Therapeutic vaccines for NSCLC: From mechanisms to clinical translation Authors: Weijing Liu, Jiefei Han, Kailun Fei, Yiting Dong, Man Li, Jie Zhao, Mingzhao Wang, Zhuoran Li, Siyuan Chen, Jiachen Xu, Amit Jain Published: 2026 ABSTRACT Despite advances in targeted therapy and immune checkpoint inhibitors (ICIs), durable disease control remains elusive for most patients with non-small cell lung cancer (NSCLC), particularly those with advanced-stage disease. Therapeutic cancer vaccines offer the potential for sustained immune surveillance by inducing tumor-specific effector and memory T-cell responses. However, their clinical benefits have been inconsistent to date. In advanced disease, rapid tumor progression, an immunosuppressive tumor microenvironment (TME), defective antigen presentation, and immune escape collectively impair the infiltration, persistence, and cytotoxic function of vaccine-induced T cells. These biological factors suggest that vaccine activity may depend on a therapeutic window characterized by lower tumor burden, preserved antigen presentation, and limited immunosuppression. Combination strategies with ICIs, chemotherapy, or radiotherapy are being investigated to increase antigen release and reduce TME-mediated immunosuppression. Vaccine development has also expanded beyond shared tumor-associated antigens to include personalized and shared tumor-specific antigens, as well as multi-epitope designs intended to improve immunogenicity and patient selection. Nonetheless, antigen heterogeneity, immune editing, manufacturing timelines, and the safety of combination regimens remain key translational challenges. Therapeutic vaccines may therefore be viewed as immunomodulatory agents whose long-term value in NSCLC will depend on antigen selection, treatment timing, and integration with multimodal therapy. Introduction Lung cancer remains one of the most commonly diagnosed cancers and the leading cause of cancer-related mortality worldwide, with non-small cell lung cancer (NSCLC) accounting for approximately 80–85% of cases.1,2 Patients with early-stage NSCLC have a relatively favorable prognosis after surgery with or without adjuvant therapy, with 5-year overall survival (OS) rates of 50–70%, whereas the 5-year survival rate in advanced NSCLC remains below 15%.3 The identification of actionable oncogenic driver alterations, such as epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), ROS proto-oncogene 1 receptor tyrosine kinase (ROS1) and MET proto-oncogene receptor tyrosine kinase (MET), together with the characterization of tumor microenvironment (TME) features such as high tumor mutational burden (TMB) and programmed death ligand 1 (PD-L1) expression, has guided the development of targeted therapy and immunotherapy. These approaches have substantially improved clinical outcomes, extending median overall survival (mOS) in EGFR-driven disease from <10 months to nearly 40 months, while 5-year OS rate exceeds 40% in patients with high PD-L1 expression.4 However, primary and acquired resistance remain common, and durable disease control is not achieved in many patients. The efficacy and durability of targeted therapy are limited by primary resistance related to tumor heterogeneity and by acquired resistance mechanisms such as secondary EGFR T790M/C797S mutations and MET amplification.5, 6, 7 Moreover, many patients lack actionable driver alterations. Although targeted agents are available for less common alterations such as rearranged during transfection (RET) and neurotrophic tyrosine receptor kinase (NTRK) fusions and B-Raf proto-oncogene (BRAF) V600E, these alterations occur in only a small proportion of patients.8, 9, 10 Immunotherapy, especially immune checkpoint inhibitors (ICIs), also faces several challenges, including initial nonresponse in immune-cold tumors, which are characterized by low TMB or limited expression of immune checkpoint biomarkers. In addition, there is an immunosuppressive TME characterized by infiltration of regulatory T cells (Tregs) and elevation of transforming growth factor-β (TGF-β) signaling. Thus, the objective response rate (ORR) in unscreened patients remained approximately 20% to 40% with or without combination therapy.11, 12, 13, 14 Current biomarkers also lack sufficient consistency and specificity, leading to suboptimal patient selection and compromising the long-term efficacy of both approaches. Collectively, targeted therapy and ICIs remain dependent on existing molecular vulnerabilities or baseline anti-tumor immunity. Therapeutic cancer vaccines differ from targeted therapies and checkpoint blockade because they actively induce or expand tumor-specific immune responses rather than only inhibiting oncogenic signaling or restoring pre-existing anti-tumor immunity.15 The approach has evolved from empirical immune stimulation and whole-cell vaccines to defined antigen-based platforms, including neoantigen and messenger RNA (mRNA) vaccines (Fig. 1). Unlike prophylactic vaccines targeting exogenous viral antigens, therapeutic vaccines initiate and expand a durable repertoire of tumor-specific T cells, thereby mediating long-term immune surveillance. After vaccination, vaccine-derived antigens are processed and presented by antigen-presenting cells (APCs), particularly dendritic cells (DCs), to prime tumor-specific CD4+ and CD8+ T-cell responses and promote the formation of cytotoxic T lymphocytes (CTLs) and memory T cells.15 A potential advantage of therapeutic vaccination is its capacity to induce or broaden tumor-specific immune responses beyond those present at baseline, including in patients whose tumors lack actionable driver alterations. In addition, advances in mRNA technology during the coronavirus disease 2019 pandemic have accelerated the development of cancer vaccines by enabling efficient antigen delivery through multifunctional vector systems, often combined with immune adjuvants to enhance innate immune activation.16Fig. 1Historical evolution of therapeutic cancer vaccines and renewed development in NSCLC. The timeline summarizes major developments in therapeutic cancer vaccine research, from Coley’s toxins and early autologous vaccine approaches to the first FDA-approved therapeutic cancer vaccine, Sipuleucel-T, and the oncolytic virus therapy T-VEC. Subsequent progress in checkpoint blockade, neoantigen discovery, and mRNA-LNP delivery has supported a shift toward personalized, platform-based, and combination vaccine strategies. In NSCLC, negative phase III results with shared tumor-associated antigen vaccines, including MAGE-A3 and tecemotide, redirected the field toward more precise antigen selection and better-defined clinical settings. CTLA-4, Cytotoxic T-lymphocyte-associated protein 4; FDA, Food and Drug Administration; LNP, Lipid nanoparticle; MAGE-A3, Melanoma-associated antigen A3; mRNA, Messenger RNA; NSCLC, Non-small cell lung cancer; TAA, Tumor-associated antigen.Fig 1 dummy alt text Historical evolution of therapeutic cancer vaccines and renewed development in NSCLC. The timeline summarizes major developments in therapeutic cancer vaccine research, from Coley’s toxins and early autologous vaccine approaches to the first FDA-approved therapeutic cancer vaccine, Sipuleucel-T, and the oncolytic virus therapy T-VEC. Subsequent progress in checkpoint blockade, neoantigen discovery, and mRNA-LNP delivery has supported a shift toward personalized, platform-based, and combination vaccine strategies. In NSCLC, negative phase III results with shared tumor-associated antigen vaccines, including MAGE-A3 and tecemotide, redirected the field toward more precise antigen selection and better-defined clinical settings. CTLA-4, Cytotoxic T-lymphocyte-associated protein 4; FDA, Food and Drug Administration; LNP, Lipid nanoparticle; MAGE-A3, Melanoma-associated antigen A3; mRNA, Messenger RNA; NSCLC, Non-small cell lung cancer; TAA, Tumor-associated antigen. According to ClinicalTrials.gov, as of June 2026, there were nearly 200 clinical trials investigating therapeutic cancer vaccines in NSCLC. Current clinical developments have broadly followed several directions. First, vaccines are increasingly being evaluated in combination with ICIs, chemotherapy, radiotherapy, or other treatments to address the limited activity of vaccine monotherapy.17, 18, 19, 20, 21 Second, both personalized and ``off-the-shelf'' approaches are being pursued to balance antigen specificity with manufacturing feasibility.21,22 Third, investigators are placing greater emphasis on defined clinical contexts, including maintenance therapy after induction treatment, perioperative or neoadjuvant settings, and late-line disease control.23, 24, 25 Although most candidates remain in early-phase development, accumulating data support the feasibility and immunogenicity of these strategies. This review summarizes recent progress in therapeutic cancer vaccines for NSCLC and examines how immunological mechanisms, platform characteristics, and antigen selection may inform their evaluation across different treatment settings. Mechanisms Immunological basis of therapeutic cancer vaccines in NSCLC Therapeutic cancer vaccines are designed to induce or amplify tumor-specific T-cell immunity by delivering tumor antigens together with immune-stimulatory signals. The goal is to induce antigen-specific CTLs and durable immune memory that may contribute to the control of recurrence and tumor evolution. In NSCLC, vaccine efficacy depends on efficient antigen processing and presentation by APCs, effective T-cell priming, intratumoral T-cell infiltration, and sustained cytotoxic activity within an immunosuppressive TME (Fig. 2).Fig. 2Immunological sequence and resistance barriers of therapeutic cancer vaccines in NSCLC. (A) Different vaccine platforms deliver antigens through distinct intracellular or exogenous routes. (B) Mature APCs, including endogenous DCs or ex vivo antigen-loaded DC vaccine products, prime naïve CD8+ and CD4+ T cells in tumor-draining lymph nodes. (C) Vaccine-elicited CTLs must infiltrate NSCLC lesions, recognize tumor pMHC-I and retain cytotoxic function. APC, Antigen-presenting cell; APM, Antigen-processing and presentation machinery; β2M, β2-microglobulin; CAF, Cancer-associated fibroblast; CD, Cluster of differentiation; CTL, Cytotoxic T lymphocyte; CXCL9/10, C-X-C motif chemokine ligand 9/10; CXCL12, C-X-C motif chemokine ligand 12; DC, Dendritic cell; DNA, Deoxyribonucleic acid; ECM, Extracellular matrix; ER, Endoplasmic reticulum; HLA-I, Human leukocyte antigen class I; IFN-γ, Interferon-γ; IFNGR, Interferon-γ receptor; IL-2, Interleukin 2; JAK, Janus kinase; LAG-3, Lymphocyte-activation gene 3; MDSC, Myeloid-derived suppressor cell; MHC, Major histocompatibility complex; mRNA, Messenger RNA; NSCLC, Non-small cell lung cancer; PD-1, Programmed cell death protein 1; pMHC, Peptide–MHC complex; STAT, Signal transducer and activator of transcription; TAM, Tumor-associated macrophage; TAP, Transporter associated with antigen processing; TCR, T-cell receptor; TIM-3, T-cell immunoglobulin and mucin-domain containing-3; TNF-α, Tumor necrosis factor-α.Fig 2 dummy alt text Immunological sequence and resistance barriers of therapeutic cancer vaccines in NSCLC. (A) Different vaccine platforms deliver antigens through distinct intracellular or exogenous routes. (B) Mature APCs, including endogenous DCs or ex vivo antigen-loaded DC vaccine products, prime naïve CD8+ and CD4+ T cells in tumor-draining lymph nodes. (C) Vaccine-elicited CTLs must infiltrate NSCLC lesions, recognize tumor pMHC-I and retain cytotoxic function. APC, Antigen-presenting cell; APM, Antigen-processing and presentation machinery; β2M, β2-microglobulin; CAF, Cancer-associated fibroblast; CD, Cluster of differentiation; CTL, Cytotoxic T lymphocyte; CXCL9/10, C-X-C motif chemokine ligand 9/10; CXCL12, C-X-C motif chemokine ligand 12; DC, Dendritic cell; DNA, Deoxyribonucleic acid; ECM, Extracellular matrix; ER, Endoplasmic reticulum; HLA-I, Human leukocyte antigen class I; IFN-γ, Interferon-γ; IFNGR, Interferon-γ receptor; IL-2, Interleukin 2; JAK, Janus kinase; LAG-3, Lymphocyte-activation gene 3; MDSC, Myeloid-derived suppressor cell; MHC, Major histocompatibility complex; mRNA, Messenger RNA; NSCLC, Non-small cell lung cancer; PD-1, Programmed cell death protein 1; pMHC, Peptide–MHC complex; STAT, Signal transducer and activator of transcription; TAM, Tumor-associated macrophage; TAP, Transporter associated with antigen processing; TCR, T-cell receptor; TIM-3, T-cell immunoglobulin and mucin-domain containing-3; TNF-α, Tumor necrosis factor-α. Effective vaccination depends on a common sequence of immune events. Mature DCs must first prime naïve CD8+ and CD4+ T cells in lung tumor-draining lymph nodes.26 Effective priming is required for CTL expansion, CD4+ helper responses, and memory T-cell formation, rather than a transient peripheral immune response alone. Activated CTLs must then enter the NSCLC TME, recognize tumor-derived peptide–MHC class I complex (pMHC-I) complexes, and retain cytotoxic function through perforin/granzyme release and effector cytokines such as inteferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α), which together promote tumor-cell killing and local anti-tumor activity. Tumor cell killing releases additional antigens, which can be captured and processed by APCs, leading to secondary immune activation and epitope spreading.27 This process may broaden the anti-tumor T-cell repertoire beyond the initially selected vaccine antigens. However, several mechanisms in NSCLC can restrict epitope spreading and the amplification of vaccine-induced immunity. Loss of HLA-I expression or defects in antigen-processing and presentation machinery (APM), including β2-microglobulin (B2M) or transporter associated with antigen processing (TAP) abnormalities, impair tumor antigen presentation.28 Although CTL-derived IFN-γ can contribute to anti-tumor control, tumor-cell IFN-γ pathway resistance, including impaired interferon-γ receptor–Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling, can reduce HLA-I upregulation, weaken C-X-C motif chemokine ligand (CXCL) 9/10-mediated T-cell recruitment, and blunt adaptive immune pressure.29 Myeloid-derived suppressive cells (MDSCs) and tumor-associated macrophages (TAMs) suppress T-cell function, whereas cancer-associated fibroblasts (CAF)-derived CXCL12 signaling and extracellular matrix (ECM) remodeling restrict CTL infiltration, resulting in CTL exclusion and dysfunction.30 In addition, chronic antigen stimulation and signaling through inhibitory pathways, including programmed cell death protein 1 (PD-1)/programmed death ligand 1 (PD-L1), lymphocyte-activation gene 3 (LAG-3), and T-cell immunoglobulin and mucin-domain containing-3 (TIM-3), promote T-cell dysfunction and exhaustion. Thus, a measurable systemic immune response does not necessarily indicate effective intratumoral immunity.31,32 Peripheral immune readouts alone are therefore insufficient to assess vaccine activity. Antigen-specific CD4+ and CD8+ T-cell frequencies, effector cytokine production, and in vitro cytotoxicity reflect systemic immune activation but do not fully capture intratumoral T-cell infiltration, antigen recognition, or durable anti-tumor activity. This distinction is particularly important in