Expression of PD-L1 and other immunotherapeutic targets in thymic epithelial tumors

Introduction The thymus is a critical organ for the development of the adaptive immune system and thymic epithelial tumors (TETs; thymomas and thymic carcinomas) are often associated with auto-immune paraneoplastic conditions. However, the immunobiology of TETs is not well described. An evaluation of the tumor microenvironment, with particular focus on expression of immunotherapeutic targets, may facilitate and prioritize development of immunotherapy strategies for patients with TETs. Methods Tumor tissues from 23 patients with WHO Type B2/B3 thymoma (n = 12) and thymic carcinoma (n = 11) were identified and clinical outcomes were annotated. The expression of membranous PD-L1 on tumor cells, CD3+ and CD8+ tumor infiltrating lymphocytes (TILs), co-stimulatory (CD137, GITR, ICOS), and co-inhibitory immune checkpoint molecules (PD-1, CTLA-4, TIM-3) were assessed semi-quantitatively using immunohistochemistry. Results PD-L1 positivity (≥ 25% of tumor membrane expression) was frequent in TETs (15/23, 65%), more common in thymomas compared to thymic carcinomas (p<0.01), and was associated with longer overall survival (p = 0.02). TIM-3 and GITR were expressed in all TETs, including 18/23 and 12/23 with at least moderate/high expression, respectively. Moderate/high CD137 expression correlated with CD8+ (p = 0.01) and moderate/high GITR expression co-associated with PD-1 (p = 0.043). Conclusions TETs are characterized by frequent PD-L1 expression and PD-L1 is associated with improved survival, suggesting PD-L1 signaling may be biologically important in TETs. Robust expression of markers of immune activation and immunotherapeutic target molecules in TETs emphasizes the potential for development of anti-PD-1/PD-L1 therapies.


Introduction
The thymus is a critical lymphoid organ in the development of the adaptive immune system during childhood, but through gradual involution becomes largely atrophic in adults. In relatively rare cases, residual epithelial tissue can become neoplastic. Thymic epithelial tumors (TETs) are classified as thymomas (WHO types A, AB, B1, B2, and B3) or thymic carcinomas (WHO type C) based on the morphology of epithelial tumor cells and proportion of lymphocytic involvement. [1] While thymomas are frequently associated with a wide range of autoimmune paraneoplastic disorders [2] these disorders are rarely seen in patients with thymic carcinoma. [3] Standard initial treatment regimens for patients with advanced TETs include those with anthracycline and/or platinum [4][5][6]. After progression of disease, targeted therapies have been evaluated but are associated with modest response rates. [7][8][9] Given the importance of the thymus in the development of the adaptive immune system, there has been interest in examining the potential role of immune checkpoint therapies in TETs. However, as these therapies can also be associated with potentially severe auto-immune toxicity, there we sought to assess the immune microenvironment prior to more empiric approaches to evaluating these therapies which may enable more efficient drug development.
We hypothesized that evaluation of the tumor microenvironment of TETs, with a particular focus on targets with therapeutic agents currently in development, may reveal subgroups of distinct immunophenotypes and inform the safe and effective development of relevant immunotherapy agents for patients with TETs. We examined this hypothesis by evaluating the presence of PD-L1 using a clinically validated antibody, assessing for membranous staining. In addition to PD-L1, other co-inhibitory and co-stimulatory molecules were evaluated, as well as quantifying subsets of tumor-infiltrating lymphocytes (TILs). We also sought to characterize the interplay between these multiple markers and T cell subsets including CD3 and CD8 expression. This analysis was performed in a series of TETs with histologic features [10] associated with high risk for recurrence and poor prognosis thereby focusing on a group most likely to require systemic therapy, and be enrolled in future clinical trials with immunotherapeutic agents.

Methods
Following approval by the Memorial Sloan Kettering Cancer Center institutional review board (IRB), pathologic specimens of high grade TETs (WHO Type B2 and B3 thymoma and thymic carcinoma) from patients treated at Memorial Sloan Kettering Cancer Center were identified. Diagnosis was based on WHO classification (2004 edition) and was confirmed by a thoracic pathologist (A.M., N.R.).
Associations between clinicopathologic factors, survival, immune checkpoint molecule expression, and TIL populations were described. Overall survival following initial diagnosis was estimated using Kaplan-Meier methodology and compared between groups by log-rank test. Association between categorical variables was assessed by Fisher's exact test.

Results
The clinical and pathologic features of patients with TETs and their associated tumors are summarized in Table 1 with individual details outlined in S1 Table. Most thymomas (9/12) were WHO Type B3. The majority of patients presented with Masaoka Stage III, IVA, or IVB TETs at the time of initial diagnosis (17/23). 2 patients (both with a diagnosis of thymoma), had a history of myasthenia gravis.
Sixty-five percent of TETs (15/23, 95% CI: 43 to 84%) were PD-L1+ using a cutoff PD-L1M-score ! 25%. PD-L1 positivity was more common in thymomas compared with thymic carcinomas (11/12 vs. 4/11, p<0.01) as summarized in Table 2. There was no significant difference in PD-L1 positivity when comparing samples from patients according to stage, tumor size, or site of sample (primary vs metastatic). There was also no significant difference in PD-L1 positivity, CD8 or CD3 + TILs, or expression of other co-stimulatory and co-inhibitory markers when comparing samples from patients prior to chemotherapy or after chemotherapy exposure.

