Predictive role of tumor-infiltrating lymphocytes and immune phenotype for pembrolizumab in relapsed or refractory thymic carcinoma

Introduction

Thymic carcinoma is a rare cancer that exhibits a poor prognosis owing to its limited response to chemotherapy and a higher tendency to metastasize to distant sites, such as the liver (1). Surgical resection offers a potential cure for residual thymic carcinoma (2). Nonetheless, recurrence or metastasis occurs in a significant number of patients and platinum-based chemotherapy is considered the standard treatment (3). However, a standard treatment following the failure of platinum-based chemotherapy has not yet been established.

Several studies have been conducted on treatment options after the failure of platinum-based chemotherapy. The binding of programmed cell death protein 1 (PD-1), which is mainly present on activated T cells, to its ligand programmed death-ligand 1 (PD-L1) on cancer cells results in the suppression of cytotoxic T-cell activity (4). Since the thymus is associated with T-cell development, the high expression of PD-L1 in thymic carcinoma has raised expectations for anti-PD-1/PD-L1 agents (5). Two phase 2 trials evaluated pembrolizumab, an anti-PD-1 antibody, demonstrating a promising objective response rate of approximately 20% (6,7). However, thymic carcinoma has lost the function of inducing T-cell development, and caution is warranted due to the higher risk of immune-related adverse events (irAEs) associated with the use of pembrolizumab in this setting (8).

Considering the concerns regarding irAEs and the high cost of pembrolizumab treatment, predictive biomarkers are required to achieve optimal treatment for these patients. Tumor-infiltrating lymphocytes (TIL) evaluated using Lunit SCOPE IO, an artificial intelligence (AI)-powered analyzer of TIL, have shown prognostic value in predicting outcomes with immune checkpoint inhibitor (ICI) in various cancers (9-12). Additionally, we explored the potential predictive value of AI-powered TIL analysis for locally advanced thymic epithelial tumors (TET) (13). Based on these considerations, we conducted this analysis to assess the predictive value of AI-powered TIL analysis for pembrolizumab in platinum-failed, relapsed or refractory (R/R) thymic carcinomas. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-526/rc).

Methods

Participants and Study design

This was a single-center, retrospective cohort study. Patients with histologically confirmed, R/R thymic carcinoma who received pembrolizumab at Seoul National University Hospital between January 2016 and December 2021 were included in the analysis. Patients who experienced at least one failure of platinum-based chemotherapy before receiving pembrolizumab were eligible. In addition, for TIL analysis using Lunit SCOPE IO, only patients with hematoxylin and eosin (H&E)-stained archival tumor tissues were eligible. All histopathological slides were reviewed by two independent pathologists (J.K. and K.C.J.) to confirm the diagnosis. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Seoul National University Hospital (No. H-2201-005-1285), and individual consent for this retrospective analysis was waived.

Initially, 12 patients were identified; however, one patient without archival tumor tissue was excluded, leaving 11 patients for TIL analysis using Lunit SCOPE IO. Among them, one patient did not pass the Lunit SCOPE IO quality control due to an insufficient grid count for TIL density calculation, resulting in 10 patients included in the final analysis (Figure 1). Electronic medical records were reviewed for baseline characteristics, response to pembrolizumab, and survival outcomes. Tumor response was assessed using RECIST version 1.1 (14). We analyzed the association between TIL density or immune phenotype (IP) of tumor specimens and tumor response or survival outcomes. For heavily treated patients who might have had multiple tissue samples from previous assessments, we selected the most recent sample prior to the administration of pembrolizumab for TIL analysis. If the specimen consisted of more than one slide, including the surgical specimen, the representative slide with the largest tumor area was selected for analysis.

Figure 1 Details of cohort assembly strategy. IP, immune phenotype; TIL, tumor-infiltrating lymphocyte; R/R, relapsed or refractory.

