La combinación del inhibidor de KRAS(G12C) adagrasib con inmunoterapia anti-PD-1 mejora la supervivencia global y previene la recurrencia en modelos preclínicos de metástasis cerebrales.

Cell Culture: CT26KRASG12C cells were gifted by Mirati Inc. and KPARKRASG12C were purchased from Cancer  Tools.^[18] Cells were tagged with firefly-luciferase-mCherry (FmC) using lentiviral transduction to generate CT26KRASG12C-FmC and KPARKRASG12C-FmC. Cells were maintained in culture as described in the Supplementary Material.

Dose-response assay: CT26 and KPAR cell lines were cultured in 96-well plate and tested with different concentrations of adagrasib to identify its IC50. After 72 h incubation with the compound, cell viability was assessed as luminescence using the CellTiterGlo kit.

Animal Studies: All animal studies were approved by the IACUC at Mass General Brigham (MGB) and conducted on 6-8-week-old female C57BL/6 and BALB/c mice from Charles River Laboratories. Mice were housed at the MGB Center for Comparative Medicine animal facility in a 12-h light-dark cycle with unlimited access to food and water.

Immunotypic Models of BM: We evaluated anti-tumor immune responses using previously described immunotypic models of BM, where mice were implanted with dual subcutaneous and intracranial tumors.^[19],^[20] CT26KRASG12C and KPARKRASG12C cell suspensions were injected subcutaneously (100,000 and 200,000 cells, respectively) three days before the intracranial tumor injections (50,000 CT26 cells; 100,000 KPAR cells). Cell doses were experimentally determined for each cell line, thus overall survival depends on the intracranial tumor growth. For tumor rechallenge, 200,000 and 400,000 cells of CT26 and KPAR, respectively, were injected in the left brain hemisphere, contralateral to the first tumor implantation. Intracranial tumor growth was monitored using bioluminescence imaging (BLI) at the MGB Center for System Biology Imaging Core as previously described.^[21]

Drug Administration: Animals received 100 mg/kg adagrasib or vehicle orally twice a day (BID) for three weeks starting from 5 days after intracranial tumor injection. Anti-PD-1 or isotype control IgG was administered at 10 mg/kg intraperitoneally every 3 days.

Gene Expression: KPAR and CT26 cells were plated in 6-well plates and treated with adagrasib at 500 nM and 1 µM concentrations or DMSO for 24 h. RNA purified from the cells was used to evaluate gene expression via RT-qPCR using the TaqMan probes described in the Supplementary Material.

T cell isolation and downstream analysis: T cells were obtained and activated from C57BL/6 or BALB/c mouse spleens, adapting the published STAR Protocol.^[21] Spleens were mechanically dissociated and filtered to obtain single cells. Red blood cells were removed using the Gibco ACK Lysis buffer. The remaining cells were counted and plated in pre-coated wells with anti-CD3e and CD28 antibodies for T cell activation. After 24 h, T cells were expanded by adding IL-2 and then passaged every 2 days till day 7, when they were used in co-culture with CT26 or KPAR luciferase-tagged tumor cells at different dilutions (1:1 or 1:100) with adagrasib (500 nM), anti-PD-1 (10 µg/mL), or their combination. Tumor cell viability was assessed by the One-Step Luc Assay kit.

Statistical Analysis

Statistical analyses used for each experiment are described in the figure legend. Generally, tumor growth experiments were analyzed using linear mixed models with log10-transformed tumor volume as the outcome.  The models included fixed effects for time, treatment group, and their interaction. Survival experiments were analyzed using Kaplan-Meier method. Multiple pairwise comparisons were adjusted using  Tukey-Kramer or Benjamini-Hochberg methods. SAS 9.4 and GraphPad Prism v.10.6.1 were used for statistical analyses.

There are no deposited data associated with this manuscript.

Statistical Analysis

Statistical analyses used for each experiment are described in the figure legend. Generally, tumor growth experiments were analyzed using linear mixed models with log10-transformed tumor volume as the outcome.  The models included fixed effects for time, treatment group, and their interaction. Survival experiments were analyzed using Kaplan-Meier method. Multiple pairwise comparisons were adjusted using  Tukey-Kramer or Benjamini-Hochberg methods. SAS 9.4 and GraphPad Prism v.10.6.1 were used for statistical analyses.

