Characteristics of patients with EGFR-mutant non-small-cell lung cancer who benefited from immune checkpoint inhibitors

Abstract

Objectives

Immune checkpoint inhibitors (ICIs) are less effective in non-small-cell lung cancer (NSCLC) patients with epidermal growth factor receptor (EGFR) mutations. However, a small percentage of patients with EGFR-mutant NSCLC do respond, and the characteristics of these patients are not known. Here, we identify the characteristics of patients who may respond to ICI therapy for EGFR-mutant NSCLC.

Patients and methods

The medical records of NSCLC patients with EGFR mutations who received PD-1/PD-L1 antibody monotherapy at nine institutions were reviewed.

Results

In total, 58 patients with EGFR-mutant NSCLC were analyzed. Various clinical factors such as smoking history and EGFR mutation type were not associated with progression-free survival (PFS) of ICIs, while the PFS of prior EGFR tyrosine kinase inhibitors (TKIs) was inversely associated with that of ICIs. Patients who responded to prior EGFR TKIs for > 10 months exhibited a significantly shorter response to ICIs compared to those who had responded for ≤ 10 months (PFS of ICI: 1.6 vs. 1.9 months; hazard ratio: 2.54; 95% confidence interval 1.26–5.12; p = 0.009). However, patients who responded to ICIs for > 6 months responded to prior EGFR TKIs for significantly shorter periods compared to those who responded to ICIs for ≤ 6 months (PFS of prior EGFR TKI: 5.3 vs. 12.1 months; log-rank test: p = 0.0025).

Conclusion

The duration of response to prior EGFR TKIs could be a predictive marker of ICI therapy in EGFR-mutant NSCLC patients.

Electronic supplementary material

The online version of this article (10.1007/s00262-020-02662-0) contains supplementary material, which is available to authorized users.

Introduction

Immune checkpoint inhibitors (ICIs) such as anti-programmed death-1 (PD-1) and anti-programmed death-ligand 1 (PD-L1) antibodies have been approved for the treatment of various malignant tumors, including non-small-cell lung cancer (NSCLC) [1, 2]. One of the most promising effects of ICIs is a long-term response potentially leading to a cure, and PD-L1 expression is clinically used as a predictive biomarker [3]. Unfortunately, ICIs rarely exert tumor inhibition in NSCLC patients harboring epidermal growth factor receptor (EGFR) mutations when compared to those with wild-type EGFR [4], so the clinical significance of ICIs in EGFR-mutant NSCLC patients remains unclear. However, a recent exploratory analysis of a phase III study revealed that adding ICI to chemotherapy significantly prolonged progression-free survival (PFS) and overall survival (OS) in patients with EGFR-mutant NSCLC [5], implying a potential role for ICIs even in EGFR-mutant NSCLC. In clinical practice, we have found that a small percentage of patients with EGFR-mutant NSCLC respond to ICI therapy. Identification of the characteristics of patients who potentially respond to ICI therapy in EGFR-mutant NSCLC is therefore needed.

In the current study, we investigated the associations between clinical factors and the efficacy of ICIs in patients with EGFR-mutant NSCLC using a multi-institutional, retrospective, cohort approach.

Materials and methods

Patients

This study was conducted using our retrospective database (Okayama Lung Clinical Study Group-Immunotherapy Database; OLCSG-ID), which included the medical records of all NSCLC patients who received PD-1/PD-L1 antibody monotherapy at nine institutions between December 2015 and May 2018. This study was approved by the ethics committee of each institution. Using this database, we found 58 patients who were treated with ICIs for the treatment of EGFR-mutant NSCLC.

Response assessment

Responses were reevaluated for this study by each investigator according to the response evaluation criteria in Solid Tumors (version 1.1).

Statistical analysis

OS was evaluated as the period from the beginning of ICI therapy to the day of death from any cause. PFS was evaluated as the period from the beginning of ICI or EGFR TKI treatment to the day of disease progression or death from any cause using the Kaplan–Meier method. A multivariate analysis was conducted using a Cox proportional hazards model with the following variables: sex, performance status (PS), smoking history, histology, EGFR mutation type, and PFS of prior EGFR TKI treatment. Differences were statistically significant at p < 0.05. All statistical analyses were conducted using STATA software, version 11.0 (Stata Corp., College Station, TX, USA).

