EARLY ONLINE RELEASE

EARLY ONLINE RELEASE Note: This article was posted on the Archives Web site as an Early Online Release. Early Online Release articles have been peer reviewed, copyedited, and reviewed by the authors. Additional changes or corrections may appear in these articles when they appear in a future print issue of the Archives. Early Online Release articles are citable by using the Digital Object Identifier (DOI), a unique number given to every article. The DOI will typically appear at the end of the abstract. The DOI for this manuscript is doi: 10.5858/arpa.2017-0041-RA The final published version of this manuscript will replace the Early Online Release version at the above DOI once it is available. 2017 College of American Pathologists

Review Article Is There a Role for Programmed Death Ligand-1 Testing and Immunotherapy in Colorectal Cancer With Microsatellite Instability? Part II The Challenge of Programmed Death Ligand-1 Testing and Its Role in Microsatellite Instability-High Colorectal Cancer Esmeralda Celia Marginean, MD, FRCPC; Barbara Melosky MD, FRCPC Context. The world of oncology has changed dramatically in the past few years with the introduction of checkpoint inhibitors and immunotherapy. The promising findings of a small, phase 2 clinical trial that led to the US Food and Drug Administration breakthrough designation and approval of the anti programmed death receptor-1 (PD-1) drug pembrolizumab (Keytruda, Merck, Kenilworth, New Jersey) to treat metastatic/refractory microsatellite instability high colorectal cancer (CRC) has significantly boosted interest in immunomodulatory therapies in microsatellite instability high CRC. Objectives. To review the immune response to cancer and the role of immune checkpoints, focusing on the technical and interpretation challenges of PD-1/programmed death ligand-1 (PD-L1) testing by pathologists and the clinical implications of the test and the therapeutic potential of treating CRC with checkpoint inhibitors. Data Sources. A PubMed review was performed of articles pertaining to CRC, microsatellite instability and mismatch repair systems, molecular classification, immune response, PD-1/PD-L1, and immunotherapy. Conclusions. Exciting success with anti PD-1/PD-L1 and anticytotoxic T-lymphocyte associated protein 4 (CTLA4) checkpoint inhibitors has already been reported in melanoma and in lung and renal carcinomas. Recently, microsatellite instability high CRCs, expressing PD-L1 by immunohistochemistry, regardless of the level of that PD- L1 expression, appeared to respond to checkpoint blockades with anti PD-1 or anti PD-L1 agents, whereas microsatellite-stable tumors were much less responsive. With microsatellite instability routinely tested by most centers, studies that include larger cohorts are required to study the predictive role of PD-1/PD-L1 expression in microsatellite instability high CRC, to assess which immunohistochemistry antibodies to use, to refine the scoring criteria, and to critically analyze the interpretation pitfalls. (Arch Pathol Lab Med. doi: 10.5858/arpa.2017-0041-RA) Although the success of many checkpoint inhibitors has changed the therapeutic paradigm for several cancers, such as melanoma, lung non small cell carcinoma (NSCLC), and renal clear cell carcinoma, many tumors remain highly resistant to immunotherapy. A recent study 1 has demonstrated that a well-known subpopulation of patients with solid tumors achieved excellent and durable responses to Accepted for publication April 26, 2017. From the Department of Pathology, University of Ottawa, Ottawa, Ontario, Canada (Dr Marginean); the Gastrointestinal Pathology Section, The Ottawa Hospital, Ottawa (Dr Marginean); the Department of Medical Oncology, University of British Columbia, Vancouver, Canada (Dr Melosky); and the Department of Oncology, British Columbia Cancer Agency, Vancouver (Dr Melosky). The authors have no relevant financial interest in the products or companies described in this article. Presented in part at the Canadian Anatomic and Molecular Pathology conference; February 3 4, 2017; Whistler, British Columbia, Canada. Reprints: Esmeralda Celia Marginean, MD, FRCPC, Gastrointestinal Pathology Section, The Ottawa Hospital, 501 Smyth Rd, CCW-Room 4251, Ottawa, ON K1H 8L6, Canada (email: cmarginean@toh.on.ca). these new therapies. Those tumors demonstrated alterations in the mismatch repair (MMR) pathway that leads to high levels of microsatellite instability high (MSI-H). In part I of this article, we reviewed MMR deficiency in colorectal cancer (CRC), its definition, diagnostic tools to assess MSI, and the prognostic and the predictive value MSI-H status has in CRC care. In this part, we review briefly the immune response to cancer and the role of immune checkpoints and focus on the technical and interpretation challenges of programmed death receptor-1 (PD-1)/programmed death ligand-1 (PD-L1) testing by pathologists, as well as the clinical implications of the test in CRC, especially in the MSI-H subset, and the therapeutic potential of treating CRC with checkpoint inhibitors. IMMUNE RESPONSE TO CANCER, TUMOR IMMUNOEDITING, AND CHECKPOINT INHIBITION An important role of the immune system is to identify and eliminate tumors. Transformed cells of tumors express antigens, called tumor-associated antigens (TAAs), which are not found on healthy cells. The immune system Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky 1