There are no deposited data associated with this manuscript.

Results

Efficacy of the KRASG12C Inhibitor Adagrasib, in Combination With Anti-PD-1 Immunotherapy in an Immunotypic Mouse Model of KRASG12C Colorectal Cancer BM

To evaluate KRASG12C targeted therapy in combination with anti-PD-1 immunotherapy, we identified murine cell lines that harbor the KRASG12C mutation, including the colorectal cancer cells, CT26-KrasG12C and the lung cancer cells, KPARKrasG12C (Figure 1A). We first tested their sensitivity to the selective KRASG12C inhibitor, adagrasib, in vitro using CellTiterGlo assay. The dose-response assay demonstrated that both cell lines have a similar sensitivity to this compound, with an IC50 of 311 nM in CT26 cells and 399.8 nM in KPAR cells, respectively (Figure 1A-C).

We then evaluated whether adagrasib, at two different concentrations (500 nM and 1 µM), altered gene expression in a set of immune-related genes in CT26 and KPAR cell lines. We confirmed downregulation of KRAS-dependent MAPK targets following adagrasib treatment and found upregulation of MHC class I genes as well as cytokines, including Cxcl10, in both models (Supplementary Figure S1A and B). We detected increased transcript levels of Cd274 encoding PD-L1 and receptor tyrosine kinase Axl only in KPAR cells, suggesting that adagrasib treatment may promote a more suppressive tumor immune microenvironment in the KPAR model (Supplementary Figure S1B).

Next, we assessed the efficacy of adagrasib in combination with anti-PD-1 therapy in syngeneic mouse models that mimic BM immunology (immunotypic models) using BALB/c or C57BL/6 mice. We generated BM models that involve subcutaneous (SC) and intracranial (IC) tumor implantation to recapitulate the immunologic features in BM.^[19],^[20] First, we established a CT26 dual subcutaneous intracranial model of BM, injecting CT26-KrasG12C cells SC, followed by IC implantation of CT26-KrasG12C-FmC tagged cells three days later. All treatments were initiated 5 days after intracranial tumor injection (Figure 1D). Adagrasib was administered twice a day (BID) via oral gavage at 100 mg/kg for three weeks, and anti-PD-1 antibody was injected intraperitoneally every three days (Q3D) at 10 mg/kg for five doses total. We monitored mice for body weight, which remained constant during the entire treatment period, suggesting good tolerability of the treatments (Supplementary Figure S2A). In addition, subcutaneous and intracranial tumor growth were assessed, quantifying subcutaneous tumor volume and intracranial bioluminescence (BLI), respectively (Figure 1E and F and Supplementary Figure S2B). Although mice receiving anti-PD-1 presented with significantly smaller SC tumor volume compared to those receiving vehicle (control group), tumor growth was completely abrogated in mice treated with adagrasib alone or in combination with anti-PD-1 (Figure 1E and Supplementary Figure S2B-E). Bioluminescence imaging showed similar anti-tumor effects intracranially, with a trend for the greatest benefit with the combination treatment (Figure 1F and Supplementary Figure S2F).

We performed an additional BLI follow-up at 3 weeks post-treatment (Supplementary Figure S2G). Seven out of 10 animals were intracranial tumor-free in the adagrasib-only and the combination groups, as compared with 5 out of 10 and 1 out of 10 animals in the anti-PD-1 group and vehicle group, respectively.### Treatment Responders Rejected Intracranial Rechallenge With CT26 Tumor Cells

Because of the observed CNS tumor regression, we continued long-term animal follow-up (Figure 2A). We assessed the long-term efficacy of the treatments with overall survival (Figure 2B and C) and intracranial tumor growth (Supplementary Figure S3A, week 0) at the 9-week timepoint after treatment discontinuation. We observed that, in the vehicle group, only 10% of the animals were still alive, whereas in the anti-PD-1 group, 50% survived; in the adagrasib and combination groups, 60% and 70% of animals survived, respectively (Figure 2B and C).

