Immunotherapy for patients with EGFR-TKI-resistant non-small-cell lung cancer: Potential mechanisms, efficacy predictors, and therapeutic integration.

Introduction
Lung cancer is the leading cause of cancer-related mortality globally.1 Non-small-cell lung cancer (NSCLC) constitutes more than 85% of all lung cancers.2 The histological types of NSCLC mainly include adenocarcinoma, squamous cell carcinoma, and large cell carcinoma.3 Among the most prevalent druggable alterations in NSCLC are mutations in the epidermal growth factor receptor (EGFR). These mutations, occurring almost exclusively in lung adenocarcinoma, are detected in approximately 50% of NSCLC patients from East Asia.4
The introduction of EGFR tyrosine kinase inhibitors (EGFR-TKIs) has profoundly extended the survival outcomes for patients with advanced NSCLC positive for EGFR mutations. Three generations of EGFR-TKIs have been evaluated in clinical trials. First-generation TKIs, such as gefitinib and erlotinib, exert anti-tumor effects through reversible competitive inhibition of ATP binding to the tyrosine kinase domain in EGFR.5 Compared with platinum-based chemotherapy in treatment-naïve advanced EGFR-mutant NSCLC, these TKIs have demonstrated marked improvements in progression-free survival (PFS), ranging from 9.5 to 11.0 months.6,7 Second-generation TKIs, such as dacomitinib and afatinib, function as irreversible multitargeted inhibitors of the erythroblastic leukemia viral oncogene homolog (ERBB) receptor family,5 exhibiting a PFS of 11.1 to 14.7 months in this patient population.8,9 Third-generation EGFR-TKIs, such as furmonertinib and osimertinib, impede tyrosine kinase activity and demonstrate efficacy against tumors positive for the EGFR T790M mutation, with median PFS reaching 17.8–20.8 months.10, 11, 12
Despite these therapeutic advances, acquired resistance to EGFR-TKIs inevitably emerges, resulting in disease progression.13 Both mechanisms of EGFR-dependent and EGFR-independent resistance have been identified. The EGFR T790M mutation represents the predominant EGFR-dependent resistance mechanism to first- and second-generation TKIs, occurring in approximately 55% of patients,14 while the most acquired mutation mediating EGFR-dependent resistance mechanisms to third-generation TKIs is EGFR C797S mutation, accounting for 22–40%.15 The EGFR-independent resistance mechanisms mainly include bypass activation (such as HER2/MET amplification), oncogenic fusions (such as BRAF and RET), histological transformation and other acquired mutations (such as PIK3CA mutation).16 Notably, the mechanisms underlying resistance remain unidentified in a substantial subset of patients.
Currently, effective treatment options for EGFR-mutated NSCLC patients who progress on EGFR-TKI therapy remain limited. In contrast, for patients with advanced or metastatic NSCLC lacking driver gene mutations, the therapeutic landscape has expanded considerably over the past decade, particularly with the advent of immune checkpoint inhibitors (ICIs).17 By inhibiting the interaction between programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1), ICIs enhance the ability of infiltrating T cells to recognize tumor cells.18 Several studies have suggested potential clinical benefits from immunotherapy in patients with EGFR mutations.19,20 However, others have reported more discouraging outcomes.21,22 The efficacy of immunotherapy largely relies on the interplay between tumor cells and the surrounding tumor microenvironment (TME). TME is a dynamic ecosystem comprising diverse immune cell types (such as tumor-infiltrating lymphocytes [TILs], tumor-associated macrophages [TAMs]), cancer-associated fibroblasts (CAFs), endothelial cells, extracellular matrix, and various additional tissue-resident cell types.23,24
However, mechanisms underlying the relationship between TME heterogeneity after EGFR-TKI resistance and the efficacy of subsequent immunotherapy remain unclear. Furthermore, there is a lack of adequate predictive biomarkers for immunotherapy after EGFR-TKI resistance. Consequently, this review primarily focuses on the immunological dimensions of the TME, with an emphasis on its dynamic evolution in EGFR-mutant NSCLC, from treatment initiation to the development of acquired resistance to EGFR-TKIs. Furthermore, we have reviewed ongoing clinical trials concerning ICIs after EGFR-TKI resistance, and identified potential biomarkers for guiding subsequent immunotherapy.

