Introduction

Epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) are the standard of care for patients with advanced non-small cell lung cancer (NSCLC) harboring sensitizing EGFR mutations; however, almost all patients inevitably develop acquired resistance to EGFR-TKIs.1 Immune checkpoint inhibitor (anti-PD-L1/PD-1) monotherapy was not found to have superior survival compared with chemotherapy in patients with previously treated EGFR-mutant NSCLC.2 Immunotherapy-based combination regimens have been explored to improve clinical benefits in EGFR-TKI-resistant patients with NSCLC. However, the sustained benefits of immunotherapy-based regimens have been debated. Neither Checkmate-722 nor Keynote-789 reached the prespecified statistical level of progression-free survival (PFS) or overall survival (OS) when adding immunotherapy to chemotherapy. It has been proposed that bevacizumab enhances the T cell-mediated cancer cell-killing action of immunotherapy by reversing VEGF-mediated immunosuppression.3 Orient-31 and ATTLAS trials aimed to explore the efficacy of immunotherapy plus bevacizumab with chemotherapy in TKI-resistant, EGFR-mutant, metastatic NSCLC.4 5 However, randomized controlled trials (RCTs) that directly compare the efficacy of immunotherapy plus bevacizumab with chemotherapy versus immunotherapy plus chemotherapy in EGFR-TKI-resistant NSCLC are lacking. We conducted an indirect comparative meta-analysis using chemotherapy as a common comparator to classify the relative efficacy of these two immunotherapy-based regimens. In addition, the critical question regarding who would benefit from an immunotherapy-based regimen remains unclear. Therefore, this meta-analysis provides insight into the aforementioned question and clinical decision-making for immunotherapy-based regimen selection in TKI-resistant EGFR-mutant NSCLC.

Materials and methodsSearch strategy

This meta-analysis was conducted using a predetermined protocol following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension statement for network meta-analysis. A systematic literature search of electronic databases, including PubMed, EMBASE, and the Cochrane Central Register of Controlled Trials (until October 20, 2023), was performed. Meeting abstracts from ASCO, ESMO, AACR, WCLC were also reviewed. This prospective protocol was created in advance and registered in the Prospective Register of Systematic Reviews (ID: CRD42023492329).

Study selection and data extraction

Studies that met the following criteria were included (1) trials that compared two or more different regimens for patients with EGFR-mutated NSCLC who progressed after EGFR-TKI treatment; (2) trials that had at least one arm using an immune-chemotherapy strategy; and (3) trials that reported data on PFS or OS. Studies not adhering to the inclusion criteria were excluded, and other exclusion criteria were as follows: (1) trials in which the control group was not standard-of-care chemotherapy and (2) trials that contained insufficient data for meta-analysis. The primary variables of interest were HRs with 95% CIs for PFS or OS. The methodological quality and risk of bias of individual studies were evaluated using the tool recommended by the Cochrane Handbook for Systematic Reviews of Interventions (online supplemental figure 1).

Statistical analysis

Direct and indirect evidence were synthesized to compare different treatments in terms of PFS and OS. The clinical benefit of immunotherapy plus chemotherapy (IO+Chemo) or immunotherapy plus bevacizumab with chemotherapy (IO+Bev+Chemo) was compared with chemotherapy alone (Chemo alone). Furthermore, an indirect comparison of IO+Chemo and IO+Bev+Chemo was conducted using Bayesian network meta-analysis. A subgroup meta-analysis was also performed on patients receiving an immunotherapy-based regimen and controls. Furthermore, the trial-specific HRs for indirect comparison (19del vs L858R, with vs without T790M, and one-line vs two-line previous TKI treatment) were calculated according to a previously reported method,6 which were also pooled in this meta-analysis. Heterogeneity across included trials and publication bias were also assessed (online supplemental table 1). All analyses were performed using Comprehensive meta-analysis software V.3.3 (Biostat, Englewood Cliffs, New Jersey, USA). The significance level was set at α=0.05 for statistical tests.

