Learning lessons from MVA85A, a failed booster vaccine for BCGBMJ 2018; 360 doi: https://doi.org/10.1136/bmj.k66 (Published 10 January 2018) Cite this as: BMJ 2018;360:k66
All rapid responses
13 August 2019
Since the correspondence (1), questioning the evidence for efficacy of the influenza vaccine MVA-NP+M1 claimed by Lillie et al, there have been some developments that should be recorded. I am grateful to Thomas Evans MD, CEO Vaccitech, for discussion and fact checking this letter.
1) The claim for efficacy in humans of the MVA-NP+M1 in the Lillie study that I challenged (1) has now been formally corrected in the form of an “erratum” published in the Journal Clinical Infectious Diseases, acknowledging that the p value quoted in the article was not correct (https://academic.oup.com/cid/article/69/1/195/5497806). The path to obtaining publication of this “erratum” was prolonged and tiresome: my observations were published in the BMJ on 23 January 2018, and it was not until 23 May 2019 that the “erratum” appeared in C.I.D. After a prolonged correspondence, and provision of independent reviews by a senior statistician invited by the Oxford Vaccine Oversight Committee, and the C.I.D statistics editor, a form of words was eventually found.
2) Despite being dismissive in their response to my request for performing a properly powered double blind challenge study with MVA-NP+M1 (1), Vaccitech have now been funded by BARDA $8.5 million for such a study conducted by SGS Life Sciences in Belgium (clinicalTrials.gov Identifier: NCT03883113). The objective pre-defined primary end point is viral shedding. BARDA's statistician and modelers will review the statistical plan before and after data analysis and unblinding. This should establish once and for all whether MVA-NP+M1 has any efficacy in humans as a “booster vaccine” for influenza.
3) The poorly designed “INVICTUS” study (ClinicalTrials.gov Identifier: NCT03300362), that I criticised (1) and referred to the Oxford Vaccine Oversight Committee because it combined standard subunit vaccination with MVA-NP+M1, but lacked any virological end-point, or specificity controls, has been “terminated” with the reason given as “change to recommended seasonal flu vaccine in UK”. This refers to the recommendation to upgrade from quadrivalent to MF59 adjuvanted trivalent subunit vaccine in those aged 65 and over (https://www.nhs.uk/conditions/vaccinations/flu-influenza-vaccine/).
4) Without waiting for the results of the challenge study NCT03883113 to establish whether MVA-NP+M1 has any efficacy, the INVICTUS study has been replaced by a similar but much larger phase 2b field study, involving 6,000 participants (NCT03880474) to be conducted in Australia, but this time at least with an objective primary end-point of “incidence rate of laboratory confirmed influenza using reverse transcription polymerase chain reaction (RT-PCR) on deep nasal swab samples to record confirmed cases of influenza.” This has improved the design, and steps have also been taken to administer the MVA-NP+M1 and the subunit vaccine in different deltoids with the aim of minimising the nonspecific adjuvant effect of MVA.
My view (1) is that this field trial should have been postponed until the evidence for efficacy of MVA-NP+M1 from the statistically sound challenge study NCT03883113 was available, before exposing another 1100 participants to this recombinant Pox Virus this year. At least the Patient Information Leaflet should have explained to volunteers that to date this vaccine has not shown any efficacy in humans, and minimal evidence in animals (1), that a similar design of vaccine failed to prevent TB (1, 2), and that if they wished they could postpone their decision to participate until the following influenza season, when the results of the efficacy study NCT03883113 should be available. I have repeatedly requested to see the Patient Information Leaflet for the field trial NCT03880474, but have been declined access by Thomas Evans MD, CEO at Vaccitech, on the basis that it is “proprietary” to the company.
(1) Rapid Responses to Editorial Learning lessons from MVA85A, a failed booster vaccine for BCG (https://www.bmj.com/content/360/bmj.k66/rapid-responses).
