Fig. 2: Tumor effector T cell changes predict outcomes to immunotherapy re-challenge after radiotherapy and TPE.

A PBMCs were isolated from each patient and subpopulations were measured by a broad flow cytometry panel and change in percent of each subpopulation is reported for each patient on a log scale. B Cox proportional hazards regression was performed for changes in each subpopulation. Hazard ratios with 95% confidence intervals are shown; box size indicates proportion of patients with increasing immune cell subpopulation. C Overall Survival (OS) is predicted by change in GZMB⁺CX3CR1⁺CD11a^high CD8⁺ (T_TR) T cells from baseline to the second cycle of ICI re-challenge (log rank p = 0.002; HR 0.15 [95% CI 0.04-0.6], p = 0.008). D Overall Survival (OS) is predicted by change in CD25⁺FOXP3⁺ CD4⁺ (T_reg) T cells from baseline to the second cycle of ICI re-challenge (log rank p = 0.0009; HR 9.72 [95% CI 2-47.3], p = 0.005). E Overall Survival (OS) by change in NKG7⁺CD11a^high CD8⁺ T cells from baseline to the second cycle of ICI re-challenge is shown (log rank p = 0.05; HR 0.32 [95% CI 0.1-1.07], p = 0.064). F Overall Survival (OS) is predicted by change in Bim⁺ T cells from baseline to the second cycle of ICI re-challenge (log rank p = 0.006; HR 5.5 [95% CI 1.4-21], p = 0.014). CI confidence interval, GZMB Granzyme B, HR Cox proportional hazard ratio, PBMCs peripheral blood mononuclear cells. All statistical tests were two-sided. See also Supplementary Tables 2–5, Supplementary Figs. 6, 7.