Therapeutic anti-cancer vaccines: a systematic review of prospective intervention trials for common hematological malignancies

Summary

Background

This review comprehensively assesses all prospective trials on anti-cancer vaccines for common hematological malignancies by analyzing trial designs, endpoints, and whether these endpoints are met across these trials.

Methods

We included onco prospective clinical trials involving therapeutic anti-cancer vaccines for hematological malignancies published up to May 2025. We excluded retrospective cohort studies, case reports, non-research opinion publications, and studies not related to hematological malignancies. Information sources: Embase, MEDLINE, Web of Science Core Collection, and ClinicalTrials.gov. All the included RCTs were assessed for bias using the Cochrane Handbook for Systematic Review of Interventions, version 6.2 and Cochrane risk-of-bias tool. Results were synthesized descriptively, using frequencies (%) and medians (interquartile range [IQR]). The study protocol utilized was recorded in the PROSPERO database (Registration ID. CRD42024504780).

Findings

Out of 2856 studies screened, a total of 187 studies were included. The median sample size was 18 (IQR = 20), and 33/187 (18%) studies were randomized. Most utilized primary endpoints were translational and safety, of which the endpoint was met 65/81 (80%) and 51/74 (69%) of the time, respectively. In 35/187 (19%) of the total studies included, the primary endpoint was a clinical efficacy endpoint (PFS, OS, duration of remission, cancer response) of which, 11/35 (31%) of studies met their clinical primary endpoint. Of the 33 randomized studies, 24 measured a clinical endpoint as their primary endpoint. Besides BCG administration in AML, no vaccine trial met a clinical efficacy endpoint in a randomized trial. There was not a single instance in which a vaccine product demonstrated an improvement in overall survival. The risk of bias assessment for RCTs showed most studies at low or intermediate risk of bias.

Interpretation

This systematic review of all therapeutic anti-cancer vaccine trials in common hematological malignancies shows that although vaccines generally demonstrate immunogenicity, they have mostly failed to show consistent anti-cancer activity. Limitations include a lack of quantitative synthesis due to heterogeneity of assessed interventions, small sample sizes in most studies, and a lack of clear endpoint description in some studies.

Funding

None. PROSPERO Registration ID. CRD42024504780.

Research in context.

Evidence before this study

Therapeutic cancer vaccines represent a promising approach in immunotherapy. While research on these vaccines has shown potential in various malignancies, their efficacy in hematological cancers remains unclear. Previous reviews have typically focused on specific cancer types or vaccine technologies, leaving a gap in our understanding of their overall effectiveness across different hematological malignancies. This systematic review aimed to provide a comprehensive assessment of prospective trials on anti-cancer vaccines for major hematological malignancies, evaluating their efficacy in achieving primary endpoints across various study designs. We searched Embase, MEDLINE, ClinicalTrials.gov, and the Web of Science Core Collection databases with no language restriction for articles published from the inception of the database up to May 2025 and found no previous study that had performed such an assessment on vaccines as a therapeutic modality for hematological malignancies.

Added value of this study

Our analysis suggests most vaccine trials have met translational and safety endpoints. Amongst the minority of studies that do assess clinical endpoints such as cancer response, progression free survival (PFS), overall survival (OS), or duration of remission, the majority of them did not meet their endpoint. Besides the Bacillus Calmette-Guérin (BCG) vaccine administration for acute myeloid leukemia (AML), there was no instance of a vaccine product meeting a clinical efficacy endpoint in a randomized trial. Furthermore, there was not a single reported instance in which a vaccine product demonstrated an improvement in OS.

Implications of all the available evidence

Our findings indicate that anti-cancer vaccine trials thus far have mostly focused on translational endpoints and generally have failed to show an improvement in clinical endpoints. Future research should prioritize larger vaccine studies measuring clinical endpoints and new vaccine platforms.

