T790M mutation is the most frequently reported mechanism of acquired resistance and accounts for approximately 50–60 % of cases [43, 48, 52, 68]. As mentioned previously, apart from altering EGFR TKI binding to the EGFR, T790M also increases ATP binding affinity leading to EGFR TKI resistance [33]. In most patients, it is uncertain whether T790M exist de novo or whether it is acquired through clonal selection induced during EGFR TKI therapy. Other reported point mutations in EGFR associated with acquired resistance to EGFR TKI include D761Y and L747S on exon 19 and T854A on exon 21, but together, they account for less the 5 % of acquired resistance mechanisms [39–41].

MET amplification is the second commonest acquired EGFR TKI resistance mechanism and is found in about 5–20 % of patients with EGFR acquired resistance [17, 42, 43]. PI3CA mutations were detected in 5 % of patients with acquired resistance [34, 42]. There are many other less common alterations that bypass the signaling of the EGFR reported, for example, HER2 amplification, v-raf murine sarcoma viral oncogene homolog B (BRAF) mutation, ERK amplification, and increased AXL expression [34, 43, 47]. More potential alterations are to be identified, but they are likely to be less clinically significant in view of their infrequencies.

Small cell transformation and epithelial to mesenchymal transition (EMT) are phenotypical changes that have been found to be associated with EGFR TKI-acquired resistance [42, 43, 48].

The management of patients who developed progression of disease while on EGFR TKI should be highly individualized. Premature termination of EGFR TKI can lead to rapid deterioration of the disease and this “disease flare” phenomenon is well recognized [49]. The role of EGFR TKI beyond progression is being investigated in large randomized study [IMPRESS NCT01544179, ASPIRATION NCT01310036]. Anecdotally, it is reasonable to continue with the EGFR TKI for those with oligo-progression (i.e., four or fewer sites of progression) or asymptomatic disease with slow disease tempo [50]. For those sites of oligo-progression, the addition of local therapy such as radiotherapy has shown improvement in survival in retrospective studies [51–53]. It is thought that most resistant tumors are composed of mixed population of EGFR TKI-sensitive and EGFR TKI-resistant cancer cells (tumor heterogeneity) [34]. Continuation of EGFR TKI in combination with chemotherapy after the acquisition of resistance ideally would avoid the disease flare while the resistant tumor can be dealt with by chemotherapy [56, 57]. An ongoing phase 3 study is investigating this approach. For patients with rapid systemic progression on EGFR TKI, switching to chemotherapy is usually required, especially for those with documented histologic transformation.

As mentioned before, T790M mutation is the commonest acquired mutation. Irreversible EGFR TKIs like the second-generation afatinib and dacomitinib were developed with improved properties to intensify EGFR inhibition and have potential activity against T790M mutation in vitro [17, 54]. Apart from forming covalent bonds with ATP binding site leading to permanent inhibition of EGFR, afatinib also inhibits human epidermal growth factor receptor 2 (HER2) while dacomitinib is a pan-HER inhibitor. Whether second generation can overcome acquired resistance caused by T790M is yet to be confirmed clinically [58]. A single arm study showed promising response rate (>30 %) when afatinib was combined with cetuximab for enhanced intra- and extracellular blockage of EGFR in patients with acquired resistance to EGFR TKI [37, 59]. Recent early clinical phase I studies with third-generation EGFR TKIs including CO 1686, AZD 9291, and HM61713 which covalently and selectively bind the sensitizing as well as T790M resistance mutations, sparing the EGFR wild-type receptor, have been very promising, in particular in the subgroup of patients with a confirmed T790M mutation where response rates of up to 28–64 % have been reported [60–62].

Many novel compounds targeting different genetic oncogenic mutations are in the development pipeline. Depending on the alternative pathways being activated, switching to and combining EGFR TKI with other targeted kinase inhibitor (e.g., MET inhibitors, PI3KCA inhibitors) are potential strategies that may overcome resistance [63].

The role of adjuvant TKI therapy in patients with EGFR mutations has yet to be defined. In the RADIANT trial, 973 stage IB–IIIA patients were randomized into 2 years of erlotinib versus placebo after surgery and standard adjuvant chemotherapy (if indicated). All patients were EGFR positive either by immunohistochemistry (IHC) or fluorescence in situ hybridization (FISH). The subgroup of 161 EGFR-mutated patients had a disease-free survival (DFS) improvement of 17.9 months with adjuvant erlotinib (HR 0.61, 28.5 vs. 46.4 months, HR 0.61) [64, 65]. However, these findings were not statistically significant due to hierarchical testing with the primary end point, i.e., DFS of the whole study population, being negative. In addition, preliminary OS was not different in the subgroup of patients with EGFR mutation (HR 1.09).

A single arm trial in the USA enrolled 100 patients to receive 2 years of adjuvant erlotinib after standard chemotherapy also demonstrated a favorable DFS of 76 % at 2 years and even 91 % in the 28 patients with stage IIIA disease [66]. The vast majority of relapses in the erlotinib arm occurred after completion of therapy; however, it is currently unclear, whether a longer duration of treatment improves outcomes. These results are in line with previous observations [37]. On the other hand, a very small exploratory subgroup analysis of 15 EGFR-mutated patients of a randomized trial which enrolled 503 unselected patients assessing the role of 2 years of adjuvant gefitinib versus placebo did not suggest any benefit of adjuvant TKI therapy (HR for DFS 1.84, p 0.395; HR for OS 3.16, p = 0.15) [67].

Taken together, these results suggest a prolongation of DFS with the use of erlotinib by delaying recurrence with uncertain effect on OS. In addition, the significantly higher rate of brain metastases (40 % vs. 12.9 %) observed in the erlotinib arm indicates that the disease course potentially could be altered in an unfavorable manner [65]. Further studies are warranted before adjuvant EGFR TKI therapy is recommended.