PD-L1 expression and tumor mutation burden (TMB) are the most popularly applied biomarkers for predicting immunotherapy response. NSCLC patients with a high expression of PD-L1 usually obtain better benefits from anti-PD-1/PD-L1 therapy (29, 30). NSCLC patients with a higher tumor tissue TMB (tTMB) are correlated with a more effective response to PD-1/PD-L1 inhibitors by leading to tumor-specific neoantigen formation to elevate tumor immunogenicity (31–34). Therefore, any intrinsic and extrinsic factors affecting the expression of PD-L1 or/and TMB in tumor cells may be the mechanism for resistance to immunotherapy. Next-generation sequencing (NGS) applications found that innate drive gene mutations affect PD-L1 expression and TMB, which may contribute to primary or adaptive resistance to immunotherapy in NSCLC. EGFR is a well-studied drive gene in NSCLC patients, the mutation status of which can induce PD-L1 expression (35, 36). In EGFR-mutated NSCLC cells, activated EGFR was found to elevate PD-L1 expression via the IL-6/JAK/STAT3 signaling or p-ERK1/2/p-c-Jun pathway (35, 36). Moreover, NSCLC patients with uncommon EGFR mutation, including G719X, L861Q, S768I, and Ex20 ins, had more abundant PD-L1 expression accompanied by CD8⁺ tumor-infiltrating lymphocytes(TILs) infiltration (37). In addition, mutations of RET and HER2 attenuated PD-L1 expression, while mutations of ALK, ROS, and MET enhanced PD-L1 expression but downregulated TMB and TIL infiltration leading to resistance to ICIs (38–41). TP53 or NFE2L2/KEAP1 mutations could increase TMB and PD-L1 expression to enhance the sensibility to immunotherapy (42, 43). However, NSCLC subsets with co-occurrence of EGFR, HER2, ALK, ROS1, RET, and MET mutations get minimal benefit from ICIs despite high PD-L1 expression. These findings confirmed a TMB/PD-L1-independent effect on response sensitivity to ICIs for specific drive genes mutations (44). KRAS mutation, another frequently mutated type in NSCLC, usually co-occurs with different gene mutations. Various KRAS mutation subtypes have distinct TMB and PD-L1 expressions, and KRAS G12C is the most common subtype with a high rate of PD-L1 positive (45). Mutated-TP53 was more prevalent in KRAS wild-type NSCLC, while mutated-STK11 was more frequently found in KRAS-mutated NSCLC (45). In KRAS-mutant lung adenocarcinoma, STK11 mutation was identified as the most common genomic background for primary resistance to PD-1/PD-L1 inhibitors (46). Interestingly, the acquired resistance after initial response to ICIs in NSCLC patients showed the landscape of genomic changes characterized by loss of putative mutation-associated neoantigens through eliminating tumor subclones or deleting specific chromosomal regions (47). Moreover, High TMB and neoantigen burden(APOBEC, IFNGR1, or VTCN1 mutation) were associated with enhanced efficacy response in NSCLC immunotherapy, but PTEN mutation was associated with non-response to immunotherapy (34).

Impaired antigen processing and presentation have been confirmed as a mechanism for lung cancer with acquired resistance to ICIs (48, 49). Class I human leukocyte antigen (HLA-I/MHC-I) plays a leading role in neoantigen presentation to improve tumor recognition by T cell receptors. HLA gene loss damages the process of neoantigen presentation resulting in immune evasion for tumors (50). The germline HLA-I evolutionary divergence was strongly associated with the survival benefit of metastatic melanoma or NSCLC patients treated with anti-CTLA-4 or anti-PD-1/-PD-L1 (51). 40% of NSCLC carry allele-Specific HLA Loss(HLA-I LOH), which is related to a raised neoantigen burden, PD-L1 positivity, and poor response to ICIs treatment (52, 53). Loss of the B2M gene, an essential chaperone for HLA-I-mediated antigen presentation, formed an immunosuppressive TME characterized by reduced TIL infiltration and conferred resistance to ICIs in NSCLC (49, 54, 55). Interferon-gamma (IFN-γ) has also been identified to stimulate the expression of HLA on NSCLC cells (56, 57). Except for the role of IFN-γ in HLA regulation, IFN-γ signaling is also critical for the initiation of PD-L1 expression in cancer and host cells (58, 59). An analysis of gene expression profiles for pembrolizumab-treated patients found that IFN-γ-related mRNA profile could predict clinical response to PD-1 inhibitor (60). A prospective study about cytokine profiles in NSCLC patients receiving ICI treatment in the second line revealed that patients with elevated expression of IFN-γ significantly benefit from PD-1 blockers (61). Moreover, attenuated antigen presentation was also found in NSCLC with compromised LKB1 and AMPK activity (62). Low expression of KAT2B concurrent with a higher frequency of somatic genes mutation was associated with lower response efficacy to ICIs in NSCLC patients, while KAT2B was linked to IFN-γ regulation, antigen processing, and presentation (63).