NSCLC, where immune-excluded tumors, stromal and myeloid immunosuppression, and defects in antigen presentation can limit the in vivo anti-tumor activity of vaccine-induced T cells.33 Vaccine design dimensions The key vaccine design dimensions discussed in this section are summarized in Fig. 3, including technological platform, antigen specificity, and product availability. Platform selection affects antigen delivery, innate immune activation, CD4+/CD8+ bias, safety, and manufacturing complexity; antigen specificity affects tumor selectivity, immune escape, and HLA dependence; product format determines whether vaccination can be standardized or requires individualized manufacturing.Fig. 3Classification of therapeutic cancer vaccines based on technological platform, antigen specificity, and product availability. The left panel shows four major vaccine platforms: nucleic acid-based vaccines, peptide/protein vaccines, cell-based vaccines, and viral-vector vaccines. The middle panel classifies antigen specificity into TAA-based vaccines targeting shared self-derived antigens and TSA/neoantigen-based vaccines targeting mutation-derived peptides. The right panel contrasts off-the-shelf products based on standardized shared antigen sets with personalized products requiring tumor sampling, sequencing, neoantigen prediction, and individualized manufacturing. CEA, Carcinoembryonic antigen; circRNA, Circular RNA; DC, Dendritic cell; DNA, Deoxyribonucleic acid; EGFR, Epidermal growth factor receptor; HER2, Human epidermal growth factor receptor 2; indel, Insertion or deletion; KRAS, Kirsten rat sarcoma viral oncogene homolog; MAGE-A3, Melanoma-associated antigen A3; MHC, Major histocompatibility complex; mRNA, Messenger RNA; MUC1, Mucin 1; MYA, Myxoma virus Ankara; NSCLC, Non-small cell lung cancer; NY-ESO-1, New York esophageal squamous cell carcinoma 1; pMHC, Peptide–MHC complex; RNA-seq, RNA sequencing; SLP, Synthetic long peptide; SNV, Single-nucleotide variant; TAA, Tumor-associated antigen; TP53, Tumor protein p53; TSA, Tumor-specific antigen; WES, Whole-exome sequencing.Fig 3 dummy alt text Classification of therapeutic cancer vaccines based on technological platform, antigen specificity, and product availability. The left panel shows four major vaccine platforms: nucleic acid-based vaccines, peptide/protein vaccines, cell-based vaccines, and viral-vector vaccines. The middle panel classifies antigen specificity into TAA-based vaccines targeting shared self-derived antigens and TSA/neoantigen-based vaccines targeting mutation-derived peptides. The right panel contrasts off-the-shelf products based on standardized shared antigen sets with personalized products requiring tumor sampling, sequencing, neoantigen prediction, and individualized manufacturing. CEA, Carcinoembryonic antigen; circRNA, Circular RNA; DC, Dendritic cell; DNA, Deoxyribonucleic acid; EGFR, Epidermal growth factor receptor; HER2, Human epidermal growth factor receptor 2; indel, Insertion or deletion; KRAS, Kirsten rat sarcoma viral oncogene homolog; MAGE-A3, Melanoma-associated antigen A3; MHC, Major histocompatibility complex; mRNA, Messenger RNA; MUC1, Mucin 1; MYA, Myxoma virus Ankara; NSCLC, Non-small cell lung cancer; NY-ESO-1, New York esophageal squamous cell carcinoma 1; pMHC, Peptide–MHC complex; RNA-seq, RNA sequencing; SLP, Synthetic long peptide; SNV, Single-nucleotide variant; TAA, Tumor-associated antigen; TP53, Tumor protein p53; TSA, Tumor-specific antigen; WES, Whole-exome sequencing. Technological platform NSCLC vaccines can be developed using several platform types. These platforms differ in antigen source, delivery route, innate immune activation, personalization potential, and manufacturing burden. Nucleic acid-based vaccines Nucleic acid-based vaccines include DNA, mRNA, and emerging circular RNA platforms. After cellular uptake, plasmid DNA must enter the nucleus for transcription. The encoded antigen is subsequently translated in the cytoplasm and can be processed through the proteasome–TAP–MHC-I pathway for presentation to CD8+ T cells. Released or secreted antigens may also be taken up by APCs and presented through MHC-II to CD4+ helper T cells. Unmethylated cytosine-phosphate-guanine motifs in plasmid DNA can activate Toll-like receptor 9 signaling, increasing inflammatory cytokine production and costimulatory molecule expression and thereby supporting CD4+ and CD8+ T-cell priming.34 However, cellular uptake and nuclear entry remain inefficient in vivo, limiting antigen expression and immunogenicity. More than 95% of intramuscularly injected DNA remains in the muscle interstitium and is rapidly degraded. Only a small fraction reaches the cytoplasm, and nuclear entry is further limited by plasmid size, resulting in low antigen expression and weak immunogenicity. Some physical and nanoparticle-based delivery methods can improve DNA uptake, but these approaches add complexity and their in vivo transfection efficiency remains limited.35 Among RNA-based platforms, mRNA vaccines are currently the most widely developed form. Unlike DNA, mRNA can be translated directly in the cytoplasm without nuclear entry and can produce antigen in both dividing and nondividing cells. After translation, the encoded antigen is degraded by the proteasome, transported by TAP, and loaded onto MHC-I molecules. Meanwhile, released or secreted antigens can be taken up by APCs and presented through MHC-II. mRNA platforms generally produce antigen more rapidly and at higher levels than DNA platforms and avoid the risks of genomic integration and insertional mutagenesis. Lipid nanoparticle (LNP) delivery improves mRNA stability and cellular uptake and has supported the clinical development of mRNA cancer vaccines.36 Circular RNA may provide greater stability and more sustained antigen expression than linear mRNA because of its covalently closed structure. Preclinical studies have shown increased nuclease resistance and prolonged protein expression, but the clinical development of circular RNA cancer vaccines remains at an early stage.37 Nucleic acid platforms support rapid sequence design and modular manufacturing, allowing the incorporation of shared tumor-associated antigens (TAAs) or personalized neoantigens, although delivery efficiency and platform-specific reactogenicity remain important considerations. Protein/peptide-based vaccines Protein vaccines have become mature platforms for the prevention of infectious diseases by delivering complete recombinant proteins or protein domains containing multiple linear and conformational epitopes. This broad epitope coverage can effectively induce CD4+ T-cell and humoral immune responses. However, intact proteins enter the cytosol inefficiently and are poorly cross-presented, which limits CD8+ T-cell activation.38 Peptide-based cancer vaccines are usually composed of epitopes derived from tumor antigens. These epitopes are selected according to predicted immunogenicity and compatibility with HLA alleles. Peptide length critically determines the quality of the immune response. Short peptides (8–11 amino acids) can bind directly to MHC-I molecules and activate CD8+ T cells, but rapid degradation, HLA restriction, and limited CD4+ T-cell support may reduce the durability of the response. By contrast, synthetic long peptides (SLPs, 15–30 amino acids) require internalization and processing by DCs, enabling presentation via both MHC-II to CD4+ T cells and cross-presentation via MHC-I to CD8+ T cells. This dual presentation can coordinate CD8+ effector responses with CD4+ helper support and favor memory formation. Because protein and SLP vaccines lack intrinsic immunostimulatory signals, they are commonly formulated with adjuvants to support DC activation and T-cell priming.39, 40, 41 Cell-based vaccines Cell-based vaccines mainly include whole-tumor cell vaccines (WTCVs) and DC vaccines and may broaden antigen coverage through the presentation of multiple tumor antigens. Autologous or allogeneic inactivated WTCVs can provide multiple tumor antigens for uptake and cross-presentation by DCs. However, they may carry intrinsic immunosuppressive signals and allogeneic features, which can reduce cross-presentation efficiency and make these platforms highly dependent on potent adjuvants.42 Genetic modification methods, such as TGF-β2 antisense oligonucleotides, have been used to attenuate immunosuppressive signaling in allogeneic inactivated tumor cells.43 In addition to direct infusion of tumor cells, DCs loaded with antigens or tumor lysates in vitro can be used to achieve multi-epitope antigen presentation, promote pMHC-I/II presentation, and simultaneously provide three signals to activate T cells.44 This approach increases preparation complexity and limits large-scale production.45,46 Viral vector-based vaccines Viral vector (VV)-based vaccines use recombinant viruses to deliver tumor antigen-encoding genes into host cells. Replication-defective adenoviral (Ad) vectors and poxviral vectors such as modified vaccinia Ankara virus (MVA) express the encoded antigens without productive viral replication, supporting intracellular processing and MHC-I presentation to CD8+ T cells.47,48 In contrast, oncolytic viruses retain tumor-selective replication competence, enabling dual functionality: direct oncolytic tumor cell killing and in situ immune activation through immunogenic cell death and local antigen release.49 One of the main limitations of the VV platforms is pre-existing or treatment-induced anti-vector immunity, which can attenuate antigen delivery upon repeated administration. Strategies to mitigate this include heterologous prime-boost regimens and engineered chimeric vectors (e.g., Ad5/35) designed to circumvent pre-existing neutralizing antibodies.50 Antigen specificity Vaccine efficacy is strongly influenced by the selected antigen. TAA-based vaccines target self-derived antigens that are overexpressed or aberrantly expressed in tumor cells but may also be present at low levels in normal tissues. Based on expression patterns and tissue distribution, TAAs can be broadly classified into several categories.51 Overexpressed self-proteins, including EGFR, human epidermal growth factor receptor 2 (HER2), and survivin, may carry a risk of on-target, off-tumor immune effects because they are also expressed at lower levels in normal tissues.52, 53, 54 Cancer-testis antigens, such as melanoma-associated antigen A3 (MAGE-A3), New York esophageal squamous cell carcinoma 1 (NY-ESO-1) and preferentially expressed antigen in melanoma (PRAME), are normally confined to immune-privileged germline tissues but are aberrantly activated in a wide range of cancers, rendering them attractive vaccine targets.55, 56, 57 Tissue differentiation antigens, such as carcinoembryonic antigen (CEA) and mucin 1 (MUC1), are also expressed in normal epithelial tissues but are frequently overexpressed or aberrantly expressed in tumors.58,59 However, because TAAs are not strictly tumor-specific, both central and peripheral tolerance mechanisms may constrain the magnitude and durability of TAA-directed immune responses, representing a fundamental immunological limitation of this antigen class. In contrast, tumor-specific antigen (TSA)-based vaccines target antigens that are absent from normal tissues. These include mutation-derived neoantigens arising from nonsynonymous single-nucleotide variants (SNVs), insertions or deletions, frameshift mutations, and gene fusions, as well as tumor-specific peptides generated by aberrant splicing or other non-canonical processes. Candidate neoantigens are usually selected by integrating tumor–normal whole-exome sequencing (WES), RNA sequencing, and HLA typing: WES identifies somatic mutations, RNA sequencing confirms mutant transcript expression, and HLA typing supports prediction of peptide–MHC binding. Because TSAs are non-self antigens, central tolerance is less likely to delete high-avidity TSA-reactive T cells, allowing more tumor-specific responses than most TAAs.60 TSA-based vaccines include personalized neoantigen vaccines tailored to each patient’s mutational profile and shared neoantigen vaccines targeting selected recurrent driver mutations, such as KRAS alterations, in molecularly defined patient subsets.19,61 Their use remains limited by imperfect prediction, tumor heterogeneity, HLA restriction, manufacturing time, and scalability. Product availability and manufacturing timelines For clinical translation, off-the-shelf and personalized vaccines differ mainly in manufacturing time, quality control, and feasible treatment setting. Off-the-shelf vaccines, including most TAA-based peptide vaccines and recombinant subunit vaccines, are based on predefined antigens, standardized formulations, and scalable manufacturing. They are therefore easier to produce, store, distribute, and combine with other treatments.62 However, off-the-shelf vaccines usually rely on shared TAAs, making them vulnerable to interpatient and intratumoral heterogeneity. Their self-antigen origin also means that central and peripheral tolerance may limit the frequency and avidity of vaccine-reactive T cells. Personalized vaccines, including autologous DC therapies and patient-specific neoantigen mRNA or peptide vaccines, are designed around the antigenic profile of an individual tumor. For personalized neoantigen vaccines, production requires tumor sampling, sequencing, computational antigen prediction, and individualized manufacturing. The resulting turnaround time typically ranges from 4 to 12 weeks, with mRNA-based platforms currently offering the shortest timelines of approximately 4–6 weeks.63 This delay may limit their use in patients with rapidly progressive disease. Immune heterogeneity in NSCLC shapes vaccine responsiveness NSCLC exhibits significant immunologic heterogeneity across disease stage, histologic subtype, smoking status, and oncogenic driver subtypes, all of which may influence vaccine responsiveness.64 Compared with advanced disease, early-stage NSCLC generally has a lower tumor burden and may have a less immunosuppressive TME. However, tumor-infiltrating lymphocytes may already show heterogeneous functional states, ranging from partially exhausted effector cells to quiescent or poorly cytotoxic populations.65, 66, 67, 68 In this setting, lower tumor burden and a less suppressive TME may allow more time for vaccine-induced T-cell expansion and memory formation against minimal residual disease (MRD) after definitive local therapy.22,69,70 As NSCLC progresses, tumor burden and intratumoral heterogeneity often increase, and the immune microenvironment becomes more suppressive. At the cellular level, the accumulation of Tregs and myeloid-derived suppressor cells (MDSCs), together with progressive effector T-cell exhaustion, contributes to an increasingly suppressive TME.71, 72, 73 At the molecular level, disease progression is associated with HLA downregulation and increased secretion of immunosuppressive cytokines such as TGF-β.73, 74, 75, 76 In this setting, vaccine-induced T cells face multiple barriers to tumor infiltration, persistence, and cytotoxic function. Histologic subtype is also associated with differences in tumor immunogenicity. Lung squamous cell carcinoma (LUSC) generally has a higher TMB than lung adenocarcinoma (LUAD) and