Classic EGFR mutations

The discovery of mutations in the epidermal growth factor receptor (EGFR) gene as oncogenic driver in lung cancer patients has changed both the diagnostic process and treatment of such patients. The EGFR mutations are detected in ∼10% of Caucasian patients with non-squamous non-small-cell lung cancer (NSCLC) and in up to 50% of Asian NSCLC patients (Dearden et al, 2013). In addition to the higher prevalence in people from Asian descent, there is a higher prevalence of EGFR mutations in women, nonsmokers and adenocarcinoma patients (Barlesi et al, 2016). The vast majority of EGFR mutations comprise microdeletions in exon 19 (45–50%) and the Leu858Arg (L858R) substitution, resulting from a point mutation in exon 21 (40–45%; Murray et al, 2008). These mutations are so-called sensitising EGFR mutations and hereafter referred to as ‘classic EGFR mutations’ (Supplementary Figure 1). The beneficial effect of treatment with EGFR tyrosine kinase inhibitors (TKIs) in NSCLC patients who harbour a classic EGFR mutation in their tumour is well established (Lynch et al, 2004; Maemondo et al, 2010; Mitsudomi et al, 2010; Fukuoka et al, 2011; Zhou et al, 2011; Han et al, 2012; Rosell et al, 2012; Sequist et al, 2013; Wu et al, 2014). However, resistance is inevitable and median progression-free survival (PFS) on EGFR-TKI treatment for NSCLC patients with a classic EGFR mutation is 8.0–13.1 months (Mok et al, 2009; Maemondo et al, 2010; Mitsudomi et al, 2010; Fukuoka et al, 2011; Zhou et al, 2011; Han et al, 2012; Rosell et al, 2012; Sequist et al, 2013; Wu et al, 2014, 2015).

The T790M mutation

The T790M mutation is a distinct EGFR mutation that is located in exon 20. It interferes with binding of EGFR-TKI to EGFR, thereby prohibiting the inhibitory effect of these agents. Detection of the T790M mutation before EGFR-TKI treatment is rare (0.5%; Yu et al, 2014), although the detection rate of pretreatment T790M is higher with more sensitive detection methods (Rosell et al, 2011). However, the T790M mutation is detected in ∼60% of EGFR-mutated NSCLC patients on or post treatment with an EGFR-TKI showing renewed tumour growth (Yu et al, 2013).

Non-classic EGFR mutations

The EGFR mutations other than the classic EGFR mutations and exon 20 T790M mutations are less prevalent (hereafter referred to as ‘non-classic EGFR-mutations’) (Supplementary Figure 1). The most prevalent non-classic EGFR mutations are insertions or duplications in EGFR exon 20 (further referred to as ‘EGFR exon 20 insertions’) that are detected in ∼2.2–5.0% of NSCLC patients (Wu et al, 2008a; Arcila et al, 2013; Oxnard et al, 2013; Beau-Faller et al, 2014). In the study of Arcila et al (2013), EGFR exon 20 insertions were furthermore mutually exclusive with mutations in other genes, such as KRAS and BRAF, except for PIK3CA and there were no associations with age, sex, race or stage. Patients with EGFR exon 20 insertions generally have a lower response rate to EGFR-TKI treatment and a poorer prognosis compared with NSCLC patients with classic EGFR mutations (Wu et al, 2008a; Oxnard et al, 2013). Other non-classic EGFR mutations include so-called uncommon mutations (Supplementary Figure 1), for example, in EGFR exon 18 (e.g., G719X; X=A, S or C), EGFR exon 20 (e.g., S768I) and EGFR exon 21 (e.g., L861Q). The proportion of uncommon EGFR mutations among EGFR-mutated NSCLC patients might be as high as 14%, but varies in different studies (Yokoyama et al, 2006; Zhang et al, 2007; Wu et al, 2008b, 2011; Hata et al, 2010; De Pas et al, 2011; Arcila et al, 2013; Kobayashi et al, 2013; Keam et al, 2014).

Multiple uncommon EGFR mutations or an uncommon EGFR mutation in combination with a classic EGFR mutation may co-exist in the same tumour. These so-called ‘double’ (or complex, or compound) mutations are reported to occur in 6.6% of EGFR-mutated NSCLC patients (Hata et al, 2010).

