Lung cancer vaccines to improve immune checkpoint blocker treatment: current evidence and future outlook

The concept of cancer immunotherapy dates back to 1863, when Dr. Rudolf Virchow proposed the “chronic irritation theory,” linking chronic inflammation to tumor development. Although not a direct immunotherapy, this observation laid the foundation for understanding the immune system’s role in cancer. In the 1890 s, Dr. William B. Coley treated inoperable sarcomas with a mixture of heat-killed Streptococcus pyogenes and Serratia marcescens (Coley’s toxins), achieving notable clinical successes [23]. In the 1970 s, the intratumoral use of nonvirulent Mycobacterium bovis (Bacillus Calmette-Guérin, BCG) was shown to induce immune responses against various cancers, including cutaneous melanoma, partly through trained immunity [24]. In the post-genomic era, the first FDA-approved autologous dendritic cell (DC) vaccine, Sipuleucel-T, emerged in 2010 for metastatic castrate-resistant prostate cancer. Patient DCs are incubated with a fusion protein of prostatic acid phosphatase and GM-CSF, then reinfused to stimulate anti-tumor immunity. Talimogene laherparepvec (T-VEC), the first FDA-approved oncolytic virus in 2015, acts as an in-situ cancer vaccine in melanoma. Intratumoral injection lyses cancer cells, releasing tumor-associated and tumor-specific antigens (TAAs and TSAs) that, together with the virus, activate the immune system.

Virally induced cancers account for 15–20% of human malignancies, including HPV-related cervical and oropharyngeal cancers, HBV/HCV-related hepatocellular carcinoma, and EBV-associated lymphomas. Preventive vaccines against HPV and HBV have successfully reduced the incidence of these cancers, but such vaccines are not effective against established tumors, especially non-virus-related cancers like lung cancer [25]. Therapeutic vaccines targeting TAAs or TSAs from nonviral cancers have also shown limited efficacy when used alone.

Lessons from the COVID-19 pandemic demonstrate that most lung cancer patients can mount adequate innate and adaptive immune responses to vaccines [26]. Vaccinology, the science of developing vaccines, has increasingly been applied to cancer treatment [27]. Therapeutic cancer vaccines are designed to elicit de novo T-cell responses by targeting one of two major types of cancer antigens [28, 29]: (1) Tumor-Associated Antigens (TAAs): These are expressed at higher levels on cancer cells compared to normal cells and are relatively restricted to tumor cells; and (2) Tumor-Specific Antigens (TSAs) or Cancer Neoantigens: These result from genetic alterations that accumulate exclusively in cancer cells but not in normal cells during tumorigenesis or through epigenetic processes. Next-generation sequencing (NGS) and computational bioinformatics enable rapid identification of patient-specific neoantigens and assessment of immune responses. However, only ~10% of somatic mutations produce immunogenic peptides [30, 31]. To elicit an effective response, neoantigens must: [1] alter protein expression [2], be properly processed and presented on human leukocyte antigen (HLA) or human major histocompatibility complex (MHC) molecules for T-cell receptors (TCRs) recognition, and [3] exist as clonal or trunk rather than subclonal mutations. Compared with TAAs, neoantigens can bypass central tolerance and elicit stronger tumor immunity [28, 32].

Cancer vaccines train the immune system to recognize and attack tumor cells (Fig. 2). Antigenic epitopes, either public, shared, or personalized, are selected, combined with adjuvants, and administered via defined routes. Antigens are phagocytosed by dendritic cells (DCs), which present them on HLA-I and HLA-II molecules to prime CD8 + cytotoxic and CD4 + helper T cells in lymphoid tissues. CD4 + T cells facilitate maturation of CD8 + T cells and B cells, while B cells produce antibodies that mediate antibody-dependent cellular cytotoxicity (ADCC) via NK cells, releasing additional tumor antigens. Booster vaccinations amplify memory T and B cells, sustaining immune responses. This coordinated activation enhances anti-tumor immunity, which can be further potentiated by ICIs.

