Structure-based classification of EGFR mutations in operable pre-invasive and invasive non-small cell lung cancer: a cross-sectional study.

Background: It has been reported that the structure-based approach for defining functional groups of epidermal growth factor receptor (EGFR) mutations predicts the efficacy of EGFR inhibitors better than the traditional exon-based approach in the advanced stage. However, less is known about this structure-based classification of EGFR mutations in operable early-stage lung adenocarcinoma.
Methods: Non-small cell lung cancer (NSCLC) patients with pathological stage I-III or adenocarcinoma in situ (AIS) who had EGFR mutations identified in next-generation sequencing (NGS) testing were recruited. Both exon-based and structure-based groupings of EGFR mutations were compared between the AIS and stage I-III patients using Fisher's exact test.
Results: In total 1,012 patients including 66 AIS and 946 stage I-III patients were analyzed in the study. A total of 1185 EGFR mutations were identified in the 1,012 NSCLC patients, of whom 84.39% harbored a single EGFR mutation and 15.61% harbored complex EGFR mutations. As expected, L858R was more common than 19del in our population (39.33% vs. 35.67%). Interestingly, concurrent L858R and 19del mutations were identified in 9 patients (0.89%), and all these patients were diagnosed with multiple primary lung cancer. A higher percentage of atypical EGFR mutations was identified in the AIS cohort than in the stage I-III NSCLC cohort (33.33% vs. 21.66%, P=0.03). According to the structure-based classification of EGFR mutations, 86.07%, 7.11%, 5.04%, and 1.78% of the EGFR mutations were classified as classical-like, P-loop and α C-helix compressing (PACC), exon 20 insertions (Ex20ins), and T790M-like mutations, respectively. The composition of EGFR mutations was different between patients <65 and ≥65 years (P=0.0267) but similar between patients with AIS and stage I-III NSCLC (P=0.1436). However, a higher percentage of Ex20ins occurred in younger (<65 years) patients, nonsmoking patients, and patients with AIS (6.7% vs. 2.5%, P=0.003; 5.8% vs. 0.8%, P=0.0107; and 10.6% vs. 4.7%, P=0.0423, respectively). Conclusions: This large cross-sectional study delineated the structure-based classification of EGFR mutations in patients with operable NSCLC. While the traditional exon-based EGFR grouping showed difference between AIS and stage I-III NSCLC cohort, no difference was identified in the structural approach. Which approach had better prediction of targeted therapy efficacy in adjuvant settings warrants further investigation. 2022 Journal of Thoracic Disease. All rights reserved.

Introduction

Lung cancer is the most common cause of cancer death. It is broadly divided into non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC), with 80–85% classified as NSCLC (1). Although up to 30% of patients with NSCLC can be diagnosed early and undergo curative surgery, disease recurrence is still common in early-stage disease (2,3). Nearly half of patients with stage IB NSCLC and more than three-quarters of patients with stage IIIA NSCLC experience recurrence within 5 years (4). Adjuvant treatment is recommended for patients to reduce the risk of postoperative recurrence (5-7).

Epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) are a targeted first-line treatment for patients with EGFR mutation–positive advanced NSCLC (8-11). Several studies have shown that EGFR-TKIs as adjuvant therapy improve the prognosis of early-stage patients with EGFR exon 19 deletions and L858R mutations (12-14). In addition to classical EGFR mutations, atypical EGFR mutations have been identified in 10–30% of patients with NSCLC (15-17). However, the use of EGFR-TKI treatment for patients with atypical EGFR mutations has not been well-studied. To understand the effect of atypical EGFR mutations on patient outcome, a recent study proposed a structure-based approach for improving the prediction of drug sensitivity in patients with atypical EGFR mutations (18). Four EGFR mutation subgroups were identified based on structure–function using a drug sensitivity assay and an in silico prediction model: (I) classical-like mutations that were distant from the ATP-binding pocket and were predicted to have little effect on the overall structure of EGFR, including L861Q, T725M, and EGFR classical mutations. These mutations were sensitive for all classes of EGFR-TKIs; (II) exon 20 insertions (Ex20ins), insertions in the loop at the C-terminal end of the α C-helix in exon 20, which can be subdivided into near-loop (NL) and far-loop (FL) insertions based on in vitro sensitivity. Second-generation TKIs and Ex20ins-active TKIs were more sensitive in Ex20ins-NL than in Ex20ins-FL; (III) mutations on the interior surface of the ATP-binding pocket or C-terminal end of the α C-helix, which were predicted to be P-loop and α C-helix compressing (PACC), including G719A and E709A. PACC mutations were more sensitive to second-generation TKIs than any other TKI class; (IV) T790M-like mutations in the hydrophobic core, which were mostly composed of complex mutations combined with T790M mutations. T790M-like mutations consist of 2 subgroups of third-generation TKI-sensitive (T790M-like-3S) and third-generation TKI-resistant (T790M-like-3R) mutations.

