Targeted therapy in advanced BRAF-mutated colorectal cancer: systematic review and network meta-analysis

Bao-Dong Qin; Xiao-Dong Jiao; Zhan Wang; Ke Liu; Yan Ling; Ying Wu; Yuan-Sheng Zang

doi:10.1136/bmj-2025-086026

. 2025 Nov 19;391:e086026. doi: 10.1136/bmj-2025-086026

Targeted therapy in advanced BRAF-mutated colorectal cancer: systematic review and network meta-analysis

Bao-Dong Qin ¹, Xiao-Dong Jiao ¹, Zhan Wang ¹, Ke Liu ¹, Yan Ling ¹, Ying Wu ¹, Yuan-Sheng Zang ^1,^✉

PMCID: PMC12628362 PMID: 41355781

Abstract

Objective

To investigate the individual and comparative efficacy and safety of targeted therapy based strategies in advanced BRAF-mutated colorectal cancer.

Design

Systematic review and network meta-analysis.

Data sources

PubMed, Embase, Cochrane Library, and ClinicalTrials.gov from inception to 31 May 2025, and international conference proceedings.

Eligibility criteria for selecting studies

Clinical trials and multicentre, real world studies investigating the efficacy and safety of targeted therapies for advanced BRAF-mutated colorectal cancer with at least one of the clinical outcomes of interest.

Data synthesis

The primary endpoint was overall survival in the first line and second or later line settings. Secondary endpoints included progression-free survival, objective response rate, disease control rate, and grade ≥3 adverse events. Single arm meta-analysis, pairwise meta-analysis, and network meta-analysis were performed to pool hazard ratios with 95% credible intervals (CrIs) for overall survival and progression-free survival, and odds ratios with 95% CrIs for objective response rate, disease control rate, and adverse events. The rankogram and surface under the cumulative ranking curve (SUCRA) evaluated the relative superiority of regimens in the network meta-analysis.

Results

60 studies involving 4633 patients with advanced BRAF-mutated colorectal cancer were included. Pooled estimates indicated that patients could benefit from an anti-EGFR (epidermal growth factor receptor)/BRAF based regimen. Doublet chemotherapy (DCT)-anti-EGFR/BRAF was associated with the best overall survival, providing significant benefits compared with DCT-anti-VEGF (vascular endothelial growth factor) (hazard ratio 0.49, 95% CrI 0.36 to 0.66), triplet chemotherapy-anti-VEGF (0.51, 0.33 to 0.80), and anti-EGFR/BRAF (0.70, 0.51 to 0.96) regimens in the first line setting. Chemotherapy-anti-EGFR/BRAF showed significant superiority in overall survival (SUCRA=0.94 for DCT-anti-EGFR/BRAF, 0.90 for single agent chemotherapy (SCT)-anti-EGFR/BRAF) and progression-free survival (0.93 for DCT-anti-EGFR/BRAF and 0.92 for SCT-anti-EGFR/BRAF) among all first line targeted therapy strategies. In the second or later line setting, anti-EGFR/BRAF with or without an additional inhibitor (anti-MEK (mitogen-activated protein kinase kinase) or anti-PI3K (phosphoinositide 3-kinase)), exhibited better efficacy compared with other alternative strategies, ranking highest across study endpoints based on rank probability and SUCRA.

Conclusions

For initial treatment of advanced BRAF-mutated colorectal cancer, combining doublet chemotherapy with anti-EGFR/BRAF therapy offers the best survival benefit. For patients who have had previous treatment, anti-EGFR/BRAF regimens (with or without a MEK inhibitor) are the most effective and tolerable options.

Systematic review registration

PROSPERO CRD420250653959.

