Influenza vaccines and immunopathology

Keywords::

• immunopathology
• influenza
• narcolepsy
• vaccination

In order to defend us from infections with viruses such as influenza A, the immune system has to perform a difficult balancing act in a very short space of time. Peak influenza symptoms occur 48 h after infection. To prevent severe disease, the immune system may deploy virus-neutralizing antibodies to mop up free virions before they can infect cells of the host’s respiratory tract, and cytotoxic T cells to destroy cells that have already been infected before they release a further round of free virions capable of infecting more cells. In some rather specific circumstances, this destruction of infected cells can lead to immunopathology. There has been some expression of concern that influenza vaccines designed to increase T-cell immunity to influenza may, therefore, result in immunopathology once the vaccinated subjects are exposed to influenza virus [1].

These concerns were first prompted by the unusual W-shaped mortality curve seen in the 1918 influenza pandemic. In the USA 1918, there was a high mortality rate in those aged 18–39 years as well as in the very young and the very old. It is hypothesized that older adults had previously been exposed to an influenza virus that induced antibodies with some cross-reactivity to the 1918 virus, providing protection, but those aged 18–39 years had been exposed only to viruses of a different subtype. They therefore lacked virus-neutralizing antibodies, but possessed a strong cytotoxic T-cell response that caused immunopathology and predisposed the individual to secondary bacterial infection which was the major cause of death, rather than death resulting from influenza pneumonia [2]. However, this is only one way of interpreting the available data. More recent surveys of T-cell responses to influenza in children have found that strong T-cell responses are generally present within the first decade of life [3,4], which is difficult to reconcile with the low mortality rate in those aged 5–15 years in 1918. Other interpretations of the data from 1918 regard these cross-reactive T-cell responses as protective, albeit only for short periods of time following the last prepandemic influenza virus exposure. The transient nature of these responses have been described [5], resulting in the waves of infection observed in influenza pandemics [6], as those who are exposed but protected by naturally acquired immunity during the first wave lose their immunity and become susceptible again by the time of the second wave.

In 2009, cross-reactive antibodies acquired during early life appear to have had a role in protecting those born before 1950 [7], but younger individuals were also protected if they had recently been infected with seasonal influenza, and a high level of immunity was noted despite low levels of antibody [8]. The requirement for recent influenza infection plus discrepancy between antibody levels and protection again points to a protective role for cross-reactive T-cell responses.

Animal models of influenza can provide an opportunity to manipulate immune responses and examine their role in protection or immunopathology. There are numerous examples of induction of protective T-cell responses to influenza nucleoprotein (NP) in particular, in the absence of antibodies to other influenza antigens [9–11]. The partially protective effect of cross-reactive T cells induced by seasonal influenza infection leading to reduced mortality following H5N1 infection has also been modeled [12,13]. However, there are two studies in which immunopathology does appear to have been induced, both following very high dose administration of influenza virus. One study used mice expressing a transgenic T cell receptor (F5) capable of recognizing an NP epitope, which was present on 90% of the T cells [14]. The mice were also RAG-1^-/-, and therefore lacking all other B and T cells. Perhaps not surprisingly, influenza infection of all doses of influenza virus from 10² to 10⁷ PFU was lethal in RAG-1^-/- mice. F5RAG-1^-/- mice were protected by their transgenic T cells against doses of up to 10⁵ PFU, but at the highest challenge dose (10⁷ PFU), immunopathology resulted.

In a study in pigs receiving a DNA vaccine designed to induce antibodies against M2e and T cells against NP, following aerosol administration of 10⁸ TCID₅₀ influenza virus, vaccinated pigs fared worse than the control animals [15]. Bronchoalveolar lavage was performed 4 days prior to influenza challenge as well as at multiple time points after challenge. Three out of six vaccinated animals died, two of them during the lavage procedure. While these animals had a fully functioning immune system, bronchoalveolar lavage pre- and post-challenge may have contributed to lung damage or inflammation following a very high dose virus challenge delivered to the lungs.