To determine whether treatments induced anti-tumor immune memory in long-term responders, we rechallenged the tumor-cleared mice, injecting the same tumor cells in the opposite brain hemisphere at a cell number four times higher than the original implantation (Figure 2A). A group of 4 mice that had never been injected with tumor cells (tumor naïve) served as a control for tumor growth. We monitored intracranial tumor growth using BLI (Figure 2D and Supplementary Figure S3A). As expected, after 2 weeks from the rechallenge, all the tumor naïve mice died (Supplementary Figure S3B), due to rapid tumor growth (Supplementary Figure S3A). In contrast, most of the rechallenged mice that rejected the original implantation remained alive with stable tumor regression during the 12-week follow-up period after the rechallenge (Figure 2B and C). More specifically, in both anti-PD-1 and adagrasib monotherapy groups, 50% of the animals survived, in the combination therapy group, 60% were still alive, and in the control group, only 10% were alive.

Together, these data suggest that adagrasib alone or in combination with immunotherapy can provide a statistically significant long-term benefit in overall survival and prevent long-term intracranial recurrences in BM derived from CT26 colorectal cancer cells.### Adagrasib Showed Intracranial Benefit in Combination With Anti-PD-1 Therapy in an ICI-Resistant KRASG12C Lung Cancer BM Model

Because CT26 cells are known to be relatively immunogenic, we tested adagrasib in combination with anti-PD-1 therapy in a model established with a KRASG12C-mutant cancer cell line, KPAR, which was reported to be ICI-resistant.^[14],^[18] We injected KPAR-KrasG12C cells in C57BL/6 mice to generate an immunotypic BM model and evaluated the same therapies as described above (Figure 3). Here, we used the same drug administration protocol and readouts adopted in the CT26-KrasG12C model (Figure 3A), including body weight (Supplementary Figure S4A) as well as subcutaneous and intracranial tumor growth (Figure 3B and C). Despite a reduction in tumor volumes in mice receiving anti-PD-1, in this model, it did not reach statistical significance compared to the vehicle group. Consistent with our results with CT26 cells, we again observed a complete remission of subcutaneous tumors in animals treated with both adagrasib alone or in combination with ICI (Figure 3B and Supplementary Figure S4B-E). Intracranially, a trend for a reduction in tumor growth was detected after 2 weeks of drug administration for both adagrasib and the combination groups compared to the control, which was statistically significant at 5 weeks (Figure 3C and Supplementary Figure S4F). In addition, anti-PD-1 treatment significantly controlled intracranial tumor growth compared to the vehicle (Figure 3C and Supplementary Figure S4F), but not to the extent observed for adagrasib and the combination groups.

We conducted the long-term follow-up of the surviving animals (Figure 4A). At week 13, while none of the animals in the vehicle group survived, 3 out of 10 animals in the ICI group were still alive, while 6 out of 9 in the adagrasib group, and 9 out of 10 in the combination group were alive (Figure 4B). However, we observed tumor regrowth in 2 of the 3 mice in the anti-PD-1 group, showing that in this model, tumor recurrences occurred after immunotherapy discontinuation, resulting in the exclusion of these animals from subsequent rechallenge studies.

We rechallenged the remaining animals with a higher dose of KPAR lung cancer cells as previously done for the CT26 model and monitored intracranial tumor growth and overall survival for 11 weeks (Figure 4C and D and Supplementary Figure S5A). We observed a clear distinction in the overall survival between different treatments, as 10%, 44%, and 80% of animals in the anti-PD-1, adagrasib alone, and combination group, respectively, achieved long-term survival (Figure 4B and C). The cumulative incidence of death analysis, assessed using Gray’s test, showed that the combination treatment conferred a significant survival benefit compared with the vehicle and anti-PD-1 alone groups as well as a trend for a greater survival compared to single agent adagrasib (Figure 4C and Supplementary Figure S5B).