(2) Cochrane Database Syst Rev. 2019. MVA85A vaccine to enhance BCG for preventing tuberculosis. Kashangura R, Jullien S, Garner P, Johnson S.(https://www.ncbi.nlm.nih.gov/pubmed/31038197#)
Competing interests: I lead a group developing a single cycle influenza virus "S-FLU" as a broadly protective vaccine.
Monday 26 March 2018
Before responding to their clarifications I would like to establish that Gilbert and colleagues do not quibble with my conclusion that
“Booster” vaccination by i.m. injection with MVA-NP+M did not have a significant effect on either the duration or quantity of viral shedding after challenge with influenza in humans.
Given this tacit acknowledgement I now insist that the claim in the abstract to their paper (1) that
“there was a significant reduction in the number of days of virus shedding in those vaccinees who developed influenza (mean, 1.09 days in controls, 0.45 days in vaccines, P =. 036).”
should be formally retracted from the journal Clinical Infectious Diseases and replaced by an appropriate statement explaining that on review the data do not show a statistically significant reduction in symptoms or viral shedding.
In addition, all public references to MVA-NP+M as a “universal influenza vaccine” should be modified by explaining that this term is an aspiration, without at present a foundation in fact.
Finally, this paper should never again be referenced as evidence for efficacy of MVA-NP+M in any publication, or application to grant awarding agencies, or ethics or regulatory committees.
In answer to the clarifications provided by Gilbert and colleagues:
1. They claim “the vaccine (MVA-NP+M) was designed to boost T cell responses in humans with prior influenza A infection”. MVA-NP+M was not “designed” as a booster vaccine – it failed as a vaccine on its own, and “boosting” was all that was left for it. It was modelled on the recombinant Vaccinia viruses developed by Smith et al, and his suggestion in 1983 that “It also should be possible to produce more attenuated forms of vaccinia virus (i.e. MVA) that produce milder primary vaccination reactions” (2, 3). These recombinant vaccinia viruses were studied by Bennink and Yewdell (4, 5) for their ability to stimulate CTL responses more than 30 years ago. “Design” did not come into it.
2. I agree it is difficult to test MVA-NP+M as a “booster” in ferrets in this context, because influenza infection by itself provides robust heterotypic immunity lasting at least eighteen months (6). Recombinant Vaccinia vaccines expressing various influenza proteins failed to protect ferrets (7). By contrast, the M2SR single cycle influenza virus, under development by FluGen as a universal influenza vaccine, induced clear heterotypic immunity in ferrets (8).
3. The geometric mean virus calculation I quoted in my letter was as an illustration, because I could not include a diagram. The point is that 3 vaccinees and 5 controls shed virus. A comparison of the amount of virus shed by these two groups reveals no significant difference in the amount of virus shed per individual. There is no reduction in the vaccinated shedders. This is quite different to what is seen in animal models of T cell protection (for example (8)), and human challenge studies (9), where as a rule the T cell response reduced the amount of virus shed per individual by between ~10 fold- to undetectable. The T cell response does not provide an “all or nothing” protective effect. One would have to postulate that MVA-NP+M had an all or nothing effect in the Lillie study for the results to mean anything, which is biologically implausible.
4. I agree that in large placebo controlled randomised double blind trials or cohort studies, where symptomatic volunteers present themselves for a diagnostic PCR, it is of course standard to report symptoms. In the Lillie study the definition of “Laboratory Confirmed Infection” requiring symptoms was introduced post-hoc, and led to the inappropriate classification of a vaccinated volunteer who shed virus for two days as “not infected” because they were asymptomatic. Up to 75% of all influenza infections are asymptomatic (10), but that does not mean they can be ignored when it suits one's purpose.
5. I firmly disagree, given the lack of proven efficacy of MVA-NP+M, “that an appropriate way forward to assess the utility of MVA NP+M1 is to conduct a Proof of Concept study in the affected population”- i.e. the INVICTUS study. Gilbert and colleagues state that the “product development hypothesis tested in this trial is that both high antibody and T cell responses are more likely to provide enhanced protection.” This is based on a false premise, because the Lillie study shows that in fact the T cell responses induced by MVA-NP+M do NOT provide any significant protection in humans.
