PD-L1 expression in tumor tissues is the most extensively studied predictive biomarker for clinical response to PD-1/PD-L1 inhibitors. Status of tumoral PD-L1 expression is now routinely used for selecting patients who could benefit the most. The contribution of PD-L1 expression to poor efficacy of anti-PD-1/PD-L1 therapy in EGFR mutant NSCLC is controversial. A number of studies have reported that high PD-L1 expression is found more frequently in EGFR-mutant than EGFR-wild type lung tumor tissues (44, 54–56). The activation of the PD-1 pathway is thus believed to contribute to immune escape in EGFR-driven NSCLC. In contrast, another study has found a decreased PD-L1 expression in tumor tissues from NSCLC patients bearing EGFR mutation (57). More recently, a pooled analysis of 15 clinical studies also revealed that patients with EGFR mutation have decreased PD-L1 expression (58). Analysis of The Cancer Genome Atlas (TCGA) and the Guangdong Lung Cancer Institute (GCLI) cohort have also confirmed an inverse correlation between EGFR mutation and PD-L1 expression in tumor tissues (58). On the other hand, other studies have reported a lack of correlation between the expression of PD-L1 and PD-L2 in patients with different EGFR mutation status (59). With these conflicting findings, the use of PD-L1 expression alone could not adequately predict and explain why EGFR-mutant NSCLC exhibits poor response to PD-1/PD-L1 inhibitors.

In a recent study investigating the impact of TMB on clinical outcomes of NSCLC patients treated with EGFR TKIs, TMB was found to be remarkably lower in EGFR-mutated tumors (n = 153) than EGFR wild-type tumors (n = 1,849) (median 3.77 versus 6.12 mutations/Mb; P < 0.0001) (60). To this end, the association of higher TMB with tobacco smoking leading to better outcomes with PD-1/PD-L1 ICIs is well documented (9, 11, 34, 37, 46, 61). A recent meta-analysis also revealed that never smokers are less responsive to PD-1/PD-L1 inhibitors (62). Interestingly, EGFR mutant NSCLC is more enriched in the never smoking population (63).

Among the more common sensitizing EGFR mutations, TMB in the exon 19 deletion cohort was found to be lower than that in the L858R cohort (60). Hastings et al. has recently reported that clinical outcomes (OS and ORR) with PD-1 ICIs were worse in patients with exon 19 deletion than patients harboring L858R mutation (53). PD-L1 and smoking status were similar in the two patient subpopulations. Further TMB analyses of the two patient cohorts suggested that the higher TMB in EGFR L858R mutation could contribute to the differential responses to PD-1 ICIs.

In fact, specific subset of EGFR-mutant NSCLC patients have been shown to preferentially benefit from PD-1/PD-L1 ICIs to some extent (64–68). In a retrospective study evaluating NSCLC patients from the IMMUNOTARGET registry, the response of 125 pre-treated EGFR-mutated patients with ICI monotherapy was compared among patients with different EGFR mutation subgroups (69). The ORR was not notably affected by PD-L1 expression levels, smoking status or previous lines of treatment, but was significantly different in the various EGFR mutation subgroups (3.7%, 9.5%, 20.8% and 11.8%, respectively, in EGFR T790M, exon 19 deletion, L858R, and other EGFR mutation subgroups) (69). This is in line with the aforementioned findings of Hastings et al., demonstrating more favorable outcomes of L858R (coincident with higher TMB) than exon 19 deletion (coincident with lower TMB). The combination of nivolumab and erlotinib in 21 EGFR mutated NSCLC patients has been evaluated by Gettinger et al. (64). Intriguingly, patients bearing the EGFR L858R mutation were achieving longer survival benefit than those harboring other EGFR mutations (64). While TMB in the tumor tissues was not assessed in the study, the findings are consistent with the fact that TMB in EGFR L858R mutated NSCLC favors better response to PD-1 ICIs.

