Higher cytolytic score correlates with an immunosuppressive tumor microenvironment and reduced survival in glioblastoma

Abstract

Cytolytic score (CYT), calculated from mRNA expression levels of granzyme and perforin, positively correlates with CD8+ T cell infiltration/activity in a variety of cancers. Unlike other cancers, higher CYT has been associated with worse prognosis in glioblastoma (GBM). To address this discrepancy, we sought to investigate the relationship between CYT and immune checkpoint gene score (ICGscore), as well as their correlation with patient survival and tumor immune cell infiltration. Clinical and RNA-sequencing data for patients with newly diagnosed GBM were obtained from The Cancer Genome Atlas. Maximally-selected rank statistics was used to dichotomize subgroups. CIBERSORT was used to estimate abudence of immune cell-types. Spearman correlation was used to characterize the relationship between CYT and ICGscore. Kaplan-Meier curves were generated for survival analysis. Overall, 28/151 patients had high CYT. High CYT was associated with a mesenchymal subtype (p < 0.001) and worse survival (7.45 vs. 12.2 months, p < 0.001). There were no differences in patient demographics, IDH/MGMT mutation status, or treatment. On subgroup analysis, patients with high CYT/ICGscore had significantly increased CD8+ infiltration (p < 0.001), as expected, and worse survival (HR 0.445, p < 0.01). Furthermore, CYT strongly correlated with ICGscore (R_S = 0.675, p < 0.001). The high CYT/ICGscore subgroup was associated with greater infiltration of M2 macrophages (p = 0.011) and neutrophils (p = 0.055). Our study highlights a multidimensional immunosuppressive GBM microenvironment in patients with higher CYT and potentially identifies patients with high CYT/ICGscore as a subgroup that may particularly benefit from multi-faceted immunotherapies, given their already elevated tumor CD8+ T cell levels.

Figure 1

Survival analysis of GBM patients with high vs. low CYT. High CYT was determined using maximally selected rank statistics. Patients with high CYT had reduced survival when compared to those with low CYT (7.45 vs. 12.2 months, p < 0.001).

Figure 2

Survival analysis of CYT/ICG subgroups. Kaplan–Meir curve of CYT/ICG subgroups across all ICGs including (A) CTLA4, (B) IDO, (C) PD1, (D) TIM3, (E) PDL1, (F) LAG3 assessing their impact on patient survival. (G) Kaplan–Meir curve of the composite ICGscore subgroups is also displayed (HR 0.445, p < 0.01).

Figure 3

Spearman’s rank order correlation of ICG and CYT. Correlation between CYT and ICGs including CTLA4, IDO, LAG3, PD1, PDL1, and TIM3. The correlation between CYT and composite ICGscore is also highlighted (R_S = 0.675, p < 0.001).

Figure 4

Immune cell infiltration between CYT/ICGscore subgroups. Immune cell populations differ between HighCYT/HighICGscore and LowCYT/LowICGscore subgroups. CD8+ T cells have increased infiltration in patients with HighCYT/HighICGscore (p < 0.001). M2 Macrophages are increased in patients with HighCYT/HighICGscore (p = 0.0107). Resting natural killer cells are lower in patients with HighCYT/HighICGscore (p = 0.0289). Neutrophils are increased in patients with HighCYT/HighICGscore, approaching a significant difference (p = 0.055).

Figure 5

Validation study using CGGA GBM cohort. (A) Kaplan–Meier curve comparing high CYT patients with low CYT patients (19.1 vs. 25.7 months, p = 0.05). (B) Kaplan–Meier curve of CYT/composite ICGscore subgroups (HR = 0.673, p = 0.066). (C) Spearman’s rank order correlation between CYT and composite ICGscore (R_S = 0.8291, p < 0.001). (D) Comparison of immune cell infiltration between CYT/ICGscore subgroups. CD8 T-Cell (p < 0.001) and M2 Macrophage (p = 0.017) populations were both increased in patients with HighCYT/HighICGscore.