Procedures

The Lunit SCOPE IO (Lunit Inc., Seoul, Republic of Korea) is an AI-powered spatial TIL analyzer that detects and measures TILs within the cancer epithelium (intratumoral TILs; iTIL) and stroma (stromal TIL; sTILs) from H&E-stained whole-slide images (WSIs). It comprises two convolutional neural networks: one segments the cancer area (CA) and cancer-related stroma (CS), while the other identifies TILs. The Lunit SCOPE IO was originally trained and optimized using 2.8×10⁹ µm² of H&E-stained tissue regions, containing 6.0×10⁵ TILs, extracted from 3,166 WSI assorted from 25 different tumor types, including thymic carcinoma, and annotated by board-certified pathologists (9). The model used in this study was updated via further training using 1.4×10¹⁰ µm² of CA and CS, including 6.23×10⁵ TILs, extracted from 18,679 H&E-stained WSI of 17 different solid tumor types including thymic carcinoma.

The detailed process of TIL analysis using Lunit SCOPE IO has been described in our previous study (13). Specifically, the model segmented the WSIs into CA and CS, and identified and quantified the TILs in each area. The model estimated the density of iTILs or sTILs per 1 mm² of the corresponding tissue area in each case. Moreover, the model used the densities of iTILs and sTILs in 0.25 mm²-sized grids to derive IP of each grid: inflamed-grids having iTIL density of ≥130/mm²; immune-excluded-grids having sTIL density of ≥260/mm² and iTIL density of <130/mm²; and immune-desert-grids having iTIL and sTIL densities of <130/mm² and <260/mm², respectively. The inflamed score (IS), immune-excluded score (IES), and immune-desert score (IDS) of the WSIs were defined as the number of grids annotated to a certain IP divided by the total number of grids analyzed in the WSI. Finally, for each WSI, the representative IP was classified as inflamed if the IS was ≥20.0%; as immune-excluded if the IES was ≥33.3% and IS was <20.0%; and as immune-desert in all other cases. The thresholds for TIL and IS used to define IP classifications were pre-established based on their optimal ability to predict high interferon-γ-responsive gene signature levels in a cohort of The Cancer Genome Atlas (TCGA) pan-carcinoma tumor samples (n=7,454) (15,16).

Statistical analysis

Descriptive statistics presented as numbers with percentiles or medians with ranges were used to summarize the data. The Clopper-Pearson exact 95% confidence interval (CI) for the proportion of patients who responded was calculated using an exact binomial calculation. The Kaplan-Meier method was used to estimate progression-free survival (PFS) and overall survival (OS). Patients who were alive without progression were censored at the time of last follow-up. The cutoff for discriminating between low and high iTIL and sTIL levels was defined as the point with the lowest P value for OS by the log-rank test for all possible levels for each biomarker. The log-rank test was used to assess differences between the groups in terms of PFS and OS. Factors associated with PFS were evaluated using the Cox proportional hazards model, and hazard ratios (HRs) with 95% CIs were calculated. We performed a backward-selection multivariable Cox proportional hazards model adjusted for covariates (P<0.2) in univariable analyses. The statistical software ‘R’ version 4.3.1 (www.r-project.org) was used for all statistical analyses. P<0.05 was considered statistically significant.

Results

Patient and disease characteristics

The baseline characteristics of the study subjects are presented in Table 1. The median age was 57 years (range, 40–76 years), and seven patients (70%) were men. Overall, 7 patients (70%) had squamous cell carcinoma. Five patients (50%) had extrathoracic metastases. Six patients (60%) had undergone prior surgical resection for curative intent, and three patients (30%) had received prior radiation therapy for the primary mass. The median value of the previous lines of systemic treatment was three, indicating heavily treated patients.

Efficacy of pembrolizumab

The best response to pembrolizumab was complete response (CR) in one patient (10%), partial response in one patient (10%), stable disease in three patients (30%), and progressive disease in five patients (50%) (Table S1). The overall response rate was 20.0% (95% CI: 2.5–55.6%) and the disease control rate was 50.0% (95% CI: 18.7–81.3%). With a median follow-up of 34.0 months (95% CI: 12.0–not determined), the median PFS for the entire cohort was 5.0 months (95% CI: 1.0–not determined) (Figure S1A). The 6-month PFS rate was 40.0% (95% CI: 18.7–85.5%). The 12-month OS rate was 65.6% (95% CI: 40.2–100%), with a median OS of 15.0 months (95% CI: 9.0–not determined) (Figure S1B).

