Comprehensive genomic profiling by liquid biopsy portrays metastatic colorectal cancer mutational landscape to predict antitumor efficacy of FOLFIRI plus cetuximab in the CAPRI-2 GOIM trial

Abstract

Background: Limited evidence is currently available on the role of liquid biopsy (LBx) in predicting the efficacy of anti-epidermal growth factor receptor (EGFR) therapies in metastatic colorectal cancer (mCRC).

Methods: The CAPRI-2 GOIM is a phase II trial investigating the use of LBx-comprehensive genomic profiling (CGP)-guided, cetuximab-based treatment through three subsequent lines of therapy in patients with RAS/BRAF wild-type (WT) mCRC. LBx-CGP is carried out at baseline and at progressive disease to first- and second-line therapies. In case of RAS/BRAF WT circulating tumor DNA at progressive disease, EGFR therapeutic blockade is continued by combining cetuximab with a different chemotherapy backbone. The primary endpoint is overall response rate (ORR) by RECIST 1.1 criteria. Tumor molecular characteristics by LBx-CGP are correlated with treatment efficacy.

Results: One hundred and ninety-two RAS/BRAF WT microsatellite stable mCRC patients treated with FOLFIRI plus cetuximab with baseline LBx-CGP and assessable for response were included in the analysis. One hundred and thirty-seven patients with WT tumors for potential anti-EGFR drug resistance genes (RAS/BRAF/EGFR/PIK3CA/MAP2K1/MET/RET/ALK/ROS1/NTRK/NF1/FGFR, and HER2 amplification; 'negatively hyper-selected' cases) had 78.1% ORR compared with 54.5% ORR for patients with mutations [odds ratio 2.95, 95% confidence interval (CI) 1.44-6.10, P = 0.001]. 'Negatively hyper-selected' patients had median progression-free survival of 12.35 months (95% CI 10.58-15.4 months) compared with 8.68 months (95% CI 4.87-12.1 months) for patients with mutations (hazard ratio 0.64, 95% CI 0.44-0.92, P = 0.017). High cancer cell clonality of pathogenic variants (PVs) correlated with worse median progression-free survival (3.55 months, 95% CI 2.57 months to NE) compared with low cancer cell clonality of PV (9.63 months, 95% CI 7.16 months to NE, P = 0.21). After first-line therapy failure, approximately one out of five patients had acquired PVs of potential anti-EGFR drug resistance genes, whereas RAS/BRAF WT circulating tumor DNA was maintained in most patients (78.5%).

Conclusions: These results support the integration of LBx-CGP for implementing the efficacy and for optimizing the use of anti-EGFR therapies in RAS/BRAF WT mCRC.

Keywords: cetuximab; colorectal cancer; comprehensive genomic profiling; liquid biopsy.

Figure 1

Study diagram of the CAPRI-2 GOIM study. ctDNA, circulating tumor; mCRC, metastatic colorectal cancer; MSI, microsatellite instability; WT, wild-type.

Figure 2

Overall response rate and progression-free survival according to baseline anti-EGFR resistance molecular pathogenic variants. Overall response rate in the overall study cohort (A) and according to RAS/BRAF (B) and to ‘negatively hyper-selected’ tumors (C). A higher response rate was observed for WT tumors compared with RAS/BRAF^V600E mutated [OR 6.52 (95% CI 1.96-25.31), P = 0.0006] and ‘positive hyper-selected’ mutated tumors [OR 2.95 (95% CI 1.44-6.10), P = 0.001]. Kaplan–Meier curves of PFS according to RAS/BRAF (D) and to ‘negatively hyper-selected’ tumors (E), with RAS/BRAF^V600E [HR 3.03 (95% CI 1.79-5.22), P < 0.0001] and ‘positive hyper-selected’ mutated tumors [HR 1.55 (95% CI 1.08-2.23), P = 0.017] have lower PFS compared with WT tumors. CI, confidence interval; EGFR, epidermal growth factor receptor; HR, hazard ratio; MUT, mutant; OR, odds ratio; PFS, progression-free survival; WT, wild-type.

Figure 3

Prevalence and impact of baseline anti-EGFR resistance molecular pathogenic variants clonality. Oncoprint of baseline anti-EGFR resistance molecular PVs in 64/192 (33.3%) patients (A). Cancer cell clonality of RAS/BRAF^V600E and other anti-EGFR resistance PVs, with RAS/BRAF^V600E having higher clonality compared with other anti-EGFR resistance PVs (B). Association between PFS and cancer cell clonality of PV (C and D). Using an optimal cut-off of 27% for RAS/BRAF^V600E PVs, tumors with high-clonal RAS/BRAF^V600E PVs had lower PFS compared with tumors with low-clonal RAS/BRAF^V600E PVs (P = 0.02) (C). Using an optimal cut-off of 39% for all anti-EGFR drug resistance PVs, a similar negative impact of high-clonal PVs on PFS was observed (P = 0.21) (D). CI, confidence interval; EGFR, epidermal growth factor receptor; NE, not evaluable; PFS, progression-free survival; PV, pathogenic variant; SNV, single nucleotide variant.

Figure 4

Emergence of anti-EGFR drug resistance molecular pathogenic variant at disease progression to FOLFIRI plus cetuximab. Oncoprints comparing the frequency of anti-EGFR drug resistance PV at baseline (left) and at disease progression (right) among 106 cases with matched baseline and post-progression F1L CDx test, with tumor sidedness, TMB, and ctDNA TF annotations on top of the oncoprints matrix (A). Correlation between ctDNA TF and number of anti-EGFR drug resistance PVs: higher numbers of PVs were observed in case with higher ctDNA (Kendall’s tau 0.25, P = 0.02) (B). Prevalence of anti-EGFR drug resistance PV between baseline and post-progression samples. A statistically significant enrichment in post-progression compared with baseline samples was observed for KRAS (P < 0.001, q = 0.002) and EGFR ECD (P = 0.01, q = 0.22). Horizontal dotted line represents an alpha value of 5% (C). Kaplan–Meier plot of PFS for cases with acquired RAS/BRAF^V600E PVs: multiple RAS/BRAF^V600E PVs had trend for worse PFS (P = 0.15). Cases with baseline RAS/BRAF^V600E PVs were excluded from the analysis (D). Association between cancer cell clonality and clearance or persistence of PVs between baseline and disease progression time-points. For cases with cleared anti-EGFR drug resistance PVs, low cancer cell clonality was observed compared with high cancer cell clonality of PVs which were maintained at disease progression [0.65% (interquartile range 0.3% to 4.4%) versus 65.3% (interquartile range 6.0% to 100%), P = 0.018] (E). CI, confidence interval; ctDNA, circulating tumor DNA; EGFR, epidermal growth factor receptor; NE, not evaluable; PFS, progression-free survival; PV, pathogenic variant; TF, tumor fraction; TMB, tumor mutational burden.