6. As presently designed, the INVICTUS study is testing two actions of the MVA-NP+M simultaneously: the action as an adjuvant, and the action of acting as a broadly protective influenza vaccine through the induction of broadly reactive T cells. It lacks a virological end-point, and lacks the specificity controls to distinguish these actions. The whole point of the MVA-NP+M is that it is supposed to be able to reduce viral shedding - something that can be detected with a simple and objective test. Why not use it? As only ~14% of upper respiratory illnesses are caused by influenza (11), the study will be beset by “noise”. Symptoms alone are fraught with difficulties in influenza - and open to placebo effects - which in this trial might arise because the MVA-NP+M injections are clearly more painful than the saline placebo (Antrobus 2013).
7. I believe it is not ethical to continue with this muddled trial without any evidence for efficacy of MVA-NP+M in humans. The evidence supporting MVA-NP+M as a “booster” vaccine should be bolstered by a properly designed and adequately powered human challenge study to establish once and for all whether it has any efficacy. This should be done BEFORE clouding the issue with mixtures of this unproved vaccine with TIV. In a properly powered study it may show some effect. Grant agencies would then be able to compare MVA-NP+M with other, perhaps more effective candidates, for funding. In their response, Gilbert and colleagues dismiss human challenge studies. This is not appropriate - FluGen are planning a 100-volunteer double blind challenge study to test their “universal vaccine” candidate M2SR (FluGen web site). FluGen has a similar level of funding to Vaccitech; if FluGen can afford to do this reliable test, so can Vaccitech.
(1) Lillie PJ, Berthoud TK, Powell TJ, Lambe T, Mullarkey C, Spencer AJ, et al. Preliminary assessment of the efficacy of a T-cell-based influenza vaccine, MVA-NP+M1, in humans. Clin Infect Dis 2012 Jul;55(1):19-25.
(2) Smith GL, Murphy BR, Moss B. Construction and characterization of an infectious vaccinia virus recombinant that expresses the influenza hemagglutinin gene and induces resistance to influenza virus infection in hamsters. Proc Natl Acad Sci U S A 1983 Dec;80(23):7155-7159.
(3) Smith GL, Levin JZ, Palese P, Moss B. Synthesis and cellular location of the ten influenza polypeptides individually expressed by recombinant vaccinia viruses. Virology 1987 Oct;160(2):336-345.
(4) Bennink JR, Yewdell JW, Smith GL, Moller C, Moss B. Recombinant vaccinia virus primes and stimulates influenza haemagglutinin-specific cytotoxic T cells. Nature 1984 Oct 11-17;311(5986):578-579.
(5) Yewdell JW, Bennink JR, Smith GL, Moss B. Influenza A virus nucleoprotein is a major target antigen for cross-reactive anti-influenza A virus cytotoxic T lymphocytes. Proc Natl Acad Sci U S A 1985 Mar;82(6):1785-1789.
(6) Yetter RA, Barber WH, Small PA,Jr. Heterotypic immunity to influenza in ferrets. Infect Immun 1980 Aug;29(2):650-653.
(7) Jakeman KJ, Smith H, Sweet C. Mechanism of immunity to influenza: maternal and passive neonatal protection following immunization of adult ferrets with a live vaccinia-influenza virus haemagglutinin recombinant but not with recombinants containing other influenza virus proteins. J Gen Virol 1989 Jun;70 ( Pt 6)(Pt 6):1523-1531.
(8) Hatta Y, Boltz D, Sarawar S, Kawaoka Y, Neumann G, Bilsel P. M2SR, a novel live influenza vaccine, protects mice and ferrets against highly pathogenic avian influenza. Vaccine 2017 Jul 24;35(33):4177-4183.
(9) McMichael AJ, Gotch FM, Noble GR, Beare PA. Cytotoxic T-cell immunity to influenza. N Engl J Med 1983 Jul 7;309(1):13-17.
(10) Hayward AC, Fragaszy EB, Bermingham A, Wang L, Copas A, Edmunds WJ, et al. Comparative community burden and severity of seasonal and pandemic influenza: results of the Flu Watch cohort study. Lancet Respir Med 2014 Jun;2(6):445-454.
