Summary and future perspectives 239 7 a single oncogenic alteration or other autonomous features of cancer cells for which tyrosine kinase inhibitor therapy or chemotherapy is available. Therefore, we hypothesized that a predictive model should contain more than one biomarker to capture the complex interplay of different components within the TME. Previous studies have already shown the additive value of combining PD-L1 with tumor mutational burden (TMB)24,25 and PD-L1 with CD8 TILs26,27. As demonstrated in chapter 2, combining PD-1T TILs with either 50% or 1% PD-L1 TPS did not improve predictive accuracy. A more comprehensive analysis was performed in chapter 3, where we evaluated individual and pairs of biomarkers as continuous variables. The tested biomarkers included various T cell markers (CD8 TILs, PD-1T TILs, CD3 TILs), a B-cell marker (CD20), TLS, PD-L1 TPS and the tumor inflammation signature (TIS). The TIS is an mRNA signature that has demonstrated predictive potential in different cancer types treated with PD-1 blockade monotherapy28–30. Our findings in chapter 3 indicated that composite biomarkers did not provide improved predictive performance compared to the use of PD-1T TILs or TIS alone. In contrast to PD-1T TILs, a distinct tumor-reactive TIL subset7, we assessed general TIL density as this biomarker has been correlated to response to PD-1 blockade monotherapy in different cancer types26,31–34. In addition, the specific localization of CD8 TILs within the tumor and the peritumoral compartment (i.e. stroma) has been proposed as a biomarker for response to PD-blockade monotherapy31,35. In our study, combining CD8 or CD3 density with intratumoral localization of CD8 showed limited discriminatory ability in the validation cohort. Notably, PD-1T TILs showed the highest predictive performance among all the biomarkers, particularly in identifying patients with no long-term benefit. These findings align with other studies that have observed that not all TILs are in a state to recognize and eliminate tumor cells7,36,37. Chapter 3 also demonstrates that combining PD-1T TILs with any of the other biomarkers did not improve predictive capacity. This observation can be attributed to the interdependence between biomarkers, as they all reflect aspects of the antitumor immune response. Consequently, we propose combinations of PD-1T TILs with alternative biomarkers that capture distinct facets of the anti-tumor immune response. For example, TMB can serve as an indicator of immunogenic neoantigens resulting from somatic mutations38. Additionally, mutations in STK11/LKB1 and KEAP1 are associated with a TME characterized by T-cell exclusion, low or absent PDL1 expression, downregulation of MHC class II in STK11 mutated tumors, and downregulation of type I interferon and other cytokines in KEAP1 mutated tumors. These subgroups have been linked with primary resistance to PD-1 blockade monotherapy
RkJQdWJsaXNoZXIy MTk4NDMw