Figure 1. A, Cancer-immunity cycle. This cycle can be divided into 7 major steps, starting with the release of antigens from the cancer cell and ending with the killing of cancer cells. B, Cancer-immunity, cycle-targeting opportunities. Modified from 3 Chen DS, Mellman I. Oncology meets immunology: the cancer-immunity cycle. Immunity. 2013;39(1):1 10 with permission from Elsevier. Abbreviations: CD, cluster of differentiation; CTLA4, cytotoxic T-lymphocyte-associated protein 4; GM-CSF, granulocyte macrophage colony-stimulating factor; IDO, indoleamine 2,3- dioxygenase; IFN, interferon; IL, interleukin; LAG3, lymphocyte activation gene 3; PD-1, programmed death receptor-1 (programmed death-ligand 1 receptor); PD-L1, programmed death ligand-1; TLR, toll-like receptors; VEGF, vascular endothelial growth factor. recognizes those antigens as not self and mounts an immune response against the tumor cells. There are 3 key cells involved in the immune response: (1) antigen-presenting cells, such as dendritic cells (DCs), which identify and uptake foreign antigens and present them to T cells; (2) T lymphocytes, which become activated by antigen-presenting cells to recognize and destroy the cells containing foreign TAAs; and (3) B lymphocytes, which produce antibodies specific to TAAs. Other immunomodulatory cells include T regulatory cells, natural killer cells, macrophages, and myeloid-derived suppressor cells. All these immune cells, together with stromal cells surrounding tumor cells, form the tumor microenvironment and are involved in immune surveillance of cancers. 2 The generation of immunity to cancer is a cyclic process that can be self-propagating, leading to an accumulation of immune-stimulatory factors that, in principle, should amplify and broaden T cell responses. The cycle is also characterized by inhibitory factors that lead to immune regulatory feedback mechanisms, which can halt the development or limit the immunity. 3 That cycle can be divided into 7 major steps (Figure 1, A), starting with the 2 Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky

release of antigens from the cancer cells, which are captured by DCs (step 1); presentation by DCs of tumor antigens on major histocompatibility complex (MHC) to T cells for processing (step 2), resulting in the priming and activation of effector T cell responses against the cancer-specific antigens (step 3). Activated effector T cells traffic to the tumor bed (step 4), infiltrate the tumor bed (step 5), specifically recognize and bind to cancer cells through the interaction between their T-cell receptors and their antigens bound to the MHC (step 6), and kill their target cancer cells (step 7). 3 The DCs are characterized by expression of MHC class I, class II, and costimulatory molecules (B7, ICAM-1, LFA-1, LFA-3, and CD40). The DCs capture and process exogenous antigens, such as TAAs, which are then presented by MHC class I molecules to cytotoxic CD8 þ T cells or are presented by MHC class II molecules to T helper (Th) CD4 þ cells. The activated cytotoxic T cells kill the tumor cells through effector molecules such as perforin and granzyme B. Th cells secrete interferon (IFN)-c (Th1 cells), and interleukins Il-4 and IL-10 (Th2 cells), which can further sensitize tumor cells to cytotoxic T cells. Efficient antigen presentation by MHC class I and class II molecules on DCs is essential for evoking tumor-specific immune responses. Therefore, one of the aims of immunotherapy is to simultaneously activate CD8 þ cytotoxic T cells (which recognize TAAs) and CD4 þ Th cells. 4 Despite tumor immune surveillance, tumors develop several mechanisms to escape immune recognition and still develop and grow in the presence of a functioning immune system. The updated concept of tumor immunoediting is a more-complete explanation for the role of the immune system in tumor development. 5 The tumor immunoediting concept is divided into 3 phases: designated elimination, equilibrium, and escape. In the elimination phase of cancer, the immune system detects and eliminates tumor cells that have developed because of failed, intrinsic tumor-suppressor mechanisms. The elimination phase can be complete, when all tumor cells are cleared, or incomplete, when only a portion of the tumor cells are eliminated. In the case of incomplete tumor elimination, the theory of immunoediting is that a temporary state of equilibrium can then develop between the immune system and the developing tumor. During that period, the tumor cells most likely either remain dormant or continue to evolve, accumulating further changes (such as DNA mutations or changes in gene expression). As that process continues, the immune system exerts selective pressure by eliminating susceptible tumor clones where possible. The pressure exerted by the immune system during that phase is sufficient to control tumor progression, but eventually, if the immune response still fails to completely eliminate the tumor, the process results in the selection of tumor-cell variants that are able to resist, avoid, or suppress the antitumor immune response, leading to the escape phase. During the escape phase, the immune system is no longer able to contain tumor growth, and a progressively growing tumor results. 6 This process of immunoediting is regulated through a series of checkpoint receptors, including cytotoxic T- lymphocyte associated protein 4 (CTLA4), the programmed death-1 pathway (including PD-1 and its ligand PD-L1), and the lymphocyte activating 3 (LAG3) protein. 7 Immunecheckpoint pathways can modulate T-cell responses by influencing communication between T cells and antigenpresenting cells. Tumor cells can evade the body s immune system by turning it off, just as it begins to mount a response against the tumor cells. 