Combination therapies promote immune-mediated bystander killing of KRAS-G12C inhibitor resistant cells

Although initial clinical responses to KRAS-G12C inhibitors (G12Ci) have been encouraging, their efficacy is limited by the rapid development of resistance. To extend therapeutic benefit, strategies capable of limiting or preventing resistance are required. However, given the heterogeneity of resistance mechanisms identified, designing targeted combination therapies remains challenging. Here, we hypothesised that a more broadly applicable approach would be to exploit the ability of KRAS-G12C inhibition to partially relieve the KRAS-induced immunosuppressive tumour microenvironment, thereby promoting immune-mediated attack on the inhibitor-resistant cancer cells. We developed a preclinical model to mimic development of resistance to G12Ci by co-engrafting reporter-traced isogenic cells harbouring KRAS-G12C (G12Ci-responsive) with a minor subpopulation of KRAS-G12D (G12Ci-resistant) cells. KRAS G12C inhibition as monotherapy using either the RAS(ON) G12C-selective inhibitor RMC-4998 or the KRAS G12C(OFF) inhibitor adagrasib, led to a rapid outgrowth of the G12Ci-resistant subpopulation. However, when combined with therapies that enhance anti-tumour immune responses, such as SHP2 inhibition or PD-1 blockade, the G12Ci-resistant cells were eliminated, even though these treatments do not affect their growth in the absence of G12Ci-sensitive cells. Notably, this bystander killing of G12Ci-resistant cells and the resulting complete responses were dependent on the adaptive immune system. These combination therapies led to a profound remodelling of the tumour immune microenvironment, with an increase of inflammatory macrophages and an influx of NK and T cells, including CD8 + T cells recognising the Emv2 endogenous retroviral protein, which represents the major tumour-associated antigen shared between the G12Ci-resistant and G12Ci-sensitive cancer cells. Moreover, transcriptional profiling suggested an enhanced interferon response in the G12Ci-resistant cells caused by the effect of G12Ci on the G12Ci-sensitive cells. Loss of IFNgamma receptor rendered the resistant cells less susceptible to immune-mediated bystander killing, indicating that this process was at least partly dependent on their tumour cell-intrinsic ability to respond to IFNgamma. Overall, our preclinical results demonstrate that appropriate combinations can elicit anti-tumour immune responses capable of bystander elimination of G12Ci-resistant subclones, providing a paradigm for the development of therapeutic combinations with greater potential to prevent or counteract the emergence of inhibitor resistance.