Data on results of EGFR-TKI treatment in Caucasian patients with non-classic EGFR mutations are scarce as they are commonly reported in small series, whereas the larger series typically include patients of Asian descent. We therefore evaluated a cohort of Dutch (i.e., predominant Caucasian) EGFR-mutated NSCLC patients retrospectively. The purpose of this study was to evaluate the prevalence and genotype distribution of non-classic EGFR mutations in this cohort, as well as clinical characteristics and outcome on EGFR-TKI treatment.

MATERIALS AND METHODS

Patients

All NSCLC patients in whom an EGFR mutation was detected in the VU University Medical Center (VUmc) between May 2006 and November 2014 (N=240) were retrospectively evaluated. As the VUmc is a diagnostic referral centre, some patients were diagnosed at our centre, but follow-up and treatment were performed in other hospitals. Patients with missing data on follow-up were excluded from analysis of clinical characteristics and outcome on EGFR-TKI treatment. For all other patients, data on demographics, clinical and tumour-related features, treatments and clinical outcomes was extracted from the medical records.

Mutation analysis

All mutation analyses were part of the routine diagnostic procedures in VU University Medical Center, Amsterdam, The Netherlands. The molecular diagnostic modalities for EGFR mutation analysis included Sanger sequencing, HRM sequencing and cancer panel multiplexed targeted resequencing (Janmaat et al, 2006; Heideman et al, 2009; Sie et al, 2014). All assays are designed to identify deletions or insertions in EGFR exons 19 and 20, and hot spot mutations in EGFR exons 18 through 21.

For analytical purposes, deletions in EGFR exon 19 and the L858R point mutation in EGFR exon 21 are referred to as classic EGFR mutations. Among non-classic EGFR mutations, a distinction between exon 20 insertions and ‘uncommon EGFR-mutations’ was made (Supplementary Figure 1). The post-treatment T790M mutations are not included in our analyses, nor are common EGFR polymorphisms. All alterations that were detected were checked in Alamut Visual version 2.7 (Interactive Biosoftware, Rouen, France), the mycancergenome database (www.mycancergenome.org; accessed 1 April 2016) and the Cosmic database (cancer.sanger.ac.uk/cosmic; accessed 23 April 2016).

Treatment and outcomes

Patients who were alive at closing date (26 November 2015) or who were lost to follow-up were censored at the last date of follow-up. The EGFR-TKI treatment included treatment with erlotinib, gefitinib or afatinib in patients with advanced-stage disease. Survival was calculated from date of diagnosis of advanced-stage (stage IIIB or IV) disease until date of death. Objective response rate (ORR) on EGFR-TKI treatment was calculated as the proportion of patients with complete or partial response according to Response Evaluation Criteria in Solid Tumours (RECIST) 1.1 (Eisenhauer et al, 2009). Disease control rate (DCR) on EGFR-TKI treatment was calculated as the proportion of patients with an objective response or stable disease (for at least 6 weeks) according to the RECIST 1.1 criteria (Eisenhauer et al, 2009). Progression-free survival on EGFR-TKI treatment was calculated as the time from first day of treatment until progression of disease or date of death (from any cause). Patients who had not progressed at data cutoff were censored at the last day of follow-up. Overall survival (OS) on EGFR-TKI treatment was either calculated as the time from the first day of EGFR-TKI treatment until date of death (from any cause), or patients were censored at last follow-up.

Statistical analyses

Comparison of categorical variables was performed with Pearson’s χ2 test. Comparison of three or more continuous variables was performed with one-way ANOVA. The Kaplan–Meier method was used for survival analyses and the log rank test was used to test for significance. Two-sided P-values of ⩽0.05 were considered significant and confidence intervals (CIs) were calculated at a 95% CI. The SPSS for Windows (version 20; SPSS Inc., Chicago, IL, USA) was used for statistical analyses. The medical ethical committee of VU University Medical Center (Amsterdam, The Netherlands) approved the protocol.

Results

Classic EGFR mutations

In 186 out of 240 patients (77.5%), a classic EGFR mutation was detected (Figure 1): 134 patients (72.0%) with an exon 19 deletion and 52 patients (28.0%) with an exon 21 L858R point mutation.

Figure 1
figure 1

Flowchart. *No treatment and follow-up in VUmc.