Introduction

Mutations of the BRAF gene represent critical molecular alterations in colorectal cancer, with the BRAF V600E variant being the most frequently observed subtype.¹ ² Notably, BRAF-mutated colorectal cancers account for about 8-15% of all patients with colorectal cancer, establishing it as a clinically distinct subtype characterised by highly aggressive tumours, including poor prognosis and intrinsic resistance to conventional chemotherapy.³ ⁴ ⁵ This subgroup of patients presents substantial challenges for treatment in advanced disease settings, particularly due to the limited efficacy of current regimens and the urgent need for optimal biomarker driven therapeutic strategies. Currently, anti-VEGF agents in combination with chemotherapy are often recommended as first line treatment for advanced BRAF-mutated colorectal cancer.⁶ ⁷ However, some important controversies persist about key clinical decisions, including optimal intensity of chemotherapy (see box 1 for definitions of abbreviations)—doublet (eg, FOLFOX or FOLFIRI) versus triplet (FOLFOXIRI) chemotherapy, and the selection of targeted therapy agents (anti-VEGF versus anti-EGFR (epidermal growth factor receptor) agents.⁸ ⁹ ¹⁰ ¹¹ ¹² ¹³ These debates stem from insufficient evidence, particularly the lack of large scale studies on this molecularly defined population.

Box 1. Definition of abbreviations.

BRAF—B-Raf proto-oncogene, serine/threonine kinase
EGFR—epidermal growth factor receptor
ERK—extracellular signal-regulated kinase
FOLFIRI—folinic acid, 5-fluorouracil, and irinotecan
FOLFOX—folinic acid, 5-fluorouracil, and oxaliplatin
FOLFOXIRI—folinic acid, 5-fluorouracil, oxaliplatin, and irinotecan
MEK—mitogen-activated protein kinase kinase
PD-1—programmed cell death protein-1
PI3K—phosphoinositide 3-kinase
VEGF—vascular endothelial growth factor
WNT—wingless/integrated

Owing to the limited efficacy of conventional therapeutic strategies for BRAF-mutated colorectal cancer, BRAF targeted therapies have been explored to improve therapeutic efficacy. BRAF inhibitor monotherapy has not yielded clinically significant improvements,¹⁴ ¹⁵ however, likely due to compensatory feedback reactivation of EGFR mediated mitogen-activated protein kinase signalling pathways, a well documented mechanism for resistance in preclinical models.¹⁶ ¹⁷ Subsequently, the finding of EGFR mediated feedback activation has driven the exploration of dual EGFR/BRAF blockade in BRAF-mutated colorectal cancer. Trials, including BEACON, have shown the efficacy of EGFR/BRAF co-inhibition in second or later line settings, whereas BREAKWATER and SWOG1406 further established the superiority of anti-EGFR/BRAF regimens combined with chemotherapy in both first line and later line treatments.¹⁸ ¹⁹ ²⁰ Although targeted therapy has been shown to improve outcomes, heterogeneity in the design of clinical trials raises important questions about the optimal line of therapy and combination strategies, including the optimal combination regimen among available strategies targeting VEGF and EGFR signalling pathways; the sufficiency of targeted therapy alone versus its combination with chemotherapy, and, if combined, whether single agent or doublet chemotherapy provides superior efficacy; the survival benefit of first line versus later line initiation of targeted based therapy; and the comparative safety profiles of targeted therapy based regimens, specifically whether they show a manageable increase in toxicity while maintaining acceptable safety.

To clarify these key clinical uncertainties, we conducted a systematic review and meta-analysis with two primary objectives: to integrate pooled efficacy and safety data for individual regimens through conventional meta-analysis, overcoming the limitations of small-sample studies, and to perform direct and indirect comparisons of targeted therapies through pairwise and network meta-analyses, addressing the scarcity of head-to-head trials. We identified current optimal first line and later line therapeutic strategies for BRAF-mutated colorectal cancer while guiding future trial design by prioritising high value combination approaches based on comparative effectiveness and safety profiles.

Methods

This systematic review, meta-analysis, and network meta-analysis was conducted using a predetermined protocol following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension statement for network meta-analyses (see supplementary table S1).²¹ The study protocol was registered in the Prospective Register of Systematic Reviews (PROSPERO).

Search strategy

We performed a systematic literature search of PubMed, Embase, the Cochrane library, and ClinicalTrials.gov for eligible studies investigating the efficacy and safety of targeted therapy based strategies for advanced BRAF-mutated colorectal cancer (last updated on 31 May 2025). We also searched abstracts and presentations from international conferences, such as the American Society of Clinical Oncology, European Society for Medical Oncology, and American Association for Cancer Research. The reference lists of relevant articles were checked for additional studies. Supplementary table S2 provides the complete search details.

