Sir,

We read with interest the meta-analysis by Rowland et al addressing the role of BRAF V600E mutation as predictor of benefit from anti-EGFR monoclonal antibodies (mAbs) in metastatic colorectal cancer.

Authors conclude that there is insufficient evidence to definitively state that RAS WT/BRAF MT individuals attain a reduced benefit from anti-EGFR mAbs compared with RAS WT/BRAF WT ones. Their conclusion is based on the lack of a significant interaction between BRAF mutational status and the effect of the addition of an anti-EGFR mAb to standard therapies (Rowland et al, 2015).

In our opinion some considerations are needed to properly put these results in the clinical perspective, as pointed out in our previous work (Pietrantonio et al, 2015).

First, it should be noted that in terms of PFS, where the confounding effect of subsequent lines of treatment is absent, the P-value for interaction is equal to 0.07. Of note, an alfa-error up to 0.10 is often considered reasonable for interaction tests. In any case, it should be considered that these analyses are based on the retrospective, unplanned evaluation of subgroups of patients included in randomized trials and are therefore definitely underpowered to evidence a statistically significant difference. Although the global number of patients included in the analysis is high, the low incidence of BRAF V600E mutation weakens the power of this analysis. In the meta-analysis, OS comparison included 3096 patients (89% BRAF wild-type and 11% BRAF mutated). Even if 100% of events had been observed – that is a clear overestimation, especially with respect to OS data – the statistical power to detect a significant interaction between BRAF mutational status and the effect of anti-EGFR mAbs (assuming hazard ratio 0.8 in BRAF wt and hazard ratio 1.0, that is, absence of effect, in BRAF mutant patients) would have been as low as about 50%. Therefore, even if the lack of statistical significance of the interaction test for OS is a matter of fact, the relevant risk of a false negative result should be properly acknowledged.

Second, results from FIRE-3 trial, comparing first-line FOLFIRI plus cetuximab with FOLFIRI plus bevacizumab were not included in the metanalysis by Rowland et al. In their discussion, authors elegantly argue that FIRE-3 is not sufficiently comparable to the other included trials, as bevacizumab use in the control arm is associated with a significant benefit, as compared with chemotherapy alone. We totally agree with that observation, but, again, by a practical perspective it should be recognised that first-line chemotherapy plus bevacizumab is one of the most common choices worldwide. From a clinical point of view, the decision of adding an anti-EGFR mAb to chemotherapy in patients with BRAF mutation, based on the absence of interaction between BRAF status and treatment efficacy, would be totally reasonable in the absence of therapeutic alternatives. Given that an alternative is actually available, the use of an anti-EGFR mAb, instead of bevacizumab, should be probably reserved to those patients who may actually derive benefit from these drugs, with a different and often less acceptable toxicity profile. To this purpose, the metanalysis by the same authors highlighting the role of panRAS mutations as predictors of resistance to anti-EGFR mAbs, also including results from FIRE-3, is of special interest (Pietrantonio et al, 2015; Sorich et al, 2015). Unfortunately, results in the RAS WT/BRAF WT subgroup of the FIRE-3 trial have not been provided yet, thus preventing from including this trial in the present analysis. As information about BRAF mutational status is also lacking from the other head-to-head randomized trials PEAK and CALGB80409, we recognise that the question about the ‘best’ biologic agent to be combined with a first-line chemotherapy doublet in BRAF mutant individuals is far from being answered.

Third, as BRAF mutant patients are often unable to receive subsequent lines of therapy (Seligmann et al, 2015), the choice of the upfront treatment is of paramount importance. Although results with doublets plus a biologic are disappointing (Stintzing et al, 2014), increasing evidences support the choice of FOLFOXIRI plus bevacizumab as a preferred option for fit patients (Fakih, 2015; Loupakis et al, 2014).

Nevertheless, more targeted approaches will hopefully enter the clinical scenario in the next future, based on promising results of early phase trials investigating BRAF ±MEK and EGFR inhibitors in molecularly selected patients (Atreya et al, 2015). Knowing BRAF status is today crucial to allow BRAF MT patients to enter clinical trials with those targeted agents.

In conclusion, although the negative predictive power of BRAF V600E mutation with respect to anti-EGFR mAbs will never be formally demonstrated in properly designed, wide and expensive clinical trials, BRAF testing is today recommended by major guidelines. In our opinion, irrespectively of the personal choice of treating physicians to expose BRAF mutant patients to anti-EGFR mAbs, BRAF clearly stands as a molecular marker able to inform clinical decisions in the daily practice, and hopefully its role in treatment decisions will be better defined in the near future.