Immunological basis: tumor microenvironment and response to ICIs
Tumors can be classified into three immune phenotypes according to the spatial distribution of cytotoxic immune cells within the TME: the immune-inflamed, immune-excluded, and immune-desert phenotypes.25 Immune-inflamed tumors, commonly referred to as “hot” tumors, are characterized by highly active immune cells. As ICIs exert their immunotherapeutic effects by reactivating T cell-mediated anti-tumor immune responses, these “hot” tumors demonstrate heightened sensitivity to immunotherapy.25,26 Mechanistically, these tumors primarily show elevated levels of CD8+ T cell infiltration and monocytes, increased interferon-γ (IFN-γ) signaling, a high tumor mutation burden (TMB), and high PD-L1 expression. They also feature increased B cell infiltration and intact antigen presentation, such as major histocompatibility complex I (MHC-I) expression on tumor cell surfaces.27 Collectively, these characteristics enhance the sensitivity of “hot” tumors to ICIs.28
In contrast, immune-excluded and immune-desert tumors, collectively known as “cold” tumors, exhibit limited responsiveness to ICIs.28 Both phenotypes are marked by reduced IFN signaling and low MHC-I expression.27 On one hand, a key characteristic of immune-excluded tumors is the retention of immune cells inside the tumor stroma, while failing to infiltrate the tumors. For example, CD8+ T cells are trapped in the stroma due to the presence of dense fibers and CAFs surrounding the tumor.29,30 Emerging evidence highlights the interaction between CAFs and immune cells in tumor tissue. Insulin-like growth factor 2 (IGF2), predominantly secreted by CAFs, suppresses the infiltration and cytotoxic function of CD8+ T cells, thereby facilitating immune escape.31 Another study shows that the limited effectiveness of PD-L1 inhibitors may result from the activation of the transforming growth factor-β (TGF-β) pathway, leading to exclusion of T cells.32 Despite the activation of T cells following immunotherapy, the inability to induce T cell infiltration results in poor clinical response. These tumors present high TGF-β signaling, myeloid-derived suppressor cells (MDSCs), and tumor angiogenesis.27 Furthermore, abnormal galectin expression and excessive consumption of vital nutrients, such as glucose and glutamine, have been documented in these tumors.33, 34, 35
Immune-desert tumors, on the other hand, exhibit a deficiency of T cells within the TME, resulting in a poor response to ICIs.36,37 Immunosuppressive cells like MDSCs and regulatory T cells (Tregs) are present and may inhibit dendritic cells (DCs) from presenting antigens.33 The immunosuppressive TME of “cold” tumors is further reinforced by metabolic reprogramming. These tumors are characterized by highly proliferative tumor cells exhibiting elevated fatty acid metabolism or neuroendocrine features.27 Meanwhile, although lactic acid production as a byproduct of glycolysis is a common metabolic feature of tumors, the immunosuppressive microenvironment driven by lactic acid accumulation is highly consistent with the characteristics of immune-desert tumors, indicating that lactic acid metabolism may serve as a potential mechanistic bridge between metabolic reprogramming and the immune “cold” phenotype. Specifically, the effect of lactic acid is mediated by hypoxia-inducible factor 1α, which promotes tumor progression by upregulating vascular endothelial growth factor (VEGF) and polarizing TAMs toward the immunosuppressive M2 phenotype.38 Besides, tumor-derived lactic acid also reduces the cytotoxic activity of T cells and natural killer (NK) cells by inhibiting nuclear factor of activated T cells (NFAT) signaling.39 It also impairs the differentiation of DCs, accelerates antigen degradation, and weakens antigen cross-presentation, thereby reducing the ability to trigger anti-tumor responses.40,41 In summary, lactic acid extensively inhibits anti-tumor immune responses in the TME, promoting tumor progression and immune evasion (Fig. 1).
Consequently, strategies to convert “cold” tumors into “hot” tumors for expanding the efficacy of immunotherapy are pivotal and actively explored. Multiple fundamental strategies are involved. For example, research has found that tumor-infiltrating ST2+ Tregs are increased in response to anti-PD-L1. Thus, blocking the interleukin (IL)-33/ST2 signal augments the anti-tumor efficacy of ICIs. Targeting both IL-33 and PD-L1 may enhance anti-tumor activity by reducing immunosuppressive factors such as Tregs and increasing cytotoxic T lymphocyte cells so that “cold” tumors can be converted to “hot” tumors.42 Theoretically, targeting immunosuppressive cells such as Tregs may relieve the immune suppression in the TME.43 Another study demonstrated that cell-intrinsic downregulation of bridging integrator 1 (BIN1) in NSCLC impairs CD8⁺ T cell infiltration and cytotoxicity by stabilizing Ras GTPase-activating protein-binding protein 1 (G3BP1) and promoting signal transducer and activator of transcription 1 (STAT1) degradation, thus contributing to tumor immune escape. Notably, this immunosuppressive microenvironment can be reversed by targeting the BIN1/G3BP1/STAT1 signaling pathway via STAT1 agonists like SB02024, G3BP1 inhibition or BIN1 restoration, which are potential therapeutic strategies.44 Additionally, inhibition of histone deacetylase has been shown to suppress the immunosuppressive activity of MDSCs and drive the conversion of tumors from “cold” to “hot” status.45,46
Parallel approaches aim to enhance immune effector function within the TME. For example, research has shown that the immune response can be activated by augmenting the cross-presentation capability of DCs, regulating chemokine/chemokine receptor axis to promote immune cell infiltration, and boosting T cell function through cytokine-based fusion proteins.25,47, 48, 49 Moreover, targeting tumor-specific antigens or inducing immunogenic cell death can enhance tumor immunogenicity.50,51 Furthermore, it is crucial to recognize cancer as a systemic disease. Modulating systemic factors, such as the gut microbiome and overall metabolism, offers novel approaches for reshaping the TME.25 In addition, treatment with histone deacetylase inhibitors can enhance tumor immunotherapy. Mechanistically, these inhibitors upregulate the expression of MHC molecules, enhance the recognition of tumor antigens by cytotoxic T lymphocytes, and increase the expression of ligands for NK activating receptors.52 Clinical trials have demonstrated that histone deacetylase inhibitors, such as mocetinostat and vorinostat, exhibit anti-tumor activity when combined with ICIs in patients with NSCLC who have progressed on prior ICI therapy.53,54
Therefore, immunotherapy outcomes are determined by immune phenotypes. TME of “cold” tumors and “hot” tumors is essential for understanding the efficacy of immunotherapy.