may therefore harbor a larger pool of candidate neoantigens.77 However, TMB does not necessarily reflect the immune context. Gene-expression analyses showed low immune-cell expression in proximal-proliferative adenocarcinoma and higher expression in secretory squamous tumors, while nonsilent mutation burden was not correlated with immune-cell expression across molecular subtypes.78 Clinical evidence for differential vaccine responsiveness remains limited. A recent meta-analysis reported an OS benefit in LUSC but not in LUAD, although the included studies differed in vaccine platform and treatment setting.24 Smoking history and oncogenic driver status further shape the immune context of NSCLC. Smoking-associated tumors often carry higher TMB and may generate more candidate neoantigens, whereas oncogene-driven tumors, such as EGFR-mutant or ALK-rearranged NSCLC, frequently show lower T-cell infiltration, reduced inflammatory signaling, and a more immune-excluded phenotype.79 Vaccine response may also be influenced by host-level and tissue-context factors, including baseline immune competence, prior antibiotic exposure, microbiome-related immune tone, metastatic organ site, and the degree of myeloid- or neutrophil-dominant inflammation.80 Beyond canonical oncogenic drivers, tumor-suppressor alterations further shape the immune context of NSCLC. STK11/LKB1 loss is associated with neutrophil recruitment, reduced tumor-infiltrating lymphocytes, impaired T-cell activity and lower PD-L1 expression, while KEAP1 alterations are linked to a myeloid-enriched TME with reduced CD8+ T-cell infiltration.81, 82, 83 These genotype-associated features may limit the intratumoral accumulation, persistence and cytotoxic function of vaccine-induced T cells. Therefore, STK11 and KEAP1 alterations may be considered potential modifiers of vaccine responsiveness. Future genotype-stratified studies could examine whether strategies targeting suppressive myeloid populations or supporting CD4+ and CD8+ T-cell activity modify vaccine responses in these molecular subgroups. Immune kinetic mismatch constrains clinical efficacy Therapeutic cancer vaccines are also limited by immune kinetics. Vaccine-elicited T-cell responses can be detected relatively early after vaccination, but their translation into effective intratumoral cytotoxicity and clinical benefit usually takes longer.84,85 By comparison, targeted therapy such as osimertinib and cytotoxic chemotherapy such as docetaxel can produce radiographic tumor responses within weeks.86,87 This timing gap is especially relevant in advanced NSCLC with high tumor burden or rapidly progressive disease, where tumor growth may outpace the development of a sufficiently strong vaccine-induced antitumor response. Vaccine monotherapy may be insufficient when tumor progression outpaces the development of vaccine-induced immunity. In settings with lower tumor burden or controlled disease, such as the postoperative stage or stable disease after induction therapy, vaccine-induced T-cell responses may have more time to expand, mature, and persist.88 From biological mechanisms to clinical application in NSCLC The clinical efficacy of therapeutic cancer vaccines in NSCLC depends on the interaction between tumor burden, immune function, antigen profile, and disease dynamics. These biological parameters vary across disease stages and may inform the evaluation of vaccination in different clinical settings (Fig. 4).Fig. 4Biological window hypothesis for therapeutic cancer vaccine evaluation in NSCLC. Clinical contexts are positioned by biological barriers to vaccination and the time available for immune priming relative to disease kinetics. Perioperative/MRD settings may provide a more permissive window, whereas maintenance, oligometastatic, and advanced first-line settings are conditional on disease control and immune context. Later-line progressive disease represents a more challenging window. MRD, Minimal residual disease; NSCLC, Non-small cell lung cancer.Fig 4 dummy alt text Biological window hypothesis for therapeutic cancer vaccine evaluation in NSCLC. Clinical contexts are positioned by biological barriers to vaccination and the time available for immune priming relative to disease kinetics. Perioperative/MRD settings may provide a more permissive window, whereas maintenance, oligometastatic, and advanced first-line settings are conditional on disease control and immune context. Later-line progressive disease represents a more challenging window. MRD, Minimal residual disease; NSCLC, Non-small cell lung cancer. In early-stage and perioperative settings, lower tumor burden and a less established immunosuppressive microenvironment may favor durable vaccine-mediated immune memory against MRD. As disease progresses, increasing tumor burden and TME suppression can restrict endogenous and vaccine-induced T-cell activity. Combination strategies may therefore be needed to reduce tumor mass, relieve immune suppression, or increase antigen release.89 ICIs may complement vaccination by relieving inhibitory signaling and sustaining vaccine-primed T-cell activity. Chemotherapy and radiotherapy may also support vaccination by inducing tumor-cell death, antigen release, and transient TME remodeling, although their timing and intensity require careful evaluation because treatment-related lymphocyte depletion may impair vaccine-induced immunity.90 In oncogene-driven NSCLC, targeted therapy can rapidly reduce tumor burden and may create a window for vaccination, but resistant tumors often undergo clonal diversification through secondary resistance mutations, bypass pathway activation, histologic transformation, and changes in antigen presentation or immune contexture.19,61,91 Although original driver alterations are often retained, resistant subclones may acquire additional molecular and immunological features that reduce the durability of single-target treatment. Therefore, multi-epitope and personalized vaccine strategies may broaden immune pressure across multiple tumor antigens and help limit evolutionary escape.92,93 These mechanistic considerations provide a framework for selecting clinical settings in which vaccination is biologically plausible. The following sections review clinical evidence according to treatment setting. Current clinical landscape of cancer vaccines in NSCLC Clinical development of therapeutic cancer vaccines in NSCLC has expanded substantially over the past decade, driven by advances in antigen discovery, platform technology, and combination immunotherapy strategies. Trials now span second- and later-line treatment, first-line therapy, perioperative treatment, and maintenance. Early-phase studies have mainly established safety and immunogenicity, whereas more recent programs have begun to test defined clinical endpoints in selected populations. The following sections summarize completed and active clinical evidence organized by treatment setting. Table 1 provides a comprehensive overview of ongoing and upcoming clinical trials listed as Recruiting, Not yet recruiting, or Active, not recruiting.Table 1Ongoing and upcoming clinical trials of therapeutic cancer vaccines in NSCLC across lines of therapy.Table 1 dummy alt textNCT numberStudy statusPhaseClinical scenarioVaccine nameTargetVaccine typeDiseaseTreatment regimen(s)Efficacy and safety outcomesCountryNCT06928922RecruitingEarly Phase ISecond-/later-lineBMD006Tumor tissue expression of ≥1 of nine TAAs: NY-ESO-1, MAGE-A3, TPTE, tyrosinase, KK-LC-1, MAGE-A4, CLDN6, PRAME, or MAGE-C1Inhaled mRNA dry powder vaccineAdvanced lung cancer (driver gene negative) or advanced solid tumors with lung metastasis; TAA expression requiredBMD006 alone; BMD006 + PD-1 antibody; BMD006 + PD-1/VEGF antibodyEfficacy: Not yet available; Safety: Not yet availableChinaNCT06472245RecruitingPhase IIISecond-/later-lineOSE2101 (Tedopi)P53, HER2, CEA, MAGE-A2, MAGE-A3 (+PADRE)Peptide vaccine (HLA-A2-restricted epitopes; Montanide ISA 51 adjuvant)Metastatic NSCLC (squamous or non-squamous), HLA-A2 positive, secondary resistance to immune checkpoint inhibitorOSE2101 monotherapy vs. docetaxelEfficacy: Not yet available; Safety: Not yet availableMultinational (Canada, Hungary, Greece, Germany, Italy, …)NCT06329908RecruitingEarly Phase ISecond-/later-lineNeo-DCVac (LG002)Personalized neoantigensDendritic cell vaccine (neoantigen-loaded)Advanced lung cancer resistant to ICIs after 1L PD-1/PD-L1 immunochemotherapyNeo-DCVac + immune checkpoint inhibitor(s)Efficacy: Not yet available; Safety: Not yet availableChinaNCT05950139RecruitingPhase I/IIMaintenanceALK peptide vaccineMutant ALK (acquired resistance mutations)Peptide vaccineStage IV ALK+ NSCLC on ALK TKI with stable disease ≥4 monthsALK peptide vaccine + ongoing ALK targeted therapyEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT05886439RecruitingPhase ISecond-/later-lineLK101 injectionPersonalized neoantigenDendritic cell vaccine (personalized neoantigen-pulsed)Stage IIIB–IV NSCLC after anti-PD-1/PD-L1 failure (plus ES-SCLC cohort; plus maintenance cohort)LK101 + PD-1/PD-L1 mAb (pembrolizumab/durvalumab/tislelizumab)Efficacy: Not yet available; Safety: Not yet availableChinaNCT05751798RecruitingPhase I/IIFirst-lineOSE2101 (Tedopi)TAAs (HLA-A2-restricted peptides)Peptide vaccineMetastatic (stage IV) NSCLC, PD-L1 TPS ≥50% (HLA-A2± cohorts)OSE2101 + OSE-279 (anti-PD-1 mAb)Efficacy/Safety: No NSCLC-specific data available for the OSE-279 + OSE2101 cohort. OSE-279 monotherapy Part A: 1 CR, 4 PRs, and 7 SD among 20 advanced solid-tumor patients208Multinational (Belgium, France, Spain)NCT05269381RecruitingPhase I/IISecond-/later-line, adjuvantPersonalized neoantigen peptide-based vaccine (PNeoVCA)NeoantigensPeptide neoantigen vaccineAdvanced solid tumors including metastatic NSCLC (stage III/IV)Personalized neoantigen peptide vaccine + pembrolizumabNo NSCLC-specific efficacy/safety data availableUnited StatesNCT05254184RecruitingPhase IFirst-linePooled mutant KRAS-targeted long peptide vaccineKRAS mutations (G12C/G12V/G12D/G12A/G13D/G12R)Long peptide vaccineUnresectable stage III/IV KRAS-mutated NSCLCKRAS peptide vaccine + poly-ICLC + nivolumab + ipilimumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT05142189RecruitingPhase IFirst-line, later-line, neoadjuvant/adjuvant, maintenanceBNT116CLDN6, KK-LC-1, MAGE-A3, MAGE-A4, MAGE-C1, PRAMEmRNA vaccineAdvanced/metastatic NSCLC (multi-cohort)BNT116 alone; BNT116 + cemiplimab; BNT116 + docetaxel (and other cohorts)First-line (Frail, PD-L1 ≥ 1%)21: Efficacy: ORR 45.0% (9/20), DCR 80.0% (16/20), mPFS 9.9 months. Safety: BNT116 TRAEs 100.0% (Grade ≥3: 15.0%), cemiplimab TRAEs 65.0% (Grade ≥3: 10.0%). Second/later-Line (post-PD-1, PD-L1 ≥ 50%)165: Efficacy: ORR 10.0% (2/20), DCR 80.0% (16/20), mPFS 5.5 months. Safety: Overall TEAEs 100.0% (20/20), Grade 3 TEAEs 55.0% (11/20), no Grade 4/5 events. Second/Later-Line (post-PD-1 + Chemo, BNT116 + docetaxel)209: Efficacy: ORR 35.0% (7/20), DCR 85.0% (17/20). Safety: Serious TEAEs 50.0% (10/20; 3 related to BNT116, 3 to docetaxel), no DLTs.Consolidation (post-CRT)210: Efficacy: 6-month EFS 72.1%, 6-month OS 95.0%; median follow-up 6.3 months. Safety: BNT116 TRAEs 100.0% (20/20), cemiplimab TRAEs 75.0% (15/20), discontinuation due to AEs 25.0% (5/20).Late-Line (>3 prior lines; BNT116 ± sequential cemiplimab)211: Efficacy: ORR 0 (0/10 evaluable), DCR 60.0% (6/10), median duration of disease control 1.53 months. Safety: Overall TEAEs 100.0% (18/18), serious TEAEs 27.8% (5/18), no DLTsMultinational (Hungary, Turkey, Spain, Poland, …)NCT05104515RecruitingPhase ISecond-/Later-lineOVM-200SurvivinRecombinant overlapping peptide vaccineLocally advanced or metastatic NSCLC (≥1 prior line)OVM-200 aloneSecond-/later-Line (mixed solid-tumor cohort; n = 12, including 1 patient with NSCLC): Efficacy: SD 50.0% (6/12). Safety: no DLTs or serious ADRs; all reported AEs were Grade 1 injection-site reactionsUnited KingdomNCT05098210RecruitingPhase ISecond-/later-lineNeoantigen peptide vaccine (personalized)NeoantigenPeptide neoantigen vaccineStage III–IV NSCLC (pretreated; plus melanoma/breast cohorts)Neoantigen peptide vaccine + poly-ICLC + nivolumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT04147078RecruitingPhase IAdjuvantNeoantigen-primed dendritic cell vaccineNeoantigenDendritic cell vaccine (peptide-primed)Postoperative locally advanced NSCLC (with standard adjuvant treatment)Neoantigen-primed DC vaccine (subcutaneous) after standard adjuvant treatmentNo NSCLC-specific efficacy/safety data availableChinaNCT01720836RecruitingPhase I/IIAdjuvantMUC1 100-mer peptide vaccine + poly-ICLC (Hiltonol)MUC1Peptide vaccineStage IA–IIIB NSCLC (localized/locally advanced)MUC1 peptide vaccine + poly-ICLC after surgery ± adjuvant chemotherapy or after chemoradiationEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT00722228RecruitingPhase I/IISecond-/Later-lineAutologous/allogeneic whole cell cancer vaccine (DNP-conjugated, irradiated tumor cells)Whole tumor cells (autologous/allogeneic)Cell-based whole tumor cell vaccineMetastatic epithelial cancers including lung cancer; failed ≥1 chemotherapyAutologous or allogeneic tumor cell vaccine (subcutaneous)Efficacy: Not yet available; Safety: Not yet availableIsraelNCT07561723RecruitingPhase ISecond-/Later-lineIMP3-saRNA (YMN-136)IMP3Self-amplifying RNA-LNP vaccineIMP3-positive advanced NSCLC after failure of, intolerance to, or refusal of standard treatmentIMP3-saRNA vaccine aloneEfficacy: Not yet available; Safety: Not yet availableChinaNCT07513376RecruitingPhase IIIAdjuvantIntismeran (mRNA-4157/V940)Personalized neoantigensPersonalized mRNA neoantigen vaccineCompletely resected high-risk stage I NSCLC (adjuvant)Intismeran + pembrolizumab coformulated with berahyaluronidase alfa; Intismeran alone; placeboEfficacy: Not yet available; Safety: Not yet availableMultinational (Australia, Canada, Greece, Hungary, United States, …)NCT07652125RecruitingNAFirst-lineRGL-270Personalized neoantigensPersonalized neoantigen mRNA vaccineAdvanced NSCLC without actionable driver alterationsRGL-270 + PD-1/PD-L1 inhibitor; physician’s-choice standard of care (real-world cohort)Efficacy: Not yet available; Safety: Not yet availableChinaNCT06685653Not yet recruitingEarly Phase IAdjuvant, Second-/Later-lineRGL-270Personalized neoantigensmRNA vaccineResectable stage IIB–IIIB (N2) NSCLC post neoadjuvant chemotherapy + ICI with no MPR; recurrent/metastatic NSCLC after 1 L platinum ± ICI (EGFR/ALK excluded)Adebrelimab + RGL-270; RGL-270 alone (some cohorts)Efficacy: Not yet available; Safety: Not yet availableChinaNCT06756295Not yet recruitingEarly Phase IAdjuvant, second-/later-linePersonalized tumor FAST vaccine combo (with adjuvant) (FAST-CR)Autologous tumor antigens (whole-cell)Autologous whole-cell vaccine (FAST), with adjuvantLung cancer; melanoma; breast cancer; colorectal cancer (mixed)Standard therapy + FAST vaccine; RT + FAST vaccine (non-surgical cohort)Efficacy: Not yet available; Safety: Not yet availableChinaNCT06735508Not yet recruitingEarly Phase IAdjuvantmRNA neoantigen vaccine (personalized)Personalized neoantigensmRNA vaccineResected NSCLC (adjuvant), EGFR/ALK negativemRNA neoantigen vaccine + adebrelimab (PD-L1 antibody)Efficacy: Not yet available; Safety: Not yet availableChinaNCT07683832Not yet recruitingEarly Phase