BRIEF REPORT: Compound EGFR mutations and response to EGFR tyrosine kinase inhibitors

Abstract

INTRODUCTION

Non-small-cell lung cancers (NSCLCs) containing epidermal growth factor receptor (EGFR) mutations are exquisitely sensitive to EGFR tyrosine kinase inhibitors (TKIs). This is the case of the most common EGFR mutations affecting exon 18 (G719X), 19 (inframe deletions) and 21 (L858R and L861Q). However, the frequency of compound (i.e., double or complex) EGFR mutations - where an EGFR TKI sensitizing or other mutation is identified together with a mutation of unknown clinical significance – and their pattern of response/resistance to EGFR TKIs are less well described.

METHODS

We analyzed the EGFR mutation pattern of 79 cases of NSCLC harboring EGFR mutations, and compiled the genotype-response data for patients with NSCLCs with compound EGFR mutations treated with EGFR TKIs.

RESULTS

Out of the 79 EGFR mutated tumors identified, 11 (14%) had compound mutations. Most involved the EGFR TKI-sensitizing G719X (n=3, plus S768I or E709A), L858R (n=4, plus L747V, R776H, T790M or A871G), L861Q (n=1, plus E709V) and delL747_T751 (n=1, plus R776H). 8 patients received an EGFR TKI: 3 cases with G719X plus another mutation had partial responses (PR) to erlotinib; out of 3 cases with L858R plus another mutation, 2 displayed PRs and 1 (with EGFR-L858R+A871G) progressive disease to erlotinib; 1 NSCLC with EGFR-L861Q+E709A and 1 with delL747_T51+R776S had PRs to EGFR TKIs.

CONCLUSIONS

Compound EGFR mutations comprised 14% of all mutations identified during routine sequencing of exons 18–21 of EGFR in our cohort. Most patients with an EGFR TKI sensitizing mutation (G719X, exon 19 deletion, L858R and L861Q) in addition to an atypical mutation responded to EGFR TKIs. Reporting of the genotype-response pattern of NSCLCs with EGFR compound and other rare mutations, and the addition of this information to searchable databases will be helpful to select the appropriate therapy for EGFR mutated NSCLC.

INTRODUCTION

Lung cancer is ranked as the most common cause of cancer death in the United States and many developed countries. Non-small-cell lung cancers (NSCLCs) are the most prevalent form of this disease, and the understanding of its genetic diversity has led to the potential for molecularly-targeted interventions. In particular, mutations in the epidermal growth factor receptor (EGFR) gene and translocations involving the anaplastic lymphoma kinase (ALK) or the proto-oncogene tyrosine kinase c-ROS1 (ROS1) genes play a pivotal role in carcinogenesis of some lung cancers. Alterations in these actionable kinases predict the clinical effectiveness of tyrosine kinase inhibitors (TKIs) that target these aberrant oncogenes (1).

There are a multitude of somatic mutations in the epidermal growth factor receptor (EGFR) that have been described in NSCLC samples, and most cluster within the tyrosine kinase domain of this receptor tyrosine kinase. The most frequent – and clinically-significant – mutations include inframe deletions involving amino acids LREA of exon 19 and the exon 21 L858R. Together, these “classic” mutations account for almost 85% of all EGFR mutations. Gefitinib and erlotinib, oral EGFR TKIs, have been extensively studied in clinical trials that included NSCLCs with classic EGFR mutations, and both drugs lead to superior response rates (RRs) and progression-free survivals (PFSs) when compared to standard platinum-doublet cytotoxic chemotherapies (1;2). RRs to gefitinib/erlotinib exceed 60–70% in these trials, with median PFSs of more than 9–10 months and overall survival times beyond 20 months (1). Other EGFR mutations have also been associated with enhanced effects of EGFR TKIs. These include the less prevalent exon 18 G719X mutations (~3% of reported EGFR mutations) and the exon 21 L861Q (~2% of all EGFR mutations). In aggregate, G719X and L861Q mutated NSCLCs have been described to have RRs that exceed 50% and PFSs of >5 months in gefitinib/erlotinib-treated patients (3). On the contrary, other classes of EGFR mutations can be associated with lack of response to gefitinib or erlotinib. This is the case of the most prevalent EGFR exon 20 inframe insertion mutations (~5% of EGFR mutations) following the regulatory C-helix of EGFR (4).