Changes in the TME before and after EGFR-TKI resistance

Immunosuppressive TME landscape in EGFR-mutant NSCLC
EGFR-mutated NSCLC generally exhibits reduced PD-L1 expression and TMB, which are fundamental biomarkers for the response to immunotherapy, leading to the immunosuppressive TME and weaker immunogenicity.55, 56, 57 Research has confirmed that IFN-γ activates interferon regulatory factor 1 (IRF1) via the Janus kinase (JAK)-signal transducer and activator of transcription (STAT) pathway, thereby upregulating PD-L1 expression.58 However, studies have confirmed the decrease of IFN-γ expression and IRF1 in EGFR-mutant NSCLC tumors, ultimately downregulating PD-L1 expression.59,60 These tumors downregulate the expression of MHC-I through the mitogen-activated extracellular protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) pathway, resulting in poor response to immunotherapy.61 Meanwhile, TGF-β expression is upregulated in EGFR-mutant NSCLC patients via EGFR–ERK1/2–p90 ribosomal S6 kinase (p90RSK) signaling. Increased TGF-β expression restrains the anti-tumor function and infiltration of CD8+ T cells, thereby fostering the development of “cold” tumors.62 Furthermore, EGFR mutations upregulate CD47 via the ERK/c-Myc and protein kinase B (AKT)/nuclear factor (NF)-κB pathways, mediating innate immune escape by impairing macrophage phagocytosis.63
EGFR mutations can also influence the level of immunosuppressive cytokines. One analysis shows that EGFR-mutant tumor cells exhibit elevated levels of immunosuppressive cytokines, such as C-C chemokine ligand (CCL) 18, CCL4, C-X-C chemokine ligand (CXCL) 1, CXCL17 and IL-1β.64 These enriched cytokines primarily accelerate the recruitment of MDSCs while impairing the recruitment of effector T cells. Conversely, EGFR-negative tumor cells strongly express pro-inflammatory cytokines, such as CXCL5, IL-7 and IL-32.64,65 Besides, EGFR signaling in EGFR-mutated NSCLC inhibits CXCL10 which recruits CD8+ T cells and promotes the production of CCL22 which recruits Treg cells.59 Inhibiting EGFR increases the expression of MHC I and II.66 Moreover, elevated level of CD73/adenosine pathway has been observed in EGFR-mutant lung cancer. CD73 inhibition could significantly decrease the proportion of Treg cells and inhibit tumor development.60 Promisingly, a study using an EGFR-mutant NSCLC mouse model showed that inhibiting CD73 in combination with anti-PD-L1 markedly suppressed tumor proliferation and increased the infiltration of CD8+ T cells and the release of IFN-γ and tumor necrosis factor-α.67 This provides a promising strategy for transforming the “cold” tumors to the “hot” ones (Fig. 2).

Differential immunogenicity across EGFR mutation subtypes
Different EGFR subtypes exhibit heterogeneous responses to immunotherapy. This difference may be attributed to a higher TMB in the EGFR 21L858R cohort than in the EGFR 19del cohort.68 Consistent with this lower TMB, 19del tumors maintain an immunosuppressive TME with decreased CD8+ T cells.69 Additionally, patients with the EGFR 21L858R mutation exhibited a significant increase in endothelial cells within the cancer area. Conversely, patients with 19del or acquired T790M mutations demonstrated a decrease in TILs in the tumor area.70 Meanwhile, patients harboring uncommon EGFR mutations, particularly the G719X mutation, demonstrated higher TMB, PD-L1 expression and M1 macrophages than those with classical EGFR mutations.71
Collectively, the limited efficacy of immunotherapy in EGFR-mutant NSCLC is attributable to its intrinsically immunosuppressive TME. Consequently, identifying patients who may benefit from immunotherapy based on mutation subtype and TME characteristics remains a critical challenge and a priority for ongoing research.

