Applying principles of vaccine development to oncomicrobial vaccines Open Access

Vaccination is a cornerstone of public health, estimated to save 4 to 6 million lives globally per year.¹ However, until recently, vaccination has mainly been directed toward preventing infectious disease. As we develop a greater appreciation of the complexity of host-microbe interactions, opportunities to apply vaccination to other aspects of public health, including cancer, have become evident.

In 1911, Peyton Rous provided the first evidence of infectious etiology of cancer by showing cell-free tumor filtrates could induce tumors by transmitting a virus.^2,3 Since then, the ability of some microbes to exert oncogenic effects has become well established, with an estimated 13% to 18% of the global cancer burden attributable to known oncogenic infectious agents, or oncomicrobes.⁴ Hepatitis B virus (HBV), hepatitis C virus, human papillomavirus (HPV), Epstein-Barr virus (EBV), and the paradigmatic oncobacterium, Helicobacter pylori, are responsible for ∼97% of this estimated cancer burden attributable to infection,^4,5 and the strong evidence of their oncogenicity has led to each being considered a class 1 carcinogen by the International Agency for Research on Cancer.⁶ These established agents and additional, recently identified oncomicrobes provide opportune vaccine targets for reduction of cancer burden. In this review, we summarize key concepts and principles relevant to oncomicrobe vaccine design, and highlight important successes of oncomicrobe vaccination, remaining challenges, and novel emerging targets.

Vaccination is exposure to an agent, or parts thereof, to evoke immunity and establish immune memory. This protects the vaccinated individual from future infection (through establishing circulating antibodies and effector cells, and/or by enabling a more rapid and effective anamnestic response) and, if the number of individuals vaccinated is sufficient to disrupt chains of transmission, also protects unvaccinated members of the population (ie, herd effect).^7,8 To develop effective vaccines, key variables including vaccine type, antigen selection, administration route, and vaccine timing must be appropriately tailored for the specific microbe target of interest.

Most clinically approved vaccines can be categorized into 5 types: live attenuated, killed/inactivated, recombinant protein, viral vector, and nucleic acid (Figure 1). Use of a whole organism as the immunogen is represented by the first 2 categories: live attenuated (“weakened”) vaccines and inactivated vaccines. Some live attenuated vaccines use microbes that are related to, but less virulent than, the pathogen of interest to induce crossprotective immunity. Indeed, such crossprotective application was used in the first application of vaccination, wherein cowpox vaccination was used to confer protection from smallpox⁹ (Box 1). Alternatively, the pathogen of interest can be directly attenuated, such as by desiccation or mutation during serial passaging in an atypical host.^10,11 In some instances, live attenuated vaccines have natural and directed attenuation, such as Bacillus Calmette–Guérin (BCG) vaccines against tuberculosis (caused by Mycobacterium tuberculosis) that contain mutated Mycobacterium bovis virus with reduced invasion of lung tissue.¹² 

Inactivated vaccines, or killed vaccines, use a whole organism rendered noninfectious (typically by heat or chemical treatment),¹¹ making them considerably safer than live attenuated vaccines, especially for immunocompromised individuals.¹³ Current polio vaccines in Canada and the United States, for example, are inactivated instead of live attenuated,¹⁴ to eliminate disease risk due to inadequate attenuation. However, the downside of killed vaccines is that they may be less effective than live vaccines. For example, live attenuated vaccines for viruses and invasive bacteria can maintain their ability to access the cytosol, stimulating CD8⁺ T-cell responses against intracellular antigens,¹⁵ whereas killed vaccines elicit CD8⁺ T-cell immunity only via crosspresentation.¹⁵ In vaccination settings for which CD8⁺ T-cell responses are important, this may be a critical distinction.

