Uncovering tyrosine kinase inhibitor resistance mutations in EGFR mutant lung adenocarcinoma using whole-exome sequencing

Abstract

Lung cancer mortality is the highest of all cancer types worldwide. Epidermal growth factor receptor (EGFR) is a commonly mutated gene in lung adenocarcinomas (LUAD), especially of Asian non-smoking females. Tyrosine kinase inhibitors targeting EGFR (EGFR-TKI) are used to treat metastatic LUAD with EGFR sensitizing mutations, but outcomes are not uniform and acquired drug resistance develops after long term treatment. Though aberrations conferring EGFR-TKI resistance such as EGFR T790M and MET amplification have been identified, the mechanisms in around 30% of cases remain unknown. Mutation screening of post-treatment samples should reveal potential resistant targets but such biopsies are often unavailable or small. More efficient approaches to decipher complex genomic changes are need.

This study was performed to identify candidate resistant mutations in resected, pre-treatment LUAD using whole-exome sequencing (WES). Data from 41 patients were analyzed, including 18 with activating EGFR substitutions and 23 with exon 19 deletions. There were 22 responders who showed complete radiological resolution or clinically significant partial tumour shrinkage at 3 months after the start of TKI; and 19 resistant patients who showed lack of clinical response or had continually enlarging or progressing tumours. Single nucleotide variations (SNV), insertions and deletions (INDEL), and copy number variations (CNV) were compared. Mutations were deemed to be resistance candidates when they were present in clinically resistant cancers only. They were categorized into those known to be clinically actionable according to published databases or literature, and those with inferred or potential importance based on signaling pathway analysis. Results showed there were significantly more coding region mutations of clinically actionable genes of the PI3K/AKT-WNT/ß-catenin signaling network in the resistant compared to the responsive group (Fishers’ exact test, p=0.004). Mutations of PIK3CA, PTEN and NF1 were only detected in resistant patients. Besides, recurrent amplifications of CCND1 and SMARCB1 were detected only in EGFR-TKI-resistant tumours. Gene set enrichment analysis showed genes mutated only in the resistant group were more significantly enriched for EGFR-related pathways.

Formalin-fixed paraffin-embedded (FFPE) tissues are the most convenient clinical samples for sequencing studies. However, results are affected by DNA artefacts induced by formalin fixation. To find solutions to tackle the problem of using FFPE for WES studies, comparisons were made between 3 types of tumour-normal tissue pairs, namely, FFPE-FFPE, FFPE-frozen/peripheral blood (PB) and frozen-frozen/PB tissues. Results showed FFPE-FFPE and frozen-frozen/PB pairs had similar SNV abundance which were significantly less than the FFPE-frozen/PB combination, attributed to the randomness of formalin-induced single base artefacts in paired tumour and normal tissues of the same case being eliminated during data processing. However, similar offset effects were not observed for INDEL mutations. Nevertheless, most formalin-induced mutations were not located in coding regions and negative effects on identifying key resistant mutations were minimal.

In summary, this study showed mutated resistant pathways concurrent with EGFR somatic mutations could be uncovered by WES in excised lung cancers prior to TKI treatment. The identified candidate resistant mutations should be tested by target deep sequencing in an extended cohort for verification and application to treatment planning. (496 words)