The hidden complexity of adjuvant selection

A vaccine’s antigen defines what the immune system is asked to recognise. Its adjuvant helps determine how that antigen is sensed, presented and remembered. Two formulations built around the same antigen can therefore produce different immune profiles—even when their adjuvants share a broad technological category.

This is more than a question of “more” or “less” immunogenicity. A useful comparison asks which responses changed, how large and durable those changes were, whether they were functional, and whether they translated into protection. Human head-to-head studies of influenza, malaria and tuberculosis vaccines illustrate both the potential importance of adjuvant choice and the limits of what immunological differences can tell us.

Beyond a Th1–Th2 shorthand

Adjuvants can activate innate immune pathways and influence the recruitment, activation and antigen presentation of dendritic cells and other antigen-presenting cells. The resulting cytokine and co-stimulatory environment helps shape the adaptive response. These events are context-dependent: the antigen, its physical presentation, dose, route, schedule and host all contribute [4,6].

Figure 1. Conceptual overview of how antigen and adjuvant contribute to vaccine immunity. Original Noulin BioConsulting illustration; scientific context: references 4, 6 and 7. View full-size figure 1

Th1 responses, often described through interferon-γ (IFN-γ) and related cellular functions, can be valuable against intracellular pathogens. Th2-associated cytokines such as IL-4, IL-5 and IL-13 are often linked to antibody-associated responses. But these labels are not a scorecard for protection. Antibodies can be essential against viruses, while CD4+ and CD8+ T cells may be important in other settings; the relevant balance depends on the pathogen and vaccine objective.

T follicular helper (Tfh) cells support germinal-centre reactions, where B cells undergo selection and maturation. That process can affect antibody affinity, epitope targeting, class switching, memory B-cell formation and the generation of long-lived antibody-secreting cells. Neutralisation is one important measure of antibody function, but it does not capture every useful Fc-mediated activity, breadth or tissue-specific response. Th17 and mucosal responses may also matter for some pathogens and routes of exposure. Human lymph-node sampling after influenza vaccination has directly documented Tfh-cell expansion, maturation and germinal-centre activity, while mechanistic work with AS01 has shown how innate activation can shape downstream cellular responses [6,7].

The practical distinction is between response magnitude and response quality. A higher titre may be useful, but it does not by itself establish better protection, broader recognition or longer-lasting immunity. Likewise, a cellular signature that looks biologically attractive is not automatically a validated correlate of clinical benefit. Adjuvant selection should be judged against the intended product profile and the evidence needed for the next development decision.

Figure 2. Formulation platforms and components of the adjuvants compared in this article. Original Noulin BioConsulting illustration. View full-size figure 2

Case 1 — Two oil-in-water emulsions, different H7N9 antibody responses

A randomised, double-blind phase II trial enrolled 980 adults aged 19–64 years to assess an inactivated, monovalent H7N9 vaccine with different doses and adjuvant schedules. The most informative direct comparison used the same H7N9 antigen, the same nominal 15-µg haemagglutinin dose and the same two-dose, 21-day schedule, with either AS03 or MF59 mixed with the vaccine at the study site.

Both adjuvants are oil-in-water emulsions, but they are not the same formulation. AS03 contains squalene and α-tocopherol; MF59 is a squalene emulsion stabilised with polysorbate 80 and sorbitan trioleate. After dose two, 84% of the AS03 group reached a haemagglutination-inhibition titre of at least 40, compared with 57% of the MF59 group. The corresponding geometric mean titres were 103.4 (95% CI 78.7–135.9) and 29.0 (95% CI 22.4–37.6); the between-group comparisons were statistically significant (p<0.001).

The trial also reported a dose-sparing signal for AS03: within its AS03 groups, the measured response was not significantly different across the tested antigen doses. This is a signal, not evidence that the doses were equivalent; the study was not designed as an equivalence or non-inferiority comparison across antigen doses. One technical detail also matters for interpretation: subsequent SRID testing found that the administered haemagglutinin content was approximately 50% higher than the nominal values used to label the groups. The randomised comparison between the matched nominal-dose groups remains informative, but “15 µg” was a nominal label: SRID confirmed 22.5 µg of haemagglutinin in those vaccine formulations.