Subunit vaccines deploy specific components of an agent rather than the whole organism. Subunit vaccines may use protein purified from the agent or recombinantly produced protein as antigens. Subunit vaccines often lack pathogen-associated molecular patterns, and thus typically cannot sufficiently initiate innate immune responses that drive the development of protection.¹⁰ Therefore, adjuvants (which stimulate stronger immune responses to antigens) are required for subunit vaccines. Adjuvants, the most common of which are alum salts, enhance innate immune engagement through sustained release and a variety of other mechanisms, including Toll-like receptor 4 (TLR4) stimulation, cytokine/chemokine production, and immune cell recruitment.¹⁶ Subunit vaccines can comprise different types of subunits. For example, toxoid vaccines are subunit vaccines that use as antigen an inactivated form of an otherwise disease-causing microbial toxin, such as tetanus toxin or diphtheria toxin.¹³ Other subunit vaccines, such as HPV vaccines, are engineered to form self-assembling virus-like particles.¹⁷ A polysaccharide can also serve as the subunit; however, polysaccharides are poorly immunogenic on their own¹³ and are, therefore, most often formulated as protein-polysaccharide conjugate vaccines, in which the protein component generates T-cell help to promote antipolysaccharide B-cell responses.¹³ Finally, viral vector and nucleic acid vaccines are subunit vaccine modalities that facilitate the translation of protein immunogens by host cells, thus eliciting immunity to this specific component of an agent.

Viral vector vaccines are engineered viruses that exploit viral host cell entry to deliver nucleic acids that encode for an antigen of interest. Viral vectors used for vaccination are typically highly attenuated or replication-incompetent to maximize safety.^18,19 In addition, some viral vector platforms, such as lentivirus, can integrate into the genome and therefore pose a risk of insertional mutagenesis.²⁰ Integration-defective lentiviral vectors, which are maintained as an episome, circumvent this issue.¹⁹ Importantly, preexisting immunity to the virus type used can hamper viral vector vaccine immunogenicity. Higher doses and/or additional doses can be attempted (which risks exacerbating the problem), or viral vector types with lower seroprevalence may be used.¹⁸ Furthermore, heterologous prime-boost strategies can be used to circumvent antivector immunity induced by the prime dose.¹⁸ 

Nucleic acid vaccines contain DNA or messenger RNA (mRNA) encoding an antigen of interest, and must deliver their DNA or mRNA cargo into host cells. Encapsulation of nucleic acids in lipid nanoparticles (LNPs) enables effective nucleic acid delivery, as evidenced by severe acute respiratory syndrome coronavirus 2 mRNA LNP vaccines (mRNA-1273 and BNT162b2). LNP-based nucleic acid vaccines require cell uptake by endocytosis and endosomal escape to deliver their cargo.²¹ This is initiated by the ionizable lipid of LNP formulations, which becomes protonated under the low pH during endosome maturation, and leads to fusion of the LNP and endosomal membranes, releasing the cargo into the cytoplasm.^21,22 In the case of mRNA vaccines, mRNA cargo can be directly translated in the cytoplasm following delivery. Alternatively, in the case of DNA vaccines, DNA cargo must reach the nucleus for transcription.²³ This can be facilitated by nuclear localization signals or other mechanisms;²³ however, DNA vaccines have shown limited success compared with mRNA vaccines, potentially due to this additional requirement for efficient antigen expression. mRNA used in nucleic acid vaccines requires several design considerations to optimize antigen expression, particularly involving capping, 5ʹ and 3ʹ untranslated region (UTR) sequences, and the poly-A tail.²⁴ Furthermore, mRNA in these vaccines typically have modified nucleotides incorporated, such as N¹-methylpseudouridine, which can improve immunogenicity by decreasing innate intracellular sensing that inhibits antigen translation.^25-27 

Viral vector vaccines and LNP-encapsulated vaccines stimulate innate immunity and are, therefore, “self-adjuvanting,” avoiding the need for additional adjuvants in their formulation.^18,28,29 These modalities, however, may be limited in applicability for some bacterial antigen targets. For example, data suggest mammalian-derived expression (such as that of viral vector and nucleic acid vaccines) of a bacterial antigen can result in limited native antigen recognition due to nonnative glycosylation and/or folding of bacterial antigen.^30,31 This is in contrast to mammalian expression of viral antigen targets, such as severe acute respiratory syndrome coronavirus 2 Spike mRNA LNP vaccines, which would be expected to exhibit native glycosylation and structure, because viruses naturally rely on host machinery to propagate.