Carcinogens such as tobacco smoke, airborne particulate matter smaller than 2.5 microns (PM2.5), and chronic lung inflammation can induce somatic mutations in the bronchi, bronchioles, or alveoli. Gain-of-function mutations in oncogenes, including epidermal growth factor receptor (EGFR), Kirsten rat sarcoma viral oncogene homolog (KRAS), anaplastic lymphoma kinase (ALK), and human epidermal growth factor receptor 2 (HER2, also known as ERBB2), and loss-of-function mutations in tumor suppressor genes such as tumor protein p53 (TP53), RB transcriptional corepressor 1 (RB1), phosphatase and tensin homolog (PTEN), and cyclin-dependent kinase inhibitor 2 A (CDKN2A) drive malignant transformation. These driver mutations can accumulate over years or decades in histologically normal lung tissue or precancerous lesions, preceding the development of lung adenocarcinoma (LUAD) from atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), or minimally invasive adenocarcinoma (MIA), alongside numerous passenger mutations.

Mutations that occur in antigenic epitopes can generate neoantigens that are recognized as intolerable by the immune system. All nucleated cells, including cancer cells, express class I human leukocyte antigen (HLA-I) molecules, which present intracellular peptides (HLAp) on the cell surface. While epitopes from normally expressed proteins are tolerated by the immune system due to thymic selection, ectopically expressed proteins, such as cancer-testis antigens like New York esophageal squamous cell carcinoma 1 (NY-ESO-1), or neoantigens presented by HLA-I are recognized as non-self. Consequently, adaptive T cells mount immune responses against cancer cells. Innate immune natural killer (NK) cells also contribute by inducing apoptosis in stressed cancer cells [41].

Targetable oncogenic mutations are identified in approximately 60% of lung adenocarcinoma patients in Western populations and 80% in Asian populations [43, 44]. Commonly identified oncogenes in NSCLC include EGFR, ALK rearrangements, KRAS, ROS1, BRAF V600E, NTRK1/2/3, MET exon 14 skipping mutations, RET, and ERBB2 (HER2). Targeted therapies using tyrosine kinase inhibitors (TKIs) have achieved substantial clinical success in these patient populations; however, therapeutic resistance typically develops within months to years of treatment [45]. Evidence indicates that patients harboring these oncogenic mutations generally exhibit minimal response to ICIs, likely due to low tumor mutation burden and reduced immunogenicity [46–49]. These observations highlight the need for novel therapeutic strategies in oncogene-driven NSCLC. The widespread adoption of NGS has facilitated the rapid identification of targetable oncogenes, positioning them as promising candidates for the development of “off-the-shelf” vaccine therapies.

Current delivery platforms for personalized vaccines mirror those of off-the-shelf approaches and include peptide-based, nucleic-based, cell-based, or engineered viral vector-based formats. However, the manufacturing process can be more complex. It involves biospecimen acquisition, NGS mutation profiling, neoantigen identification with immunogenicity prediction, vaccine synthesis, and final product release. The typical turnaround time from sample collection to vaccine administration is approximately 3–4 months, posing significant challenges in patients with rapidly progressing disease [33, 60].

The mucosal immune system constitutes the most extensive peripheral immune network and serves as a frontline defense against microbial and dietary antigens. The respiratory tract represents the body’s second-largest mucosal surface area, following the gastrointestinal tract; consequently, mucosal immunity plays a critical role in the prevention and treatment of lung infections [72, 73]. Recent studies have shown that mucosal (intranasal) vaccination outperforms systemic intramuscular vaccination in controlling head and neck tumors, primary lung tumors, and lung metastases from breast cancer [74, 75]. A central mechanism underlying this enhanced efficacy of mucosal immunity is the robust induction of tissue-resident memory T cells (Trm) within the respiratory tracks and lung parenchyma, which can serve as a key determinant of vaccine response [76]. These Trm cells provide a new subset of long-lived CD8 + memory T cells. They differ from circulating memory T cells in their tissue location, gene expression, and function, providing long-term, rapid local protection against reinfection and cancer but can also cause autoimmune conditions by attacking healthy tissue. In addition, Trm cells are also predictive of response to anti–PD-1 therapy [77, 78]. Furthermore, intranasally delivered cancer vaccines can prevent tumor onset in spontaneous KRAS-mutant lung cancer mouse models [79]. However, this protection appears substantially diminished in metastatic settings, where disseminated tumors often arise outside the lung’s mucosal immune compartment. Further studies are needed to define the optimal clinical contexts and combinatorial strategies for effective mucosal vaccination.

