HIER 64 min CC1 (ph 8.5) 30 min TRS (ph 9) Blocking Peroxidase block 3% hydrogen peroxide Ab concentration 1/50 in Dako antibody diluent with background-reducing components 1/75 in Ventana disco diluent 1/100 in Ventana disco diluent Ventana Medical Systems Inc, Tucson, Arizona 1/4 in Ventana disco diluent Ab incubation condition 368C, 60 min Room temperature, 60 min Amplification 8 min Ventana amplification 15 min Dako rabbit linker 15 min Dako rabbit linker Detection system Ventana OptiView 20 min Dako Flex Envision DAB enhancer... 5 min Abbreviations: Ab, antibody; CC, cell condition; DAB, 3,3 0 -diaminobenzidine; HIER, heat-induced epitope retrieval; IgG1, immunoglobulin G1; Phe, phenylalanine; Thr, threonine; TRS, target retrieval solution. 984 Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al

Figure 1. Overview of immunohistochemistry staining of serial sections in tonsil. Immune cell characterization (CD11c, CD68, CD163, CD4, and CD8) in tonsil deep crypt epithelium (A through E, respectively) and in germinal center (F through J, respectively). PD-L1 staining pattern using the kit protocols (22C3, 28-8, SP142, and SP263) in tonsil deep crypt epithelium (K through N, respectively) and germinal center (O through R, respectively). The staining pattern of PD-L1 with 22C3, 28-8, and SP263 kits is more macrophage-like in contrast to apparent lymphocyte staining by SP142 kit (insets) (original magnifications 35 [A through R]) and 360 [insets O through R]). Abbreviation: PD-L1, programmed death ligand-1. performed on a Lab Vision autostainer 480S (Thermo Scientific, Waltham, Massachusetts). PD-L1 IHC Evaluation All stained slides were scanned as whole slide images by using a digital slide scanner (3DHISTECH, Budapest, Hungary). The PD-L1 staining images were compared in identical regions of serial slides. A pathologist (24 years in practice), trained and qualified for scoring the PD-L1 assays, blinded for the applied protocol, scored the slides sequenarch Pathol Lab Med Vol 142, August 2018 tially for percentage of PD-L1 positive immune cells (ICs) and percentage of PD-L1 positive tumor cells (TCs). Only the membranous PD-L1 staining at all intensities was scored in the complete tumor area. Statistical Evaluation Statistical computations were performed with MedCalc (version 12.3.00, Ostend, Belgium). Spearman correlations (q) above 0.85 are considered as strong. Different Detection Methods for Evaluation of PD-L1 Schats et al 985

Figure 2. Comparison of PD-L1 immunohistochemistry staining patterns in NSCLC (A through D; I through L) and melanoma (E through H; M through P) tumor samples, using PD-L1 kits 22C3, 28-8, SP142, and SP263. High PD-L1 expression on TCs or ICs is equally detected in some cases (A through H). In other cases, sensitivity for TC or IC staining is different (I through P) (original magnifications 35 [A through H] and 310 [I through P]). Abbreviations: IC, immune cell; NSCLC, non small cell lung cancer; PD-L1, programmed death ligand-1; TC, tumor cell. RESULTS Implemented PD-L1 Kits Despite published validation details, all PD-L1 kits underwent an implementation validation. 3 5 The 22C3, 28-8, SP142, and SP263 PD-L1 kits demonstrate comparable staining of syncytiotrophoblasts in placenta, interferon c stimulated peripheral blood mononuclear cells, and Hodgkin lymphoma 2 (HDLM-2) cell pellets, while none stained the prostate cancer 3 (PC-3) cell pellets (supplementary Figure 1, A through T; see supplemental digital content, containing 2 figures at www.archivesofpathology.org in the August 2018 table of contents). In tonsil (Figure 1, A through R), the deep crypt epithelium was positively stained with all assays but at different intensities (Figure 1, K through N). The SP263 assay, in particular, shows different intensity grades in contrast to the rather homogenous intensity of the 22C3, 28-8, and SP142 assays. As expected, all clones show dominant immunoreactivity in the germinal centers (Figure 1, O through R) and some scattered cells in the paracortical regions. However, within the germinal center there are differences in the number and cell types that are positively stained with each PD-L1 assay. The SP142 clone tends to stain a combination of macrophages/dendritic cells (positive for CD68 and CD11c, respectively) and lymphocytes (positive for CD4 and CD8). Contrarily, the 22C3, 28-8, and SP263 clones stain predominantly macrophage- or dendritic cell like cells (Figure 1, A through J). Within the tumor samples, some regions show strong immunoreactivity and similar staining patterns (TCs and/or ICs) with all 4 PD-L1 kits (Figure 2, A through H), while other regions show clear differences for TC and IC staining (Figure 2, I through P). The PD-L1 TC staining is not always detected with the SP142 assay, while it is clearly visible with the other assays (22C3, 28-8, SP263). Conversely, the SP142 assay tends to stain more ICs and with a sharper membrane delineation. The 22C3 and 28-8 assays show a less intense, fainter IC staining. Moreover, the dominant TC staining obtained with 22C3 and 28-8 assays seems to mask the intraepithelial PD-L1 positive ICs (Figure 2, I through L), rendering them difficult to distinguish. In accordance with these observations, a good correlation between the 22C3, 28-8, and SP263 kits was observed for TC as well as IC scoring (all q. 0.80) (supplementary Figure 2, A), while SP142 IC/TC correlations with the other kits were less than 0.80. 986 Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al