IPerioperative/AdjuvantABO2109Personalized tumor antigens (not specified)Personalized cancer vaccine (platform not specified)Resected NSCLC (perioperative/adjuvant setting)ABO2109 alone; ABO2109 + toripalimabEfficacy: Not yet available; Safety: Not yet availableNot yet listedNCT07614646Not yet recruitingPhase IFirst-lineEVM14Five TAAs (not disclosed)Off-the-shelf multi-TAA mRNA-LNP vaccineUntreated unresectable Stage IIIB–IIIC or Stage IV squamous NSCLC; PD-L1 TPS ≥1%EVM14 + ivonescimabEfficacy: Not yet available; Safety: Not yet availableNot yet listedNCT07543172Not yet recruitingPhase IISecond-/Later-lineCIMAvax-EGFEGFProtein conjugate vaccineAdvanced KRAS G12C-mutated NSCLCCIMAvax-EGF + KRAS G12C inhibitorEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT02955290Active, not recruitingPhase I/IISecond-/later-line, first-line, maintenanceCIMAvax-EGF (recombinant human EGF-rP64K/Montanide ISA 51)EGFProtein vaccine (EGF conjugate + adjuvant)Advanced/metastatic NSCLC (multiple cohorts including post-platinum progression; also 1 L PD-L1 ≥50% and maintenance cohorts)CIMAvax + nivolumab or CIMAvax + pembrolizumab (includes maintenance regimens)Second-Line (post-Chemotherapy, + Nivolumab)212,213: Efficacy: ORR NR, DCR 47.6% (10/21), mOS 11.9 months (3-year OS rate 29%). Safety: NR in abstract (Note: Longest mOS of 31.7 months observed in KRAS wild-type).United StatesNCT05557591Active, not recruitingPhase IIFirst-lineBNT116 (FixVac Lung)CLDN6, KK-LC-1, MAGE-A3, MAGE-A4, MAGE-C1, PRAMEmRNA vaccineAdvanced NSCLC, PD-L1 ≥50%, first-lineBNT116 + cemiplimab vs. cemiplimab monotherapyEfficacy: Not yet available; Safety: Not yet availableMultinational (Georgia, the Republic of Korea, Germany, Turkey, United States, Spain)NCT05242965Active, not recruitingPhase IIMaintenanceSTEMVAC (CD105/YB-1/SOX2/CDH3/MDM2 polyepitope plasmid DNA vaccine)CD105, YB-1, SOX2, CDH3, MDM2DNA plasmid vaccineStage IV NSCLC after carboplatin + pemetrexed + pembrolizumab induction; on maintenance therapySTEMVAC + sargramostim during pembrolizumab + pemetrexed maintenanceEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT05195619Active, not recruitingPhase IMaintenanceAutologous dendritic cell vaccine loaded with personalized peptides (PEP-DC vaccine)Personalized peptides (neoantigen-based)Dendritic cell vaccine (cell-based)Advanced/metastatic NSCLC (IIIA non-curative to IVB) on SOC maintenance/continuation without progressionPEP-DC vaccine + low-dose cyclophosphamide (with SOC maintenance/continuation)Efficacy: Not yet available; Safety: Not yet availableSwitzerlandNCT03970746Active, not recruitingPhase I/IIMaintenance, First-linePDC*lung01NY-ESO-1, MAGE-A3, MAGE-A4, Multi-MAGE-A, MUC1, survivinPlasmacytoid dendritic cell vaccine (allogeneic, peptide-loaded)HLA-A*02:01+ NSCLC; cohorts include resected NSCLC and metastatic stage IV PD-L1 ≥50%PDC*lung01 alone or with anti-PD-1 (pembrolizumab)First-Line18: Efficacy: confirmed ORR 51.0% (n = 45), 9-month PFS rate 47.0%, mPFS 9.0 months. Safety: most AEs were mild to moderate; related DLT 2.0% (1/45) in cohort B2.Multinational (Belgium, France, Netherlands, Germany, Poland)NCT03546361Active, not recruitingPhase ISecond-/Later-lineAd-CCL21-DC (CCL21-gene modified dendritic cell vaccine)CCL21 (immune chemokine)Dendritic cell vaccine (adenovirus CCL21 gene-modified)Stage IV NSCLC (includes cohorts after PD-1/PD-L1 or after EGFR/ALK TKIs)Ad-CCL21-DC + pembrolizumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT02432963Active, not recruitingPhase ISecond-/later-linep53MVA (modified vaccinia Ankara vaccine expressing p53)p53 (TP53)Viral vector vaccine (MVA)Refractory advanced solid tumors including NSCLC; TP53 overexpression/mutation populationp53MVA + pembrolizumabNo NSCLC-specific efficacy/safety data availableUnited StatesNCT06253520Active, not recruitingPhase ISecond-/later-lineGRT-C903/GRT-R904KRAS G12D/G12VAdenoviral vector + mRNA vaccineMetastatic solid cancers including NSCLC with KRAS G12D or G12V mutationAutologous KRAS TCR-transduced PBL + GRT-C903/GRT-R904 (with lymphodepletion/IL-2)Efficacy: Not yet available; Safety: Not yet availableUnited StatesNCT06015724Active, not recruitingPhase IISecond-/Later-lineKRAS vaccine (TG-01/Stimulon QS-21)KRAS mutant peptidesPeptide vaccineAdvanced refractory NSCLC (also includes PDAC)KRAS vaccine + daratumumab + nivolumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesADR, Adverse drug reaction; AE, Adverse event; ALK, Anaplastic lymphoma kinase; CCL21, C—C motif chemokine ligand 21; CEA, Carcinoembryonic antigen; CLDN6, Claudin 6; CR, Complete response; CRT, Chemoradiotherapy; DC, Dendritic cell; DCR, Disease control rate; DLT, Dose-limiting toxicity; EFS, Event-free survival; EGF, Epidermal growth factor; EGFR, Epidermal growth factor receptor; ES-SCLC, Extensive-stage small-cell lung cancer; FAST, Fragmented autoantigen stimulated T-cell; FAST-CR, Fragmented autoantigen stimulated T-cell-immunotherapy combined with radiotherapy; HER2, Human epidermal growth factor receptor 2; HLA, Human leukocyte antigen; ICI, Immune checkpoint inhibitor; IMP3, Insulin-like growth factor 2 mRNA-binding protein 3; KK-LC-1, Kita-Kyushu lung cancer antigen 1; LNP, Lipid nanoparticle; mAb, Monoclonal antibody; MAGE-A2, Melanoma-associated antigen 2; MAGE-A3, Melanoma-associated antigen A3; MAGE-A4, Melanoma-associated antigen A4; MAGE-C1, Melanoma-associated antigen C1; MPR, Major pathological response; mOS, Median overall survival; mPFS, Median progression-free survival; mRNA, Messenger RNA; NA, Not applicable; NR, Not reported; NSCLC, Non-small cell lung cancer; NY-ESO-1, New York esophageal squamous cell carcinoma 1; ORR, Objective response rate; OS, Overall survival; PADRE, Pan HLA-DR-binding epitope; PBL, Peripheral blood lymphocyte; PD-1, Programmed death 1; PD-L1, Programmed death ligand 1; PDAC, Pancreatic ductal adenocarcinoma; PR, Partial response; PRAME, Preferentially expressed antigen in melanoma; RT, Radiotherapy; saRNA, Self-amplifying RNA; SD, Stable disease; SOC, Standard of care; TAA, Tumor-associated antigen; TCR, T-cell receptor; TEAE, Treatment-emergent adverse event; TKI, Tyrosine kinase inhibitor; TPS, Tumor proportion score; TPTE, Transmembrane phosphatase with tensin homology; TRAE, Treatment-related adverse event; VEGF, Vascular endothelial growth factor; WES, Whole-exome sequencing. Table 1Ongoing and upcoming clinical trials of therapeutic cancer vaccines in NSCLC across lines of therapy.Table 1 dummy alt textNCT numberStudy statusPhaseClinical scenarioVaccine nameTargetVaccine typeDiseaseTreatment regimen(s)Efficacy and safety outcomesCountryNCT06928922RecruitingEarly Phase ISecond-/later-lineBMD006Tumor tissue expression of ≥1 of nine TAAs: NY-ESO-1, MAGE-A3, TPTE, tyrosinase, KK-LC-1, MAGE-A4, CLDN6, PRAME, or MAGE-C1Inhaled mRNA dry powder vaccineAdvanced lung cancer (driver gene negative) or advanced solid tumors with lung metastasis; TAA expression requiredBMD006 alone; BMD006 + PD-1 antibody; BMD006 + PD-1/VEGF antibodyEfficacy: Not yet available; Safety: Not yet availableChinaNCT06472245RecruitingPhase IIISecond-/later-lineOSE2101 (Tedopi)P53, HER2, CEA, MAGE-A2, MAGE-A3 (+PADRE)Peptide vaccine (HLA-A2-restricted epitopes; Montanide ISA 51 adjuvant)Metastatic NSCLC (squamous or non-squamous), HLA-A2 positive, secondary resistance to immune checkpoint inhibitorOSE2101 monotherapy vs. docetaxelEfficacy: Not yet available; Safety: Not yet availableMultinational (Canada, Hungary, Greece, Germany, Italy, …)NCT06329908RecruitingEarly Phase ISecond-/later-lineNeo-DCVac (LG002)Personalized neoantigensDendritic cell vaccine (neoantigen-loaded)Advanced lung cancer resistant to ICIs after 1L PD-1/PD-L1 immunochemotherapyNeo-DCVac + immune checkpoint inhibitor(s)Efficacy: Not yet available; Safety: Not yet availableChinaNCT05950139RecruitingPhase I/IIMaintenanceALK peptide vaccineMutant ALK (acquired resistance mutations)Peptide vaccineStage IV ALK+ NSCLC on ALK TKI with stable disease ≥4 monthsALK peptide vaccine + ongoing ALK targeted therapyEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT05886439RecruitingPhase ISecond-/later-lineLK101 injectionPersonalized neoantigenDendritic cell vaccine (personalized neoantigen-pulsed)Stage IIIB–IV NSCLC after anti-PD-1/PD-L1 failure (plus ES-SCLC cohort; plus maintenance cohort)LK101 + PD-1/PD-L1 mAb (pembrolizumab/durvalumab/tislelizumab)Efficacy: Not yet available; Safety: Not yet availableChinaNCT05751798RecruitingPhase I/IIFirst-lineOSE2101 (Tedopi)TAAs (HLA-A2-restricted peptides)Peptide vaccineMetastatic (stage IV) NSCLC, PD-L1 TPS ≥50% (HLA-A2± cohorts)OSE2101 + OSE-279 (anti-PD-1 mAb)Efficacy/Safety: No NSCLC-specific data available for the OSE-279 + OSE2101 cohort. OSE-279 monotherapy Part A: 1 CR, 4 PRs, and 7 SD among 20 advanced solid-tumor patients208Multinational (Belgium, France, Spain)NCT05269381RecruitingPhase I/IISecond-/later-line, adjuvantPersonalized neoantigen peptide-based vaccine (PNeoVCA)NeoantigensPeptide neoantigen vaccineAdvanced solid tumors including metastatic NSCLC (stage III/IV)Personalized neoantigen peptide vaccine + pembrolizumabNo NSCLC-specific efficacy/safety data availableUnited StatesNCT05254184RecruitingPhase IFirst-linePooled mutant KRAS-targeted long peptide vaccineKRAS mutations (G12C/G12V/G12D/G12A/G13D/G12R)Long peptide vaccineUnresectable stage III/IV KRAS-mutated NSCLCKRAS peptide vaccine + poly-ICLC + nivolumab + ipilimumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT05142189RecruitingPhase IFirst-line, later-line, neoadjuvant/adjuvant, maintenanceBNT116CLDN6, KK-LC-1, MAGE-A3, MAGE-A4, MAGE-C1, PRAMEmRNA vaccineAdvanced/metastatic NSCLC (multi-cohort)BNT116 alone; BNT116 + cemiplimab; BNT116 + docetaxel (and other cohorts)First-line (Frail, PD-L1 ≥ 1%)21: Efficacy: ORR 45.0% (9/20), DCR 80.0% (16/20), mPFS 9.9 months. Safety: BNT116 TRAEs 100.0% (Grade ≥3: 15.0%), cemiplimab TRAEs 65.0% (Grade ≥3: 10.0%). Second/later-Line (post-PD-1, PD-L1 ≥ 50%)165: Efficacy: ORR 10.0% (2/20), DCR 80.0% (16/20), mPFS 5.5 months. Safety: Overall TEAEs 100.0% (20/20), Grade 3 TEAEs 55.0% (11/20), no Grade 4/5 events. Second/Later-Line (post-PD-1 + Chemo, BNT116 + docetaxel)209: Efficacy: ORR 35.0% (7/20), DCR 85.0% (17/20). Safety: Serious TEAEs 50.0% (10/20; 3 related to BNT116, 3 to docetaxel), no DLTs.Consolidation (post-CRT)210: Efficacy: 6-month EFS 72.1%, 6-month OS 95.0%; median follow-up 6.3 months. Safety: BNT116 TRAEs 100.0% (20/20), cemiplimab TRAEs 75.0% (15/20), discontinuation due to AEs 25.0% (5/20).Late-Line (>3 prior lines; BNT116 ± sequential cemiplimab)211: Efficacy: ORR 0 (0/10 evaluable), DCR 60.0% (6/10), median duration of disease control 1.53 months. Safety: Overall TEAEs 100.0% (18/18), serious TEAEs 27.8% (5/18), no DLTsMultinational (Hungary, Turkey, Spain, Poland, …)NCT05104515RecruitingPhase ISecond-/Later-lineOVM-200SurvivinRecombinant overlapping peptide vaccineLocally advanced or metastatic NSCLC (≥1 prior line)OVM-200 aloneSecond-/later-Line (mixed solid-tumor cohort; n = 12, including 1 patient with NSCLC): Efficacy: SD 50.0% (6/12). Safety: no DLTs or serious ADRs; all reported AEs were Grade 1 injection-site reactionsUnited KingdomNCT05098210RecruitingPhase ISecond-/later-lineNeoantigen peptide vaccine (personalized)NeoantigenPeptide neoantigen vaccineStage III–IV NSCLC (pretreated; plus melanoma/breast cohorts)Neoantigen peptide vaccine + poly-ICLC + nivolumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT04147078RecruitingPhase IAdjuvantNeoantigen-primed dendritic cell vaccineNeoantigenDendritic cell vaccine (peptide-primed)Postoperative locally advanced NSCLC (with standard adjuvant treatment)Neoantigen-primed DC vaccine (subcutaneous) after standard adjuvant treatmentNo NSCLC-specific efficacy/safety data availableChinaNCT01720836RecruitingPhase I/IIAdjuvantMUC1 100-mer peptide vaccine + poly-ICLC (Hiltonol)MUC1Peptide vaccineStage IA–IIIB NSCLC (localized/locally advanced)MUC1 peptide vaccine + poly-ICLC after surgery ± adjuvant chemotherapy or after chemoradiationEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT00722228RecruitingPhase I/IISecond-/Later-lineAutologous/allogeneic whole cell cancer vaccine (DNP-conjugated, irradiated tumor cells)Whole tumor cells (autologous/allogeneic)Cell-based whole tumor cell vaccineMetastatic epithelial cancers including lung cancer; failed ≥1 chemotherapyAutologous or allogeneic tumor cell vaccine (subcutaneous)Efficacy: Not yet available; Safety: Not yet availableIsraelNCT07561723RecruitingPhase ISecond-/Later-lineIMP3-saRNA (YMN-136)IMP3Self-amplifying RNA-LNP vaccineIMP3-positive advanced NSCLC after failure of, intolerance to, or refusal of standard treatmentIMP3-saRNA vaccine aloneEfficacy: Not yet available; Safety: Not yet availableChinaNCT07513376RecruitingPhase IIIAdjuvantIntismeran (mRNA-4157/V940)Personalized neoantigensPersonalized mRNA neoantigen vaccineCompletely resected high-risk stage I NSCLC (adjuvant)Intismeran + pembrolizumab coformulated with berahyaluronidase alfa; Intismeran alone; placeboEfficacy: Not yet available; Safety: Not yet availableMultinational (Australia, Canada, Greece, Hungary, United States, …)NCT07652125RecruitingNAFirst-lineRGL-270Personalized neoantigensPersonalized neoantigen mRNA vaccineAdvanced NSCLC without actionable driver alterationsRGL-270 + PD-1/PD-L1 inhibitor; physician’s-choice standard of care (real-world cohort)Efficacy: Not yet available; Safety: Not yet availableChinaNCT06685653Not yet recruitingEarly Phase IAdjuvant, Second-/Later-lineRGL-270Personalized neoantigensmRNA vaccineResectable stage IIB–IIIB (N2) NSCLC post neoadjuvant chemotherapy + ICI with no MPR; recurrent/metastatic NSCLC after 1 L platinum ± ICI (EGFR/ALK excluded)Adebrelimab + RGL-270; RGL-270 alone (some cohorts)Efficacy: Not yet available; Safety: Not yet availableChinaNCT06756295Not yet recruitingEarly Phase IAdjuvant, second-/later-linePersonalized tumor FAST vaccine combo (with adjuvant) (FAST-CR)Autologous tumor antigens (whole-cell)Autologous whole-cell vaccine (FAST), with adjuvantLung cancer; melanoma; breast cancer; colorectal cancer (mixed)Standard therapy + FAST vaccine; RT + FAST vaccine (non-surgical cohort)Efficacy: Not yet available; Safety: Not yet availableChinaNCT06735508Not yet recruitingEarly Phase IAdjuvantmRNA neoantigen vaccine (personalized)Personalized neoantigensmRNA vaccineResected NSCLC (adjuvant), EGFR/ALK negativemRNA neoantigen vaccine + adebrelimab (PD-L1 antibody)Efficacy: Not yet available; Safety: Not yet availableChinaNCT07683832Not yet recruitingEarly Phase IPerioperative/AdjuvantABO2109Personalized tumor antigens (not specified)Personalized cancer vaccine (platform not specified)Resected NSCLC (perioperative/adjuvant setting)ABO2109 alone; ABO2109 + toripalimabEfficacy: Not yet available; Safety: Not yet availableNot yet listedNCT07614646Not yet recruitingPhase IFirst-lineEVM14Five TAAs (not disclosed)Off-the-shelf multi-TAA mRNA-LNP vaccineUntreated unresectable Stage IIIB–IIIC or Stage IV squamous NSCLC; PD-L1 TPS ≥1%EVM14 + ivonescimabEfficacy: Not yet available; Safety: Not yet availableNot yet listedNCT07543172Not yet recruitingPhase IISecond-/Later-lineCIMAvax-EGFEGFProtein conjugate vaccineAdvanced KRAS G12C-mutated NSCLCCIMAvax-EGF + KRAS G12C inhibitorEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT02955290Active, not