Specimen analysis using Lunit SCOPE IO

The results of the analysis using Lunit SCOPE IO are listed in Table 2. Among the 10 patients who underwent analysis, six samples (60%) were obtained via needle biopsy, with two originating from the primary mediastinal mass and the remaining four from metastasis. Of the four samples (40%) obtained through surgery, three originated from primary mediastinal masses. The median iTIL and sTIL densities in the 10 patients were 32.85/mm² (range, 0.76–1,125.16/mm²) and 226.12/mm² (range, 20.59–1,319/mm²), respectively. The IP of each 1 mm² grid was determined for spatial analysis of TIL distribution. The median IS, IES, and IDS values for all patients were 4.29%, 18.77%, and 73.33%, respectively. Finally, two patients (20%) had an inflamed IP, one (10%) had immune-excluded IP, and the remaining seven (70%) had immune-desert IP. Notably, the patient with the highest iTIL density and inflamed IP showed a CR to pembrolizumab, which was sustained until the last follow-up (CR was maintained for 41 months). Figure 2 illustrates the Lunit SCOPE IO-inferred images of the patient’s specimen as representative images for analysis using Lunit SCOPE IO.

Figure 2 Representative image of Lunit SCOPE IO. Representative image of H&E original image (A) and Lunit SCOPE IO-inferred segmentation of cancer epithelium (orange), cancer stroma (grass green), and TIL (cyan blue) (B). H&E, hematoxylin and eosinTIL, tumor-infiltrating lymphocyte.

TIL density, IP and survival outcomes

Using the log-rank test, the optimal cut-off thresholds for high and low iTIL and sTIL densities were determined to be 27.23/mm² and 252.54/mm², respectively. Using these thresholds, six (60%) and five (50%) patients were categorized into the high iTIL and sTIL groups, respectively. Patients with higher iTIL density (>27.23/mm²) exhibited longer OS (Figure 3A, median OS not determined vs. 4.0 months, P=0.03) and PFS (Figure 3B, median PFS 9.5 vs. 1.5 months, P=0.03). In addition, patients with a higher sTIL density (>252.54/mm²) exhibited longer OS (Figure 3C, median OS not determined vs. 9.0 months, P=0.01) and PFS (Figure 3D, median PFS 10.0 vs. 1.0 months, P=0.006).

Figure 3 Survival outcomes based on TIL density. (A) Kaplan-Meier curves of OS based on intratumoral TIL density. (B) Kaplan-Meier curves of PFS based on intratumoral TIL density. (C) Kaplan-Meier curves of OS based on stromal TIL density. (D) Kaplan-Meier curves of PFS based on stromal TIL density. OS, overall survival; PFS, progression-free survival; TIL, tumor-infiltrating lymphocyte.

Table 3 presents the factors associated with PFS determined using the Cox proportional hazards model. Histology, prior radiotherapy, iTIL, and sTIL density were associated with PFS (P value less than 0.2 in the univariable analysis. Using these four factors, backward-selection multivariate analysis resulted in a model consisting of radiotherapy and sTIL density. Unlike radiotherapy, which did not show statistically significant differences, Low sTIL density was associated with shorter PFS (HR: 33.89; 95% CI: 1.94–591.54; P=0.01).

To clarify whether IP is also a predictive biomarker of pembrolizumab treatment, such as TIL density, we analyzed survival outcomes according to IP. Patients with inflamed IP exhibited significantly longer PFS than those with non-inflamed IP (Figure 4A; median PFS, 10.0 vs. 3.0 months; P=0.046). In contrast, no significant difference was observed between patients with and without inflamed IP in terms of OS (Figure 4B, median OS 15.0 months vs. not determined, P=0.75).

Figure 4 Survival outcomes based on IP. (A) Kaplan-Meier curves of progression-free survival based on IP. (B) Kaplan-Meier curves of overall survival based on IP. IP, immune phenotype.

