Supplementary Figure 7. Gastroesophageal cancers are enriched in KRAS wild-typeWT amplified tumors. Gastroesophageal cancers (n=3464) were selected from the AACR Genie MSK and DFCI cohort (version v16.0-public). Only patients with any alterations in the above listed genes are shown (1498 unaltered patients are not shown). Co-alterations are ranked by frequency. KRAS amplified samples are defined with a GISTIC score of 2 according to AACR GENIE as exact copy number thresholds are not available from the AACR GENIE cohort.

Supplementary Figure 1: Cellular sensitivity to BI-2493 integrated with CRISPR and RNAi gene dependency data. (Top) CRISPR: Drug-target associations show selectivity of BI-2493 for KRAS but not for HRAS and NRAS. Panel left, mid and right show gene effect (or dependency) scores derived from Chronos for KRAS, HRAS and NRAS, respectively on the x-axis. A low gene effect score indicates that a cell line is likely to depend on a given gene. A score equal to or close to 0 indicates genes that are non-essential, whereas a score of -1 is defined as the median of all common essential genes and is commonly used as a threshold for gene dependency (indicated by a red dotted vertical line). The y-axis shows drug sensitivity values reported as 1-AUC for BI-2493 derived from the PRISM screen. Larger values indicate higher sensitivity. KRAS Pearson R=-0.385, P=2.23e-23; HRAS Pearson R=0.079, P=0.0483; NRAS Pearson R=0.184, P=3.69e-06; (Bottom) RNAi: Drug-target associations show selectivity of BI-2493 for KRAS but not for HRAS and NRAS. Panel left, mid and right show gene effect (or dependency) scores derived from Demeter2 for KRAS, HRAS and NRAS, respectively on the x-axis. KRAS Pearson R=-0.393, P=2.44e-18; HRAS Pearson R=-0.0015, P=0.9; NRAS Pearson R=0.0698, P=0.14.

Supplementary Figure 9: BI-2493 and BI-2865 treatment induces cell cycle arrest and apoptosis in KRAS wild-type amplified cancer cell lines. (A) Upper panel: Representative flow blots of cell cycle states determined by EdU incorporation into newly synthesized DNA and total DNA content staining by FxCycle of indicated cell lines treated for 48 h with DMSO (mock), 3 µM BI-2493, 3 µM BI-2865 and 0.3 µM trametinib. Cells were pre-gated based on scattering properties and DNA content. Numbers indicate frequency of parent population. Lower panel: Impact of 48 h treatment with DMSO (mock), BI-2493, BI-2865 or trametinib at the indicated concentrations on cell cycle states of the indicated cell lines (N=3 (assay was run in triplicates), means + SD) (B) Upper panel: Representative flow blots of induction of apoptosis after 48 h of treatment of the indicated cell lines with DMSO (mock), 3 µM BI-2493, 3 µM BI-2865, 0.3 µM trametinib and 2 µM Camptothecin. Cells were pre-gated based on their scattering properties. Apoptotic cells were defined by loss of inner mitochondrial membrane potential (Δψm) and detection of phosphatidylserine by Annexin V staining. Numbers indicate frequency of parent population. Lower panel: Induction of apoptosis in the indicated cell lines by treatment with DMSO (mock), BI-2493, BI-2865, trametinib or Camptothecin at the indicated concentrations after 24 and 48 h determined by flow cytometry (N=3 (assay was run in triplicates), means +SD).

Abstract KRAS is the most frequently mutated oncogene, with high prevalence in indications with significant unmet clinical needs, such as lung, pancreatic and colorectal cancer. The most frequent alterations result in an amino acid exchange at position 12 from Glycine to Aspartic Acid (G12D), Valine (G12V) or Cysteine (G12C). The approval of mutant specific KRAS G12C inhibitors by the FDA opened a new field of treatment options for patients with KRAS G12C mutation and additional KRAS targeted therapy approaches are currently being clinically tested in the hope to provide benefit to cancer patients with cancers harboring other KRAS mutant alleles. These therapies include pan-RAS, pan-KRAS and allele selective inhibitors as well as degraders. For patients with advanced solid tumors harboring a KRAS G12D mutation, several selective inhibitors have entered clinical trials (MRTX1133, RMC-9805, QTX3034/46, LY3962673...) and RMC-6236, a pan-RAS inhibitor from Revolution Medicines, has recently shown promising early clinical data. However, alongside the great hopes placed in (K)RAS targeting therapy, resistance is likely to occur, as it was already observed in patients relapsing in response to KRAS G12C inhibitors. We aimed to use this pre-clinical study to predict and understand potential resistance mechanisms that may arise during targeted treatment against (K)RAS in the CRC setting. To this aim, the colorectal cancer cell line GP2d, expressing KRAS G12D, was continuously treated with a pan-RAS inhibitor to generate resistance. Once the resistance was confirmed, individual outgrowing clones were subsequently profiled in a series of assays to identify potential mechanism(s) and cross-tested with other inhibitors. Interestingly, cells were cross resistant to both pan-RAS (RMC-6236) and KRAS G12D (RMC-9805) inhibitors from Revolution Medicine, highlighting a common mechanism of action, while remaining sensitive to other KRAS G12D inhibitors. These findings were then confirmed in an orthogonal assay. Collectively our in vitro preclinical study identified a resistance mechanism to Tri-complex inhibitors in colorectal cancer cells. This resistance can still be addressed by other KRAS G12D inhibitors, opening options for patients harboring this characteristic post RMC-therapy. Future investigations may focus on characterizing tumors from patients who relapse on RMC-6236 and RMC-9805 to validate these findings clinically. Citation Format: Sabine Jurado, Simone Lieb, Marco H. Hofmann, Mark Pearson, Phillipp Schmalhorst, Krzysztof Zak. Acquired resistance to Tri-complex inhibitors in colorectal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1877.

