Expert consensus on the benefits of neuraminidase in conventional influenza vaccines: a Delphi study

Abstract

Background

Seasonal vaccination is the mainstay of human influenza prevention. Licensed influenza vaccines are regularly updated to account for viral mutations and antigenic drift and are standardised for their haemagglutinin content. However, vaccine effectiveness remains suboptimal. Neuraminidase (NA) evolves more gradually than hemagglutinin and has been demonstrated to provide added clinical benefits. However, NA is not currently a mandated or standardised component of influenza vaccines.

Methods

Here, we collated expert opinions on the importance of NA in influenza vaccines in a two-stage Delphi survey. Nine statements about NA were formulated by a steering committee based on a targeted literature review. In the survey’s first round, panellists recruited from three continents were requested to report on their agreement with each statement and estimate the strength of evidence for each statement. Panellists were also requested to explain their choice of answer and suggest revisions to the statements. Consensus was considered reached if ≥ 75% of panellists agreed with a statement. If consensus was not reached for a statement, this statement was revised and included in the survey’s second round.

Results

Nine panellists with a broad range of NA-related expertise, including clinical, research, and public health experience, completed the survey. They agreed that anti-NA responses acquired via natural infection or vaccination are associated with protective immunity independently of haemagglutinin and that NA provided additional advantages including improving disease severity metrics. The experts identified several knowledge gaps concerning heterologous cross-reactivity of vaccine-induced anti-NA antibodies, correlations between anti-NA titres and reduced transmission or infection risks, and differences in anti-NA responses to seasonal influenza vaccines.

Conclusions

NA is an important influenza vaccine component and is associated with specific benefits. These benefits would likely be greater if NA content were standardised. Additional research is needed to optimise vaccines for anti-NA effects.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-024-10277-4.

Background

Influenza is an acute respiratory disease caused by human influenza A, B, and C viruses. Annual influenza epidemics are estimated to account for 1 billion infections, 3 to 5 million cases of severe illness, and up to 650,000 respiratory deaths per year [1, 2].

Seasonal vaccination is a long-established mainstay of human influenza prevention. To date, vaccine development has typically focused on eliciting immune responses to the viral receptor-binding and membrane fusion glycoprotein haemagglutinin (HA). Licensed inactivated and recombinant vaccines are thus currently standardised for their HA content [3]. To account for the antigenic drift and shift of influenza viruses, the World Health Organization (WHO) makes biannual surveillance-based recommendations for the strain to be included in seasonal influenza vaccines (two ‘influenza seasons’ occur per year: one in the Northern and one in the Southern Hemisphere) [4]. However, because of continuous changes in circulating influenza strains, the effectiveness of these HA-focused vaccines is often suboptimal [5], and vaccines that can elicit stronger and broader protection are needed [6].

Another influenza virus surface glycoprotein, neuraminidase (NA), has attracted attention as a potential complementary standardised component of next-generation influenza vaccines, as it experiences slower antigenic drift and can evolve discordantly to HA [3, 7–12]. NA is one of three membrane proteins expressed on human influenza A and B virions, is critical to viral replication and transmission, and is an important target of the host immune system [10]. Natural human influenza virus infections induce substantive, long-lasting homologous and cross-reactive anti-NA responses [13–15], even without corresponding anti-HA responses [16]. Anti-NA antibodies to previously circulating strains may thus provide long-term protection against newer strains [17, 18]. Anti-NA antibodies are considered an independent predictor of various disease severity metrics. Increased anti-NA titres have been associated with reduced viral shedding [19, 20] and lower infection rates [9, 21]. Further, individuals with higher vaccine-induced or naturally-acquired baseline anti-NA titres are less likely to experience symptomatic influenza disease [22, 23], severe symptoms [24, 25], or prolonged illness [20, 26].

