Pipe dream to pipeline: Journey of cancer vaccines and the road ahead

Summary

Over the past decade, clinical trials of therapeutic cancer vaccines have evolved substantially in scope and design, renewing interest in this immunotherapy modality. Advances in sequencing technologies and an improved understanding of the tumor microenvironment have enabled precise targeting of tumor neoantigens, accelerating the development of personalized cancer vaccines. In this review, we critically evaluate the current landscape of cancer vaccines, particularly neoantigen-based approaches, in light of recent clinical trial data. Although cancer vaccines have historically demonstrated limited efficacy as monotherapies, growing evidence suggests enhanced clinical benefit when combined with other anti-cancer treatments, including immune checkpoint inhibitors. Drawing on these findings, we summarize key lessons from past efforts and highlight persistent knowledge gaps that limit broader clinical success. Finally, we discuss emerging strategies to overcome these challenges, with the aim of improving vaccine efficacy and facilitating the integration of cancer vaccines into standard-of-care treatment for cancer patients.

Graphical abstract

Onkar et al. discuss the current landscape of therapeutic cancer vaccines in the context of recent clinical trials, with a special focus on neoantigen vaccines. They also highlight challenges and scope for improvement such as the need for enhanced strategies for target identification, validation, and immune-monitoring practices.

Introduction

The discovery of vaccines that can promote robust and long-lasting immune responses against a host of infectious diseases has proven to be a turning point in the story of human survival. The leap from vaccination against infectious diseases to treating cancer was first made by Dr. Coley over a century ago through the use of a cocktail of inactivated bacteria, widely known as Coley’s toxins, to elicit an anti-tumor response.¹ Since then, for decades, cancer vaccines, designed to augment immune recognition and targeting of tumor antigens, have been the subject of active and intensive investigation. The primary goal of a cancer vaccine is to initiate a cascade of immune activation events leading to the priming of antigen-specific T cells, which can then mount effective tumor control and potentially generate long-term memory.^2,3 In the past, numerous cancer-targeted vaccine approaches have been tested in early-phase clinical studies. Yet, only one vaccine called sipuleucel-T (Sip-T), an autologous cellular vaccine targeting prostatic acid phosphatase (PAP) in patients with metastatic prostate cancer, has received Food and Drug Administration (FDA) approval. However, despite getting FDA approval, Sip-T failed to induce long-lasting clinical benefits.⁴ Other antigen-agnostic immune-modulating agents, namely, talimogene laherparepvec (T-VEC), an oncolytic virus delivered intratumorally in patients with unresectable melanoma, and Bacillus Calmette-Guérin (BCG), a live attenuated strain of Mycobacterium bovis injected directly into the bladder of patients with non-muscle invasive bladder cancer, are also currently approved by the FDA. Despite indications of immunogenicity, with the exception of BCG, these therapies induce only a modest overall survival (OS) (4 months with Sip-T) and progression-free survival (PFS) benefit (29.5% with T-VEC).^4,5

In contrast, over the last 15 years, immunotherapy with immune checkpoint inhibitors (ICI) has led to impressive complete responses in various types of hematological and solid malignancies. However, despite the remarkable efficacy rates of ICIs, such as up to 100% complete clinical response rate in some patients with locally advanced, mismatch repair-deficient colorectal cancer, the treatment overall only benefits a small fraction of patients (∼10%–15%), mostly those harboring tumors with high mutational loads and T cell infiltration.⁶

Thus, there is a need to further investigate the complexities of the tumor microenvironment (TME) and develop complementary treatment modalities such as vaccines that can augment and prolong the benefits of ICI universally. While ICIs function by potentiating pre-existing T cell responses, cancer vaccines offer the ability to prime and activate tumor-specific T cells de novo. Even though cancer vaccines have shown to have limited benefit as a monotherapy, there is an opportunity to unleash their potential by combining vaccine therapy with ICIs or other modalities like chimeric antigen receptor, T cell receptor (TCR) T cell therapies, and other modalities such as antibody-drug conjugates. This, along with our enhanced understanding of how the choice of antigen and delivery systems impact vaccine-induced immunity, has reinvigorated an interest in cancer vaccines.

In this review, we will describe the mechanisms mediating cancer vaccine immunogenicity and discuss recent advancements in vaccine design enhancing the efficacy of cancer vaccines, with a special focus on neoantigen-targeted vaccines.

Factors impacting cancer vaccine efficacy: TME components and immune resistance

The operating principle behind the design of cancer vaccines is derived from the cancer immunity cycle wherein dead or dying tumor cells release antigens that are taken up by dendritic cells (DCs) or other professional antigen-presenting cells (APCs).^7,8,9 These DCs then present the tumor antigens and activate naive CD4⁺ and CD8⁺ T cells in the draining lymph node, leading to their migration into the TME guided by chemokines such as CXCL9/10/11 and CCL5 secreted by tissue-resident DCs, macrophages, endothelial cells, and even tumor cells.^10,11,12 Within the TME, antigen-specific cytotoxic CD8⁺ T cells recognize tumor cells through major histocompatibility complex class I (MHC-I) and TCR interactions and mediate killing by releasing perforin and granzymes.^13,14 There is growing literature supporting the role of CD4⁺ T cells in controlling antitumor responses, not only as a helper subset to amplify the CD8⁺ T cell-mediated killing and promote DC maturation via CD40L-CD40 interactions but also as cytotoxic cells within the TME.^15,16,17 While T cells were the primary targets for cancer vaccines initially, emerging preclinical and translational studies have highlighted the complex network of immune cells that orchestrate both endogenous and vaccine-induced antitumor immunity. For example, lately, B cells have been highlighted for their activity within the tertiary lymphoid structures (TLSs) and their role as APCs within the TME as another key modulator of adaptive immune response.^18,19,20,21 Similarly, expansion of cross-presenting DCs (CD141⁺ cDC1 subset) in the TME and DC-natural killer (NK) cell crosstalk has been shown to regulate the response to immunotherapy (vaccines and ICI) by enhancing CD8⁺ T cell activation and mediating an influx of tumor-specific T cells into the TME.^{22,23,24,25,26} However, the TME is also enriched for immunosuppressive cells like regulatory T cells, myeloid-derived suppressor cells (MDSCs), and pro-tumorigenic macrophages, which have been shown to hamper the effectiveness of vaccination.^{27,28,29,30,31} The interplay between these immune cell subsets may lead to tumor immune resistance. In fact, in retrospect, armed with modern insight into regulation of immune networks by the tumor, it is possible to attribute previous lack of durable cancer vaccine efficacy to tumor-induced immune suppression that is now well known to actively suppress both endogenous and vaccine-induced immunity.^32,33,34,35

