Abstract

Background

   Currently, survival benefit of immunotherapy in advanced non‐small cell
   lung cancer (NSCLC) with EGFR exon 20 insertions (ex20ins) is
   controversial, though it generally indicates poor response and
   activity. Compared with standard chemotherapy in combination with
   bevacizumab, first‐line chemotherapy plus immune checkpoint inhibitor
   (ICI) in advanced NSCLC with EGFR ex20ins remains elusive and lacks
   real‐world evidence.

Patients and Methods

   A retrospective real‐world study was conducted to evaluate clinical
   outcomes of chemotherapy alone (C), chemotherapy plus ICI (C + I), or
   chemotherapy plus angiogenesis inhibitors (C + A) as first‐line
   strategies for advanced NSCLC patients with EGFR ex20ins.
   Investigator‐assessed response and survival outcomes were compared
   between subgroups. Kyoto Encyclopedia of Genes and Genomes (KEGG)
   analysis was conducted to reveal concomitant alterations and explore
   the molecular landscape of ex20ins.

Results

   A total of 164 patients were screened, identifying 35 kinds of ex20ins,
   and 122 cases treated with C, C + I, and C + A were finally included in
   the first‐line analysis. C + A achieved much better objective response
   rate (ORR, 38.1% vs. 18.2%) and significant progression‐free survival
   (PFS) benefit compared with C (median, 7.73 vs.5.93 months, HR = 0.60,
   95% CI: 0.40–0.90, p = 0.014), and it showed similar ORR (38.1% vs.
   40.0%), but higher disease control rate (DCR, 96.8% vs. 80.0%) and
   numerically longer median PFS (7.73 vs. 6.53 months, HR = 0.83, 95% CI:
   0.44–1.56, p = 0.30) than C + I. There was no PFS difference between
   C + I and C, despite of PD‐L1 expression or tumor mutational burden.
   KEGG analysis revealed concomitant upregulation of PI3K/AKT signaling
   might mediate intrinsic resistance to ICI in ex20ins.

Conclusion

   First‐line chemotherapy plus angiogenesis inhibitors might yield more
   survival benefits than chemotherapy alone for NSCLC with EGFR ex20ins,
   whereas, it suggests that chemotherapy in combination with ICI might
   not obtain a better survival benefit for this subset of patients.
   Activation of PI3K/AKT signaling might mediate intrinsic
   immunosuppression in NSCLC with EGFR ex20ins.

   Keywords: angiogenesis inhibitor, chemotherapy, EGFR, exon 20
   insertion, immune checkpoint inhibitor, non‐small cell lung cancer
     __________________________________________________________________

   (1) Chemotherapy plus ICI as first‐line therapy might not yield a
   better survival benefit than chemotherapy alone or in combination with
   angiogenesis inhibitors in advanced NSCLC with EGFR ex20ins. (2)
   Activation of PI3K/AKT signaling might mediate intrinsic resistance to
   immunotherapy in NSCLC patients with EGFR ex20ins.

   graphic file with name CAM4-12-335-g003.jpg

1. INTRODUCTION

   Tyrosine kinase inhibitor (TKI) for the management of advanced
   non‐small cell lung cancer (NSCLC) with epidermal growth factor
   receptor (EGFR)‐sensitizing mutations has been established in the
   first‐line setting, which has become the treatment paradigm for certain
   EGFR aberrations including exon 19 deletions and exon 21 L858R
   mutations.[44] ^1 , [45]^2 With the development of genomic sequencing,
   more uncommon EGFR mutations have been identified, in which EGFR exon
   20 insertions (ex20ins) constitute a heterogeneous subset of
   EGFR‐activating alterations, ranking third with a prevalence of 4%–10%
   in EGFR‐mutant NSCLC.[46] ^3 , [47]^4 , [48]^5 Almost 90% of ex20ins
   variants occur in the loop region following the C‐helix, generally
   insensitive to clinically approved EGFR TKIs owing to their unique
   conformations and mechanistic differences.[49] ^4 , [50]^6 , [51]^7 ,
   [52]^8 , [53]^9