Results

Patient characteristics

Fifty-eight patients with NSCLC harboring EGFR mutations were treated with PD-1/PD-L1 inhibitors. The patients’ characteristics are shown in Table 1. The median age was 68 years. The male and female percentages were approximately the same, and half of the patients were non-smokers (29/58). The majority of the patients had a good performance status (PS) (50/58), adenocarcinoma histology (54/58), and a common EGFR mutation type such as the exon 19 deletion and exon 21 L858R (51/58). PD-L1 expression was determined in 19 of the 58 patients. All 58 patients had been treated with EGFR TKIs and the median PFS of EGFR TKIs prior to ICI treatment was 10.3 months [95% confidence interval (CI) 7.7–13.4]. Further details of the treatments prior to ICIs are given in Supplementary Table 1. The median number of regimens prior to ICI therapy was four; most patients had undergone cytotoxic chemotherapy (54/58).

Table 1.

Patient characteristics

	n = 58
Median age, years (range)	68 (37–87)
Sex (male/female)	31/27
Smoking history (yes/no)	29/29
PS (0–1/2–4)	50/8
Histology (ad/sq)	54/4
EGFR mutation type (del/L858R/other)	33/18/7
PD-L1 expression (≥ 50%/< 50%/unknown)	8/11/39
Median PFS to EGFR TKI month (95% confidence interval)	10.3 (7.8–13.4)

EGFR epidermal growth factor receptor, TKI tyrosine kinase inhibitor, PS performance status, advanced incurable stage III/IV, rec postoperative recurrence, sq squamous carcinoma, del exon 19 deletion, L858R exon 21 L858R, PFS progression-free survival, PD-L1 anti-programmed death-ligand 1

Efficacy of ICIs

The median PFS and OS were 1.9 months (95% CI 1.4–2.0; Fig. 1a) and 10.1 months (95% CI 5.9–16.6; Fig. 1b), respectively. The tumor responses to ICIs were determined in 52 of the 58 patients, and the overall response rate (ORR) was 9.6%; the complete response rate was 0/52 (0%), the partial response rate was 5/52 (9.6%), the stable disease rate was 10/52 (19.2%), and the progressive disease rate was 37/52 (71.2%).

Fig. 1.

Progression-free survival (PFS) and overall survival (OS) among the 58 patients. a PFS of all 58 patients. b OS of all 58 patients

Inverse association between the PFS of EGFR TKIs and ICI

Although most of the patients did not benefit from ICI therapy, some did. To determine the characteristics of those who benefited, we examined the associations between clinicopathological factors and the PFS of ICIs (Table 2). In our univariate and multivariate analyses, sex, age, PS, smoking history, histology, and EGFR mutation type were not associated with the PFS of ICIs. However, the PFS of patients with prior EGFR TKI treatment when we used 10 months as a cut-off value (i.e., the median PFS of EGFR TKIs in this cohort; Table 1) was inversely associated with that of ICIs [univariate analysis, hazard ratio (HR) 2.36, 95% CI 1.30–4.29, p = 0.005; multivariate analysis, HR 2.87, 95% CI 1.39–6.00, p = 0.005] (Table 2). Kaplan–Meier plots clearly separated the curves (EGFR TKI > 10 vs. ≤ 10 months; median PFS, 1.6 vs. 1.9 months; log-rank test, p = 0.0025) (Fig. 2a). We found no association with regimen number or a history of cytotoxic chemotherapy prior to ICI therapy, (Supplementary Table 2). We then classified the patients into two groups: the ICI benefit group, in which the patients continued using ICIs without disease progression for more than 6 months, and the ICI non-benefit group, in which the patients’ NSCLC progressed within 6 months (Table 3). We used 6 months as a cut-off value because the 6-month PFS rate has been reported to be associated with OS using ICI therapy [6]. Consistent with the results in Table 2, the median PFS of the prior EGFR TKI was significantly shorter in the ICI benefit group than in the non-benefit group (5.3 vs. 12.1 months; log-rank test; p < 0.001), while there was no significant deviation in sex, PS, smoking history, histology, or EGFR mutation type between the ICI benefit and non-benefit groups (Table 3; Fisher’s exact test). Finally, we sought associations between clinicopathological factors and OS (Supplementary Table 3). Consistent with earlier PFS analysis, the PFS of prior EGFR TKI treatments was inversely associated with OS of ICI, although statistical significance was not attained (HR 2.16, p = 0.062).