Dynamic evolution of the TME during EGFR-TKI treatment and upon acquired resistance
In patients with EGFR-sensitive mutations, EGFR-TKIs regulate the TME through a variety of mechanisms to stimulate an anti-tumor immune response. For example, short-term EGFR-TKI treatment is associated with elevated levels of cytotoxic CD8+ T cells, DCs and IFN-α, downregulated immunosuppressive proteins such as laminin subunit γ2, suppression of M2-like macrophage polarization, and elimination of Foxp3+ Tregs, all of which collectively enhance anti-tumor immunity.72,73 Besides, EGFR-TKIs influence tumor plasticity and immune-mediated cytotoxicity in EGFR-mutant NSCLC cells, resulting in a significant increase in tumor lysis facilitated by innate NK cells and antigen-specific T cells.74 Consistent with the reactivation of anti-tumor immunity, several studies found that EGFR-TKIs might lead to the down-regulation of PD-L1 expression in EGFR-mutant NSCLC cells,75, 76, 77 suggesting that the therapy not only enhances immune effector function but also relieves tumor-intrinsic immunosuppressive signals.
TILs, as primary anti-tumor immune effectors, exhibit functional changes linked to EGFR-TKI resistance. Single-cell RNA sequencing has revealed that EGFR-TKI resistance is associated with an increase in eomesodermin+ CD8+ T cells, potentially related to tissue-resident memory T cell conversion and metabolic reprogramming, thus offering important insights into TME heterogeneity.78 However, the elevated expression of PD-L1 indicates a potential basis for ICIs in patients with EGFR-mutant NSCLC and resistance to osimertinib.79 Furthermore, CAFs secrete inflammatory cytokines such as IL-6, IL-8, and hepatic growth factor to facilitate EGFR-TKI resistance.80 Despite these immunosuppressive features, upregulation of class I and class II antigen-presenting proteins has been observed in “hot” tumors following EGFR-TKI resistance, suggesting a potential enhancement of tumor immunogenicity that may influence the response to subsequent immunotherapy (Fig. 3).69
Previous studies have detected that elevated levels of CD25 on DCs and central memory CD8+ T cells in patients with EGFR-TKI-resistant NSCLC are markedly associated with the improved effectiveness of immunotherapy. This observation suggests that combining ICIs with CD25 inhibitors may improve the efficacy of immunotherapy in this patient population.81
Collectively, tumors may transition to a “hot” phenotype after EGFR-TKI resistance, offering new hope for immunotherapy in NSCLC patients with acquired TKI resistance. However, this complex and remodeled TME offers both opportunities and challenges for immunotherapy. Future research should aim to identify the key “time window” for immunotherapy and clarify the mechanisms underlying distinct TME dynamics, thereby providing a foundation for the selection of optimal therapeutic strategies and intervention timing.

Advances in immunotherapy after EGFR-TKI resistance
For nearly a decade, ICIs targeting PD-1 and PD-L1 have achieved profound clinical outcomes in individuals with advanced or locally advanced NSCLC and negative driver genes. Conversely, for patients with NSCLC with EGFR mutations, efficacy has been limited. A comparison of the immunotherapy trials for EGFR-mutated NSCLC after TKI resistance is shown in Table 1.

Mono-immunotherapy
Studies had shown that EGFR-mutant NSCLC patients derived minimal advantages from immunotherapy after developing resistance to TKIs.82,83 Notably, in the WJOG8515L study, nivolumab exhibited limited efficacy compared with chemotherapy in NSCLC patients with TKI resistance (median PFS: 1.7 months vs. 5.6 months, P < 0.001, hazard ratio [HR] = 1.92; median overall survival [OS]: 20.7 months vs. 19.9 months, P = 0.517, HR = 0.88).84 Additionally, this study showed that tumors that exhibited clinical benefits from nivolumab had higher levels of the T-cell-inflamed gene expression profile score and expression of genes which was linked to cytotoxic T lymphocytes or their recruitment.84 Similarly, the ATLANTIC study assessed the efficacy of durvalumab treatment in patients with EGFR mutations or ALK rearrangements. In patients with at least 25% of tumor cells with PD-L1 expression, the objective response rate (ORR) was 12.2%, which is higher than the 3.6% observed in those with fewer than 25% of tumor cells expressing PD-L1. However, both rates were lower than those in EGFR wild-type patients (16.4% in the PD‑L1 expression ≥25% group vs. 7.5% in the PD‑L1 expression <25% group).85
Although the safety of mono-immunotherapy after EGFR-TKI treatment is acceptable, its overall efficacy remains limited.84 Therefore, it is essential to explore a more optimized combination therapy to overcome the existing challenges.