6 The new inhibitors of these immune checkpoints (including US Food and Drug Administration [FDA] approved ipilimumab [Yervoy, Bristol-Myers Squibb, Princeton, New Jersey], nivolumab [Opdivo, Bristol-Myers Squibb], and pembrolizumab [Keytruda, Merck & Company, Whitehouse Station, New Jersey]) interrupt that pathway, leading the body s immune system to avoid getting turned off by immune checkpoints and thereby stimulate antitumor immunity. The immune checkpoints inhibit the T cells at different phases of the immune response: the CTLA4 checkpoint is active in the T-cell activation phase in the lymph nodes (early or priming phase), whereas PD-1/PD-L1/PD-L2 regulate T-cell activity during the effector phase in tissues (Figure 1, B). CTLA4 is expressed only on T cells after they are activated by the MHC T-cell receptor interaction, and they are not expressed on resting T cells. CTLA4 is homologous to the T-cell costimulatory protein CD28, and both molecules bind to CD80 (B7-1) and CD86 (B7-2) on antigen-presenting cells (Figure 2, A). CTLA4 binds CD80 and CD86 with greater affinity and avidity than CD28, thus enabling it to outcompete CD28 for its ligands. CTLA4 transmits an inhibitory signal to T cells, whereas CD28 transmits a stimulatory signal. 8 Targeting CTLA4 by ipilimumab blocks its binding to CD80/CD86 and turns off its inhibitory mechanism, thus allowing T cells to destroy the tumor cells. PD-1/PD-L1/ PD-L2 checkpoints block the interaction of tumor-expressed ligands PD-L1 and PD-L2 or T cell expressed PD-1 receptor (Figure 2, B). This binding of PD-1 to PD-L1 and/or PD-L2 inactivates T cells and allows the tumor cells to potentially evade immune surveillance. 9 In addition to T cells, PD-1 receptor is also expressed on the surface of macrophages and B cells; its 2 ligands-pd-l1 and PD-L2 are expressed by many tumor and inflammatory cells. PD-L1 expression correlates with poor clinical outcome but also with the likelihood of response to targeted immune-checkpoint inhibition. Several therapeutic monoclonal antibodies inhibiting either PD-1 (nivolumab, pembrolizumab) or PD-L1 (MPDL3280A [atezolizumab, Tecentriq, Genentech Oncology, Roche, Nutley, New Jersey], MEDI4736 [durvalumab, AstraZeneca, Cambridge], BMS-936559 [Bristol-Meyers Squibb]) have been developed and are now used for the treatment of various malignancies (eg, metastatic melanoma, non small cell lung carcinoma, renal cell carcinoma, bladder carcinoma, and Hodgkin lymphoma). 2,10,11 14 Compared with previously tested immunotherapies, which included peptide vaccines, DC vaccines, whole tumor vaccines, viral vector vaccines, adoptive cell-transfer therapy, monoclonal antibody immunotherapy, and cytokine therapy, 4 PD-1/PD-L1 inhibitors, when used as monotherapies, appear to shrink tumors in a much greater proportion of patients, resulting in longer response durations, and to have fewer high-grade toxicities. PD-1/PD-L1 IMMUNOTHERAPY IN MSI-H CRC In CRC, the efficacy of immune checkpoint therapy is limited. 15,16 In contrast to microsatellite stable (MSS) CRCs, tumors exhibiting MSI-H have an active immune microenvironment infiltrated by cytotoxic (CD8 þ ) T lymphocytes and activated Th1 cells (CD4 þ ). 2 The MSI-H tumors are thought to possess greater tumor-infiltrating lymphocyte (TILs) Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky 3

Figure 2. A, CTLA4 pathway. CTLA4 primarily regulates T cells during the priming phase (early stage) of activation in the lymph nodes and is thought to function as an off switch, shutting down T-cell activity. Activation of naive T cells requires both TCR (signal 1) and CD28 (signal 2) signaling. Under conditions of suboptimal costimulation, low level expression of CTLA4 could inhibit activation of the T cell, either by directly competing with CD28 for ligand binding and/or by generating inhibitory signals. Under conditions of optimal costimulation, low-level expression of CTLA4 becomes limiting and full T cell activation follows. Surface expression of CTLA4 then increases until it effectively competes with CD28 and mediates inhibitory signals to terminate the T-cell response. Targeting CTLA4 by ipilimumab blocks its binding to CD80/CD86 and turns off its inhibitory mechanism thus allowing T cells to destroy the tumor cells. B, PD-1/PD-L1 pathway. PD-1 regulates T-cell activity during the effector phase of immune response in peripheral tissues. PD-L1 and PD-L2 expressed by tumor cells engage the PD-1 receptor on T cells and thus inactivates them, allowing the tumor cells to escape the immune response. Targeting PD-1 receptor by inhibitors (pembrolizumab, nivolumab) allows the immune cells to destroy tumor cells. Abbreviations: CD, cluster of differentiation; CTLA4, cytotoxic T-lymphocyte-associated protein 4; MHC, major histocompatibility complex; PD-1, programmed death receptor-1 (programmed death-ligand 1 receptor); PD-L1, programmed death ligand-1; PD- L2, programmed death ligand-2; TAA, tumor associated antigen; TCR, T-cell receptor. 4 Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky

densities compared with MSS tumors because of the presence of numerous neoantigens (mutated proteins) created by the high mutational load, typically between 10 and 50 times that of DNA repair-sufficient (MSS) tumors. 1,17 Despite such a hostile microenvironment full of immune cells, MSI-H tumor cells are not eliminated by the immune system because of the cancer-specific upregulation of inhibitory checkpoints, including PD-1, CTLA4, LAG3, and indoleamine 2,3-dioxygenase (IDO). 2,18 This indicates that their active immune microenvironment is counterbalanced by immune-inhibitory signals that resist tumor elimination. 2 These data indicate that MSI-H CRCs are potentially good candidates for checkpoint immunotherapy, as recently shown in a small, phase 2 clinical trial that included 41 patients with metastatic cancer refractory to all treatment, including 11 patients (27%) with MSI-H CRC. 1 Three cohorts of patients (N ¼ 41) were enrolled and treated with pembrolizumab 10 mg/kg every 2 weeks. Cohort A included 11 patients (27%) with MSI-H CRC. Cohort B included 21 patients (51%) with MSS CRC. Cohort C included 9 patients (22%) with MMR-deficient cancers other than CRC. The primary endpoint for cohorts A and B were immune-related objective response rates and the 20-week progression-free survival (PFS). The primary endpoint for cohort C was PFS at 20 weeks. In cohort A (MSI-H CRC), the immune objective response rate was 40% (4 of 10 patients; 95% CI, 12 74), with a PFS at 20 weeks of 78% (7 of 9 patients; 95% CI, 40 97). In cohort C (MSI-H other cancers), the immune objective response rate was 71% (5 of 7 patients; 95% CI, 29 96), with a PFS at 20 weeks of 67% (4 of 6 patients; 95% CI, 22 96). In cohort B (MSS CRC), the immune objective response rate was 0% (0 of 18 patients; 95% CI, 0 20), with a PFS at 20 weeks of 11% (2 of 8 patients; 95% CI, 1 35). Interestingly, the patients in cohort C (MSI-H other cancers) had faster response rates than patients in cohort A. Equally fascinating, the 6 of 11 patients (55%) in cohort A (MSI-H CRC) with sporadic cancer (non Lynch syndrome) all responded (100%; 6 of 6 patients) compared with the 27% response rate (3 of 11 patients) in the tumors associated with Lynch syndrome. It was suggested that the greatly increased number of mutation-associated neoantigens resulting from MMR deficiency was the basis for the enhanced anti PD-1 responsiveness of this genetically defined subset of cancers. The conclusion by the authors was that MMR status predicted a clinical benefit of immune-checkpoint blockade with pembrolizumab. This led the FDA to grant, in November 2015, a breakthrough-therapy designation to the anti PD-1 drug pembrolizumab for the treatment of metastatic/refractory MSI-H CRC. PATHOLOGY ASSESSMENT FOR PD-1/PD-L1 BY IHC The PD-1 receptor is a type-1 membrane protein of the immunoglobulin superfamily, and as mentioned before, it is expressed on the surface of T and B cells, natural killer cells, DCs, and macrophages and is overexpressed on the surface of exhausted T cells. The binding of PD-L1 and PD-L2 ligands to the PD-1 receptor blocks T-cell mediated immune response to the tumor. Among the ligands belonging to the B7 family (which include PD-L1, PD-L2, B7-H3, and B7-H4), PD-L1 is one of the most important membrane-inhibitory ligands and the most studied in lung cancer clinical trials. The recent approval of pembrolizumab by the FDA to treat patients with metastatic NSCLC, based on PD-L1 IHC expression, lead to an exponential increase in pathology IHC testing with various PD-1/PD-L1 antibodies. Of particular relevance to pathologists, one study 12 showed that patients with NSCLC whose tumors had membranous PD-L1 expression (using PD-L1 22C3, Dako, Carpinteria, California) in 50% or more of the malignant cells by IHC were significantly more likely to respond to pembrolizumab than those with less than 50% expression. In this study, out of 495 patients who received pembrolizumab, 182 patients were assigned to the training group and 313 patients to the validation group to define a PD-L1 cutoff. After evaluation of several methods for pathological assessment, membranous PD-L1 expression in at least 50% of tumor cells (proportion score, 50%) was selected as the cutoff on the basis of the ease of use and ROC analysis. Among patients with a PD-L1 proportion score of at least 50% in the validation group (n ¼ 73), the response rate was 45.2%; median progression-free survival (PFS) was 6.3 months (95% CI, 2.9 to 12.5) for all patients, 6.1 months (95% CI, 2.1 to 12.5) for previously treated patients, and 12.5 months (95% CI, 2.4 to 12.5) for previously untreated patients. PFS and OS were shorter among patients with a PD-L1 proportion score of 1 49% or a score of,1% than among those with a score.50%. The median duration of response was 12.5 months (range, 2.1 to 23.3) for a PD-L1 proportion score of at least 50%, 7.2 months (range, 1.4 to 8.3) for a proportion score of 1 to 49%, and not reached (range, 1.0 to 10.8) for a proportion score of less than 1%. The FDA approved the combination of antibody clone (PD-L1 22C3 pharmdx test, Dako) and detection system (Autostainer Link 48, Dako) as companion diagnostic tests for selecting patients with lung cancer for pembrolizumab therapy. In contrast, another study 19 showed that response rates to nivolumab were significantly greater in patients with nonsquamous NSCLC, showing only 1% or greater tumor cell PD-L1 positivity using the same Dako detection system but a different antibody clone (28-8, Abcam, Cambridge, Massachusetts). Response rates in PD-L1 þ patients in that trial were 31% to 52%, but notably up to 16% of PD-L1 patients also showed treatment response, indicating that PD-L1 expression enriches for responders, but the absence of expression was not an absolute indicator of the lack of benefit. Published abstracts from trials of other PD-L1 inhibitors, such as