Table 2.

Univariate and multivariate analyses of the factors associated with PFS following ICI treatment

	Univariate analysis		Multivariate analysis
	HR (95% CI)	p	HR (95% CI)	p
Sex female vs. male (n = 27 vs. 31)	1.55 (0.89–2.67)	0.12	0.67 (0.25–1.75)	0.41
Age > 70 vs. ≤ 70 (n = 25 vs. 33)	0.84 (0.49–1.45)	0.53	0.68 (0.37–1.23)	0.20
PS 2–4 vs. 0–1 (n = 8 vs. 50)	1.29 (0.60–2.76)	0.51	0.97 (0.43–2.23)	0.95
Smoking history yes vs. no (n = 29 vs 29)	0.61 (0.36–1.06)	0.078	0.53 (0.21–1.32)	0.18
Histology sq vs. ad (n = 4 vs 54)	0.67 (0.27–1.70)	0.40	1.32 (0.46–3.83)	0.60
EGFR mutation uncommon vs. common (n = 7 vs 51)	0.57 (0.23–1.45)	0.24	0.75 (0.27–2.08)	0.58
PFS to prior EGFR TKIs > 10 vs. ≤ 10 months (n = 27 vs. 30)	2.36 (1.30–4.29)	0.005	2.87 (1.39–6.00)	0.005

PFS progression-free survival, ICI immune checkpoint inhibitor, HR hazard ratio, EGFR epidermal growth factor receptor, TKI tyrosine kinase inhibitor, PS performance status, advanced incurable stage III/IV, rec postoperative recurrence, sq squamous carcinoma, common EGFR exon 19 deletion and exon 21 L858R, uncommon EGFR mutations other than common ones

Fig. 2.

Inverse association of immune checkpoint inhibitor (ICI) and prior epidermal growth factor receptor tyrosine kinase inhibitor (EGFR TKI). a Progression-free survival (PFS) of ICIs in patients who responded to EGFR TKIs for ≤ 10 months (solid line) and those who responded for > 10 months (dashed line). b PFS of prior EGFR TKIs in the ICI non-benefit group (solid line) and ICI benefit group (dashed line)

Table 3.

Characteristics of patients who did and did not benefit from ICIs

	Benefit N = 9	No benefit N = 49
Median age, years	65 (46–72)	69 (37–87)
Sex (male/female)	7/2	24/25	p = 0.15
PS (0–1/2–4)	9/0	41/8	p = 0.33
Smoking history (yes/no)	6/3	23/26	p = 0.48
Histology (ad/sq)	8/1	46/3	p = 1.00
EGFR mutation type (common/uncommon)	8/1	44/5	p = 0.30
Median PFS to prior EGFR TKI, months	5.3 (0.5–8.8)	12.1 (8.3–14.2)	p < 0.001

PS performance status, advanced incurable stage III/IV, rec postoperative recurrence, sq squamous carcinoma, EGFR epidermal growth factor receptor, TKI tyrosine kinase inhibitor, common EGFR exon 19 deletion and exon 21 L858R, uncommon EGFR mutations other than common ones

These data imply that ICIs are effective in EGFR-mutant NSCLC patients who show a relatively short response to prior EGFR TKI treatments.

The significance of EGFR mutation type in ICI therapy

We then investigated the potential impact of EGFR mutation type in ICI therapy in more detail. Among the 58 patients, the EGFR exon 19 deletion, L858R mutation, and other mutations were present in 33, 18, and seven patients, respectively (Table 1). There was no significant difference between the EGFR exon 19 deletion and L858R in terms of ORR (14% vs. 0%; Fisher’s exact test; p = 0.29) and PFS (1.8 vs 1.9 months, log-rank test; p = 0.063). Comparing common EGFR mutations (exon 19 deletion and L858R) to uncommon ones, there was also no significant difference in ORR (8% vs. 25%; Fisher’s exact test; p = 0.34) and PFS (1.9 vs 1.9 months; log-rank test; p = 0.22).