Full size image

Sixty-two patients with a classic EGFR mutation were not treated at our centre and were excluded from further analysis. Clinical characteristics of the remaining 124 EGFR-mutated NSCLC patients are described in Table 1. Median follow-up was 31.6 months (95% CI, 26.1–27.3). The EGFR-TKI treatment was started in 111 patients (89.5%) (Supplementary Table 1). Clinical outcome on EGFR-TKI treatment of this group of patients is described in Table 2. Supplementary Tables provide more detailed data on start and/or progression on EGFR-TKI treatment (Supplementary Table 2), survival after EGFR-TKI treatment (Supplementary Table 3) and response setting (Supplementary Table 4).

Table 1 Patient characteristics
Full size table
Table 2 Median PFS, OS, ORR and DCR on EGFR-TKI treatment in advanced-stage NSCLC patients with classic EGFR mutations
Full size table

Non-classic EGFR mutations

A total of 54 patients (22.5%) harbouring a non-classic EGFR mutation were identified: 23 patients (9.6%) with an exon 20 insertion and 31 patients (12.9%) with an uncommon EGFR mutation in exons 18, 19, 20 and/or 21. In one patient, both an exon 20 insertion and an EGFR exon 20 V769L point mutation were detected. This patient was categorised in the EGFR exon 20 insertion group. All EGFR exon 20 insertions concerned insertions located on regions V769-N771 or H773-V774.

Of the group with uncommon EGFR mutations, 15 patients (6.3%) had a single uncommon EGFR mutation (Table 3) and 16 patients (6.7%) were identified with double uncommon EGFR mutations (Table 4). In three patients (1.3%) with a single uncommon EGFR mutation, a KRAS mutation was also detected (Table 3). In four patients (1.7%) with double EGFR mutations, one of these mutations concerned the classic EGFR mutation L858R on exon 21 (Table 4). There were two patients with a single G719X EGFR mutation and two patients with a single L861Q EGFR mutation (Table 3). Nine patients were identified with a double EGFR mutation that included G719X, L861Q and/or S768I (further referred to as ‘double G719X/L861Q/S768I’ EGFR mutations; Table 4).

Table 3 Patients with single uncommon EGFR mutations
Full size table
Table 4 Patients with double EGFR mutations (and at least one uncommon EGFR mutation)
Full size table

Of the patients with non-classic EGFR-mutations, four were not treated in our centre and excluded from further analysis (i.e., one with an exon 20 insertion, one with L858R+V834L mutation, one with an exon 19 insertion and one with L861Q mutation).

Clinical characteristics of the remaining 22 patients with an EGFR exon 20 insertion and 28 patients with an uncommon EGFR mutation are described in Table 1. Median follow-up of these patients was 29.4 months (95% CI, 19.6–39.3). Baseline demographic characteristics were similar between the three groups, except for smoking (P<0.01).

EGFR-TKI treatment in patients with an EGFR exon 20 insertion

Sixteen patients with advanced-stage disease and an exon 20 insertion received EGFR-TKI treatment. Seven patients (43.8%) received EGFR-TKI as first-line treatment, but most patients received EGFR-TKI treatment as second-, third- or fourth-line treatment (Supplementary Table 5). Median PFS on EGFR-TKI treatment was 2.9 months (95% CI, 2.3–3.6). Median OS on EGFR-TKI treatment was 9.7 months (95% CI, 0.00–21.1). Both PFS and OS on EGFR-TKI treatment were significantly shorter in patients with an EGFR exon 20 insertion compared with patients with a classic EGFR mutation (P<0.01 and P=0.01, respectively; Figure 2A and B). The ORR was 0.0% and DCR was 56.3%.

Figure 2
figure 2

The PFS and OS on EGFR-TKI treatment in patients with a classic EGFR mutations vs EGFR exon 20 insertions or uncommon EGFR mutations. Difference between classic EGFR mutations vs EGFR exon 20 insertions in PFS (A) and OS (B) and between classic EGFR mutations and uncommon EGFR mutations in PFS (C) and OS (D). A full colour version of this figure is available at British Journal Of Cancer online.

Full size image

EGFR-TKI treatment in patients with an uncommon EGFR exon 18, 19, 20 and 21 mutation

Twenty patients with an uncommon EGFR mutation received EGFR-TKI treatment. Sixteen patients (80%) received EGFR-TKI treatment as first-line and four patients (20%) as second-line treatment. For two patients, there was no registered date of progression. Median PFS on EGFR-TKI treatment for the remaining 18 patients (all advanced-stage disease) was 6.4 months (95% CI, 0.0–17.6). This was not significantly different compared with median PFS in patients with a classic EGFR mutation (P=0.39). Median OS on EGFR-TKI treatment in patients with an uncommon EGFR mutation was 20.2 months (95% CI, 0.0–41.7). This was significantly shorter compared with the median OS on EGFR-TKI treatment in patients with a classic EGFR mutation (P=0.04).

