EGFR Exon 20 Insertion in Metastatic Non-Small-Cell Lung Cancer: Survival and Clinical Efficacy of EGFR Tyrosine-Kinase Inhibitor and Chemotherapy

Simple Summary EGFR exon 20 insertions are rare genetic alterations in non-small-cell lung cancers (NSCLCs) that are usually unresponsive to approved EGFR tyrosine kinase inhibitors (TKIs), but data is limited about this population. We aim to describe clinical features, survival, and response to chemotherapy and TKIs in this patient population. Abstract EGFR exon 20 insertions are rare genetic alterations in non-small-cell lung cancers (NSCLCs) that are usually unresponsive to approved EGFR tyrosine kinase inhibitors (TKIs). In this paper, we describe the clinical characteristics, efficacy of EFGR TKIs and chemotherapy, and resulting survival in this population. We retrospectively collected patients with EGFR exon 20 insertions (Exon20ins) from 11 French genetic platforms and paired them (1:2 ratio) with classic Exon 19/21 EGFR mutation patients (controls). Between 2012 and 2017, 35 Exon20ins patients were included. These patients were younger at diagnosis than the controls. All Exon20ins patients who were treated with first-line EGFR TKIs (n = 6) showed progressive disease as the best tumor response. There was no significant difference in the tumor response or the disease control rate with first-line platinum-based chemotherapy between the two groups. A trend towards shorter overall survival was observed in Exon20ins vs. controls (17 months (14—not reach(NR) 95% confidence interval(CI) vs. 29 months (17–NR 95%CI), p = 0.09), respectively. A significant heterogeneity in amino acid insertion in EGFR exon 20 was observed. EGFR exon 20 insertions are heterogeneous molecular alterations in NSCLC that are resistant to classic EGFR TKIs, which contraindicates their use as a first-line treatment.


Introduction
Epidermal growth factor receptor (EGFR) mutations, mainly exon 19 deletions and L858R exon 21 mutations, are present in about 12% of lung adenocarcinomas tested in France [1]. Besides these canonical mutations, in-frame insertion mutations in EGFR exon 20 (Exon20ins) are the third-most frequent subtype of EGFR mutation, accounting for up to 4% of EGFR mutations [2]. Exon20ins represent a heterogeneous group of insertions and/or duplications affecting the C-helix domain and the following loop, usually between amino acids 762 and 775. These mutations do not affect the ATP-binding pocket without modifying the affinity of early-generation tyrosine kinase inhibitors (TKIs), compared to wild-type EGFR receptors. Moreover, these insertions/duplications decrease the noncovalent binding of these TKIs through steric hindrance [3]. Exon20ins are known to be generally unresponsive to standard TKIs (erlotinib, gefitinib, afatinib), and their heterogeneity has slowed down the characterization of each alteration and their sensitivity to those TKIs. Recent progress has permitted a comprehensive approach to these molecular alterations and the development of new targeted treatments in both early and advanced phases. A new TKI (mobocertinib) and a specific antibody (amivantamab) recently received FDA breakthrough therapy designations for previously treated Exon20ins NSCLC patients. On-going phase III trials (EXCLAIM-2, NCT04129502 for mobocertinib; and PAPILLON, NCT04538664 for amivantamab) aim to compare these innovative therapies against usual platinum-based chemotherapy. In this period of rapid progress, we aim to improve the characterization of this rare population.
The purpose of our study was to describe EGFR Exon20ins NSCLC patients and evaluate their response to first-line historical ECGF TKIs or chemotherapy.

Clinical and Molecular Features at Baseline
Between 2012 and 2017, 35 patients with metastatic EGFR Exon20ins NSCLC were included and paired with 76 patients with classic EGFR exon 19/21 deletion/mutation metastatic NSCLC (control group).
Clinical features at baseline are summarized in Table 1. Exon20ins patients were younger than patients in the control group at diagnosis (mean age 63.8 vs. 69.6 years old, respectively; p = 0.02). There was no significant difference in sex, smoking status (active/former smokers vs. non-smokers) or clinical pattern (number of metastatic sites, presence of brain metastases, pleural or pericardial effusion) between the two groups.
EGFR exon 20 insertions are heterogeneous at the molecular level. We found 19 different mutations in the 35 patients harboring EGFR exon 20 insertions, with p.Ala767_Val769dup (n = 6) and p.Ser768_Asp770dup (n = 6) being the most frequent. No patient harbored any insertions before alanine in position 767, usually considered proximal insertions. Fifty-six patients in the control group had an EGFR exon 19 deletion, and twenty patients had an EGFR exon 21 L858R mutation.
Baseline molecular features are summarized in Table 2. Figure 1 shows comparative data between the CBio Cancer Genomics Portal data and data from the present study. The CBio Cancer Genomics Portal is an open-access database providing access to the genomics data of thousands of tumors from hundreds of cancer studies [4,5]. For NSCLC cancer, the CBio Portal aggregates 20 studies, gathering more than 6000 samples.