A major advantage of viral vector and nucleic acid vaccines is their ability to be rapidly developed using a “pipeline” approach, in which any antigen of interest may be fed into the manufacturing platform once its sequence is identified. The benefit of rapid development is well exemplified by the recent COVID-19 pandemic, for which the Moderna mRNA-1273 mRNA LNP vaccine went from sequence to clinical trial injections in only 66 days.³² 

Lastly, vaccines may be monovalent (protecting against a single strain or pathogen) or multivalent (protecting against multiple strains or pathogens, also called combination vaccines). For example, current HPV vaccines are multivalent, designed for protection against multiple HPV strains (ie, quadrivalent Gardasil targeting HPV6/11/16/18).³³ MMR vaccines are also multivalent, protecting against measles, mumps, and rubella.

Antigen selection is a critical aspect of effective vaccine design. For live attenuated vaccines and inactivated/killed vaccines, where the pathogen itself defines the vaccine antigen content, representation of circulating variants of that pathogen is important. For example, each year, based on surveillance data, the World Health Organization selects the influenza strains predicted to be most prevalent for vaccine formulation, thereby maximizing overall protection.³⁴ It can also be advantageous for the vaccine to represent the most infectious stage of a pathogen’s replication cycle. For example, live attenuated vaccines for malaria have been developed using Plasmodium falciparum sporozoites with promising initial results.³⁵ Although the diverse antigen content delivered by whole-organism vaccines can provide broad coverage, a potential drawback is that responses may be primarily directed toward antigens that are poorly suited for establishing protection. For example, H pylori undergoes autolysis, releasing immunogenic intracellular proteins that may act as “decoys” and diminish responses to surface antigens.³⁶ Additionally, immunodominant antigens may not be well conserved between strains, which can be a major barrier to effective protection for targets with significant strain diversity, such as Fusobacterium nucleatum.

Subunit vaccines allow rational selection of antigen target(s). However, antigen selection must be made with careful consideration of several factors. First, the antigen should be expressed on the surface of the agent, to be accessible to antibodies; second, the antigen should be conserved among the agents being targeted; third, the antigen should be consistently expressed, particularly during relevant phases of infection, such that the target is present/sufficiently abundant for targeting; fourth, the antigen should be specific to the pathogenic targets of interest, to minimize off-target effects (eg, disturbing commensal microbe populations and leading to dysbiosis) and, finally, the antigen should be functionally important to allow protection mediated by antibody-mediated neutralization. Indeed, many virus-targeted vaccines are protective via neutralization of proteins that enable infection.^37-39 In the case of bacteria, neutralization can also provide protection through inhibition of key pathogenic functions of the antigen target.

Antigen avidity (the cumulative affinity achieved through multiple interactions) significantly affects immunogenicity following vaccination, with increased valency associated with greater B-cell activation and antibody titers.^40,41 Recombinant subunit vaccines are advantageous in this regard as they support inclusion of multimerization domains, enabling multimeric antigen generation. For example, the multimeric nature of HPV virus-like particles is proposed as a primary reason for their exceptional potency.¹⁷ For viral vector and nucleic acid vaccines, surface display of antigen can effectively increase avidity by antigen localization/concentration on the cell surface, and has been shown to enhance antibody titers by approximately an order of magnitude compared with secreted monomeric antigen.⁴² 

Infectious agent vaccines are often effective prophylactics but generally ineffective therapeutics. Therapeutic vaccination is typically only viable in rare circumstances of long pathogen incubation periods (weeks to months), such as for rabies and HBV, during which vaccine-induced immunity can be established before widespread infection and acute disease. Aside from these exceptions, it is necessary that vaccination is timed to establish immunity before exposure to the microbe of interest. Therefore, understanding transmission underpins ideal vaccine timing for successful protection.

