Standard first-line treatment for unresectable RAS wild-type (WT) metastatic colorectal cancer (mCRC) involves chemotherapy combined with either an anti-epidermal growth factor receptor (EGFR) antibody (e.g., panitumumab or cetuximab) or an anti-vascular endothelial growth factor (VEGF) antibody (bevacizumab). The phase 3 PARADIGM trial showed that first-line panitumumab plus modified FOLFOX6 (mFOLFOX6) resulted in longer overall survival (OS) compared to bevacizumab plus mFOLFOX6 in patients with left-sided primary tumors and in the overall population. However, exploratory analyses revealed poorer survival in patients with right-sided tumors. Genetic alterations and right-sided tumor location are associated with resistance to anti-EGFR treatment. This study aimed to evaluate the association between baseline circulating tumor DNA (ctDNA) gene alterations and treatment efficacy, focusing on a broad panel of alterations associated with resistance to EGFR inhibition. The use of ctDNA as a minimally invasive method for identifying patients most likely to benefit from anti-EGFR therapy is a promising approach, offering a potential alternative to tissue biopsy. Previous studies have explored a testing panel encompassing various rare genomic alterations linked to primary resistance to EGFR inhibition, demonstrating that the detection of these alterations in tumor biopsy samples predicted clinical outcomes. This study analyzed ctDNA in baseline plasma samples from patients with RAS WT mCRC within the PARADIGM trial to investigate the utility of ctDNA-based negative hyperselection in predicting treatment outcomes.

Several studies have established links between specific molecular alterations and resistance to anti-EGFR therapy in mCRC. The BRAF V600E mutation and high microsatellite instability (MSI-H) are well-known examples. Guidelines from the American Society of Clinical Oncology (ASCO) emphasize the importance of considering primary tumor location and testing for BRAF and RAS mutations and deficient mismatch repair (dMMR) or MSI status when deciding to initiate anti-EGFR treatment. Other less common alterations, including mutations in PTEN and EGFR extracellular domain (ECD), amplifications of HER2 and MET, and fusions of ALK, RET, and NTRK1, have also been implicated in primary resistance. While single molecular markers have guided drug development and patient selection, testing for combinations of multiple markers holds promise for more precise therapy selection. To achieve molecular negative hyperselection, studies have developed panels including various genomic alterations associated with resistance. These studies showed that the presence of these alterations, along with primary tumor sidedness, predicted clinical outcomes on anti-EGFR therapy. Liquid biopsy, using ctDNA, is a less invasive alternative to tissue biopsy for identifying patients who may benefit from anti-EGFR therapy.

This prespecified exploratory biomarker analysis of the phase 3 PARADIGM trial (NCT02394834) included 733 patients (91.4%) from the 802 patients in the PARADIGM efficacy analysis set (NCT02394795) who provided informed consent and had evaluable baseline plasma ctDNA samples. Patient ctDNA was assessed for various alterations (mutations, amplifications, rearrangements) in mCRC-related genes using a custom next-generation sequencing (NGS)-based panel. Negative hyperselection was defined as plasma ctDNA being negative for all prespecified gene alterations associated with resistance to anti-EGFR therapy. The study evaluated the association of negative hyperselection status (all negative versus gene altered) with overall survival (OS), progression-free survival (PFS), response rate, depth of response, and curative resection rate. Efficacy outcomes were also evaluated according to RAS, BRAF (V600E), and MSI status. OS was defined as the time from randomization to death from any cause, PFS was the time from randomization to disease progression or death (or surgery if no progression), and response rate was the percentage of patients with complete or partial response. Kaplan-Meier analysis was used for OS and PFS. Cox proportional hazard models were used for HR calculations, and logistic regression for ORs of response and resection rates. Statistical tests were two-sided without multiple comparisons adjustment. An additional exploratory analysis assessed genetic alterations of MSS/MSI status and RAS/BRAF mutations based on guideline recommendations.