It has been reported that the aberrations of MAPK, PI3K, WNT, and IFN signaling pathways may be implicated in the resistance mechanisms of lung cancer immunotherapy (12). As discussed above, genetic mutation EGFR influences tumor immunogenicity by regulating PD-L1 expression and TMB. MAPK and PI3K pathways were involved in the immunotherapy resistance mechanism mediated by EGFR mutation-induced PD-L1 increasing (64). MAPK and PI3K pathways were downstream of the RAS signaling, the activation of which supported intrinsic PD-L1 expression (65, 66). BRAF mutation is a rare form of NSCLC and is part of the MAPK pathway. And BRAF mutation seems to be more associated with high expression of PD-L1(PD-L1 expression ≥50%) than other subtypes (67). In addition, the activity of the MAPK pathway was significant for EGF- and IFNγ-induced PD-L1 expression, contributing to improving response to ICIs in NSCLC (68). The PI3K/AKT/mTOR pathway plays critical roles in multiple biological functions or processes of cancers, and it can be activated by the genetic mutation of EGFR or KRAS in NSCLC (69–71). Significantly, uncontrolled activation of the PI3K/AKT/mTOR pathway can modulate the response to ICIs by driving PD-L1 expression and remodeling the infiltration and function of TIL (71, 72). WNT signaling alterations in NSCLC were also associated with PD-L1 negativity but this altered PD-L1 expression without predictive value for ICIs (73). However, a recent study identified that SCD1-related fatty acids in serum were correlated with the response efficiency of NSCLC patients treated with a PD-1 inhibitor, while WNT signaling was significantly involved in the immunomodulatory function of SCD1 (74). The loss of IFN signaling in tumor cells has been highlighted as a mechanism of the primary and acquired resistance to ICIs in cancer patients (75–77). The production of IFN in TME could induce PD-L1 expression on the surface of tumor cell lines, including NSCLC (78–80). There is still a lack of evidence about the relation between the genomic alterations in IFN signaling and response to ICIs treatment in NSCLC patients.

Macrophages within the TME are designated as tumor-associated macrophages (TAMs). TAMs are the fundamental components within the TME, which perform a crucial function in the immunity remodeling of TME and affect response to immunotherapy (82, 83). A recent single-cell RNA sequencing revealed that the temporal and spatial distribution of macrophages was diverse in the TME of NSCLC and assisted tumor immune escape by initiating regulatory T-cell response (84). There are two classically polarized phenotypes of macrophages, including M1 (immune-activated type) and M2 (immunosuppressive type), and the latter is the main type of TAMs (72). TME enriched with more M2 macrophages is significantly associated with a worse response rate and prognosis for NSCLC patients receiving immunotherapy (85). However, TAMs can be re-engineered into M1-type to increase the response to ICIs treatment (86). In NSCLC, TAMs could promote tumor cell glycolysis by TNFα secretion and facilitate tumor hypoxia by increasing AMPK and PGC-1α, leading to decreased PD-L1 of tumor cells and T-cell infiltration in TME to cause immunotherapy resistance (87). Surprisingly, NSCLC patients enriched with PD-L1⁺ TAM in TME represented better survival for receiving PD-1/PD-L1 blockers (88). Additionally, another study suggested that PD-L1 mainly plays an effect in forming an immunosuppressive M2-type TAM (89). However, M2-type TAM could be reprogrammed into an immune-activated type by anti-PD-L1 treatment but not anti-PD-1 (89). Many genes or signaling pathways are involved in targeting TAM recruitment, activation, and survival, which can be applied as a targeting strategy to improve tumor immunotherapy (90, 91).