Booster vaccinations are also relevant to vaccine timing, and can be applied in 2 distinct manners. Firstly, boosters can be used to increase immunity above protective thresholds in cases where a single immunization elicits insufficient immunity. In this scenario, booster-induced immunity must therefore be established before exposure, as previously described. Alternatively, boosters can be used to maintain immunity above protective thresholds after waning. Timing of these boosters must be appropriately spaced to ensure constant levels above protective thresholds. For example, infants have generally weaker and shorter-lived immune responses to vaccination. Thus, booster series (often ≥2 vaccinations) are commonly used from infancy to early childhood to establish and maintain protective levels of immunity.⁴³ 

Vaccines can be administered through different routes, such as intramuscular, intranasal, IV, intradermal, subcutaneous, or oral. Route of administration can dictate the immune cell populations engaged and thus impact protection. Ideally, the route of vaccine delivery mimics the natural route of infection.¹⁵ For example, a major advantage of oral poliovirus vaccination is that oral administration can stimulate better mucosal immunity than intramuscular inactivated poliovirus vaccination, which may contribute to its greater protection against viral shedding.^44,45 Similarly, it has been demonstrated that IV administration of malaria vaccines induces greater CD8⁺ and CD4⁺ T-cell responses in both peripheral blood and liver, and better protection compared with subcutaneous administration.^35,46 

Vaccinating against oncomicrobes as a means of cancer control has long been a major interest of the scientific community, particularly for oncomicrobes with significant burden (Table 1). In this subsection, the successes, remaining challenges, and future directions of such efforts are discussed.

HBV was the first oncomicrobe for which an anticancer vaccine was successfully developed. Chronic HBV infection is estimated to be responsible for 50% to 80% of hepatocellular carcinoma (HCC),⁵⁷ the dominant form of primary liver cancer (∼80%).⁵⁸ Infants perinatally infected are more susceptible to developing chronic HBV infection (80%-90% of infections become chronic) than children infected <6 years of age (20%-30%) or healthy individuals infected as adults (<5%),⁵⁹ thus it is important to establish immunity shortly after birth.⁶⁰ Accordingly, HBV vaccination consists of multiple doses given within the first 6 months.

Plasma-derived, first-generation HBV vaccines protect against infection with high efficacy.⁶¹ However, the long interval between HBV infection and HCC manifestation has made direct evaluation of protection against HCC difficult until recently. The first mass HBV vaccination program for infants began in 1984 in Taiwan,^62,63 and reports comparing incidence before and after vaccine program implementation have shown early indications of protection.⁶⁰ Perhaps the best current example is a randomized case-control study of infants vaccinated between 1983 and 1990 in China, which demonstrated protective efficacy of 72% and 70% against liver cancer incidence and mortality, respectively.⁴⁸ Continued monitoring of these cohorts will be needed to evaluate protection into older age, as well as confirmation of similar results for modern recombinant HBV vaccine formulations; however, based on data available to date, HBV vaccination has the potential to dramatically decrease liver cancer incidence worldwide.

HPV is responsible for almost all (>99%) cervical cancers. Moreover, before HPV vaccine rollout, it was estimated that >90% of men and >80% of women, with ≥1 opposite sex partner, will acquire ≥1 HPV infection in their lifetime.⁶⁴ Although most HPV infections spontaneously resolve (67% within 12 months), chronic infection is associated with greater risk of cervical cancer.^65,66 Risk varies with HPV subtype, for example, HPV subtypes 16 and 18 are considered high risk for cervical cancer, whereas HPV subtypes 6 and 11 are considered low risk for cervical cancer (though are associated with ∼90% of genital warts).^67,68 Because HPV is sexually transmitted, vaccination programs can be successfully applied prophylactically, before onset of sexual activity. Given the association of HPV with cervical cancer, initial HPV vaccination programs focused on vaccinating girls aged 9 to 14 years. However, HPV is also associated with lower-incident cancers, such as oral cavity, penile, anal, vulvar, vaginal, oropharyngeal, and laryngeal.⁴ Therefore, immunization of boys is now recommended as a secondary target group.⁶⁹ These vaccinations should also increase herd effects and further benefit girls as well.