ICIs coupled with chemotherapy
For patients with EGFR-mutant NSCLC who have developed resistance to TKIs, combining ICIs with chemotherapy has emerged as a key clinical strategy. The CheckMate-722 study reported that in NSCLC patients with EGFR-TKI resistance, treatment with nivolumab plus chemotherapy yielded an ORR of 31.3%, median PFS of 5.6 months, and median OS of 19.4 months.22 Similarly, in the KEYNOTE-789 study, patients receiving pembrolizumab plus chemotherapy achieved an ORR of 29%, median PFS of 5.6 months, and median OS of 15.9 months.86 However, both trials showed that the addition of ICIs to chemotherapy did not markedly prolong survival compared to placebo plus chemotherapy in this patient population.
Despite these negative outcomes, several studies have suggested that certain subgroups may derive greater benefit from immunotherapy plus chemotherapy. One analysis showed that among patients with resistance to first- or second-generation EGFR TKIs, the ORR of the combination of immunotherapy and chemotherapy surpassed that of the immune monotherapy group (25.0% vs. 9.1%, respectively). Additionally, patients without the T790M mutation exhibited notably longer PFS than those with the T790M mutation (4.23 months vs. 1.70 months, P = 0.019).87 Another study similarly found that T790M negative patients had superior PFS compared with T790M positive patients when treated with the combination of immunotherapy and chemotherapy (7.63 months vs. 3.42 months, P = 0.028).88 Collectively, these data indicate that immune–chemotherapy can improve ORR, PFS, and OS in select patient population.89,90
Although findings across studies are heterogeneous, the combined use of immunotherapy with chemotherapy may offer clinical benefits for some patients with EGFR-mutated NSCLC after EGFR-TKI failure. Mechanistically, there is a synergistic effect between ICIs and chemotherapy. Chemotherapy results in the release of antigens which will modulate the TME and potentially enhance the efficacy of ICIs.91,92 Nonetheless, certain studies have reported no significant survival benefits. Consequently, due to the varying outcomes of studies on the effectiveness of combining ICIs with chemotherapy, future research should focus on identifying related mechanisms.

ICIs, chemotherapy and antiangiogenic drugs
The combination of ICIs, chemotherapy, and antiangiogenic drugs has emerged as a promising strategy for patients with EGFR TKI resistant NSCLC, with several trials demonstrating clinical benefit. The phase III ORIENT 31 trial confirmed that adding bevacizumab and chemotherapy to sintilimab significantly improved outcomes in this population compared with chemotherapy alone. The combined therapy markedly prolonged PFS (7.2 months vs. 4.3 months; HR = 0.51, 95% CI: 0.39-0.67]; P < 0.001) and OS (21.1 months vs. 19.2 months; HR ranging from 0.79 to 0.84 after adjusting for crossover).93 This therapy was approved for use in patients with EGFR-mutant late-stage NSCLC who experienced TKI failure by National Medical Products Administration in China. Moreover, the IMpower150 study showed that compared with bevacizumab plus chemotherapy, the combination of atezolizumab, bevacizumab, and chemotherapy offers enhanced survival advantages for EGFR-mutant NSCLC patients who had received prior TKI therapy. The median OS was 27.8 months versus 18.1 months (HR: 0.74, 95% CI: 0.38–1.46).94
The mechanisms for this combination are well established. Tumors secrete angiogenic factors such as VEGF, TGF-β and prostaglandin E2 and reduce effector T cells function, fostering an immunosuppressive environment.95 Notably, VEGF further promotes immune evasion by inhibiting the maturation of DCs, reducing T cell infiltration, and increasing immunosuppressive cells such as MDSCs and Tregs. Anti-VEGF antibodies, such as bevacizumab, can reverse these immunosuppressive effects by blocking VEGF, thereby mediating synergistic effects when combined with ICI.96 Specifically, by normalizing blood vessels, increasing cytotoxic T-lymphocytes responses and activating DCs, anti-angiogenic drugs may therefore theoretically prevent or delay acquired ICI resistance.97 One research indicated that in EGFR-mutated NSCLC patients with resistance to EGFR-TKIs, the administration of antiangiogenic drugs before immunotherapy demonstrated superior PFS compared to immune monotherapy (3.42 vs. 1.58 months, P = 0.027).98
However, not all trials have yielded consistent results. The IMpower151 study, which evaluated atezolizumab in combination with chemotherapy and antiangiogenic drugs, failed to meet its primary endpoints,99 a finding that diverges from the outcomes of IMpower150. This discrepancy may be attributed to the following factors. IMpower151 exclusively enrolled Chinese patients characterized by a high EGFR mutation rate and a relatively small sample size. A significant proportion of EGFR-mutant patients had received third-generation EGFR-TKIs.99 In contrast, IMpower150 enrolled a more diverse patient population from multiple ethnic groups with a larger sample size and a lower EGFR mutation rate. Patients mainly received first- or second-generation TKIs in first-line treatment.100 These differences, particularly in driver gene mutation profiles and previous treatment history, might influence immunotherapy response and be key factors that prevented IMpower151 from replicating the successful results seen in IMpower150.
Additionally, in-depth analysis of clinical trial data reveals that the heterogeneity in treatment efficacy is systematically associated with other factors. Firstly, the molecular characteristics of tumors indicate that those with EGFR 21L858R mutation and T790M negative status typically exhibit a more favorable immune microenvironment, resulting in increased sensitivity to immune combination therapy.93,101 Secondly, the patient treatment history indicates that prolonged exposure to EGFR-TKIs may lead to T cell depletion, M2 macrophage polarization, and metabolic reprogramming, such as lactic acid accumulation. These changes collectively promote immunosuppression within the TME, thereby impairing the efficacy of subsequent immunotherapy.72 In contrast, patients with fewer lines of front-line treatment may possess a TME that shows greater potential for immune mobilization.93,102 These factors may partly explain the wide interpatient variability in immunotherapy responses, as demonstrated by the distinct efficacy of immune monotherapy and combination immunotherapy.
In summary, although ICI monotherapy has demonstrated limited efficacy in EGFR mutant NSCLC after TKI resistance, ICI based combination strategies may represent a more promising approach. Despite encouraging outcomes in some studies, these combinations are associated with increased toxicity, which has led to premature trial termination in certain cases.103 For patients with EGFR mutant NSCLC progressing on EGFR-TKIs, re-biopsy is recommended to assess biomarkers related to the TME and guide treatment selection. Future research should focus on identifying predictive biomarkers and refining combination strategies to maximize efficacy while minimizing toxicity.