Taken together, these findings highlight the efficacy of this combination strategy in long-term preclinical models of BM harboring KRASG12C mutations.

To understand whether adagrasib and anti-PD-1 enhanced T cell killing activity, we used 48h co-cultures of activated T cells and CT26 or KPAR cells at two different target-to-effector ratios (1:1 and 1:100) in the presence of adagrasib alone or in combination with anti-PD-1 and evaluated tumor cell viability (Supplementary Figure S6A-C). Our data showed that in both CT26 and KPAR models, we observed a significant decrease in tumor cell viability in the co-cultures treated with adagrasib alone or in combination with anti-PD-1 (Supplementary Figure S6B and C). Of note, although we detected the highest T cell killing of CT26 cells in the combination treatment at the 1:100 tumor-T cell ratio (Supplementary Figure S6B), we did not observe this with KPAR cells (Supplementary Figure S6C). Since our in vivo data showed the benefit of the combination treatment in both models, these data suggest different mechanisms of action between models that may not be fully recapitulated and detected in tumor-T cell co-cultures in vitro.

Efficacy of the KRASG12C Inhibitor Adagrasib, in Combination With Anti-PD-1 Immunotherapy in an Immunotypic Mouse Model of KRASG12C Colorectal Cancer BM

To evaluate KRASG12C targeted therapy in combination with anti-PD-1 immunotherapy, we identified murine cell lines that harbor the KRASG12C mutation, including the colorectal cancer cells, CT26-KrasG12C and the lung cancer cells, KPARKrasG12C (Figure 1A). We first tested their sensitivity to the selective KRASG12C inhibitor, adagrasib, in vitro using CellTiterGlo assay. The dose-response assay demonstrated that both cell lines have a similar sensitivity to this compound, with an IC50 of 311 nM in CT26 cells and 399.8 nM in KPAR cells, respectively (Figure 1A-C).

We then evaluated whether adagrasib, at two different concentrations (500 nM and 1 µM), altered gene expression in a set of immune-related genes in CT26 and KPAR cell lines. We confirmed downregulation of KRAS-dependent MAPK targets following adagrasib treatment and found upregulation of MHC class I genes as well as cytokines, including Cxcl10, in both models (Supplementary Figure S1A and B). We detected increased transcript levels of Cd274 encoding PD-L1 and receptor tyrosine kinase Axl only in KPAR cells, suggesting that adagrasib treatment may promote a more suppressive tumor immune microenvironment in the KPAR model (Supplementary Figure S1B).

Next, we assessed the efficacy of adagrasib in combination with anti-PD-1 therapy in syngeneic mouse models that mimic BM immunology (immunotypic models) using BALB/c or C57BL/6 mice. We generated BM models that involve subcutaneous (SC) and intracranial (IC) tumor implantation to recapitulate the immunologic features in BM.^[19],^[20] First, we established a CT26 dual subcutaneous intracranial model of BM, injecting CT26-KrasG12C cells SC, followed by IC implantation of CT26-KrasG12C-FmC tagged cells three days later. All treatments were initiated 5 days after intracranial tumor injection (Figure 1D). Adagrasib was administered twice a day (BID) via oral gavage at 100 mg/kg for three weeks, and anti-PD-1 antibody was injected intraperitoneally every three days (Q3D) at 10 mg/kg for five doses total. We monitored mice for body weight, which remained constant during the entire treatment period, suggesting good tolerability of the treatments (Supplementary Figure S2A). In addition, subcutaneous and intracranial tumor growth were assessed, quantifying subcutaneous tumor volume and intracranial bioluminescence (BLI), respectively (Figure 1E and F and Supplementary Figure S2B). Although mice receiving anti-PD-1 presented with significantly smaller SC tumor volume compared to those receiving vehicle (control group), tumor growth was completely abrogated in mice treated with adagrasib alone or in combination with anti-PD-1 (Figure 1E and Supplementary Figure S2B-E). Bioluminescence imaging showed similar anti-tumor effects intracranially, with a trend for the greatest benefit with the combination treatment (Figure 1F and Supplementary Figure S2F).