   In the current international guidelines, platinum‐based chemotherapy
   alone or in combination with the angiogenesis inhibitor (Ai)
   bevacizumab is considered as the standard first‐line option for
   advanced NSCLC with EGFR ex20ins.[54] ^8 , [55]^10 , [56]^11 , [57]^12
   Recently, specifically designed inhibitors against ex20ins including
   EGFR‐MET bispecific monoclonal antibody amivantamab (JNJ‐61186372) and
   novel inhibitor mobocertinib (TAK‐788) showed high selectivity and
   potency, and have been approved by the Food and Drug Administration
   (FDA) of U.S.A. for the treatment of metastatic NSCLC patients with
   EGFR ex20ins failed to prior platinum‐based chemotherapy.[58] ^13 ,
   [59]^14 In addition, poziotinib, CLN‐081 (TAS6417), and DZD9008 as
   other promising inhibitors targeting ex20ins are under ongoing clinical
   trials with favorable activities.[60] ^15 , [61]^16 , [62]^17 However,
   it is currently a dilemma in China for advanced NSCLC patients with
   EGFR ex20ins at the time of disease progression on chemotherapy when no
   validated targeted agents are available, and what's more, chemotherapy
   has already hit a plateau.[63] ^8

   Immunotherapy with programmed cell death‐1 (PD‐1) or programmed cell
   death ligand‐1 (PD‐L1) monoclonal antibodies has displayed glorious
   clinical benefit for wild‐type (WT) advanced NSCLC, but the efficacy of
   PD‐1/PD‐L1 inhibitor was much limited for patients with oncogenic
   drivers including EGFR mutations or ALK fusions.[64] ^18 , [65]^19 To
   date, there is no direct evidence on first‐line immunotherapy for
   advanced NSCLC with EGFR ex20ins, even though the post‐analysis of
   IMpower150 study has indicated clinical benefit of ICI combined with
   chemotherapy and bevacizumab for this subset of patients.[66] ^20

   Although some uncommon EGFR mutations were reported to derive clinical
   benefit from immunotherapy,[67] ^21 , [68]^22 the correlation between
   NSCLC with EGFR ex20ins and its response to ICI was controversial by
   several studies, covering a possible trend that ICI might not be a
   favorable option for ex20ins patients.[69] ^8 , [70]^10 , [71]^12 ,
   [72]^22 , [73]^23 , [74]^24 , [75]^25 , [76]^26 Yet, there lacks
   comprehensive study evaluating the exact efficacy of combination
   therapy with chemotherapy plus Ai or ICI for advanced NSCLC with EGFR
   ex20ins in China. Besides, we acquired little about the heterogeneously
   molecular landscape of EGFR ex20ins in NSCLC. Therefore, we conducted
   this retrospective real‐world study to compare first‐line clinical
   outcomes of chemotherapy alone, chemotherapy plus ICI, or chemotherapy
   plus Ai in advanced NSCLC patients with EGFR ex20ins. We also performed
   bioinformatics analysis to fully investigate the concomitant
   alterations with ex20ins and explore their molecular landscape.

2. PATIENTS AND METHODS

2.1. Study design and patients

   This real‐world study retrospectively reviewed metastatic NSCLC
   patients with EGFR ex20ins at the National Cancer Center/Cancer
   Hospital, Chinese Academy of Medical Sciences between December 2015 and
   September 2021. The inclusion criteria were as follows: (1) Stage IV
   disease at initial diagnosis; (2) ≥18 years of age; (3) histologically
   or cytologically confirmed NSCLC; (4) EGFR ex20ins mutations confirmed
   at initial diagnosis by next‐generation sequencing (NGS) or polymerase
   chain reaction (PCR) with tumor tissues or liquid biopsy samples; (5)
   documented with available data of first‐line therapies in medical
   records. PD‐L1 expression was detected by Dako 22C3 pharmDx test kit.
   The PD‐L1 tumor proportion score (TPS) was calculated as the percentage
   of ≥100 viable tumor cells with complete or partial membrane staining.
   Tumor mutational burden (TMB), defined as the number of somatic, coding
   base substitutions, and short insertions and deletions per megabase of
   genome examined, was assessed by formalin‐fixed, paraffin‐embedded
   tissue samples or blood using NGS method covering various gene panels.
   The NGS testing in this study was performed in institutional
   laboratories or qualified third‐party genetic testing companies as
   documented in medical records, which had acquired the national quality
   system certification in China, and all of the NGS testings were carried
   out based on the Illumina sequencing system. All clinical data were
   extracted from electronic records. This retrospective study was
   approved by the Ethics Committee of National Cancer Center/Cancer
   Hospital, Chinese Academy of Medical Sciences, and written informed
   consent was waived.