Results

Four RCTs (ATTLAS, ORIENT-31, KEYNOTE-789, and CheckMate-722) involving 1474 patients with EGFR mutations were deemed eligible for inclusion after detailed evaluation (online supplemental figure 2). ORIENT-31, KEYNOTE-789, and CheckMate-722 compared immunotherapy plus chemotherapy to chemotherapy alone. The ATTLAS and ORIENT-31 trials compared immunotherapy, bevacizumab plus chemotherapy, to chemotherapy alone (table 1). IMpower150 and IMpower151 were excluded because EGFR-TKI treatment-naïve NSCLC patients were enrolled.

Table 1

Characteristics of patients in the included trials

Direct and indirect comparison analysis

Compared with chemotherapy alone, immunotherapy plus chemotherapy reduced the risk of progression (HRIO+Chemo/Chemo=0.77, 95% CI 0.67 to 0.88), and death (HRIO+Chemo/Chemo=0.86, 95% CI 0.75 to 0.99), among patients with TKI-resistant, EGFR-mutated NSCLC (figure 1A). Compared with chemotherapy alone, immunotherapy and bevacizumab plus chemotherapy reduced the risk of progression (HRIO+Bev+Chemo/Chemo=0.54, 95% CI 0.44 to 0.67), but not death (HRIO+Bev+Chemo/Chemo=0.98, 95% CI 0.77 to 1.25) (figure 1A).

Figure 1

Direct and indirect comparisons among IO+Bev+Chemo, IO+Chemo, and chemo alone (A). Forest plot of HR directly comparing PFS between immunotherapy-based regimen (IO+Bev+Chemo and IO+Chemo) with chemo alone (B). Indirect comparison analysis of HR comparing PFS from immunotherapy-based regimen between different characteristics (EGFR mutation type: L858R vs19del, T790M: negative vs positive, previous TKI treatment: one line vs two lines) (C). Bev, bevacizumab; Chem, Chemotherapy; EGFR, epidermal growth factor receptor; IO, immunotherapy; PFS, progression-free survival; TKI, tyrosine kinase inhibitor.

The indirect comparison showed that immunotherapy and bevacizumab plus chemotherapy had a significantly better PFS (HRIO+Bev+Chemo/IO+Chemo=0.71, 95% CI 0.55 to 0.91) than that of immunotherapy plus chemotherapy (figure 1A). There was no significant difference in OS between patients who received these two immunotherapy-based regimens (HRIO+Bev+Chemo/IO+Chemo=0.88, 95% CI 0.66 to 1.16).

Stratified analysis for immunotherapy-based regimen versus chemotherapy alone

We conducted a subgroup analysis to identify patients who could benefit from an immunotherapy-based regimen (figure 1B). The results showed that patients harboring EGFR L858R have significantly longer PFS (HR 0.44, 95% CI 0.34 to 0.57) from immunotherapy-based regimen, but not those harboring EGFR 19del (HR 0.86, 95% CI 0.69 to 1.07). Also, patients without T790M mutation (HR 0.49, 95% CI 0.41 to 0.60) and patients who only received previous one-line TKI treatment (HR 0.56, 95% CI 0.48 to 0.66) could obtain more benefit from the immunotherapy-based regimen. Conversely, patients with T790M mutation or who previously received two-line TKI treatment did not have significantly better PFS from immunotherapy-based regimen than chemotherapy alone.

To further determine the influence of EGFR mutation type, T790M mutation, and previous TKI treatment on the PFS of immunotherapy-based treatments, an indirect comparison analysis (19del vs L858R, with vs without T790M, and one-line vs two-line previous TKI treatment) was conducted (figure 1C). Compared with their counterparts, patients with L858R (HR 0.52, 95% CI 0.37 to 0.72), without T790M mutation (HR 0.50, 95% CI 0.35 to 0.71), and receiving one-line TKI treatment (HR 0.47, 95% CI 0.34 to 0.65) tended to obtain significantly more benefit from immunotherapy-based regimens.

Discussion

Among patients with NSCLC without driver mutations, immunotherapy-based treatment has demonstrated superior survival to that of chemotherapy alone.7 Whether this efficacy is in line with that of patients with EGFR-mutated NSCLC who progressed to EGFR-TKI remains unclear. There is a great need to answer two questions: (1) Can adding a VEGF inhibitor to immunotherapy further improve efficacy in this patient population? and (2) Which patients would benefit from immunotherapy combination treatment?