Whereas experimental NA-specific vaccines elicit protective anti-NA responses in a dose-dependent manner [22, 27, 28], conventional vaccines are not standardised to NA content [29] and vary widely in their NA immunogenicity [30–34]. Currently licensed recombinant vaccines do not include NA at all [35]. Vaccines with no NA cannot elicit robust anti-NA responses and may provide less protection than those containing meaningful quantities of NA. While some mRNA-based seasonal influenza vaccines currently under clinical investigation encode only membrane-bound HA [36], numerous groups and companies are currently investigating mRNA vaccination with multiple influenza antigens [37–40].

Until recently, large-scale, accurate assessment of NA content of influenza vaccines has been technically challenging [41]. Advances that now enable accurate measurement of NA content and immunogenicity [9, 11] have sparked renewed interest in NA as a vaccine component [3, 9–12]. However, there is currently no consensus on whether NA should be included and standardised in influenza vaccines. Here, using Delphi methodology [42], we examined the opinions of experts about the benefits of NA in influenza vaccines. Our aim was to determine the level of consensus on benefits of including NA as a standardised component of seasonal vaccines and to stimulate further global research and informed debate amongst vaccine developers about the potential of NA as a vaccine component.

Methods

Study design

The Delphi study was conducted between June and November 2023. It was not registered prospectively. The Delphi method, which assumes that group judgments are more valid than individual judgments, is a widely used structured communication technique for achieving convergence of opinion from a panel of experts [42]. Here, the procedure involved a two-stage survey of expert opinion on the importance NA in current influenza vaccines (Fig. 1). This article complies with the ACCORD reporting guideline for consensus methods in biomedicine [43].

Fig. 1.

Delphi methodology flow diagram. The Delphi survey was run in two rounds. In the first round, panellists were asked to report on the extent to which they agreed with nine survey statements and to estimate the strength of evidence supporting each statement. Panellists were also requested to explain their choice of answer and suggest statement revisions. After the survey’s first round, percentages of panellists agreeing with each statement were calculated; consensus was considered reached if ≥ 75% of panellists agreed with a statement. If consensus was not reached for a statement, this statement was revised and included in the survey’s second round. Percentages of panellists agreeing with the revised statement(s) were again calculated and consensus was considered reached at ≥ 75%. Abbreviations: NA, neuraminidase

Participant selection

The consensus exercise was directed by a team of scientists (led by PS and MH) from Evidera, a company contracted by the sponsor to conduct the study. The Evidera team invited 37 influenza experts from North America, Australia, and Europe to act as panellists via email. To participate, panellists had to have a background and expertise in public health, immunology, vaccinology, or virology, and to be an author of peer-reviewed publications or a member of relevant organisations. The experts signed contracts detailing the way they would support the study and the time needed for the study, and were compensated for their time based on a predetermined country-specific rate. Two experts (JPB and MM) acted as lead experts and served in a scientific steering committee together with the Evidera team.

Development of survey statements

The contents of the Delphi survey were based on key topics previously identified in a targeted literature review (MEDLINE and Embase searches performed in June 2022) and discussions with the two lead experts. The targeted literature review identified and synthesised evidence on clinical benefits from human NA exposure through natural infection, challenge studies, or vaccination; animal studies were excluded (Supplementary Materials: Table S1, Fig. S1). Based on the targeted literature review, the steering committee developed nine statements to serve as the “backbone” of the Delphi survey; these were the statements on which consent from the full expert panel was later sought (Table 1). The steering committee was further responsible for any revision of the statements following the survey’s first round.

Table 1.

Key topics and statements developed for Delphi survey

Key topics	Nine statements developed for Delphi survey
Anti-NA antibody responses that are associated with natural influenza infection	1. Natural infection is associated with homologous and cross-reactive anti-NA response. 2. Anti-NA antibodies are associated with protective responses independently of neutralizing antibodies to HA.
Anti-NA antibody responses that are associated with influenza vaccination	3. NA-containing influenza vaccines can elicit anti-NA responses. 4. There is limited evidence on the cross-reactivity of vaccine-induced anti-NA antibodies against heterologous influenza strains.
Variation in the anti-NA responses to vaccination	5. There are limited data on how specific differences in seasonal influenza vaccines (e.g., vaccine type, formulation, or content) affect anti-NA responses.
Relationship between anti-NA antibody response and influenza infection/disease	6. Compared to individuals with lower anti-NA titres, those with higher anti-NA titres were less likely to develop symptomatic infection and/or experienced shorter or less severe illness. 7. There is suggestive evidence that anti-NA antibodies acquired through natural infection or vaccination may confer protection against heterologous strains of the same NA subtype. 8. There is no direct evidence on whether pre-infection anti-NA titres correlate with a reduced risk of transmitting influenza. 9. Anti-NA titres are a useful correlate of protection against infection.