2.2. Treatment and assessment

   Patients who met the above inclusion criteria were collected for
   molecular analysis, and those who were treated with different therapy
   patterns of chemotherapy alone (C), chemotherapy plus Ai (C + A), or
   chemotherapy plus ICI (C + I) were included for first‐line efficacy
   analysis. Patients receiving the above treatment options were at a
   standard dose according to clinical guidelines in practice. Baseline
   images of measurable target lesions were obtained with computed
   tomography of the chest and abdomen, and the magnetic resonance imaging
   of brain. Guideline from the Response Evaluation Criteria in Solid
   Tumors (RECIST) version 1.1 was performed to identify a response of
   complete response (CR), partial response (PR), stable disease (SD), or
   progressive disease (PD). Progression‐free survival (PFS) was defined
   as the time from initiation of first‐line therapy to date of documented
   disease progression or death from any cause. The objective response
   rate (ORR) was calculated as the percentage of confirmed CR and PR.
   Disease control rate (DCR) was defined as the percentage of CR, PR, and
   SD. Overall Survival (OS) was the time from initiation of first‐line
   therapy to death. All these endpoints were assessed by investigators
   responsible for this study.

2.3. Statistical analysis

   Statistical analyses were conducted using SPSS version 23.0 (SPSS
   Inc.). Continuous variables were summarized using medians and ranges,
   and categorical variables were described using frequency and
   percentage. Comparison among the subgroups was performed using analysis
   of variance or Chi‐Square test accordingly. PFS and OS were analyzed
   using the Kaplan–Meier method. PFS between different subgroups were
   compared using the log‐rank test (two‐sided), and the corresponding
   hazard ratio (HR) and 95% confidence interval (CI) were estimated using
   the Cox proportional regression model. p < 0.05 was considered
   statistically significant. p values for these analyses are nominal, and
   all are two‐sided. The Kyoto Encyclopedia of Genes and Genomes (KEGG)
   pathway enrichment analysis was performed in R version 4.0.5 with DAVID
   version 6.8 to reveal differentially expressed concomitant mutations
   with EGFR ex20ins.

3. RESULTS

3.1. Patient characteristics

   As the flow chart showed (Figure [77]S1), a total of 164 NSCLC patients
   who met the above inclusion criteria were screened. Excluded for 38
   patients initiated with EGFR TKIs, one patient with chemotherapy plus
   TKI, and 3 patients with ICI single agent or plus Ai, finally, 122
   cases receiving chemo‐based therapies were included for first‐line
   analysis. Among them, 59.8% (n = 73) were females, 69.7% (n = 85) were
   never‐smokers, 20.5% (n = 25), and 7.4% (n = 9) presented with baseline
   central nervous system (CNS) and liver metastases, respectively. Almost
   all of them (n = 116, 95.1%) received genomic testing with tumor
   tissues, and 89.3% (n = 109) were treated with pemetrexed/platinum
   chemotherapy. In subgroup C, most of the patients (86.4%) received
   regimens of pemetrexed /platinum (including cisplatin, carboplatin, and
   nedaplatin), and five patients were treated with paclitaxel/platinum,
   with one of gemcitabine/cisplatin. In subgroup C + A, 58 patients
   (92.1%) received regimens of pemetrexed /platinum plus Ai (52 cases
   with bevacizumab, three cases with VEGFR2‐targeted TKI apatinib, and
   two with recombinant human endostatin). Three patients received
   paclitaxel/platinum plus bevacizumab, and two were treated with
   docetaxel/platinum plus bevacizumab. In subgroup C + I, 13 patients
   (86.7%) were treated with pemetrexed/platinum plus PD‐1 or PD‐L1
   inhibitors, and two patients received nab‐paclitaxel/carboplatin plus
   pembrolizumab. Among 15 patients receiving C + I, except for one case
   with pemetrexed/carboplatin plus PD‐L1 inhibitor atezolizumab, the
   others all received platinum‐based chemotherapy plus PD‐1 inhibitors
   including pembrolizumab (n = 9), sintilimab (n = 3), camrelizumab
   (n = 1), and toripalimab (n = 1). Baseline clinicopathological
   characteristics of EGFR ex20ins patients receiving chemo‐based
   therapies in first‐line setting was demonstrated in Table [78]1.