For 15 patients with uncommon EGFR mutations, data on response on EGFR-TKI treatment could be retrieved from the medical records: ORR was 53.3% and DCR was 86.7%.

Ten patients with single or double G719X/L861Q/S768I EGFR mutations were treated with an EGFR-TKI. Median PFS on EGFR-TKI treatment for patients with a double G719X/L861Q/S768I EGFR mutation (N=7) was 6.4 months (95% CI, 0.0–17.6), and this was significantly longer (P=0.02) than for patients with single-mutant status at these loci (N=3; 1.6 months (95% CI, 1.5–1.7)). Median OS on EGFR-TKI treatment for patients with a double G719X/L861Q/S768I EGFR mutation was 28.6 months (95% CI, 11.3–45.8), and 3.9 months (95% CI, 0.5–7.4) for those with a single G719X/L861Q/S768I EGFR mutation (P=0.02).

Discussion

Targeted agents are being developed rapidly and their clinical use is increasing in NSCLC patients. Considering the toxicities and costs of these drugs, their usage should be restricted to patients who truly benefit from them. In lung cancer, the efficiency of EGFR-TKIs is well known for classic EGFR mutations, but less data are available for patients with non-classic EGFR mutations. Moreover, most studies were performed in Asian populations. This study, among Dutch EGFR-mutated NSCLC-patients, adds to the current knowledge on non-classic EGFR mutations and the outcome on EGFR-TKI treatment in this subgroup of lung cancer patients.

In our cohort of 240 EGFR-mutated NSCLC patients, 54 patients (22.5%) were identified with a non-classic EGFR-mutation: 23 patients (9.6%) with an EGFR exon 20 insertion and 31 patients (12.9%) with an uncommon EGFR mutation. Previous studies on EGFR exon 20 insertions in predominantly non-Asian EGFR-mutated NSCLC patients reported a rate of 9%, 4.0% and 9.2% (Arcila et al, 2013; Oxnard et al, 2013; Beau-Faller et al, 2014), hence incidence of EGFR exon 20 insertions in our cohort is approximately in line with these studies. The incidence of uncommon EGFR mutations among non-Asian EGFR-mutated NSCLC patients varies between 5.9% and 20.4% (Pallis et al, 2007; De Pas et al, 2011; Beau-Faller et al, 2014; Stone et al, 2014; Arrieta et al, 2015; Lohinai et al, 2015), and this is also in accordance with results from our study. However, comparison to other studies is difficult, as there is a large variance in ethnicity of patients included, detection method of EGFR mutations and categorisation of non-classic EGFR mutations.

Interestingly, we detected a numerical difference in PFS and OS between patients with a classic EGFR mutation (exon 19 vs exon 21 mutation) that was significantly different for OS in favour of patients with an EGFR exon 19 deletion. Although originally it was thought that there was no difference between these two subtypes of classical EGFR mutations (Igawa et al, 2014), a meta-analysis detected a difference in PFS in favour of patients with an EGFR exon 19 deletion (Zhang et al, 2014). We did not detect a significant difference in PFS, but we did detect a difference in OS between these groups in accordance with a recent study (Rossi et al, 2016). Further investigation is warranted.

Several studies reported a higher prevalence of EGFR exon 20 insertions among women, nonsmokers and Asians (Huang et al, 2004; Kosaka et al, 2004; Shigematsu et al, 2005; Sasaki et al, 2007; Wu et al, 2008a), although another study did not find a significant difference in age, sex, ethnic origin or stage at diagnosis when compared with both patients with a classic EGFR mutation as in patients lacking a mutation (Arcila et al, 2013). Survival of NSCLC patients with an EGFR exon 20 insertion has generally been reported to be poor (Oxnard et al, 2013). Most exon 20 insertions are insensitive for treatment with both reversible and irreversible EGFR-TKIs (except for the EGFR exon 20 insertion A763_Y764insFQEA; Yasuda et al, 2013). This insensitivity is probably the reason of the poorer survival of this category of patients compared with NSCLC patients with classic EGFR mutations (Wu et al, 2008a; Lund-Iversen et al, 2012; Woo et al, 2014). In our cohort, PFS of patients with an EGFR exon 20 insertion on EGFR-TKI treatment was 2.9 months, comparable to the PFS of 1.5–2.0 months on erlotinib or gefitinib (Wu et al, 2008a; Jackman et al, 2009) and 2.7 months on afatinib (Yang et al, 2015b) that were reported previously. This suggests that these patients should preferably be treated with cytotoxic chemotherapy instead of first- and second-generation EGFR-TKIs. Recently, favourable results of a clinical study with AUY922 in NSCLC patients with an EGFR exon 20 insertion were reported and may hopefully provide a better treatment option for EGFR-mutated NSCLC patients with exon 20 insertion (Piotrowska et al, 2015).