A typical tumor mass consists of not only a heterogeneous population of cancer cells but also a variety of neighboring host cells, tumor-infiltrating lymphocytes, extracellular matrix proteins, and other secreted factors, which collectively referred to as the TME. Tumors have been classified into 4 different TME types according to the presence or absence of tumor-infiltrating lymphocytes (TIL) and PD-L1 expression (Type I: TIL⁺, PD-L1⁺; Type II: TIL^-, PD-L1^-; Type III: TIL^-, PD-L1⁺; and Type IV: TIL⁺, PD-L1^-) (70). Type I tumors were found to the only subtype that responds to PD-1/PD-L1 inhibitor (70). Therefore, besides PD-L1 expression, high level of TIL is critical to allow PD-1/PD-L1 inhibitors to work. Similarly, Chen et al. divided tumors into different immunity phenotypes (immune-desert phenotype; immune-excluded phenotype; and inflamed phenotype) according to a set of tumor, host, and environment factors (71). Using Chen’s classification, tumors with the immune-desert and immune-excluded phenotypes are resistant to PD-1/PD-L1 inhibitors (71). Importantly, there is a close correlation between EGFR mutations and an uninflamed TME with immunological tolerance and weak immunogenicity (48, 58). This correlation may explain the inferior response of EGFR-mutant NSCLC to PD-1 blockade therapy. The overexpression of CD73 has also been reported in EGFR mutant NSCLC, thus resulting an immunosuppressive TME and reduced IFN gamma signature (72). Figure 1 depicts the characteristic composition and function of the immunosuppressive TME in EGFR-mutated NSCLC cells.

EGFR-mutant NSCLC is characterized by its aberrant activation of the EGFR signaling pathway. To this end, activation of EGFR signaling has been reported in numerous studies to participate in immunosuppression and immune escape. Regulatory T cells (Tregs) play a critical role in suppressing the immune response to self and foreign particles, which help prevent autoimmune disease. They generally suppress the induction and proliferation of effector T cells. Amphiregulin (AREG) is an EGF-like growth factor and it is frequently upregulated in tumors (73). AREG is a known ligand of EGFR (74). Importantly, AREG is critical for Treg function in vivo, thus providing a mechanistic link between the EGFR signaling and regulation of the immune system (75). Wang et al. reported that AREG maintains the suppressive function of Tregs via the EGFR/GSK-3β/Foxp3 machinery in vitro and in vivo, thus confirming the importance of EGFR signaling in the regulation of Tregs (76). Recently, a long noncoding RNA lnc-EGFR has been shown to stimulate Treg differentiation and promote immune invasion of hepatocellular carcinoma via an EGFR-dependent signaling pathway (77). Moreover, the inhibition of EGFR signaling by gefitinib has been shown to alter the immune environment of the targeted cancer in vitro and in vivo, probably by reducing the number of Tregs in the tumors (78).

Myeloid-derived suppressor cells (MDSCs) are immature myeloid cells that suppress immune responses. MDSCs expand during cancer, infection and inflammatory diseases. A recent study reported that EGFR TKI therapy alters the TME in EGFR mutant NSCLC and elevates the level of mononuclear MDSCs (79). The serum level of inflammatory factors IL-10 and CCL-2 was also found to be increased in vivo after EGFR TKI treatment (79). The increase in MDSC and inflammatory factors associated with EGFR TKI treatment has been proposed to explain why most EGFR TKI-resistant NSCLC patients are also refractory to anti-PD-1/PD-L1 ICIs (79). Moreover, MDSCs are known to inhibit IL-2 and anti-CD3/CD28 mAb-induced T cell amplification and Th1 polarization but induce apoptosis in T cells in an IDO-dependent manner (80). To this end, the activation of STAT3 (an important downstream signaling molecule of the EGFR pathway) is required for IDO expression (80). Therefore, STAT3 activation is essential for immune suppression of MDSCs. In fact, persistent activation of STAT3 has been shown to promote MDSC-mediated immune suppression in lung cancer (81).