TABLE 1.

   Clinicopathological characteristics of EGFR exon 20 insertion patients
   in the first‐line setting
   Characteristics C (n = 44) C + A (n = 63) C + I (n = 15) Overall
   (N = 122) p‐value
   Age (years) 54.8 ± 8.41 52.7 ± 10.60 51.4 ± 9.36 53.3 ± 9.73 0.39
   Gender
   Male 17 (38.6%) 26 (41.3%) 6 (40.0%) 49 (40.2%) 0.97
   Female 27 (61.4%) 37 (58.7%) 9 (60.0%) 73 (59.8%)
   Pathology 0.12
   Adenocarcinoma 41 (93.2%) 62 (98.4%) 13 (86.6%) 116 (95.0%)
   Squamous carcinoma 2 (4.5%) 0 (0%) 1 (6.7%) 3 (2.5%)
   Adenosquamous carcinoma 1 (2.3%) 1 (1.6%) 1 (6.7%) 3 (2.5%)
   Smoking history 0.76
   Never 30 (68.2%) 43 (68.3%) 12 (80.0%) 85 (69.7%)
   Current/former 14 (31.8%) 20 (31.7%) 3 (20.0%) 37 (30.3%)
   Brain metastases 0.74
   Absence 36 (81.8%) 50 (79.4%) 11 (73.3%) 97 (79.5%)
   Presence 8 (18.2%) 13 (20.6%) 4 (26.7%) 25 (20.5%)
   Liver metastases 0.79
   Absence 42 (95.5%) 57 (90.5%) 14 (93.3%) 113 (92.6%)
   Presence 2 (4.5%) 6 (9.5%) 1 (6.7%) 9 (7.4%)
   NGS specimen 0.72
   Tumor tissue 41 (93.2%) 60 (95.2%) 15 (100%) 116 (95.1%)
   Plasma 3 (6.8%) 3 (4.8%) 0 (0.0%) 6 (4.9%)
   TP53 mutation 0.06
   None 8 (18.2%) 10 (15.9%) 2 (13.3%) 20 (16.4%)
   Yes 5 (11.4%) 19 (30.2%) 7 (46.7%) 31 (25.4%)
   NA 31 (70.4%) 34 (54.0%) 6 (40.0%) 71 (58.2%)
   PD‐L1 expression[79] ^† 0.02
   Negative 4 (9.1) 9 (14.3%) 2 (13.3%) 15 (12.3%)
   1 ≤ TPS < 50% 4 (9.1%) 4 (6.3%) 6 (40.0%) 14 (11.5%)
   TPS≥50% 2 (4.5%) 2 (3.2%) 1 (6.7%) 5 (4.1%)
   NA 34 (77.3%) 48 (76.2%) 6 (40.0%) 88 (72.1%)
   TMB value (Mb/Muts) 0.86
   <10 4 (9.1%) 9 (14.3%) 1 (6.7%) 14 (11.5%)
   ≥10 2 (4.5%) 2 (3.2%) 0 (0.0%) 4 (3.3%)
   NA 38 (86.4%) 52 (82.5%) 14 (93.3%) 104 (85.2%)
   Chemotherapy regimens 0.16
   Platinum/pemetrexed 38 (86.4%) 57 (90.5%) 10 (66.7%) 105 (86.1%)
   Other regimens 6 (13.6%) 6 (9.5%) 5 (33.3%) 17 (13.9%)
   [80]Open in a new tab

   Abbreviations: C, chemotherapy alone; C + A, chemotherapy plus
   angiogenesis inhibitors; C + I, chemotherapy plus immune checkpoint
   inhibitors; NA, not available; TMB, tumor mutational burden; TPS, tumor
   proportion score.
   ^^†

   There were no differences among the groups except PD‐L1 expression
   (p = 0.02).