Treatment
Treatment characteristics are summarized in Table 3 and Table 4.

Response to EGFR TKIs
Tumor responses to EGFR TKIs are summarized in Table 5. The response to EGFR TKIs differed significantly between the two groups, with a 0% objective response rate (ORR) in the Exon20ins group vs. 68% in the control group (p < 0.01). Disease control (DC) with first-line EGFR TKIs was 0% in the Exon20ins group vs. 84% in the control group with a complete response (CR) in 3%, a partial response (PR) in 65%, and stable disease (SD) in 16% (p < 0.01).
Moreover, among Exon20ins patients receiving first-or second-generation EGFR TKIs at any line (n = 19 with evaluable response), none showed ORR.

Response to Platinum-Based Chemotherapy
Tumor response after platinum-based chemotherapy is summarized in Table 5. There was no significant difference in tumor response with the first-line platinumbased chemotherapy between the two groups. The DC was 82% in the Exon20ins group vs. 64% in the control group (p = 0.3).

Overall Survival
There was a trend towards shorter overall survival (OS) in the Exon20ins group compared to the control group (median survival 17 (14-NR, 95%CI) months vs. 29 (27-NR, 95%CI) months, respectively (p = 0.09)), as shown in Figure 2. The response to EGFR TKIs differed significantly between the two groups, with a 0% objective response rate (ORR) in the Exon20ins group vs. 68% in the control group (p < 0.01). Disease control (DC) with first-line EGFR TKIs was 0% in the Exon20ins group vs. 84% in the control group with a complete response (CR) in 3%, a partial response (PR) in 65%, and stable disease (SD) in 16% (p < 0.01).
Moreover, among Exon20ins patients receiving first-or second-generation EGFR TKIs at any line (n = 19 with evaluable response), none showed ORR .

Response to Platinum-Based Chemotherapy
Tumor response after platinum-based chemotherapy is summarized in Table 5. There was no significant difference in tumor response with the first-line platinumbased chemotherapy between the two groups. The DC was 82% in the Exon20ins group vs. 64% in the control group (p = 0.3).

Outlier Long Survivors in Exon 20 Insertion Group
Four patients in the Exon20ins group had a markedly longer OS ≥ 24 months. The types of insertions and the treatment these patients received are presented in Table 6. All other patients (n = 31) had OS ≤ 18 months, with a median OS of 17 months. Among the

Outlier Long Survivors in Exon 20 Insertion Group
Four patients in the Exon20ins group had a markedly longer OS ≥ 24 months. The types of insertions and the treatment these patients received are presented in Table 6. All other patients (n = 31) had OS ≤ 18 months, with a median OS of 17 months. Among the long survivors, patient #1 (Asn771_Pro772insCysAlaTyr) received afatinib for 9 months with stable disease, patients #2 and #3 were heavily treated with three and five lines of chemotherapy/immunotherapy, respectively, followed by one course of EGFR TKI. The last patient (#4) had SD for 18 months under first-line chemotherapy with carboplatinpemetrexed/pemetrexed maintenance. None of these mutations are known to be associated with a better prognosis.

Discussion
The aim of our study was to describe Exon20ins NSCLC patients and their response to historical EGFR TKIs and platinum-based chemotherapy. Between 2012 and 2017, thirty-five Exon20ins patients were included in our study. These patients were quite comparable with classic EGFR mutation patients except they had younger ages at diagnosis.
Overall survival appeared to be longer in the classic exon 19/21 group than in the Exon20ins group with a 12 month gap in median OS (29 months vs. 17 months). However, this was not statistically significant (p = 0.09), probably due to the small number of patients and the few outlier long survivors in the Exon20ins group. All four Exon20ins long survivors harbored a different insertion, forbidding any molecular explanation for this outcome The 17 month OS reported here is consistent with previous observations [6,7]. Although they are controversial, some retrospective studies and post-hoc analyses in small groups of patients, such as the study by Naidoo et al. (n = 46), found no difference in OS between patients with exon 19/21 mutations and Exon20ins (31 months vs. 26 months, respectively, p = 0.53) [8].

