British Journal of Cancer (2009) 101, 374–375. doi:10.1038/sj.bjc.6605131 www.bjcancer.comPublished online 7 July 2009& 2009 Cancer Research UKSirWe read with great interest the paper by Molinari et al (2009),recently published in the British Journal of Cancer. The studycompares findings in primary colorectal cancer and pairedmetastases in 38 patients, showing differences with respect toEGFR pathway deregulation, which may imply a different responseto anti-EGFR monoclonal antibodies. As a consequence, theauthors suggest that the analysis of metastatic lesions should beconsidered both for patient management and when planningclinical trials with anti-EGFR drugs.Although this statement can be considered as a hypothesistenable in principle, it seems not to be supported by the resultsprovided by the study itself, for the following reasons:(1) Statistical analysis is lacking in terms of study questiondefinition and sample size calculation and therefore all theconsiderations made are not supported by a formal hypothesisverification;(2) Comparisons are made in terms of concordance, using thekappa index, instead of exploiting discrepancies between thetwo paired sets of data. However, even looking at concordanceindexes, kappa values were equal to 0.83, 1, 0.73, 1, 0.49 forKRAS mutational status, BRAF mutational status, PTENprotein expression, EGFR protein expression and EGFR genestatus, respectively, showing that only for EGFR gene statusthe concordance did not reach a ‘good’ value. As no intra-sample concordance results are available, the questionwhether these differences are because of a true discordancebetween primary tumour and metastases or rather due to anintra-sample variability (i.e., variability (a) between differentareas within the same sample because of biological hetero-geneity (Al-Mulla et al, 1998) or (b) to technical reproduci-bility) remains unanswered.(3) Looking at Table 2,(3.1) for KRAS the data show two patients mutated in primarytumour whose result became wild type in distantmetastatic sites, whereas one patient wild type inprimary tumour became mutated;(3.2) only for three patients with PTEN-negative metastasesbut positive primary tumours, the finding of non-response is consistent with biological hypothesis andother retrospective proof of principles (Loupakis et al,2009);(3.3) for BRAF no modification was seen;(3.4) full concordance was detected for EGFR IHC analysis,whereas for EGFR gene status the problems related tothe reproducibility are well known.Even looking at these data in a ‘qualitative’ way, the overallpicture does not seem to suggest a clear trend towards a negativepredictive effect of metastases, and only for PTEN there is someevidence towards a negative effect.In the clinical practice, the biopsy of the metastatic lesion couldbe an invasive procedure, not free from risks and causes delays intreatment start, other than understandable anxiety for the patient.Therefore, costs and advantages should be well balanced. More-over, it should be considered (1) that the only universally accepteddeterminant of resistance to anti-EGFR MoAbs are KRASmutations, (2) that post hoc analyses of randomised studies, whichled to the regulatory restriction for anti-EGFR MoAbs to KRASwild-type patients were conducted almost exclusively on primarytumours (Amado et al, 2008; Karapetis et al, 2008) and (3) thatother series greater than the present found a degree of correlationbetween primary tumours and related metastases in terms of KRASmutational status that does not justify, at present, the clinical needfor biopsies of metastases (Artale et al, 2008; Santini et al, 2008;Loupakis et al, 2009). Given the retrospective nature of the study,which exposes the results to selection and verification biases, andgiven the small sample size, we do think that no definitive clinicalimplication can be driven from this study. Rather, it is importantto stress the need for prospective, properly powered studies, aimedto evaluate the importance of tumoral sampling at time oftreatment’s start for the molecular prediction of benefit fromanti-EGFR MoAbs, but, at the same time, there are no data forsupporting the need for biopsies of metastases in the routinarypractice.