In addition to the immunosuppressive cellular milieu, several other tumor-intrinsic and -extrinsic mechanisms of immune resistance may emerge as the TME evolves. These include disruptions in antigen processing and presenting machinery, MHC-I downregulation,^36,37 immune-suppressive cytokine secretion,³⁸ tumor neoangiogenesis,^39,40 etc. Cumulatively, these hamper T cell recognition of target cells via MHC-I-TCR interaction and limit T cell infiltration into the TME. Another key resistance mechanism is immunoediting.^41,42,43 Either spontaneously or due to immune pressure, tumor cells may alter their mutational profile by accumulating mutations that improve their fitness and by downregulating or losing tumor antigens, especially those that are highly immunogenic and derived from non-essential passenger mutations. Given the dynamic nature of tumor evolution, often one of the biggest challenges in vaccine design is optimizing the selection of tumor antigens, prioritizing those that are highly immunogenic, abundantly expressed, and presented exclusively on the tumor cells’ MHC-I molecules. In the context of antigen immunogenicity, a key strategy frequently employed for boosting immunogenicity over tolerance by an antigen is the co-administration of adjuvants such as incomplete Freund’s adjuvant (e.g., Montanide ISA-720 and Montanide ISA-51) or TLRagonists (poly IC:LC and CpG ODN) that enhance DC activation and maturation for effective CTL and Th1 T cell priming in vaccine settings. To date, several clinical trials have leveraged this ability of adjuvants for engaging the innate immune system and amplifying the adaptive response to the target antigen.^44,45,46 Indeed, careful selection of vaccine targets and adjuvants can potentially bypass tolerance, trigger strong immune activation, and synergize with other therapeutic approaches to overcome resistance. In the next section, we discuss the different classes of tumor antigens, with an emphasis on neoantigens, which have emerged as particularly promising targets for next-generation cancer vaccines.

Targets for cancer vaccines

An effective vaccine enables the delivery of most promising antigens, to the appropriate cellular (DCs) and physiological compartments (lymph nodes) for robust priming and activation of T cell response. Tumor antigens are generally classified into tumor-associated antigens (TAAs), tumor-specific antigens (TSAs) including neoantigens, and viral antigens that are known drivers of certain tumors (like the HPV for head and neck or cervical cancer).

TAAs represent unmutated self-proteins that are abnormally overexpressed in cancer cells compared to normal cells. Examples of TAAs that have shown promise in early clinical trials include Wilms tumor 1 (WT1) in leukemia and pancreatic ductal adenocarcinoma (PDAC),^47,48 Her2/neu in HER2+ breast cancer,⁴⁹ prostate-specific antigen (PSA) and PAP in prostate cancer,^50,51 NY-ESO in melanoma,⁵² and MUC1 in non-small cell lung carcinoma (NSCLC)⁵³ (discussed in additional detail elsewhere).⁵⁴ Even though TAAs are preferentially enriched on tumor cells, their ubiquitous presence, in some cases, on normal cells makes them less-attractive targets due to potential for off-target toxicity. Additionally, being self-proteins, T cells reactive against such antigens have been subjected to thymic central tolerance limiting potential antigen-specific TCRs that can be activated and expanded post-vaccination. Controlled clinical trials have indicated a lack of durable clinical efficacy bestowed by TAA-targeting vaccines. A noteworthy example is the phase 3 program that led to FDA approval of the autologous cellular vaccine Sip-T, composed of peripheral blood mononuclear cells pulsed ex vivo with a fusion protein of PAP (a prostate TAA) and GM-CSF, for the treatment of metastatic castration-resistant prostate cancer. In a series of placebo-controlled phase 3 trials, Sip-T was found to be immunogenic and even bestow survival benefits in some cases. However, Sip-T’s failure to improve PFS, achieve disease regression, and lack of information about its mechanism of action eventually led to lack of enthusiasm in this platform for treating advanced disease.^4,55,56 More recently, randomized, double-blind placebo-controlled vaccine trials targeting MAGE-A3 as a TAA in patients with NSCLC (MAGRIT trial) and melanoma (DERMA trial) were conducted in adjuvant settings. In both trials, the cohorts that received the vaccine (n = 1,515 NSCLC, n = 895 melanoma) showed no significant clinical benefits compared to the placebo cohort (n = 757 NSCLC, n = 450 melanoma), limiting the prospect of further development of TAAs as vaccine candidates.^57,58

In contrast, neoantigens, which are a type of TSA, arise as a result of somatic mutations, including nonsynonymous single-nucleotide variations, insertion-deletions, splicing defects, defective DNA repair machinery leading to frameshift mutations, chromosomal rearrangements, as well as aberrant gene activations as a result of epigenetic changes that are restricted to tumor cells.⁵⁹ Since T cells against neoantigens are not subjected to central tolerance, these antigens represent potential highly immunogenic targets, making them attractive candidates for vaccination. Neoantigens, for use in vaccines, are selected based on various parameters including their predicted binding affinities to MHC-I alleles, expression levels, prioritization of driver mutations over passenger mutations, and manufacturability of the peptides or RNA.^44,60,61,62 In fact, over the last decade, there has been a notable shift toward personalization of vaccines through identification of neoantigens from individual tumors.

Nevertheless, to date, clinical trials targeting shared TAAs either alone or in combination with neoantigens continue to be the mainstay of vaccine trials landscape as shown in Figure 1. Broadly, these can be categorized based on the form of antigen delivery into cellular (DCs pulsed with specific peptides or whole tumor lysates used as antigen), peptide-based (antigen delivered as a peptide), DNA- or RNA-based (antigen delivered as nucleic acids), or viral vaccines (using viral vectors to deliver antigen). However, the tumor specificity and lack of immune tolerance to neoantigens have fueled a growing interest in expanding neoantigen-based vaccine therapies. In the next section, we will discuss recent advances in vaccine development, with a focus on personalized and shared neoantigen vaccines, which encompass several of such upgrades.

Figure 1.

Clinical trial landscape over the last 5 years (from 2019 to 2024)

Sunburst plot depicting ongoing and completed vaccine clinical trials in North America and Europe. Pie slices represent the proportional share of target antigens (middle layer) and vaccine type (outer layer) for each year (inner layer).