(11) Hayward AC, Wang L, Goonetilleke N, Fragaszy EB, Bermingham A, Copas A, et al. Natural T Cell-mediated Protection against Seasonal and Pandemic Influenza. Results of the Flu Watch Cohort Study. Am J Respir Crit Care Med 2015 Jun 15;191(12):1422-1431.
(12) Antrobus RD, Berthoud TK, Mullarkey CE, Hoschler K, Coughlan L, Zambon M, et al. Coadministration of seasonal influenza vaccine and MVA-NP+M1 simultaneously achieves potent humoral and cell-mediated responses. Mol Ther 2014 Jan;22(1):233-238.
Competing interests: I have worked on the T cell response to Influenza since 1979, and am trying to develop our own version of a broadly protective attenuated influenza vaccine. I have stated this in the second paragraph of my first response.
The letter concerning the MVA-NP+M1 vaccine candidate by Townsend raises several points that we would like to clarify. The vaccine was designed to boost T cell responses in humans with prior influenza A infection, and this capacity of the vaccine has been shown convincingly in multiple studies to date [1-5]. Higher level T cell responses have been associated with a degree of protection in prospective assessments in humans [6, 7]. Thus, we are testing the hypothesis that boosting T cells to a level higher than those seen in the “naturally protected” population will also enhance protection using a safe and well-established immunization approach.
The ferret model is not validated for use in the setting of prior influenza exposure , and we do not know how to appropriately mimic the complexity and heterogeneity of the T cell responses observed in humans, which vary widely in magnitude and specificities, and which are acquired after a lifetime of repeated exposures to influenza viruses. In addition, the availability of truly standardized T cell reagents for use in ferrets remains limited .
The human challenge study  used as the primary endpoint safety, a secondary endpoint of immunogenicity, and a tertiary endpoint of the level of viral shedding as a measure of potential efficacy. Viral shedding is a standard measure of efficacy in influenza challenge studies. This was a small exploratory study to provide sufficient information to determine next steps. The geometric mean calculation presented by Townsend was not specified as a trial endpoint, as the geometric means for those with positive titres could be equivalent even if 2/100 vaccinated and 100/100 placebo were positive for viral shedding. It is standard in all challenge studies to report on laboratory confirmed illnesses , and almost all studies are powered with this endpoint. The level of viral shedding is reported for every day for every participant , although the statistical methods should have been included in the manuscript. The symptoms are shown, as is standard for such studies, and they are not significantly different between the two groups but trend towards a reduction in the vaccinated group. This pilot study confirmed the safety of boosting T cell responses in people who are subsequently exposed to influenza virus. The low take rate and symptom severity in the unvaccinated control group limited the ability to detect changes in clinical illness between groups, as has been observed in other recent challenge studies [10, 11].
We felt, and the Research Ethical Committee and Regulatory Agency agreed, that an appropriate way forward to assess the utility of MVA NP+M1 is to conduct a Proof of Concept study in the affected population, which continues to be poorly protected by present vaccines. In contrast, an appropriately powered challenge study would not provide information about protection against natural infection, would not address efficacy in the target elderly population, and would involve exposing over 100 healthy individuals to live influenza virus.
We agree that there is a possibility that MVA with no antigen insert, when co-administered with standard influenza vaccine, could boost antibody levels to surface proteins of influenza virus and provide some level of increased protection, but the product development hypothesis tested in this trial is that both high antibody and T cell responses are more likely to provide enhanced protection. We do not aim here to define which is the more important of these two mechanisms, although some immunological analyses may help to address this, and other trials are planned to address the efficacy of T cell induction alone. An initial Proof of Concept trial always tests the hypothesis with the lowest rate of a false negative, which is the case here, because any added efficacy contributed by enhanced antibodies to the potent T cell responses induced by MVA NP+M1 would increase power to detect an efficacy signal.