Discussion

In this study, we investigated several factors, including density and spatial distribution of TILs, and their relationship with survival outcomes in patients with heavily treated thymic carcinoma who were treated with pembrolizumab. First, higher iTIL and sTIL densities were associated with a longer OS. Although the cutoff value requires further validation owing to the small number of patients in this study, the higher iTIL and sTIL groups with cutoff values determined by OS analysis also had longer PFS. In addition, univariate and multivariate Cox analysis confirmed that a higher sTIL density was associated with longer PFS. Second, IP and the spatial distribution of TILs was also correlated with PFS. Specifically, an inflamed IP was associated with longer PFS, indicating a potential link between an inflamed immune microenvironment and improved disease progression outcomes.

In our study, pembrolizumab demonstrated an overall response rate of 20%, which is consistent with outcomes from previous phase 2 trials (6,7). Furthermore, both PFS and OS closely aligned with the findings of a previous study (6). These two exploratory studies provide insights into the predictive value of PD-L1 expression for the efficacy of PD-1 blockade. Since high expression of PD-L1 has been reported in a significant portion of thymic carcinomas (17), extensive research has been conducted to assess its role as a biomarker. However, the value of PD-L1 as a prognostic biomarker in thymic carcinoma is inconsistent (5,17,18). Katsuya et al. found that PD-L1 expression was not associated with ICI efficacy (19). The difficulty in determining various testing methods and cutoff values for PD-L1 may have contributed to these inconsistent results. Moreover, there is no consensus on the most appropriate assay among the various PD-L1 tests for thymic carcinoma, and different assays were used in each study (20). In addition, the intratumoral heterogeneity of PD-L1 expression has been documented in various cancer types, including thymic carcinomas (20,21). Sato et al. investigated PD-L1 expression in whole section slides to overcome this heterogeneity, but reported that due to lack of reproducibility, the significance of PD-L1 as a biomarker in thymic carcinoma may be ambiguous (22). Therefore, there is an unmet need for predictive biomarkers of the efficacy of ICI in thymic carcinoma.

In addition to PD-L1 expression, several efforts have been made to predict the efficacy of ICI using other factors such as tumor mutation burden (TMB) and TILs in various cancer types (23). TMB has been considered a factor that can predict the ICI response, as a higher TMB can promote immune cell infiltration through neoantigen generation (24). Recent studies have consistently shown an association between high TMB and favorable response to ICI (25,26). TET is one of the cancers with the lowest TMB and shows a favorable response to ICI, raising questions regarding the predictive value of TMB for ICI in this cancer type (6,7,27). However, considering the impact of the tumor microenvironment, composed of TILs, cancer cells, and cancer stroma, on cancer progression, along with the unique features of the thymus related to T cell development, it is believed that the favorable effects of ICI in thymic carcinoma may be associated with the tumor microenvironment (28). The present study reports the importance of TIL infiltration in the use of ICI for thymic carcinoma, particularly in the cancer stroma. Considering the spatial distribution of TILs, we partitioned the tissue into CA and CS, evaluated the TIL density in each compartment, and verified that sTILs were an independent predictor of PFS and OS. These results are consistent with the previously reported findings. Shim et al. discovered a significant correlation between high sTIL levels and improved survival in thymic carcinoma (29). Bocchialini et al. found that iTIL decreased as the disease progressed and identified sTIL as a pivotal outcome predictor in their study on thymic carcinoma (30). In another study reporting the significance of sTIL in thymic carcinoma, high reproducibility of TILs was observed, and they were found to have lower intratumoral heterogeneity than PD-L1 expression (22). In our previous study, we performed a TIL analysis on multiple slides of surgical specimens from thymic carcinoma using Lunit SCOPE IO and reported an association between TIL density and survival outcomes, suggesting that TIL density assessed using Lunit SCOPE IO can partially overcome intratumoral heterogeneity (13). Furthermore, attempts have been made to categorize the spatial pattern of TIL distribution into three distinct IPs: inflamed, excluded, and desert, with the aim of using them as biomarkers for ICI treatment (9,31). We observed that the use of pembrolizumab resulted in a longer PFS in patients with inflamed IP than in those with non-inflamed IP. To the best of our knowledge, this study represents the first report of IP as a predictive biomarker for ICI treatment in thymic carcinoma, and further validation via larger-scale studies is warranted.