In the overall biomarker-evaluable population, median OS was 35.6 months with panitumumab + mFOLFOX6 and 31.6 months with bevacizumab + mFOLFOX6 (HR 0.87; 95% CI, 0.73–1.02). For patients meeting negative hyperselection criteria (no gene alteration detected), OS was significantly longer with panitumumab versus bevacizumab (40.7 vs 34.4 months; HR 0.76; 95% CI, 0.62–0.92), This was also observed in patients with left-sided tumors (42.1 vs 35.5 months; HR 0.76; 95% CI, 0.61–0.95) and there was a trend towards longer OS in those with right-sided tumors (38.9 vs 30.9 months; HR 0.82; 95% CI, 0.50–1.35). For patients with any gene alteration, median OS was similar or inferior with panitumumab. Progression-free survival (PFS) was similar with panitumumab and bevacizumab in negative hyperselected patients, but shorter with panitumumab in gene-altered patients. Among negative hyperselected patients, response rates were higher with panitumumab in the left-sided population (83.3% vs 66.5%; OR 2.52; 95% CI, 1.61–3.98) and there was a similar trend in the right-sided population (71.4% vs 66.0%; OR 1.29; 95% CI, 0.51–3.37). The overall negative hyperselected population also showed a higher response rate with panitumumab (81.5% vs 66.8%; OR 2.19; 95% CI, 1.47–3.29). Median depth of response was greater with panitumumab in negative hyperselected patients. Curative resection rates were higher with panitumumab in negative hyperselected patients with left-sided tumors (19.8% vs 10.6%; OR 2.10; 95% CI, 1.23–3.66) and a similar trend was observed for right-sided tumors. In patients stratified by RAS/BRAF and microsatellite stability status, median OS was longer with panitumumab in the RAS/BRAF WT and MSS/MSI-L group (left-sided: 40.6 vs 34.8 months; HR 0.79; 95% CI, 0.64–0.97, overall: 39.0 vs 34.1 months; HR 0.79; 95% CI, 0.66–0.96) but similar or inferior in the MSI-H and/or RAS/BRAF mutation group.

This study suggests that negative hyperselection using a comprehensive ctDNA panel can identify patients with mCRC who may benefit from first-line panitumumab plus chemotherapy, regardless of tumor sidedness. The improved OS with panitumumab in negative hyperselected patients with left-sided tumors is potentially explained by higher response rates, curative resection rates, and depth of response. Even in patients with right-sided tumors, a numerical improvement in OS was observed in the negative hyperselected group. The high prevalence of genetic alterations in right-sided tumors (49.7% vs 26.0% in left-sided tumors) is consistent with previous reports. The study demonstrates that tumor sidedness may not be the sole determinant of response and that a comprehensive ctDNA panel may provide more precise patient selection compared to current guideline recommendations that rely on RAS/BRAF and MSS status alone. Although the ctDNA assay did not detect all alterations with high sensitivity and concordance rates between tumor and ctDNA were not perfect, the results suggest that using this combined approach offers improved patient selection for anti-EGFR antibody therapy.

This study demonstrates the potential of ctDNA-based negative hyperselection using a comprehensive panel of gene alterations for identifying patients with mCRC who may benefit from first-line panitumumab plus chemotherapy, regardless of tumor sidedness. The findings suggest that this approach may improve patient selection compared to current guidelines, although further studies are needed to confirm these results, particularly in the right-sided mCRC population. Future research should focus on validating these findings in independent cohorts and investigating the potential role of other biomarkers in predicting treatment response.

This study has several limitations. Detailed analyses of tumor specimens were not conducted. Discordance existed between tumor tissue and ctDNA results for RAS mutations. The reasons for this discordance (spatial heterogeneity, timing differences, assay sensitivity) require further investigation. The detection of gene fusions and amplifications in ctDNA was technically challenging, and some mutations may have been missed in ctDNA due to low ctDNA shedding. The study was not powered for comparisons between specific subgroups. Therefore, additional studies are needed to validate the findings.