211P - Somatic KRAS mutations and resistance to EGFR-targeted therapies: is KRAS ready to include as a reflex test with EGFR in NSCLC? An evidence synthesi...

Date 30 September 2012
Event ESMO Congress 2012
Session Poster presentation II
Topics Non-Small-Cell Lung Cancer, Metastatic
Translational Research
Presenter Samuel Murray
Authors S. Murray1, F. Siannis2, D. Bafaloukos3, P. Kosmidis4, H. Linardou5
  • 1Oncology, GeneKor SA, 15344 - Athens/GR
  • 2Department Of Mathematics, University of Athens, Athens/GR
  • 331st Department Of Medical Oncology, Metropolitan Hospital, Athens/GR
  • 4Data Office, Hellenic Cooperative Oncology Group (HECOG), Athens/GR
  • 5Medical Oncology, Metropolitan Hospital, GR-185 47 - Athens/GR



The intrinsic nature of the constitutive KRAS mutational activation in predicting an absence of response to anti-EGFR tyrosine kinase inhibitors (TKIs: gefitinib and erlotinib) in non-small-cell lung cancer (NSCLC) has recently been demonstrated. We questioned the validity of the inclusion of KRAS in reflex stratification of NSCLC patients.

Material and methods

We searched PubMed and MEDLINE for publications to 31st August 2011 (last search 17/09/2011) pertaining to KRAS mutational status in patients with NSCLC receiving TKIs. Eligible studies reported complete (CR) and partial responses (PR), stratified by KRAS. Potential between-study heterogeneity was accommodated by use of random-effects models for bivariate meta-analysis of sensitivity and specificity. Positive and negative likelihood ratios (+LR and –LR) of KRAS and EGFR for predicting an absence (or presence) of response were calculated by use of pooled estimates for sensitivity and specificity.


For non-response, 26 studies (n = 1957; KRAS = 335, 17.1%) demonstrated a +LR = 40.65, and -LR = 0.82 (specificity = 0.99 [95% CI: 0.95-1.00]; sensitivity = 0.19 [0.15-0.23]). For EGFR, 70 studies (n = 5346; EGFR = 1919, 35.9%) demonstrated a +LR = 5.6 and –LR = 0.25 (specificity = 0.86 [0.82-0.89]; sensitivity = 0.78 [0.74-0.82]) for response. In a three way analysis the apparent difference in the likelihood of a patient with an EGFR mutation responding compared to KRAS is summarized in LR of +422.46 [95%CI: 56.98-3132.43, p < 0.001]; and LR +32.91 fold compared to WT patients [18.51-55.01, p < 0.001]. The likelihood of a ‘no-detectable’ mutation (WT) patient responding compared to a KRAS is LR +13.24 [1.76-99.65, p = 0.012].


Although several studies examined KRAS and EGFR, insufficient data addresses the effect of KRAS on outcomes. The three-way stratification suggests significant differences in survival outcomes due to the high specificity of non-response to KRAS, and response to EGFR. It appears that reflex stratification incorporating KRAS with EGFR treatment decisions in NSCLC is ready for clinical implementation, prospective data on survival outcomes are still however welcome.


All authors have declared no conflicts of interest.