Outside the lung cancer, cancer vaccines are generally more effective when administered in the adjuvant setting, following definitive surgery or other curative-intent therapy, than in the metastatic setting [86, 87]. Several biological and immunological principles support this paradigm. First, following resection of early-stage NSCLC, patients typically have either no or low tumor burden, i.e., minimal residual disease (MRD). In this context, the immune-mediated eradication of residual tumors in micrometastases is more effective than targeting bulky or disseminated metastatic lesions [88]. Second, the adjuvant setting is characterized by an absent or minimal presence of immunosuppressive TME, with low level of regulatory T cells, MDSCs, and suppressive cytokines in the immunosuppressive TME. This environment allows vaccines to more effectively prime, expand, and differentiate tumor-specific T cells capable of eliminating residual disease [89, 90]. In contrast, metastatic lung cancer is marked by chronic antigen exposure, which drives profound T-cell exhaustion and diminishes cytotoxic function, proliferative capacity, and responsiveness to vaccination [91]. Third, in the adjuvant setting, vaccines can be administered alongside or after treatments that enhance antigen release (e.g., surgery, radiation, some chemotherapies), providing a natural boost to vaccine-induced responses. Conversely, patients with metastatic disease often receive therapies that impair immune function, further limiting vaccine efficacy. Fourth, early-stage lung cancers harbor fewer genomic mutations and immune-evasion mechanisms than metastatic lesions, which frequently downregulate antigen presentation pathways or disrupt interferon signaling to escape immune surveillance [92]. Fifty, clinical experience with ICIs in NSCLC further supports this framework: ICIs consistently demonstrate greater benefit in early-stage and perioperative settings than in metastatic disease [86]. Finally, a recent study demonstrated that SARS-CoV-2 mRNA vaccines can sensitize tumors to ICIs, even in immunologically “cold” cancers, highlighting the potential of broadly applicable, off-the-shelf immunomodulatory strategies to overcome key barriers such as low immunogenicity, immune suppression, and limited T-cell infiltration [93]. If validated in prospective clinical trials, this approach could rapidly transform standard-of-care immunotherapy by enhancing both its effectiveness and accessibility. Together, these findings indicate that the adjuvant setting provides a uniquely favorable immunological window in which cancer vaccines are more likely to induce durable, protective antitumor immunity, reduce recurrence risk, and improve long-term outcomes in lung cancer. In the metastatic setting, cancer vaccines alone have modest clinical antitumor effect and are frequently given in combination with effective therapeutics for a rapid, robust, and effective antitumor effects needed to control the cancer.”

ICI therapy has shown limited efficacy in oncogene-driven NSCLC, with no significant survival benefit observed in patients with EGFR-mutated tumors [94]. Most oncogene-driven NSCLC, with the exception of KRAS-mutant and BRAF V600E-mutant subtypes, tend to have low TMB, resulting in reduced immunogenicity [48, 49, 95, 96]. Consequently, small molecule TKIs remain the standard first-line treatment for EGFR-mutated and ALK-rearranged NSCLC. However, resistance to TKIs is almost universally inevitable [97, 98]. Preclinical studies suggest that the poor response to ICIs in ALK-driven NSCLC may be due to inadequate CD8 + T cell priming against ALK antigens. Interestingly, this limitation could potentially be overcome by targeted vaccination strategies aimed at restoring tumor immunogenicity [53]. Nonetheless, the feasibility of personalized cancer vaccines in oncogene-driven NSCLC remains uncertain. A recent personalized vaccine trial in pancreatic cancer, another low-TMB tumor type, revealed that low TMB was associated with poor vaccine efficacy and significant manufacturing challenges.

Accordingly, the application of personalized vaccine strategies in oncogene-driven NSCLC presents several unresolved obstacles. Optimal combination approaches incorporating oncogenic peptides have not yet been defined. Ongoing trials are evaluating ALK peptide vaccines in combination with ALK inhibitors as front-line therapy, notably without concurrent ICIs. In parallel, CIMAvax-EGF is being investigated in combination with ICIs following the development of acquired TKI resistance (NCT05950139↗; NCT06095934↗).

