Baseline ctDNA gene alterations as a biomarker of survival after panitumumab and chemotherapy in metastatic colorectal cancer

For unresectable RAS wild-type mCRC, first-line therapy typically combines chemotherapy with either an anti-EGFR antibody (panitumumab or cetuximab) or an anti-VEGF antibody (bevacizumab). PARADIGM previously showed longer overall survival with panitumumab+mFOLFOX6 versus bevacizumab+mFOLFOX6 in left-sided primary tumors and overall, with poorer survival in right-sided tumors. Differences in anti-EGFR outcomes are attributed to genomic profiles associated with primary resistance (for example, BRAF V600E and MSI-H). Guidelines recommend choosing anti-EGFR therapy based on primary tumor location and testing for RAS/BRAF and mismatch repair/MSI. This study’s research question: can ctDNA-based negative hyperselection (absence of a broad panel of resistance-associated alterations) predict which RAS WT mCRC patients benefit from first-line panitumumab, potentially beyond the predictive value of tumor sidedness and standard biomarkers?

Prior studies and guidelines underpin selection for anti-EGFR therapy. ASCO/ESMO guidelines advise testing for RAS (KRAS/NRAS) and BRAF and assessing MSI/MMR. Additional, less common alterations linked to anti-EGFR primary resistance include PTEN and EGFR ECD mutations, HER2/MET amplifications, and ALK/RET/NTRK1 fusions. “Negative hyperselection” panels combining multiple rare resistance alterations in tissue have predicted outcomes on anti-EGFR therapy and refined patient selection. Liquid biopsy (ctDNA) offers a minimally invasive alternative with high concordance to tissue for RAS detection and utility in tracking resistance evolution. This analysis extends negative hyperselection to ctDNA in a large phase 3 cohort.

Design: Prespecified exploratory biomarker analysis within PARADIGM (randomized, open-label, phase 3; NCT02394795) conducted at 197 sites in Japan (May 2015–Jan 2022). Patients with unresectable RAS WT mCRC (age 20–79; ECOG 0–1; measurable disease) were randomized 1:1 to panitumumab+mFOLFOX6 or bevacizumab+mFOLFOX6 with stratification by site, age (20–64 vs 65–79), and liver metastasis. Primary endpoint of PARADIGM: OS (tested hierarchically: left-sided then overall). Secondary endpoints: PFS, response rate, duration of response, curative resection; exploratory: depth of response. Biomarker analysis population included 733/802 (91.4%) with evaluable baseline plasma ctDNA who consented. ctDNA assay: Baseline plasma ctDNA (>10 ng/mL and >10 nmol DNA) analyzed with custom NGS panel (PlasmaSELECT-R 91; 90 mutations, 26 amplifications, 3 rearrangements; MSI assessment; targeted regions 250 kb). Analytical sensitivity: sequence mutations 92% (0.10% MAF) and 83% (0.20% MAF); amplifications (high-level): 100% at 20% purity; translocations: 100% at 0.10% purity; MSI: 100% at 0.50% purity. Specificity 99.9998%–100%. Definitions: Negative hyperselected = no prespecified alterations detected (KRAS, NRAS, BRAF V600E, PTEN, EGFR ECD exons 1–16 mutations; HER2 or MET amplification; ALK, RET, NTRK1 fusions). Gene altered = any of these detected. An additional exploratory grouping followed guidelines: MSS/MSI-L and RAS/BRAF WT vs MSI-H and/or RAS/BRAF mutation. Statistics: Kaplan–Meier for OS and PFS. Cox proportional hazards models (without stratification factors) for HRs and 95% CIs; interaction P values for treatment-by-biomarker status. Logistic regression for odds ratios of response and curative resection. Wilcoxon rank-sum for depth of response comparisons. Two-sided tests without multiple-comparison adjustment. Analyses in R 4.0.5 with standard packages. Follow-up: Median follow-up ~61 months in both arms at data cutoff (14 January 2022).