Abbreviations: HA Haemagglutinin, NA Neuraminidase

Delphi survey rounds

The Delphi survey was run in two rounds. The questionnaires for both rounds are provided as Supplementary Materials (ADD-NA Statements Round 1 and 2). In the first round, hosted online between June 20 and July 3, 2023, the panellists were assigned random identifiers and were required to provide information pertaining to their demographic details, primary area of expertise (NA expert, vaccine expert, virologist, pharmacologist, other), time spent working in this area of expertise, percentage of time spent in various domains (direct patient care, research, teaching, other), and the setting of their primary practice or research (private practice, academic teaching hospital, non-academic hospital, other). They were then presented with the nine statements previously developed with the lead experts and were asked to anonymously report on the extent to which they agreed with each of the nine statements on a five-point Likert scale (strongly disagree; disagree; undecided; agree with reservation; strongly agree). They were further invited to estimate the strength of evidence (very low – very high) supporting each statement. All experts were also requested to provide free text responses on the reasoning for their choice of answer, as well as suggestions for revising the statements.

All answers were processed after the survey’s first round and the percentages of panellists voting for each category were calculated. Consensus was considered reached if ≥ 75% of panellists (n = 7/9) agreed, either strongly or with reservation, with a statement. If consensus was not reached for a statement, these results were discussed with the lead experts and the statement was revised.

In the survey’s second round, hosted between November 1 and November 28, 2023, all experts were again presented with statement 9, the only statement for which consensus had not been achieved after the first round, and which had been revised to read “There is currently insufficient evidence to conclude that anti-NA antibodies titres correlate with protection against influenza virus infection.” Consensus or non-consensus for this statement was again calculated (as described above).

Results

Delphi panel composition

The full Delphi study panel consisted of nine experts from the USA, Canada, Australia, Spain, and the Netherlands (Table S2). Four panellists identified as vaccine experts, two each as NA experts or virologists, and one as an immunologist, having worked in their areas of research for 17 to 65 years (mean 34 years) (Table 2). All devoted ≥ 50% of their professional time to research. The panellists worked in academia, in a non-academic hospital, for various public health research institutes, and for the WHO’s Global Influenza Surveillance and Response System.

Table 2.

Key demographics relating to the expertise of the nine Delphi panellists

Expert ID	Primary area of expertise	Time spent working in area of expertise (years)		Percentage of professional time spent in			Setting of primary practice/research
				Research	Teaching	Other
1	Vaccine expert		30	100	0	0	Public health research institute
2	Vaccine expert		55	90	0	10 (Training)	Public health
3	Immunologist		30	90	10	0	Research institute
4	Vaccine expert		17	80	5	15 (Administration)	University
5	Vaccine expert		65	50	50	0	Academic teaching hospital
6	Virologist		35	50	40	10 (Advising)	Flu and respiratory virus surveillance at NIC (GIRS/WHO)
7	NA expert		35	90	10	0	Research institute
8	NA expert		15	80	10	10 (Management)	Veterinary faculty
9	Virologist		24	90	10	0	Non-academic teaching hospital

Abbreviations: ID Identifier, NA Neuraminidase, NIC (GIRS/WHO) National Influenza Centre (Global Influenza Surveillance and Response System/World Health Organization)

Delphi survey results

In the survey’s first round, consensus was reached for eight statements (statements 1–8). For six of these statements, all nine experts indicated their agreement (exceptions: statements 4, 8, and 9). All but one expert agreed with the statement that there is limited evidence on the cross-reactivity of vaccine-induced anti-NA antibodies against heterologous influenza strains (statement 4). The expert who disagreed with statement 4 commented that “vaccination with adjuvanted recombinant NA induces broad heterologous, but not heterosubtypic, cross-protection against influenza virus infection in mice”. In support of this statement, the expert referred to a publication by Wohlbold et al. [29], which describes the use of recombinant NA to assess NA-based immunity in mice (the baculovirus-expressed NA immunogens used in these experiments constitute neither a licensed product nor one in late-phase clinical development).

