Combined BRAF, EGFR, and MEK Inhibition in Patients with BRAFV600E-Mutant Colorectal Cancer

Abstract

Although BRAF inhibitor monotherapy yields response rates >50% in BRAF^V600-mutant melanoma, only approximately 5% of patients with BRAF^V600E colorectal cancer respond. Preclinical studies suggest that the lack of efficacy in BRAF^V600E colorectal cancer is due to adaptive feedback reactivation of MAPK signaling, often mediated by EGFR. This clinical trial evaluated BRAF and EGFR inhibition with dabrafenib (D) + panitumumab (P) ± MEK inhibition with trametinib (T) to achieve greater MAPK suppression and improved efficacy in 142 patients with BRAF^V600E colorectal cancer. Confirmed response rates for D+P, D+T+P, and T+P were 10%, 21%, and 0%, respectively. Pharmacodynamic analysis of paired pretreatment and on-treatment biopsies found that efficacy of D+T+P correlated with increased MAPK suppression. Serial cell-free DNA analysis revealed additional correlates of response and emergence of KRAS and NRAS mutations on disease progression. Thus, targeting adaptive feedback pathways in BRAF^V600E colorectal cancer can improve efficacy, but MAPK reactivation remains an important primary and acquired resistance mechanism.Significance: This trial demonstrates that combined BRAF + EGFR + MEK inhibition is tolerable, with promising activity in patients with BRAF^V600E colorectal cancer. Our findings highlight the MAPK pathway as a critical target in BRAF^V600E colorectal cancer and the need to optimize strategies inhibiting this pathway to overcome both primary and acquired resistance. Cancer Discov; 8(4); 428-43. ©2018 AACR.See related commentary by Janku, p. 389See related article by Hazar-Rethinam et al., p. 417This article is highlighted in the In This Issue feature, p. 371.

Figure 1

Targeting adaptive feedback signaling in BRAF^V600E CRC. A, Model of adaptive feedback signaling in BRAF^V600E CRC. Left, In the absence of drug, MAPK activity is driven by mutant BRAF, and ERK-dependent negative feedback signals constrain RTK-mediated activation of RAS. Center, BRAF inhibitor alone leads to transient inhibition of MAPK signaling and loss of ERK-dependent negative feedback signals, allowing RTK-mediated reactivation of the MAPK pathway through RAF dimers (including BRAF and CRAF). Right, Combined inhibition of BRAF, EGFR, and MEK is hypothesized to prevent adaptive feedback reactivation and maintain MAPK pathway suppression. B, Trial schematic showing treatment arms and dosing cohorts for treatment of patients with BRAF^V600E CRC. Note that patients treated at doses of dabrafenib 150 mg BID, trametinib 2 mg QD, and panitumumab at 6 mg/kg or dabrafenib 150 mg BID, trametinib 2 mg QD, and panitumumab at 4.8 mg/kg were enrolled into the dose escalation and dose expansion phases of the trial.

Figure 2

Efficacy of D+P, T+P, and D+T+P in patients with BRAF^V600E CRC. A-C, Waterfall plots showing best response by RECIST in the D+P (A), T+P (B), and D+T+P (C) cohorts. Dotted lines represent the 30% threshold for PR. Bar color represents the best confirmed response by RECIST. D, PFS for the D+P, T+P, and D+T+P cohorts. Median PFS with 95% CIs are shown for each treatment arm.

Figure 3

Pharmacodynamic analysis of paired tumor biopsy specimens. A, H-scores for pERK in paired baseline and day 15 on-treatment tumor biopsy specimens from patients treated with D+P, T+P, and D+T+P. P values represent paired t test. B, The percentage change in pERK H-score in the on-treatment tumor biopsy specimen relative to the baseline biopsy specimen in individual patients according to treatment. The percentage change in pERK H-score in paired on treatment biopsy specimens for patients with BRAF^V600E CRC treated with D+T and BRAF^V600-mutant melanoma treated with dabrafenib alone are shown for comparison. Horizontal bars represent the median.

Figure 4

Serial cfDNA analysis to define correlates of response and resistance. A, Percentage change in BRAF^V600E mutation levels in cfDNA (week 4 vs baseline) or CEA levels (week 6 vs baseline) for patients achieving CR/PR, stable disease, or progressive disease (PD). CEA analysis was limited to patients with baseline levels above the upper limit of normal. P values represent CR/PR vs stable disease/PD by 2-tailed t test. B, Scatterplot of correlation between change in BRAF^V600E mutation levels in cfDNA (week 4 vs baseline) or CEA levels (week 6 vs baseline) vs best percentage tumor change. Color of dots indicates the level of response achieved. C, Spider plots showing BRAF^V600E mutation levels in cfDNA or CEA levels during therapy for patients achieving CR/PR, stable disease, or PD. D, Three representative patients treated with D+T+P with serial cfDNA monitoring of BRAF^V600E mutation levels and hot spot KRAS and NRAS mutations at baseline, at week 4 of therapy, and at time of PD, showing emergence of 1 or more KRAS or NRAS mutations.