Other EGFR mutations and tumors with multiple EGFR mutations have not been completely characterized. This is the case of compound EGFR mutations where an EGFR TKI-sensitizing mutation (such as G719X, exon 19 deletions, L858R or L861Q) coexists with uncommon mutations involving other residues of the tyrosine kinase domain of EGFR. Herein, we report the frequency of and responses to EGFR TKIs of our center’s cohort of compound EGFR mutated NSCLCs and provide a review of the literature on the pattern of response to EGFR TKIs of these mutation types. The date presented here will enhance efforts, such as Vanderbilt University’s DNA-mutation inventory to refine and enhance cancer treatment (DIRECT) database (http://www.mycancergenome.org/direct.php), to compile a searchable database for oncologists treating patients genotyped for EGFR mutations and other genetic alterations (5).

MATERIALS AND METHODS

Patient selection

Patients with a diagnosis of NSCLC and whose tumors were genotyped for EGFR mutations up to August 1st 2012 were identified through an ongoing Institutional Review Board (IRB) approved protocol at Beth Israel Deaconess Medical Center (BIDMC2009-P-000182).

Tumor genotype

EGFR mutation analysis was performed using standard DNA sequencing techniques with direct sequencing of exons 18 to 21 of EGFR. In brief, DNA was isolated from the sample, quantified and amplified by polymerase chain reaction using primers to exons 18–21 of EGFR (6). PCR products were analyzed by bi-directional direct DNA sequencing. Tumor genotype was performed in baseline diagnostic specimens prior to patient exposure to EGFR TKIs.

Data collection

Clinical, pathologic, tumor genotyping for EGFR mutation status, and response to EGFR TKIs was collected using the available electronic medical records of BIDMC and by direct review of radiographic studies. Response was calculated using RECIST (Response Evaluation Criteria In Solid Tumors) v1.1. Study data were collected and managed using REDCap electronic data capture tools hosted at BIDMC.

Statistical methods

The reporting of parameters involving clinical, pathological, radiographic, response data and tumor genotypes used descriptive methods.

RESULTS

Frequency of EGFR mutations

Table 1 summarizes the frequency of EGFR mutations identified during routine clinical genotype of our patients’ tumor specimens. Most mutations were single mutations (67/79, 84.75%) with exon 19 deletions (34/79, 43%) and L858R (24/79, 30.5%) being the most prevalent mutation types. EGFR exon 20 insertion mutations (6.3%), plus single mutations involving G719X (1.25%) and L861Q (2.5%) were less frequent. Out of the 79 unique tumors, 11 (14%) had compound mutations (Table 1).

Table 1.

Frequency of different EGFR mutations detected in NSCLC samples by direct sequencing of exons 18–21 of EGFR

EGFR mutation type	number	percentage of total
Single mutations	67	84.75%
exon 18 mutations	2	2.5%
E709_T710delETinsD	1	1.25%
G719A	1	1.25%
exon 19 deletions	34	43%
delE746_A750	26	33%
delE746_S752insV	2	2.5%
delL747_A750insP	1	1.25%
delL747_T751	3	3.75%
delL747_S752	1	1.25%
delL747_A753insQ	1	1.25%
exon 20 insertions	5	6.30%
A763_Y764insFQEA	1	1.25%
D770delDinsGY	1	1.25%
P772_H773dupH	2	2.5%
P772_H773dupPH	1	1.25%
exon 21 mutations	26	33%
L858R	24	30.5%
L861Q	2	2.5%
Compound mutations	11	14%
G719A+S768I	2	2.5%
G719S+E709A	1	1.25%
delL747_T751+R776S	1	1.25%
S768I+V769L	1	1.25%
H773L+V774M	1	1.25%
L858R+L747V	1	1.25%
L858R+R776H	1	1.25%
L858R+T790M	1	1.25%
L858R+A871G	1	1.25%
L861Q+E709V	1	1.25%
Other mutation	1	1.25%
G724fs	1	1.25%
Total	79	100%

EGFR compound mutations

Most (9/11, 81.8%) EGFR compound mutations involved the EGFR TKI-sensitizing G719X (n=3), L858R (n=4), L861Q (n=1) and delL747_T751 (n=1) mutations (Table 1). The partner mutation identified during sequencing was either an exon 18 atypical mutation (E709X, n=2), an exon 19 atypical mutation (L747V, n=1), an exon 20 atypical mutation (S7681, n=2; R776S, n=2; T790M, n=1) or an exon 21 atypical mutation (A871G, n=1). Table 1 summarizes how these combinations paired in each specific tumor.