The significance of PD-L1 expression in EGFR-mutant NSCLC patients

Although PD-L1 expression was determined in less than one-third of this cohort (19/58), we investigated the significance of PD-L1 expression in EGFR-mutant NSCLC patients. The PFS of ICIs was significantly longer in PD-L1 ≥ 50% NSCLC patients compared with those with PD-L1 < 50% (PD-L1 ≥ 50% vs. < 50%; 6.4 months vs. 1.4 months; HR 6.72; p = 0.007; Fig. 3a), implying that PD-L1 expression is a predictive marker for ICI therapy in EGFR-mutant NSCLC patients. In addition, PD-L1 expression was inversely associated with the PFS of prior EGFR TKI therapy (5.3 vs. 14.2 months; HR 0.15; p = 0.006; Fig. 3b).

Fig. 3.

Impact of anti-programmed death-ligand 1 (PD-L1) expression on epidermal growth factor receptor tyrosine kinase inhibitors (EGFR TKIs) and immune checkpoint inhibitors. a Progression-free survival (PFS) of ICIs in NSCLC with PD-L1 < 50% (solid line) and PD-L1 ≥ 50% (dashed line). b PFS of prior EGFR TKIs in non-small cell lung cancer (NSCLC) patients with PD-L1 < 50% (solid line) and PD-L1 ≥ 50% (dashed line)

Discussion

We investigated the characteristics of EGFR-mutant NSCLC patients who benefited from ICI therapy. The duration of response to prior EGFR TKIs was inversely associated with that to ICIs in these patients.

Although the precise mechanisms of why short responders to EGFR TKIs benefited from ICIs remain unclear, the tumor mutation burden (TMB) might be involved. A history of smoking, which potentially increases the TMB, was associated with a trend of a longer response to ICIs in our cohort (Table 2). TMB is a predictive marker for ICIs; patients with NSCLC with a high TMB respond well to ICIs. However, it was recently reported that TMB was negatively associated with the outcome of EGFR TKI treatment in EGFR-mutant NSCLC patients [7]. Our current finding that a shorter response to prior EGFR TKI treatment was associated with a longer response to ICIs is consistent with these reports.

A retrospective study by Hastings et al. [8] investigating the efficacy of ICIs in EGFR-mutant NSCLC patients showed that L858R had a similar ORR to ICIs when compared with EGFR wild-type patients, while an exon 19 deletion lowered the ORR. The difference between L858R and the exon 19 deletion was not determined for the PFS in that study. Another study reported that patients with an uncommon EGFR mutation had significantly better outcomes when compared to those with common mutations [9]. In the present study, there was no significant difference between exon 19 deletions and L858R mutants or between common mutations and uncommon mutations. These discrepancies are probably because of the relatively small scale of these studies; thus, additional studies are needed.

PD-L1 served as a predictive marker even in patients with EGFR-mutant NSCLC in the current study. Another group reported that EGFR TKI treatment was associated with increased PD-L1 expression [10], indicating that ICIs may be more effective if such therapy is preceded by EGFR TKI therapy in patients with EGFR-mutant NSCLC.

There are some limitations to this study, which was a retrospective study conducted with heterogeneous data regarding patient cohorts and follow-up patterns, implying that the study results are speculative and not definitive. Also, our sample size was small; we lacked the power to detect potentially predictive factors; some subanalyses included fewer than ten patients/group (Table 2). These issues could have biased the present findings, and our results should therefore be interpreted cautiously. Another major limitation is that data from the use of ICI/chemotherapy combinations were not included in this analysis because combination therapy had just been approved in clinical practice.

Although we should consider these limitations when interpreting our study, this is the first report showing a potential inverse association between the efficacy of EGFR TKIs and that of ICIs in EGFR-mutant NSCLC patients. The duration of the response to prior EGFR TKIs and PD-L1 expression could be predictive markers in ICI therapy for EGFR-mutant NSCLC patients.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Conception and design: EI and NH; Data acquisition: EI, DH, KI, TS, SH, DK, SH, NO, NO, and NH; Data analysis and interpretation: EI, KH, YM, and KK; Writing and review of manuscript: EI, DH, KI, TS, SH, DK, SH, NO, NO, NH, KH, YM, and KK.

Compliance with ethical standards

Conflict of interest

EI received honoraria from Chugai Pharmaceutical. EI received additional research funding from MSD. KH received honoraria from Taiho Pharmaceutical and Chugai Pharmaceutical. KH received additional research funding from MSD and Chugai Pharmaceutical. TM received honoraria from Chugai Pharmaceutical and Bristol-Myers Squibb. KK received honoraria from Chugai Pharmaceuticals. All other authors declare no conflict of interest regarding this study.