This is a striking within-study example of two related emulsion platforms yielding different antibody magnitudes. It is not evidence that AS03 would outperform MF59 for every antigen or vaccine. Nor did the trial measure prevention of H7N9 infection or disease: the HAI threshold is an immunogenicity endpoint, not a clinical efficacy result. Adjuvanted groups also reported more local reactions than unadjuvanted groups, underscoring that increased immune stimulation and reactogenicity need to be assessed together. [1]

Case 2 — RTS,S with AS01B or AS02A: an immune difference and a challenge result

The RTS,S malaria vaccine contains circumsporozoite-protein sequences fused to hepatitis B surface antigen. In a randomised, double-blind phase IIa study, 102 malaria-naïve adults received three 50-µg doses at months 0, 1 and 2, formulated with either AS01B or AS02A, before controlled Plasmodium falciparum challenge. The adjuvants share the immunostimulants MPL and QS-21, while their delivery platforms differ: AS01 is liposomal and AS02 is an oil-in-water emulsion.

The reported vaccine efficacy against infection after challenge was 50% for RTS,S/AS01B (95% CI 32.9–67.1) and 32% for RTS,S/AS02A (95% CI 17.6–47.6). The difference was not statistically significant (two-sided p=0.11). Before challenge, the AS01B group showed higher CSP-specific antibody and CD4+ T-cell responses, including IFN-γ-producing responses. These findings are consistent with an immunological difference between formulations, but the small study does not establish that the measured difference caused the numerical efficacy gap.

At rechallenge five months later, four of nine participants in each vaccine group remained completely protected. This small follow-up result offers no evidence of a persistence advantage for either formulation. The study is valuable because it connects a direct adjuvant comparison to a controlled protection endpoint, while showing why the numerical direction of a result must not be presented as proof of superiority. It supports further comparison; it does not settle the clinical question. [2]

Case 3 — M72 tuberculosis vaccine: a difference in Th1 CD4+ responses

An observer-blind phase I/II study evaluated M72, a recombinant tuberculosis antigen, with AS01 or AS02 in 110 adults who were naïve to Mycobacterium tuberculosis. The two principal adjuvanted M72 groups received the same 40-µg antigen dose on a two-dose, 0- and 1-month schedule. Smaller comparator groups received the related Mtb72F construct with AS02, M72 without adjuvant, or AS01 alone.

Both M72-adjuvanted formulations generated substantial antigen-specific antibody and polyfunctional CD4+ T-cell responses that persisted through follow-up. The highest CD4+ responses were observed with M72/AS01, and the authors reported significantly higher vaccine-specific Th1 CD4+ responses than with M72/AS02. Solicited adverse events were generally mild to moderate and transient; no vaccine-related serious adverse events were reported.

This gives the comparison a different emphasis from H7N9: the key reported contrast was cellular response, not an HAI titre or a protection endpoint. The matched antigen dose and schedule make the comparison informative, but the study was small and was designed to assess safety and immunogenicity. It cannot show that the higher CD4+ response would prevent more tuberculosis. A later efficacy result for an AS01-formulated M72 vaccine cannot be used as a head-to-head efficacy comparison against AS02, because that later trial did not randomise participants between the two adjuvants. [3]

Figure 3. Summary of the three human comparisons and the limits of their conclusions. Original Noulin BioConsulting illustration based on references 1–3. View full-size figure 3

What these comparisons mean for development

Together, the studies show that adjuvants from related formulation families can produce meaningfully different immune readouts with the same antigen. They also show that the consequences depend on the endpoint: AS03 and MF59 differed on H7N9 HAI titres; AS01 and AS02 differed on tuberculosis-specific CD4+ responses; and the RTS,S challenge trial observed a numerical efficacy difference that was not statistically significant.