Adenocarcinoma in situ (AIS) is a subtype of NSCLC, which exhibits early-stage growth patterns but can develop into invasion (19). The 10-year recurrence-free survival of AIS is 100% with appropriate therapy, and the 10-year overall survival is 98.1% (20). However, the 5‐year overall survival rate of advanced NSCLC patients is less than 7% (21). While there was a great amount of data about EGFR mutations in advanced NSCLC, less is known about AIS. In the study of genomic and immune profiling of pre-invasive lung adenocarcinoma, 28 AIS patients were included (22). Another study analyzed the mutational profile of Chinese NSCLC patients of adenocarcinoma including 21 AIS patients (23). Nakamura et al. (24) reported the EGFR mutation rates in earliest phases of lung adenocarcinoma with 22 AIS and 21 minimally invasive adenocarcinoma (MIA). In this study, we were able to include 66 AIS patients as a subgroup of pre-invasive NSCLC for the structural analysis of EGFR mutations.

In total, we analyzed 1,012 NSCLC patients with pathological stage I–III or AIS to evaluate relationship of the clinical characteristics and EGFR mutations stratified either by traditional exon-based method or the structure-based approach. We present the following article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-1054/rc).

Methods

Patients and tissue samples

To systemically analyze the structure-based classification of EGFR mutations in operable NSCLC patients, we performed this retrospective descriptive cross-section study. After excluding patients with stage IV or without EGFR mutations at stage I–III were excluded, 1,012 stage I–III NSCLC patients with EGFR mutation tested with next-generation sequencing (NGS) from Tongxiang First People’s Hospital, First Medical Center of PLA General Hospital and Affiliated Qingdao Central Hospital between May 2018 and October 2021 were recruited. Clinical data were collected from the medical records of each patient, including gender, age, smoking status, histological subtype, and disease stage at diagnosis. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Institutional Review Board of the Tongxiang First People’s Hospital (No. 2022-002-01) and informed consent was taken from all the patients.

DNA extraction and NGS

Formalin-fixed paraffin-embedded (FFPE) slides were stored at room temperature. Genomic DNA was extracted from FFPE tumor samples using a QIAamp DNA FFPE Tissue Kit according to the manufacturer’s protocol (Qiagen GmbH, Hilden, Germany). DNA from leukocytes was extracted using a DNeasy Blood Kit (Qiagen). DNA concentration and size distribution were estimated using a Qubit Fluorometer and a Qubit dsDNA HS Assay Kit according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA, USA).

Sequencing library preparation and sequencing protocol were conducted as described previously (25,26). Briefly, genomic DNA libraries were constructed with a KAPA DNA Library Preparation Kit (Kapa Biosystems, Wilmington, MA, USA). The capture probe design was based on approximately 1.45 Mb genomic regions of 1,021 genes frequently mutated in solid tumors. DNA sequencing was performed using Gene + Seq-2000 (Geneplus, Beijing, China) with paired-end reads. Matched peripheral blood was sequenced as a control to filter germline variation.

EGFR mutation analysis

Terminal adaptor sequences and low-quality data were removed. The clean reads were aligned to the human genome build GRCh37 using the Burrows-Wheeler Aligner (version 0.7.12-r1039; http://bio-bwa.sourceforge.net/). Single nucleotide variants (SNVs) and insertions or deletions (indels) were identified using GATK (version 3.4-46-gbc02625; Broad Institute, Cambridge, MA, USA) and MuTect (version 1.1.4; Broad Institute). All final candidate variants were verified using an integrative genomics viewer browser. The exon-based EGFR mutation types were categorized into 8 subgroups: Ex19del, L858R, T790M, classical + T790M, Ex20ins, other atypical, complex atypical, and Ex19del + L858R. Using the structure-based approach (18), the following 4 EGFR mutation subgroups were established: classical-like group, PACC group, Ex20ins group, and T790M-like group.