Advances in neoantigen-based vaccination trials

Peptide-based personalized neoantigen vaccines

Data from a series of clinical trials of peptide-based personalized neoantigen cancer vaccines first initiated between 2014 and 2016 have reliably demonstrated their safety and immunogenicity in numerous solid tumors like melanoma, glioblastoma (GBM), NSCLC, head and neck cancer, multiple myeloma, bladder, and renal cell carcinoma (RCC).^{2,44,45,60,63,64,65} These were all predominantly phase 1 trials, utilizing pools of synthetic long peptides (10–25 neoantigen peptides per patient) and investigating the safety and immunogenicity of personalized neoantigen vaccines administered in adjuvant settings either as monotherapy or in combination with other treatments like ICI, chemotherapy, or radiation therapy. It is noteworthy that the choice of adjuvants and combinations that enhance vaccine immunogenicity has evolved considerably from early strategies (e.g., GM-CSF) toward the use of more potent and mechanistically defined immune stimulants such as TLR agonists.

Recently, PGV001 personalized neoantigen vaccine that included up to 10 neoantigens, administered as a monotherapy in the adjuvant setting to patients with resected solid and hematological malignancies (post adoptive stem cell therapy), was reported to induce T cell (both CD4⁺ and CD8⁺) and B cell (primarily IgG) immunity against most vaccine antigens, highlighting the potential impact of cancer vaccines beyond T cell-mediated immunity.⁴⁴ Further, in a large clinical study, where 173 patients at various stages of GBM were treated with personalized neoantigen peptide vaccine (up to 19 neoantigens per vaccine, 9- to 17-mer peptides) in combination with standard of care (SoC) treatments, nearly 90% of the patients exhibited vaccine-induced T cell responses. Notably, improved overall survival was correlated with vaccine-driven immunity.⁶⁶

Peptide-based neoantigen vaccines have also been safely utilized in combination with ICIs to treat patients with bladder cancer, melanoma, NSCLC, and RCC, in both adjuvant and metastatic settings. Administration of anti-PD-1 prior to neoantigen vaccination (Neo-PV-01, comprising up to 20 neoantigens/vaccine) in patients with metastatic disease (melanoma, NSCLC, and bladder cancer) was well tolerated and elicited cytotoxic T cell responses against 48% of all immunizing epitopes and in some patients showed induction of immune responses against neoantigens not included in the vaccine, a phenomenon referred to as epitope spreading,⁶⁴ where vaccine- or therapy-induced target-specific T cells drive the killing of tumor cells, thereby releasing additional tumor antigens and leading to diversification of the immune response by activating new T cell clones that can mount a response to secondary epitopes distinct from the initial target. Combination treatment with NEO-PV-01, chemotherapy, and anti-PD1 antibodies as first-line therapy in advanced non-squamous NSCLC induced T cell responses to non-immunizing driver neoantigens KRAS G12V and KRAS G12C, further indicating the potential of neoantigen peptide vaccines triggering epitope spreading, thus amplifying the scope of vaccine coverage.⁶⁷ Notably, neoantigen vaccine (median of 15 neoantigen peptides per patient) administered intradermally or subcutaneously with poly IC:LC as adjuvant and anti-CTLA-4 antibody in patients with fully resected advanced stage RCC, elicited T cell immunity, driven primarily by CD4⁺ T cells. All 9 treated patients were reported to be recurrence free at 36-month follow-up, demonstrating the potential for prolonged clinical benefits through peptide vaccines and combination therapies.⁴⁵ More recently, a trial administering combination of PGV001 neoantigen vaccine given concurrently with anti-PD-L1 antibodies to patients with resected or metastatic bladder cancer reported vaccine-induced CD4⁺ and CD8⁺ T cells, indicating applicability of the peptide-based neoantigen vaccine platform across multiple indications and disease stages.⁶⁸

In general, peptide-based neoantigen vaccines have been found to prompt the generation of neoantigen-specific polyfunctional (IFNg⁺, TNFa⁺, and IL2⁺) and cytotoxic (CD107a⁺ and Granzyme B⁺) T cell responses.^{44,60,63,64,68} However, it should be noted that the majority of observed immune responses have been CD4⁺ T cell biased, likely due to inefficient uptake and cross-presentation of antigens by the DCs to CD8⁺ T cells.^2,44,60,63 Finally, although these studies were not designed to assess efficacy, exploratory analyses indicated that improved clinical benefits were generally correlated with the breadth of T cell responses to immunizing neoantigens and epitope spreading.^{2,44,64,65,66}

RNA- and DNA-based personalized neoantigen vaccines

To address the limitations associated with the delivery of exogenous peptides, such as poor intracellular uptake and preferential presentation via MHC class II pathways, alternative strategies have been developed to facilitate intracellular antigen expression. In this regard, nucleic acid-based vaccines using DNA and RNA have emerged as promising platforms for neoantigen delivery. DNA and RNA vaccines deliver their cargo directly into the cells and promote intracellular translation of nucleic acids into peptides and proteins, enhancing their presentation on MHC-I for CD8⁺ T cell activation.^69,70 Moreover, nucleic acid vaccines have the advantage of delivering multiple neoantigens in a single construct and simultaneously encoding immune stimulatory adjuvants like cytokines and chemokines together with target antigens.

Historically, messenger RNA (mRNA) vaccines were first developed and successfully tested in the context of cancer, despite their rise to prominence during the COVID-19 pandemic.⁶¹ The first-in-human trial using mRNA encoding personalized neoantigen epitopes delivered intranodaly in 13 patients with advanced melanoma in 2013 demonstrated the induction of polyfunctional T cell responses. Vaccine-induced T cells were predominantly CD4⁺, but cytotoxic CD8⁺ T cells were also present in the TME displaying an activated PD-1⁺ memory phenotype.^{61,71,72,73,74} Interestingly, in this trial, despite measurable responses against 5 vaccine neoantigens ex vivo, one patient displayed rapidly progressing disease but was reported to achieve complete response after compassionate anti-PD-1 treatment. Analysis of T cells and PD-L1 expression in the newly developed lesions showed presence of PD-1⁺ neo-epitope-specific CD8⁺ T cells, highlighting the potential for synergy between vaccination and ICI treatment.^61,75