Figure 3. Comparison of PD-L1 staining pattern in deep crypt epithelium (A through D) and germinal center (E through H) of tonsil, stained with the PD-L1 kit protocol (A and B; E and F) and the adapted Linker þ Envision Flex (LþF) protocol (C and D; G and H) for both SP142 and SP263 (original magnification 35 [A through H]). Abbreviation: PD-L1, programmed death ligand-1. Adapted PD-L1 Detection Methods Lab Vision Staining Platform: Linker þ Envision Flex Detection. The SP142 and SP263 primary antibodies were tested on the Lab Vision staining platform using the Linker þ Envision Flex (LþF) detection method, which is used in the 22C3 and 28-8 kits. In tonsil (Figure 3, A through H), less intense IC and epithelial crypt staining is detected with the SP142 adapted protocol, while the opposite accounts for the SP263 adapted protocol. In tumor (Figure 4, A through P), as evident from Figure 4, C through P, the LþF detection method predisposed both the SP142 and SP263 kits toward detection of PD-L1 on TCs with the sharpest effect seen with SP142. The crisp IC staining with SP142 in its original kit format was no longer visible with the SP142 LþF detection method, because of improved PD-L1 TC staining, leading to shielding of the intratumoral PD-L1 positive ICs (Figure 4, A through N). The adapted SP142 assay resulted in TC and IC scoring more comparable with that of 28-8 and 22C3 kit assays (improved correlation; q approximately 0.8 for TCs and 0.9 for ICs). For SP263, the correlation with 28-8 and 22C3 scoring remained high for both TCs and ICs (q approximately 0.9) (supplementary Figure 2, B). Ventana Staining Platform: OptiView þ Amplification Detection. Pairing the Dako antibodies (22C3 and 28-8) with the Ventana OptiView þ Amplification (OþA) detection method remarkably improved the detection of PD-L1 on ICs (Figures 5, A through L; and 6, A through R). In both tonsil and tumor samples, a more and sharper delineated IC staining was observed. When compared with SP142 in particular, the IC correlation coefficients increased (supplementary Figure 2, C). On the other hand, the OþA application to the Dako antibodies did not alter TC staining. In addition, the effect of the amplification technology on the SP263 antibody was tested and showed an increased sensitivity of SP263 (Figure 6, C and F). DISCUSSION This pilot study aims to unravel the differences among the PD-L1 kits in order to assist pathology laboratories in diagnostic test selection, being either an alternative PD-L1 kit or laboratory-developed test. Unsurprisingly, we confirmed that 22C3, 28-8, SP142, and SP263 PD-L1 kits have different TC and IC sensitivities. The 4 PD-L1 IHC kits (22C3, 28-8, SP142, and SP263) revealed different IC staining patterns with 22C3, 28-8, and SP263 detecting more macrophage- and dendritic cell like cells, while SP142 identified more lymphocyte-like cells. The SP142 assay is clearly more sensitive for IC detection in tonsil and tumor. These results are in line with the publication of Scheel et al, 6 Adam et al, 7 and the FDA blueprint results. 8 In contrast, TCs are best detected with 22C3, 28-8, and SP263 kits. It is noteworthy that these differences in IC and/or TC sensitivity could not be observed on cell controls or placenta tissue (often suggested as staining control). Thus, validation of PD-L1 IHC staining requires additional inclusion of tonsil and tumor samples. Unexpectedly, the applied detection method (platform and IHC reagents) influences the identification of PD-L1 positive TCs or ICs and alludes to a major role of the detection method regardless of the PD-L1 antibody clone. Differences between the kits, for the most part, could be circumvented by switching to another detection method. The SP142 LþF protocol showed more TC staining than the kit insert and was more concordant with the 22C3 and 28-8 kits. The PD-L1 IC staining with the SP142 LþF assay also correlated better with both Dako kits, presumably because the increased TC staining shielded PD-L1 detection on the intratumor ICs. Our group 9 previously demonstrated that a Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al 987