The present study demonstrated the utility of TIL density and IP analyzed using an AI-powered spatial TIL analyzer as predictive biomarkers that can aid in the selection of patients likely to respond to ICI treatment. In patients with platinum-failed, R/R thymic carcinoma, ICIs or antiangiogenic agents are treatment options, although a standard treatment has not yet been established. Attempts have been made to enhance treatment efficacy by combining ICIs with antiangiogenic agents, with favorable outcomes being reported (32,33). Angiogenesis-related factors induce immune suppression within the tumor microenvironment and antiangiogenic agents can activate TIL function through their immunomodulatory abilities (34). Furthermore, a previous study has demonstrated that TIL density can serve as a predictive biomarker for antiangiogenic agents (35). Considering these points, TIL density or IP could serve as valuable predictive biomarkers not only in ICI treatment but also in the use of antiangiogenic agents or their combination therapy in thymic carcinoma; therefore, further research in this area is required.

Nonetheless, this study has several limitations. First, the sample size was small. Although the predictive value of sTIL density was confirmed through multivariate Cox analyses, the small sample size limits its reliability. Additionally, the optimal cut-off value for TIL densities requires further validation. Therefore, larger studies are required to validate the relationship between TILs, IP, and ICI efficacy. However, since this study is exploratory in nature, investigating the predictive value of TILs and IP in the use of ICI, even a small-scale study, can still offer valuable findings. Second, we were unable to investigate PD-L1 expression. Therefore, analysis of TIL distribution and PD-L1 expression in the future might offer a broader understanding of the tumor microenvironment and the effects of ICI. Lastly, although thymic carcinoma samples were used in the initial development and updates of the model, its reliability was not specifically validated for thymic carcinoma.

Despite these limitations, our study is the first to explore a potential predictive biomarker of ICI treatment using an AI-powered model for patients with platinum-failed, R/R thymic carcinoma. Although there is no established method for quantifying TILs in thymic carcinoma, we previously confirmed that this AI model is a reliable TIL analyzer capable of effectively distinguishing TILs from cancer cells in TET. Furthermore, this model, which analyzes H&E-stained slide images without the need for additional staining, offers a straightforward and rapid process, making it advantageous for clinical applications.

Conclusions

In conclusion, to the best of our knowledge, this is the first study to explore predictive value of TIL and IP for ICI treatment in thymic carcinoma. TIL density and IP can be used as a predictive biomarker for ICI in patients with R/R thymic carcinoma. High iTIL (>27.23/mm²) or sTIL (>252.54/mm²), and inflamed IP may serve as indicators of longer PFS. Further research is warranted to validate the predictive roles of TIL density and IP in ICI treatment for R/R thymic carcinoma.