The limited clinical success of cancer vaccines is partly attributable to suboptimal neoantigen selection strategies. McGranahan and colleagues have outlined several key principles for effective neoantigen selection: (1) Clonality: Neoantigens should be clonal, meaning they are present in a large proportion of cancer cells. This helps prevent immune evasion through T cell–mediated selection against subclonal neoantigens. (2) Distinction to Self: Neoantigens should be highly distinct from self or known antigens to avoid immune tolerance and enhance immunogenicity. (3) Antigenicity: Neoantigens must be both transcribed and translated to effectively stimulate an immune response. (4) Genomic Stability: Neoantigens should arise in regions with a low likelihood of deletion or alteration during tumor evolution, which can be estimated based on their genomic context. (5) Broad HLA Binding: Neoantigens capable of binding to multiple HLA alleles are preferred to reduce the risk of immune resistance due to HLA loss [99]. In addition, some TAAs harbor epitopes shared with other human proteins expressed in normal tissues, which may increase the risk of off-target T-cell reactivity and compromise both vaccine safety and efficacy. Accordingly, rigorous epitope filtering strategies to exclude shared sequences within TAAs have been proposed as an important consideration in neoantigen selection [100].

Accurately identifying neoantigens that are both presented by MHC molecules and capable of inducing an immune response remains a major challenge [101]. This process depends not only on peptide–MHC binding affinity but also on the subsequent immunogenicity of the bound peptide [102]. The most commonly used approach involves in silico prediction of binding affinities between thousands of peptides and a limited subset of MHC class I alleles [103, 104]. However, these methods often oversimplify the complexity of antigen presentation. Indeed, in vivo studies have shown that fewer than 6% of predicted neoantigens elicit a functional T-cell response, and even peptides with low predicted affinity can be immunogenic [105, 106]. Alternative strategies include direct identification of mutated peptides presented on MHC molecules from primary tumor tissue, though this technique is limited by low sensitivity. Another approach involves isolating neoantigens that have already triggered a T cell response, though this is restricted to pre-existing immune reactivity [105, 107]. Additionally, prioritizing neoantigens that emerge before a whole-genome doubling (WGD) event may enhance vaccine efficacy, as neoantigens arising post-WGD are more likely to be lost during tumor progression [54]. Refining these selection methods is critical to enhance the immunogenicity and clinical efficacy of personalized cancer vaccines.

Host genetic variation in HLA haplotypes plays a crucial role in shaping antigen presentation, defining immune responses, generating autoantibodies during cancer immunotherapy, and determining the association with irAEs [108, 109]. HLA Class I molecules (HLA-A, -B, and -C) primarily present intracellular peptides to CD8 + T cells, whereas HLA Class II molecules (HLA-DR, -DQ, and -DP) present extracellular peptides to CD4 + T cells (Fig. 6). The loss of expression of HLA molecules and molecules related to antigen processing has been contributing to the low efficacy of therapeutic cancer vaccine in the metastatic disease. 30% of NSCLCs show selective downregulation on HLA-I tumor cells (one or more β2M, HLA-A, HLA-B/C subunits) [110–114]. HLA loss can arise through multiple mechanisms, including irreversible genetic alterations that may require gene-editing approaches for restoration, as well as more common reversible transcriptional, post-transcriptional, or epigenetic changes that can potentially be restored through pharmacologic interventions such as interferon signaling or epigenetic modulation [113, 115, 116]. Furthermore, better quality of antigen presentation in adjuvant setting than the metastatic setting: After tumor removal, antigen-presenting cells can process tumor antigens released during surgery or minimal residual disease in a context that favors activation rather than suppression. In metastatic disease, chronic exposure to tumor antigens leads to T-cell exhaustion and suboptimal antigen presentation. There are unmet needs to incorporate empirical HLA class I immunopeptidomics into neoantigen selection pipelines, as computationally predicted binders do not always correspond to peptides that are naturally processed and presented on tumor cells [117]. However, this is challenging due to the diverse pool of HLA haplotypes that vary to age, race and ethnicity.

Although neoantigens hold strong potential as targets for T cell-mediated immune responses, immune activation does not consistently translate into clinical benefit. Higher neoantigen loads are generally associated with improved outcomes, and neoantigen-targeted therapies have demonstrated tumor-killing activity in both preclinical and clinical models. Nonetheless, many cancer vaccine trials have reported underwhelming results. Despite eliciting robust immune responses, many of these trials report limited or no improvements in clinical outcomes.