Study selection and data extraction

To be included, studies had to meet the following criteria: phase 1, 2, and 3 clinical trials or multicentre, large, real world studies (published and unpublished); enrolled participants with histologically or cytologically confirmed advanced BRAF-mutated colorectal cancer; evaluated targeted therapy based regimens on at least one of the clinical outcomes of interest, including overall survival, progression-free survival, objective response rate, disease control rate, and grade ≥3 adverse events, or with sufficient data to enable calculation of these outcomes.²² ²³ We excluded studies that included patients with advanced BRAF-mutated colorectal cancer but did not independently report relevant data for the population with the mutation, contained insufficient published data or original data for meta-analysis, and focused on populations with non-BRAF V600E mutations. If studies reported updated data multiple times, we included information from the most recent version. All treatment regimens were classified according to their drug class or mechanism of action (eg, doublet chemotherapy, triplet chemotherapy, EGFR inhibitor (anti-EGFR), VEGF inhibitor (anti-VEGF), BRAF inhibitor (anti-BRAF), MEK inhibitor (anti-MEK)) irrespective of the specific agent type. This classification approach was motivated by both the well established therapeutic equivalence among regimens within the same drug class, as shown by clinical evidence and treatment guideline, and the necessity to maintain analytical robustness by preventing network fragmentation and preserving statistical validity. Two reviewers (B-DQ and X-DJ) independently screened titles and abstracts and then evaluated the full texts of potentially eligible studies for final inclusion. A third reviewer (Y-SZ) resolved any disagreements.

Two reviewers (B-DQ and XD-J) independently extracted relevant data from each study using a prespecified form, including study characteristics (first author, publication year, country, study identification, trial registration number, study design), demographic information (sample size, data source), treatment (regimen, treatment line), and outcomes (hazard ratios for overall survival and progression-free survival, and the number of participants with objective response rate, disease control rate, and grade ≥3 adverse events). We applied several requirements during data extraction: although the study reported data on multiple BRAF mutation subtypes, we prioritised the analysis of BRAF V600E variants; we preferred blinded independent central review assessed data derived from the intention-to-treat population; we preferred treatment related grade ≥3 adverse events; we prioritised the latest updated data from repeated reports of individual trials with varying follow-up periods for inclusion; we deemed targeted therapy based regimens with fewer than 10 patients to be insufficient for statistical analysis; and we exclusively used the data from single arm studies for single arm meta-analyses of objective response rate, disease control rate, and grade ≥3 adverse events, whereas they were not incorporated into pairwise meta-analysis or network meta-analyses as these studies lacked comparator arms. During data extraction, study characteristics were categorised according to prespecified classifications, including data source (trial level data versus subgroup level data), BRAF mutation type (BRAF V600E versus mixed type), treatment history (uniform treatment versus mixed treatment), and publication type (published literature versus conference report).

Risk assessment of bias across included studies

We evaluated the methodological quality of the included studies using validated assessment tools: the Cochrane Risk of Bias tool version 2 (RoB 2) for randomised controlled trials,²⁴ the Methodological Index for Non-randomised Studies for single arm interventional studies and non-randomised controlled trials,²⁵ the Assessment of Real-World Observational Studies framework and Risk Of Bias In Non-randomized Studies of Interventions for real world evidence investigations²⁶ (see supplementary tables S3-S6). In addition, the Confidence in Network Meta-Analysis (CINeMA) framework was used to evaluate the confidence in effect estimates across six domains: within study bias, reporting bias, indirectness, imprecision, heterogeneity, and incoherence. Two reviewers (B-DQ and X-DJ) independently evaluated the methodological quality and risk of bias of the individual studies, with a third reviewer (Y-SZ) resolving any disagreements by consensus.

Statistical analysis

The enrolled studies were stratified by treatment line: first line therapy versus second or later line therapy. The primary outcome analysed was overall survival, and the secondary outcomes analysed were progression-free survival, objective response rate, disease control rate, and grade ≥3 adverse events. Data analyses were performed in R (version 4.3.2) by implementing three core methods: single arm meta-analysis and pairwise meta-analyses using the meta package, and bayesian network meta-analysis using the gemtc package.