1. Antrobus, R.D., et al., Coadministration of seasonal influenza vaccine and MVA-NP+M1 simultaneously achieves potent humoral and cell-mediated responses. Mol Ther, 2014. 22(1): p. 233-8.
2. Antrobus, R.D., et al., Clinical assessment of a novel recombinant simian adenovirus ChAdOx1 as a vectored vaccine expressing conserved Influenza A antigens. Mol Ther, 2014. 22(3): p. 668-674.
3. Antrobus, R.D., et al., A T cell-inducing influenza vaccine for the elderly: safety and immunogenicity of MVA-NP+M1 in adults aged over 50 years. PLoS One, 2012. 7(10): p. e48322.
4. Berthoud, T.K., et al., Potent CD8+ T-cell immunogenicity in humans of a novel heterosubtypic influenza A vaccine, MVA-NP+M1. Clin Infect Dis, 2011. 52(1): p. 1-7.
5. Lillie, P.J., et al., Preliminary assessment of the efficacy of a T-cell-based influenza vaccine, MVA-NP+M1, in humans. Clin Infect Dis, 2012. 55(1): p. 19-25.
6. Hayward, A.C., et al., Natural T Cell-mediated Protection against Seasonal and Pandemic Influenza. Results of the Flu Watch Cohort Study. Am J Respir Crit Care Med, 2015. 191(12): p. 1422-31.
7. Sridhar, S., et al., Cellular immune correlates of protection against symptomatic pandemic influenza. Nat Med, 2013. 19(10): p. 1305-12.
8. Houser, K.V., et al., Impact of prior seasonal H3N2 influenza vaccination or infection on protection and transmission of emerging variants of influenza A(H3N2)v virus in ferrets. J Virol, 2013. 87(24): p. 13480-9.
9. Thangavel, R.R. and N.M. Bouvier, Animal models for influenza virus pathogenesis, transmission, and immunology. J Immunol Methods, 2014. 410: p. 60-79.
10. Memoli, M.J., et al., Validation of the wild-type influenza A human challenge model H1N1pdMIST: an A(H1N1)pdm09 dose-finding investigational new drug study. Clin Infect Dis, 2015. 60(5): p. 693-702.
11. Fullen, D.J., et al., Accelerating Influenza Research: Vaccines, Antivirals, Immunomodulators and Monoclonal Antibodies. The Manufacture of a New Wild-Type H3N2 Virus for the Human Viral Challenge Model. PLoS One, 2016. 11(1): p. e0145902.
Competing interests: AH and SG are co-founders of Vaccitech, a spin-out company from the University of Oxford which is developing the MVA-NP+M1 influenza vaccine. TE is the CEO of Vaccitech.
Re: Learning lessons from MVA85A, a failed booster vaccine for BCG: comparison to MVA-NP+M for Influenza
The recent discussion in the BMJ concerning the efficacy of MVA85A as a “booster” vaccine for TB has led me to question the evidence for the efficacy of the MVA-NP+M as a “booster” vaccine for Influenza. I find that the evidence is flawed.
To avoid accusations that I am trying to sabotage a competitor, I admit to having worked for nearly forty years on the T cell response to influenza, to have been deeply involved with identifying the antigens recognised and the mechanism of antigen presentation, and in trying to develop our own broadly protective attenuated vaccine.
MVA-NP-M is a genetically modified version of Modified Vaccinia Ankara containing a pair of conserved proteins from Influenza A called NP+M (similar in principle to the MVA-85A for TB). There is good evidence stretching back 40 years that T cells to the conserved proteins of influenza can be broadly protective in mice (1), and good epidemiological evidence that similar T cells in humans induced by prior infection with influenza virus, are associated with reduced viral shedding after experimental challenge (2), and a milder clinical course with reduced viral shedding in natural influenza infection (3,4). So, it was reasonable to ask if T cells induced by i.m. injection of MVA-NP+M could protect animals or humans from influenza. However, the result is not a forgone conclusion because i.m. injection of MVA-NP-M is very different to natural infection with influenza.