More recently, in a combination treatment trial, up to 20 neoantigens per patient were delivered utilizing the mRNA-lipoplex platform (autogene cevumeran) in conjunction with anti-PD-L1 antibody (atezolizumab) followed by SoC chemotherapy (mFOLFIRINOX) in patients with resected pancreatic ductal adenocarcinoma (PDAC). The treatment induced robust CD8⁺ T cell responses in 50% of the treated cohort (n = 16), the breadth and magnitude of which correlated with prolonged recurrence-free survival (RFS) in an otherwise lethal disease.⁶⁹ A follow-up study reported additional evidence for CD8⁺ T cell-mediated vaccine-induced long-term protection using sophisticated techniques for T cell clone tracking and estimation of clone life spans. The analysis demonstrated that 86% (median) of the neoantigen-specific clones per patient persisted at 3 years post vaccination in responders, and longitudinal transcriptional analysis showed transition from proliferative (during priming dose) to effector (post final vaccine) to a tissue-resident memory T_RM-like (at 3 years) T cell phenotype.⁷⁶ Notably, assessment of differences between responders and non-responders in this study indicated better vaccine and clinical responses in patients with more clonal tumors, which are potentially better suited for enhanced and more widespread immune recognition of target antigens, as well as intact lymphoid organs like spleen, for effective uridine-mRNA-lipoplex-based antigen uptake and T cell stimulation. A randomized phase 2 clinical trial (IMCODE 003, NCT05968326) is currently underway and is anticipated to provide additional evidence for these key observations in a larger cohort. Another randomized two-arm phase 2 clinical trial testing the efficacy of anti-PD1 treatment alone or in combination with personalized neoantigen mRNA vaccine (lipoplex containing mRNA-4157) encoding 34 neoantigens in patients with resected high-risk melanoma is currently underway.⁷⁷ Interim analysis from the trial showed prolonged RFS in patients who received combination therapy (n = 107) compared to anti-PD-1 monotherapy (n = 50) (79% vs. 62% at 18 months). Despite the relatively short follow-up, this trial is the first randomized trial to provide statistically backed evidence for the role of neoantigen vaccines in enhancing clinical benefits of ICI. The study also noted improvement in distant metastasis-free survival at 18 months (92% vs. 77%) in combination treatment group compared to monotherapy group. Across the two cohorts, the study found a similar distribution of grade 3–4 immune-related adverse events in ∼35% of cohort, underscoring the safety of adding neoantigen vaccine therapy to ICI treatment for high-risk patients.⁷⁷

A few recent clinical studies have evaluated the feasibility and immunogenicity of DNA vaccines as well.^70,78 Notably, a clinical trial demonstrated that a DNA vaccine (GNOS-PV02) encoding 40 neoantigens and interleukin (IL)-12 combined with anti-PD-1, delivered in the neoadjuvant setting to a cohort of 11 patients with advanced-stage hepatocellular carcinoma (HCC), was able to induce an immune response despite the various mechanisms of immune suppression at play due to tumor burden. The treatment achieved an objective response rate of 30.6%, which corresponded with neoantigen-specific T cell responses. Overall, while DNA vaccines can induce promising immunity, whether DNA-based vaccines can induce a clinically relevant and CD8⁺ T cell-biased immunity remains to be confirmed. Together, these studies have established the ability of personalized neoantigen vaccines to induce T cell responses that are robust, lasting, potentially correlated with clinical benefit, and with promising evidence of synergies with other immunotherapies. A summary of ongoing clinical trials utilizing various platforms for target neoantigen delivery either as monotherapy or in combination with other therapies is listed in Table 1.

Table 1.

Select ongoing early phase 1/2 neoantigen clinical trials (from clinicaltrials.gov)

Antigen type	Vaccine type	Target neoantigen	Combination	Disease type/setting	Current (estimated enrollment)	Trial registration number
Shared neoantigens	Peptide	mCalR	Monotherapy	MPN	10 (10)	NCT05025488
		mKRAS	Monotherapy	High-risk pancreatic cancer	37 (25)	NCT05013216
		mKRAS	Nivolumab (anti-PD-1), ipilimumab (anti-CTLA4)	Colorectal/pancreatic cancer	27 (30)	NCT04117087
		mRAS	balstilimab (anti-PD-1)	Pancreatic cancer	24 (24)	NCT05638698
		mKRAS	botensilimab (anti-CTLA4), balstilimab (anti-PD-1)	Colorectal/pancreatic cancer	54 (50)	NCT06411691
		mALK	Monotherapy	NSCLC stage IV with ALK fusion protein expression	12 (12)	NCT05950139
		mKRAS, NRAS	Monotherapy	Pancreatic/colorectal cancer	158 (156)	NCT05726864
		mKRAS	Daratumumab (anti-CD38), nivolumab (anti-PD-1)	Pancreatic cancer/refractory non-small cell lung cancer	54 (54)	NCT06015724
		DNAJB1-PRKACA fusion	Atezolizumab (anti-PD-L1)	Fibrolamellar hepatocellular carcinoma	20 (20)	NCT05937295
		DNAJB1-PRKACA fusion	Nivolumab (anti-PD-1), ipilimumab (anti-CTLA4)	FLC	56 (12)	NCT04248569
		mKRAS	Monotherapy	Pancreatic/colorectal/ovarian/non-small cell lung carcinoma/cholangiocarcinoma/bile duct/gallbladder carcinoma	25 (159)	NCT04853017
		mKRAS	Nivolumab (anti-PD-1), ipilimumab (anti-CTLA4)	Non-small cell lung cancer	12 (12)	NCT05254184
		Neoantigen HSP	Zalifrelimab (anti-CTLA4), balstilimab (anti-PD-1)	Diffuse intrinsic pontine glioma/diffuse midline glioma, H3 K27M-mutant	36 (36)	NCT04943848
		H3K27M mutation	Atezolizumab (anti-PD-L1)	Newly diagnosed H3-mutated glioma	15 (15)	NCT04808245
		mKRAS/NRAS	Monotherapy	Multiple myeloma/smoldering multiple myeloma	20 (20)	NCT05841550
	Viral	Nouscom, undisclosed	Ipilimumab (anti-CTLA4)	Myeloproliferative neoplasms	14 (60)	NCT05444530
		∗GRT-C901/-R902 mKRAS	Atezolizumab (anti-PD-L1), ipilimumab (anti-CTLA4), chemo, bevacizumab (anti-VEGFA)	Colorectal neoplasms	700 (665) ∗Phase 2/3	NCT05141721
		Nous 209- undisclosed	Pembrolizumab (anti-PD-1)	Solid tumor, adult	115 (34)	NCT04041310
		Nous 209- undisclosed	Monotherapy	Colorectal carcinoma/Lynch syndrome	45 (45)	NCT05078866
		GRT-C903/-R904 mKRAS	Aldesleukin (IL-2), chemo, KRAS-TCR-transduced PBL	Metastatic solid cancers/colorectal/breast/ovarian/non-small cell lung cancer/gastrointestinal/genitourinary cancer	210 (210)	NCT06253520
		HER2 overexpression	Pembrolizumab (anti-PD-1)	Breast cancer/HER2+ breast cancer	8 (39)	NCT03632941
	Cellular	mESR1 loaded DCs	Elacestrant (SERD)	Breast cancer metastatic breast cancer/HER2-negative breast cancer	18 (18)	NCT06691035
		H3 G34 mutation pulsed DCs	Monotherapy	Diffuse hemispheric glioma, H3 G34 mutant	6 (6)	NCT06342908
Personalized neoantigens	DNA	Individualized private mutations	Anti-PD-1 and anti-PD-L1	Solid tumor	20 (20)	NCT06631079
			Anti-PD-1 and chemotherapy	Triple-negative breast cancer	8 (8)	NCT06631092
			Monotherapy	Pediatric recurrent brain tumor	7 (10)	NCT03988283
			Anti-PD-1 and anti-PD-L1	Solid tumors, adult	6 (6)	NCT05354323
			Plasmid-encoded IL-12	Glioblastoma	9 (30)	NCT04015700
			Durvalumab (anti-PD-L1)	Extensive-stage small cell lung cancer	6 (40)	NCT04397003
			Retifanlimab (anti-PD-1)	Unmethylated glioblastoma	12 (12)	NCT05743595
	RNA	Individualized private mutations	Atezolizumab (anti-PD-L1), chemotherapy	Pancreatic cancer	29 (20)	NCT04161755
			Pembrolizumab (anti-PD-1)	Melanoma	267 (150)	NCT03897881
	Peptide	Individualized private mutations	Pembrolizumab (anti-PD-1)	Advanced cancer	30 (10)	NCT03568058
			Standard of care, TTF	Glioblastoma	13 (20)	NCT03223103
			Sotigalimab (anti-CD40), pembrolizumab (anti-PD-1)	Metastatic colorectal/pancreatic cancer	150 (40)	NCT02600949
			Radiation therapy, anti-PD-1	Advanced NSCLC	10 (10)	NCT06751901
			Pembrolizumab (anti-PD-1) and chemotherapy	Locally advanced and metastatic solid tumors	36 (36)	NCT05269381
			Chemotherapy	Pancreatic cancer	33 (30)	NCT05111353
			Ipilimumab (anti-CTLA4)	Kidney cancer	19 (20)	NCT02950766
			Lenalidomide	Smoldering plasma cell myeloma	30 (30)	NCT03631043
			CDX-301 (FLT3L)/GM-CSF	Prostate cancer	27 (27)	NCT05010200
			Monotherapy	Ewing sarcoma/rhabdomyosarcoma/synovial sarcoma	30 (30)	NCT06094101
			Pembrolizumab (anti-PD-1) and chemotherapy	Lymphocytic leukemia	15 (10)	NCT03219450
			Nivolumab (anti-PD-1)	Locally advanced and metastatic breast, non-small cell lung cancer, and melanoma	25 (20)	NCT05098210
			Nivolumab (anti-PD-1)	Ovarian cancer	22 (30)	NCT04024878
			Sacituzumab govitecan, chemotherapy, tremelimumab (anti-CTLA4)	Metastatic breast cancer/triple-negative breast carcinoma	70 (70)	NCT03606967
			Pembrolizumab (anti-PD-1) and rituximab (anti-CD20)	Follicular lymphoma	20 (20)	NCT03361852
			Nivolumab (anti-PD-1), ipilimumab (anti-CTLA4)	Cutaneous melanoma	11 (20)	NCT03929029
			Pembrolizumab (anti-PD-1) or nivolumab (anti-PD-1) and CDx-301 (FLT3L)	Melanoma/metastatic melanoma	30 (20)	NCT04930783
	Cellular	Individualized private mutations	Monotherapy	Breast cancer/triple-negative breast cancer	16 (16)	NCT06435351
			Chemotherapy	Non-small cell lung cancer	16 (16)	NCT05195619
			Nivolumab (anti-PD-1) and SOC chemo	Pancreatic adenocarcinoma	14 (12)	NCT04627246
			Monotherapy	Non-small cell lung cancer	6 (6)	NCT04078269

