Supplementary Material for: Characterization of pleural mesothelioma by hierarchical clustering analyses using immune cells within tumor microenvironment.

Introduction: Over the past decade, classifications using immune cell infiltration have been applied to many types of tumors; however, mesotheliomas have been less frequently evaluated. Methods: In this study, 60 well-characterized pleural mesotheliomas (PMs) were evaluated immunohistochemically for the characteristics of immune cells within tumor microenvironment (TME) using 10 immunohistochemical markers CD3, CD4, CD8, CD56, CD68, CD163, FOXP3, CD27, PD-1 and TIM-3. For further characterization of PMs, hierarchical clustering analyses using these 10 markers were performed. Results: Among the immune cell markers, CD3 (P < 0.0001), CD4 (P = 0.0016), CD8 (P = 0.00094), CD163+ (P = 0.042) and FOXP3+ (P = 0.025) were significantly associated with unfavorable clinical outcome. Immune checkpoint receptor expressions on tumor-infiltrating lymphocytes such as PD-1 (P = 0.050), CD27 (P = 0.014) and TIM-3 (P = 0.0098) were also associated with unfavorable survival. Hierarchical clustering analyses identified three groups showing specific characteristics and significant associations with patient survival (P = 0.011): the highest number of immune cells (ICHigh); the lowest number of immune cells, especially CD8+ and CD163+ cells (ICLow); and intermediate number of immune cells (ICInt). ICHigh tumors showed significantly higher expression of PD-L1 (P = 0.00038). Cox proportional hazard model identified ICHigh [hazard ratio (HR) = 2.90] and ICInt (HR = 2.97) as potential risk factors compared with ICLow. Tumor CD47 (HR = 2.36), tumor CD70 (HR = 3.04) and tumor PD-L1 (HR = 3.21) expressions were also identified as potential risk factors for PM patients. Conclusion: Our findings indicate immune checkpoint and/or immune cell-targeting therapies against CD70-CD27 and/or CD47-SIRPA axes may be applied for PM patients in combination with PD-L1-PD-1 targeting therapies in accordance with their tumor immune microenvironment characteristics.