atezolizumab (Tecentriq) in the Patients With NSCLC Who Progressed on Post-Platinum (POPLAR) trial 20 and durvalumab (MEDI4736), 21 describe the use of different PD-L1 IHC testing platforms (Roche and Ventana Medical Systems [Tucson, Arizona], respectively) and different antibody clones (SP142 and SP263, respectively). Furthermore, the POPLAR trial adds yet another complexity to the biomarker-scoring approach by suggesting that PD- L1 expression on tumor-infiltrating immune cells, not only tumor cells, may also predict response. The Table summarizes the PD-1/PD-L1 antibodies currently used, their scoring criteria, and their diagnostic cutoffs. The study that led to the quick FDA approval of anti-pd-1 drug pembrolizumab for the treatment of metastatic/ refractory MSI-H CRC assessed expression of CD8 and PD-L1 with a cutoff of 5% for PD-L1 of tumor cells and assessed the percentage of PD-L1 þ TILs and macrophages within the tumor and at the invasive fronts. 1 They identified membranous PD-L1 expression only in patients with MMRdeficient cancers, prominent on TIL and tumor-associated Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky 5

Programmed Death-Ligand 1 Assays and Inhibitors Criteria Pembrolizumab a Nivolumab b Durvalumab c Atezolizumab d Therapeutic target PD-1 PD-1 PD-L1 PD-L1 PD-L1 IHC assay Dako 22C3 Dako 28-8 Ventana SP263 Ventana SP142 Cell type scored NSCLC-TC, UC-TC/IC, NSCLC-TC NSCLC-TC NSCLC-TC/IC, UC-IC CRC-TC/IC Cutoff criteria (NSCLC) TC ¼ 1% 49%; TC 50% TC 1%; TC 5%; TC 10% TC 25% TC or IC 1%; TC or IC 5%; TC 50% or IC 10% Cutoff criteria (UC) 1% TC or any N/A N/A IC 10%; IC 5%; IC 1% stromal staining Cutoff criteria (MSI-H CRC) TC, 5%; TC 5%; IC any positivity (%) N/A N/A N/A Abbreviations: IC, immune cells infiltrating the tumor or at invasive front; IHC, immunohistochemistry; MSI-H CRC, colorectal cancer with high microsatellite instability; NSCLC, non small cell lung carcinoma; PD-1, programmed death receptor-1 (programmed death-ligand 1 receptor); PD- L1, programmed death ligand-1; TC, tumor cell; UC, urothelial carcinoma; N/A, not available. a Merck, Kenilworth, New Jersey. b Bristol Meyers Squibb, Princeton, New Jersey. c AstraZeneca/MedImmune, Cambridge, United Kingdom. d Genentech Oncology, Roche (Nutley, New Jersey), on a Ventana Medical Systems (Tucson, Arizona) or Dako (Carpinteria, California) platform. macrophages, located at the invasive fronts of the tumor (Figure 3, A through H). The expression of CD8 and PD-L1 was not significantly associated with PFS or overall survival but was predictive of response to pembrolizumab therapy. They also found that the number of somatic mutations in all tumors tested was positively correlated with PFS, suggesting that mutation load may serve as an important biomarker for MSI-H CRC immunotherapy outcomes. Another predictive correlate of MSI-H CRC with pembrolizumab response was seen when comparing patients who had germline MMR mutations (eg, Lynch syndrome) versus those who did not; all 6 patients (100%) with sporadic MSI-H CRC had an objective response, whereas only 3 of 11 patients (27%) with germline MSI-H CRC (Lynch syndrome) had treatment responses. This could perhaps be due to the finding that germline MSI-H CRCs generally average a lower number of frameshift mutations than do other MSI-H CRCs but may also be related to the differing pathogeneses (eg, methylation patterns). The authors also postulated that other MSI-H cancers, including carcinomas of the uterus, stomach, bile duct, pancreas, ovary, prostate, and small intestine, may benefit from anti PD-1 therapy, and patients whose tumors contain other DNA repair deficiencies, such as those with mutations in POLD, POLE, or MYH, might benefit as well. 22 The hypothesis that MSI-H tumors stimulate the immune system is not a new idea; it has been supported by observations of the dense, immune infiltration and Th1- Figure 3. A through H. Immunohistochemistry of CD8 and PD-L1 expression in microsatellite instability high (MSI-H) and microsatellite stable (MSS) colorectal cancer (CRC). The invasive front (yellow dashed line) from an MSI-H CRC (patient 16 [A through C], patient 19 [D]) and MSS CRC (patient 3, bottom, [E through H]). The yellow dashed line separates tumor (T) and normal (N) tissue. There is marked expression of PD-L1 (blue arrows) and CD8 (brown dots) in the patient with an MSI-H tumor (top panels, [B and C]), whereas there is very little expression of either marker in the MSS tumor (bottom panels, [F and G]). Representative images of tumor-infiltrating lymphocytes in another MSI-H CRC (patient 19, top [D]) and MSS CRC (patient 3, bottom [H]), immunolabeled with an antibody to CD8 (brown dots). Note the infiltration of CD8 cells in the MSI-H. From 1 Le DT, Uram JN, Wang H, et al. PD-1 blockade in tumors with mismatch-repair deficiency. N Engl J Med. 2015;372(26)(supplementary appendix, figure S6):2509 2520. Reprinted with permission from Massachusetts Medical Society. Abbreviations: CD, cluster of differentiation; MMR, mismatch repair; PD-1, programmed death receptor-1 (programmed death-ligand 1 receptor); PD-L1, programmed death ligand-1 (hematoxylin-eosin [H&E], original magnification 310 [A and E]; PD-L1, original magnification 310 [B and F]; CD8, original magnifications 310 [C, D, and G] and 320 [H]). 6 Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky

Figure 4. PD-L1 expression in microsatellite instability high (MSI-H) colorectal cancer (CRC). A and B, Strong membranous expression is shown, especially along the tumor-stromal interface. C and D, Diffuse membranous expression within the tumor, with some heterogeneity noted (PD-L1 clone E1L3N, Cell Signaling Technology). Immunostains performed on in-house MSI-H colon cancers at PhenoPath Laboratories (Seattle, Washington), courtesy of Allen Gown, MD. Abbreviation: PD-L1, programmed death ligand-1, (original magnifications 32 [A and C] and 320 [B and D]). associated, cytokine-rich environment in MMR-deficient tumors.1 A recently published study23 of 389 CRC obtained from 395 patients, of which 17% (n = 68) were MSI-H, investigated the patterns and prognostic relevance of PD1/PD-L1 IHC expression. They used different clones from the ones mentioned before (PD-L1 clone E1L3N, Cell Signaling Technology, Danvers, Massachusetts; and PD-1 clone NAT105, Cell Marque, Rocklin, California on the Ventana Benchmark Ultra platform), and they assessed PD1 positivity in TILs and stromal lymphocytes (scores 0 3, based on number of TILs/10 high-power field) and PD-L1 in tumor cells and immune cells. For both markers, both membranous and cytoplasmic staining was regarded as positive. They observed that tumor cell staining for PD-L1 occurred in 2 patterns: primarily along the tumor-stromal interface (84%) (Figure 4, A and B) or diffusely (16%) (Figure 4, C and D). They also noted that MSI-H status correlated significantly with PD-1 and PD-L1 expression by both staining intensity and the percentage of cell staining (P,.001 for both). By maximizing the sensitivity and specificity of PD-L1 or PD-1 expression in predicting MMR status through receiver-operating characteristic curve analyses, they derived the following criteria to separate Arch Pathol Lab Med tumors into high and low PD-L1 and PD-1 expression categories. High-level PD-L1 corresponded to 41% of tumor cells staining with 2þ intensity; high PD-1þ TILs corresponded to a measure of more than 1.43 TILs of 1þ or 2þ intensity per square millimeter. With those criteria, 18% and 50% of MSI-H tumors (n ¼ 68) exhibited high-level PD-L1 and high PD-1þ TILs, respectively; on the other hand, those rates for MMR-proficient/MSS tumors were only 2% and 13%, respectively (P,.001 for both). Overall, from all 389 tumors assessed by IHC, the rate of high PD-1þ TILs was 19% (n ¼ 76 patients), and high-level PD-L1 expression in the tumor cells was 5% (n ¼ 19 patients). In their study,23 high PD-1þ TILs and high-level PD-L1 tumor cells correlated with the various clinicopathologic parameters that were more often associated with MSI-H status, including medullary morphology, right-sided location, numerous TILs, marked peritumoral Crohn-like lymphoid reaction, and younger age. They also noted that, in MSI-H CRC, the combined high-level tumor PD-L1 expression and high-level PD-1þ TILs were associated with a significantly worse recurrence-free survival. However, the positive prognostic effect of PD-1þ TILs in patients with MSI-H CRC was negated by the presence of high-level PDL1 in the tumor cells. Neither PD-1 nor PD-L1 expression PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky 7

was correlated with recurrence-free survival in MSS colon cancer. To summarize, there are now several PD-L1 antibodies and IHC assays well described in the literature and used in various cancers: Dako 22C3, Dako 28-8, Cell Marque NAT105, Ventana SP263, Ventana SP142, and Cell Signaling E1L3N. The therapeutic target is either the PD-1 receptor on tumor cells (for pembrolizumab and nivolumab) or the PD-L1 ligand on T cells (for durvalumab and atezolizumab) (Table). In the face of this proliferation of PD-L1 IHC clones, staining platforms, and various scoring criteria, pathologists must decide on the feasibility of introducing a newly approved companion diagnostic assay that may require purchase of not only a specific antibody kit but also a particular staining platform. Given that the FDAapproved Dako 22C3 companion diagnostic test became available only in October 2015 and that the 22C3 clone cannot be purchased apart from the approved kit, many laboratories may find it more practical and economical to use institutional laboratory-developed tests, which are readily available, can be validated in house, and are far less costly than commercial antibodies. 24 Alternatively, individual laboratories may choose to offer a single, companion diagnostic assay tailored to the practices and preferences of their requesting oncologists and retain the option of sending samples to other laboratories that offer different assays. To reduce the chance of false-negative results with PD-L1 immunostaining, more concerted efforts to cross-compare performance of available antibodies and protocols are needed. Because of the FDA American Association for Cancer Research American Society of Clinical Oncology sponsored workshop Complexities in Personalized Medicine: Harmonizing Companion Diagnostics Across a Class of Targeted Therapies (March 24, 2015) a blueprint project to evaluate the comparability of the various PD-L1 assays in 39 cases was recently published. 