3.2. Molecular subtypes of exon 20 insertions

   Among 161 patients who had received NGS testing for ex20ins detection
   at initial diagnosis, it identified 35 kinds of distinct ex20ins
   subtypes, in which V769_D770insASV (A767_V769dup) was the most common
   variant (n = 38, 23.6%), followed by D770_N771insSVD (S768_D770dup,
   n = 25, 15.5%), D770delinsGY (n = 14, 8.7%), A763_Y764insFQEA (n = 12,
   7.5%), H773_V774insNPH (N771_H773dup, n = 8, 5.0%), P772_H773insH
   (H773dup, n = 6, 3.7%), V774_C775insHV (H773_V774dup, n = 4, 2.5%), and
   other subtypes (n = 33, 20.5%). In addition, 3 patients (1.8%) were
   identified to be ex20ins positive by PCR assay, with uncertain
   insertion subtypes (Figure [81]1).

FIGURE 1.

   FIGURE 1
   [82]Open in a new tab

   Molecular subtypes of NSCLC patients with distinct EGFR exon 20
   insertions

3.3. Efficacy

   The cutoff time of this study was September 30, 2021, with a median
   follow‐up of 32.40 months. One‐hundred and twenty‐two patients were
   available for first‐line efficacy analysis, and 107 had PFS events. The
   ORR, DCR, median PFS, and median OS of C were 18.2% (8/44), 84.1%
   (37/44), 5.93 months (95% CI: 2.70–9.17), and 32.03 months (95% CI:
   17.55–46.52), respectively. Sixty‐three patients (51.6%) were treated
   with C + A, and the ORR, DCR, median PFS, and median OS were 38.1%
   (24/63), 96.8% (61/63), 7.73 months (95% CI: 6.40–9.06), and
   30.57 months (95% CI: 19.90–41.23), respectively. Fifteen patients
   (12.3%) administered C + I, and the ORR, DCR, and median PFS were 40.0%
   (6/15), 80.0% (12/15), and 6.53 months (95% CI: 5.06–8.01),
   respectively. At the cutoff time, the median OS of C + I was immature
   because of this treatment pattern was used in more recent times with
   only three events of deaths, but not for better efficacy. The PFS by
   the above three treatment patterns was listed in Figure [83]2.

FIGURE 2.

   FIGURE 2
   [84]Open in a new tab

   Kaplan–Meier curves of PFS in patients harboring EGFR ex20ins with
   chemotherapy alone (C) versus chemotherapy plus angiogenesis inhibitors
   (C + A) versus chemotherapy plus ICI (C + I) as first‐line therapies

   It was observed significant PFS difference between C + A and C (median,
   7.73 vs. 5.93 months, HR = 0.60, 95% CI: 0.40–0.90, p = 0.014). No
   difference in PFS was observed between C + I and C (median, 6.53
   vs.5.93 months, HR = 0.73, 95% CI: 0.38–1.39, p = 0.48), or C + A and
   C + I (median, 7.73 vs.6.53 months, HR = 0.83, 95% CI: 0.44–1.56,
   p = 0.30). Besides, we also analyzed the impact of metastatic organs on
   efficacy. However, the patients who were diagnosed with baseline CNS
   metastases did not identify an inferior PFS outcome by first‐line
   chemo‐based therapies (median, 7.20 vs. 6.73 months, HR = 0.90, 95% CI:
   0.57–1.44, p = 0.66). And in the patients with baseline liver
   metastases, we observed similar PFS trend (median, 6.50 vs.
   6.90 months, HR = 1.50, 95% CI: 0.72–3.09, p = 0.27).

3.4. PD‐L1 and TMB as biomarkers for efficacy evaluation of ICI

   Among 15 patients receiving C + I as first‐line treatment and available
   for efficacy evaluation, nine cases had molecular data of PD‐L1
   expression. For two patients with TPS negative, the PFS was 6.53 and
   7.83 months. Among another six patients who carried 1 ≤ TPS < 50%, the
   ORR was 33.3% (2/6), with a median PFS of 6.57 months (95% CI:
   0.53–12.61). There was only one patient with TPS of 60%, and he
   achieved PR with a PFS of 5.40 months. For five patients who detected
   TMB values before first‐line therapy, one with TMB of 2.10 Muts/Mb
   showed SD to PD‐1 inhibitor with an ongoing PFS of 6.53 months. The
   other four patients all presented TMB < 10, and the PFS were 10.27,
   4.10, 6.03, and 2.43 months, respectively. Their responses to C + I
   were SD, SD, SD, and PR, respectively.