Supplementary Table 1: PRISM data integrated with omics data from depmap Supplementary Table 2:Top 20 sensitive cell lines (PRISM) Supplementary Table 3:Frequency of KRAS WT AMP in AACR Genie v 15.1 (MSK cohort) Supplementary Table 4: Mututal exclusivity for KRAS WT AMP (results downloaded from AACR Genie v15.1) Supplementary Table 5 IC50 values from CTG assays for KRAS wt amplified cell lines

Supplementary Figure 8: KRAS wild-type amplified cancer cell lines are sensitive to pan-KRAS inhibitors BI-2493 and BI-2865. (A) Inhibition of pERK by BI-2493, BI-2865 and trametinib at the indicated timepoints and for the indicated cell lines (n=2, means ± SD). Control cell lines, cell lines with KRAS wild-type CN<7 and cell lines with KRAS wild-type CN>7 are colored in black, blue, and red, respectively. (B) Quantification of down regulation of pERK from (A) at 1M concentration of BI-2493 and BI-2865. P-values were calculated using two-way ANOVA, followed by Tukey’s multiple comparisons test.

Supplementary Figure 5. Anti-proliferative activity of RMC-7977 across different KRAS altered cell lines. (Left) Anti-proliferative activity of RMC-7977 (2) across different KRAS mutant or KRAS wild-type amplified cell lines. Cell lines are sorted by median sensitivity across KRAS alleles. Note: AUC values are relative measures of drug sensitivity and are therefore suitable to compare drug sensitivity across cell lines for a single compound but do not allow for a comparison across compounds. (Right) Comparison of sensitivity values for RMC-7977 (2) for cell lines with dependency on either KRAS, HRAS or NRAS. Cell lines with a Chronos score (gene effect score) of less than -1 were considered dependent. Sensitivity means between groups were tested for significance using a one-sided Wilcoxon-test.

Supplementary Figure 6: Relationship between KRAS wild-type amplification and KRAS oncogenic activity in TCGA patient data. RAS activation signatures MPAS (3), RAS_addiction and Ras84 (1). Enrichment scores were estimated using single sample enrichment (ssGSEA) in TCGA patient data. A one-sided Wilcox-test was used to test for significance between KRAS relative copy number of 2-7 or >7.

Supplementary Figure 4. Correlation between signatures for KRAS activation and sensitivity to BI-2493 across the KRAS wild-type amplified cell panel. RAS activation signatures were obtained from East et al. (1). Correlation coefficient between enrichment scores and sensitivity to BI-2493 were estimated using a Pearson R.

Supplementary Figure 10. BI-2493 treatment in animal models is tolerated. % bodyweight change in xenograft models treated with control vehicle or BI-2493. Data represent the mean % bodyweight change +/- SEM of mice grafted with: (A) DMS 53 cells (N=7); (B) MKN1 cells (N=7). One animal in the BI-2493 treated group had to be sacrificed earlier (d11) due to bodyweight loss. (C) ES11082 PDX model (N=8). Two animals in the control vehicle treated group and three animals in the BI-2493 treated group had to be sacrificed earlier (d18, d18, d15, d12, d7, respectively). (D) GA6871 PDX model (N=8). One animal in the control vehicle treated group and 2 animals in the BI-2493 treated group had to be sacrificed earlier (d7 and d16, d30, respectively) due to bodyweight loss.

KRAS WT–amplified cancer cell lines are sensitive to pan-KRAS inhibitors BI-2493 and BI-2865. A, Cell lines tested for sensitivity to pan-KRAS inhibitors were ranked according to KRAS WT relative CN. Tumor of origin and KRAS WT gene expression [log2(TPM + 1)] are indicated. B, In vitro sensitivity of cell lines shown in A to BI-2493, BI-2865, and trametinib (n = 3, means ± SD). Control cell lines, cell lines with KRAS WT CN < 7, and cell lines with KRAS WT CN > 7 are colored in black, blue, and red, respectively. C, Downregulation of DUSP6 mRNA by BI-2493, BI-2865, and trametinib at the indicated timepoints and for the indicated cell lines (n = 2, means ± SD). D, Quantification of downregulation of DUSP6 mRNA from C at 1 µmol/L concentration of BI-2493 and BI-2865. P values were calculated using two-way ANOVA, followed by the Tukey multiple comparisons test. NS, not significant.