recruitingPhase I/IISecond-/later-line, first-line, maintenanceCIMAvax-EGF (recombinant human EGF-rP64K/Montanide ISA 51)EGFProtein vaccine (EGF conjugate + adjuvant)Advanced/metastatic NSCLC (multiple cohorts including post-platinum progression; also 1 L PD-L1 ≥50% and maintenance cohorts)CIMAvax + nivolumab or CIMAvax + pembrolizumab (includes maintenance regimens)Second-Line (post-Chemotherapy, + Nivolumab)212,213: Efficacy: ORR NR, DCR 47.6% (10/21), mOS 11.9 months (3-year OS rate 29%). Safety: NR in abstract (Note: Longest mOS of 31.7 months observed in KRAS wild-type).United StatesNCT05557591Active, not recruitingPhase IIFirst-lineBNT116 (FixVac Lung)CLDN6, KK-LC-1, MAGE-A3, MAGE-A4, MAGE-C1, PRAMEmRNA vaccineAdvanced NSCLC, PD-L1 ≥50%, first-lineBNT116 + cemiplimab vs. cemiplimab monotherapyEfficacy: Not yet available; Safety: Not yet availableMultinational (Georgia, the Republic of Korea, Germany, Turkey, United States, Spain)NCT05242965Active, not recruitingPhase IIMaintenanceSTEMVAC (CD105/YB-1/SOX2/CDH3/MDM2 polyepitope plasmid DNA vaccine)CD105, YB-1, SOX2, CDH3, MDM2DNA plasmid vaccineStage IV NSCLC after carboplatin + pemetrexed + pembrolizumab induction; on maintenance therapySTEMVAC + sargramostim during pembrolizumab + pemetrexed maintenanceEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT05195619Active, not recruitingPhase IMaintenanceAutologous dendritic cell vaccine loaded with personalized peptides (PEP-DC vaccine)Personalized peptides (neoantigen-based)Dendritic cell vaccine (cell-based)Advanced/metastatic NSCLC (IIIA non-curative to IVB) on SOC maintenance/continuation without progressionPEP-DC vaccine + low-dose cyclophosphamide (with SOC maintenance/continuation)Efficacy: Not yet available; Safety: Not yet availableSwitzerlandNCT03970746Active, not recruitingPhase I/IIMaintenance, First-linePDC*lung01NY-ESO-1, MAGE-A3, MAGE-A4, Multi-MAGE-A, MUC1, survivinPlasmacytoid dendritic cell vaccine (allogeneic, peptide-loaded)HLA-A*02:01+ NSCLC; cohorts include resected NSCLC and metastatic stage IV PD-L1 ≥50%PDC*lung01 alone or with anti-PD-1 (pembrolizumab)First-Line18: Efficacy: confirmed ORR 51.0% (n = 45), 9-month PFS rate 47.0%, mPFS 9.0 months. Safety: most AEs were mild to moderate; related DLT 2.0% (1/45) in cohort B2.Multinational (Belgium, France, Netherlands, Germany, Poland)NCT03546361Active, not recruitingPhase ISecond-/Later-lineAd-CCL21-DC (CCL21-gene modified dendritic cell vaccine)CCL21 (immune chemokine)Dendritic cell vaccine (adenovirus CCL21 gene-modified)Stage IV NSCLC (includes cohorts after PD-1/PD-L1 or after EGFR/ALK TKIs)Ad-CCL21-DC + pembrolizumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesNCT02432963Active, not recruitingPhase ISecond-/later-linep53MVA (modified vaccinia Ankara vaccine expressing p53)p53 (TP53)Viral vector vaccine (MVA)Refractory advanced solid tumors including NSCLC; TP53 overexpression/mutation populationp53MVA + pembrolizumabNo NSCLC-specific efficacy/safety data availableUnited StatesNCT06253520Active, not recruitingPhase ISecond-/later-lineGRT-C903/GRT-R904KRAS G12D/G12VAdenoviral vector + mRNA vaccineMetastatic solid cancers including NSCLC with KRAS G12D or G12V mutationAutologous KRAS TCR-transduced PBL + GRT-C903/GRT-R904 (with lymphodepletion/IL-2)Efficacy: Not yet available; Safety: Not yet availableUnited StatesNCT06015724Active, not recruitingPhase IISecond-/Later-lineKRAS vaccine (TG-01/Stimulon QS-21)KRAS mutant peptidesPeptide vaccineAdvanced refractory NSCLC (also includes PDAC)KRAS vaccine + daratumumab + nivolumabEfficacy: Not yet available; Safety: Not yet availableUnited StatesADR, Adverse drug reaction; AE, Adverse event; ALK, Anaplastic lymphoma kinase; CCL21, C—C motif chemokine ligand 21; CEA, Carcinoembryonic antigen; CLDN6, Claudin 6; CR, Complete response; CRT, Chemoradiotherapy; DC, Dendritic cell; DCR, Disease control rate; DLT, Dose-limiting toxicity; EFS, Event-free survival; EGF, Epidermal growth factor; EGFR, Epidermal growth factor receptor; ES-SCLC, Extensive-stage small-cell lung cancer; FAST, Fragmented autoantigen stimulated T-cell; FAST-CR, Fragmented autoantigen stimulated T-cell-immunotherapy combined with radiotherapy; HER2, Human epidermal growth factor receptor 2; HLA, Human leukocyte antigen; ICI, Immune checkpoint inhibitor; IMP3, Insulin-like growth factor 2 mRNA-binding protein 3; KK-LC-1, Kita-Kyushu lung cancer antigen 1; LNP, Lipid nanoparticle; mAb, Monoclonal antibody; MAGE-A2, Melanoma-associated antigen 2; MAGE-A3, Melanoma-associated antigen A3; MAGE-A4, Melanoma-associated antigen A4; MAGE-C1, Melanoma-associated antigen C1; MPR, Major pathological response; mOS, Median overall survival; mPFS, Median progression-free survival; mRNA, Messenger RNA; NA, Not applicable; NR, Not reported; NSCLC, Non-small cell lung cancer; NY-ESO-1, New York esophageal squamous cell carcinoma 1; ORR, Objective response rate; OS, Overall survival; PADRE, Pan HLA-DR-binding epitope; PBL, Peripheral blood lymphocyte; PD-1, Programmed death 1; PD-L1, Programmed death ligand 1; PDAC, Pancreatic ductal adenocarcinoma; PR, Partial response; PRAME, Preferentially expressed antigen in melanoma; RT, Radiotherapy; saRNA, Self-amplifying RNA; SD, Stable disease; SOC, Standard of care; TAA, Tumor-associated antigen; TCR, T-cell receptor; TEAE, Treatment-emergent adverse event; TKI, Tyrosine kinase inhibitor; TPS, Tumor proportion score; TPTE, Transmembrane phosphatase with tensin homology; TRAE, Treatment-related adverse event; VEGF, Vascular endothelial growth factor; WES, Whole-exome sequencing. Ongoing and upcoming clinical trials of therapeutic cancer vaccines in NSCLC across lines of therapy. ADR, Adverse drug reaction; AE, Adverse event; ALK, Anaplastic lymphoma kinase; CCL21, C—C motif chemokine ligand 21; CEA, Carcinoembryonic antigen; CLDN6, Claudin 6; CR, Complete response; CRT, Chemoradiotherapy; DC, Dendritic cell; DCR, Disease control rate; DLT, Dose-limiting toxicity; EFS, Event-free survival; EGF, Epidermal growth factor; EGFR, Epidermal growth factor receptor; ES-SCLC, Extensive-stage small-cell lung cancer; FAST, Fragmented autoantigen stimulated T-cell; FAST-CR, Fragmented autoantigen stimulated T-cell-immunotherapy combined with radiotherapy; HER2, Human epidermal growth factor receptor 2; HLA, Human leukocyte antigen; ICI, Immune checkpoint inhibitor; IMP3, Insulin-like growth factor 2 mRNA-binding protein 3; KK-LC-1, Kita-Kyushu lung cancer antigen 1; LNP, Lipid nanoparticle; mAb, Monoclonal antibody; MAGE-A2, Melanoma-associated antigen 2; MAGE-A3, Melanoma-associated antigen A3; MAGE-A4, Melanoma-associated antigen A4; MAGE-C1, Melanoma-associated antigen C1; MPR, Major pathological response; mOS, Median overall survival; mPFS, Median progression-free survival; mRNA, Messenger RNA; NA, Not applicable; NR, Not reported; NSCLC, Non-small cell lung cancer; NY-ESO-1, New York esophageal squamous cell carcinoma 1; ORR, Objective response rate; OS, Overall survival; PADRE, Pan HLA-DR-binding epitope; PBL, Peripheral blood lymphocyte; PD-1, Programmed death 1; PD-L1, Programmed death ligand 1; PDAC, Pancreatic ductal adenocarcinoma; PR, Partial response; PRAME, Preferentially expressed antigen in melanoma; RT, Radiotherapy; saRNA, Self-amplifying RNA; SD, Stable disease; SOC, Standard of care; TAA, Tumor-associated antigen; TCR, T-cell receptor; TEAE, Treatment-emergent adverse event; TKI, Tyrosine kinase inhibitor; TPS, Tumor proportion score; TPTE, Transmembrane phosphatase with tensin homology; TRAE, Treatment-related adverse event; VEGF, Vascular endothelial growth factor; WES, Whole-exome sequencing. Second- and later-line treatment In patients with NSCLC who progress after first-line chemoimmunotherapy or targeted therapy, later-line options remain limited and prognosis is poor, with mOS typically ranging from 8 to 12 months. These tumors often show clonal evolution, an immunosuppressive TME, and impaired endogenous anti-tumor immunity.12,94 Thus, this setting has often served as an initial clinical testing context for cancer vaccines, where safety, immunogenicity, and preliminary anti-tumor activity can be assessed before movement into earlier disease. At the same time, second‑ or later‑line disease is biologically less favorable for vaccination, as these patients typically present with higher tumor burden, greater antigenic heterogeneity, and more profound immune dysfunction. Earlier vaccine experience Earlier vaccine experience in second-/later-line NSCLC was dominated by shared-antigen approaches. Peptide-based vaccines directed against TAA represent one of the first platforms explored for late-line NSCLC therapy and are one of the most extensively studied to date. Their clinical experience illustrates a recurring pattern: peripheral antigen-specific immune responses can be induced, but these responses often fail to translate into objective tumor regression. Early trials of MAGE-A family peptides (NCT00020267), mutant RAS peptide (NCT00019006), and HER2 or NY-ESO-1 protein vaccines (NCT00291473) mainly established feasibility and safety. Subsequent HLA-restricted multi-peptide studies, including HLA-A*24:02-restricted approaches (NCT00633724, NCT00874588, NCT00676949) and up-regulated in lung cancer 10 (URLC10)/cell division cycle associated 1 (CDCA1)/kinesin family member 20A (KIF20A)-based vaccines (NCT01069640, NCT01069575, NCT01950156), further supported immunogenicity, but clinical activity was generally limited, with stable disease more common than tumor shrinkage.95, 96, 97, 98 The telomerase-derived multi-epitope vaccine UCPVax (NCT02818426) showed acceptable tolerability and induced CD4+ T helper 1 (Th1) responses in refractory metastatic NSCLC, including patients previously treated with platinum chemotherapy and PD-(L)1 blockade.99 Additional peptide approaches include the multi-cancer-testis-antigen vaccine S-488210/S-488211 (NCT04316689), arginase 1 (ARG1)-targeted vaccination (NCT03689192), and the T-cell epitopes associated with impaired peptide processing 24 (TEIPP24) long-peptide vaccine targeting non-classical MHC-I antigens in ICI-resistant tumors (NCT05898763).25,100,101 Viral vector-based vaccines provide stronger innate immune stimulation but are constrained by vector immunogenicity and repeated-dosing limitations. TG4010, an MVA vector encoding MUC1 and interleukin 2 (IL-2), was evaluated with nivolumab in previously treated metastatic non-squamous NSCLC (NCT02823990), but clinical interpretation remains limited by the small sample size and early-phase, single-arm design.102,103 The Ad-MAGEA3/MG1-MAGEA3 prime-boost strategy was tested in MAGE-A3-expressing solid tumors and subsequently with pembrolizumab in NSCLC (NCT02285816 and NCT02879760), supporting feasibility of viral-vector priming in combination with PD-1 blockade.104, 105, 106 ETBX-011/061/051 (NCT03384316), an Ad5-based multi-antigen vaccine targeting CEA, MUC1, and Brachyury, represents a related multi-antigen approach in advanced solid tumors including lung cancer.107 Among cell-based vaccines, allogeneic WTCVs have been the most clinically explored in NSCLC. Viagenpumatucel-L (HS-110), an allogeneic LUAD cell-line vaccine modified to express gp96-Ig, showed favorable tolerability and induction of broad CD8+ T-cell and IFN-γ responses when combined with nivolumab in previously treated NSCLC (NCT02439450).108 However, the survival signal remains hypothesis-generating and requires randomized confirmation. Additional approaches such as AST-VAC2 (NCT03371485), DRibble (NCT00850785), Vigil (NCT01061840), and ongoing DC-based combination strategies (e.g., Ad-CCL21-DC, NCT03546361) further support the feasibility of cellular or DC-based vaccination.109, 110, 111 Early mRNA vaccine development in NSCLC also followed the shared-antigen paradigm. CV9201 encoded NY-ESO-1, MAGE-C1, MAGE-C2, survivin, and 5T4 (NCT00923312). Immune responses against these antigens were detected in >60% of patients, yet objective tumor shrinkage was difficult to achieve with single-agent therapy.112 CV9202 (BI 1361849, NCT03164772), which added MUC1 to the CV9201 antigen set, was evaluated in combination with durvalumab and tremelimumab in phase I/II trials, and preliminary results indicated favorable immunogenicity and tolerability.17,113 Ongoing personalized mRNA vaccine trials are now primarily focused on neoantigen-based designs and combination strategies. Personalized strategies Personalized neoantigen vaccines have become a major focus because mutation-derived epitopes are less constrained by central tolerance and can be matched to the antigenic profile of individual tumors. Platforms based on self-amplifying RNA, synthetic long peptides (SLPs), mRNA, or DC loading have demonstrated robust immunogenicity, particularly when combined with ICIs. GRT-C901/R902 (NCT03639714) induced polyclonal T-cell responses and durable remissions when combined with nivolumab and ipilimumab.114,115 GEN-009 (NCT03633110) and NEO-PV-01 (NCT02897765) similarly supported the feasibility of inducing neoantigen-specific T-cell responses, with evidence of epitope spreading or tumor trafficking in selected cases.116, 117, 118 Integrated analyses of VB10.NEO trials (NCT03548467, NCT05018273) demonstrated near-universal induction of neoantigen-specific T-cell responses, with neoantigen quality correlating positively with survival outcomes.119, 120, 121 Semi-personalized off-the-shelf strategies such as GRT-C903/R904 (SLATE series, NCT03953235) showed scalable CD8+ T-cell induction when combined with dual ICIs, although fully personalized vaccines generally elicited more durable immune responses.61,122 However, most studies remain early-phase, nonrandomized, and combined with ICIs, making it difficult to determine the incremental contribution of vaccination to survival outcomes. Ongoing trials Current studies are testing several distinct hypotheses rather than simply extending older vaccine platforms. Randomized peptide-vaccine studies are evaluating whether selected shared-antigen vaccines can outperform standard chemotherapy in defined populations, as illustrated by OSE2101/TEDOPI vs. docetaxel in HLA-A2-positive metastatic NSCLC with secondary resistance to ICIs (NCT06472245). Other studies are evaluating combinations intended to address immune resistance, including daratumumab plus TG01 and nivolumab in refractory NSCLC and pancreatic cancer (NCT06015724). Personalized and cell-based platforms are also being investigated to improve antigen specificity and T-cell priming. Ongoing examples include personalized neoantigen peptide vaccines combined with ICIs or immune adjuvants (NCT05098210), neoantigen-loaded DC vaccines such as Neo-DCVac (NCT06329908) and LK101 (NCT05886439), and neoantigen vaccination combined with adoptive KRAS-specific T-cell therapy (NCT06253520). Additional studies are evaluating an insulin-like growth factor 2 mRNA-binding protein 3 (IMP3)-targeted self-amplifying RNA vaccine after standard-treatment failure or intolerance (NCT07561723) and CIMAvax-EGF combined with a KRAS G12C inhibitor in KRAS G12C-mutated NSCLC (NCT07543172). Inhaled mRNA approaches should be considered separately: BMD006 (NCT06928922) is an off-the-shelf, TAA-encoding inhaled mRNA vaccine rather than a personalized neoantigen product. Thus far, evidence in the second- and later-line setting