Efficacy of ICI in various TNBC syngeneic mice models
Four syngeneic mice models were produced by xenografting each mouse TNBC cell line, such as 4T1 (n = 20), E0771 (n = 16), JC (n = 20) and EMT6 cells (n = 20) (Table 1). In mice subcutaneously inoculated with indicated TNBC cells, the PD-1 inhibitor was injected six times, bleeding was performed three times, and tumors were obtained after euthanizing each mouse for profiling cytokines and immune cells as planned (Fig. 1a). Differential response to the PD-1 inhibitor was observed in four syngeneic mice models (Fig. 1b, Table 1). The JC model showed the highest tumor response rate (40%, 4/10) of syngeneic models: 4T1 (36%, 4/11), EMT6 (36%, 4/11), or E0771 model (23%, 3/13).
Table 1 Summary of outcome of efficacy of PD-1 inhibitor in syngeneic mice
Abbreviation: PD-1i: PD-1 inhibitor, vol: volume, SD: Standard deviation
Early change of tumor size predicted final efficacy of PD-1 inhibitor treatment.
To predict the PD-1 inhibitor efficacy in TNBC syngeneic mouse model, tumor sizes were compared at early time point (7 days post PD-1 inhibitor treatment) and the final point (at the time of mice sacrifice; at 15 days for 4T1; at 29 days for E0771 and JC; at 24 days for EMT6 xenograft mice model).
In the control group ‘relative tumor size’ (calculated by dividing tumor size at day N by the tumor size at day 0 in each group) increased more rapidly compared to the responder group as shown in the Fig. 2a. We also analyzed the ‘relative tumor size difference between responder and control group’ at 7 days post PD-1 inhibitor treatment and the final day (calculated by subtracting relative tumor size at day N in control from the relative tumor size at day N in responder). In all syngeneic models, the relative tumor size differences at 7 days post PD-1 inhibitor treatment were negative numbers, suggesting that tumor response already began to appear at early time point in responders, and the relative tumor size differences became bigger at the final point (Fig. 2b). Finally, to compare relative tumor size differences among 4 syngeneic models at 7 days post PD-1 inhibitor treatment and the final day, we transformed 4 separate bar graphs from each syngeneic model in Fig. 2b into one curved-line graph (Fig. 2c). E0771 model showed that the greatest relative tumor size difference among 4 syngeneic models at 7 days post PD-1 inhibitor treatment and also the final day, whereas 4T1 model demonstrated that the smallest relative tumor size difference among 4 syngeneic models at 7 days post PD-1 inhibitor treatment and also the final day. JC and EMT6 models showed moderate relative tumor size difference both at 7 days post PD-1 inhibitor treatment and the final day. Accordingly, relative tumor size difference at early point (7 days post PD-1 inhibitor injection) predicted the final difference of relative tumor size at sacrifice, which suggests change of tumor size at 7 days post PD-1 inhibitor treatment can be an early biomarker of final efficacy of PD-1 inhibitor in our preclinical model.
Early biomarkers using peripheral blood T cells for the PD-1 inhibitor efficacy
It may be very advantageous that we can identify predictive biomarkers for ICI efficacy at an earlier time. Therefore, we further investigated peripheral blood mononuclear cell markers at 7 days post-PD-1 inhibitor treatment to correlate the final efficacies of the PD-1 inhibitor in each syngeneic mouse model. Collected blood samples were stained with T cell markers, such as CD8+ and CD4+ T cells and proliferation marker Ki67 and analyzed by flow cytometry. Fig. 3a shows the gating strategy for flow cytometry analysis.
At 7 days post-PD-1 inhibitor treatment, earlier changes in the proportion of CD4+ and CD8+ T cells in the peripheral blood are shown in Fig. 3b. Representatively in the JC model, CD8+ T cells were 3.2-fold higher in responders compared to PBS treatment control (ANOVA, p =0.077 among the three groups; LSD, p <0.029), and CD4+ T cells were also 2.2 fold higher in responders compared to PBS treatment control (ANOVA, p =0.066 among the three groups; LSD, p <0.038; Fig. 3b left panel). However, in the E0771 model, CD8+ or CD4+ T cells were not different among the three groups of control, responders, and nonresponders at 7 days post-PD-1 inhibitor treatment. As a proliferation marker Ki67 was incorporated in the analysis, CD8+Ki67+ T cells (11% vs. 2.5% in responders and controls, respectively; ANOVA p =0.001) and CD4+ Ki67+ T cells (24% and 10% in responders and controls, respectively; ANOVA p =0.033) were markedly increased in responders compared to PBS treatment control (Fig. 3b right panel).