Safety
The safety of immunotherapy for NSCLC after EGFR-TKI resistance is closely influenced by the treatment regimen employed. Table 2 summarizes treatment related adverse events observed across key trials. Interstitial lung disease (ILD) represents a significant adverse event, particularly in patients with EGFR mutations after failure of EGFR-TKIs. In the TATTON study, five patients (22%) receiving the combination of osimertinib and durvalumab developed ILD, with two patients progressing to grade 3 or higher severity.104 The CAURAL study enrolled patients with advanced NSCLC who were positive for EGFR T790M and had experienced disease progression after standard EGFR-TKI treatment. However, because of the increased incidence of ILD observed in the TATTON trial, enrollment in the CAURAL study was terminated early. Only 14 patients had been enrolled in the osimertinib plus durvalumab group, and 1 patient developed grade 2 ILD.103 Given the increased risk of ILD associated with the combination of immunotherapy and EGFR-TKIs, this combination is currently not recommended for patients with EGFR-TKI resistance.103
In patients receiving immunotherapy, the cornerstone of preventing immune-related ILD (ir-ILD) lies in risk stratification through comprehensive pre-treatment assessment and early monitoring of high-risk individuals. Although no standardized prophylactic intervention has been established, a previous study identified impaired spirometry, reduced percentages of total lung capacity and diminished functional residual capacity as risk factors associated with ir-ILD. Furthermore, a combined assessment of percent predicted forced vital capacity and percent predicted forced expiratory volume in one second has been proposed to predict ir-ILD in patients with NSCLC treated with ICIs.105 The integration of these markers allows for the stratification of patients into various risk categories, thus facilitating the identification of subgroups that necessitate intensified monitoring. Evidence suggests that more than half of ir-ILD cases, including all severe events, occur within the first 60 days of ICI initiation.105 Patients with impaired pulmonary function should be closely monitored during the early phase of immunotherapy. Therefore, clinical decisions must carefully assess the therapeutic benefits against the risks of toxicity.

Novel treatments and opportunities
Emerging therapeutic strategies such as bispecific antibodies and personalized neoantigen vaccines have demonstrated clinical promise in the management of EGFR-mutant NSCLC. Bispecific antibodies are a novel therapeutic strategy. By concurrently targeting two distinct antigens, these antibodies enhance anti-tumor immune responses. For instance, ivonescimab (a PD-1/VEGF bispecific antibody) demonstrated improved efficacy in patients with advanced NSCLC and EGFR mutations after progression on EGFR-TKIs. The HARMONi-A trial first established that ivonescimab plus chemotherapy brought survival benefits (PFS: 6.8 months vs. 4.4 months, HR = 0.52, P < 0.001; OS: 16.8 months vs. 14.0 months, HR = 0.79, P = 0.057).106 The safety profile was tolerable and manageable.107 Accordingly, National Medical Products Administration approved this therapy for the treatment of EGFR-mutant advanced NSCLC after TKI-resistance.
Personalized neoantigen vaccines also show promise in EGFR-mutant NSCLC. Research has shown that EGFR-directed neoantigens induced robust T-cell-mediated immunity, thus influencing clinical outcomes.108,109 The efficacy could be further elevated by combining with other therapies, such as ICIs. Therefore, combining neoantigen vaccines with ICIs may enhance response, though further validation is needed.
Future work should integrate these approaches with computational models to accelerate the development of personalized therapies. Identifying optimal treatment sequences and patient selection criteria will be essential to translate these novel treatments into meaningful clinical benefits for patients with EGFR mutant NSCLC after TKI failure.