Figure 4. Comparison of PD-L1 staining pattern (IC and TC staining) in identical regions of interest of tumor samples stained with the PD-L1 kit protocol (A through D) or the adapted Linker þ Envision Flex (LþF) protocol (E through H). In addition, the PD-L1 IC staining patterns were compared to CD8 (I through L) and CD163 (M through P) staining patterns (original magnification 310 [A through P]). Abbreviations: IC, immune cell; PD-L1, programmed death ligand-1; TC, tumor cell. sensitive detection method (eg, the OþA system) is needed for the detection of PD-L1 positive ICs. Thus, detection of PD-L1 on ICs in both tonsil and tumor specimens could now be observed with 22C3 and 28-8 antibodies when they were paired with the OþA detection method. In this configuration, both 22C3 and 28-8 retained their sensitivity for PD-L1 TC expression. Finally, when the SP263 antibody was paired with the OþA method, we noted an increase in the detection of PD-L1 on ICs. Many laboratories are accustomed to the assembly of their own laboratory-developed assays, each selecting its preferred antibodies and detection methods, based on available platforms and costs. 10 With PD-L1 IHC, this practice might create unintended results. To our knowledge, this pilot study shows for the first time that the detection method used in the PD-L1 IHC assay determines the staining pattern of a specific cell type or cellular compartment. Changing the detection method has important consequences on TC and/or IC sensitivity, affecting the clinical cutoff values and subsequently, patients PD-L1 positivity status. This study has shown that the different PD-L1 kits are not interchangeable. Therefore, it is recommended that the kits be used in accordance to their insert with proven clinical utility. Alternatively, when opting for an alternative PD-L1 kit or laboratory-developed test, these data stress comparison with the original kit for which clinical utility was established in combination with the therapeutic compound. Finally, optimal IC or TC detection may require different IHC detection methods; for instance, an LþF protocolfor sensitive TC detection and an OþA for sensitive IC detection. This obviously has broad implications for patient selection and warrants further investigation, in combination with the different scoring methods, to determine whether this approach might lead to an improved predictive value of the PD-L1 IHC assays for PD-1/PD- L1 targeted therapies. On the other hand, a single biomarker like PD-L1 might not be sufficient to predict response to immune therapy. There might be a need for multiple biomarkers to properly evaluate the complex tumor microenvironment. This project was funded by the IWT O&O grant No. 130894, awarded by the Agency for Innovation by Science and Technology (IWT) Belgium. 988 Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al

Figure 5. Comparison of the PD-L1 staining pattern in deep crypt epithelium (A through F) and germinal center (G through L) of tonsil, stained with PD-L1 kit protocol (A through C; G through I) and the adapted OptiView þ Amplification (OþA) protocol (D through F; J through L) for 22C3, 28-8, and SP263 (original magnification 35 [A through L]). Abbreviation: PD-L1, programmed death ligand-1. Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al 989

Figure 6. Comparison of the PD-L1 staining pattern (IC and TC staining) in identical regions of interest in tumor samples stained with PD-L1 kit protocol (A through C; M through O) versus the adapted OptiView þ Amplification (OþA) protocol (D through F; P through R). In addition, the PD- L1 IC staining patterns were compared to the CD8 (G through I) and CD163 (J through L) staining patterns (original magnification 35 [A through R]). Abbreviations: IC, immune cell; PD-L1, programmed death ligand-1; TC, tumor cell. 990 Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al

References 1. Reck M, Rodríguez-Abreu D, Robinson AG, et al. Pembrolizumab versus chemotherapy for PD-L1 positive non small-cell lung cancer. N Engl J Med. 2016;375(19):1823 1833. 2. Rosenberg JE, Hoff J, Powles T, et al. Atezolizumab in patients with locally advanced and metastatic urothelial carcinoma who have progressed following treatment with platinum-based chemotherapy: a single-arm, multicentre, phase 2 trial. Lancet. 2016; 387(10031):1909 1920. 3. Roach C, Zhang N, Corigliano E, et al. Development of a companion diagnostic PD-L1 immunohistochemistry assay for pembrolizumab therapy in non-small-cell lung cancer. Appl Immunohistochem Mol Morphol. 2016;24(6): 392 397. 4. Rebelatto MC, Midha A, Mistry A, et al. Development of a programmed cell death ligand-1 immunohistochemical assay validated for analysis of nonsmall cell lung cancer and head and neck squamous cell carcinoma. Diagn Pathol. 2016;11(1):95. 5. Phillips T, Simmons P, Inzunza HD, et al. Development of an automated PD-L1 immunohistochemistry (IHC) assay for non-small cell lung cancer. Appl Immunohistochem Mol Morphol. 2015;23(8):541 549. 6. Scheel AH, Dietel M, Heukamp LC, et al. Harmonized PD-L1 immunohistochemistry for pulmonary squamous-cell and adenocarcinomas. Mod Pathol. 2016;29(10):1165 1172. 7. Adam J, Rouquette I, Damotte D, et al. PL04a.04: Multicentric French Harmonization study for PD-L1 IHC testing in NSCLC. J Thorac Oncol. 2017; 12(1):S11 S12. 8. Hirsch FR, Mcelhinny A, Stanforth D, et al. PD-L1 immunohistochemistry assays for lung cancer: results from phase 1 of the Blueprint PD-L1 IHC Assay Comparison Project. J Thorac Oncol. 2017;12(2):208 222. 9. Schats KA, Van Vré EA, De Schepper S, et al. Validated PD-L1 immunohistochemistry assays (E1L3N and SP142) reveal similar immune cell staining patterns in melanoma when using the same sensitive detection system. Histopathology. 2017;70(2):253 263. 10. Yaziji H, Taylor CR. PD-L1 assessment for targeted therapy testing in cancer: urgent need for realistic economic and practice expectations. 2017;25(1):1 3. Arch Pathol Lab Med Vol 142, August 2018 Different Detection Methods for Evaluation of PD-L1 Schats et al 991