We performed an additional BLI follow-up at 3 weeks post-treatment (Supplementary Figure S2G). Seven out of 10 animals were intracranial tumor-free in the adagrasib-only and the combination groups, as compared with 5 out of 10 and 1 out of 10 animals in the anti-PD-1 group and vehicle group, respectively.

Treatment Responders Rejected Intracranial Rechallenge With CT26 Tumor Cells

Because of the observed CNS tumor regression, we continued long-term animal follow-up (Figure 2A). We assessed the long-term efficacy of the treatments with overall survival (Figure 2B and C) and intracranial tumor growth (Supplementary Figure S3A, week 0) at the 9-week timepoint after treatment discontinuation. We observed that, in the vehicle group, only 10% of the animals were still alive, whereas in the anti-PD-1 group, 50% survived; in the adagrasib and combination groups, 60% and 70% of animals survived, respectively (Figure 2B and C).

To determine whether treatments induced anti-tumor immune memory in long-term responders, we rechallenged the tumor-cleared mice, injecting the same tumor cells in the opposite brain hemisphere at a cell number four times higher than the original implantation (Figure 2A). A group of 4 mice that had never been injected with tumor cells (tumor naïve) served as a control for tumor growth. We monitored intracranial tumor growth using BLI (Figure 2D and Supplementary Figure S3A). As expected, after 2 weeks from the rechallenge, all the tumor naïve mice died (Supplementary Figure S3B), due to rapid tumor growth (Supplementary Figure S3A). In contrast, most of the rechallenged mice that rejected the original implantation remained alive with stable tumor regression during the 12-week follow-up period after the rechallenge (Figure 2B and C). More specifically, in both anti-PD-1 and adagrasib monotherapy groups, 50% of the animals survived, in the combination therapy group, 60% were still alive, and in the control group, only 10% were alive.

Together, these data suggest that adagrasib alone or in combination with immunotherapy can provide a statistically significant long-term benefit in overall survival and prevent long-term intracranial recurrences in BM derived from CT26 colorectal cancer cells.

Adagrasib Showed Intracranial Benefit in Combination With Anti-PD-1 Therapy in an ICI-Resistant KRASG12C Lung Cancer BM Model

Because CT26 cells are known to be relatively immunogenic, we tested adagrasib in combination with anti-PD-1 therapy in a model established with a KRASG12C-mutant cancer cell line, KPAR, which was reported to be ICI-resistant.^[14],^[18] We injected KPAR-KrasG12C cells in C57BL/6 mice to generate an immunotypic BM model and evaluated the same therapies as described above (Figure 3). Here, we used the same drug administration protocol and readouts adopted in the CT26-KrasG12C model (Figure 3A), including body weight (Supplementary Figure S4A) as well as subcutaneous and intracranial tumor growth (Figure 3B and C). Despite a reduction in tumor volumes in mice receiving anti-PD-1, in this model, it did not reach statistical significance compared to the vehicle group. Consistent with our results with CT26 cells, we again observed a complete remission of subcutaneous tumors in animals treated with both adagrasib alone or in combination with ICI (Figure 3B and Supplementary Figure S4B-E). Intracranially, a trend for a reduction in tumor growth was detected after 2 weeks of drug administration for both adagrasib and the combination groups compared to the control, which was statistically significant at 5 weeks (Figure 3C and Supplementary Figure S4F). In addition, anti-PD-1 treatment significantly controlled intracranial tumor growth compared to the vehicle (Figure 3C and Supplementary Figure S4F), but not to the extent observed for adagrasib and the combination groups.

We conducted the long-term follow-up of the surviving animals (Figure 4A). At week 13, while none of the animals in the vehicle group survived, 3 out of 10 animals in the ICI group were still alive, while 6 out of 9 in the adagrasib group, and 9 out of 10 in the combination group were alive (Figure 4B). However, we observed tumor regrowth in 2 of the 3 mice in the anti-PD-1 group, showing that in this model, tumor recurrences occurred after immunotherapy discontinuation, resulting in the exclusion of these animals from subsequent rechallenge studies.

