Ligand-activated EGFR/MAPK signaling but not PI3K, are key resistance mechanisms to EGFR-therapy in colorectal cancer

Abstract

Understanding mechanisms of resistance to active therapies is crucial for developing more effective treatments. Here, we investigate resistance to anti-EGFR and anti-VEGF plus chemotherapy treatment in colorectal cancer (CRC) patients from the IMblaze370 trial (NCT02788279). While anti-VEGF does not select for secondary mutations, anti-EGFR leads to simultaneous mutations in EGFR and MAPK, but not PI3K pathway genes. Notably, we observe frequent acquired mutations in the EGFR extracellular but not intracellular domain and that patients with higher baseline expression of EGFR-ligands are prone to acquire resistant mutations. This data reveals a ligand-activated EGFR/MAPK-signaling dependency in CRC. We also observe enrichment for 8q gains in anti-EGFR treated patients, potentially linked to MYC amplification, a finding further supported by baseline expression analysis. This work adds to the evidence supporting broader evaluation of EGFR and pan-KRAS inhibitor combinations in CRC patients. It also underscores the utility of EGFR ligands as anti-EGFR efficacy biomarkers and provides a rationale for developing ligand blockers to complement and/or improve conventional anti-EGFR therapies in CRC.

Fig. 1. Comparison of genomic alterations detected in plasma samples at baseline (2/3 L + ) and in archival tumor tissues from IMblaze370 patients.

a, b Scatterplot showing the relationship between gene-level alteration frequencies in ctDNA and archival tissues from (a) 141 patients who received prior anti-VEGF therapies, and (b)113 patients who received prior anti-EGFR therapies. The effect size was estimated using Pearson’s correlation coefficient(r) calculated in GraphPad Prism, and the coefficient of determination (r²), representing the proportion of variance explained by the relationship, is displayed on the scatter plot. c Comparison of percentage of patients with emerging alterations between patients with prior anti-EGFR therapies and those with prior anti-VEGF therapies. Alterations that were absent from tumor tissue samples but detected in plasma samples were defined as emerging alterations. (All genes with at least one emerging alteration listed). Only known and likely oncogenic alterations in each gene were considered for the analysis. P-value was calculated using a two-tailed Fisher’s exact test to compare the percentage of patients with emerging alterations between those who received anti-EGFR therapies and those who received anti-VEGF therapies. Source data are provided as a Source Data file.

Fig. 2. Comparison of specific mutant allele frequencies of EGFR/MAPK pathway genes in plasma and tissue samples from IMblaze370 patients following anti-EGFR therapies (N = 113, top) or following anti-VEGF therapies (N = 141, bottom).

a KRAS. b EGFR. c MAP2K1. Only known and likely pathogenic alterations in each gene were considered for the analysis. P-value determined using a two-tailed Fisher’s exact test. Source data are provided as a Source Data file.

Fig. 3. KRAS^Q61H mouse colon organoids selectively activate MAPK but not PI3K signaling, and KRAS^Q61H CRC cell lines are more sensitive to MEK or pan-RAF inhibitor compared to cell lines with other KRAS mutations.

a Western blots to detect p-ERK1/2, P-S6, and P-AKT levels in organoids with different KRAS mutants. The data shown are representative of three independent experiments. b, c Comparison of relative viability of KRAS^Q61H CRC cell line with KRAS^G12D and KRAS^G13D CRC cell lines in response to MEK inhibitor GDC-0973 (b) and pan-RAF inhibitor GDC-5573 treatment (c). Data are presented as mean ± SEM and are representative of three independent experiments. Source data for panels are provided as a Source Data file.

Fig. 4. Landscape of gains and losses in IMblaze370 study.

Fractions of anti-EGFR (red) and anti-VEGF (blue) post-treatment patients with gains or losses based on 1 MB tiles. Source data are provided as a Source Data file.

Fig. 5. Identification of intrinsic biological characteristics associated with the activation of EGFR/MAPK pathway following anti-EGFR therapies.

a Cophenetic coefficient at various values of k test (k = 2–8). K = 2, with a cophenetic coefficient of 0.9752, was chosen. b Consensus patient similarity matrix depicting clusters (k = 2). The pie chart depicts the fraction of patients in each subtype with acquired mutations. c Heatmap patient samples sorted by NMF cluster assignment in the column, and the most significantly (FDR corrected Krueskal Wallis p-value < 0.05) NMF associated MSigDb hallmark pathways and xCell cell types clustered in row. P-value determined using a Kruskal-Wallis rank sum test. d Changes in tumor volume from baseline to 3 weeks post-cetuximab treatment in 244 PDX models with NMF 2.1(N = 221) and NFM2.2 (N = 23) tumors. The box and whisker plot shows the median, quartiles (hinges), whiskers, and individual outliers. P-value determined using a two-sided Wilcoxon rank sum test. Source data for panels (c and d) are provided as a Source Data file.

Fig. 6. Differentially expressed genes and pathways levels in RAS/BRAF-wild type anti-EGFR treated patients from the IMblaze370 study with acquired EGFR/MAPK pathway gene alterations and those without acquired alterations.

a Volcano plot representing differentially expressed genes between patients with and without acquired alterations. Genes with FDR-corrected p < 0.05. P-value determined using a Limma-Voom moderated t test. b Comparison of EGFR ligands (EREG and AREG) expression levels between the two groups. Data are presented as mean +/− SD. The p-value was generated using a two-tailed, unequal variances two-sample t test. c Heatmap patient samples sorted by acquired mutation status in the column and the most significantly (FDR corrected Krueskal Wallis p-value < 0.05) acquired mutation status associated MSigDb hallmark pathways and xCell cell types clustered in row. P-value determined using a Kruskal-Wallis rank sum test. d Changes in tumor volume from baseline to 3 weeks post-cetuximab treatment in 244 PDX models with EREG high (above median, n = 122) and EREG low (below median n = 122). The box and whisker plot shows the median, quartiles (hinges), whiskers, and individual outliers. P-value determined using a two-sided Wilcoxon rank sum test. Source data are provided as a Source Data file.