Although not clearly understood, the low tumor mutation burden, low PD-L1 expression, and decreased density of tumor-infiltrating lymphocytes, which might be related to the relatively high VEGF expression in EGFR-mutated NSCLC, may explain why patients with EGFR-mutated NSCLC respond poorly to immunotherapy.8 Thus, it was suggested that combining a VEGF inhibitor with PD-1/L1 antibody may reprogram the immunosuppressive microenvironment into an immunostimulatory microenvironment, further promoting antitumor immunity. Owing to the lack of head-to-head comparison analyses, the effect of the addition of VEGF inhibitors is not well defined. This is the first pooled analysis to compare the efficacy of immunotherapy plus chemotherapy with and without VEGF inhibitors in TKI-resistant EGFR-mutated NSCLC. Pooled analysis showed that the addition of a VEGF inhibitor significantly prolonged PFS compared with immunotherapy plus chemotherapy. In addition, the safety of immunotherapy-based regimen in EGFR-mutated NSCLC patients warrant caution, especially concerning the interval between immunotherapy and EGFR-TKI use.

Another issue concerns efficacy-associated factors that could be used to determine which patients would benefit from immunotherapy-based treatment for TKI-resistant EGFR-mutant NSCLC, as there remains a great unmet need for this patient population. Biomarker analysis revealed a significant difference in PFS stratified by EGFR mutation type, suggesting a greater benefit from immunotherapy-based regimens in patients with EGFR L858R than 19del. This might be explained by more CD8+PD-1+T cells infiltrating the L858R-mutant tumor.9 Given the poor clinical outcomes associated with L858R relative to 19del, this combinatorial approach may be a reasonable treatment option for patients with L858R mutation after EGFR-TKI failure. Furthermore, several preclinical studies have reported higher PD-L1 expression in T790M-negative individuals than T790M-positive individuals after EGFR-TKI failure.10 Consistent with these findings, our data showed that patients with NSCLC without T790M mutation benefited more from immunotherapy-based regimens than those with T790M mutation. Correspondingly, patients receiving previous two-line EGFR-TKIs tend to benefit less from immunotherapy-based regimens. In addition, the stratified analysis also explored the effect of other factors on clinical benefit from immunotherapy-based regimen. For example, previous evidence demonstrated that increased smoking exposure had a significant association with improved clinical benefit among advanced NSCLC treated with immunotherapy independent of PD-L1 expression.11 Our analysis showed these patients could obtain significant benefits from immunotherapy-based regimen than chemotherapy, whether smoker or non-smoker. These findings would offer the implications for clinical trial design.

Obviously, these results should be interpreted with caution for several reasons. First, the sample size limits the strength of the study. We could not conduct a subgroup analysis to identify which patients would benefit more from IO+Bev+Chemo than IO+Chemo due to inadequate data. Alternatively, stratified analysis was performed to identify those who could benefit more from immunotherapy-based regimens (IO+Bev+Chemo and IO+Chemo) than chemotherapy alone. Although there was no significant heterogeneity across prospective studies with more than 1400 cases, a small sample size with imbalanced baseline characteristics may affect the accuracy and reliability of these conclusions. Second, the data on who could benefit from immunotherapy-based regimens were mostly obtained from the subgroup analysis of included RCTs. Additional RCTs specifically addressing this issue are required to validate our conclusions. Third, owing to the design of this study, our analysis was limited to assessing the impact of documented factors on clinical benefit within the included trials. Consequently, the effects of additional factors on the efficacy of immunotherapy-based regimens were unable to ascertain. However, as such a prospective study has not been conducted recently, the present meta-analysis meets the current clinical needs and may inform future investigations.

In conclusion, the present study is the first comparative analysis to identify the optimal immunotherapy-based regimen for patients with TKI-resistant EGFR-mutant NSCLC. Our findings suggest that the addition of a VEGF inhibitors to immunotherapy and chemotherapy is the preferred treatment for this patient population. In addition, patients with L858R mutation, who are T790M-negative, may derive more clinical benefits from immunotherapy-based regimens.

Ethics statementsPatient consent for publication

Not applicable.

Acknowledgments

We thank the patients for their participation in clinical trial, and the investigators for releasing the clinical data.