All but one expert agreed with the statement that there is no direct evidence on whether pre-infection anti-NA titres correlate with a reduced risk of transmitting influenza (statement 8). The dissenting expert commented that “there is direct evidence in animal models for such a correlation,” without providing further support for this claim.

Two experts judged the strength of evidence to be very high for statements 2 and 3 (Fig. 2). These were the only statements where any of the experts judged the strength of evidence to be very high. Most experts judged the strength of evidence to be moderate or higher for statements 1 to 5 (n = 8/9 experts) and statements 7 and 8 (n = 7/9), but only four did so for statement 6. Furthermore, five experts considered the strength of the evidence supporting statement 6 to be low.

Fig. 2.

Extent of agreement with and estimated strength of evidence for statements included in the Delphi survey’s first round. All nine Delphi panellists (9/9; 100%) reported their extent of consent (strongly agree – strongly disagree; top left panel) with nine statements (right panel). They further estimated the strength of evidence (very low – very high; bottom left panel) supporting each statement. Percentages of panellists voting for each category are reported

Consensus was not reached for the statement that anti-NA titres are a useful correlate of protection against infection (statement 9): two experts disagreed, two were undecided, four agreed with reservations, and only one respondent agreed strongly. One of the experts who disagreed stated that “NA-inhibiting antibodies are negatively associated with illness after infection, but not with infection per se.” The other cited “insufficient clinical data” as their reason for disagreeing with statement 9. One undecided expert was unsure whether currently available data support statement 9. The other undecided expert provided more detail, stating that influenza virus infection does not equate to clinical influenza and further explaining that anti-NA antibodies, while proven to hinder viral spread and clinical progression to some extent, have not definitively been proven to prevent infection through the nasopharynx. This expert was furthermore unaware of papers that clearly establish the threshold of an individual’s anti-NA antibody level for protection against mild or severe clinical disease. While two experts judged the strength of evidence supporting statement 9 to be high, six considered it to be either moderate or low (n = 3/9 per category), and one expert judged it to be very low. Statement 9 was reformulated and included in the Delphi survey’s second round.

In the survey’s second round, consensus was reached for the revised statement 9. All but one expert agreed that there is currently insufficient evidence to conclude that anti-NA titres correlate with protection against influenza virus infection (Fig. 3). The expert who disagreed with this statement reasoned that, in opposition to previously prevailing opinion, several results from a recent study indicate that anti-NA titres protect against influenza infection [44]. One other expert also suggested referring to recent studies, while a third commented that “insufficient evidence might be better contextualized if limited evidence indicates protection from illness occurs.” A fourth expert, while agreeing with statement 9, added the caveat that NA antibodies probably reduce virulence of infection.

Fig. 3.

Extent of agreement with the revised statement included in the Delphi survey’s second round. The nine Delphi panellists (9/9; 100%) reported their extent of consent (strongly agree – strongly disagree) with the revised statement 9. Percentages of panellists voting for each category are reported

Discussion

NA has been a non-standardised component of influenza vaccines since egg-based vaccines have been in use. It provides varied benefits as an immune antigen that complement those of HA [7, 8, 10, 19–21, 24, 25]. However, despite the advantages of anti-NA immunity, NA remains non-standardised in licensed vaccines (some do not include NA at all), its independent benefits have not been studied in large, randomized control trials, and there are limited data on how specific differences in seasonal influenza vaccines (e.g., vaccine type, formulation, or content) affect anti-NA responses (see also statement 5) [35].

Recent technical advances enabling accurate NA titration [9, 11] have sparked renewed interest in broadening the immune response of influenza vaccines via a complementary, standardised NA component. Against this backdrop, we aimed to investigate expert opinions on the benefits of NA in the context of current influenza vaccines and raise awareness of the advantages of its continued inclusion and standardisation.