Furthermore, we analyzed the early dynamic changes of CD8+, CD4+, and Ki67+ T cells from the pre-PD-1 inhibitor treatment to 7 days post-treatment based on responsiveness. In responders, CD8+Ki67+ and CD4+Ki67+ T cells were further increased in the PD-1 inhibitor treated mice at 7 days compared with pre-PD-1 inhibitor treatment (4.5-fold, p <0.001) for CD8+Ki67+ T cells and (2.5-fold, p <0.01) for CD4+Ki67+ T cells) compared to nonresponders (Fig. 3c).
Early serum cytokine changes to predict PD-1 inhibitor efficacy.
After confirming that peripheral blood CD8+, CD4+, and proliferative T cells started to increase early after a PD-1 inhibitor treatment in the responder group, we subsequently analyzed mouse serum cytokine concentrations to determine those early immunological reactions as represented by an early release of serum cytokines could predict the efficacy of PD-1 inhibitor treatment.
Fig. 4a shows the overview of how we did serum cytokine profiling in our mouse model. In order to use blood for both cytokine profiling and PBMC analysis, we collected approximately 200ul blood each time up to 3 times with retro-orbital bleeding technique. During and after the blood collection, we did not observe any adverse events in all mice, suggesting it’s a safe and reliable technique. Then, serum cytokine was analyzed by using bead-based assay, LEGENDplex, which relies upon sandwich immunoassay as shown in Fig. 4a. Thanks to the detection method of cytokines using flowcytometry, small amount of sample of 75 ml was sufficient.
Fig. 4b shows the early serum cytokine changes at 7 days post-PD-1 inhibitor treatment. Representatively in the E0771 model, the serum IFN-γ level was increased at 7 days post-PD-1 inhibitor treatment, and its expression was higher in responders compared to nonresponders (ANOVA,p =0.133 among the three groups; LSD,p =0.053) (Fig. 4b left panel). A similar expression pattern of IFN-γ was also observed in the JC model (Fig. 4b right panel). However, the TNF-α, IL-2, IL-6, and IL-10 levels did not statistically change (data not shown).
T cell infiltration in the tumor and proliferation in the spleen based on responsiveness after a PD-1 inhibitor treatment
To investigate the T lymphocyte infiltration into the tumor tissue and T cell proliferation in the spleen based on PD-1 inhibitor responsiveness, tumor tissue and spleen of E0771 syngeneic mice were analyzed by flow cytometry and IHC. Flow cytometry of tumor tissues showed a trend toward increased CD8+ and CD4+ T cells in responders compared to nonresponders (Fig. 5a left panel). Likewise, IHC analysis of infiltrated CD8+ or CD4+ T cells in tumor tissues at the time of sacrifice showed a significantly higher infiltration of CD8+ T cell (ANOVA,p =0.015 among the three groups; LSD,p < 0.019) and CD4+ T cells (ANOVA,p <0.001 among the three groups; LSD p <0.001) in responders compared to control group. Fig. 5a right panel). In the 4T1 syngeneic mouse model, we found a consistent pattern of higher tumor infiltration of CD8+ T cells in responders compared to nonresponders (Supplementary Fig. s1). Similarly, tumor IHC analysis showed that both CD8+ and CD4+ T cells were significantly higher in responders in comparison with nonresponders (ANOVA,p <0.001 among the three groups; LSD p <0.001) for CD8+ T cells and (ANOVA,p =0.016 among the three groups; LSD p =0.022) for CD4+ T cells between responders and nonresponders) (Supplementary Fig. s1b).
Regarding the T cell proliferation in the spleen by the responsiveness of PD-1 inhibitor therapy, CD4+ and CD4+Ki67+ T cells showed a trend toward a higher proportion in responders than in nonresponders (p =0.388 and p =0.588, Fig. 5b left panel). Similarly, in IHC analysis, more infiltrated CD8+ and CD4+ T cells were observed in responders compared to nonresponders by IHC (ANOVA,p =0.001 among the three groups; LSD p =0.001) for CD4+ T cells and (ANOVA,p =0.009 among the three groups; LSD p =0.007) for CD4+ T cells; Fig. 5b right panel).