Discussion
The aim of our study was to describe Exon20ins NSCLC patients and their response to historical EGFR TKIs and platinum-based chemotherapy. Between 2012 and 2017, thirtyfive Exon20ins patients were included in our study. These patients were quite comparable with classic EGFR mutation patients except they had younger ages at diagnosis.
Overall survival appeared to be longer in the classic exon 19/21 group than in the Exon20ins group with a 12 month gap in median OS (29 months vs. 17 months). However, this was not statistically significant (p = 0.09), probably due to the small number of patients and the few outlier long survivors in the Exon20ins group. All four Exon20ins long survivors harbored a different insertion, forbidding any molecular explanation for this outcome.
The 17 month OS reported here is consistent with previous observations [6,7]. Although they are controversial, some retrospective studies and post-hoc analyses in small groups of patients, such as the study by Naidoo et al. (n = 46), found no difference in OS between patients with exon 19/21 mutations and Exon20ins (31 months vs. 26 months, respectively, p = 0.53) [8].
The PFS was significantly lower with first-line TKIs than with chemotherapy in the Exon20ins group. None of our Exon20ins patients who received first-or second-generation EGFR TKIs at any line (n = 19 with evaluable response) showed ORR. At first line, all of those patients (n = 6) showed PD as the best response at first assessment. Among the four Exon20ins patients experiencing an OS ≥ 24 months, one (p.Asn771_Pro772insCysAlaTyr) showed an unusual nine month SD with afatinib. In our study, eight patients received afatinib (n = 3 at first-line treatment and n = 5 at following lines). Among these patients, five showed PD as the best response (including the three first-line patients), two patients showed stable disease and one had a non-evaluable response). These results are consistent with Yang et al.'s combined post-hoc analysis of patients treated with afatinib in LUX-Lung 2, LUX-Lung 3 and LUX-Lung 6 (n = 23 patients, PFS = 2.7 months, and disease control rate = 15%), and other retrospective studies [7][8][9][10][11][12]. These data support the well-known data indicating that early-generation TKIs are not effective in EGFR Exon20ins NSCLC.
None of the Exon20ins patients in our cohort received osimertinib due to the period of study. In vitro and in vivo models showed promising antitumor activity of osimertinib in Exon20ins patients [13]. Pharmacological studies suggest that higher dosing may be required to achieve clinical efficacy in these populations [14]. A recent prospective phase II trial, ECOG-ACRIN EA5162, evaluated a 160 mg daily dosing of osimertinib in previously treated Exon20ins patients. In this trial, Piotrowska et al. reported a DCR of 82% and 9.6 months median PFS [15].
Among Exon20ins patients, PFS appeared to be longer in those treated with chemotherapy + bevacizumab vs. chemotherapy alone ( Figure S2). This is consistent with a very large retrospective Chinese study [16].
In our study, PFS in the control group who received first-line chemotherapy (PFS = 2.5 months) seemed lower than in large phase III trials that showed PFS ranging from 4.6 to 5.8 months [17][18][19][20], probably due to a smaller group (n = 6) and unselected population. For these reasons, we can allow neither further interpretation on PFS, nor a frontal comparison between groups on first-line efficacy based on our study.
Different targeted strategies have been developed. The bi-specific anti-EGFR and anti-MET antibody amivantamab showed exciting results (PFS = 8.3 months and ORR = 40%) with an apparently acceptable safety profile) [27]. Amivantamab received FDA breakthrough therapy designation in March of 2020 and became the first treatment to be granted FDA approval in May of 2021 for patients with locally advanced or metastatic NSCLC with EGFR exon 20 insertion mutations after platinum-based chemotherapy.
In present study, all insertions occurred in the loop following the C-helix, after amino acids at position 767. None of the patients harbored insertions before the 764 amino acid position, which could have conferred sensitivity to approved EGFR TKIs for canonical exon 19 and 21 mutations, in particular the p.Ala763_Tyr764insPheGlnGluAla insertion (A736-Y764insFQEA) [7,18]. Among 35 patients, we collected 18 different Exon20ins variants. We found p.Ala767_Val769dup (n = 6) and p.Ser768_Asp770dup (n = 6) as the more frequent insertion mutations. This data is consistent with the CBio Cancer Genomics Portal database, as presented in Figure 1.
Given the heterogeneity of Exon20ins identified in our study and in previous reports, the efficacy of targeted treatments may vary across different variants even if early trial data did not distinguish a clear correlation between Exon20ins variants and the efficacy of mobocertinib or amivantamab. Further studies of the behavior variability of new targeted therapies against different variants are needed in order to provide better comprehension of acquired resistances after Exon20ins targeted therapies.
New TKIs and specific antibodies are already effective therapies for exon 20 insertion NSCLC patients. The large phase III trials EXCLAIM-2 and PAPILLON will respectively clarify the position of mobocertinib and amivantamab vs. standard chemotherapy. The relevance of immune-checkpoint inhibition alone or in combination with platinum-based chemotherapy is largely unknown, and future trials may enrich treatment options in this population.