Predictive biomarkers for immunotherapy after EGFR-TKI resistance
Reliable biomarkers for identifying patients with a higher likelihood of benefiting from ICIs are essential in the subsequent treatment of advanced EGFR-mutant NSCLC. The PD-1/PD-L1 signaling pathway plays a critical role in tumor immunosuppression by inhibiting T lymphocyte activation and promoting immune tolerance in tumor cells, thereby facilitating tumor immune escape.110 PD-L1 itself exhibits a highly heterogeneous distribution pattern within the tumor microenvironment. Beyond its canonical membrane-bound form on tumor cells, where it binds to PD-1 on effector T cells to trigger exhaustion and immune escape, PD-L1 is also widely expressed on immune cells such as macrophages and DCs, localized in the cytoplasm of tumor cells, and carried in tumor-derived exosomes. For instance, tumor-derived exosomal PD-L1 exerts a systemic immunosuppressive effect by inhibiting functions of T cells. Cytoplasmic PD-L1 enhances DNA damage repair in cancer cells, thereby promoting tumor proliferation and progression. Unfortunately, current anti-PD-1/PD-L1 immunotherapies mainly target membrane-bound PD-L1 on tumor cells. They fail to fully downregulate PD-L1 expression on immune cells, cannot effectively block the immunosuppressive activity of tumor-derived exosomal PD-L1, and are unable to interfere with the tumor-promoting functions of cytoplasmic PD-L1. Therefore, the dynamic expression and heterogeneous localization of PD-L1 are key factors contributing to the limited efficacy and acquired resistance of anti-PD-1/PD-L1 immunotherapy.110,111 In this context, identifying patients whose tumors exhibit higher overall PD-L1 expression, including both membrane and non-membrane forms, becomes particularly important. Subgroup analysis from the ATLANTIC study indicated that the ORR of EGFR-TKI resistant NSCLC patients with PD-L1 expression ≥25% was greater than that of the PD‑L1 expression <25% group (12.2% vs. 3.6%). This finding supports the potential role of PD-L1 as a biomarker for evaluating the possibility of immunotherapy in advanced NSCLC patients with EGFR mutations.85
TMB has also emerged as a candidate biomarker for immunotherapy response, with higher TMB generally associated with increased immunogenicity and improved clinical outcomes.112 Unexpectedly, the WJOG8515L study found no clear association between TMB and nivolumab efficacy in EGFR-mutant NSCLC patients after the failure of EGFR-TKI.84 In contrast, another study reported that among NSCLC patients treated with immunotherapy after TKI-resistance, elevated TILs in cancer area were correlated with improved ORR and PFS.70 Studies demonstrated that patients with lung adenocarcinoma harboring EGFR mutations exhibited markedly lower TMB levels than individuals with wild-type EGFR, partly explaining the reason for the reduced immunogenicity and ICI effectiveness.56,113, 114, 115
Furthermore, the specific EGFR mutation subtype appears to modulate immunotherapy outcomes. Tumors harboring the EGFR 21L858R mutation and EGFR 19del exhibit distinct patterns of response to immunotherapy. A comparative clinical analysis indicated that the subtype of EGFR mutation significantly influenced the efficacy of ICI. Patients with the EGFR 21L858R mutation exhibited similar benefit from ICIs to those with wild-type tumors (ORR: 16% vs. 22%, P = 0.42; OS: HR = 0.917, 95% confidence interval: 0.597–1.409, P = 0.69). In contrast, individuals with the EGFR 19del exhibited significantly poorer outcomes from ICIs than those with EGFR wild-type tumors (ORR: 7% vs. 22%, P = 0.002; OS: HR = 0.69, 95% CI: 0.493–0.965, P = 0.03).116 Another study showed a longer median PFS in patients with EGFR 21L858R than those with EGFR 19del after receiving immunotherapy following TKI resistance (6.4 vs. 5.0 months) (Table 3).69
By integrating biomarkers such as PD-L1 expression, TMB, and EGFR mutation subtype, it may be possible to identify patients with EGFR-mutant NSCLC who are more likely to benefit from immunotherapy after TKI failure. However, current biomarkers remain limited. Future research should focus on overcoming the limitations of existing TME-based biomarkers and developing more effective strategies to guide immunotherapy in this challenging population.