supports safety and immunogenicity, whereas data on tumor regression or survival benefit from randomized trials remain limited. First-line treatment First-line treatment for NSCLC is primarily determined by driver alteration status and PD-L1 expression. Because effective standard regimens are already available, including ICIs with or without chemotherapy, first-line vaccine trials generally use combination designs intended to complement established therapy rather than replace it. Earlier vaccine experience Early first-line vaccine studies largely focused on shared TAAs, with TG4010 representing one of the most extensively evaluated examples. TG4010 is an MVA vector encoding MUC1 and IL-2, designed to activate CTLs and natural killer cells against MUC1-overexpressing tumors. In its initial phase IIB evaluation (NCT00415818), TG4010 combined with cisplatin/gemcitabine in 148 patients with MUC1-positive stage IIIB/IV NSCLC produced a modest improvement in 6-month progression-free survival (PFS) (43.2% vs. 35.1%) without OS benefit, but a post hoc analysis identified patients with low triple-positive activated lymphocytes (TrPAL) as a subgroup with more pronounced clinical benefit (HR 0.74).123 The subsequent TIME trial (NCT01383148) further supported this biomarker-selected signal, particularly in patients with low TrPAL levels.102,124 The later emergence of PD-1/PD-L1 inhibitors changed the first-line treatment landscape, and chemotherapy-only control arms became less clinically relevant. In a subsequent single-arm phase II study, TG4010 was combined with platinum/pemetrexed and nivolumab in patients with PD-L1-low advanced non-squamous NSCLC (NCT03353675). Because the study lacked a comparator, the incremental contribution of TG4010 to contemporary chemoimmunotherapy could not be determined. These findings illustrate the difficulty of translating shared TAA-directed immune responses into incremental benefit in the modern first-line setting. Cell-based and personalized vaccine strategies Subsequent first-line vaccine development has moved from shared TAA monotherapy toward strategies intended to improve antigen presentation or increase antigen specificity. DCVAC/LuCa (NCT02470468), an autologous DC vaccine loaded with tumor antigens, was evaluated in combination with carboplatin and paclitaxel in advanced NSCLC. Compared with chemotherapy alone, the combination showed longer OS, supporting the feasibility of DC-based priming in the first-line setting.125 However, its current relevance is limited by the absence of validation against contemporary chemoimmunotherapy backbones. NEO-PV-01 (NCT03380871) represents the first major effort to evaluate a personalized neoantigen vaccine combined with a modern chemo-ICI regimen. In this phase Ib study, patient-specific neoantigens identified by tumor sequencing were administered with pemetrexed, carboplatin, and pembrolizumab in metastatic non-squamous NSCLC. The regimen showed acceptable safety and induced neoantigen-specific immune responses. In the intention-to-treat population, the ORR and clinical benefit rate were 34% and 71%, respectively. Among patients who received vaccination, the corresponding rates were 69% and 100%. Median PFS was 6.3 months in the intention-to-treat population and 7.2 months among vaccinated patients, while mOS was 16.8 and 20.0 months, respectively.19 Although outcomes appeared more favorable among vaccinated patients than in the overall enrolled population, this observation is subject to selection bias because vaccination required adequate tissue, successful vaccine manufacture, sufficient clinical stability and enough time to proceed to treatment. These data therefore support feasibility and provide a hypothesis-generating clinical signal, but they do not establish vaccine-attributable benefit in the absence of randomization. Ongoing trials As of 2026, ongoing first-line vaccine development in advanced NSCLC is focused mainly on combination strategies rather than vaccine monotherapy. BNT116 (NCT05142189) is a fixed-antigen RNA-lipoplex vaccine encoding six TAAs, including claudin 6 (CLDN6), Kita-Kyushu lung cancer antigen 1 (KK-LC-1), MAGE-A3, MAGE-A4, MAGE-C1, and PRAME. In the phase I LuCa-MERIT-1 trial, BNT116 plus cemiplimab showed preliminary activity in frail patients with advanced NSCLC who were not suitable for platinum-based chemotherapy, with a confirmed ORR of 45%, disease control rate (DCR) of 80%, and median PFS of 9.9 months in a small cohort.21 An additional phase II cohort is evaluating BNT116 plus cemiplimab vs. cemiplimab alone in first-line advanced NSCLC (NCT05557591). Mutation-directed vaccination is also being explored. A pooled mutant KRAS-targeted long-peptide vaccine with polyinosinic-polycytidylic acid stabilized with poly-L-lysine and carboxymethylcellulose (poly-ICLC) is being evaluated with nivolumab and ipilimumab in locally advanced, unresectable, or metastatic KRAS-mutant NSCLC (NCT05254184), although clinical efficacy data have not yet been reported. Earlier KRAS-directed mRNA vaccine development included mRNA-5671/V941 (NCT03948763), which encoded KRAS G12D, G12V, G13D, and G12C, and was evaluated in KRAS-mutant advanced solid tumors, including NSCLC, either alone or with pembrolizumab, extending mRNA vaccine development toward molecularly defined subgroups. Other first-line programs are evaluating both shared-antigen and personalized vaccine strategies. OSE2101/Tedopi, an HLA-A2-restricted multi-epitope peptide vaccine, is being evaluated with OSE-279 in first-line metastatic NSCLC (NCT05751798). EVM14, an off-the-shelf mRNA-LNP vaccine encoding five TAAs, is being evaluated with ivonescimab in untreated unresectable stage IIIB/IIIC or stage IV LUSC with PD-L1 tumor proportion score (TPS) ≥1% (NCT07614646). Personalized mRNA vaccination combined with PD-1/PD-L1 blockade is also being studied in advanced NSCLC without actionable driver alterations (RGL-270, NCT07652125). CIMAvax-EGF (NCT02955290) continues to be investigated with PD-1 inhibitors across advanced NSCLC settings, reflecting a growth-factor neutralization strategy rather than antigen-specific T-cell priming. First-line NSCLC is clinically important but difficult for therapeutic vaccine development because established ICI-based regimens already provide substantial benefit. Current trials have therefore concentrated on selected populations, including patients who are not candidates for chemotherapy and those with molecularly defined tumors. Moreover, first-line evidence is still derived mainly from early-phase studies of safety, feasibility, and immunogenicity, and randomized data demonstrating an additional survival benefit are not yet available. Perioperative treatment The perioperative setting may provide a more permissive context not only because tumor burden is lower, but also because the primary tumor and tumor-draining lymph nodes may remain available during neoadjuvant therapy. This preserves an endogenous antigen source and a local site for T-cell priming before surgical disruption of lymphatic architecture.126 After resection, vaccination may maintain or broaden tumor-specific immunity against micrometastatic disease, where tumor-associated hypoxia, stromal exclusion, and myeloid suppression may be less pronounced.127 Early exploration in this setting included adjuvant or post-definitive-treatment vaccine approaches. Representative examples include the allogeneic 1650-G cellular vaccine formulated with granulocyte-macrophage colony-stimulating factor in resected early-stage NSCLC (NCT00654030), the MUC1 peptide-poly-ICLC vaccine in localized or locally advanced NSCLC (NCT01720836), and the autophagosome-based DPV-001 vaccine as adjuvant treatment after definitive therapy for stage IIIA/IIIB NSCLC (NCT01909752).128, 129, 130, 131, 132 These studies generally supported feasibility, acceptable safety, and vaccine-induced immune responses. However, small sample sizes, heterogeneous treatment settings, and the absence of randomized long-term survival data have prevented these approaches from establishing a clear clinical role. More recent perioperative development has shifted toward individualized antigen selection and mRNA-based platforms. Neo-mDC (NCT04078269), an autologous monocyte-derived DC vaccine loaded with mRNA encoding patient-specific neoantigens, demonstrated the technical feasibility of postoperative individualized vaccination in resected early-stage NSCLC. Phase I data showed favorable safety and sustained neoantigen-specific T-cell responses for up to 19 months. However, the small cohort size and recurrence in some patients despite detectable immune responses indicate that these findings require validation in larger studies with recurrence-based endpoints.133, 134, 135 Indirect support for this concept comes from ELI-002, a lymph-node-targeted amphiphile mutant KRAS vaccine tested mainly in pancreatic and colorectal cancer.136 Because NSCLC-specific efficacy data are not yet available, these results should be cited only as biological support for KRAS-directed vaccination in MRD rather than as clinical evidence for resected NSCLC. Current perioperative vaccine programs include mRNA platforms combined with immune checkpoint blockade. BNT116, previously discussed in the first-line setting, is being evaluated in dedicated neoadjuvant and adjuvant cohorts for resectable stage II–III NSCLC within the LuCa-MERIT-1 multi-cohort program (NCT05142189), with perioperative efficacy data pending. During neoadjuvant treatment, the presence of the primary tumor and intact tumor-draining lymph nodes may support antigen presentation and T-cell priming, whereas adjuvant vaccination may help sustain immunity against residual microscopic disease. Other ongoing programs include personalized mRNA neoantigen vaccination combined with adebrelimab as adjuvant treatment for resected NSCLC (NCT06735508). Whether these strategies can improve upon the disease-free survival benefit already achieved with perioperative chemo-ICI regimens remains uncertain. Perioperative vaccine development in NSCLC has therefore moved from broad shared-antigen or whole-cell approaches toward individualized neoantigen, mRNA-based, and mutation-directed strategies. Whether these immune responses can reduce recurrence or improve survival requires further confirmation in additional randomized trials. Maintenance treatment The maintenance setting provides a distinct context for cancer vaccine therapy, in which tumor burden is reduced but residual disease remains. In this setting, vaccines may help sustain or broaden tumor-specific immune responses after frontline treatment, particularly when prior therapy has reduced tumor-mediated immunosuppression but microscopic disease persists. The early maintenance vaccine landscape was largely shaped by tecemotide (L-BLP25), a MUC1-derived liposomal peptide vaccine. In the phase III START trial (NCT00409188), tecemotide administered after chemoradiotherapy for unresectable stage III NSCLC did not significantly improve OS in the overall population (mOS 25.6 vs. 22.3 months, HR 0.88, P = 0.123). However, a predefined subgroup analysis suggested a survival benefit in patients who had received concurrent rather than sequential chemoradiotherapy (mOS 30.8 vs. 20.6 months, HR 0.78, P = 0.016).90,137,138 This finding suggested that maintenance vaccination may depend on treatment timing and on the immune context established by prior therapy. A Japanese phase I/II study subsequently supported the tolerability of tecemotide in an Asian population, but did not demonstrate a clear survival advantage.139,140 The phase II NCT00828009 trial combined tecemotide with bevacizumab as an anti-angiogenic maintenance strategy, but efficacy data were limited. Beyond tecemotide, the phase III STOP trial (NCT00676507) evaluated belagenpumatucel-L (Lucanix), an allogeneic NSCLC cell-line vaccine modified with TGF-β2 antisense gene, as maintenance therapy after first-line platinum-based chemotherapy in advanced NSCLC. The trial failed to meet its primary endpoint and showed no significant OS improvement in the overall population. Subgroup analyses suggested improved survival in patients randomized within 12 weeks after completion of chemotherapy and in those who had received prior radiotherapy. These findings suggest that the timing of maintenance vaccination and prior treatment context may influence clinical benefit, although no significant survival benefit was demonstrated in the overall population.43 Beyond shared TAA and whole-cell platforms, maintenance vaccine development has also explored alternative technologies. In a phase II trial (NCT01159288), the DC-derived exosome vaccine Dex2 was evaluated as maintenance immunotherapy after induction chemotherapy in advanced unresectable NSCLC. The predefined PFS target was not met, but correlative analyses showed enhancement of NK-cell function in a subset of patients, supporting the biological feasibility of a cell-free vaccine platform.141 NCT05195619 is a single-center phase Ib study evaluating the safety and feasibility of combining a personalized autologous DC vaccine (PEP-DC) with low-dose cyclophosphamide and standard maintenance or continuation therapy in patients with advanced or recurrent metastatic NSCLC without disease progression. The maintenance setting in NSCLC is one of the few contexts in which earlier shared-antigen vaccine platforms generated phase III subgroup signals, although neither START nor STOP established definitive benefit in the overall study population. These findings suggest that maintenance vaccination may be biologically plausible when administered after tumor-debulking therapy and before extensive immune escape has re-emerged. Current development has therefore moved toward more individualized vaccine strategies and combinations that account for the residual immunosuppressive environment. Whether these newer approaches can translate earlier subgroup signals into reproducible clinical benefit remains uncertain. Special clinical settings Beyond conventional therapy lines, several NSCLC subgroups warrant focused evaluation for cancer vaccines. These include oligometastatic disease, in which local treatment may reduce tumor burden and antigenic diversity, elderly or physically vulnerable patients, who may benefit from lower-toxicity immune approaches, and oncogene-driven NSCLC, in which standard immune checkpoint inhibitors offer limited efficacy. Although clinical evidence remains limited, these scenarios are best regarded as hypothesis-generating contexts in which the risk–benefit profile of vaccination may differ from that in unselected advanced disease. Oligometastatic NSCLC Oligometastatic NSCLC is commonly operationalized as disease with a limited number of metastatic lesions, often up to five, that are amenable to local control. Treatment strategies are generally based on local consolidative therapy, including surgery, stereotactic body radiation therapy, or ablation, integrated with systemic therapy. Randomized phase II data support a survival benefit for local consolidative therapy in selected patients without progression after initial systemic treatment.142,143 By contrast, exploration of therapeutic vaccines remains very limited and is based mainly on the hypothesis that vaccines may enhance the