3.5. Genomic analysis for concomitant alterations with exon 20 insertions

   The UpSet plot was constructed for the interactive sets between the top
   10 concomitant genes together with ex20ins among 73 patients (45.3%) at
   initial diagnosis (Figure [85]3A). The most frequent co‐mutation
   identified in this study was TP53 aberration (n = 38, 52.1%), followed
   by EGFR amplification (n = 26, 35.6%), mutations of PIK3CA (n = 8,
   11.0%) and RB1 (n = 6, 8.2%), BIM deletion polymorphism (n = 6, 8.2%),
   alterations of PTEN (n = 4, 5.5%) and CREBBP (n = 4, 5.5%).

FIGURE 3.

   FIGURE 3
   [86]Open in a new tab

   The top 10 concomitant genes together with EGFR ex20ins in NSCLC
   patients detected at primary diagnosis in the UpSet plot (A). KEGG
   analysis of concomitant signaling pathways detected in patients with
   EGFR ex20ins (B). Possible mechanism for immunosuppression in NSCLC
   with EGFR ex20ins mediated by activation of PI3K/AKT signaling (C)

   In addition, a total of 112 genes were detected in ex20ins patients and
   involvement of these genes were overlaid on the 20 significant KEGG
   pathways (Figure [87]3B). The genes were significantly enriched in
   cancer‐related pathways, PI3K/AKT, microRNAs, Rap1 signaling pathway,
   proteoglycans, cell cycle genes, and so on.

4. DISCUSSION

   This study provided comprehensive real‐world evidence regarding
   clinical efficacy of chemo‐based combination therapies for advanced
   NSCLC with EGFR ex20ins alterations in the first‐line setting in China.
   We examined heterogeneous ex20ins subtypes and revealed full‐scale
   concomitant alterations with EGFR ex20ins.

   Notably, this is the first Chinese study to directly compare the
   clinical outcomes of ICI or Ai in combination with chemotherapy as
   first‐line strategies for advanced NSCLC with EGFR ex20ins. We observed
   a much better ORR (38.1% vs. 18.2%) and significant PFS benefit
   (median, 7.73 vs. 5.93 months, HR = 0.60, p = 0.014) of C + A compared
   with chemotherapy alone, which was much similar to that reported in our
   previous nationwide study, indicating a median PFS of 7.5 and
   5.6 months respectively for NSCLC patients with EGFR ex20ins.[88] ^8 In
   addition, C + A showed similar ORR (38.1% vs. 40.0%), but achieving a
   higher disease control rate (DCR, 96.8% vs. 80.0%) and numerically
   longer PFS (median, 7.73 vs. 6.53 months, HR = 0.83, 95% CI: 0.44–1.56,
   p = 0.30) compared with C + I. There was no PFS difference on C + I
   compared with C or C + A in first‐line setting. Similarly, series of
   studies reported worldwide also issued that ICI plus chemotherapy
   failed to improve survival benefits for ex20ins patients compared with
   conventional chemotherapy,[89] ^10 , [90]^12 , [91]^23 , [92]^25 ,
   [93]^28 with an overall ORR of 6.7%–33.3%, and median PFS of
   2.0–7.0 months (Table [94]2). Especially, ICI monotherapy was reported
   with much limited activity for ex20ins, with an overall ORR of
   3.1%–50%, and PFS of 1.9–4.8 months.[95] ^10 , [96]^12 , [97]^22 ,
   [98]^24 , [99]^26 , [100]^27 , [101]^29 , [102]^30 , [103]^31 ,
   [104]^32

TABLE 2.