As with HBV vaccination, time was required for the effects of HPV vaccination on cancer incidence to become evaluable. Recent assessments indicate an 88% reduction in cervical cancer incidence in adolescent girls vaccinated with quadrivalent HPV vaccines before 17 years of age.⁴⁷ The continued effect of HPV vaccines on reduction of cervical cancer incidence over time, as these cohorts age, and the anticipated protection against other HPV-associated cancers, will indicate the full potential of these vaccines. Although HPV16 and HPV18 are covered by all HPV vaccines, and were initially prioritized in bivalent and quadrivalent designs on the basis they account for ∼70% of total cervical cancers,⁴⁷ the more recent 9-valent HPV vaccination programs protect against additional high-risk cancer-related HPV strains,⁷⁰ and are expected to further reduce the incidence of HPV-associated cancers.

In the 1980s, Barry Marshall famously drank a culture of H pylori and developed gastritis, supporting, anecdotally, a causative relationship between H pylori infection and gastric inflammation that has since been verified.⁷¹ Over 40% of the global adult population is estimated to be infected with H pylori, though there is extensive geographic variation in carriage rates.⁷² Infection is believed to predominantly occur during childhood.⁷³ Although some individuals can clear H pylori infection without treatment, as was the case for Barry Marshall,⁷⁴ most infections persist for life unless treated.⁷³H pylori can be routinely detected (eg, by urea breath test) and treated with antibiotics.⁷⁵ However, most infections are asymptomatic,⁷³ and therefore many infected patients do not seek treatment and remain chronically infected. Concerningly, a high prevalence of pretreatment isolates (≥10%) show resistance to common antibiotics.⁷⁶ Additionally, high rates of reoccurrence, either through reinfection (particularly in developing countries with higher prevalence) or through recrudescence (ie, incomplete elimination), further highlight that antibiotics have limited effectiveness in disease control, and emphasizes the need for alternative interventions that confer long-lived immune memory and protection.^77,78 

H pylori has been estimated to be associated with 75% to 80% of gastric cancers.⁴H pylori can survive the acidic stomach lumen during colonization due, in part, to its production of urease, which converts urea to ammonia,⁷⁹ neutralizing stomach acid and creating a more favorable environment for the microbe. Further, H pylori is highly motile, and can embed itself in the gastric mucosa, adhere to epithelial cells, and exhibits various immune evasion mechanisms that support its long-term persistence in the stomach.⁷⁹ 

H pylori elimination can improve cancer-related outcomes. For example, a recent systematic review of H pylori eradication therapy reported a relative risk of 0.54 for gastric cancer incidence in treated healthy individuals compared with placebo or no treatment healthy control subjects.⁸⁰ However, given the widespread prevalence and often asymptomatic nature of H pylori infection, and the potential for further emergence of antibiotic-resistance strains, early childhood vaccination has the best potential for significant reduction of gastric cancer incidence. Unfortunately, despite the now long-standing recognition of the role of H pylori in gastric cancer, and the significant effort toward developing such vaccines, an effective, approved, H pylori vaccine has remained elusive due to immune evasion mechanisms, significant genetic diversity, and, as mice are not natural H pylori hosts, model systems that do not fully emulate human infection.⁸¹ Nevertheless, some H pylori vaccine clinical trials have shown promise. For example, an orally administered H pylori vaccine using a fusion protein of urease B subunit of H pylori and heat-labile enterotoxin B subunit of Escherichia coli, the latter of which acts as a mucosal adjuvant, was evaluated in a phase 3 clinical trial.⁸² This strategy demonstrated 71.8%, 55.0%, and 55.8% efficacy against H pylori infection in naïve individuals at 12, 24, and 36 months, respectively.⁸² Unfortunately, however, the vaccine has since been discontinued,⁸³ and therefore additional investigations of H pylori vaccine approaches, ideally with greater long-term protection, are needed.