immunogenic effects of local therapy or act synergistically as a maintenance strategy. Current evidence is limited to case-level observations or small studies in surgically treated or locally controlled NSCLC, rather than trials specifically designed for oligometastatic disease. For example, neoantigen dendritic cell vaccination after surgery has been described in resected NSCLC, but controlled data and oligometastasis-specific analyses are lacking.133 Similarly, clinical trials of therapeutic vaccines such as CIMAvax-EGF and OSE2101 (TEDOPI) have enrolled broader populations with advanced or metastatic NSCLC, mainly in maintenance, second- or later-line, or post-immunotherapy settings. Although these studies included patients with different metastatic burdens, oligometastasis-specific analyses and independent prognostic impact of vaccination in this population have not been reported.144, 145, 146 Elderly/poor performance status patients Elderly patients (typically ≥70 years) and those with poor Eastern Cooperative Oncology Group performance status ≥2 often have multiple comorbidities and limited tolerance for intensive treatment, making individualized and lower-toxicity strategies particularly relevant. Therapeutic vaccines have potential advantages in this setting because of their generally favorable tolerability, although current evidence is derived mainly from subgroup analyses of phase II–III trials and real-world studies. Among available data, CIMAvax-EGF has been evaluated most extensively in elderly patients and those unsuitable for chemotherapy. A phase IV real-world study in Cuba suggested favorable safety and potential clinical benefit in patients aged 70 years and older, with some subgroup reports showing prolonged survival and improved quality of life.147 More recently, BNT116 has been explored with cemiplimab in frail patients with advanced NSCLC considered unsuitable for first-line platinum-based chemotherapy. Preliminary phase I data from this small cohort showed an ORR of 45.0%, DCR of 80.0%, and median PFS of 9.9 months, with grade ≥3 TRAEs attributed to BNT116 and cemiplimab reported in 15% and 10% of patients, respectively.21 Molecularly defined subsets Driver gene-positive NSCLC Driver-gene-positive NSCLC is usually managed with targeted therapy.148 Compared with smoking-associated driver-negative NSCLC, EGFR- and ALK-driven tumors generally have lower TMB, lower neoantigen load, and a relatively immune-cold TME, which partly accounts for the limited clinical benefit of ICIs in these subtypes.149,150 In this setting, therapeutic cancer vaccines may be more relevant as a means of controlling residual tumor clones during targeted therapy rather than as a primary strategy for inducing tumor regression. On the one hand, the rapid reduction in tumor burden and transient remodeling of the TME induced by tyrosine kinase inhibitors (TKIs) may provide a biological rationale for evaluating vaccination in combination with targeted therapy.33,151,152 On the other hand, shared or public neoantigen vaccines may also be explored by focusing on recurrent driver mutations in the context of specific HLA restrictions, thereby enabling approaches closer to off-the-shelf formulations. Further studies are needed to determine how the temporal remodeling of the TME by TKIs might inform the timing of vaccination. Early evidence supports the immunogenicity of selected driver-derived epitopes. In EGFR-mutant NSCLC, mutation-derived peptides can elicit vaccine-specific CD4+ and CD8+ T-cell responses.153 Combination trials in patients with acquired resistance to EGFR TKIs have also begun to emerge. For example, NCT06095934 evaluated a neoantigen peptide vaccine combined with anti-PD-1 therapy and chemotherapy in EGFR-mutated advanced NSCLC after EGFR-TKI progression. In a small cohort of 11 vaccinated patients, the ORR was 45% and the DCR was 100%, and no severe TRAEs were reported. Vaccine-specific T-cell responses were detected in 6 of 7 evaluable patients.154 Because treatment also included anti-PD-1 therapy and chemotherapy, the contribution of vaccination cannot be determined from this single-arm study. However, clonal driver neoantigen loss under selective pressure from targeted therapy and immune responses indicates that even driver-derived vaccine targets may not be evolutionarily stable in all cases.155 KRAS-directed vaccine development includes clinical and preclinical programs.156,157 Supporting external evidence comes from ELI-002, an amphiphile-conjugated mutant KRAS (mKRAS) vaccine evaluated mainly in pancreatic and colorectal cancers, which induced mKRAS-specific T-cell responses in 84% of treated patients with activation of both CD4+ and CD8+ compartments and establishment of durable immune memory.136 Given that KRAS-mutant NSCLC shares the same hotspot mutations (particularly G12C, G12D, and G12V) addressed by this platform, these non-NSCLC data provide a mechanistically plausible rationale for extending mutation-directed KRAS vaccination to lung cancer. A representative clinical test of the resistance prevention paradigm is NCT05950139, a phase I/II trial evaluating an ALK-derived peptide vaccine for preventing acquired resistance after ALK-TKI treatment, grounded in preclinical evidence that ALK fusion proteins can elicit specific CD8+ T-cell responses capable of restoring anti-tumor immunity and promoting residual tumor clearance in this molecularly defined subgroup.158 In addition, NSCLC contains a “long tail” of low-frequency driver subgroups, including HER2 kinase domain mutations (around 2–4%, most commonly exon 20 insertions), MET exon 14 skipping (3–4%), ROS1 and RET fusions (around 0.9–2.6% and 1–2%, respectively), and PIK3CA hotspot mutations (around 3–4%).159, 160, 161, 162, 163, 164 Although these mutations or fusion breakpoints may generate novel antigenic epitopes, their practical use as vaccine targets is limited. In addition to the low frequency of these genomic events, HLA restriction further narrows the eligible patient population. The development of standardized off-the-shelf vaccines for such small subpopulations requires careful evaluation of marginal clinical benefit, particularly because targeted therapeutic options for these alterations are relatively well established. More feasible strategies are often to incorporate these alterations into personalized neoantigen vaccine frameworks or to leverage them as antigen sources within combination immunotherapy approaches. Across these programs, however, practical development remains constrained by HLA polymorphism, the efficiency of mutant epitope presentation, and the need for experimental validation of immunogenicity in appropriately selected patient populations. Driver gene-negative NSCLC Driver gene-negative NSCLC is generally more responsive to immunotherapy, particularly in smoking-associated tumors with higher TMB. In advanced disease, therapeutic cancer vaccines are being evaluated in combination with chemotherapy and/or ICIs to broaden and sustain tumor-specific T-cell responses. In perioperative and adjuvant settings, their potential role in controlling residual microscopic disease is also under investigation. Clinical trials of shared TAA-based vaccines did not improve overall outcomes in pivotal phase III studies in NSCLC, including MUC1-based tecemotide and MAGE-A3 immunotherapy.55,137 Most recently, vaccine strategies have converged along two major directions. First, personalized neoantigen vaccines have been explored in combination with frontline therapy, such as NEO-PV-01 administered alongside chemotherapy and anti-PD-1 therapy in advanced non-squamous NSCLC.19 Second, greater emphasis has been placed on scalable off-the-shelf platforms, including early clinical data evaluating multi-TAA-encoding mRNA vaccines combined with PD-1/PD-L1 blockade, with HLA-restricted multi-epitope products serving as complementary approaches.21,146,165 Challenges and future perspectives Biological barriers to vaccine efficacy Tumor heterogeneity and immunoediting Vaccine efficacy is limited by both tumor-cell evolution and suppressive components of the TME.166 NSCLC shows substantial neoantigen heterogeneity, and differences in clonal composition across lesions and over time make it difficult for a small set of shared antigens to cover the entire tumor.167 Clonal neoantigens remain reasonable targets for personalized vaccines because they are shared by most tumor cells and can elicit long-lived CD8+ T-cell responses against antigen-positive clones.84 Even so, immune pressure may select for escape, including depletion of expressed neoantigens, loss of target epitopes, clonal replacement, and relapse.166 Defects in antigen processing and presentation Immune evasion can be further reinforced by defects in antigen presentation. Allele-specific HLA class I loss of heterozygosity, B2M mutations, and inactivation of the IFN-γ–JAK/STAT pathway can reduce MHC-I expression and impair neoantigen presentation, leaving tumor cells functionally less visible to vaccine-induced T cells.168, 169, 170 MHC-I downregulation and impaired peptide presentation are therefore likely to be proximal drivers of vaccine inefficacy, not only downstream resistance markers. Effective vaccination requires cross-presentation by APCs during T-cell priming and stable presentation of the target epitope on tumor cells.171 TAA vaccines (e.g., MAGE-A3) also illustrate that peripheral immune activation may be insufficient when antigen expression, immune tolerance, and tumor-site effector function are not adequately controlled.55 TME barriers: stromal, vascular, and myeloid Within the TME, immune-excluded or immune-desert phenotypes in NSCLC are often shaped by stromal remodeling, cancer-associated fibroblasts, dense extracellular matrix, abnormal vasculature, and myeloid-mediated suppression. These barriers may determine intratumoral entry, persistence, and cytotoxic function of vaccine-induced T cells despite measurable peripheral expansion. Fibroblast- and extracellular matrix-associated physical exclusion, impaired extravasation across dysfunctional vessels, and suppressive myeloid populations such as MDSCs collectively create a tumor-site environment that is poorly permissive for T-cell-mediated killing.172 Vaccine-induced T cells may therefore remain measurable in blood while having limited function at the tumor site. Vaccine design and translational challenges In current NSCLC vaccine development, the translational bottleneck appears to be broader than manufacturing alone. Clinical translation therefore requires the selection of antigens that are presented by tumor cells, delivery systems that support T-cell priming, platform-appropriate adjuvants, and manufacturing processes that meet clinical timelines and quality-control requirements. Antigen prediction and T-cell recognition Affinity-based screening methods such as NetMHCpan primarily estimate the equilibrium strength of pMHC binding and do not fully account for antigen processing, cell-surface presentation, pMHC stability, or T-cell recognition. pMHC stability reflects how long a peptide remains presented on the cell surface and may correlate more closely with T-cell immunogenicity than binding affinity alone.173,174 Tools like NetMHCstabpan incorporate predicted pMHC dissociation rates and can improve the prioritization of stable MHC-I ligands.175 At the algorithmic level, deep learning models trained on mass spectrometry-derived HLA ligand data provide a more direct representation of peptides selected during intracellular processing and presentation. MHCflurry and BigMHC use such ligand data to improve the prediction of naturally presented MHC-I peptides rather than relying only on in vitro binding measurements.176,177 More recent models extend prediction beyond peptide presentation to TCR–antigen binding specificity, supporting direct analysis of TCR–pMHC recognition, although their performance across HLA alleles and tumor contexts remains to be validated.178 HLA-II presentation remains more difficult to predict than HLA-I presentation because the open-ended peptide-binding groove accommodates peptides of different lengths and multiple binding registers, and available HLA-II ligand datasets are smaller and more heterogeneous.179 Models such as MARIA integrate naturally presented HLA-II ligands with gene-expression and antigen-processing features and have improved presentation prediction, but their performance across less well-represented HLA-II alleles remains uncertain.180 Non-canonical antigen sources Conventional neoantigen discovery has largely focused on nonsynonymous somatic mutations detected by exome sequencing. This strategy is biologically intuitive, because mutation-derived peptides are absent from normal tissues and may escape central tolerance. However, mutation-centered pipelines do not capture the full repertoire of tumor-presented antigens. Beyond single-nucleotide variants and small insertions or deletions, aberrant transcription, splicing, and translation can also generate peptides that are presented by MHC molecules and recognized by T cells. Alternative splicing (AS) is one such source of non-SNV antigens. Multi-omic analyses in a 279-patient hepatocellular carcinoma cohort showed that AS events occurred more frequently than somatic mutations and generated a broad repertoire of potentially immunogenic peptides. An mRNA vaccine encoding selected AS-derived candidates increased neoantigen-reactive T-cell infiltration and reduced tumor growth in vivo.181 These data support further investigation, but they remain largely preclinical and tumor-type specific, and NSCLC clinical validation is still lacking. AS-derived antigens also point to a broader limitation of mutation-centered antigen discovery: many tumor-presented peptides are not captured by conventional exome-based neoantigen pipelines. This broader non-canonical antigen space is often discussed under the term “dark genome”, which refers to immunogenic candidates derived from genomic or transcriptomic regions outside conventional protein-coding annotations. Candidate antigens may arise from endogenous retroviruses and transposable elements, exon-transposable element chimeric transcripts, intron retention, cryptic splice-site usage, RNA editing, unannotated open reading frames, upstream open reading frames, long non-coding RNA-encoded open reading frames, or proteasomal peptide splicing.182 Recent proteogenomic studies have made this concept directly relevant to NSCLC. In melanoma and NSCLC, many MHC-presented tumor antigens were found to originate from non-mutated genomic regions.183 NSCLC-focused work has also identified recurrent immunogenic antigens produced at exon-transposable element junctions.184 These studies broaden the antigen-discovery space beyond mutation-derived neoantigens, but they should not be interpreted as clinical validation of dark-genome-based vaccines. MHC presentation alone does not establish therapeutic immunogenicity, and candidate antigens still require validation for productive T-cell recognition, tumor-selective expression, recurrent or clonal presentation, and acceptable safety in normal tissues. Dark-genome-derived antigens may eventually complement SNV-based vaccine strategies, particularly in tumors with low mutational burden but