25 The results of that feasibility study (the Blueprint Project) indicated that there were both similarities and differences in the 4 PD-L1 assays for dynamic ranges, cell types stained, and overall staining characteristics. Overall, 3 of the assays (28-8, 22C3, and SP263) were similar in analytic staining performance, as assessed by percentage of tumor cells showing tumor cell membrane staining. The SP142 assay generally stained fewer tumor cells. However, that study did not address the specificity and sensitivity of the assays or provide clinicaloutcome comparisons from classifications of the PD-L1 status of the patient s tumor. The relative similarities among 3 of the 4 assays does not mean the assays are interchangeable for clinical use because PD-L1 IHC was tested in only a few cases and was read by experts in the field at a single test site. They concluded that larger studies with more cases and more pathologists were warranted. Based on those limited data, the authors concluded that (1) the thresholds used to select a specific treatment should never be interchanged, and (2) more data are required to inform on the validity of using alternative staining assays upon which to read different, specific therapy-related PD-L1 cut points. In a separate effort, the National Comprehensive Cancer Network is collaborating with Bristol-Myers Squibb to assess variability across assays, heterogeneity within individual samples, and concordance of pathologist interpretation. 26 Besides the availability and cost of various antibodies, pathologists are faced with various cutoffs for positivity; very low cutoffs may fail to maximize the differences in response between positive and negative groups. Sampling errors, tumoral heterogeneity, or testing of tissue obtained at diagnosis, rather than at time of progression or relapse, may underestimate or overestimate the percentage of neoplastic cells showing PD-L1 expression. Moreover, PD-L1 expression is dynamic and may change after targeted therapy and/ or chemotherapy/radiation therapy. Another challenge is tissue availability for IHC testing, which may be an issue in NSCLC, where most samples are obtained by fine-needle aspirate and testing prioritizes for epidermal growth factor receptor and anaplastic lymphoma kinase standard-of-care testing before PD-L1, which may deplete the tissue; this is not an issue in resected CRC. CONCLUSIONS The world of oncology has changed dramatically in the past few years with the introduction of checkpoint inhibitors and immunotherapy. The promising findings of the small, phase 2 clinical trial that lead to FDA approval of the anti PD-1 drug pembrolizumab to treat metastatic/refractory MSI-H CRC has significantly boosted interest in immunomodulatory therapies in MSI-H CRC. 1 Among the many immune inhibitory ligand-receptor pairs (checkpoints) and metabolic enzymes discovered to date, CTLA4, PD-1/PD-L1, LAG3, and IDO are of particular interest because inhibitory antibodies or drugs (in the case of IDO) are currently in active clinical testing to enhance therapeutic antitumor responses. Of note, all of those checkpoint molecules were greatly elevated in all 3 compartments (TIL, stroma, and invasive front) of MSI-H tumors, whereas MSS tumors and their infiltrating and invasive front lymphocytes express very little of those checkpoint molecules. 2 What is strikingly different in MSI- H CRC from melanoma, renal cancer, or lung cancer is that there is little PD-L1 expressed by the tumor cells; rather, the PD-L1 is expressed predominantly by infiltrating immune cells, T cells, and myeloid cells (which may be identified by coexpressing CD163). 2 Regardless of the level of PD-L1 IHC expression, MSI-H tumors appear to respond to checkpoint blockade with agents such as anti PD-1 or anti PD-L1 antibodies, whereas MSS tumors are much less responsive. 1,2 The increased expression of non PD-1/PD-L1 immune checkpoints is suggestive that other immune checkpoint inhibitors can have favorable effects in the treatment of MSI-H CRC. Currently, several clinical trials testing anti PD-1 antibodies in patients with MSI-H stage IV CRC have been initiated, using checkpoint inhibitors, such as pembrolizumab, durvalumab, atezolizumab, nivolumab, in combination with anti-tla4 ipilimumab as well as combinations of pembrolizumab with other therapies, such as p53 vaccines, Janus kinase 1 (JAK1) inhibitors, and/or phosphoinositide 3-kinase (PI3K-d) inhibitors. 9 Further elucidation of immunotherapy in MSI-H CRC in clinical trials that include larger cohorts may be applicable to other malignancies as well, including MSI-H gastric, ovarian, endometrial, prostate, bile duct, and pancreatic cancers. Now, membranous IHC expression of PD-L1 in NSCLC is widely viewed as the best biomarker for identifying patients who will respond to PD-1/PD-L1 checkpoint inhibitors. 8 Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky

Recently, the FDA approved a companion diagnostic test in association with pembrolizumab therapy in advanced NSCLC (Dako PD-L1 clone 22C3 pharmdx test and detection system Autostainer Link 48), and, a novel concept, a complementary diagnostic test in association with nivolumab therapy in advanced squamous and nonsquamous NSCLC (Dako PD-L1 clone 28-8), although PD-L1 expression analysis by IHC was not required in this last setting. Although the conventional companion diagnostics are typically linked to a specific drug within its approved label, complementary diagnostics are associated more broadly, usually not with a specific drug but with a class of drugs and are not confined to specific uses by labeling (can be used off label ), with consequent ramifications for economic, regulatory, and clinical considerations. Testing of MSI-H CRC by IHC with various PD-L1 antibodies, unlike lung cancer, has not been as extensively evaluated. However, a few studies showed that PD-L1 testing by IHC is feasible in CRC, and their high-level expression correlated with microsatellite status and histologic features usually associated with MSI-H and predicted clinical response to anti PD-1/PD-L1 inhibitors. 