EBV has been found to infect >90% of the global population,⁸⁴ establishing a life-long latency. EBV has been linked to numerous conditions, including infectious mononucleosis (IM), nasopharyngeal cancers (85% of incidences), gastric cancer (7.7% of incidences), and Hodgkin (50% of incidences) and non-Hodgkin lymphomas (55% of incidences),^4,49,85 including Burkitt lymphoma.⁸⁶ 

In the 1970s, Anthony Epstein proposed the development of an EBV vaccine to reduce the burden of EBV-associated cancers.⁸⁷ Since then, most vaccine development efforts have focused on the most abundant glycoprotein (gp) expressed on EBV’s envelope, gp350.⁸⁸ An adjuvanted gp350 monomer progressed to phase 2 clinical trials and demonstrated 98.7% seroconversion in vaccinated participants. Although EBV infection was not prevented, rates of IM were reduced.^89,90 This is an important finding, as an EBV vaccine that controls viral load may protect against associated cancers even without achieving sterilizing-immunity.⁸⁴ No further reports on this vaccine have been made. However, a gp350-ferritin nanoparticle vaccine, which elicited robust neutralizing antibody responses in mice and nonhuman primates, and demonstrated efficacy in a recombinant virus challenge murine model,⁹¹ is now in phase 1 clinical testing (ClinicalTrials.gov identifier: NCT04645147).⁹² Other EBV vaccines have targeted the membrane fusion molecules gB and gH/gL.⁸⁸ In rabbits vaccinated with trimeric or monomeric gH/gL, trimeric gB, and tetrameric gp350, higher neutralizing titers were observed relative to monomeric gp350.⁹³ Additionally, Moderna has launched a phase 1 trial for a multivalent mRNA vaccine encoding gH, gL, gp42, and gp220 to target IM (NCT05164094),⁸⁸ and a therapeutic candidate for long-term sequalae (NCT05831111) and, more recently, a phase 1 study evaluating an adjuvanted gH/gL/gp42-ferritin nanoparticle vaccine has been launched (NCT06908096).

As emphasized by the >10 years of development between the classification of HPV as a group 1 carcinogen and approvals for HPV vaccinations, and the lack of approved H pylori vaccines, now at >30 years from its classification as a group 1 carcinogen,⁸² vaccine development requires considerable research investment toward identifying underlying features of pathogenesis and protective immunity, and efficacious formulations. Nevertheless, as now evidenced by HBV and HPV vaccines, successful development of such vaccines can have significant impacts on reducing cancer incidence and, therefore, continued efforts are needed.

In recent years, enterotoxigenic Bacteroides fragilis (ETBF), polyketide synthase–positive (pks⁺) E coli, and F nucleatum have emerged as novel oncomicrobes in the setting of colorectal cancer (CRC). In this subsection, the biology and roles of ETBF, pks⁺E coli, and F nucleatum in cancer are discussed, and potential strategies for vaccine development against these agents are proposed.

B fragilis is a normal inhabitant of the healthy human colon. Two subtypes exist: nontoxigenic B fragilis (NTBF) and ETBF, and it is specifically the ETBF subtype that is enriched in sporadic and hereditary CRC.^94,95 Further, ETBF, but not NTBF, has been shown to potentiate the development of colonic tumors in APC^min/+ mouse models.⁹⁶ The distinguishing feature between these 2 subtypes is BFT, a 20-kilodalton toxin present in ETBF and absent in NTBF. BFT cleaves E-cadherin,⁹⁷ stimulating β-catenin signaling and cell proliferation,⁹⁸ and can disrupt epithelial cell barriers,^99,100 and lead to recruitment of myeloid cells to the distal colon, which contributes to tumorigenesis in murine models.¹⁰¹ As the procarcinogenic action of ETBF can be directly attributed to BFT, a toxoid vaccine deploying a mutated, recombinant variant of the toxin that lacks toxin functionality but elicits antitoxin immunity has potential. Indeed, targeted mutagenesis to ablate E-cadherin cleavage has been previously determined.¹⁰² 

E coli variants bearing a 50-kilobase pathogenicity island encoding a pks¹⁰³ are enriched in tumor specimens of spontaneous and hereditary CRC.^95,104 The product of this pks, colibactin, is genotoxic, causing DNA double-strand breaks, interstrand crosslinks, and chromosomal aberrations.^105-107 Intestinal organoids repeatedly exposed to pks⁺E coli have been used to identify 2 colibactin-specific mutational signatures.¹⁰³ These signatures, either individually or in combination, were found to be present in 2% to 8.8% of primary and metastatic CRC samples. Moreover, the colibactin target motif represented 2.4% of driver mutations in CRC, supporting a direct initiating role of colibactin in at least a small fraction of CRC.¹⁰³ 