They do not establish a universal ranking of adjuvants. The H7N9 result cannot be generalised to other antigens; the M72 cellular signal is not a protection correlate; and the RTS,S challenge result is not a field-efficacy comparison. Plausible explanations include differences in innate stimulation, antigen uptake and presentation, local recruitment of immune cells, and the timing or magnitude of downstream responses. These are mechanistic hypotheses unless measured directly in the study at issue.

The most useful development lesson is operational: treat the adjuvant as part of the vaccine candidate, not as a plug-in component assumed to be interchangeable. When comparing formulations, hold antigen identity and dose, route, schedule, population and assay methods constant wherever possible. Measure more than one dimension of immunity when the product profile requires it: binding and neutralising antibodies, antibody quality or breadth, relevant T-cell functions, durability and reactogenicity. Then test whether these readouts connect to protection in an appropriately designed challenge or field study.

The evidence also argues for restraint. In a separate randomised study of 180 children in Gabon, RTS,S/AS01E and RTS,S/AS02D produced similar anti-circumsporozoite antibody responses and similar safety profiles. That result is a useful reminder that related adjuvants can yield comparable responses in a different population and formulation context. [5]

Conclusion

Adjuvant choice can alter the magnitude and type of vaccine-induced immunity, even when the antigen is held constant and the platforms share broad features. But a higher titre, stronger Th1 signal or numerical challenge-efficacy difference is not automatically a better vaccine. The decision requires a defined target product profile, a direct and controlled comparison, and endpoints that link immune function to the protection the vaccine is meant to deliver.

For vaccine developers, the practical question is not simply whether an adjuvant “boosts” an antigen. It is which immune functions the formulation adds, for whom, for how long, at what tolerability cost, and with what evidence of clinical relevance.

References

1. Jackson LA, Campbell JD, Frey SE, et al. Effect of varying doses of a monovalent H7N9 influenza vaccine with and without AS03 and MF59 adjuvants on immune response: a randomized clinical trial. JAMA. 2015;314(3):237–246. doi:10.1001/jama.2015.7916 · PMID: 26197184.

2. Kester KE, et al. Randomized, double-blind, phase 2a trial of falciparum malaria vaccines RTS,S/AS01B and RTS,S/AS02A in malaria-naive adults: safety, efficacy, and immunologic associates of protection. J Infect Dis. 2009;200(3):337–346. doi:10.1086/600120 · PMID: 19569965.

3. Leroux-Roels I, et al. Improved CD4+ T cell responses to Mycobacterium tuberculosis in PPD-negative adults by M72/AS01 as compared to the M72/AS02 and Mtb72F/AS02 tuberculosis candidate vaccine formulations: a randomized trial. Vaccine. 2013;31(17):2196–2206. doi:10.1016/j.vaccine.2012.05.035 · PMID: 22643213.

4. Burny W, Callegaro A, Bechtold V, et al. Different adjuvants induce common innate pathways that are associated with enhanced adaptive responses against a model antigen in humans. Front Immunol. 2017;8:943. doi:10.3389/fimmu.2017.00943 · PMID: 28855902.

5. Lell B, Agnandji S, von Glasenapp I, et al. A randomized trial assessing the safety and immunogenicity of AS01 and AS02 adjuvanted RTS,S malaria vaccine candidates in children in Gabon. PLoS One. 2009;4(10):e7611. doi:10.1371/journal.pone.0007611 · PMID: 19859560.

6. Coccia M, et al. Cellular and molecular synergy in AS01-adjuvanted vaccines results in an early IFNγ response promoting vaccine immunogenicity. npj Vaccines. 2017;2:25. doi:10.1038/s41541-017-0027-3 · PMID: 29263880.

7. Schattgen SA, et al. Influenza vaccination stimulates maturation of the human T follicular helper cell response. Nat Immunol. 2024;25:1742–1753. doi:10.1038/s41590-024-01926-6 · PMID: 39164477.