Population: 733 ctDNA-evaluable patients (554 left-sided, 169 right-sided, 10 both sides). Negative hyperselection achieved in 530/733 (72.3%): 79.4% of left-sided (440/554) vs 50.3% of right-sided (85/169). Among 203 (27.7%) gene-altered patients, most frequent alterations: BRAF V600E 10.6%, KRAS 6.0%, PTEN 5.5%; co-mutations were uncommon, more frequent in right-sided. Overall survival (OS): - All biomarker-evaluable: median OS 35.6 months with panitumumab vs 31.6 with bevacizumab; HR 0.87 (95% CI 0.73–1.02). - Negative hyperselected: panitumumab improved OS vs bevacizumab in left-sided (42.1 vs 35.5 months; HR 0.76; 95% CI 0.61–0.95), showed a favorable trend in right-sided (38.9 vs 30.9; HR 0.82; 95% CI 0.50–1.35), and improved overall (40.7 vs 34.4; HR 0.76; 95% CI 0.62–0.92; interaction P=0.037). - Gene altered: OS similar or inferior with panitumumab vs bevacizumab: left-sided 24.2 vs 26.4 months (HR 1.08; 95% CI 0.71–1.64); right-sided 14.1 vs 18.5 (HR 1.33; 95% CI 0.84–2.11); overall 19.2 vs 22.2 (HR 1.13; 95% CI 0.83–1.53). Progression-free survival (PFS): - Negative hyperselected: similar PFS between treatments (left-sided 14.0 vs 12.8 months; HR 0.91; 95% CI 0.73–1.13; right-sided 13.2 vs 11.3; HR 1.08; 95% CI 0.66–1.77; overall 13.6 vs 12.8; HR 0.92; 95% CI 0.75–1.12). - Gene altered: PFS similar in left-sided (9.3 vs 9.9; HR 1.45; 95% CI 0.94–2.23), but shorter with panitumumab in right-sided (6.3 vs 10.3; HR 2.25; 95% CI 1.36–3.70) and overall (7.8 vs 9.8; HR 1.68; 95% CI 1.23–2.29). Response rate (RR): - Negative hyperselected: higher with panitumumab vs bevacizumab (left-sided 83.3% vs 66.5%; OR 2.52; 95% CI 1.61–3.98; right-sided 71.4% vs 66.0%; OR 1.29; 95% CI 0.51–3.37; overall 81.5% vs 66.8%; OR 2.19; 95% CI 1.47–3.29). - Gene altered: similar or lower with panitumumab (left-sided 67.7% vs 73.5%; OR 0.76; 95% CI 0.33–1.70; right-sided 41.9% vs 65.9%; OR 0.37; 95% CI 0.15–0.89; overall 57.8% vs 69.1%; OR 0.61; 95% CI 0.34–1.09). Depth of response (median best change in target lesions): - Negative hyperselected: panitumumab superior (left-sided −60.2% vs −43.6%; right-sided −56.4% vs −39.4%; overall −60.2% vs −43.6%). - Gene altered: similar overall (−46.0% vs −45.1%); in right-sided, panitumumab inferior (−30.0% vs −53.3%). Curative resection rate: - Negative hyperselected: higher with panitumumab (left-sided 19.8% vs 10.6%; OR 2.10; 95% CI 1.23–3.66; overall 18.9% vs 11.1%; OR 1.87; 95% CI 1.15–3.09). - Gene altered: similar overall (11.0% vs 8.5%; OR 1.33; 95% CI 0.53–3.54). Standard biomarker subgroup (MSS/MSI-L and RAS/BRAF WT vs MSI-H and/or RAS/BRAF mut): - Among MSS/MSI-L and RAS/BRAF WT (81.6%), panitumumab tended to improve OS vs bevacizumab (overall 39.0 vs 34.1 months; HR 0.79; 95% CI 0.66–0.96) with higher RR and deeper responses; in MSI-H and/or RAS/BRAF mutation group, outcomes were inferior or similar with panitumumab (overall OS HR 1.27; 95% CI 0.88–1.84; PFS also tended shorter with panitumumab).

This large, prespecified ctDNA biomarker analysis demonstrates that negative hyperselection—defined as absence of a broad set of resistance-associated alterations in ctDNA—identifies patients who benefit most from first-line panitumumab plus mFOLFOX6, independent of primary tumor sidedness. Although right-sided disease is clinically associated with poorer outcomes and a higher prevalence of resistance alterations, negatively hyperselected right-sided patients still showed numerically longer OS, higher response, and greater depth of response with panitumumab. While individual alterations were too infrequent to draw definitive conclusions on single-gene effects, combining them clarified therapeutic impact and reflected their mutual exclusivity as oncogenic drivers. Compared with selection by tumor sidedness or by current standard biomarkers (RAS/BRAF and MSI), the broader ctDNA panel improved prediction of OS benefit for panitumumab, supporting ctDNA-based negative hyperselection as a more informative strategy for treatment selection in RAS WT mCRC. Safety profiles were comparable across biomarker strata. The findings suggest some right-sided patients, when negatively hyperselected, may derive benefit from anti-EGFR therapy; however, confirmation in additional studies is warranted.

Negative hyperselection using a validated ctDNA panel encompassing multiple resistance-associated gene alterations can guide first-line treatment choice in RAS WT mCRC, identifying patients—regardless of tumor sidedness—who derive improved overall survival, response rates, and depth of response with panitumumab plus mFOLFOX6 versus bevacizumab plus mFOLFOX6. The approach appears more predictive than tumor sidedness alone and may enhance selection beyond standard RAS/BRAF/MSI testing. Future research should validate these findings prospectively in diverse populations, refine detection of fusions/amplifications in ctDNA, address tissue–plasma discordance, and explore integration with additional biomarkers and dynamic ctDNA monitoring.

- Tumor tissue analyses were not comprehensively performed alongside plasma; discordance observed between locally assessed tissue RAS WT and ctDNA-detected RAS mutations (potentially due to spatial heterogeneity, timing of sampling, or assay sensitivity differences). - Detection of gene fusions and amplifications in ctDNA is technically challenging; copy number/fusion calls may be limited. - Clonal hematopoiesis–related variants were not specifically filtered beyond an algorithm to reduce false positives; some CHIP-related mutations may remain. - Possible misclassification in negative hyperselection for patients with low ctDNA shedding (undetectable mutations). Nevertheless, 87% had maximum VAF ≥1.0%. - Subgroup analyses were exploratory and underpowered for specific alterations or small strata (e.g., right-sided subgroups), with wide confidence intervals. - The study population was from a single country (Japan), which may affect generalizability.