YAP is a major mediator of the Hippo pathway and it has been shown to promote cancer progression, drug resistance and metastasis in NSCLC (82). Accumulating evidence suggests that YAP also plays critical role in cancer immunity. YAP interacts with interferon regulatory factor 3 to negatively regulate innate immunity (83). YAP has also been reported to regulate tumor–associated immune cells (including MDSCs, macrophages, and Tregs) in the TME (83–85). In NSCLC tumor specimens, high nuclear YAP staining is associated with positive PD-L1 expression (86). Genetic knockdown or chemical inhibition of YAP was shown to reduce mRNA and protein expression of PD-L1 in NSCLC cell lines (86). On the other hand, forced expression of YAP was shown to increase PD-L1 protein expression in NSCLC A549 cells (87). Recently, a gefitinib-resistant PC9 cell line has been shown to express higher protein level of YAP and PD-L1 than the parental cells (88). Importantly, YAP knockdown could reduce PD-L1 expression in gefitinib-resistant PC9 cells (88). It is also noteworthy that the EGFR signaling pathway has crosstalk with the Hippo/YAP pathway, which positively regulating the YAP oncogenic function in various cancers including NSCLC (89).

EGFR TKIs have been reported to cause immunogenic apoptosis of tumor cells (124), and subsequently releasing aberrant intracellular antigens and recruiting T cells via interferon-γ-induced major histocompatibility complex class I presentation (120). A few clinical studies have been conducted to evaluate the combination of PD-1-based immunotherapy and EGFR targeted therapy in EGFR TKI-naïve and/or pretreated EGFR-mutant NSCLC patients ( Table 3 ).

CheckMate 012 (NCT01454102) is a multi-arm Phase 1 study evaluating the combination of nivolumab and different agents including erlotinib in advanced NSCLC with small sample size (n = 20) (125). At least 4 out of the 20 patients with acquired resistance to EGFR TKI were shown to achieve clear benefit from the addition of nivolumab to EGFR TKI therapy [objective response rate (ORR) = 15%, including 1 complete response (CR)] (125). The combination of atezolizumab and erlotinib has been evaluated in another Phase I clinical trial in EGFR TKI-naïve and –treated NSCLC patients (NCT02013219; n = 28) (126). Atezolizumab plus erlotinib was shown to demonstrate a tolerable safety profile and favorable efficacy compared with prior reports of erlotinib monotherapy. ORR was 75% and median PFS was 15.4 months (126).

On the other hand, the combination of EGFR TKIs with PD-1/PD-L1 inhibitors did not demonstrate favorable clinical efficacy in EGFR-mutated NSCLC patients in a few other trials. In the Phase I/II KEYNOTE-021 trial (NCT02039674), the combination of erlotinib (n = 12) or gefitinib (n = 7) with pembrolizumab was evaluated in EGFR-mutant advanced NSCLC patients (129). While pembrolizumab plus erlotinib produced similar adverse events as erlotinib monotherapy, pembrolizumab plus gefitinib combination caused grade 3/4 liver toxicity in five out of seven patients and resulting in premature treatment discontinuation. Disappointingly, pembrolizumab plus erlotinib did not improve ORR compared with previous EGFR TKI monotherapy studies (129). Another open-label multicenter Phase I trial (NCT02088112) has also been recently conducted to evaluate the combination of gefitinib and duvalumab in patients with EGFR-mutant and EGFR TKI-naïve NSCLC (127). Durvalumab and gefitinib in combination was shown to produce higher toxicity than either drug alone. There was no significant improvement in PFS or ORR compared with gefitinib monotherapy previously reported in similar patient populations. To the best of our knowledge, no other Phase III trials investigating the combination EGFR TKI and PD-1 inhibitors in EGFR TKI-naïve patients are currently planned or actively recruiting.

In preclinical studies, the 3^rd generation EGFR TKI, osimertinib, has been shown to enhance the antitumor efficacy of PD-1/PD-L1 blockade therapy by increasing CD8+ T cell infiltration in tumors (130). However, in a Phase Ib trial TATTON (NCT02143466) investigating the safety and tolerability of combining osimertinib and durvalumab, 38% of patients developed serious interstitial pneumonitis and thus the poor safety profile renders the combination not feasible (102). In an animal study using EGFR mutated tumor-bearing mouse model, osimertinib (but not gefitinib) combined with anti-PD-L1 therapy was shown to cause pneumonitis and injury to lung tissues (124).

