Mutation analysis of epidermal growth factor receptor (EGFR) gene is essential for treatment selection in non-small cell lung cancer (NSCLC). Treatment with EGFR inhibitors is indicated for patients with EGFR activating mutations, whereas T790M mutation of the same gene is associated with resistance to these treatments. Since tumor biopsy/cytology implies invasive procedures and difficulties to obtain enough sample, we have evaluated the utility of EGFR analysis in cell-free DNA (cfDNA) obtained from plasma.
We focused our study in the most prevalent activating mutations of EGFR, L858R and delE746-A750, and in the resistance-related mutation T790M. We looked for these mutations retrospectively in cfDNA from NSCLC patients in whom EGFR mutations were detected in tumor samples by ARMS technology. We used droplet digital PCR (ddPCR), a sensitive technique recently developed, as one problem with tumor cfDNA is its low relative abundance in blood. With this technique, we could detect these mutations even when diluted as much as 0.005% in wild-type DNA.
In the case of advanced stage patients, the highest concordance between cytology and cfDNA results was obtained for L858R mutation (87%). Interestingly, we could detect in cfDNA, mutations and double mutations not previously detected in cytological samples, in 34% and 17% of patients respectively. This could be explained by the higher analytical sensitivity of ddPCR when compared with ARMS technology, but also to the fact that cfDNA could be a more representative sample for molecular analysis due to the tumor heterogeneity, a known characteristic of NSCLC and other tumors. Certainly, heterogeneity could present mutations expressed by some tumoral cell clones that might be absent in the biopsy. Also, not all lesions are accessible for biopsy. On the contrary, mutational analysis in plasma cfDNA could be more informative of the existence of different clones present in tumor metastases. In fact, T790M mutation was detected in 13% of cfDNA samples but in none of the corresponding cytological samples. This agrees with other recent studies suggesting that this mutation could be much more prevalent in tumors not treated than previously thought. On the contrary, no EGFR mutation was detected in cfDNA from early stage patients.
Molecular analysis of EGFR in baseline cfDNA was also useful in the prognosis. High concentrations of mutated copies in plasma of advanced NSCLC patients were associated with reduced survival. Interestingly, we observed for the first time that high concentrations of total EGFR copies (mutated plus wild-type copies) were also associated with reduced survival. Even more, greater total EGFR copy burden increased with tumor stage, suggesting an EGFR amplification in tumor that was reflected in cfDNA.
Resistance to treatment due to T790M mutation could also be monitored in cfDNA, since this mutation was detected at least once during the treatment follow-up in 43% of patients.
In summary, EGFR analysis in cfDNA is useful for NSCLC patients, overcoming some of the problems associated with tumor heterogeneity and the invasiveness of procedures to obtain tumor cytologic studies, and allowing frequent re-analysis during the follow-up to evaluate the appearance of T790M–related resistance. Besides its utility to detect EGFR mutations related with treatment election, the quantification of both total and mutated EGFR copies presents prognostic value.