Introduction

Advances in current therapies for several common hematological malignancies, including multiple myeloma (MM),¹ acute myeloid leukemia (AML),² chronic lymphocytic leukemia (CLL),³ and diffuse large B-cell lymphoma (DLBCL),⁴ have significantly improved patient outcomes. These treatments have extended survival times, and in some cases, even offered curative potential. Despite these advancements, relapses are frequently observed, highlighting the need for novel therapeutic approaches with anti-cancer vaccines serving as one of the potential avenues to address this problem. Prophylactic cancer vaccines such as the human papillomavirus vaccine (HPV) have shown effectiveness in preventing certain types of cancers. However, the potential of therapeutic vaccines as a treatment modality against cancer has been limited so far, with only two vaccines approved to date across all cancers: Bacillus Calmette-Guérin (BCG) and sipuleucel-T.⁵

Although there are no anti-cancer vaccines currently approved for hematological malignancies, numerous trials have shown varying degrees of benefit.6, 7, 8 Nevertheless, the true clinical impact of these cancer vaccines on the trajectory of cancer for patients with hematological malignancies is still unclear, as no previous systematic review has thoroughly reviewed the landscape of anti-cancer vaccine trials to ascertain endpoints utilized and whether these endpoints are met.^9,10

We conducted this systematic review to evaluate all prospective, therapeutic anti-cancer vaccine intervention trials conducted for the most common hematologic malignancies, including MM, AML, CLL, and DLBCL. Specifically, this review aimed to assess the primary endpoints utilized, the success rate of these trials in meeting their predefined endpoints, and the variation in outcomes across different trial phases.

Methods

Protocol and registration

A systematic review analysis was conducted utilizing publicly available data according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Extension for Scoping Reviews guidelines.¹¹ The study protocol utilized was recorded in the Prospective Registry of Systematic Reviews, PROSPERO database (Registration ID. CRD42024504780) on February 22, 2024.

Eligibility criteria

Clinical trials reported or published prior to May 2025 as manuscripts in a peer-reviewed journal or as abstracts in conference proceedings were included. All prospective clinical trial studies were included with no restrictions on the type of therapeutic anti-cancer vaccines, participants age, sex, race, or country of origin, and the language in which the original study was published in.

Studies reporting only in vitro and/or in vivo results, only collecting patient samples or cells, or involved only in vaccine generation rather than administration were excluded. Studies with vaccines tested against non-hematological malignancies and studies evaluating non-cancer preventive or therapeutic vaccines were also excluded. Additionally, case reports, case series, review articles, case–control, retrospective cohort, and single-arm studies were excluded. In addition, non-research opinion publications such as editorials, analyses, and commentaries, were excluded as well.

Information sources and search strategy

The search strategy was restricted to clinical trial studies assessing prospective, therapeutic, anti-cancer vaccine trials on patients with an established history of hematological cancer such as MM, AML, CLL, and DLBCL. A search strategy was developed using keywords: (“acute myelocytic leukemia” OR “multiple myeloma” OR “diffuse B-cell lymphoma” OR “chronic lymphocytic leukemia” OR AML OR CLL OR DLBCL) AND vaccine] to be searched across varying databases, including Embase, MEDLINE, and Web of Science (WoS) Core Collection (refer to Supplementary Appendix). The search date was May 6, 2025. To further enrich the search results, we searched ClinicalTrials.gov, using the same search terms, to find studies that may not have been found on other databases of published trials.

Study selection and screening

To screen the peer-reviewed studies obtained from the strategy search, four reviewers (DS, VS, KS, and MA) worked independently to identify abstracts and manuscripts meeting the inclusion criteria. This approach ensured that each study underwent evaluation by at least two reviewers, facilitating the resolution of any discrepancies upon consultation with a fifth reviewer (GRM) when necessary.

In addition, the ClinicalTrials.gov registry was searched. Two reviewers (DS and VS) independently screened the prospective clinical trials to identify the ones meeting the inclusion criteria and the senior reviewer (GRM) was consulted to resolve any discrepancies. Clinical trials with designations, “Actively recruiting,” “Active, not recruiting,” “Completed,” as well as “Unknown status” were included in our criteria. Trials that were “Terminated” or “Terminated with Results” were excluded from this search.

Data extraction and study outcomes

The independent data extraction was performed by five independent reviewers (DS, VS, KS, MA, and PS). At least two independent reviewers were assigned to each study for accurate data extraction. A senior reviewer (GRM) was consulted to resolve any discrepancies and/or uncertainties. The following data from each study was collected: title, publication year, ClinicalTrials.gov identifier (if applicable), type of publication (abstract or full manuscript), sponsorship (pharmaceutical vs. non-pharmaceutical), trial location (in the US or outside the US), specific cancer type studied (MM, AML, CLL, or DLBCL), trial design, whether patients were required to be in remission or response at the time of investigational agent administration, trial phase (phases 1–3), number of participants, the median age of participants, type of investigational agent, duration of treatment, and whether the vaccine was administered in an adjuvant/combination setting (if applicable). A vaccine was defined as a biological preparation that was administered to stimulate the body's response against cancer, and by utilizing that definition oncolytic viruses and modified T cell therapy were excluded from our study. The endpoints included were overall survival (OS), progression free survival (PFS), duration of remission, cancer response, translational/immunogenicity, and safety. OS, PFS, cancer response, and duration of remission were defined as clinical efficacy endpoints in this analysis.