Currently, there is no universally accepted gold standard for measuring vaccine-induced immunogenicity. Surrogate markers and the timing of immune assessments vary widely across studies. Traditional approaches often measure cytokine production (e.g., IFN-γ, TNF-α, IL-2) or quantify T cell frequencies [66, 118, 119]. While informative, these metrics frequently lack specificity for vaccine-induced responses and may not accurately reflect functional antitumor activity. More recent studies aim to improve specificity by identifying neoantigen-specific antibodies or T cells. Although these markers offer greater precision, they still fail to capture the full complexity and functional dynamics of the immune response. For example, central memory T cells support durable immunity through long-term persistence and recall capacity, while cytotoxic CD4⁺ T cells are increasingly recognized as important contributors to antitumor immune responses [120–123]. Figure 7 illustrates the time course of the adaptive immune responses to cancer vaccines.

Functional assays are being employed to evaluate T-cell trafficking and cytotoxic potential at tumor sites, providing more detailed insights into immune activity. However, these methods capture only a fraction of the broader immune response. The immune system is a highly complex, multilayered network, and accurately quantifying its components remains a significant challenge. Practical limitations further complicate these efforts, including the limited availability of patient-derived samples and the need for sophisticated bioinformatics and high-dimensional data integration. These barriers continue to hinder the establishment of standardized, universally accepted metrics for evaluating vaccine-induced immunity.

Similar to ongoing debates regarding the optimal duration of ICI therapy, whether treatment should be limited to 1–2 years or continued indefinitely, the ideal timing and duration of cancer vaccine administration remain unresolved [124]. Vaccine regimens vary considerably across clinical trials, typically involving three to ten total doses. These are often coordinated with concurrent chemotherapy or ICI therapies to reduce patient burden and streamline logistics.

Another critical consideration is the choice of prime-boost strategy. Most clinical studies employ a homologous prime-boost approach, repeatedly delivering the same immunogen to amplify the pool of memory B and T cells generated by the initial dose [70, 125]. However, repeated administration of the same vaccine may lead to diminished cellular responses over time, largely due to the formation of neutralizing antibodies against the vaccine vector or its components [126]. To overcome this limitation, heterologous prime-boost strategies, using different vaccine platforms or vectors for priming and boosting, have been explored. These approaches have demonstrated robust and durable immune responses in several viral vector-based trials, although their superiority over homologous strategies has not been definitively established [71]. Finally, the sequencing of cancer vaccines and ICIs is under active investigation. Optimizing timing to permit effective CD8 + T cell priming while maintaining long-term memory responses may be critical to overcoming resistance and enhancing the durability of antitumor immunity [88, 127, 128].

Despite strong scientific rationale, therapeutic cancer vaccines have historically shown modest immunogenicity and limited clinical benefit in lung cancer. Several intrinsic biological and immunological barriers limiting the efficacy of lung cancer vaccines that have historically achieved limited clinical activity. Overcoming these foundational obstacles will require combination strategies that remodel the TME, boost antigen presentation, and enhance T-cell infiltration and functionality. To date, the most widely explored approach has been the combination of cancer vaccines with ICIs, particularly in personalized treatment settings [129, 130].

Beyond anti–PD-(L)1 based combinations, additional strategies targeting key immunosuppressive pathways within the TME may further augment vaccine efficacy in lung cancer. For example, approaches aimed at reducing regulatory T cells, such as low-dose cyclophosphamide, anti-CD25 antibodies, or CCR4 antagonists, can mitigate Treg-mediated suppression and improve the expansion and function of vaccine-induced effector T cells [131]. In parallel, therapies directed against MDSCs, including CXCR2 inhibition, arginase or iNOS blockade, and PI3Kγ inhibitors, may alleviate myeloid-driven suppression and facilitate T-cell migration into lung tumors [132, 133]. Targeting the HGF/MET axis represents another promising avenue, as MET activation promotes immune exclusion, tumor invasiveness, and resistance to immunotherapy; MET inhibitors may therefore potentiate vaccine-induced infiltration and cytotoxicity [134]. Similarly, CSF1/CSF1R blockade can reprogram tumor-associated macrophages from an immunosuppressive, M2-like phenotype toward a pro-inflammatory state that enhances antigen presentation and supports vaccine-primed T-cell responses [135]. Inhibition of the Wnt/β-catenin pathway, which is frequently implicated in “immune-cold” lung tumors lacking T-cell infiltration, may further improve responsiveness by converting non-inflamed tumors into more permissive, T-cell–accessible TME [136]. Together, these combinational strategies underscore the opportunity to combine cancer vaccines with agents that remodel the suppressive lung TME, thereby amplifying T-cell priming, improving infiltration into tumor sites, and ultimately enhancing the depth and durability of anti-tumor immunity.