A single arm meta-analysis was performed to synthesise pooled estimates for objective response rate, disease control rate, and the occurrence of grade ≥3 adverse events for each regimen, based on all available arm level data from the included studies reporting quantifiable outcome measures. We restricted pairwise meta-analyses to direct comparisons between two or more controlled trials. Hazard ratios with 95% credible intervals (CrIs) were used to quantify time-to-event outcomes, including progression-free survival and overall survival, and odds ratios with 95% CrIs were used to assess binary endpoints, including objective response rate, disease control rate, and grade ≥3 adverse events. Heterogeneity across studies was evaluated using the χ² statistic based on Cochran’s Q and was quantified using the inconsistency index (I²) test. When I² was >50% or the P value based on Cochran’s Q was <0.10, we implemented random effects models. Otherwise, we used fixed effects models.

Network meta-analysis was performed in a bayesian framework using the Markov Chain Monte Carlo simulation technique, comparing any two targeted therapy based strategies by simultaneously synthesising direct and indirect evidence. We generated network plots for each outcome to illustrate the geometries and clarify which treatments were compared directly or indirectly across the included studies. A random effects consistency model was applied to allow for heterogeneity between studies in the treatment comparison effects. Markov Chain Monte Carlo sampling was performed with four parallel chains, each consisting of 20 000 iterations, after a 5000 iteration adaptation phase for algorithm tuning using JAGS.

We used trace plots, density plots, and the Brooks-Gelman-Rubin method²⁷ to evaluate convergence of the modes (see supplementary figures S1 and S2). Within the bayesian framework, the network meta-analysis estimated the overall rankings of treatments by calculating the surface under the cumulative ranking curve (SUCRA) and rankogram: 0 denoted the worst treatment regimen (certainly the least efficacious or toxic treatment) and 1 denoted the best treatment regimen.

To ensure the transitivity of the key assumptions underlying the network meta-analysis, we included studies with strict patient allocation and an optimised balance for the same conditions. The transitivity assumption underlying indirect comparisons was evaluated using bayesian meta-regression analyses, which examined the potential modifying effects of covariates, including sample size, study design, data source, mutation type, treatment history, and publication type. Inconsistency was evaluated locally by comparing the pooled estimates from network and pairwise meta-analyses, and globally by comparing the fit and parsimony of consistency and inconsistency models using the deviance information criterion value.²⁸ Inconsistencies in the entire network were also assessed using the node splitting method, in which direct and indirect evidence were compared separately.

Patient and public involvement

No patients or members of the public were directly involved in this study as it was a secondary analysis of existing trial data. Limited funding, methodological complexity of network meta-analysis, and time constraints precluded formal patient and public involvement engagement. Importantly, this research directly addresses priorities identified by patients with advanced BRAF-mutated colorectal cancer, specifically, the need for rigorous efficacy and safety evaluations of targeted regimens to optimise treatment outcomes.

Results

Figure 1 shows the flow chart of studies through this systematic review and meta-analysis. We retrieved a total of 4134 citations from the databases and identified 173 records from conference proceedings. After the removal of duplicate records and the screening of abstracts and titles, 111 studies were further assessed through a full text review. Overall, 60 studies were eligible for inclusion: 25 single arm clinical trials,⁵ ¹⁴ ²⁹ ³⁰ ³¹ ³² ³³ ³⁴ ³⁵ ³⁶ ³⁷ ³⁸ ³⁹ ⁴⁰ ⁴¹ ⁴² ⁴³ ⁴⁴ ⁴⁵ ⁴⁶ ⁴⁷ ⁴⁸ ⁴⁹ ⁵⁰ ⁵¹ 28 controlled clinical trials,⁴ ⁹ ¹⁰ ¹¹ ¹⁸ ¹⁹ ²⁰ ⁵² ⁵³ ⁵⁴ ⁵⁵ ⁵⁶ ⁵⁷ ⁵⁸ ⁵⁹ ⁶⁰ ⁶¹ ⁶² ⁶³ ⁶⁴ ⁶⁵ ⁶⁶ ⁶⁷ ⁶⁸ ⁶⁹ ⁷⁰ ⁷¹ ⁷² and seven multicentre real world studies,⁷³ ⁷⁴ ⁷⁵ ⁷⁶ ⁷⁷ ⁷⁸ ⁷⁹ comprising 4633 patients with advanced BRAF-mutated colorectal cancer (see supplementary table S7).

