ARID1A variations in EGFR mutated lung adenocarcinomas and its relation with resistance to targeted therapy

Abstract

Lung cancer causes the highest cancer mortality worldwide and amongst lung cancers, adenocarcinoma (AD) is the commonest histological type. EGFR mutation occurs in around 45% cases of lung AD which indicates responsiveness to targeted therapy by tyrosine kinase inhibitor (TKI) but drug resistance is almost inevitable. ARID1A which encodes a protein subunit in the SWI/SNF chromatin remodeling complex has been proposed to be a tumor suppressor gene in various cancers. In a preliminary study of whole exome sequencing of EGFR-mutated lung AD treated by first generation TKI, ARID1A mutations were suggested to associate with the lack of TKI response, implicating a potential novel mechanism of primary TKI resistance. To further explore this possibility, additional cases were analyzed for detection of ARID1A mutations and aberrant expressions. Thirty cases were analyzed by direct sequencing including 2 duplications from the preliminary cohort. Results showed totally 4 cases of ARID1A mutations comprising 2 missense (p.G1265S and p.V1982I) and 2 nonsense (p.S1791* and p.E2054*) mutations, yielding a mutation frequency of 13.3%. Correlation studies suggested a possible relation with the unfavourable TKI responses of static disease (SD) or progressive disease (PD), but statistical significance was not reached either in the current (p = 0.290) or combined (p = 0.158) cohort. Further comparison with the duration of TKI treatment, and the progression free survival (PFS) from the start of TKI treatment, respectively, showed no significant correlation with ARID1A mutations. To gain a further understanding, ARID1A expression was analyzed by immunohistochemistry (IHC) in totally 90 eligible lung AD including the new cohort. No significant correlation of ARID1A expression levels with the TKI response pattern, PFS or TKI treatment duration were observed. Notably, no significant loss of ARID1A expression was detected in the cases bearing ARID1A missense or nonsense mutations. Although a similar lack of inverse correlation between mutations and protein expression of ARID1A has been reported in other cancers, further investigation using other approaches or methods of expression analysis is warranted before definitive conclusions could be drawn. Overall, this study attempted to answer a novel question to which no data are available in the literature for comparison. While a possible relation between ARID1A mutations and unfavourable TKI response patterns was observed in the extended cohort (p = 0.158), no trend of association with other parameters of TKI response was detected. Nevertheless, further enlargement of the study cohort might be worthwhile for more definitive validation. If verified, ARID1A mutations, rather than protein expression, might be a useful biomarker for the prediction of TKI responses. A more thorough understanding of the biochemistry, regulation of expression and action of ARID1A in human cancers would also help to clarify the clinical observations.