Studies in animals
The evidence for a protective effect in mice is minimal (5,6). In the first paper an MVA-NP on its own, after either one or two doses, did not protect mice, but was apparently protective after one or two priming doses of an NP DNA vaccine. However, the two-dose regime of the DNA vaccine was protective by itself, and the control of a single priming dose of the DNA vaccine alone is missing from the paper. In the second paper, dual immunisation with an Adenovirus-NP+M followed by boosting with MVA-NP+M failed to protect mice from challenge with the standard mouse adapted virus A/PR/8/1934 at the minimal dose of 25 pfu (6). The combination did prevent weight loss in mice after a non-lethal challenge with X31 or “H17” influenza viruses. However, the latter is almost non-pathogenic for mice (proper name E61-13-H17), and was introduced to demonstrate the specificity of T cells for NP, not as a challenge virus (7). There are no efficacy studies with MVA-NP+M in ferrets.
Studies in humans
The key study that is often quoted as evidence for efficacy in humans is the challenge study by Lillie et al (8). The original study protocol (Registration Number NCT00993083) defined a Phase IIa open label, non-placebo controlled, non-randomised controlled challenge study, which had as primary and secondary outcome measures "Safety of a new Influenza vaccine, MVA-NP+M1, when administered to healthy volunteers" and " Cellular immune response generated by a new influenza vaccine MVA- NP+M1 when administered as a single dose to healthy adults" respectively. Although challenge with influenza virus was mentioned in the protocol, there were no predefined outcomes other than safety. Outcome measures such as "symptom scores", “laboratory confirmed infection”, and "days of viral shedding" (but not total or peak viral shedding), were all selected post-hoc.
The authors state: “Symptoms were less pronounced in vaccinated subjects”. As the subjects were not blind to their vaccination status, and self-reported their own symptoms, a placebo effect is very likely. The self-reporting of symptoms by the subjects was mentioned only in the supplementary information. The fact that the study was neither randomised nor double blind, was not mentioned anywhere in the body of the paper or in supplementary information.
Days of Viral Shedding.
They claim: “there was a significant reduction in the number of days of virus shedding in those vaccinees who developed influenza (mean, 1.09 days in controls, 0.45 days in vaccines,
P =. 036).” A description of the method of calculation for this P value is missing from the paper and the supplementary information.
The claim depends on both the post-hoc definition of laboratory confirmed influenza involving symptoms, and a questionable statistical calculation. I have recalculated the p values obtained with the original contingency tables and for a table corrected for the asymptomatic vaccinee who shed virus, and find that the p value quoted can only be obtained by an unadjusted X^2 calculation, and a one tailed p value. On enquiry, this was confirmed as the method of calculation by the senior author. Either the inclusion of the asymptomatic vaccinee who shed virus in the “infected” group, or adjustment of the X^2 value for the small numbers in the study, results in a non-significant p value.
There is no justification for avoiding a correction of X^2 for the small numbers of subjects, for misclassifying a clearly infected vaccinee, or for quoting a one tailed p value without any predefined justification in the trial protocol. As the study endpoints were all selected post-hoc, it is not clear that any standard statistical test would be valid anyway.
The quantity of viral shedding.
The geometric mean virus shed per subject in total was 7.15 log10 TCID50 for vaccinees, and 7.2 log10 TCID50 for controls (calculated from the data). This does not suggest any reduction in viral shedding. In addition, the authors say in the discussion that there was no correlation between the level of T cells detected in the circulation induced by vaccination with MVA NP+M, and viral shedding. By contrast, in innumerable studies in mice and ferrets, effective T cell inducing vaccines reduce viral replication in the lungs of challenged animals, for example a mismatched LAIV (9).
“booster” vaccination by i.m. injection with MVA NP+M did not have a significant effect on either the duration or quantity of viral shedding after challenge with influenza in humans. This lack of effect is exactly what would be predicted from animal experiments testing recombinant vaccinia expressing NP as a stand-alone influenza vaccine in the 1980s, when this approach was shown not to work (for instance 10,11).