Continuation studies from prior phases were identified within the 187 included trials by matching their NCT numbers. In some cases, different phases of the same study shared the same NCT number and were further verified using the materials and methods, as well as the authors and institutions involved. Missing statistics in studies were labeled as such and not included in statistical analysis.

Study bias assessment

All the included RCTs were assessed for bias using the Cochrane Handbook for Systematic Review of Interventions, version 6.2 and Cochrane risk-of-bias tool.^12,13 Two reviewers (GRM, KS) worked independently to assess risk of bias for each RCT. This approach ensured that each study underwent assessment by at least two reviewers, facilitating the resolution of any discrepancies upon consultation with a third reviewer (DS) when necessary.

Statistical analysis

The results were synthesized descriptively and were presented as frequencies (%) and medians (interquartile range [IQR]). Data was analyzed using Microsoft Excel. As the data utilized was publicly available and did not use individual patient data, these analyses did not require institutional review board approval. Given the substantial heterogeneity across the included studies, we determined that a meta-analysis was not methodologically appropriate for this systematic review. The studies differed widely in design, vaccine platforms, cancer types, disease states (newly diagnosed vs. relapsed/refractory), trial phases, use of adjuvant therapies, patient populations, sample sizes, outcome measures, and primary endpoints. These clinical and methodological variations introduced considerable complexity, limiting the feasibility and interpretability of a pooled quantitative analysis. Instead, we employed a narrative synthesis approach to capture the nuanced findings of each of each study, utilizing descriptive statistics to highlight key trends and insights within the evolving landscape of anti-cancer vaccine research.

Role of the funding source

There was no public, private, or non-profit agency funding source for this study.

Results

Studies that met inclusion criteria

The initial search strategy identified 2856 clinical trial studies, of which 187 were included in the final sample of studies. Table 1 lists characteristics of included studies, while Supplementary Tables S2 and S3 have individual details of all included studies. Fig. 1 highlights the screening process.

Table 1.

Characteristics of included studies.

Median sample size	18 (IQR = 20)a (Missing 3 studies)
Median age	61 (IQR = 7)a (Missing 104 studies)
Cancers studieda
Multiple myeloma	75 (40%)
Acute myelogenous leukemia	86 (46%)
Chronic lymphocytic leukemia	28 (15%)
Diffuse large B cell lymphoma	8 (4%)
Sponsorship
Pharma	24 (13%)
Non-pharma	151 (81%)
Both	12 (6%)
Trial phase
I	66 (35%)
I/II	41 (22%)
II	37 (20%)
III	4 (2%)
II/III	1 (<1%)
Missing	38 (20%)
Randomization
Yes	35 (19%)
No	152 (81%)
Studies grouped by types of vaccine
Peptide-based	74 (36%)
Autologous dendritic cell	52 (28%)
Autologous tumor cell	22 (12%)
Allogeneic dendritic cell	12 (6%)
Allogeneic tumor cell	4 (2%)
BCG	17 (9%)
Other	6 (3%)
Studies grouped by primary endpoint
Clinical: progression-free survival	6 (3%)
Clinical: cancer response	10 (5%)
Clinical: overall survival	2 (1%)
Clinical: duration of remission	17 (9%)
Translational	81 (43%)
Safety	74 (40%)
Was the vaccine given in conjunction with another cancer therapy (e.g., chemotherapy)?
Yes	37 (20%)
No	141 (75%)
Given in conjunction and as a monotherapy in separate groups	8 (4%)
Vaccine given for maintenance after induction therapy?
Yes	116 (62%)
No	56 (30%)
Given as maintenance AND/OR prevention AND/OR treatment in separate groups	14 (8%)
Did the study meet the primary endpoint(s)?
Yes	124 (66%)
No	22 (12%)
Missing	41 (22%)

^aA total of 187 studies were included. While most of the studies included only one disease type, 9 studies included more than one disease, such as investigating the role of therapeutic vaccines in AML and MM for example. 4 studies included both safety and translational/immunogenicity as the primary endpoints. The endpoint and trial design for 1 study could not be ascertained.

Fig. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-based flow diagram of the study selection process.