Table depicts a selection of currently ongoing neoantigen-based vaccine clinical trials, classified into shared and personalized neoantigens. Columns highlight the types of vaccine formulations, specific neoantigens being targeted (in case of shared neoantigens), therapeutic combinations being employed along with disease settings, and the current/estimated enrollment numbers. ∗Asterisk denotes phase 2/3 clinical trial; the rest are early phase 1/2 trials.

MPN, myeloproliferative neoplasm; FLC, fibrolamellar hepatocellular carcinoma; TTF, tumor treating field; SERD, selective estrogen receptor degrader.

Challenges in personalized neoantigen-based vaccination trials

Despite their promise, several challenges limit the widespread clinical application of neoantigen vaccines. First, neoantigens are often expressed at lower levels compared to TAAs, which may reduce the likelihood of effective T cell priming and tumor recognition. Additionally, tumors with low tumor mutational burden may harbor few neoantigens, making it more difficult to identify suitable targets. Although, recent studies have demonstrated the feasibility of generating personalized neoantigen vaccines even for tumors with modest neoantigen expression, expanding the potential applicability of this approach.^44,45,69

More significantly, the production of personalized neoantigen vaccines remains resource intensive, costly, and time consuming, typically requiring a minimum of 6–12 weeks from tumor sequencing to vaccine administration. This lag time is particularly problematic for patients with rapidly progressing disease, who may not benefit from such delayed interventions.

To address this issue, combination strategies incorporating both TAAs and personalized neoantigens have been explored. The feasibility of this strategy has been demonstrated in a trial treating patients with melanoma. This study incorporated NYESO-1-targeting TAA vaccine for some patients with NYESO-1 or tyrosinase-positive tumors, while awaiting the release of their personalized neoantigen vaccines.⁶¹ In another example, the GAPVAC trial included 15 patients with newly diagnosed GBM who received off-the-shelf GBM-specific TAA vaccines (APVAC1) while their personalized neoantigen vaccines (APVAC2) were being developed. Immunogenicity assessments revealed T cell responses to 50% of APVAC1 TAAs and up to 84% of APVAC2 neoantigens. Importantly, the trial reported a median overall survival of 29 months, with one patient surviving beyond 38 months, highlighting the feasibility and potential clinical benefit of this sequential vaccine strategy.⁶²

Shared neoantigen vaccines: Bridging personalization and accessibility

Despite the advances in their personalization, there remains a compelling need for vaccine platforms that combine the high immunogenicity and tumor specificity of neoantigen vaccines with wider accessibility and production efficiency of off-the-shelf TAA vaccines. Public neoantigens, which are shared across patients with cancers, provide a unique opportunity to bridge this gap.