Statistical analysis

Differences among subgroups stratified by stage, gender, age, and smoking status were analyzed by Fisher’s exact test, where appropriate. All analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Two-sided P values of <0.05 were considered statistically significant.

Results

Clinical characteristics and EGFR mutations

A total of 632 female and 380 male NSCLC patients with a mean age of 61 years (range, 21–96 years) were included in this study. The clinical characteristics of the patients are listed in . A total of 66 patients (6.52%) with AIS were included in this research. A total of 1,185 EGFR mutations were identified in the 1,012 patients with NSCLC, of whom 84.39% and 15.61% harbored a single EGFR mutation and complex EGFR mutations, respectively. As shown in , L858R and 19del were the major types; as expected, L858R was more common than 19del in our population (39.33% vs. 35.67%). In addition, EGFR T790M mutation was identified in 18 patients (1.78%), including 1 patient with a T790M mutation but no other EGFR mutations. Concurrent L858R and 19del mutations were identified in 9 patients (0.89%), who were all diagnosed with multiple primary lung cancer. The detailed clinical and molecular data of the 9 patients are shown in .

Table 1

Clinical characteristics of 1,012 patients with NSCLC

Clinical factors	Number (%)
Gender
Female	632 (62.45)
Male	380 (37.55)
Age (years)
Median [range]	61 [21–96]
<65	616 (60.87)
≥65	396 (39.13)
Stage
0	66 (6.52)
I–III	946 (93.48)
Smoking status
Smokers	132 (13.04)
Nonsmokers	504 (49.80)
Unknown	376 (37.16)
Histology
Adenocarcinoma	916 (90.51)
Squamous-cell carcinoma	17 (1.68)
Adenocarcinoma in situ	66 (6.52)
Other types of NSCLC	13 (1.29)

NSCLC, non-small cell lung cancer.

Figure 1

The classical and atypical EGFR mutations according to traditional classification. (A) Percentage of patients with NSCLC containing classical and atypical EGFR mutations (n=1,012 patients). (B) EGFR mutations of AIS and invasive NSCLC (stage I–III). AIS, adenocarcinoma in situ; EGFR, epidermal growth factor receptor; NSCLC, non-small cell lung cancer.

Table 2

Patients with concurrent EGFR L858R and EX19del mutations

Patient ID	Gender	Age	Stage	EGFR mutation	Tumor site	Other mutations
E0028	Female	67	I	L858R	Left upper lobe	LRP1B p.R1619H; MLL3 p.A4152V; SMO p.A324T
E0028	Female	67	I	E746_T751delinsV	Left upper lobe	TMPRSS2 p.I521L
E0155	Female	65	Ib	E746_A750del	Right upper lobe	CTNNB1 p.S37F; KRAS p.G12S; RARA p.M284V; SMAD4 p.Q334*
E0155	Female	65	Ib	L858R	Left lower lobe	ATRX p.M2492L; RAD51B p.S175I
E0183	Male	74	I	L747_T751del	Right upper lobe	ABL1 p.R1095W; ACIN1 p.E274G; MAP3K1 p.V1435G; TP53 p.H193D
E0183	Male	74	I	L858R	Right upper lobe	AXL p.A210T; GAB2 p.W661*; HNRPDL p.M120V; PIK3C2B p.D1537_P1538[2>1]; SETD2 p.T904I; SLX4 p.P1095T
E0212	Female	59	Ia	L747_T751del	Right lower lung	No
E0212	Female	59	Ia	L858R	Right upper lobe	No
E0446	Female	52	II	L858R	Right upper lobe	No
E0446	Female	52	II	E746_A750del	Right upper lobe	FANCM p.A141S; PTPRD p.L503V
E0735	Female	66	I	E746_A750del	Left upper lobe	ARAF p.K336N; ATM p.V2951D; DOT1L p.W611R; DOT1L p.E1360D
E0735	Female	66	I	L858R	Left upper lobe	RBM10 p.Q843*; TXNIP p.V54Sfs*20
E0790	Female	53	II	S752_I759del	Left lower lobe	CSF1R p.N240S; RBM10 p.E624*
E0790	Female	53	II	L858R	Right upper lobe	RBM10 p.E578*; SETD2 p.Q2334*
E0790	Female	53	II	L858R	Left lower lobe	No
E0897	Female	66	I	L858R	Right middle lobe	CYP2D6 p.D100E; MED12 p.S440T; NF1 p.D1091V; RBM10 p.E494*
E0897	Female	66	I	E746_A750del	Left upper lobe	AXIN1 p.A443V
E0966	Female	70	I	E746_A750del	Left upper lobe	WT1 p.Q238R
E0966	Female	70	I	L858R	Left upper lobe	EPHA3 p.D316G