The median sample size of the studies was 18 (IQR = 20). Of the 187 studies, 33 (18%) were randomized. The most studied cancer was AML (86/187), followed by MM (75/187), accounting for 46% and 40% of the studies, respectively. There were 9 studies that investigated more than one of the four indications. Out of the 187 studies, 37 (20%) of them were performed in a setting where the vaccine was given in conjunction with another cancer therapy (e.g., chemotherapy, immunotherapy). Of the studies that administered the vaccine standalone (n = 141), 90/141 (64%) were intended for maintenance purposes after completion of cancer directed therapy (Table 1). Eight studies out of the 187 (4%) administered the vaccine as a monotherapy and in conjunction with other therapeutics in separate groups. Fourteen studies (7%) administered the vaccine in maintenance, prevention, and treatment settings in different patient groups. Trial design could not be ascertained for one study. Fig. 2 highlights the distribution of manuscripts by year of publication. Of all 187 studies, only 39 (21%) used a vaccine standalone for the intent of therapeutic improvement instead of maintenance (Supplementary Table S3).

Fig. 2.

Distribution of Manuscript Publications by Year (n = 149). ∗38 studies were only published on clinicaltrials.gov and did not have an associated manuscript.

Types of vaccination

Peptide-based vaccines (n = 74, 40%) and autologous dendritic cell-based (n = 52, 28%) were the most common therapeutic vaccine products. Autologous tumor cell vaccines were used in 21 (11%) of studies. Allogeneic dendritic cell-based and allogeneic tumor cell vaccines were less commonly used in only 12 (6%) and 4 (2%) of the studies, respectively. BCG vaccines accounted for 17 (9%) of studies. Other vaccine types, such as genetic-based and viral-based, as well as different vaccine combinations accounted for 7 (4%) of studies.

Endpoints

35 (19%) of the included studies utilized a clinical primary endpoint (PFS, OS, duration of remission, and cancer response) and the rest 151 (81%) investigated primary outcomes such as safety and immunogenicity/translational endpoints. For one study, the primary endpoint could not be ascertained.¹⁴

Among the studies with clinical endpoints, duration of remission was the primary endpoint in 17 (9%) of trials with OS designated as the primary endpoint in two studies. Cancer response was the primary endpoint in 10 (5%), and PFS was the primary endpoint in 6 (3%) of trials.

Endpoints with clinical outcomes

Amongst the studies with clinical cancer outcomes (n = 35), 17 studies (49%) analyzed the duration of remission, 10 (29%) analyzed cancer response, 6 (17%) studies analyzed PFS, and 2 (6%) analyzed OS. The median sample size for these studies was 64, 26, 112, and 186 respectively. Treatment duration varied from 0.5 to 36 months with a median duration of 3.5 months (IQR = 6).

Likelihood of meeting endpoints

Amongst the included studies, 146 (78%) reported clearly whether the primary endpoint was met. Among those studies, 124 (85%) met their endpoint.

Of the 33 randomized trials, 26 reported clearly whether the endpoint was met, of which 14 (42%) trials met their primary endpoint. Amongst these randomized trials, 24 measured clinical endpoints of which 6 (25%) were met and 6 (25%) did not clearly state whether or not the endpoint was met or had no results available.

Overall, primary endpoints were met in 55/75 (73%) of MM studies, followed by 51/86 (59%), 20/28 (71%), and 3/8 (38%) in AML, CLL, and DLBCL studies respectively.

In MM studies, 57/69 (83%) of translational/immunogenicity studies met the primary endpoint, while 2/6 (33%) of studies measuring a clinical endpoint were successful. In AML, 43/59 (73%) of translational/immunogenicity studies met the primary endpoint, compared to 8/26 (31%) of studies with a clinical endpoint. In CLL, 20/27 (74%) of translational/immunogenicity studies met the primary endpoint, while 0/1 (0%) of studies measuring a clinical endpoint met their endpoint. In DLBCL, 2/6 (33%) of translational/immunogenicity studies and 1/2 (50%) of studies measuring a clinical endpoint met their respective endpoints (Fig. 3).

Fig. 3.

Primary endpoints of the four hematological malignancies (MM, AML, CLL, DLBCL) with respect to achievement of endpoints, with the numerator indicating how many studies met endpoints, and denominator indicating total number of studies for that subset.