The panel of nine experts achieved consensus on the importance of NA in vaccines; they determined several current knowledge gaps and, in doing so, identified targets for further global research and informed debate. The experts agreed that natural infection is associated with homologous and cross-reactive anti-NA responses (statement 1), that anti-NA antibodies are associated with protective responses independently of HA (statement 2), and that NA-containing influenza vaccines can elicit anti-NA responses (statement 3). They further concurred that anti-NA antibodies (acquired through natural infection or vaccination) may confer heterologous immunity (statement 7) and decrease disease severity and duration (statement 6). This consensus is reinforced by research on the antigenicity of NA [13–15, 17, 18, 45] and heterogenous retrospective studies showing that increased anti-NA titres are associated with improved disease severity metrics, including decreases in viral shedding [19, 20], infection rates [16, 21], and symptom severity [21, 24, 25]. It is further supported by multiple clinical trials (spanning nearly half a century) on the anti-NA response associated with conventional [32–34, 46] and NA-specific experimental [22, 27–29, 47] influenza vaccines.

While the elicitation of NA-specific responses via vaccination remains undisputed (most conventional vaccines prompt anti-NA titre increases, albeit at different levels), the experts identified several important knowledge gaps. Notably, most of them agreed that evidence of heterologous cross-reactivity of vaccine-induced anti-NA antibodies is currently limited (statement 4). Indeed, only few (contradictory) studies have examined heterologous anti-NA protection in humans. Frobert et al. demonstrated H5N1 cross-protection in patients born between 1930 and 1950 but were unable to confirm cross-protection in the sera of vaccinated individuals [48]. Newer studies have shown that both naturally-acquired [13] and vaccine-induced [33] NA antibodies are broadly reactive.

The expert panel further highlighted a dearth of data on how specific differences in seasonal influenza vaccines affect anti-NA responses (statement 5). Vaccine type e.g., appears to be an unreliable predictor of NA-immune responses: while inactivated vaccines typically elicit stronger systemic anti-NA responses than live attenuated ones [16, 32, 34, 46], live attenuated vaccines have also been associated with high NA immunogenicity [49]. Due to lack of standardisation, NA content not only varies by vaccine type, but also by manufacturer between seasons [50]. Further, while a dose-dependent relationship between vaccine NA content and anti-NA response has been suggested [30], this correlation is not always directly proportional [28] and remains to be further elucidated.

Most panellists judged evidence to be lacking on whether pre-infection anti-NA titres correlate with protection against influenza infection or reduce transmission risk (statements 8 and 9). Again, this consensus reflects pertinent literature. Several multivariate analyses have shown that naturally-acquired anti-NA titres act as independent predictors of immunity and protection from severe clinical disease, but not of influenza transmission [19, 21, 24, 25]; fewer studies have investigated the protective efficacy of anti-NA antibodies acquired through vaccination [16, 22, 34].

Recent literature reviews have highlighted the increased attention paid to NA in the field of influenza vaccine development [9–11, 51, 52]. mRNA therapeutic platforms, first granted a WHO Emergency Use Listing during the COVID-19 pandemic, are increasingly explored to address other pathogens [53] and support multi-antigen (HA and NA) influenza vaccine formulations [47, 54]. However, despite increased interest in and feasibility of NA standardisation, no general agreement has yet been reached regarding NA’s potential as a vaccine component and whether future influenza vaccines should be required to include NA.

Consensus obtained from a group of experts using formal methods is recognised as a reliable approach to develop best practice guidance [42, 43]. Here, we leveraged robust Delphi methodology to examine the opinions of experts on the importance of NA in influenza vaccines. The high level of expert agreement (only one of nine statements was submitted to a second round) supports NA’s immunological relevance. The inclusion of experts with long-standing expertise in various domains and from different geographical regions (three continents are represented) is a clear strength of this study. The experts based their responses in the two Delphi survey rounds on their expertise alone and, owing to the anonymous set-up of the online questionnaire, were unlikely to be influenced by their peers in their decision-making (potential non-anonymised interactions outside of the Delphi panel context cannot, of course, be excluded).