Public neoantigens arise from recurrent genetic alterations shared across multiple patients. These neoantigens typically originate from common driver mutations in oncogenes and tumor suppressor genes such as KRAS, IDH1, BRAF, PIK3CA, and TP53. In addition, shared neoantigens can result from hotspot mutations, such as frameshift mutations in repetitive microsatellite regions, particularly in tumors with deficiencies in the mismatch repair pathway.^79,80,81

A first-in-human, proof-of-concept trial conducted in 2021 for patients with PDAC and colorectal cancer (CRC) harboring KRAS mutations leveraged a lymph node targeting novel delivery system called amphiphiles (Amph), which are albumin-binding lipids.^82,83 In this study, 25 patients (20 with PDAC and 5 with CRC) received Amph-conjugated shared neoantigen peptides (mutant KRAS G12D or G12R) combined with escalating doses of Amph-conjugated CpG7909, a TLR9 agonist used as an adjuvant. The vaccine was shown to induce robust and sustained ex vivo CD8⁺ and CD4⁺ T cell responses in 71% of the cohort, with induction of notable cross-reactivity against non-immunizing mKRAS epitopes such as G12A, G12C, G12V, and G12S in more than 67% of the patients. This study also demonstrated a correlation between T cell immune responses (strong responders displaying more than 9.17-fold increase over baseline) and clinical benefit with improved RFS (not reached in strong responders vs. 3.02 months weak responders), and median OS (not reached in strong responders vs. 15.98 months in weak responders). Furthermore, authors report biomarker (ctDNA) clearance or reduction (100% in strong responders vs. 62% in weak responders), indicating the potential of a shared neoantigen like mKRAS to elicit a strong anti-tumor response and epitope spreading.^84,85,86

In an ongoing phase 1/2 clinical trial, self-amplifying RNA, which enables enhanced antigen expression even when administered at lower doses, is being evaluated.^87,88 The vaccine incorporates a chimpanzee adenoviral vector and self-amplifying mRNA encoding 20 shared neoantigens (GRT-C903 and GRT-R904) derived from common oncogenic driver mutations like mTP53, mKRAS, β-catenin, and BRAF, resulting in an off-the-shelf, multi-tumor vaccine.⁸⁹ Despite a majority of patients (15/19) experiencing disease progression within 2 months of enrollment, phase 1 immunogenicity analysis from this trial highlighted the critical influence of epitope immunodominance, a property of certain epitopes to dominate the immune response, while suppressing reactivity against sub-dominant antigens.^{90,91,92,93,94} In this study, the immunodominant mTP53 epitope appeared to have hampered the immune response to the sub-dominant epitopes like mKRAS. To overcome this, a follow-up research study by the same group demonstrated that increasing the antigen payload by incorporating repeated mKRAS epitope-encoding cassettes into the vaccine restored immunity against sub-dominant mKRAS.⁸⁹ This strategy is currently being evaluated in a phase 2 clinical trial.⁸⁹

These findings underscore a key challenge in multi-epitope neoantigen vaccine design. Ultimately, if a few immunodominant epitopes hijack the immune response, inclusion of additional subdominant targets may offer limited benefit. Such considerations are particularly important with the inclusion of increasingly higher number of antigens in vaccines. For example, the NOUS-209 viral vector vaccine delivers 209 shared frameshift neoantigens to cancer patients with microsatellite instability. Initial studies have reported induction of T cell immunity in all patients analyzed (n = 37) and confirmed polyfunctional T cell responses against 115 different neoantigens; it remains to be seen if the majority of immunity was driven against a small number of neoantigens.^95,96

Overall, in addition to predicted binding affinity to HLA alleles, epitope selection should also be guided by consideration of immunodominance and competition to ensure a broad and effective vaccine response.

Key considerations for improvement

Despite the recent success in cancer vaccines, several challenges remain, impacting clinical outcome and patient survival. Addressing these shortcomings is critical for improving the design, delivery, and integration of neoantigen vaccines into standard cancer treatment paradigms. In this section, we highlight the major areas where innovation and optimization are needed to realize the full potential of personalized cancer vaccination, as depicted in Figure 2.

Figure 2.

Proposed future directions for optimization of therapeutic vaccine pipeline

Figure summarizing proposed changes for developing a standardized pipeline for cancer vaccines in the future including antigen selection and validation (steps 1, 2, and 3), patient selection criteria (step 4), expansion of combination strategies deployed (step 5), and standardization of immune-monitoring parameters (step 6) (created with BioRender.com).

Evolving technology for antigen prediction

The past decade has witnessed many advances in sequencing and computational analysis tools such as VarScan2, Strelka2,⁹⁷ Mutect2,⁹⁸ and Manta⁹⁹ and pipelines like OpenVax,¹⁰⁰ NeoDisc,¹⁰¹ pVACtools,¹⁰² and others (reviewed in depth elsewhere^103,104) that have enabled the development of prediction-based neoantigen vaccines. Despite these technological advances, immune monitoring data from almost all the recent clinical trials have demonstrated that only a fraction of these predicted epitopes for each individual patient generate a response, highlighting the need for improvement in the prediction algorithms.^2,44,60,64 There may be several reasons for this seeming discrepancy: (1) some pipelines rely solely on DNA sequencing, potentially including unexpressed tumor antigens as targets, while incorporating both DNA and RNA sequencing data enables expression filtering, improving the selection of antigens highly abundant in the tumors, which leads to effective T cell priming, and (2) most of the selection criteria are guided by epitope-MHC-I binding affinity predictions using algorithms that are trained on and limited by the availability of validated peptide-MHC binding affinity datasets.^105,106 This is also compounded by lack of data availability for less-frequently represented HLAs in the non-western world, making the algorithms better at predicting binding to common HLAs like HLA-A^∗02:01, HLA-A^∗24:02, HLA-C^∗04:01, etc. However, several strides have been made recently to improve predictions by enhancing coverage for tumor antigens using highly sensitive mass spectrometry methods like field asymmetric waveform ion mobility spectrometry for class I epitopes and mono-allelic purification with tagged allele constructs for class II epitopes.^{74,107,108,109,110} Additionally, standardization and improvement in prediction pipelines to incorporate factors like RNA splicing, proteasomal degradation for antigen processing, and potential sequence homology with pathogen-derived epitopes will also be useful.

In terms of the parameters for prioritization of antigens, there are several factors that can differ between pipelines, for example, bias toward class I binding epitopes over class II epitopes for enhanced CD8⁺ T cell cytotoxicity, balance between driver vs. passenger mutations, etc.^{45,61,62,68,69,89} These can be better guided through a comprehensive meta-analysis of the existing data from various prediction pipelines, correlating them with immunogenicity data for individual targets to inform the algorithm of the parameters that led to immunogenic targets. Analysis of immunogenic neoepitopes across multiple cohorts by the global consortium named Tumor Neoantigen Selection Alliance (TESLA) to improve prediction algorithms represents one such effort in that direction.¹¹¹ More importantly, high-throughput wet lab systems for validation of predicted targets like the MANAFEST and other TCR functional assessment assays^112,113 and affordable access to platforms like RNAscope or Nanostring DSP for spatial confirmation of transcript expression directly within tumor cells are urgently needed to guide further development and optimization of the prediction algorithms.