Fig 1 — Flow chart of studies in systematic review and meta-analysis

First line setting

Of the 60 included studies, 32 focused on first line therapy in 2460 patients with advanced BRAF-mutated colorectal cancer. The regimens were categorised into 12 types (see supplementary table S8, also see box 1 for definitions of abbreviations): EGFR inhibitor based combinations (doublet chemotherapy-anti-EGFR, triplet chemotherapy-anti-EGFR), VEGF inhibitor based combinations (doublet chemotherapy-anti-VEGF, triplet chemotherapy-anti-VEGF), EGFR/VEGF inhibitor based combinations (doublet chemotherapy-anti-EGFR/VEGF), EGFR/BRAF inhibitor based combinations (single agent chemotherapy-anti-EGFR/BRAF, doublet chemotherapy-anti-EGFR/BRAF), chemotherapy only regimens (doublet or triplet chemotherapy only), targeted therapy only regimens (anti-EGFR/BRAF, anti-EGFR/BRAF/MEK), and VEGF/PD-1 inhibitor based combinations. Regimens for triplet chemotherapy only and VEGF/PD-1 inhibitor based combinations were excluded from subsequent meta-analyses owing to fewer than 10 patients in each regimen.

Single arm meta-analysis

Objective response rate—In the single arm meta-analysis, 25 studies evaluated objective response rate and 11 evaluated disease control rate in a first line setting. Overall, the pooled objective response rate across all regimens was 44.2% (95% CrI 37.7% to 50.8%). For individual regimens, the pooled objective response rates were 39.7% (30.5% to 49.7%) for anti-EGFR based combinations, 42.7% (37.5% to 47.9%) for anti-VEGF based combinations, and 65.8% (58.2% to 72.7%) for anti-EGFR/BRAF based combinations. Doublet chemotherapy-anti-EGFR/BRAF showed the highest objective response rate (67.0%, 58.3% to 74.6%), followed by single agent chemotherapy-anti-EGFR/BRAF (62.0%, 45.3% to 76.2%) (see supplementary figure S3).

Disease control rate—The pooled estimates for disease control rate were 73.2% (95% CrI 63.0% to 81.3%) for anti-EGFR based regimens, 73.1% (66.9% to 78.6%) for anti-VEGF based regimens, and 87.7% (81.1% to 92.3%) for anti-EGFR/BRAF based regimens.

Adverse events—Doublet chemotherapy-anti-EGFR/BRAF was associated with the highest risk of grade ≥3 adverse events, and anti-EGFR/BRAF was associated with the lowest risk.

Pairwise meta-analysis

A pairwise meta-analysis was conducted for the overall survival, progression-free survival, and objective response rate in the first line setting (see supplementary figure S4).

Overall survival—No statistically significant differences were observed in the available head-to-head comparison analysis for overall survival (triplet chemotherapy-anti-VEGF versus doublet chemotherapy-anti-VEGF, doublet chemotherapy-anti-EGFR versus doublet chemotherapy, doublet chemotherapy-anti-VEGF versus doublet chemotherapy, and doublet chemotherapy-anti-EGFR versus doublet chemotherapy-anti-VEGF).

Progression-free survival—Anti-EGFR/BRAF combination regimens (single agent chemotherapy-anti-EGFR/BRAF) showed significantly prolonged progression-free survival than anti-VEGF regimens (doublet chemotherapy-anti-VEGF) (hazard ratio 0.55, 95% CrI 0.35 to 0.85).

Objective response rate—Single agent chemotherapy-anti-EGFR/BRAF regimens showed a significantly higher objective response rate than doublet chemotherapy-anti-VEGF (odds ratio 5.17, 95% CrI 2.51 to 10.7) regimens.