Concluding remarks and future perspectives
This review systematically summarizes the advances in immunotherapy for EGFR-mutant NSCLC after EGFR-TKI resistance. Despite decades of research, this patient population has long faced limited treatment options following disease progression on targeted therapy, and the advent of immunotherapy offers a new hope for overcoming acquired resistance. Beginning with the dynamic evolution of the TME, this review delineates the mechanisms by which EGFR mutations drive the formation of “cold” tumors and the TME remodeling induced by EGFR-TKI treatment, thereby providing a critical opportunity for immunotherapy.
Notably, although certain combination therapies have demonstrated survival benefits in specific subgroups, the striking heterogeneity of clinical responses remains a core challenge that limits broad application. Studies such as CheckMate-722 and KEYNOTE-789 failed to reach their primary endpoints, while others such as ORIENT-31 and HARMONi-A achieved breakthrough results, which profoundly reveal the high heterogeneity in this patient population.
This heterogeneity arises from multiple factors. First, the intrinsic immunological profiles differ between EGFR mutation subtypes. EGFR 19del and 21L858R mutations exhibit distinct patterns of immune cell infiltration, PD-L1 expression, and TMB. Second, the impact of different EGFR-TKI resistance mechanisms, such as EGFR C797S mutation and MET amplification, on TME remodeling has not yet been clarified. Third, the functional heterogeneity of immune cell subsets in the TME plays a crucial role in immune response after resistance, such as metabolic reprogramming of tissue-resident memory T cells and polarization regulation of M2-type TAMs. Fourth, the cumulative impact of previous treatment lines and TKI exposure duration on the functional status of immune cells leads to differences in immune mobilization capacity among individuals.
This complex heterogeneity poses substantial challenges for clinical decision-making. Consequently, significant bottlenecks persist at the clinical translation level. First, the predictive power of existing biomarkers is insufficient. Traditional single biomarkers such as PD-L1 expression and TMB show limited sensitivity in this population, and there is a lack of comprehensive predictive models that can integrate genomic characteristics, TME phenotypes, and clinical factors. Second, toxicity management in combination therapy poses a challenge. Although immunotherapy plus chemotherapy plus anti-angiogenic therapy improves efficacy, the incidence of adverse reactions increases, and risk stratification and prevention strategies for ir-ILD urgently need to be improved. Third, the clinical evidence for novel therapies, such as bispecific antibodies and personalized neoantigen vaccines, remains limited, and the screening criteria for eligible populations are not well defined.
To address these bottlenecks and the inherent heterogeneity of this patient population, future research should focus on the following directions. Employing single-cell RNA sequencing and spatial transcriptomics to figure out the dynamic evolution of the TME before and after TKI treatment and across different resistance mechanisms will be essential for clarifying the functional remodeling of immune cell subsets and identifying key intercellular communication networks that drive immune escape.117 Besides, in-depth analysis of the crosstalk mechanisms among EGFR signaling, immune checkpoint pathways, and angiogenesis pathways and how this crosstalk activates downstream immunosuppressive mediators is necessary for precision intervention. Moreover, developing composite dynamic biomarkers that integrate genomic features, TME status, and clinical factors to construct multi-dimensional predictive models of immunotherapy responses, represent a promising strategy to overcome the limitations of single biomarkers. Additionally, advancing the development of novel therapies such as bispecific antibodies and neoantigen vaccines, and exploring their synergistic mechanisms and optimal application in combination with existing regimens will be critical. Meanwhile, exploring the role of interventions such as gut microbiota modulation and metabolic reprogramming in reshaping the TME after EGFR-TKI resistance will provide new insights to enhance therapeutic efficacy.
Much work remains to be accomplished and the following strategies might help translate these insights into clinical practice. First, for patients who present “immune-desert” or “immune-excluded” phenotypes, the addition of agents that target TME remodeling, such as TGF-β inhibitors or CD73 inhibitors, to existing combination regimens may pave the way for immune cell infiltration. For those exhibiting an “immune-inflamed” phenotype, the opportunities should be seized to explore the combination with novel immunomodulators. For example, combining CD25 inhibitors to deplete Tregs may reinforce immune responses. Second, optimization of treatment sequencing and combinations is critical. Based on dynamic TME monitoring, strategies should be refined to optimize administration sequences and establish de-escalation or escalation treatment plans based on efficacy prediction. Third, establishment of dynamic monitoring and risk stratification systems is essential. Liquid biopsy and analysis of immune phenotype of circulating tumor cells enable real-time assessment of TME changes during treatment to guide therapeutic adjustment. Besides, it is important to conduct toxicity risk stratification for patients, strengthen early monitoring for high-risk groups and explore preventive intervention measures.
In summary, overcoming the dilemma of immunotherapy after EGFR-TKI resistance lies in upgrading the current therapeutic approach to a TME-based precise therapy model. It is expected to identify those most likely to benefit from immunotherapy and match them with optimal combination regimens and the best intervention timing, ultimately maximizing survival benefits for every patient with acquired EGFR-TKI resistance.

Funding
This work was supported by National Natural Science Foundation of China (No. 82273124 to W.Z.).

CRediT authorship contribution statement
Xinwei Wu: Writing – review & editing, Writing – original draft, Conceptualization. Yangbin Shi: Writing – review & editing, Conceptualization. Danni Wang: Writing – original draft, Conceptualization. Beibei Liu: Writing – review & editing. Yanwei Zhang: Writing – review & editing, Supervision. Lele Zhang: Supervision. Fangfei Qian: Writing – review & editing. Wei Zhang: Writing – review & editing, Supervision, Conceptualization.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.