Limitations of the study pertain to the small number of participants, likely bias in subject-specific expert selection, and limited panel diversity. Although invitations were sent to 37 potential panellists, only nine experts agreed to serve on the Delphi panel. While there are no generally agreed standards for panel size, panellist numbers are recognised to impact the quality of Delphi panel recommendations [55]. The fact that consensus was already reached on most statements in the survey’s first round indicates a general homogeneity of opinion amongst NA and vaccine experts on the importance of NA in influenza prevention. The small panel size per se is hence unlikely to have significantly impacted our findings. Perhaps more importantly, the panel lacked experts from South America, Asia, and Africa, and our findings may thus not necessarily represent the opinions of experts with experience of different cultures and health systems.

Conclusions

This timely Delphi study consolidated expert opinions on the NA-specific benefits in influenza vaccines. The experts agreed that licensed vaccines including a non-standardised amount of NA are likely to provide a benefit by eliciting independent anti-NA responses. NA-specific advantages include reducing influenza virus shedding and spread, limiting symptom duration and severity, and inducing a response to a more conserved target than HA. These benefits would probably be greater if the NA content were standardised. Despite extensive past research into NA, several knowledge gaps are yet to be closed. We aimed to highlight this topic and advocate for future vaccine research to include endpoints that specifically quantify the independent benefits of NA. The influence of vaccine type, formulation, and dose on anti-NA responses – and ultimately on vaccine efficacy – must specifically be explored. Research and informed debate are required on how development, standardisation, and regulation of conventional vaccines might need to change to optimise them for anti-NA effects without compromising their traditional efficacy.

Authors’ contributions

John Youhanna: conceptualization, methodology, data curation, writing – review and editing, visualization, supervision, funding acquisition; Joan Puig-Barberà: conceptualization, validation, investigation, writing – original draft, writing – review and editing; Matthew S. Miller: conceptualization, validation, investigation, writing – original draft, writing – review and editing; Deborah Molrine: conceptualization, writing – review and editing; Monica Hadi: conceptualization, methodology, formal analysis, investigation, writing - original draft, writing - review and editing, funding acquisition; Shweta Bapat: conceptualization, methodology, data curation, writing – review and editing, visualization, project administration; Ike Iheanacho: conceptualization, methodology, validation, investigation, writing – review and editing, funding acquisition; Sophie Dodman: conceptualization, investigation, writing – review and editing, funding acquisition; Tsion Fikre: methodology, data curation, writing – review and editing, visualization, project administration; Paul Swinburn: conceptualization, methodology, investigation, writing – original draft, writing – review and editing, supervision, project administration, funding acquisition. All authors were involved in the critical review and editing of the manuscript for intellectual content. All authors approved the final version of the manuscript and agree to be accountable for the work.

Funding

This study was supported by Seqirus.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Institutional Review Board approval was not necessary according to Title 45 of the Code of Federal Regulations (CFR) Part 46, of the Health and Human Services (HHS) in the US. All study participants provided written informed consent.

Consent for publication

Not applicable.

Competing interests

Potential conflicts of interest are as follows: John Youhanna and Deborah Molrine are employees of Seqirus which funded this study. Monica Hadi, Shweta Bapat, Ike Iheanacho, Sophie Dodman, Tsion Fikre, and Paul Swinburn are employees of Evidera, which received financial support from Seqirus for work on this study. Juan Puig-Barbera has received honoraria for consultancy work from Novavax, Seqirus/CSL, HIPRA, and Pfizer. Additionally, he received payment for his participation in the current Delphi panel. Matthew Miller has received honoraria for consultancy work from Seqirus, Sanofi, Evidera, and Grifols, and research funding from Pfizer and Providence Therapeutics. Matthew Miller is a co-founder of Aeroimmune Inc. Stanley Plotkin has received consulting remuneration from Merck, Moderna, Novavax, BNT, Ntx, Valneva, Vaxart, Sanofi, and GSK.