NK cells are powerful innate immune cells. NK cells perform a direct tumor-killing effect and indirectly enhance antitumor immunity mediated by T cells. Additionally, NK cells regulate DCs, macrophages, and neutrophils through cytotoxicity and cytokine release (92). Moreover, tumor-infiltrating NK cells could trigger T-cell-mediated immunity by stimulating the recruitment of DCs into TME, conferring improved tumor immune control (93). The infiltration of NK and plasma cells has been defined as a distinct immune subset in NSCLC, contributing to the most favorable prognosis (94). The high infiltration of NK cells in tumor tissue has been confirmed as a biomarker for predicting durable response to ICIs immunotherapy in NSCLC patients (95, 96). It is worth noting that there is a negative relationship between the density of TAMs and NK- and T-cell antitumor activities in NSCLC (97, 98).

Tumor-infiltrating B cells have been identified as the most differential gene between immunotherapy responders and non-responders in patients with melanoma (99). Tertiary lymphoid structures (TLS) were also used as a marker of efficient immunotherapies due to initiating and/or maintaining local and systemic T- and B-cell mediated antitumor activity (100). B cells and plasma cells were found to co-present in TLS, and the abundance of intra-tumoral B cells was also linked to the prediction in the response efficacy of anti-PD-L1 in NSCLC (101, 102). B cells have also been found to exert antigen-presenting function to CD4⁺ TILs in TME to influence prognosis in NSCLC immunotherapy (102).

The presence of CD4⁺ T cells and CD8⁺ T cells in TME was associated with the objective clinical responses to anti-PD-1/PD-L1 blockade in NSCLC (103–107). And the predictive potency of CD4⁺ T cells and CD8⁺ T cells was more prominent in the PD-L1 positive sub-population. PD-1⁺ CD4⁺ T-cell was a negative predictor for immunotherapy availability in advanced NSCLC patients, while PD-1⁺CD8⁺ T-cell was a positive predictor (85, 103–105, 108). To complicate matters, multiple subsets or activity states of CD4⁺ T cells and CD8⁺ T cells have diverse effects on the response to immunotherapy. A single-cell sequencing analyzing T cell composition in NSCLC suggested that patients enriched with “pre-exhausted” CD8⁺ T cells (CD8-C4-GZMK), non-activated Tregs, and activated CD4⁺ cells had a much better prognosis than that in patients enriched with exhausted T cells (CD8-C6-LAYN and CD4-C7-CXCL13) and activated Tregs (109). Epigenetic changes-mediated high exhaustion of T cells is an important resistant mechanism to ICIs treatment (91, 110, 111), and therefore the “pre-exhausted” T cells might be alternative target for improving immunotherapy (109). Another study explored that NSCLC cells-derived exosomal circUSP7 could induce CD8⁺ T cell dysfunction to confer anti-PD-1 resistance (112). Recently, a phase I clinical trial confirmed the safety and feasibility of PD-1-edited T cells in NSCLC (113). Chimeric antigen receptor (CAR)-T cells, as another modified-T cell therapy, has attracted more and more interest in clinical applications as antitumor therapy for various solid tumors, including NSCLC, in recent years (114, 115). In summary, accumulated evidence validated various gene expression disorders or mutations in NSCLC cells impair the activity, tumor-cell-killing function, proliferation, and infiltration of CD8⁺ T-cells in TME, which contribute to exhausted CD8⁺ T-cells for immunotherapy resistance (91, 97, 111, 116–118).