Although colibactin is the direct mediator of this mutational signature and therefore would be an ideal vaccine target in the same manner as BFT, colibactin, as a secondary metabolite, is poorly immunogenic, which limits the potential of this strategy. A protein-conjugate vaccine approach could possibly be explored in this scenario; however, colibactin is also highly labile, which may make protein-conjugate vaccine development challenging. Recent data showing epithelial cell adhesion is required for colibactin’s mutational effects provide another potential tactic.¹⁰⁸ The 2 adhesins responsible for binding of E coli to epithelial cells, FimH and FmlH, may be effective vaccine targets, through which blockade of adhesion may prevent the localized production of colibactin that enables DNA damage by this rapidly destabilized mutagen. Consistent with this strategy, a FimH-neutralizing small molecule, siobofimloc, was effective in reducing DNA damage and tumor burden in pks⁺E coli–exposed mice.¹⁰⁸ FimH vaccines have been developed and shown to be effective for adhesion blockade of E coli in other contexts, such as urinary tract infections, and may be evaluated for benefit in CRC as well.^109,110 Given the pks mutational signature is established before 10 years of age,^103,111 such a vaccine would need to be given in the first years of life to act prophylactically.

Given the colonic niche of both BFT and pks⁺E coli, vaccines inducing strong mucosal responses in the gastrointestinal tract would be most applicable to protection. An orally administered vaccine may therefore be appropriate. However, both situations require use of specific subunit antigens to prevent targeting of related commensals (ie, NTBF, pks⁻E coli), and subunit vaccines have significant barriers to oral applications, such as stability during transit through the stomach and overcoming tolerogenic mechanisms.¹¹² Indeed, all licensed oral vaccines are formulated with live attenuated or killed whole cells.¹¹³ Furthermore, nonlive attenuated mucosal vaccines can suffer from short-lived protection, particularly in children (which would be the ideal target population for ETBF and BFT vaccination due to early colonization by these oncomicrobes). The most relevant example is Dukoral, comprised of inactivated Vibrio cholera and recombinant cholera toxin B subunit. Protection against cholera conferred by this vaccine is lower and rapidly wanes in children aged 2-5 years compared with individuals aged >5 years.¹¹⁴ Furthermore, even protection in individuals aged >5 years is relatively short-lived (decreasing from 78% to 40% after the first and third year, respectively),¹¹⁴ compared with decade or lifetime protection conferred by many systemic vaccines. Thus, novel strategies may need to be developed and used for effective, long-lived mucosal immunity in the context of BFT and pks⁺E coli vaccination.

F nucleatum is a gram-negative, anaerobic bacterium ubiquitous in the oral cavity, and acts largely as a commensal microbe in this setting. Oral bacteria such as F nucleatum can escape the oral cavity during transient, low-level bacteremia that commonly occurs from everyday activities such as toothbrushing.¹¹⁵ When F nucleatum spreads hematogenously, its adhesive properties allow it to selectively colonize peripheral niches including tumors, mainly CRC,^116,117 but also breast,⁵³ esophageal,⁵⁴ pancreatic,¹¹⁸ oral/head and neck,⁵⁵ and cervical cancers.¹¹⁹ Tumor colonization by F nucleatum is mediated by its 2 key adhesin proteins, Fap2 and RadD, which bind to GalGalNAc and CD147, respectively, which are both overexpressed on the surface of some tumor cell types.^120,121 

F nucleatum is distinct from the other oncomicrobes in this review in being a cancer promoter rather than a cancer initiator. Nonetheless, associations of F nucleatum with increased tumor growth,^53,122 chemoresistance,¹²³ decreased effectiveness of immunotherapy,^124,125 metastasis,¹²⁶ recurrence,¹²⁷ and poor patient outcomes¹²⁸ are sufficient to prompt vaccine development efforts. F nucleatum, however, is a particularly difficult vaccine target, due to the need to continuously protect against systemic dissemination from the oral cavity, and its tropism for tumors, which are immunosuppressive microenvironments and therefore can offer safe harbor from host immunity. This requires a F nucleatum vaccine to induce sufficiently robust circulating serum antibodies to intercept bloodborne F nucleatum before tumor colonization. Toward this goal, IV vaccination may be well suited for F nucleatum, for the reasons described above for malaria IV vaccination.