Response to EGFR TKIs of EGFR mutated NSCLCs

Of the 11 patients with tumor genotype revealing a compound EGFR mutation, 8 received an EGFR TKI during their clinical course. Table 2 summarizes 7 patients that only received erlotinib.

Table 2.

Clinical, pathological, molecular characteristics and response to erlotinib of NSCLCs with compound EGFR mutations

EGFR mutation	sex/age(y ears)/eth nicity/PS	smoking history (pack-years)	histology	EGFR TKI (line of therapy)	dose (for >50% course)	response RECIST	percent change target lesion(s)	PFS (months)	total time on EGFR TKI (months)	OS (from start of EGFR TKI)
G719A+S768I	female/63/White/1	former (23)	NSCLC NOS	erlotinib (first line)	100mg/day	PR	−69.3%	5	5	8
G719A+S768I	female/78/Asian/1	never (0)	adenocar cinoma	erlotinib (first line)	100mg/day	PR	−33.3%	7	7	10
G719S+E709A	female/55/Black/1	never (0)	adenocarcinoma	erlotinib (first line)	150mg/day	PR	−54.5%	8	8	11
delL747_T751+ R776S *	female/65/White/0	former (30)	adenocarcinoma	erlotinib (first line)	25mg/every other day	PR	−41.1%	20	40+	40+
L858R+A871G	female/66/Asian/3	never (0)	NSCLC NOS	erlotinib (first line)	150mg/day	PD	+91.6%	2	2	3
L858R+L747V	female/64/White/1	never (0)	adenocarcinoma	erlotinib (first line)	150mg every other day	PR	−37.3%	6+	6+	6+
L858R+R776H	female/81/White/1	never (0)	adenocarcinoma	erlotinib (first line)	100mg/day	PR	−61.3%	3+	3+	3+

EGFR, epidermal growth factor receptor; PS, ECOG performance status; TKI, tyrosine kinase inhibitor; NSCLC, non-small-cell lung cancer; NOS, not otherwise specified; RECIST, Response evaluation in solid tumors version 1.1; PFS, progression-free survival; OS, overall survival; +, ongoing response, PFS or OS;

^*initial clinical course for this patient had been detailed previously (ref 7).

Out of the 3 cases with G719X plus another mutation, all had PR to erlotinib (at concentrations of 100–150 mg/day). However, in none did the PFS exceed 8 months and in none did the survival exceed 12 months.

Out of the 3 cases with L858R plus another mutation, the patients whose tumors had L858R+L747V and L858R+R776H displayed PRs that are ongoing (Table 2). The patient (who had a poor performance status at diagnosis) whose tumor had L858R+A871G had rapid PD and died within 3 months of starting erlotinib 150 mg/day (Table 2).

One patient with an exon 19 deletion (delL747_T751)+R776S mutated NSCLC, who had her initial clinical course detailed previously (7), had a prolonged response to a low dose of erlotinib with a PFS of 20 months (Table 2). The final patient (a 68 year-old white, never smoker, with ECOG performance status of 0 and stage IV adenocarcinoma) with a NSCLC harboring L861Q+E709A had a prolonged PR (decreased in 48% of target lesions) to an experimental irreversible EGFR TKI (as first line systemic therapy) that lasted for 25 months, followed by an additional 7 months of erlotinib at a dose of 150 mg/every other day until further progression at month 32 of EGFR TKIs.

DISCUSSION

EGFR mutations are considered the most robust predictive biomarker for the clinical and radiographic responses attained by EGFR TKIs in clinical practice. Single EGFR mutations or multiple mutations can be identified during sequencing of exon 18–21 of EGFR. The clinical prevalence and significance of multiple mutations have not been completely ascertained. One of the largest cohorts of EGFR mutated NSCLCs disclosed that ~7% of the tumors harbored compound mutations involving the EGFR TKI sensitizing mutations G719X, exon 19 deletions, L858R and L861Q in combination with other EGFR mutations in Taiwanese patients (3). Others have reported frequencies lower than 4% for these mutations (8). The method of EGFR mutation detection (direct sequencing of exon 18–21 [as used in this report], mutant-enriched polymerase chain reaction [PCR], PNA-LNA PCR, PCR clamp, allele-specific reactions, high-resolution melting analysis, Applied Biosystems (ABI) Prism SNaPshot Multiplex system, among others) and the potential introduction of artifacts generated in the PCR step may account for the unequal prevalence of compound mutations in prior series (5;9). The cohort reported here indicated a higher prevalence of EGFR compound mutations, at 14% of our 79 cases. Therefore, it seems that the frequency of EGFR compound mutations is not insignificant.