Antigen selection: Expanding the potential target repertoire

Over the last decade, identifying key features of tumor antigens that elicit a robust and tumor-specific immune response has remained an area of intense investigation. The vast majority of vaccine clinical trials, until recently, have targeted canonical proteins: either neoantigens resulting from non-synonymous somatic mutations or unmutated antigens overexpressed by tumor cells. Recently, there is a growing body of evidence demonstrating the potential of other genomic alterations like RNA splice variants or products of reactivation of epigenetically silenced endogenous retroviral genes, which are currently not included in most neoantigen prediction pipelines, to generate targetable shared/public neoantigens.^{114,115,116,117}

Another avenue for targeting tumors, especially those that are deficient in antigen processing and presentation via TAP1 and TAP2, could be through the priming of T cells against a class of novel, ubiquitous and unmutated antigens known as T cell epitopes associated with impaired peptide processing (TEIPP). These antigens have been identified to be immunogenic in several preclinical studies.^118,119,120 A first-in-human vaccine clinical trial in patients with checkpoint-resistant non-small cell lung cancer (n = 24) targeting LRPAP1 peptide vaccine (TEIPP24) has shown promising results with CD8⁺ T cell response in 83% of the patients.¹²¹ Additionally, the latest discovery of the potential for non-canonical proteins to be immunogenic has opened doors to new tumor-specific shared targets for vaccine therapy. These non-canonical proteins, also known as cryptic antigens or dark matter antigens, arise from the ability of tumor cells to translate regions of the genome outside of the open reading frame like 5′ and 3′ UTRs, long non-coding RNAs, and alternate reading frames like internal open reading frames (intORFs). Three recent studies assessed the expression and immunogenicity of these cryptic antigens in PDAC,¹²² melanoma and non-small cell lung cancer,¹²³ gynecological cancers, and head and neck carcinoma.¹²⁴ Their data have shed light on the potential for some of these short-lived, rapidly degraded proteins to be presented on MHC-I and elicit a T cell response in vitro. Notably, the frequencies of these T cells were undetectable ex vivo suggesting that these cryptic antigens, some of which were shared across multiple tumor types, are not part of the endogenous tumor response, making them attractive targets for vaccine-induced priming of T cells.^124,125 This lack of endogenous response is hypothesized to be due to the short-lived nature of these peptides, making cross-presentation and uptake by DCs unlikely owing to preferential presentation of stable and abundant proteins.^126,127 While in its infancy, the potential for cryptic/dark matter antigens warrants further study and exploration. Given the early stage of discovery, it is important to recognize several challenges inherent to this antigen category. First, given their novelty the tools to reliably and uniformly identity such antigens are not well developed. Second, their abundance and expression across tumor versus normal tissues remains largely uncharacterized. Third, the immunogenicity of such antigens is yet to be broadly validated, making it hard to determine their usefulness as cancer vaccine targets. Fourth, it remains to be determined whether such antigens are truly shared among a broad patient population. Finally, there is limited evidence regarding the safety of targeting these antigens, underscoring the need for careful preclinical evaluation before their incorporation into clinical vaccine platforms.

Timing the attack: Adjuvant vs. neoadjuvant vaccination

The timing of cancer vaccine administration requires further exploration, given the continual impact of immuno-editing in the TME and the potential for sub-clinical tumor progression. Currently, the majority of cancer vaccine trials are conducted in the adjuvant setting (post-surgical resection) with patients exhibiting no or low disease burden in the form of minimal residual disease and where tumor-driven immune suppression is minimal.^{44,45,60,69,77} Immunogenicity and clinical correlation data from these vaccine trials, both as monotherapy and in combination with ICIs, have demonstrated the ability of vaccination in adjuvant setting to elicit strong immune responses and reduce the risk of recurrence.^61,69,82 However, this approach precludes patients with unresectable or metastatic disease from receiving treatment.

On the other hand, the neoadjuvant (pre-surgical) setting, despite being influenced by immunosuppressive signals from the tumor, offers a unique advantage: it allows vaccine-induced T cell responses to develop in the presence of the intact tumor. This not only enables direct tumor cell killing but also promotes the release of additional tumor antigens, facilitating a broader T cell response through epitope spreading. Pre-clinical studies comparing neoadjuvant vs. adjuvant ICI treatment in murine models of metastatic breast cancer (E0771 and 4T1) demonstrated that neoadjuvant immunotherapy induced significantly greater magnitude of neoantigen-specific CD8⁺ T cells and was more effective in eradicating distant metastasis.¹²⁸

A study in patients with resectable PDAC evaluated the benefit of sequential vaccination with GVAX, a GM-CSF-secreting whole tumor cell vaccine, alone or in combination with anti-PD1 and anti-CD137 antibodies, in neoadjuvant (pre-surgery) followed by continuing administration with SoC in the adjuvant (post definitive surgical resection) setting. Evaluations of the resected tumor revealed that neoadjuvant vaccination had induced tumor infiltration of effector and cytotoxic T cells as well as prompted formation of TLSs, which are local sites for T cell priming and activation.¹²⁹ This study provided evidence for the potential of neoadjuvant vaccines to convert a relatively “cold” tumor into “hot or inflamed” tumor. However, the relative impact of neoadjuvant vaccination on overall disease-free survival (DFS) and OS could not be ascertained. In another study, vaccination of patients with unresectable HCC with a DNA vaccine (GNOS-PV02) with concurrent administration of anti-PD1 and IL-12 encoding up to 40 personalized neoantigens was shown to be immunogenic,⁷⁸ further supporting the use of neoadjuvant vaccinations.

Despite these encouraging results, neoadjuvant vaccination faces logistical challenges. These include difficulties in manufacturing the vaccine from biopsy materials with a relatively short turnaround time, especially where surgical resection of tumor is not SoC. This constraint makes off-the-shelf formulations, such as those targeting shared neoantigens or TAAs, a more viable option in the neoadjuvant setting.