Study Design
We retrospectively collected all consecutive patients with EGFR Exon20ins metastatic NSCLC from 11 genetic platforms in France between 2012 and 2017. When available, we paired each Exon20ins patient with 2 metastatic EGFR classical L858R exon 21 mutation or exon 19 deletion NSCLC patients (controls). To ensure consistency of care, we chose controls treated in the same care facility and diagnosed in the same 3-month period. If more than 2 controls met these criteria, we randomly chose 2 of them. Exclusion criteria were localized or locally advanced NSCLC, patients in whom EGFR status was not obtained at a primary biopsy (i.e., re-biopsy at progression after systemic treatment), and the presence of concomitant active neoplasia.
EGFR TKIs or conventional platinum-based chemotherapy as a first-or subsequentline therapy was at the physician's discretion. Disease assessment was based on a CT scan every two or three months and was also at the physician's discretion.
Between 2012 and 2017, EGFR exon 20 insertions were detected from a formalin-fixed paraffin-embedded biopsy by PCR with fluorescent primers and fragment analysis (Sanger). Next-generation sequencing (NGS) was used only at the end of our period of study (2017).

Efficacy Analysis
The main objective of this study was to evaluate the efficacy of first-line EGFR TKIs and conventional chemotherapy in patients with metastatic EGFR Exon20ins NSCLC, defined by the objective response rate (ORR). Tumor response was assessed according to the Response Evaluation Criteria in Solid Tumors Version 1.1 (RECIST 1.1). Tumor responses included a complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD). The disease control rate was defined as the addition of complete responses, partial responses and stable disease (CR + PR + SD). The objective response rate (ORR) was defined as the addition of complete responses and partial responses (CR + PR).
The secondary aims of our study were to determine overall survival (OS) and progressionfree survival (PFS).

Statistical Analysis
OS and PFS were estimated using the Kaplan-Meier method. PFS was calculated from the date treatment started until death or disease progression. OS was calculated from the date of the molecular diagnosis of the EGFR mutation until death or the final follow-up (in patients that were lost to follow-up) or the date of censorship (in living patients). The date of censorship was 1 April 2018. OS and PFS were compared using the log-rank test between the two groups. All statistical analyses were performed using R-studio software version 1.1.419 (The R Foundation for Statistical Computing, Vienna, Austria).

Conclusions
EGFR Exon20ins are heterogeneous genetic alterations conferring resistance to firstand second-generation EGFR TKIs in NSCLC. Patients showed similar clinical characteristics to patients with classic EGFR exon 19/21 mutations but with a tendency to poorer overall prognosis. Platinum-based chemotherapy should be considered as the first-line treatment in these patients until the results of an ongoing trial in exon 20 targeted TKIs become available. Institutional Review Board Statement: Ethical review and approval were waived for this study since no ethics approval is needed for retrospective studies according to French legislation in force at the beginning of the study.
Informed Consent Statement: Patient consent was waived for this study. Non-opposition form for collective retrospective data was provided to all patients. None objected to data collection.

Data Availability Statement:
The datasets analyzed during the current study are available from the corresponding author upon reasonable request.