Likewise, it has been reported that patients with uncommon EGFR mutations have lower EGFR-TKI sensitivity (Arrieta et al, 2015). We did not detect a statistically significant difference between patients with uncommon and classic EGFR mutations with respect to PFS, but considering the large numerical difference (6.4 vs 12.0 months, respectively), this is probably because of the small sample size and wide variation in PFS among patients with an uncommon EGFR mutation (Tables 3 and 4).

In our study, G719X/L816Q/S768I EGFR mutations are the most frequently detected among uncommon EGFR mutations, in line with previous reports (Mitsudomi and Yatabe, 2007; Shi et al, 2014). The G719X and the L861Q EGFR mutations were reported to have a shorter OS on gefitinib compared with classic EGFR mutations (Watanabe et al, 2014), although a recent study reported a PFS and OS of 13.8 and 26.9 months, respectively, for patients with a G719X EGFR mutation on first-line afatinib (Yang et al, 2015a). Chiu et al (2015) detected a statistically significant difference in median PFS on EGFR-TKI between patients (N=161) with single and double G719X/L816Q/S768I EGFR mutations. In our cohort, patients with double G719X/L816Q/S768I EGFR mutations not only had a statistically significant longer PFS on EGFR-TKI treatment compared with patients with single-mutant status at these loci, but also a longer OS. However, groups were small in our cohort, and hence interpretation should be with caution. Further investigation on the difference between single and double G719X, L816Q and/or S768I EGFR mutations is warranted.

It has been suggested that platinum-doublet treatment might be the best first-line treatment option for patients with (both single and double) G719X/L816Q/S768I EGFR mutations (Watanabe et al, 2014). However, taking into account the durable responses on EGFR-TKI treatment of several patients with a double EGFR mutation that included G719X, L861Q and/or S768I, in our opinion EGFR-TKI treatment could be considered as first-line treatment for these patients. In addition, high response rates of patients with G719X/L816Q/S768I EGFR mutations to first-line afatinib were recently reported (Yang et al, 2015a). Prospective trials are needed to elucidate this question.

Several limitations should be taken into account for this study. Because of the retrospective design, bias cannot be excluded. In addition, in a considerable part of the patients (especially in patients with an EGFR exon 20 insertion) data on performance score and smoking could not be retrieved from the medical records and the line of EGFR-TKI treatment (i.e., first-, second-line and so on) varied. Perhaps therefore, we detected a significant difference between the groups in smoking. In addition, in the early days of EGFR testing, clinical characteristics were taken into account. Therefore, there might have been a screening bias for women and nonsmokers. However, from 2012, all stage IV adenocarcinoma patients were tested for EGFR mutations, irrespective of gender, smoking status and race. A large molecular heterogeneity existed among patients with non-classic EGFR mutations. The results of the subgroup analyses should therefore be interpreted with caution. Furthermore, in routine pathology, solely tumour tissue is evaluated, and for most cases of our study no normal DNA was available to confirm the somatic origin of the mutations identified. The R776H mutation (detected in three patients in our cohort), for example, has both been reported as somatic and germline (Nagalakshmi et al, 2013; van Noesel et al, 2013). For one of our patients, analysis of normal DNA confirmed somatic nature (data not shown), but for the others because of absence of normal DNA a germline nature cannot be excluded.

To summarise, in this cohort of Dutch EGFR-mutated NSCLC patients, the prevalence and genotype distribution of non-classic EGFR mutations was in accordance with previously published studies among non-Asian, EGFR-mutated NSCLC patients. Outcome on EGFR-TKI treatment was poor for patients with EGFR exon 20 insertions and varied widely in patients with uncommon EGFR mutations. Further (prospective) studies on patients with non-classic EGFR mutations are warranted to hopefully improve prognosis of these patients.