Acknowledgments

This study was funded through a generous donation from Ms. Soon-Hee Park. Despite losing her son to thymic carcinoma, she made a donation to support the progress of thymic carcinoma treatment. We are deeply grateful for her invaluable support.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-526/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-526/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-526/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-526/coif). Y.L. and S.S. are from Lunit Inc., Seoul, Republic of Korea. C.Y.O. holds a leadership position at Lunit and owns stocks in the company. D.W.K. receives research funding from Alpha Biopharma, Amgen, Astrazeneca/Medimmune, Boehringer-Ingelheim, Bridge BioTherapeutics, Chong Keun Dang, Daiichi-Sankyo, GSK, Hanmi, InnoN, IQVIA, Janssen, Merck, Merus, Mirati Therapeutics, MSD, Novartis, ONO Pharmaceutical, Pfizer, Roche/Genentech, Takeda, TP Therapeutics, Xcovery, Yuhan. K.J.N. is a cofounder and chief medical officer of Portrai, Inc. B.K. receives research funding from MSD, AstraZeneca, and Ono Pharmaceutical Co., Ltd., and has served as an advisor for Handok, NeoImmuneTec, Trialinformatics and ImmuneOncia outside of the current work. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Seoul National University Hospital (No. H-2201-005-1285), and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Ceative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Okuma Y, Hosomi Y, Takagi Y, et al. Clinical outcomes with chemotherapy for advanced thymic carcinoma. Lung Cancer 2013;80:75-80. [Crossref] [PubMed]
2. Zhao K, Liu Y, Jing M, et al. Long-term survival and prognosis after surgical treatment of patients with thymic carcinoma: a retrospective analysis. J Thorac Dis 2024;16:7582-91. [Crossref] [PubMed]
3. Girard N, Ruffini E, Marx A, et al. Thymic epithelial tumours: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2015;26:v40-55. [Crossref] [PubMed]
4. Ortega MA, Boaru DL, De Leon-Oliva D, et al. PD-1/PD-L1 axis: implications in immune regulation, cancer progression, and translational applications. J Mol Med (Berl) 2024;102:987-1000. [Crossref] [PubMed]
5. Katsuya Y, Fujita Y, Horinouchi H, et al. Immunohistochemical status of PD-L1 in thymoma and thymic carcinoma. Lung Cancer 2015;88:154-9. [Crossref] [PubMed]
6. Cho J, Kim HS, Ku BM, et al. Pembrolizumab for Patients With Refractory or Relapsed Thymic Epithelial Tumor: An Open-Label Phase II Trial. J Clin Oncol 2019;37:2162-70. [Crossref] [PubMed]
7. Giaccone G, Kim C, Thompson J, et al. Pembrolizumab in patients with thymic carcinoma: a single-arm, single-centre, phase 2 study. Lancet Oncol 2018;19:347-55. [Crossref] [PubMed]
8. Rajan A, Sivapiromrat AK, McAdams MJ. Immunotherapy for Thymomas and Thymic Carcinomas: Current Status and Future Directions. Cancers (Basel) 2024;16:1369. [Crossref] [PubMed]
9. Park S, Ock CY, Kim H, et al. Artificial Intelligence-Powered Spatial Analysis of Tumor-Infiltrating Lymphocytes as Complementary Biomarker for Immune Checkpoint Inhibition in Non-Small-Cell Lung Cancer. J Clin Oncol 2022;40:1916-28. [Crossref] [PubMed]
10. Jung HA, Park KU, Cho S, et al. A Phase II Study of Nivolumab plus Gemcitabine in Patients with Recurrent or Metastatic Nasopharyngeal Carcinoma (KCSG HN17-11). Clin Cancer Res 2022;28:4240-7. [Crossref] [PubMed]
11. Kim DH, Ahn JS, Kang M, et al. Comparison of Tumor Microenvironments between Primary Tumors and Lymph Node Metastases in Head and Neck Squamous Cell Carcinoma and Their Predictive Role in Immune Checkpoint Inhibitor Treatment. Cells 2024;13:1557. [Crossref] [PubMed]
12. Kim DH, Kang M, Park G, et al. Changes in the tumor microenvironment in recurrent head and neck squamous cell carcinoma and its implication on efficacy of immune checkpoint inhibitors. Discov Oncol 2024;15:686. [Crossref] [PubMed]
13. Kim DH, Lim Y, Kim S, et al. Artificial intelligence-powered spatial analysis of tumor-infiltrating lymphocytes as a biomarker in locally advanced unresectable thymic epithelial neoplasm: A single-center, retrospective, longitudinal cohort study. Thorac Cancer 2023;14:3001-11. [Crossref] [PubMed]
14. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228-47. [Crossref] [PubMed]
15. Ayers M, Lunceford J, Nebozhyn M, et al. IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest 2017;127:2930-40. [Crossref] [PubMed]
16. Chen DS, Mellman I. Elements of cancer immunity and the cancer-immune set point. Nature 2017;541:321-30. [Crossref] [PubMed]
17. Padda SK, Riess JW, Schwartz EJ, et al. Diffuse high intensity PD-L1 staining in thymic epithelial tumors. J Thorac Oncol 2015;10:500-8. [Crossref] [PubMed]
18. Yokoyama S, Miyoshi H, Nakashima K, et al. Prognostic Value of Programmed Death Ligand 1 and Programmed Death 1 Expression in Thymic Carcinoma. Clin Cancer Res 2016;22:4727-34. [Crossref] [PubMed]
19. Katsuya Y, Kitano S, Yamashita M, et al. Comprehensive biomarker analysis from phase II study of nivolumab in patients with thymic carcinoma. Front Oncol 2022;12:966527. [Crossref] [PubMed]
20. Higuchi R, Goto T, Hirotsu Y, et al. PD-L1 Expression and Tumor-Infiltrating Lymphocytes in Thymic Epithelial Neoplasms. J Clin Med 2019;8:1833. [Crossref] [PubMed]
21. Rasmussen JH, Lelkaitis G, Håkansson K, et al. Intratumor heterogeneity of PD-L1 expression in head and neck squamous cell carcinoma. Br J Cancer 2019;120:1003-6. [Crossref] [PubMed]
22. Sato J, Kitano S, Motoi N, et al. CD20+ tumor-infiltrating immune cells and CD204+ M2 macrophages are associated with prognosis in thymic carcinoma. Cancer Sci 2020;111:1921-32. [Crossref] [PubMed]
23. Yarchoan M, Albacker LA, Hopkins AC, et al. PD-L1 expression and tumor mutational burden are independent biomarkers in most cancers. JCI Insight 2019;4:e126908. [Crossref] [PubMed]
24. Chen DS, Mellman I. Oncology meets immunology: the cancer-immunity cycle. Immunity 2013;39:1-10. [Crossref] [PubMed]
25. Chalmers ZR, Connelly CF, Fabrizio D, et al. Analysis of 100,000 human cancer genomes reveals the landscape of tumor mutational burden. Genome Med 2017;9:34. [Crossref] [PubMed]
26. Klempner SJ, Fabrizio D, Bane S, et al. Tumor Mutational Burden as a Predictive Biomarker for Response to Immune Checkpoint Inhibitors: A Review of Current Evidence. Oncologist 2020;25:e147-59. [Crossref] [PubMed]
27. Hou X, Lin S, Liu Y, et al. Analysis of the tumor microenvironment and mutation burden identifies prognostic features in thymic epithelial tumors. Am J Cancer Res 2022;12:2387-96.
28. Hinshaw DC, Shevde LA. The Tumor Microenvironment Innately Modulates Cancer Progression. Cancer Res 2019;79:4557-66. [Crossref] [PubMed]
29. Shim HS, Byun CS, Bae MK, et al. Prognostic effect of stromal lymphocyte infiltration in thrymic carcinoma. Lung Cancer 2011;74:338-43. [Crossref] [PubMed]
30. Bocchialini G, Schiefer AI, Müllauer L, et al. Tumour immune microenvironment in resected thymic carcinomas as a predictor of clinical outcome. Br J Cancer 2022;127:1162-71. [Crossref] [PubMed]
31. Shen J, Choi YL, Lee T, et al. The inflamed immune phenotype (IIP): A clinically actionable artificial intelligence (AI)-based biomarker predictive of immune checkpoint inhibitor (ICI) outcomes across >16 primary tumor types. J Clin Oncol 2022;40:2621.
32. Conforti F, Zucali PA, Pala L, et al. Avelumab plus axitinib in unresectable or metastatic type B3 thymomas and thymic carcinomas (CAVEATT): a single-arm, multicentre, phase 2 trial. Lancet Oncol 2022;23:1287-96. [Crossref] [PubMed]
33. Remon J, Girard N, Novello S, et al. PECATI: A Multicentric, Open-Label, Single-Arm Phase II Study to Evaluate the Efficacy and Safety of Pembrolizumab and Lenvatinib in Pretreated B3-Thymoma and Thymic Carcinoma Patients. Clin Lung Cancer 2022;23:e243-6. [Crossref] [PubMed]
34. Fukumura D, Kloepper J, Amoozgar Z, et al. Enhancing cancer immunotherapy using antiangiogenics: opportunities and challenges. Nat Rev Clin Oncol 2018;15:325-40. [Crossref] [PubMed]
35. Kim DH, Lim Y, Ock CY, et al. Artificial intelligence-powered spatial analysis of tumor-infiltrating lymphocytes as a predictive biomarker for axitinib in adenoid cystic carcinoma. Head Neck 2023;45:3086-95. [Crossref] [PubMed]