Another major challenge to vaccinating against F nucleatum is the extraordinary strain diversity of this bacterium. Early experimentation with whole-cell vaccine approaches for F nucleatum demonstrated poor cross-reactivity, with >60% of antibodies isolated following whole-cell vaccination recognizing only a single isolate.¹²⁹ Subunit vaccines targeting conserved regions of the adhesins Fap2 and RadD may be a promising alternative strategy to prevent F nucleatum tumor colonization, analogous to targeting FimH/FmlH to disrupt the adherent properties of pks⁺E coli.

For oncomicrobes, vaccination timing can be prophylactic relative to infection or relative to tumor development. Prophylactic vaccination against ETBF and pks⁺E coli before tumor development, and ideally in early life, is well justified, because these agents are confirmed tumor initiators. This could be especially beneficial for individuals with elevated CRC risk, such as patients with ulcerative colitis (ie, 2%, 8%, or 18% of patients with ulcerative colitis develop CRC after 10, 20, or 30 years with the disease, respectively).¹³⁰ Alternatively, the aim of vaccinating against F nucleatum would be to prevent tumor promotion by this agent by blocking its spread from the commensal oral environment to an ectopic tumor environment. In addition, F nucleatum vaccination could be coordinated with standard radiotherapy and chemotherapy approaches, to prevent recolonization upon relapse, or coordinated with antibiotic therapy, to prevent recolonization after a course of antibiotics has eliminated the original bacterial population from sites of minimal residual disease.

It is also important to highlight that reduction of cancer incidence is not a viable end point for clinical trials that would lead to initial oncomicrobe vaccine approvals, due to the long delay between infection and cancer manifestation. For example, HBV vaccines were approved on the basis of effective prevention of acute hepatitis, not HCC.⁶¹ Similarly, HPV vaccines were approved based on reduction of precancerous cervical lesions, as a surrogate end point for cervical cancer.¹³¹ For H pylori and EBV, primary outcomes of prevention of gastritis and IM, respectively, could be used. Clinical evaluation of F nucleatum vaccination in combination scenarios previously described may provide a path to initial approval. Monitoring rates and outcomes of subsequent cancers in these cohorts may justify more widespread applications. Alternatively, BFT vaccination could be evaluated for prevention of diarrheal diseases, for which its association is well established.¹³² For pks⁺E coli, prevention of infection, and specifically prevention of its related mutational signature, which is established early in life, may be an ideal outcome of interest. Such preliminary data from these vaccines would be informative with respect to potential for cancer control. Given the large number of individuals that would stand to benefit from protection against these novel oncomicrobes, the potential of ETBF, pks⁺E coli, and F nucleatum vaccination demands investigation.

We are in an exciting era in which the cancer-reducing potential of HBV and HPV vaccination is being substantiated by population-scale, real-world data, validating vaccination as a viable means of controlling infectious agent-associated cancers. It is our sincere hope that such results will reinvigorate the development of vaccines against both established and novel oncomicrobial targets.

This work was supported by Cancer Research UK (C54768/A29062). This work was also supported by the Canadian Glycomics Network (GlycoNet), a member of the Networks of Centres of Excellence Canada program. C.A.D. was supported by Canadian graduate scholarships from the Canadian Institutes of Health Research.

Contribution: C.A.D., N.J.P.V., and R.A.H. wrote the manuscript.

Conflict-of-interest disclosure: C.A.D., N.J.P.V., and R.A.H. are working on developing Fusobacterium nucleatum vaccines with potential for institutional patent filings.

Correspondence: Robert A. Holt, British Columbia Cancer Research Institute, 675 W 10th Ave, Vancouver, BC V5Z 1L3, Canada; email: rholt@bcgsc.ca.