Fig. 4 highlights the likelihood of meeting a primary endpoint based on what the endpoint was. The primary endpoints were met in 11/35 (31%) of studies reporting results of clinical endpoints. Ten of those 35 studies measured cancer response as the primary endpoint, with it being met in 4 (40%) of studies. PFS was assessed as a primary endpoint in 6 studies, with the endpoint being met in none of the six. OS was a primary endpoint only twice, and no results are available for those trials (NCT00454168 & NCT04229979). Duration of remission was assessed in 17 studies as the primary endpoint, with it being met in 7 (41%) of them. Translational outcomes were the primary endpoint in 81 studies, of which the endpoint was met 65 (80%) of the time. Safety was examined as a primary endpoint in 74 studies, with the endpoint being met 51 (69%) of the time (Fig. 4). Four studies measured both safety and translational immunogenicity as the primary endpoint.

Fig. 4.

Primary endpoints of anti-cancer vaccine trials and studies that successfully achieved the categorized endpoints. ∗4 studies included both safety and translational/immunogenicity as the primary endpoints. The endpoint for 1 study could not be ascertained.

Lifecycle of an individual vaccine product

A total of 187 individual vaccine products were assessed, amongst all studies in our cohort. Of these, 88 were studied in a large subsequent phase study. 33 were studied in a randomized study of any phase, and only 3 products were studied in a randomized phase 3 study. Of these three randomized phase III trials, one trial consisted of a BCG vaccine while the other two were a peptide-based vaccine.15, 16, 17 Further details on vaccine types for each study are provided in Supplemental Table S3, and a summary of tested endpoints for each vaccine type is presented in Fig. 5.

Fig. 5.

Primary endpoints of different vaccine types with respect to achievement of endpoints, with the numerator indicating how many studies met endpoints, and denominator indicating total number of studies for that subset.

Of the total 33 vaccine products tested in randomized trials, all had been evaluated in preceding non-randomized studies.

Trials that showed benefit in a clinical endpoint

A total of 11 studies demonstrated benefit in a clinical outcome (of 35 studies with clinical outcomes as primary endpoint), with 4 of those indicating improvement in cancer response, and 7 showing improvement in duration of remission. Six of those 11 trials investigated BCG in patients with AML. BCG-based vaccines constituted a total of 17 studies, with all of those studies targeting AML. Among these, 6 (out of 17) trials demonstrated efficacy by successfully meeting their primary endpoint. Of the 17 studies, 15 were randomized, with only 6 showing benefit in clinical outcomes. Besides BCG for AML, there was no instance of vaccines meeting their endpoint of cancer efficacy in a randomized trial, and no instance in which any vaccine improved OS.

Detailed description of the 11 studies demonstrating benefit in a clinical endpoint

In a trial involving 19 patients receiving chemotherapy alone and 23 patients receiving BCG plus chemotherapy, those in the chemotherapy plus BCG group experienced prolonged remission.¹⁸ Similar findings were observed in two additional randomized trials, involving 64 patients,¹⁹ and 41 patients respectively.²⁰ Another randomized trial, which assessed various combinations and sequences of BCG in 14 patients with AML, also demonstrated benefit for regimens incorporating BCG.¹⁶ One of the largest randomized trial of BCG in AML involving 295 patients, reported a longer duration of remission in the BCG group (35 vs. 19 weeks, p < 0.10), although survival differences were not statistically significant.²¹ One non-randomized trial of 30 patients showed a potential benefit of BCG in extending remission; however, all patients in this study were allocated to receive BCG, albeit in varying frequencies.²² Another non-randomized trial involving 41 patients treated with methotrexate for AML also indicated that BCG might contribute to prolonging remission.²³

Several other vaccine products demonstrated anti-cancer efficacy by achieving their respective primary endpoints. Another study involved a single-arm trial of allogeneic dendritic cells (DCP-001) in 20 patients with AML who were in remission after initial therapy but still had measurable residual disease (MRD). The trial successfully met its primary endpoint by demonstrating improvements in MRD.²⁴ Another single-arm trial evaluated GM-CSF-based vaccine in combination with lenalidomide in 15 patients with MM, which achieved its primary endpoint of converting near-complete responses to complete responses.²⁵ Additionally, cancer responses were observed in 6 patients in a single-arm study involving 13 DLBCL patients receiving a lipid depot-based therapeutic cancer vaccine composed of survivin epitopes.²⁶ Finally, in a single-arm trial involving 27 patients with multiple myeloma post-autologous stem cell transplantation, the consolidation therapy with APC8020 (Mylovenge™) achieved an objective response rate of 96%, although the confirmatory randomized trial was negative.²⁷

Risk of bias assessment

Among the 26 randomized controlled trials with available results that were assessed for risk of bias using the Cochrane risk-of-bias tool, the majority demonstrated low to intermediate risk of bias across the evaluated domains. Specifically, 7 studies (27%) were rated as low risk of bias, 17 studies (65%) as intermediate risk, and only 2 studies (8%) demonstrated serious risk of bias (Baker et al., 1982²⁸ and Presant et al., 1980¹⁶). The most common sources of bias were related to measurement of outcomes and selection of reported results, particularly in older studies from the 1970s–1980s era.