*, translation stop codon. EGFR, epidermal growth factor receptor.

AIS differed from stage I–III NSCLC in terms of exon-based EGFR mutation classification

To comprehensively investigate the EGFR mutation characterization of early-stage NSCLC, 66 patients with AIS and 946 patients with stage I–III NSCLC were included in the study. The mutation profiles of EGFR in the AIS and stage I–III patients are listed in . A different EGFR mutation distribution was observed between AIS and stage I–III patients. As shown in , 37.88%, 27.27%, and 33.33% AIS patients harbored Ex19del, L858R, and atypical mutations, respectively. In the stage I–III group, the distribution of EGFR mutations was 35.52% (Ex19del), 40.17% (L858R), and 21.67% (atypical; P=0.03). The proportion of L858R in AIS patients was lower than that in stage I–III patients, while the proportion of atypical mutations was higher in AIS patients.

Structural-based EGFR mutation classification

Based on a previous publication, we classified EGFR mutations of patients with pre-invasive and invasive NSCLC into 4 distinct subgroups with structural features: (I) classical-like; (II) PACC; (III) Ex20ins, including Ex20ins-NL and Ex20ins-FL; and (IV) T790M-like (). Classical-like mutations were the largest subgroup of EGFR mutations (871 patients, 86.07% of the cohort), followed by PACC (72 patients, 7.11%), Ex20ins (51 patients, 5.04%), and T790M-like (18 patients, 1.78%). The patents of Ex20ins-NL group were more than the Ex20ins-FL group (3.95% vs. 1.09%). The frequency of Ex20ins-NL and Ex20ins-FL mutations is shown in the lollipop plot in . The clinical characteristics and subgroups of EGFR mutations in patients are presented in . Our analyses indicated that the composition of EGFR mutations was different between patients <65 and ≥65 years (P=0.0267) but similar between patients with AIS and patients with stage I–III NSCLC (P=0.1436). However, a higher percentage of Ex20ins occurred in younger (<65 years) patients, nonsmoking patients, and AIS patients (6.7% vs. 2.5%, P=0.003; 5.8% vs. 0.8%, P=0.0107; and 10.6% vs. 4.7%, P=0.0423 respectively).

Figure 2

Classification of EGFR mutations according to a structure-based approach. (A) Four distinct subgroups of EGFR mutations according to a structure-based approach (n=1,012). (B) Distribution of Ex20ins-NL mutations and Ex20ins-FL mutations in the cohort. Ex20ins-NL, exon 20 insertions near-loop; Ex20ins-FL, exon 20 insertions far-loop; PACC, P-loop and α C-helix compressing; EGFR, epidermal growth factor receptor.

Table 3

Patient characteristics stratified by EGFR mutation type

Clinical factors	Classical-like (N=871)	PACC (N=72)	Ex20ins (N=51)	T790M-like (N=18)	P value
Sex					0.1325
Female	550	39	35	8
Male	321	33	16	10
Age (years)					0.0267
<65	522	41	41	12
≥65	349	31	10	6
Stage					0.1436
AIS	52	5	7	2
I–III	819	67	44	16
Smoking status
Smokers	115	12	1	4	–
Nonsmokers	444	24	29	7
Unknown	312	36	21	7
Histology
Adenocarcinoma	794	64	44	14	–
Squamous-cell carcinoma	12	3	0	2
AIS	52	5	7	2
Other types of NSCLC	13	0	0	0

The associations of EGFR mutation type with clinical variables were evaluated by Fisher’s exact test. P<0.05 was considered statistically significant, and all tests were two-tailed. EGFR, epidermal growth factor receptor; PACC, P-loop and α C-helix compressing; Ex20ins, exon 20 insertions; AIS, adenocarcinoma in situ; NSCLC, non-small cell lung cancer.