Priming and re-invigorating T cells: Combinations with ICI

Data from multiple clinical trials have pointed to the limited efficacy of cancer vaccines as a monotherapy, making it crucial to consider the most appropriate combination therapy and the sequence in which the combination should be administered. ICIs like anti-PD-1, anti-PD-L1, and anti-CTLA-4 antibodies, now SoC in many cancers, are commonly combined with vaccines, given concurrently or sequentially so as to augment vaccine-induced T cell responses and mitigate exhaustion, especially in tumor that had previously developed resistance to ICIs.^44,60,61,78 This was especially notable in a few reported cases with advanced melanoma and head and neck disease relapses where patients previously unresponsive to ICI showed disease clearance and prolonged RFS/DFS upon receiving anti-PD-1 post-vaccination.^44,61,130 Additionally, analysis of T cell clonal overlap in patients receiving ICI treatment followed by neoantigen vaccine showed distinct clones being expanded after each treatment, suggesting the potential for synergistically targeting different responder subsets of T cells through these two modalities.⁶⁹ However, it is important to consider the order in which these therapies are administered. Although data from human clinical trials are limited, some studies have shown that it is possible that treatment with ICI prior to vaccination may lead to poor CD8⁺ T cell response through induction of dysfunctional, sub-optimally primed PD-1⁺CD38^hi CD8⁺ T cells, loss of immunogenic neoantigens by post-ICI immunoediting, or reduced expression of targetable epitopes through deletion of clonal populations by re-invigorated T cells pre-vaccination, which in turn can hamper vaccine efficacy.^{131,132,133,134}

Targeting the tumor and hostile microenvironment: Novel combinations

Besides ICIs, other modalities, often already in use as SoC such as chemotherapy and radiation therapy, can complement vaccination and should be carefully integrated with the vaccine regimen to maximize synergy. For example, chemotherapy and radiation therapy can induce tumor cell death promoting antigen release, DC priming, enhancing antigen presentation on tumor cells, and reducing frequencies of suppressor cells in the TME.^{28,135,136,137,138} Other combinations that can synergize with vaccines include FDA-approved tumor-targeting therapies like SLAMF7 agonist (elotuzumab), which activated NK cell-mediated tumor cell cytotoxicity¹³⁹; PARP inhibitors like olaparib and niraparib, which lead to synthetic lethality and death of tumor cells¹⁴⁰; and anti-estrogen therapy (tamoxifen) starving estrogen-driven proliferation of estrogen receptor-positive tumors,¹⁴¹ thereby reducing the tumor burden. Another avenue of synergy is through the use of an anti-angiogenic antibody like bevacizumab that can remodel tumor vasculature, enabling better immune cell infiltration into the tumor,¹⁴² as well as some investigational therapies that reduce immune suppression in the TME like transforming growth factor β and PD-L1 bi-specific antibody (YM101 or fusion protein M7824)^143,144 and all- trans-retinoic acid for depletion of MDSCs.^145,146

Harmonizing methods for immune analysis

A lack of standardized assays and methodologies to define correlates of an effective immune response induced by cancer vaccines is a major hurdle in cross-referencing results and data across different studies. Currently, the most common methods used for immune monitoring involve culturing patient’s peripheral blood T cells, in vitro, in the presence of APCs, immunizing peptides, and/or specific cytokines and adjuvants, followed by measurement of effector cytokine production by ELISpot assay or intracellular staining by flow cytometry to detect antigen-specific activated T cells pre- and post-vaccination.^44,60,61,69

More recently, high-throughput techniques such as single-cell RNA and tandem TCR sequencing, barcoded tetramer or dextramer staining for neoantigen-specific T cells, multiplexed spatial microscopy, sophisticated computational analyses (CloneTrack and MANAFEST), and functional screening of TCR libraries for a robust validation of vaccine-induced neoantigen specificity of T cells have been employed.^{69,76,112,113} However, with different groups utilizing varied benchmarks for defining immunogenicity, it is not possible to draw consistent conclusions across various studies. Hence, there is a need for a harmonized assay system with validated benchmarks to evaluate correlates of vaccine-induced T cell immunity across multiple studies.

Looking beyond T cells: Impact of vaccines on the immune system

Currently, the majority of correlative studies evaluating vaccine-induced immune changes predominantly focus on T cell responses with less emphasis on the potential role of antibodies, NK cells, and myeloid populations, including assessment of immunosuppressive subsets. Only a handful of clinical trial studies have reported comprehensive immune profiling of both myeloid and lymphoid cells through the course of vaccination.^44,69 Even fewer studies have evaluated neoantigen-specific antibodies or systemic cytokine and chemokine modulations. It is possible that such narrow focus on T cell might be overlooking other key contributors, especially since immunity is an integrated outcome of complex interaction between numerous immune cells.

Signal searching: Need for effective biomarkers of response

There is a pressing need to develop rapid and minimally invasive tests to longitudinally monitor immune biomarkers and assess evolving vaccine immune responses. Some examples include testing for immune biomarkers in urine or assessing secreted immune modulators in blood by high-throughput multiplexed assay. While surrogate methods such as radiographic assessment, ctDNA, circulating carcinoembryonic antigens, cancer antigen 19-9, and PSA have been utilized successfully as biomarkers for vaccine benefit in certain contexts,^{82,147,148,149} they offer limited insight into underlying immune mechanisms. The development and validation of additional sensitive, non-invasive biomarkers, capable of capturing immune activation, modulation, and durability over time, will be critical for optimizing vaccine design, identifying early responders, and guiding clinical decision-making.

Conclusion and perspective

Building on decades of research, cancer vaccines are re-emerging as a powerful tool in oncology. Recent advances in sequencing technologies, antigen prediction algorithms, and delivery platforms have modernized vaccine design and resulted in a surge in personalized vaccine trials across diverse tumor types, including traditionally ICI-resistant cancers like pancreatic ductal adenocarcinoma (PDAC). Unlike ICIs, which are more effective in immune-infiltrated tumors, cancer vaccines have demonstrated the capacity to prime immune responses even in tumors with low T cell infiltration or tumor mutational burden. A detailed profiling of the TME using high-resolution tools such as single-cell and spatial transcriptomics is essential to identify resistance mechanisms ranging from extracellular matrix barriers to metabolic dysfunction and immune exhaustion and guide rational combination therapies that can improve vaccine efficacy and patient outcomes.

Future trials should prioritize early-stage disease and leverage hybrid strategies such as combining shared TAAs with neoantigens, in diverse treatment settings. Finally, the majority of cancer vaccine trials to date have been early-phase, single-arm phase 1 or 2 studies. While these early studies were essential for demonstrating safety, feasibility, and initial immunogenicity, they fall short of providing robust evidence for clinical benefit. Key questions remain unresolved, including the optimal number of neoantigens required to elicit effective immunity and the dose that maximizes therapeutic response. These challenges are particularly complex in the context of personalized vaccines, where each patient presents a unique set of neoantigens and immune responses. Nevertheless, to advance the field, it is imperative to establish clear immunological and dosing benchmarks and to conduct larger, well-controlled randomized clinical trials that can rigorously assess efficacy and inform future SoC integration.

Declaration of interests

The authors declare no competing interests.