Risk of bias for all RCTs are summarized in the Supplementary, with most studies at low or intermediate risk of bias.

Discussion

Our comprehensive systematic review of anti-cancer vaccines in patients with hematological cancers indicates that although these vaccines commonly demonstrate immunogenicity, they have generally failed to show consistent anti-cancer activity. Besides BCG vaccine administration for AML, there was no instance of a vaccine product meeting a clinical efficacy endpoint in a randomized trial. Furthermore, there was not a single instance reported in which a vaccine product demonstrated an improvement in overall survival.

Our study demonstrated heterogeneity in the investigational setting, with trials often being conducted in the adjuvant/combination setting and in combination with other cancer therapies such as chemotherapy and immunotherapy. Although these clinical trials allowed for the assessment of safety and immunogenicity, such trial designs—particularly single-arm studies—make it difficult to isolate the vaccine's true effects. Trials, where vaccine products showed efficacy in an endpoint measuring cancer burden, were usually single-arm, included a small sample size, and enrolled patients who were often in remission already from previous anti-cancer therapies, limiting any definitive evaluation of the vaccine's appropriate efficacy. The findings from this study suggest that there is limited evidence to support the use of anti-cancer vaccines as an effective therapeutic tool in hematological malignancies. The results of our analysis highlight the need for larger studies with clear clinical endpoints for accurate ascertainment of clinical improvements in meaningful patient outcomes.^5,29,30

The limited clinical success of anti-cancer vaccines in hematological malignancies can be attributed to several fundamental challenges inherent to blood cancers. Unlike solid tumors, hematological malignancies typically exhibit lower mutational burdens, resulting in fewer immunogenic neoantigens available for vaccine targeting.³¹ However, this pattern is not universal among hematological malignancies; AML, for example, can present with high mutational burdens.³² Additionally, factors beyond neoantigen availability may contribute to the differential outcomes observed with anti-cancer vaccines across tumor types.

Despite these distinctions, the promise of anti-cancer vaccines remains largely unrealized in both solid tumors and hematological malignancies. Even the FDA-approved vaccine for prostate cancer, Sipuleucel-T, has faced scrutiny regarding its efficacy. Some analyses suggest that the observed survival benefit may have been confounded by inappropriate crossover in clinical trials rather than reflecting true therapeutic efficacy.³³ Additionally, these cancers often involve the immune system itself, creating inherent immunosuppression through mechanisms such as hypogammaglobulinemia in CLL and immune paresis in multiple myeloma. The systemic nature of blood cancers, combined with the immunosuppressive bone marrow microenvironment, presents unique obstacles compared to localized solid tumors. Furthermore, most vaccine trials in our analysis utilized single antigen approaches with suboptimal adjuvants.

BCG trials showed promise, with one of the largest RCTs involving 295 patients found that remission duration was extended with BCG treatment compared to chemotherapy alone, although the same study found that survival benefit did not reach statistical significance.²¹ While historical BCG trials in AML from the 1970s–1980s demonstrated some clinical benefit in extending remission duration, these approaches have limited contemporary relevance given advances in AML treatment and understanding of immune mechanisms. The non-specific immune activation achieved by BCG, while notable in our systematic analysis, does not provide a viable framework for modern vaccine development in hematological malignancies.