intact antigen-presentation machinery. At present, however, their use in NSCLC remains investigational. Clinical translation will require reproducible expression, mature proteogenomic pipelines, standardized annotation of transposable elements and endogenous retroviruses, and careful assessment of basal expression in normal tissues. In parallel, frameshift-derived neoantigens are being explored as another non-SNV antigen source. FRAME-001 (NCT04998474) is evaluating a personalized frameshift-peptide vaccine in advanced NSCLC after standard first-line pembrolizumab-based therapy in patients with stable disease.185 Self-homology in TAA-based vaccine design Alongside challenges in neoantigen prediction, TAA-based vaccines require careful assessment of self-homology and tissue expression. A recent analysis suggested that some TAAs included in clinical-stage vaccines, including CLDN6 encoded by BNT116, contain MHC-II-length sequences homologous to proteins expressed in healthy tissues.186 This raises a theoretical concern for delayed autoimmunity or regulatory T-cell skewing, although the clinical relevance for individual vaccine products remains to be established. Long-peptide epitopes with substantial homology to normal tissue proteins may also favor antigen-specific Treg induction in some contexts.187 This issue is particularly relevant for TAA or MHC-II-binding long-peptide vaccines, where self-derived CD4+ T-cell epitopes may induce regulatory rather than effector responses and thereby weaken endogenous tumor-specific immunity. In situ vaccination strategies Intratumoral in situ vaccination has been developed as a complementary approach to systemic vaccine platforms. Rather than delivering a predefined antigen, these strategies attempt to convert the tumor into a local source of antigen release, innate immune activation, and antigen presentation. Approaches include oncolytic viruses and local strategies that combine checkpoint modulation with enhanced antigen presentation, with the aim of broadening responses against endogenous tumor antigens.188 In NSCLC, this approach may be relevant to heterogeneous tumors in which a limited set of predefined antigens may not cover all malignant clones. However, most strategies remain at an early stage, and whether local immune activation can generate systemic T-cell responses rather than transient inflammation at the treated site remains uncertain. Personalized manufacturing and scalability For personalized vaccines, clinical translation is also constrained by manufacturing time and product-specific quality-control requirements.189,190 For patients with rapidly progressive disease, the interval from tumor sampling to sequencing, antigen prioritization, product release, and vaccination may exceed the available therapeutic window.191 Personalized vaccines also require individualized quality control and release testing, because each product contains a patient-specific antigen set and cannot be evaluated through conventional batch-level standardization alone. Scalable implementation will therefore depend on parallelized manufacturing, automated production workflows, standardized bioinformatic pipelines, and quality-control systems that can preserve product consistency across individualized products. Safety and long-term monitoring General tolerability and pulmonary considerations Therapeutic cancer vaccines are generally well tolerated. Most adverse events are mild to moderate and consist mainly of injection-site reactions, fever, fatigue, and flu-like symptoms.18,24 In NSCLC, vaccines are frequently combined with other treatments, and many patients have underlying pulmonary disease, making safety assessment and adverse-event attribution more difficult. These generally mild reactions may nevertheless be clinically relevant in perioperative patients or in those with limited pulmonary reserve.24,190,192 Pulmonary toxicity requires particular consideration in NSCLC. Immune-mediated pneumonitis is a particular concern in patients with prior thoracic radiotherapy or pre-existing interstitial lung disease, and its clinical consequences may be greater in those with limited pulmonary reserve.193,194 Whether vaccination increases the risk of ICI-related pneumonitis remains unclear. Safety data from other tumor types are reassuring, but NSCLC-specific data are still needed. In the randomized phase IIb KEYNOTE-942 trial, mRNA-4157/V940 combined with pembrolizumab in high-risk melanoma showed a safety profile consistent with the known effects of the individual treatments, with no new safety signals during extended follow-up.195 Future NSCLC trials should prospectively record pulmonary symptoms, radiographic changes, corticosteroid use, treatment interruption, and rechallenge outcomes. Scheduling and sequencing of combination regimens In combination regimens, treatment sequence may affect both safety and vaccine-induced immune responses. Chemotherapy may increase tumor-cell death and antigen release and reduce immunosuppressive cell populations.196 However, cytotoxic treatment can also cause lymphopenia and impair the induction and expansion of vaccine-induced T cells.197 Concurrent administration may align vaccination with treatment-induced antigen release, whereas sequential administration may allow lymphocyte recovery but move vaccination beyond this period of antigen release.198 Local irradiation may enhance antigen release and in situ immune priming, whereas larger thoracic radiation fields can cause prolonged lymphopenia.199,200 Thymic irradiation may represent an additional source of impaired T-cell recovery after thoracic radiotherapy. Preliminary data indicate that thymic radiation dose is associated with a subsequent decline in thymic health, while thymic health in adults is associated with thymic output, TCR diversity, and immunotherapy outcomes.201,202 Whether these changes affect vaccine immunogenicity or timing remains unknown. Preclinical studies suggest that PD-1/PD-L1 blockade may support vaccine-induced T-cell function.103 However, current clinical evidence does not establish whether ICIs should be administered before, during, or after vaccine priming, and overlapping immune-related toxicities may complicate adverse-event attribution. Clinical studies have used different treatment sequences. TG4010 was initiated with the first chemotherapy cycle, whereas DCVAC/LuCa was added after chemotherapy had begun.123, 124, 125 Tecemotide and belagenpumatucel-L were administered as maintenance treatment after chemoradiotherapy or platinum-based chemotherapy, respectively, in patients without disease progression.43,137,138 Personalized vaccine manufacturing creates an additional timing constraint. In the NEO-PV-01 trial, patients received four cycles of systemic therapy during vaccine production, and vaccination began after manufacturing was completed.19 Across these studies, neither concurrent nor sequential administration showed a clear increase in serious toxicity attributable to vaccination, and vaccine-related adverse events were mainly mild local or systemic reactions. However, these studies differed in vaccine platforms, combination regimens, and study designs, and none directly compared treatment schedules. The effect of treatment sequence on safety therefore remains unclear. Platform-specific safety and long-term monitoring Safety profiles differ across vaccine platforms, and monitoring should be aligned with the delivery system and mechanism of immune activation. RNA nanoparticle or lipoplex formulations can induce injection-site reactions, fever, fatigue, and other inflammatory symptoms.16,17 Preclinical data suggest that particle composition may influence the inflammatory response to mRNA formulations.203 Standard reactogenicity assessment remains appropriate in early-phase trials, with inflammatory biomarkers evaluated when systemic symptoms are severe or prolonged. Viral-vector vaccines are also affected by vector immunity. Neutralizing antibodies resulting from prior exposure or repeated vaccination may reduce subsequent antigen delivery and alter immunogenicity.47,48 In repeated-dose studies, adverse events can be assessed according to dose number, with neutralizing antibodies monitored in homologous or heterologous prime-boost regimens. Replication-defective vectors reduce risks related to viral replication but can still induce local or systemic inflammatory reactions. The safety of cell-based vaccines is influenced by both the cell source and the manufacturing process. Autologous DC vaccines have generally been associated with limited systemic toxicity in NSCLC studies.91,204 However, leukapheresis and variation in cell source, culture conditions, maturation state, and antigen loading may affect product consistency.44 Allogeneic whole-cell vaccines may elicit immune responses against non-tumor alloantigens, and variation in antigen composition and residual immunosuppressive signals may also affect consistency between production batches.42 Safety monitoring should cover procedure-related and post-vaccination adverse events, together with relevant product-release characteristics. Long-term safety remains insufficiently characterized across vaccine platforms. Earlier large NSCLC trials of shared-antigen vaccines, including adjuvant MAGE-A3 and tecemotide after chemoradiotherapy, did not show major increases in severe adverse events, but delayed or rare immune-related toxicities were not consistently captured as dedicated long-term endpoints.55,138 In the perioperative Neo-mDC study, vaccine-induced T-cell responses were detectable for up to 19 months, and all treatment-related adverse events were grade 1 or 2.133 However, only six patients received vaccination, and protocol-defined safety monitoring ended 3 weeks after the final dose.133 The study therefore cannot exclude rare or delayed adverse events. Delayed autoimmunity remains a theoretical concern when vaccine epitopes share sequence homology with proteins expressed in normal tissues.186 Trials of repeated booster vaccination or vaccine–ICI combinations should extend safety follow-up beyond the vaccination period and report delayed autoimmune and organ-specific inflammatory events separately. Future perspectives Epigenetic priming as a vaccine adjuvant strategy Epigenetic priming may extend dark-genome antigen discovery into a therapeutic strategy. DNA methyltransferase and histone deacetylase inhibitors can induce endogenous retroviruses and other normally silenced elements, creating new antigenic transcripts and activating innate immune sensing through double-stranded RNA pathways.205 Preclinical studies also suggest that cyclin-dependent kinase 2 (CDK2) inhibition can amplify endogenous retrovirus (ERV)–IFN signaling.206 In NSCLC with limited antigenicity, epigenetic priming could increase the pool of candidate vaccine targets, provided that antigen processing and presentation remain functional. Vaccination could then expand T-cell responses against the induced antigens, although this combination remains preclinical. Important uncertainties include whether the induced antigens are sufficiently tumor selective and how the timing of epigenetic treatment affects antigen expression and toxicity. MRD-guided adaptive vaccination Longitudinal circulating tumor DNA (ctDNA) monitoring may help guide vaccine timing after initial therapy. In an MRD-guided model, ctDNA positivity, increasing molecular burden, or newly detected subclones could inform decisions regarding booster vaccination, combination therapy, or antigen-panel revision in selected personalized platforms. This approach may be most feasible in perioperative or maintenance settings, where molecular relapse can precede radiographic progression and potentially provide a window for intervention. However, this window may be limited by assay sensitivity and manufacturing time. When ctDNA levels are low after initial therapy, tumor-derived variants may remain below the detection limits of current assays, reducing the time available for antigen-panel revision and personalized vaccine production before radiographic recurrence.207 Clinical implementation would therefore require sensitive longitudinal monitoring, rapid manufacturing, predefined criteria for antigen updates and regulatory pathways that allow iterative product modification. These requirements remain difficult to meet for personalized vaccines, whose production still usually takes several weeks. Regulatory considerations for shared and personalized modalities Shared and personalized vaccines raise different regulatory considerations. Shared off-the-shelf vaccines have a fixed antigen composition, allowing standardized assessment of product identity, purity, potency, and batch consistency. Patient selection nevertheless remains part of product development. For HLA-restricted vaccines, eligibility depends on both target-antigen expression and HLA type. For dark-genome targets, transcriptional or antigen-expression assays may be more informative than DNA mutation testing alone. These assays would require analytical validation and predefined thresholds, particularly when antigen expression varies with tumor state or prior treatment. Personalized vaccines pose a different challenge because antigen sequence and composition vary between patients. Regulatory evaluation would therefore need to distinguish the patient-specific antigen component from shared platform elements, including the delivery system, adjuvant, manufacturing process, and dosing schedule. Prespecified criteria would be needed for sequencing quality, antigen selection, sequence updates, and release testing, while manufacturing time must remain compatible with ongoing clinical treatment. Safety and manufacturing deviations should be assessed for each patient, whereas pooled analyses across products using the same platform may help identify uncommon adverse events and recurring quality problems. Funding This work was supported by National Natural Science Foundation of China (No. 82522069), Beijing Natural Science Foundation (No. 7242114), Chinese Academy of Medical Sciences (CAMS) Innovation Fund for Medical Sciences (CIFMS) (Nos. 2025-I2M-QN-004, and 2025-I2M-C&T-B-054), Young Elite Scientists Sponsorship Program by China Association for Science and Technology (CAST) (No. 2024QNRC001), National Natural Science Foundation of China (No. 82473366), CAMS Cancer Hospital Collaborative Fund (Beijing–Langfang–Shenzhen) Youth Project (No. CFA202503012) and 10.13039/501100005090Beijing Nova Program (No. 202604841151). CRediT authorship contribution statement Weijing Liu: Writing – review & editing, Writing – original draft, Visualization, Conceptualization. Jiefei Han: Writing – review & editing, Writing – original draft. Kailun Fei: Writing – review & editing, Writing – original draft. Yiting Dong: Writing – review & editing, Investigation. Man Li: Visualization, Investigation. Jie Zhao: Writing – review & editing. Mingzhao Wang: Visualization, Investigation. Zhuoran Li: Visualization, Investigation. Siyuan Chen: Writing – original draft, Investigation. Jiachen Xu: Writing – review & editing, Supervision, Project administration, Funding acquisition. Amit Jain: Writing – review & editing, Supervision, Project administration, 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.