IMpower150 was an open-label Phase III study (NCT02366143) comparing atezolizumab ± chemotherapy + bevacizumab (ABCP group) versus chemotherapy + bevacizumab (BCP group) in metastatic NSCLC patients who had not previously received chemotherapy (128). Bevacizumab is an anti-VEGF monoclonal antibody and its combination with chemotherapy has been approved for the treatment of metastatic NSCLC (131). Apart from the well-known antiangiogenic effects of bevacizumab (132), the inhibition of VEGF has also been shown to mediate immunomodulatory effects (14, 133, 134). Thus, the efficacy of atezolizumab may be enhanced by the addition of bevacizumab to reverse VEGF-mediated immunosuppression (134, 135). Encouragingly, the ABCP group was shown to achieve significantly longer PFS (8.3 months versus 6.8 months) and OS (19.2 months versus 14.7 months) than the BCP group, regardless of PD-L1 expression and EGFR/ALK genetic alteration status (128) ( Table 3 ).

Thermal ablation has been used for the management of localized tumors for patients not eligible for surgical resection. A growing body of evidence suggests that thermoablation could modulate both adaptive and innate immunity (154). However, the induced immune responses are mostly weak and not sufficient for the eradication of tumors or durable prevention of disease progression. In recent years, the combination of thermal ablation and ICIs therapy have been evaluated with promising results. Shi et al. reported that radiofrequency ablation (RFA) treatment of liver metastases increased not only T cell infiltration but also PD-L1 expression in primary human colorectal tumors (155). Using mouse tumor models, RFA treatment of one tumor was found to initially enhance a strong T cells mediated immune response in tumor. However, the tumor quickly overcame the immune responses by inhibiting CD8 and CD4 T cell function, subsequently driving a shift to higher regulatory T cell to Teff ratio (155). Importantly, the combination of RFA and anti-PD-1 antibodies was found to significantly enhance T cell immune responses and lead to prolonged animal survival (155).

Radiotherapy is commonly used in cancer therapy to achieve a local control of the irradiated target tumor lesions regardless of clinical stage. Numerous preclinical studies have demonstrated that radiotherapy could activate anti-tumor immune response. Irradiation is known to activate host immunity by triggering ICD, which is characterized by the release of DAMPs to activate dendritic cells and to prime antigen-specific T cells in a dose-dependent manner (156). Procureur et al. has recently published an excellent review about the enhancement of ICIs by radiotherapy-induced immunogenic cell death (157). Radiotherapy-induced systemic immune activation could cause shrinkage of distant tumor lesions outside the irradiated field, a phenomenon known as abscopal effect (158). In the past, abscopal effect was believed to be a very rare phenomenon. However, recent clinical data revealed that the combination of radiotherapy and anti-PD-1/PD-L1 ICIs could induce the abscopal effect (159).

On the other hand, the induction of immunosuppressive cytokines and chemokines by radiotherapy contribute to immunosuppressive reactions (160). The PD-1/PD-L1 axis is one of the key factors in cancer immune escape induced by radiotherapy. Moreover, upregulation of PD-L1 expression has been reported in NSCLC patients who have undergone radiotherapy with or without chemotherapy as preoperative treatment (161). In addition, EGFR signaling after irradiation leads to PD-L1 upregulation via the IL-6/JAK/STAT3 pathway (162). As anti-PD-1/PD-L1 antibodies could relieve this immunosuppression, it makes sense to combine PD-1/PD-L1 ICIs and radiotherapy (161).

The phase III PACIFIC trial (NCT02125461) provided the most remarkable clinical data to support the combination of radiotherapy and anti-PD-1/PD-L1 ICIs. In this study, progression-free survival was significantly prolonged by prescribing durvalumab as a consolidation therapy after concurrent chemoradiotherapy as compared with placebo (51). Based on this trial, durvalumab following chemoradiotherapy has been approved for the treatment of NSCLC by the US Food and Drug Administration in 2018.

Local immunotherapies such as the intratumoral injection of oncolytic compounds have been used to reinstate and enhance systemic anticancer immune responses. A recent animal study reported the use of local immunotherapy to sensitize the tumor to subsequent immune checkpoint blockade (163). LTX-401 is an oncolytic peptide designed for local immunotherapy. The sequential LTX-401 treatment combined with double checkpoint inhibition of PD-1 and CTLA-4 exhibited strong antineoplastic effects on both the primary lesions and distant tumors (163).