Among the analyzed studies, there was only one Phase 3 study with reported results. A 1980 phase 3 study incorporated BCG, traditionally a tuberculosis vaccine, in two chemotherapy regimens for AML maintenance therapy, in which both regimens included BCG, and the study did not measure the direct efficacy of BCG against leukemia but rather compared the overall effectiveness of two different chemotherapy protocols that both utilized BCG.¹⁶ A randomized phase 3 trial beginning in 2005 intended to explore whether a PR1 leukemia peptide vaccine combined with GM-CSF could improve OS of patients with AML, but the study has not posted any results, with the latest update on clinicaltrials.gov in 2014, suggesting that there may have been challenges in study completion.¹⁵ Finally, a randomized phase 3 trial that started in 2021 is exploring whether the galinpepimut-S peptide vaccine combined with montanide and GM-CSF could improve OS of AML patients in second complete remission compared to four best available active comparators used as monotherapies or in combination (e.g., venetoclax, cytarabine, azacitidine or decitabine, hydroxyurea for palliative management). The study has not uploaded any results and has a projected primary completion date of September 2025.¹⁷

Furthermore, the studies analyzed included several phase 2 trials measuring and meeting a clinical primary endpoint.24, 25, 26, 27 There was no indication of these studies moving forward to phase 3 despite successfully meeting a clinical primary endpoint. Possible reasons could be the complexity of the immune responses observed and the lack of randomized control groups,³⁴ which may hinder definitive conclusions about the vaccine effect. For example, one study focused on a combination therapy of the Maveropepimut-S (MVP-S) vaccine, Pembrolizumab, and cyclophosphamide in DLBCL.²⁶ Although there was an improvement in cancer response, the trial lacked a randomized control group, making it difficult to attribute observed clinical responses solely to the MVP-S vaccine rather than other factors like pembrolizumab or cyclophosphamide. In another instance, a phase 2 study's primary endpoint was to evaluate whether infusions of Id-KLH primed CD3/CD28 activated autologous lymphocytes mediate a more intense Id-specific immunity than non-Id-KLH primed CD3/CD28 activated autologous lymphocytes (immunogenicity). Although the study found that the Id-KLH group demonstrated a more robust immune response with significant upregulation of immune activation and effector function genes in T-cells compared to the control group, there was no significant improvement in PFS. This indicates that while the vaccine may have activated the immune system against hematological malignancy, this response did not translate into measurable clinical benefits in terms of disease progression. The absence of improvement in PFS highlights that immunological responses cannot reliably serve as surrogates for clinical activity and brings into question the utility of using solely immunological markers to predict therapeutic success.³⁵

Despite promising, albeit limited, findings of the BCG vaccine for patients with AML, the success of other therapeutic cancer vaccines in achieving clinical benefits remains extremely limited. Although the technology retains promise and consistently demonstrates immunogenicity, there is currently limited evidence to support anti-cancer vaccines as a therapeutic strategy to target hematological cancers. The high costs and stringent requirements for conducting randomized phase 3 trials often deter progression beyond the exploratory phases of vaccine development. Furthermore, immune evasion mechanisms, such as the loss of neoantigens and upregulation of immune checkpoint molecules by tumors, suggest adaptive resistance that can blunt the long-term efficacy of vaccines.^36,37 These variabilities highlight the challenges in advancing cancer vaccines from phase 2 to phase 3 trials and ultimately hinder the development of broadly effective therapeutic vaccines.

This study systematically evaluated the collective evidence on anti-cancer vaccines in hematological cancers. The strength of this systematic review lies in its exclusive focus on clinical trial studies investigating anti-cancer vaccines in patients with hematological cancers, incorporating studies from diverse geographical regions and various cancer types, which enhances the generalizability of our conclusions. Several important limitations should be acknowledged in interpreting these findings. First, we were unable to conduct a quantitative synthesis due to the sheer heterogeneity of assessed interventions, and a lack of clear description of the endpoint of a study in some instances. Given that the median sample size was 18 (IQR = 20), the majority of trials were likely underpowered, with 159 of 187 studies enrolling fewer than 50 participants. Furthermore, although we attempted to be thorough, it is possible that some trials may have been missed from our search strategy. Given that only 2 of 26 randomized studies with available results (8%) demonstrated serious risk of bias, we did not conduct a formal sensitivity analysis excluding these studies. The vast majority of randomized trials (92%) showed low to intermediate risk of bias, providing confidence in the robustness of our findings. Moreover, both studies with serious risk of bias (Baker et al., 1982 and Presant et al., 1980) were historical BCG trials from the early 1980s that supported rather than contradicted our primary finding that BCG vaccines demonstrated clinical efficacy in AML. Their exclusion would not change our key conclusion that besides BCG, no other vaccine products met clinical efficacy endpoints in randomized trials for hematological malignancies.

