Coexistence of KRAS mutation with mutant but not wild-type EGFR predicts response to tyrosine-kinase inhibitors in human lung cancer.

Sir,

EGFR and KRAS mutations occur mutually exclusively in NSCLC, suggesting functional redundancy (Kosaka ; Pao ; Shigematsu ; Tam ). However, they predict contrasting response rates to tyrosine-kinase inhibitors (TKIs) – while EGFR mutation predicts longer progression-free survival rate (Lynch ; Mok ; Fukuoka ; Chougule ), adverse prognosis is associated with patients harbouring KRAS mutations (Mao ; Ihle ). The recently reported co-occurrence of KRAS and EGFR activating mutations (Li ) in 30 of 5125 patients raises questions about the relative values of EGFR and KRAS mutation status as predictors of outcome in NSCLC. This has obvious implications for routine KRAS testing in this disease, potentially precluding EGFR TKI therapy from some patients, similar to the current practice in colorectal cancer (Lievre ).

EGFR mutations occur less frequently among Caucasians (10–15%) compared to East Asians (30–60%) (Lynch ; Paez ) in contrast to KRAS mutations for which the situation is opposite – Caucasians vs East Asians, 25–50% vs 5–15%, respectively (Mao ; Roberts ). We recently reported an intermediate frequency of EGFR mutations (23%) in a study involving 907 Indian NSCLC patients (Chougule ). In a smaller study at our centre there was 74% response to TKI among patients with tumours having EGFR mutations compared to 5% in those with wild-type EGFR (Noronha ). We performed directed sequencing of KRAS exons 2 and 3 in 86 patients from the same cohort and correlated its status with outcome after treatment with EGFR TKI (see Supplementary Table S1). There were 15 patients with KRAS G12C and 1 patient with KRAS G12V mutation for an overall mutation rate of 18.6%. Three of these 86 patients had coincident KRAS G12C and EGFR Exon-19 Del (E746-A750) mutations (Table 1), which were independently validated in each of the three samples by four orthogonal technologies (Sequenom Mass Array genotyping, Taqman Real Time PCR, Sanger Sequencing and SNaPShot PCR; see Supplementary Figure 1). All these three patients had partial response to EGFR TKI. However, only 1 of the remaining 13 patients with KRAS mutation had partial response and his tumour had wild-type EGFR with unknown copy number. This is consistent with previous studies: response to EGFR TKI in one of five patients harbouring KRAS mutation with unknown EGFR copy number or mutation status (Zhu ); 1 response in 20 KRAS mutant patients who also harboured EGFR amplification but not EGFR mutation (Gumerlock ); and response to gefitinib in three of five patients with coincident EGFR and KRAS mutations (Benesova ). Furthermore, it has been shown that when KRAS and EGFR mutations are coincident in the same tumour, the genetic lesion in the latter is almost exclusively in exon-19, with virtually no occurrence of exon-21 abnormalities (Li ). This was also true in all three cases in our series with coincident EGFR and KRAS mutations. Of note, mutations in exon-19 have been shown to predict a higher response to EGFR TKI than those in exon-21 (Mitsudomi ).

Table 1

EGFR and KRAS mutation co-occurrence in Indian NSCLC patients

Sample ID	EGFR mutation	KRAS mutation	Histology type	Gender	Smoking history	Response to gefitinib
LB-21	Exon-19 Del	G12C	Adenocarcinoma	Female	No	PR
LB-69	Exon-19 Del	G12C	Adenocarcinoma	Female	No	PR
LB-117	Exon-19 Del	G12C	Adenocarcinoma	Female	No	PR

Abbreviation: PR=partial response.

In summary, these data suggest that NSCLC patients with KRAS mutations are unlikely to respond to EGFR TKI therapy in the absence of coincident EGFR alterations. Therefore additional KRAS molecular testing may not add predictive value in selecting patients for EGFR TKI therapy and cannot be routinely recommended.