1,23 Larger studies are required to study the predictive role of PD-L1 expression in MSI-H CRC, to assess which antibody to use, to refine the scoring criteria and critically analyze the interpretation pitfalls and, eventually, to evaluate whether PD-1/PD-L1 testing will be a complementary diagnostic test in CRC as it is in lung cancer. Pathologists have to cope with the rapid expansion of immunotherapies, new novel targets, and therapeutic strategies. The surgical pathologists practice must change from the traditional morphology-based routine to a fully clinical-oriented interaction with medical oncologists, surgeons, and radiotherapists, that is, to become multidisciplinary. That change is fundamental to carefully selecting those patients who, based on clinical and tumor features, are the best candidates for immunotherapeutic approaches and should receive IHC testing for multiple targets. References 1. Le DT, Uram JN, Wang H, et al. PD-1 blockade in tumors with mismatchrepair deficiency. N Engl J Med. 2015;372(26):2509 2520. 2. Llosa NJ, Cruise M, Tam A, et al. The vigorous immune microenvironment of microsatellite instable colon cancer is balanced by multiple counter-inhibitory checkpoints. Cancer Discov. 2015;5(1):43 51. 3. Chen DS, Mellman I. Oncology meets immunology: the cancer-immunity cycle. Immunity. 2013;39(1):1 10. 4. Koido S, Ohkusa T, Homma S, et al. Immunotherapy for colorectal cancer. World J Gastroenterol. 2013;19(46):8531 8542. 5. Dunn GP, Bruce AT, Ikeda H, Old LJ, Schreiber RD. Cancer immunoediting: from immunosurveillance to tumor escape. Nat Immunol. 2002;3(11):991 998. 6. Swann JB, Smyth MJ. Immune surveillance of tumors. J Clin Invest. 2007; 117(5):1137 1146. 7. Lee V, Murphy A, Le DT, Diaz LA Jr. Mismatch repair deficiency and response to immune checkpoint blockade. Oncologist. 2016;21(10):1200 1211. 8. Krummel MF, Allison JP. CD28 and CTLA-4 have opposing effects on the response of T cells to stimulation. J Exp Med. 1995;182(2):459 465. 9. Quiroga D, Lyerly HK, Morse MA. Deficient mismatch repair and the role of immunotherapy in metastatic colorectal cancer. Curr Treat Options Oncol. 2016;17(8):41. 10. de Guillebon E, Roussille P, Frouin E, Tougeron D. Anti program death-1/ anti program death-ligand 1 in digestive cancers. World J Gastrointest Oncol. 2015;7(8):95 101. 11. Chung C, Christianson M. Predictive and prognostic biomarkers with therapeutic targets in breast, colorectal, and non small cell lung cancers: a systemic review of current development, evidence, and recommendation. J Oncol Pharm Pract. 2014;20(1):11 28. 12. Garon EB, Rizvi NA, Hui R, et al; KEYNOTE-001 Investigators. Pembrolizumab for the treatment of non small-cell lung cancer. N Engl J Med. 2015;372(21):2018 2028. 13. Hamid O, Carvajal RD. Anti-programmed death-1 and anti-programmed death-ligand 1 antibodies in cancer therapy. Expert Opin Biol Ther. 2013;13(6): 847 861. 14. Rijavec E, Genova C, Alama A, et al. Role of immunotherapy in the treatment of advanced non small-cell lung cancer. Future Oncol. 2014;10(1):79 90. 15. Brahmer JR, Tykodi SS, Chow LQ, et al. Safety and activity of anti-pd-l1 antibody in patients with advanced cancer. N Engl J Med. 2012;366(26):2455 2465. 16. Brahmer JR, Drake CG, Wollner I, et al. Phase I study of single-agent antiprogrammed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics, and immunologic correlates. J Clin Oncol. 2010; 28(19):3167 3175. 17. Smyrk TC, Watson P, Kaul K, Lynch HT. Tumor-infiltrating lymphocytes are a marker for microsatellite instability in colorectal carcinoma. Cancer. 2001; 91(12):2417 2422. 18. Gatalica Z, Snyder C, Maney T, et al. Programmed cell death 1 (PD-1) and its ligand (PD-L1) in common cancers and their correlation with molecular cancer type. Cancer Epidemiol Biomarkers Prev. 2014;23(12):2965 2970. 19. Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus docetaxel in advanced nonsquamous non small-cell lung cancer. N Engl J Med. 2015; 373(17):1627 1639. 20. Spira AI, Park K, Mazières J, et al. Efficacy, safety and predictive biomarker results from a randomized phase II study comparing MPDL3280A vs docetaxel in 2L/3L NSCLC (POPLAR) [abstract 8010]. J Clin Oncol. 2015;33(15)(suppl):8010. 21. Rebelatto MC, Mistry A, Sabalos C, et al. Development of a PD- L1companion diagnostic assay for treatment with MEDI4736 in NSCLC and SCCHN patients [abstract 8033]. J Clin Oncol. 2015;33(15)(suppl):8033. 22. Palles C, Cazier JB, Howarth KM, et al. Germline mutations affecting the proofreading domains of POLE and POLD1 predispose to colorectal adenomas and carcinomas [published correction appears in Nat Genet. 2013;45(6): 713].Nat Genet. 2013;45(2):136 144. 23. Lee LH, Cavalcanti MS, Segal NH, et al. Patterns and prognostic relevance of PD-1 and PD-L1 expression in colorectal carcinoma. Mod Pathol. 2016; 29(11):1433 1442. 24. Sholl LM, Aisner DL, Allen TC, et al; Members of Pulmonary Pathology Society. Programmed death ligand-1 immunohistochemistry a new challenge for pathologists: a perspective from members of the Pulmonary Pathology Society. Arch Pathol Lab Med. 2016;140(4):341 344. 25. 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. 26. NCCN collaborates with Bristol-Myers Squibb to study PD-L1 expression and test interpretation in lung cancer [news release]. Fort Washington, PA: National Comprehensive Cancer Network; November 30, 2015. http://www. nccn.org/about/news/newsinfo.aspx?newsid¼559. Accessed December 3, 2015. Arch Pathol Lab Med PD-L1 Testing and Role in MSI-H CRC Marginean & Melosky 9