The genotype-clinical/radiographic response pattern of EGFR compound mutations has not been completely defined and neither has there been a consorted effort to determine if these compound mutations occur in cis or in trans within EGFR alleles. We will summarize the available literature below. NSCLCs with G719X mutations in addition to other mutations are commonly reported. As an example, in the aforementioned Taiwanese cohort 8 out of the 15 NSCLCs with G719X harbored compound G719X plus other mutations and the RR for G719X mutations (single + compound) was 53.3% with a PFS of 8.1 months (3). In our dataset, 3/4 (75%) G719X mutations were compound. The EGFR-G719A+S768I mutation, which we identified in 2 patients with PR to erlotinib, has been reported by others in two NSCLCs that had lack of responses to gefitinib (10;11) and EGFR-S768I alone has been reported in a gefitinib-responsive NSCLC (12). The in vitro characterization of G719S+S768I showed that this EGFR is as sensitive to gefitinib as G719S alone (13). A NSCLC with G719S+E709A, which we identified in 1 patient with PR to erlotinib, has been previously reported as having a PR to the irreversible EGFR TKI neratinib (14). The in vitro characterization of G719C+E709A demonstrated that the resulting EGFR protein is only slightly less sensitive to gefitinib than G719C alone (15), which may not be significant at clinical doses used for erlotinib (at 150 mg/day the serum trough exceeds 1.5µM) or irreversible EGFR TKIs (16). Exon 19 deletions are less frequently identified as compound mutations, outside the setting of acquired resistance to gefitinib/erlotinib where they are frequently found with EGFR-T790M that shifts the ensuing proteins into resistant patterns to clinically achievable doses of gefitinib and erlotinib (17). It seems other compound exon 19 mutations, such as the erlotinib responsive EGFR-delL747_T751+R776H reported by our group (7), retain their responsiveness to EGFR TKIs.

L858R mutations are frequently reported a part of compound mutations, with ~10% of L858R mutated NSCLC in association with other mutations (3). We identified 5/29 (17.25%) of our EGFR-L858R as compound mutations. How these compound mutations affect the response to gefitinib and erlotinib of L858R alone is not completely known. Our group has characterized in vitro EGFR-L858R+L747S and shown that this compound mutation shifts the sensitivity curve to EGFR TKIs in a manner that would make it resistant to trough serum concentrations of a dose of gefitinib 250 mg/day (<0.5µM) but not to serum troughs (>0.5–1µM) of erlotinib at doses above 50–100 mg/day (18). The same seems true for L858R+D761Y (19). We and others have also shown that EGFR-L858R+T790M is resistant to gefitinib and erlotinib at their maximum tolerated doses (16;18). The in vitro characterization of L858R+E709A, L858R+E709G and L858R+L838V showed that these EGFRs are as sensitive to gefitinib as L858R alone (13). In this report, we show that NSCLCs with L858R+L747V and L858R+R776H had PRs to erlotinib. A prior NSCLC with L858R+R776H has also been described as having PR to gefitinib (11). The case of EGFR-L858R+A871G NSCLC reported here as having PD to erlotinib 150 mg/day may indicate that this compound mutation may confer resistance to EGFR TKIs; an assertion that will require in vitro modeling. EGFR-L858R+H870R in vitro is less sensitive to gefitinib than L858R alone (15).

L861Q mutations can be part of compound mutations in more than half of EGFR-L861Q positive cases identified. We observed 1/3 (33.3%) of cases of L861Q as a compound mutation. Others have reported RRs as high as 60% for single or compound mutant L861Q bearing tumors with PFSs of 6 months (3). We report a L861Q+E709A mutation that had a prolonged PR to an experimental irreversible EGFR TKI followed by additional disease control with erlotinib.

In summary, compound EGFR mutations comprised 14% of all mutations identified during routine sequencing of exons 18–21 of EGFR in our cohort. Most tumors with an EGFR sensitizing mutation (G719X, exon 19 deletions, L858R and L861Q) in addition to an atypical mutation responded to EGFR TKIs. Reporting of the genotype-response pattern of NSCLCs with EGFR compound and other rare mutations will help define the complete spectrum of how EGFR genotype affects the response to EGFR TKIs of NSCLCs.