Sir,

It is with great interest that we read the letter entitled ‘KRAS and BRAF mutations are prognostic biomarkers in patients undergoing lung metastasectomy of colo-rectal cancer. Variation in survival associated with proto-oncogenes is not evidence for effectiveness of metastasectomy’.

In their comment on the prognostic value of proto-oncogenes in lung metastasectomy of colo-rectal cancer (CRC) (Renaud et al, 2015), Cardillo and colleagues stated that ‘no difference in survival attributable to surgical removal of lung metastases has been shown in a control trial’ and reached the conclusion on ‘the doubt on effectiveness of metastasectomy in colorectal cancer’ (Cardillo et al, 2015). However, as stated by the authors, no randomised controlled trial has shown a survival benefit for follow-up compared with surgery. The very large majority of the published series shows that in metastatic CRC, medical treatment alone leads to poor overall survival (OS) (Rooney et al, 2015), while surgery leads to 5-y OS up to 70% (Riquet et al, 2010; Hawkes et al, 2012; Renaud et al, 2014).

At this point a word of caution is essential and we would like to emphasise several points that we consider to be important.

First, we agree that patients who underwent surgery were highly selected. However, the proper selection of candidates for surgery is the basis of surgical oncology. As an example, stereotactic ablative radiotherapy and radiofrequency are offered as alternatives to surgery in resectable NSCLC unfit for surgery, since poor medical condition is associated with worse outcomes (Boily et al, 2015; Dupuy et al, 2015). Consequently, the doubt of hyperselection can be applied to all the fields of surgical oncology.

Second, the authors relevantly stated that ‘the study shows the influence of oncogenes on survival but these are likely to be general prognostic factors’. They are right! Our aim was to identify new prognostic factors. Indeed, despite known prognostic factors, there are wide variations of OS among patients after metastasectomy of CRC. It seems that, from the primary CRC tumour, the molecular status could predict the course and the aggressivity of CRC (Renaud et al, 2015). Consequently, molecular markers might help to a better selection of patients, from the primary CRC surgery. In addition, we further think that molecular analysis could be even more helpful. Indeed, it seems that KRAS specific amino-acid substitutions is associated with different activations of downstream effectors, which can induce different behaviours (Garassino et al, 2011; Ihle et al, 2012; Izar et al, 2014; Nadal et al, 2014). In particular, in CRC cell lines, KRAS G12V is known to overexpress CXCR4, implied in metastasis process, promoting higher aggressivity (Alamo et al, 2015). These interesting preliminary results may lead in the future to a molecular prognostic classification, which may help to select the best patients for surgery.

Third, in the meta-analysis of Gonzalez et al (2013), CEA remained significant in multivariate analysis in only 9 out of 19 selected studies. Furthermore, CEA only reflects the total tumour mass. On the other hand, the absence of impact of disease free survival in our work probably reflects the proper selection of patients.

Fourth, meta-analyses cited by the authors, such as the latest one, does not add supplementary reflexion to our discussion (Gonzalez et al, 2013). Indeed, it only identifies four risk factors of poor outcome, which have been included in our analysis.

Fifth, CEA second look trial recruited patients who were randomised between surgery and follow-up according only to CEA elevation (Treasure et al, 2014). Obviously, this study cannot be generalised to lung metastasectomy of CRC: (1) patients with extra-abdominal recurrence were excluded, and different regimen of peri-operative treatment were used; (2) the use of CEA alone may have probably led to the inclusion of patients without recurrence, for whom surgery was futile. Indeed, patients were included in case of CEA >10 ng ml−1. However, when <30 ng ml−1, many conditions can induce CEA increase, in particular smoking and alcohol. Excluding these conditions by simple questioning exposed to information biases. Otherwise, Primrose et al (2014) did not reach the conclusion that treatment of recurrence was unnecessary, but that intensive follow-up after surgery, even if detecting earlier recurrence, does not significantly diminish the mortality in comparison with minimum follow-up (Primrose et al, 2014). Furthermore, they clearly stated that ‘detection of recurrence that was treatable surgically with curative intent was chosen as the main outcome measure’ (Primrose et al, 2014), and ‘that the statistical power of their trial to assess a mortality advantage of intensive follow-up was limited’ (Mant and Primrose, 2014). Finally, in their work, recurrence was treated by various regimen forbidding to conclude on the benefit of surgical treatment of recurrence.

Consequently, so far, we strongly advocate that no data allows concluding that lung metastasectomy is a futile procedure in CRC.