We read with interest the article by Lum and colleagues, titled “Impact of BRAF Gene Mutation in Nonmetastatic Colorectal Cancer on Disease Progression and Survival Outcomes,” published recently in the Journal of Gastroenterology and Hepatology (2026;41:667–677) 1. In this well-designed retrospective analysis of prospectively collected data, the authors address an important gap in early-stage colorectal cancer (CRC) research—clarifying the prognostic role of the BRAF V600E mutation in Stage II and III disease, with consideration of mismatch repair (MMR) status. Their finding that BRAF mutation independently predicts overall survival (OS) in nonmetastatic CRC offers meaningful insight that could help refine risk stratification and adjuvant therapy decisions. We would like to offer a few analytical suggestions that we believe could further strengthen the robustness and clinical applicability of this already valuable work. First, we would encourage the authors to consider incorporating cancer-specific survival (CSS) as an additional endpoint 2. The current analysis uses all-cause OS as the primary outcome, but Table 2 reveals substantial variation in causes of death across subgroups—most notably, 71.4% of deaths in the BRAFmut/dMMR group were not CRC-related, compared with only 2.6% in the BRAFwt/pMMR group. Given that the BRAFmut/dMMR subgroup also had a higher mean age and greater comorbidity burden, all-cause OS may not fully capture the tumor-specific impact of BRAF mutation. A CSS analysis would help isolate the mutation's effect on CRC progression and mortality, more directly addressing the study's central question. Second, we suggest a quantitative E-value analysis to evaluate the potential influence of unmeasured confounding 3. While the multivariable model adjusts for several important covariates, the dataset does not include other established prognostic factors such as RAS mutation status, tumor budding, perineural invasion, or perioperative factors like blood transfusion and complications. E-value analysis would help determine how strongly an unmeasured confounder would need to be associated with both BRAF status and survival to negate the observed findings. This would offer a clearer sense of the causal robustness of the conclusions. Third, we recommend stage-stratified multivariable analyses to better define the prognostic value of BRAF mutation in Stage II versus Stage III disease 4. The Kaplan–Meier curves suggest differences in OS and RFS across subgroups depending on stage—OS differences are significant only in stage III patients (p = 0.027), while RFS differences are significant only in stage II patients (p = 0.001). Conducting separate multivariable models by stage would provide more precise evidence to inform stage-specific risk stratification, which is particularly relevant given the authors' suggestion to incorporate BRAF mutation into high-risk criteria for stage II CRC. Finally, we believe it would be helpful to assess whether follow-up protocols were consistent across the cohort 5. The study spans 12 years and includes patients from two tertiary institutions, but it does not detail whether postoperative surveillance—such as imaging intervals or biomarker testing—was standardized. Variations in follow-up intensity could influence the timing of recurrence detection and recorded survival outcomes, potentially introducing surveillance bias. Adjusting for these factors or conducting sensitivity analyses among patients with uniform follow-up would help confirm that the findings are not driven by such heterogeneity. In summary, Lum and colleagues have made a meaningful contribution to the literature on BRAF mutation in early-stage CRC. The additional analyses proposed here could further enhance the methodological transparency and clinical relevance of their findings. We thank the authors for their thoughtful work and look forward to seeing how their research continues to evolve. The authors declare no conflicts of interest. The data that support the findings of this study are openly available in (repository name) at (DOI), reference number (reference number).
Huang et al. (Mon,) studied this question.