   Reported studies of ICI for advanced NSCLC with EGFR exon 20 insertion
   Author Year Sample size[105]^* Therapeutics ORR (%) DCR (%) Median PFS
   (months) Median OS (months)
   Negrao et al.[106] ^26 2018 36 (36) ICIs 25 50 2.9 NR
   Takeda et al.[107] ^29 2018 5 (12) PD‐1 inhibitor 20 40 NA NA
   Hastings et al. [108]^22 2019 28 (28) PD‐1/PD‐L1 ± CTLA‐4 inhibitors
   10.7 32.1 1.9 5.5
   Dersarkissian et al.[109] ^31 2019 29 (199) ICIs NA NA NA NA (range,
   6.1–8.0)
   Yang et al.[110] ^8 2020 10 (165) ICIs ± PbC or Ai 50 80 NA (range,
   1.5–8.7) NA
   Patil et al.[111] ^28 2020 3 (38) ICIs ± PbC 33.3 NA NA NA
   Tomaras et al.[112] ^32 2020 30 (NA) ICIs 10.0 NA 3.2 7.5
   Choudhury et al.[113] ^10 2021 15 (59) PD‐1 inhibitor NA NA 2.8 NA
   12 (59) PD‐1 inhibitor + PbC NA NA 7.0 NA
   Ou et al.[114] ^12 2021 11 (237) ICIs 9.1 NA 3.1 11.1
   16 (237) ICIs + PbC 18.8 NA 4.5 11.3
   32 (237) ICIs 3.1 NA 2.3 8.1
   Metro et al.[115] ^23 2021 15 (30) ICIs ± PbC 6.7 13.4 2.0 5.3
   Chen et al.[116] ^24 2021 9 (35) ICIs 22.2 NA 4.0 23.3
   Geng et al.[117] ^25 2021 12 (283) PD‐1 inhibitor ± PbC or Ai 16.7 83.3
   NA (range, 1.5–5.0) NA
   Lau et al.[118] ^27 2021 6 (6) PD‐1/PD‐L1 inhibitors 50 67 4.8 NA
   Morita et al.[119] ^30 2021 8 (23) PD‐1 inhibitor 25 50 3.1 NA
   [120]Open in a new tab

   Abbreviations: Ai, angiogenesis inhibitors; CTLA‐4, cytotoxic T
   lymphocyte‐associated antigen‐4; DCR, disease control rate; NA, not
   available; NR, not reached; ICIs, immune checkpoint inhibitors; ORR,
   overall response rate; OS, overall survival; PbC, platinum‐based
   chemotherapy; PD‐1, programmed cell death protein 1; PD‐L1, programmed
   cell death ligand 1; PFS, progression‐free survival.
   ^*

   Expressed as patient number with EGFR ex20ins treated with
   immunotherapy (total patient number with EGFR ex20ins).

   Our study identified and reported 35 different kinds of heterogeneous
   ex20ins subtypes, in which V769_D770insASV and D770_N771insSVD were the
   most common insertion variants, overall representing 40% of all ex20ins
   subtypes. This finding was in accordance with two previous large sample
   Chinese studies,[121] ^8 , [122]^25 and also was similar to that
   discovered in the American databases.[123] ^33 , [124]^34 Similarly, we
   also found TP53 occurred as the most frequent concomitant alteration
   together with EGFR ex20ins, with a prevalence of 52.1% in this study,
   which was much alike with those of 35%–56% reported by other
   studies.[125] ^10 , [126]^11 , [127]^23 , [128]^24 , [129]^25 ,
   [130]^33 , [131]^35 As clarified by current reports, EGFR ex20ins was
   associated with much lower levels of TMB (2.8–3.6/Muts/Mb) than WT
   EGFR, as well as with a low TPS of PD‐L1 expression, which suggested
   lack of comparable benefit of ICI in EGFR ex20ins as observed in NSCLC
   harboring common EGFR mutations.[132] ^10 , [133]^21 , [134]^22 ,
   [135]^33 In our study, only 2 patients carrying 1 ≤ TPS˂50% showed
   response to C + I (ORR, 2/6), with a median PFS of 6.57 months. Five
   patients available for TMB assessment all harbored low TMB values less
   than 10 Muts/Mb, and showed ORR of 1/5 to C + I, with PFS between 2.43
   and 10.27 months. Outcomes of C + I observed in our study were well
   correlated with previous observations reported worldwide, indicating a
   limited activity of ICI for NSCLC patients with EGFR ex20ins.[136] ^8 ,
   [137]^10 , [138]^12 , [139]^22 , [140]^23 , [141]^24 , [142]^25 ,
   [143]^26 , [144]^27 , [145]^28 , [146]^29 , [147]^30 , [148]^31 ,
   [149]^32