The disappointing clinical outcomes can be attributed to several fundamental constraints inherent in historical cancer vaccine approaches. Traditional vaccines targeting broadly expressed tumor-associated antigens (CEA, survivin, NY-ESO-1, hTERT) have demonstrated limited clinical efficacy due to multiple interconnected factors.38, 39, 40, 41 These antigens exhibit poor immunogenicity, relatively low tumor expression levels, and insufficient potency to overcome the immunosuppressive tumor microenvironment.⁴² This hostile environment is characterized by regulatory T cells, myeloid-derived suppressor cells, immune checkpoint overexpression, and inhibitory cytokines such as TGF-β.^43,44

Additional barriers include the low mutational burden of many tumors, which restricts the availability of tumor-specific neoantigens, while conventional peptide-based vaccines utilizing standard adjuvants (KLH, incomplete Freund's adjuvant) have proven inadequately immunogenic to mount effective anti-tumor responses.45, 46, 47

Despite these historical limitations, emerging strategies offer considerable promise for enhancing therapeutic outcomes in hematological malignancies. Advances in genomic sequencing now enable identification of highly immunogenic, patient-specific neoantigens that can elicit robust tumor-specific T cell responses. Novel vaccine platforms, including RNA-based, DNA-based, dendritic cell-peptide, and nanoparticle delivery systems, demonstrate superior immunogenicity compared to traditional approaches. Combination strategies that incorporate checkpoint inhibitors and stimulatory cytokines alongside vaccination may also be promising. These approaches represent a shift toward overcoming the immunosuppressive barriers that have historically limited cancer vaccine efficacy.48, 49, 50, 51

Future vaccine development could prioritize personalized neoantigen platforms leveraging advanced sequencing technologies to allow more precise targeting of cancer cells and immune responses.⁵² Other promising approaches include the utilization of mRNA-based vaccines formulated with lipid nanoparticles.⁵³ Additionally single-cell sequencing may allow deeper understanding of the tumor microenvironment which may provide new insights for optimization of future vaccine design.

In summary, this comprehensive systematic review demonstrates that current anti-cancer vaccines approaches in hematological malignancies, while immunogenic, lack consistent clinical efficacy. Moving forward, larger, randomized studies are needed to accurately ascertain whether anti-cancer vaccines offer meaningful clinical benefit. More importantly, the development of new vaccine platforms incorporating novel antigens, delivery systems, and combination strategies will be essential to achieve therapeutic benefit and improvement in clinical endpoints as the current vaccines are unlikely to yield meaningful advances in hematologic malignancies.

Contributors

D.S.: Contributed to study screening, data extraction, data access and verification, data curation, analysis, and interpretation. Assisted in writing the original draft and participated in reviewing and editing the manuscript.

V.S.: Contributed to study screening, data extraction, data access and verification, data curation, analysis, and interpretation. Assisted in writing the original draft and participated in reviewing and editing the manuscript.

K.S.: Assisted in data extraction, data access and verification, data analysis, and interpretation. Created figures and tables. Contributed to writing the original draft and participated in reviewing and editing the manuscript.

P.J.S.: Assisted in data extraction and contributed to writing the original draft.

M.A.: Assisted in data extraction and participated in reviewing the manuscript.

A.H.: Contributed to literature search, and review, as well as reviewing and editing of the manuscript.

V.P., R.C., M.H.Q.: Contributed to the conception of study design, reviewing and editing the manuscript, and provided expert advice on the content matter.

G.R.M.: Contributed to the conception of study design, accessed and verified data, participated in data review, supervised the entire project, and assisted with reviewing and editing of the manuscript.

Data sharing statement

All data was gathered from publicly available data. The dataset can be made available upon request to the corresponding author.

Declaration of interests

GRM: Receives honoraria from writing for MashupMD and Medscape and research funding to site from Janssen; RC: Consulting/Advisory Board-Alexion (AstraZeneca), Janssen, Sanofi Pasteur, and Adaptive Biotech; Research funding: AbbVie, Genentech.

MHQ: Research funding from Johnson & Johnson, Angiocrine, and Sanofi; Advisory Board of Sanofi.

VP: Receives research funding from Arnold Ventures through a grant made to UCSF, and royalties for books and writing from Johns Hopkins Press, MedPage, and the Free Press. He declares consultancy roles with OptumRX; He hosts the podcasts, Plenary Session, VPZD, Sensible Medicine, writes the newsletters, Sensible Medicine, the Drug Development Letter, and VP's Observations and Thoughts, and runs the YouTube channel Vinay Prasad MD MPH, which collectively earn revenue on the platforms: Patreon, YouTube, and Substack.

None of the other authors report any conflicts of interest.