Discussion

Research on EGFR-TKI treatment for early-stage NSCLC patients with atypical EGFR mutations is lacking. To our knowledge, this is the first study to delineate the structural classification of EGFR mutations in early-stage NSCLC and AIS using a large cohort.

In our cohort, L858R was more common than 19del, which was similar to previous studies in East Asian patients (27,28) and different to previous study in Western patient populations (18). The atypical mutations were also less common in our cohort than in Western patients (22.43% vs. 30.8%) (18). When classified with traditional exon-based method, patients with AIS exhibited a higher proportion of atypical EGFR mutations than patients with stage I–III lung cancer (33.33% vs. 21.66%; P=0.03). However, according to the structure-based method, the proportion of atypical EGFR mutations in patients with AIS was similar to those with stage I–III lung cancer. In addition, the mutation rate of Ex20ins differed in patients according to age, smoking, and invasion stage. Moreover, we reported 9 patients with Ex19del + L858R double mutations who were diagnosed with synchronous multiple primary lung cancer. As all our patients were EGFR-TKI naïve, there were no T790M-like-3R mutations, which are rare in EGFR-TKI naïve patients but common in 3rd generation EGFR-TKI treated patients (18).

Previous research has indicated that the frequency of EGFR mutation is about 27.3–52% in pre-invasive lung adenocarcinoma, suggesting that EGFR mutations may be an early genetic event in the development of lung cancer (23,24,29,30). Our study included 66 patients with AIS, and the EGFR mutation spectrum was as follows: Ex19del mutation (37.88%), L858R mutation (27.27%), and atypical mutation (33.33%). A retrospective study with 28 AIS patients identified 11 EGFR mutations in 10 samples, including 6 patients with Ex19del, 2 patients with L858R, and 2 patients with atypical mutations (30). Another study identified 4 Ex19del mutations, 4 L858R mutations, 1 Ex19del + L858R double mutation, and 1 other mutation in 10 patients with EGFR mutation-positive atypical adenomatous hyperplasia (AAH)/AIS (31). The different EGFR mutation rate between our patients and previous studies may have been caused by limited sample size or the method of EGFR detection.

Patients harboring EGFR Ex20ins exhibited poorer prognosis compared to patients with sensitizing mutations in EGFR. Ex20ins mutations are heterogenous in EGFR-TKI (32,33). Previous studies have demonstrated that patients with EGFR Ex20ins mutations are usually non-smoking females (34-37). Arcila et al. reported that among 33 patients with EGFR Ex20ins, 67% were female, 48% had never smoked, 55% were in stage I–II, and 45% were in stage III–IV (34). EGFR Ex20ins were more common among non-smoking patients (P<0.0001), and no significant difference was detected in age, sex, race, or stage. Another study with 27 patients with EGFR Ex20ins found that 19 patients were females (P=0.24), 15 patients had never smoked (P<0.001), 8 patients were in stage I–III, and 19 patients were in stage IV (P=0.05) (35). Our study further demonstrated that in patients with early-stage NSCLC, Ex20ins occur in patients who are younger, nonsmoking, and have AIS (P=0.003, P=0.017, and P=0.0423, respectively).

There were several limitations in our study. Firstly, it was a retrospective analysis, and thus no EGFR-TKI adjuvant therapy data were included. Secondly, only AIS data were available in our cohort, and thus AAH and MIA data were not included. Finally, the smoking status of some patients was not clear.

In summary, we have delineated the structural classification of EGFR mutations in early lung cancer and AIS using a large cohort. Whether this approach can improve predictions of targeted therapy efficacy in adjuvant therapy is worthy of further study.

The article’s supplementary files as