   To date, it lacks in vitro and molecular exploration for the possible
   mechanism of unfavorable activity of ICI for EGFR ex20ins. In this
   study, we observed concomitant genes with ex20ins significantly
   enriched in PI3K/AKT, microRNAs in cancer, and Rap1 signaling pathways
   by KEGG analysis, revealing the upregulation of PI3K/AKT signaling was
   much more common and important in NSCLC with EGFR ex20ins, which had
   been reported in HER2‐positive breast cancer.[150] ^36 Recent studies
   have confirmed that microRNAs are involved in tumor proliferation and
   chemotherapy resistance in NSCLC.[151] ^37 , [152]^38 The dysregulation
   of microRNAs is closely related to the remodeling of tumors and the
   immune microenvironment.[153] ^39 EGFR has a regulatory role in
   microRNAs maturation through phosphorylation of AGO2 with HIF‐1α gene
   expression.[154] ^40 , [155]^41 At the meantime, TP53 can modulate cell
   survival and proliferation by interacting with microRNAs.[156] ^38 Rap1
   plays an important role in the cell behaviors by regulating the
   function of integrins and other adhesion molecules in different cell
   types, with promoting AKT phosphorylation.[157] ^42

   Based on the annotation of the KEGG website ([158]http://www.kegg.jp)
   and pathway enrichment analysis in this study, we supposed that
   together with the activation of Rap1 and the loss of PTEN could enhance
   PI3K/AKT signaling and ultimately suppress the activation of FOXO
   proteins, thus mediating drug resistance, including poor response to
   ICI in EGFR ex20ins. In terms of possible mechanisms explaining
   immunosuppressive microenvironment in NSCLC with EGFR ex20ins, we
   explained that ex20ins could activate the downstream MAPK and PI3K/AKT
   signaling pathways, resulting in AKT1 recruitment and impairing
   TANK‐binding kinase 1 (TBK1) phosphorylation, and suppress interferon
   regulatory factor 3 (IRF3), thus disrupting the stimulator of
   interferon gene (STING) signaling and suppressed type I and II
   interferon response and antitumor immune responses (Figure [159]3C),
   which had been currently reported by in vitro researches.[160] ^43 ,
   [161]^44

   This study has several limitations. First, it inevitably exists
   selective bias due to the retrospective nature. In addition, although
   this study provides comprehensive subtypes of EGFR ex20ins with
   detailed concomitant alterations in several signaling pathways, less
   than half of the patients received large panels of NGS testing, with
   not adequate data for molecular analysis. Further, although we
   demonstrated ICI in combination with chemotherapy might not improve
   survival outcomes over chemotherapy alone or plus angiogenesis
   inhibitors in NSCLC with EGFR ex20ins, corresponding results should be
   interpreted with caution due to the small sample size receiving
   immunotherapy in this study. As well, the lack of in vitro evidence on
   tumor immunohistochemistry or flow cytometry data of PI3K/AKT pathway
   and other immunosuppressive genes among EGFR ex20ins patients were
   considered another limitation.

   In conclusion, chemotherapy plus ICI as first‐line therapy might not
   yield a better survival benefit than chemotherapy alone or in
   combination with angiogenesis inhibitors in advanced NSCLC with EGFR
   ex20ins. PI3K/AKT signaling might mediate intrinsic resistance to
   immunotherapy in this subset of patients. New therapeutics targeting
   the immunosuppressive microenvironment in NSCLC with EGFR ex20ins need
   to be explored.

AUTHOR CONTRIBUTIONS

   Guangjian Yang, Yaning Yang, and Runze Liu: conceptualization,
   methodology, software, formal analysis, project administration, data
   curation, writing ‐ original draft, writing ‐ review & editing. Weihua
   Li, Haiyan Xu, Xuezhi Hao, and Junling Li: validation, resources,
   visualization. Shuyang Zhang, Fei Xu, and Siyu Lei: investigation. Yan
   Wang: conceptualization, resources, supervision, writing ‐ review &
   editing.

CONFLICT OF INTEREST

   The authors declare no conflict of interest.

ETHICS STATEMENT

   As an observational study, this research was exempted from obtaining
   patients' informed consent without therapeutic intervention, and this
   research was approved by the Research Ethics Board of Cancer Hospital
   and conducted in accordance with the Declaration of Helsinki as well.

Supporting information

   Figure S1
   [162]Click here for additional data file.^ (103.7KB, jpg)

ACKNOWLEDGMENTS