Network meta-analysis

Overall, 20 studies (n=2138) with nine first line regimens were included in the network meta-analysis for overall survival, 19 studies (n=1691) with nine first line regimens in the network meta-analysis for progression-free survival, and 17 studies (n=1490) with nine first line regimens in the network meta-analysis for objective response rate (fig 2).

Fig 2 — Eligible comparisons for efficacy outcomes of first line and second or later line regimens in network meta-analysis. Network plots illustrate direct and indirect comparisons for overall survival, progression-free survival, and objective response rate. Nodes represent each regimen, with size proportional to number of participants. Lines represent direct comparisons between regimens, with thickness indicating number of direct comparisons. Indirect comparisons arose from combining direct comparisons within the network. BRAF=B-Raf proto-oncogene, serine/threonine kinase; Chemo=chemotherapy; DCT=doublet chemotherapy; EGFR=epidermal growth factor receptor; MEK=mitogen-activated protein kinase kinase; PI3K=phosphoinositide 3-kinase; TCT=triplet chemotherapy; VEGF=vascular endothelial growth factor

Overall survival—Anti-EGFR/BRAF based regimens were found to be the most effective therapeutic approaches. Both doublet chemotherapy-anti-EGFR/BRAF and single agent chemotherapy-anti-EGFR/BRAF had comparable overall survival benefits (hazard ratio 0.95, 95% CrI 0.52 to 1.74), showing significant superior benefit compared with other regimens. Doublet chemotherapy-anti-EGFR/BRAF yielded the best overall survival, with significant benefits compared with doublet chemotherapy (hazard ratio 0.42, 95% CrI 0.28 to 0.63), doublet chemotherapy-anti-EGFR (0.48, 0.31 to 0.74), doublet chemotherapy-anti-VEGF (0.49, 0.36 to 0.66), doublet chemotherapy-anti-EGFR/VEGF (0.42, 0.24 to 0.72), triplet chemotherapy-anti-EGFR (0.37, 0.17 to 0.76), triplet chemotherapy-anti-VEGF (0.51, 0.33 to 0.80), and anti-EGFR/BRAF (0.70, 0.51 to 0.96) regimens. Similarly, single agent chemotherapy-anti-EGFR/BRAF regimens showed a statistically significant advantage in overall survival benefits compared with the alternative treatment strategies (doublet chemotherapy: hazard ratio 0.44, 95% CrI 0.27 to 0.76, doublet chemotherapy-anti-EGFR: 0.50, 0.29 to 0.90, doublet chemotherapy-anti-VEGF: 0.51, 0.31 to 0.86, doublet chemotherapy-anti-EGFR/VEGF: 0.43, 0.22 to 0.87, triplet chemotherapy-anti-EGFR: 0.38, 0.16 to 0.94) (fig 3).

Fig 3 — Pooled efficacy estimates of multiple comparisons in first line setting based on network meta-analysis. Data are pooled hazard ratios (95% credible intervals) for overall survival and progression-free survival. Bold data indicate statistical significance. BRAF=B-Raf proto-oncogene, serine/threonine kinase; Chemo=chemotherapy; DCT=doublet chemotherapy; EGFR=epidermal growth factor receptor; TCT=triplet chemotherapy; VEGF=vascular endothelial growth factor

Progression-free survival—Anti-EGFR/BRAF based regimens (doublet chemotherapy-anti-EGFR/BRAF and single agent chemotherapy-anti-EGFR/BRAF) also showed a statistically significant improvement in progression-free survival compared with traditional anti-EGFR based or anti-VEGF based combinations, as well as anti-EGFR/BRAF (hazard ratio 0.49, 95% CrI 0.35 to 0.68; 0.50, 0.28 to 0.90; respectively) (fig 3).

Objective response rate and disease control rate—Objective response rate showed improvement with dual inhibition of EGFR and BRAF compared with the EGFR based or VEGF based combination regimens (see supplementary figures S5 and S6). A similar trend of high disease control rate was observed for the anti-EGFR/BRAF based regimen; the difference was not, however, significant.

Adverse events—Anti-EGFR/BRAF showed a more favourable safety for grade ≥3 adverse events than other strategies, although without statistical significance.