We rechallenged the remaining animals with a higher dose of KPAR lung cancer cells as previously done for the CT26 model and monitored intracranial tumor growth and overall survival for 11 weeks (Figure 4C and D and Supplementary Figure S5A). We observed a clear distinction in the overall survival between different treatments, as 10%, 44%, and 80% of animals in the anti-PD-1, adagrasib alone, and combination group, respectively, achieved long-term survival (Figure 4B and C). The cumulative incidence of death analysis, assessed using Gray’s test, showed that the combination treatment conferred a significant survival benefit compared with the vehicle and anti-PD-1 alone groups as well as a trend for a greater survival compared to single agent adagrasib (Figure 4C and Supplementary Figure S5B).

Taken together, these findings highlight the efficacy of this combination strategy in long-term preclinical models of BM harboring KRASG12C mutations.

To understand whether adagrasib and anti-PD-1 enhanced T cell killing activity, we used 48h co-cultures of activated T cells and CT26 or KPAR cells at two different target-to-effector ratios (1:1 and 1:100) in the presence of adagrasib alone or in combination with anti-PD-1 and evaluated tumor cell viability (Supplementary Figure S6A-C). Our data showed that in both CT26 and KPAR models, we observed a significant decrease in tumor cell viability in the co-cultures treated with adagrasib alone or in combination with anti-PD-1 (Supplementary Figure S6B and C). Of note, although we detected the highest T cell killing of CT26 cells in the combination treatment at the 1:100 tumor-T cell ratio (Supplementary Figure S6B), we did not observe this with KPAR cells (Supplementary Figure S6C). Since our in vivo data showed the benefit of the combination treatment in both models, these data suggest different mechanisms of action between models that may not be fully recapitulated and detected in tumor-T cell co-cultures in vitro.

Discussion

BM are common in patients with advanced KRAS mutant lung cancer, and because the CNS is associated with unique immune and drug delivery challenges, their occurrence is associated with poor clinical outcomes.^[22-24] Over the past decade, KRAS-targeted therapies, such as the KRASG12C inhibitor, adagrasib, have been approved for treating solid tumors and are now being evaluated for BM. However, resistance and tumor recurrence commonly occur after treatment discontinuation, suggesting the need for combination strategies. Since the KRASG12C oncogenic mutation is known to promote tumor immune evasion with high levels of PD-L1,^[13] the combination of immunotherapy with KRAS-targeted therapy has been explored as a promising therapeutic approach in mouse models of extracranial disease,^[4] while its intracranial activity remains unexplored. In this study, we aimed to fill this gap, investigating the efficacy of this combination therapy in in vivo models of BM.

Although intracranial mouse models of BM have been instrumental in understanding solid tumor growth in the brain, these models may not recapitulate the BM immune microenvironment. The immune landscape of BM represents an ongoing challenge for in vivo studies. During metastatic cancer progression in humans, the host immune system has already been exposed to primary tumor antigens, thus making the traditional intracranial-only models not ideal for evaluating immunotherapies in the context of CNS cancer progression. Therefore, we created immunotypic BM models that better mimic the sequential exposure of the immune system, where we first implanted tumors subcutaneously and then intracranially, obtaining dual BM tumor models.^[19] In addition, it has been shown that because of the complexity of the BM microenvironment, intracranial ICI response may be primed by the presence of extracranial disease,^[19],^[20],^[25-27] underscoring the value of using dual-tumor BM models for evaluating combination therapies with anti-PD-1. Moreover, patients with BM are more likely to have a partial response to ICI compared to patients with primary brain tumors only. Using these dual-tumor models enables the study of ICI interventions in a system that may more accurately reflect the immunological context and clinical response of BM than intracranial implantation alone.