Discussion
Immune checkpoint inhibitors are now widely used in a variety of different cancer types [14][15][16], however only a subset of patients sustain clinical responses and determining the predictors of response remains a critical, ongoing effort. Accordingly, when considering new cancer indications to explore, identifying rational ways to prioritize development of the multiple immunotherapies possible to study will be critical to the development of appropriate and efficient clinical trials. In that light, we begin to profile TETs for PD-L1 expression and additional immunotherapeutic markers. We find that PD-L1 expression is frequent in TETs, particularly Table 2

Thymoma
Thymic thymomas and those patients with PD-L1 positive TETs have distinctly improved survival compared to PD-L1 negative TETs. The finding that PD-L1 expression is associated with improved survival has been noted in other tumor types as well. For example, in patients with non-small cell lung cancers (NSCLCs), PD-L1 expression was associated with improved overall survival independent of age and stage of disease [17] and also associated with improved overall survival in early stage patients [18]. From a mechanistic perspective of the association PD-L1 and improved survival, PD-L1 expression may generally be a surrogate marker for an "inflamed" tumor microenvironment, which is consistent with an effective host response to the tumor and improved survival as a result of native anti-tumor immunity. This association between high PD-L1 expression and improved benefit from PD-1/PD-L1 blockade in other tumor types [19] raises the possibility that anti-PD-1/PD-L1 drugs may be effective therapy for patients with TETs. Indeed, clinical trials evaluating the use of these agents in the treatment of thymic carcinoma are already underway and have demonstrated promising preliminary clinical activity. [20] We further evaluated the tumor microenvironment for other immune co-stimulatory and co-inhibitory markers that may be targeted therapeutically, the first such evaluation in this disease. Overall, there was variability in the expression of these molecules consistent with the notion that the immune response to cancer is heterogenous. [21] However, some trends were evident. Although not associated with high PD-L1 expression, moderate to high TIM-3 expression was found in most TETs. TIM-3 has been implicated as an effective therapeutic target, which may be synergistic with anti-PD-1/PD-L1 blockade and therefore the absolute rate of high expression may be of interest for purposes of future drug development. [22] Moderate to  high expression of other potentially therapeutic targets in this series were also seen for CTLA-4 (52%), GITR (52%), ICOS (48%), CD137 (54%). High expression of CD137, in particular, was frequently co-expressed with high CD8+ TILs, potentially making this relevant to explore in TETs as well. CD137 expression has recently been used to identify neoantigen-specific T cells [23] and promising data of a CD137 agonist in combination with PD-1 blockade therapy was recently presented in patients with melanoma (response rate 50% [23/46] [26]. Direct comparisons between our study and these others are difficult due to differences in PD-L1 antibodies and criteria used to define PD-L1 positivity. In particular, it is noted that none of these studies based PD-L1 positivity on M-score, as was used in our study and is typical for current clinical trials. Furthermore, our report is distinct in using an antibody and scoring (including membranous expression only) that has been demonstrated to be consistent with assays with known predictive of therapeutic benefit with anti-PD-(L)1 therapy in clinical use and therefore has the greatest biologic and therapeutic applicability. A recent prospective report from described the consistency of the E1L3N PD-L1 assay with 22C3 and 28-8, which are now FDA approved and have important predictive validity [27] as opposed to other antibodies (e.g. Abcam EPR1161) which have no known cutpoint or predictive appreciation with response to anti-PD(L1) therapy and therefore may have limited clinical implications.
Thus far, there has been significant discrepancy regarding the prognostic impact of PD-L1 on survival [24][25][26]28,29]. Padda et al concluded that high PD-L1 expression associated with poorer overall survival [24] while Katsuya et al. described no survival difference associated with PD-L1 expression. [25] Meanwhile, our series suggests that high PD-L1 expression on the membrane of tumor cells is associated with improved overall survival. In contrast to the previous studies which examined TETs more broadly, we note that our series was largely comprised of patients with more aggressive histologies and advanced stage disease. Yokoyama and colleagues similarly described improved overall survival in high PD-L1 expression, [26] however this series was limited to only thymic carcinoma histology. Our series, which included patients with both thymic carcinoma and advanced stage thymoma with aggressive histologies, provides clarification on this issue. While its focus is limited to only patients with more aggressive histologies of thymic epithelial tumors (WHO Type B2, B3, and Thymic carcinoma), understanding the immunophenotype of these more aggressive tumors optimizes conclusions related to potential systemic therapies and best reflective of patients who are most likely to be treated with immunotherapies in clinical trials. These results are most reflective of the immunobiology of TETs in the setting in which immunotherapy would be clinically explored and therefore particularly valuable.
Our series is certainly limited by its size, such that larger series will be needed to more comprehensively evaluate the immunophenotypic landscape of TETs and more robustly determine associations with outcomes in a more comprehensive multivariate analysis. However, as the major intent of this analysis was to examine therapeutic targets that should be prioritized for clinical development, a modest size cohort can still provide useful insight. To this end, we hope that use of a PD-L1 assay with reliable association with clinically validated assays may help to assure its biological relevance in this disease and encourage the development of immunotherapies for patients with thymic cancers.
The development of immune checkpoint inhibitors in the treatment of TETs represents an important next step in improving outcomes in this disease. Building on studies such at this, we are hopeful that describing the presence and interconnections between immunotherapeutic targets and TILs may help to facilitate and expedite clinical trials to explore immunotherapies in TETs.
Supporting information S1 Table.