In the Patient Information Leaflet for the INVICTUS trial (NCT03300362), and on the BBC News, and on the Jenner and Vaccitech websites, the impression is given that MVA-NP-M is a “universal” influenza vaccine candidate with evidence for efficacy in humans through the induction of a cross-reactive T cell response. These statements are not justified by the available evidence and should be modified.
Before any further studies are done in humans with the MVA-NP+M vaccine, it should be tested in a rigorous efficacy trial in the best available animal model (ferrets) and the human challenge study should be repeated with properly pre-defined end points, randomisation and blinding. If these studies provided evidence for the cross-protective effect of MVA-NP+M as a vaccine, I would support the concept of the phase IIb INVICTUS trial as a logical extension. However, without this crucial evidence I believe that continuation of this trial in its present form is not ethical, because the MVA-NP+M is being given simultaneously with standard TIV seasonal vaccine without a control group of either MVA-NP+M alone or MVA + TIV. The lack of efficacy of the MVA-NP+M as a specific influenza vaccine will be masked by the well described adjuvant property of the MVA.
1. Yap KL, Ada GL, McKenzie IF. Transfer of specific cytotoxic T lymphocytes protects mice inoculated with influenza virus. A 1978 May 18;273(5659):238-239.
2. McMichael AJ, Gotch FM, Noble GR, Beare PA. Cytotoxic T-cell immunity to influenza. N Engl J Med 1983 Jul 7;309(1):13-17.
3. Sridhar S, Begom S, Bermingham A, Hoschler K, Adamson W, Carman W, et al. Cellular immune correlates of protection against symptomatic pandemic influenza. Nat Med 2013 Oct;19(10):1305-1312.
4. Hayward AC, Wang L, Goonetilleke N, Fragaszy EB, Bermingham A, Copas A, et al. Natural T Cell-mediated Protection against Seasonal and Pandemic Influenza. Results of the Flu Watch Cohort Study. Am J Respir Crit Care Med 2015 Jun 15;191(12):1422-1431.
5. Degano P, Schneider J, Hannan CM, Gilbert SC, Hill AV. Gene gun intradermal DNA immunization followed by boosting with modified vaccinia virus Ankara: enhanced CD8+ T cell immunogenicity and protective efficacy in the influenza and malaria models. Vaccine 1999 Nov 12;18(7-8):623-632.
6. Lambe T, Carey JB, Li Y, Spencer AJ, van Laarhoven A, Mullarkey CE, et al. Immunity against heterosubtypic influenza virus induced by adenovirus and MVA expressing nucleoprotein and matrix protein-1. Sci Rep 2013;3:1443.
7. Townsend AR, Skehel JJ. The influenza A virus nucleoprotein gene controls the induction of both subtype specific and cross-reactive cytotoxic T cells. J Exp Med 1984 Aug 1;160(2):552-563.
8. Lillie PJ, Berthoud TK, Powell TJ, Lambe T, Mullarkey C, Spencer AJ, et al. Preliminary assessment of the efficacy of a T-cell-based influenza vaccine, MVA-NP+M1, in humans. Clin Infect Dis 2012 Jul;55(1):19-25.
9. Baz M, Boonnak K, Paskel M, Santos C, Powell T, Townsend A, et al. Non-replicating influenza A vaccines confer broad protection against lethal challenge. mBio 2015;6:e01487-e01500.
10. Andrew ME, Coupar BE, Boyle DB, Ada GL. The roles of influenza virus haemagglutinin and nucleoprotein in protection: analysis using vaccinia virus recombinants. Scand J Immunol 1987 Jan;25(1):21-28.
11. Lawson CM, Bennink JR, Restifo NP, Yewdell JW, Murphy BR. Primary pulmonary cytotoxic T lymphocytes induced by immunization with a vaccinia virus recombinant expressing influenza A virus nucleoprotein peptide do not protect mice against challenge. J Virol 1994 Jun;68(6):3505-3511.
Competing interests: I have worked on the T cell response to Influenza since 1979, and am trying to develop our own version of a broadly protective attenuated influenza vaccine. I have stated this in the second paragraph of my response.