Herein, we demonstrate, for the first time, the intracranial efficacy of adagrasib and ICI combination using two KRASG12C dual tumor syngeneic mouse models, showing that combining adagrasib with anti-PD-1 immunotherapy can prevent BM recurrences originating from KRASG12C solid tumors. While in both BM models, we observed strong antitumor activity of adagrasib monotherapy, which effectively arrested subcutaneous tumor growth, we also reported model-specific responses to treatments. Specifically, in the colorectal BM model, both adagrasib and anti-PD-1 reduced intracranial tumor growth, and adagrasib significantly prolonged survival compared to vehicle. Importantly, in the lung cancer BM model established with KPAR cells, the combination treatment achieved a superior therapeutic response compared to monotherapy. These data are consistent with previous literature in the extracranial setting,^[5] and validate CNS penetrance and intracranial benefit of adagrasib in two different models, supporting the clinical relevance of our findings.

Another important consideration for BM patients treated with targeted therapy is the high risk of recurrence. In this context, we used tumor re-challenge experiments to mimic tumor recurrences. We demonstrated the benefit of combining KRAS targeted therapy and anti-PD-1 in preventing intracranial tumor recurrences for the long-term (up to 5 months of treatment discontinuation) in two models of BM. Although high response rates to the combination treatment were consistently noted across the two models, response to the single agents (anti-PD-1 and adagrasib only) was greater in the CT26 model. This observation may be due to differences in tumor immunogenicity and the tumor immune microenvironments (TIME) in the two models.^[14],^[18],^[28] Although both CT26 and KPAR cells were modified to have KrasG12C, they differ in genetic background and immune profile.^[29] Specifically, CT26 cells form tumors with high levels of immune infiltration characterized by TILs, such as CD8+T cells, Foxp3− CD4+T cells, and dendritic and NK cells.^[30] Consistent with our in vivo data, this tumor microenvironment likely renders this model more responsive to ICI.^[30],^[31] In contrast, KPAR cells, though highly immunogenic,^[32] engage in a broader immune response, characterized by an immune suppressive tumor microenvironment. This includes the presence of regulatory T cells and PD-L1+ myeloid cells, which promote immune evasion.^[32] The intrinsic ICI-resistant nature of KPAR tumors^[18] mirrors the immune-resistant profile often observed in BM and may explain its partial resistance to anti-PD-1 as single agent treatment, while showing encouraging response to the combination treatment. Of note, our in vitro data showed that adagrasib increased Cd274 (PD-L1) expression in KPAR cells, which may contribute to enhancing suppressive TIME and mediating an improved response to anti-PD-1. The efficacy of the combination treatment is likely due to a dual effect of adagrasib on immune and tumor cells. Adagrasib can block tumor cells from producing cytokines, arresting the recruitment of immune suppressive cell population, while promoting the recruitment of NK cells and activating cytotoxic T cells able to respond to anti-PD-1 effectively. Because the KRAS mutation induces a high expression of PD-L1^[9] and the overload of PD-1-PD-L1 interactions can overwhelm ICI alone, adagrasib-mediated reduction of PD-L1 levels on tumor cells may make ICI effective. Nevertheless, our findings suggest that pharmacological modulation of BM immune system may be needed for achieving a complete tumor remission and preventing recurrences. However, our investigation is limited by its focus on testing combination therapy for BM recurrence. Future work should explore the detailed mechanisms by which pharmacological targeting of the BM immune system influences tumor remission and relapse.

Treatment-induced reshaping of the TIME was characterized recently in preclinical extracranial models of lung cancer using spatial multiplex analysis.^[33] Our work represents a strong rationale for pursuing similar investigations intracranially. Such studies would help define the specific immune populations that mediate treatment-driven tumor clearance and clarify how single-agent or combination therapies affect these populations, using approaches such as flow cytometry and/or single-cell analysis.

Collectively, our data describe the benefit of using adagrasib in combination with anti-PD-1 immunotherapy as a therapeutic strategy to control growth and prevent recurrences of BM in two preclinical models of BM. Given the high intracranial recurrence rates in patients with BM and the limited effective treatment options, our work holds potential clinical relevance for this patient population. Our data support ongoing clinical investigations testing adagrasib in combination with pembrolizumab in KRAS-mutant lung cancer BM patients (KRYSTAL-7),^[34],^[35] thus offering new hope for the use of CNS-penetrant targeted therapies combined with ICI in a patient population that has been traditionally excluded from clinical trials.