Bayesian ranking profile

The bayesian ranking profiles of regimens were almost consistent with those of the pooled analyses using hazard ratios and odd ratios, including rankogram (fig 4), and SUCRA (fig 5). Anti-EGFR/BRAF based regimens consistently ranked among the top two in efficacy outcomes: doublet chemotherapy-anti-EGFR/BRAF showed the highest overall survival and progression-free survival (SUCRA 0.94 and 0.93, respectively) followed by single agent chemotherapy-anti-EGFR/BRAF (0.90 and 0.92, respectively), whereas single agent chemotherapy-anti-EGFR/BRAF achieved superior objective response rate and disease control rate (0.92 and 0.88, respectively). In the safety analysis, anti-EGFR/BRAF showed significant superiority, whereas doublet chemotherapy-anti-EGFR/BRAF ranked last (0.91 and 0.27, respectively) among the five regimens with reported grade ≥3 adverse events.

Fig 4 — Bayesian ranking profiles of comparable regimens on efficacy and safety outcomes in first line setting and second or later line setting. An interactive version of this graphic is available at https://public.flourish.studio/visualisation/25802694/

Fig 5 — Cumulative ranking plots for comparable regimens on efficacy and safety outcomes in first line setting and second or later line setting. An interactive version of this graphic is available at https://public.flourish.studio/visualisation/25803421/

Second or later line setting

Thirty studies investigated second or later line therapies for advanced BRAF-mutated colorectal cancer in a pooled population of 2173 patients (see supplementary table S7). The second or later line regimens were categorised into 14 types (see supplementary table S8): chemotherapy, chemotherapy-EGFR inhibitor (chemotherapy-anti-EGFR), chemotherapy-EGFR and BRAF inhibitor (chemotherapy-anti-EGFR/BRAF), chemotherapy-EGFR, BRAF, and MEK inhibitors (chemotherapy-anti-EGFR/BRAF/MEK), BRAF inhibitor (anti-BRAF), BRAF inhibitor plus MEK inhibitor (anti-BRAF/MEK), EGFR inhibitor-MEK inhibitor (anti-EGFR/MEK), EGFR inhibitor-BRAF inhibitor (anti-EGFR/BRAF), and anti-EGFR/BRAF with MEK inhibitor (anti-EGFR/BRAF/MEK), with PI3K inhibitor (anti-EGFR/BRAF/PI3K), with ERK inhibitor (anti-EGFR/BRAF/ERK), with WNT inhibitor (anti-EGFR/BRAF/WNT), and with PD-1 inhibitor (anti-EGFR/BRAF/PD-1), and BRAF inhibitor-MEK inhibitor with PD-1 inhibitor (anti-BRAF/MEK/PD-1).

Single arm meta-analyses

In the single arm meta-analysis, 29 studies evaluated objective response rate and 24 evaluated control rate in second line or later line settings. Overall, the pooled objective response rate across all regimens was 22.6% (95% CrI 19.0% to 26.6%) (see supplementary figure S7). The pooled objective response rate was highest for the anti- EGFR/BRAF/PD-1 regimen (36.9% 18.7% to 59.8%), followed by chemotherapy-anti-EGFR/BRAF (33.3%, 9.0% to 71.5%) and anti-EGFR/BRAF/MEK (27.1%, 21.8% to 33.1%). Similar trends were observed in the meta-analysis of disease control rate, with anti-EGFR/BRAF/MEK showing the highest disease control rate (76.9%, 95% CrI 70.4% to 82.3%), followed by anti-EGFR/BRAF/PD-1 (75.8%, 51.7% to 90.1%) and anti-EGFR/BRAF (69.9%, 64.4% to 75.0%).

Pooled data showed that the incidence of grade ≥3 adverse events was 39.9% (95% CrI 24.7% to 57.4%) for anti-EGFR/BRAF, compared with 48.6% (24.0% to 73.9%) for anti-EGFR/BRAF/MEK. Chemotherapy-anti-EGFR and anti-BRAF/MEK showed a relatively higher risk of grade ≥3 adverse events (54.9%, 46.8% to 62.7% and 52.0%, 36.0% to 67.7%, respectively).