Systemic combination therapies were currently clinically possible strategies for systemic/multiple progression in NSCLC treated with ICIs. BTCRC-LUN15-029(phase 2) suggested that NSCLC patients who progressed after ICIs alone or in combination with chemotherapy could benefit from pembrolizumab plus next-line chemotherapy (135). However, further research is needed to confirm the certain population that can benefit from continued immunotherapy after immunotherapy resistance. Although EGFR/ALK mutation was involved in the mechanism for ICI resistance, EGFR/ALK TKI combination with ICI usually served as a strategy for treating NSCLC patients with acquired EGFR-TKI resistance, not for patients with ICI resistance. However, limited clinical efficacy and a high incidence of treatment-related toxicities did not encourage the further application of this combination. Many clinical trials, such as CheckMate 370 (136), KEYNOTE-021 (137), GEFTREM (138), LUX-Lung-IO (139), have shown that EGFR or ALK inhibitors combined with ICIs were feasible but limited efficacy and more toxicity in treating NSCLC patients with newly diagnosed or refractory or PD after first-line therapy. Furthermore, the growing evidence supported the combination of anti-angiogenic agents and ICIs (140, 141). Ramucirumab combined with pembrolizumab showed an encouraging antitumor activity with acceptable toxicities in NSCLC patients, having progressed on one to three lines of previous therapy (142). A pooled analysis from 23 prospective studies confirmed that ICIs combined with anti-angiogenic agents with favorable antitumor activity and manageable toxic effects might be a new option for NSCLC patients, especially those who received no treatment or chemotherapy intolerance (143). However, especially for treating NSCLC patients with PD after ICI-pretreated, the application of this combination therapy is in the exploring phase. More and more prospective trials have been conducted to identify the efficiency and safety of PD-1/L1 inhibitors plus anti-angiogenesis drugs(Cabozantinib, Sitravatinib, Lenvatinib, Nintedanib) in NSCLC patients previously treated with an ICI (144–150). Cabozantinib is a multi-targeted tyrosine kinase inhibitor (TKI) and reshapes an immune-active TME by the inhibition of MET and TAM receptor kinases (TYRO3, AXL, MER) (151, 152). Sitravatinib is another multi-targeted TKI that inhibits VEGFR, TAM receptors (TYRO3, AXL, MERTK), and Split family receptors, which reduce the proliferation of immunosuppressive cells, initiate T cell infiltration into TME, decrease T cell exhaustion, promote M1 macrophage polarization (153, 154). The anti-VEGFR drug Lenvatinib was found to perform immunoregulation by reducing TAMs infiltration and increasing the activity of CD8⁺ T cells (155, 156). Another anti-angiogenic agent Nintedanib also plays an antitumor immunity by remodeling TME through increased tumoral infiltration of CD8⁺ T cells and granzyme B production (157). Anti-PD-1/PD-L1 and other non-PD-1/PD-L1 blockers are alternative therapeutic strategies for NSCLC patients with immunotherapy resistance. However, Durvalumab plus tremelimumab(anti-CTLA-4) had minimal benefits in NSCLC patients with progression after anti-PD-1 therapy (158). Other trials about combining with Vibostolimab(anti-TIGIT) (NCT02964013, NCT04725188) have not been completed. Moreover, TME modulators, such as MDM2, SEMA4D, and HDAC inhibitors, combined with ICIs were explored to apply in NSCLC patients with refractory and failure after immunotherapy, resulting in safety but limited benefits (159–162). These preliminary results need to be validated in further studies, including a larger sample, adjusted scheme, or changed time point of combined treatment. There are other ongoing clinical trials to investigate the practicability of new drugs such as AK112(dual-targeting PD-1/VEGF) or KN046(dual-targeting PD-1/CTLA-4), or SAR408701(anti-CEACAM5) in antagonizing immunotherapy resistance (163–165). Increasing clinical trials are ongoing to investigate and explore the safety and efficacy of a new combination treatment or drug in NSCLC patients with ICIs resistance ( Table 1 ). We expect these trials’ results will provide additional therapeutic options for ICI-resistant NSCLC in the short term.

Should NSCLC patients with prior anti-PD-1/PD-L1 therapy with a durable response be given a second course of ICI if their disease progresses? For PD-L⁺ NSCLC patients who had progression after 35 cycles/2 years of pembrolizumab, pembrolizumab retreatment showed that the disease control rate was over 77% (166). Another real-world setting also found evidence supporting that nivolumab retreatment could bring better responses in NSCLC patients who had a long-term response to first-course treatment (167). In addition, other reports also found that only limited patients could benefit from immunotherapy retreatment (168, 169), so it is necessary to explore effective predictive biomarkers to screen the optimum population.