In naturally transmitted influenza infections, the infectious dose is known to be very low, except in the particular case of H5N1 infections in humans. Since the H5N1 virus infects via a cellular receptor that is chiefly found in the lower, rather than upper, respiratory tract, only very high virus exposure results in infection, and pneumonia is the immediate result of infection. Controlled human challenge studies of seasonal influenza virus employ high doses delivered intranasally, and no evidence of immunopathology has been found even when T-cell responses have been deliberately boosted by vaccination [16]. Rather, there has been a protective effect of naturally acquired T-cell immunity following deliberate high-dose challenge [17].

So, the concern over immunopathology resulting from T-cell responses to NP or other conserved antigens in the absence of neutralizing antibodies in humans can be narrowed to high-dose infections with some pandemic viruses, although both in 1918 and 2009, there is evidence that T-cell responses were more likely to be responsible for protection than increasing disease severity. Furthermore, novel influenza vaccine research is moving toward the induction of both protective T cells and neutralizing antibodies with the aim of inducing the highest possible level of immune protection against all influenza A viruses [18] [Gilbert SC. Advances in the Development of Universal Influenza Vaccines (2012), Submitted]. The theoretical concerns over the safety of increasing cellular immune responses to influenza do not present a barrier to clinical testing of improved influenza vaccines, for which there is a great need.

A second manifestation of potential immunopathology following influenza vaccination has now raised its head in the wake of the 2009 pandemic. An increase in the number of cases of narcolepsy following vaccination of children with Pandemrix, an adjuvanted pandemic H1N1 vaccine, was reported in a number of northern European countries, especially Finland [19] and Sweden [20]. As reviewed by Han, narcolepsy is characterized by excessive daytime sleepiness and sleep paralysis [21]. It is an autoimmune disorder strongly associated with HLA DQB1*06:02, resulting from neuronal degradation in the lateral hypothalamus. Loss of hypocretin-producing neurons results in low levels of hypocretin-1 in the cerebrospinal fluid, and measurement of this is now recognized as diagnostic for narcolepsy, along with the presence of HLA DQB1*06:02.

Prior to the use of Pandemrix in large numbers of children, onset of narcolepsy following influenza or streptococcal infection had been noted. HLA DQB1*06:02 is present in 15–25 % of the population, and is notably frequent in northern Europe, but narcolepsy only occurs in 25–50 per 100,000 individuals. The disease generally presents during adolescence or early adulthood [22]. The widespread use of an adjuvanted influenza vaccine in northern Europe appears to have precipitated the onset of narcolepsy in small numbers of individuals. In Finland, 54 cases were diagnosed in 2010, representing a 17-fold increase over preceding years.

A study in the UK found that the adjuvanted Pandemrix vaccine was highly immunogenic in children and resulted in improved persistence of antibodies following vaccination [23]. T-cell responses to internal influenza antigens (NP, matrix protein 1 and nonstructural protein 1) were higher in children who had received the adjuvanted vaccine compared to a control group who received nonadjuvanted whole virion vaccine [24]. This highly immunogenic vaccine may therefore have, in rare cases, resulted in the induction of an autoimmune response and the onset of narcolepsy. The unusual situation in which large numbers of children were vaccinated with a pandemic-specific vaccine, whereas in most years, the majority would not have received influenza vaccine at all, and the surveillance following immunization that resulted in the reporting of these cases now presents us with an opportunity to understand this rare autoimmune disease. If the auto-cross-reactive HLA DQB1*06:02 epitope in an influenza antigen is identified, this antigen (or at least the region of the antigen containing that epitope) can be omitted from future vaccines. It is important to note that influenza infection can also trigger narcolepsy, and that this is a property of the influenza virus antigens rather than any particular vaccine or its adjuvant. Since HLA DQB1*06:02 is linked to promoter polymorphisms of the TNFA gene [25] and MX2 tends to be downregulated in subjects with HLA DQB1*06:02 [26], the explanation may prove to be complex. However, if we are to encourage vaccination, particularly of children, against influenza, it is necessary to understand why in rare cases, immunization can precipitate narcolepsy, and this opportunity should not be ignored.

Financial & competing interests disclosure

SC Gilbert is named as an inventor on patents relating to methods of vaccination, including novel influenza vaccines. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.