Immunity to Influenza. Preventing Infection and Regulating Disease

Evasion of the immune response by influenza viruses in the human population has resulted in a heavy healthcare burden and substantial mortality (1, 2). Influenza virus eludes the immune system through multiple processes, including seasonal mutations in the surface epitopes, termed “antigenic drift,” and gene segment reassortment from distinct strains generating a novel lineage, termed “antigenic shift.” Both of these processes create obstacles in clinical attempts to limit influenza infections in humans. Research in many labs is focused on developing innovative techniques to prevent influenza infection despite antigenic drift and shift, including efforts to design a universal influenza vaccine, targeting the regions of the influenza virus that are conserved across strains (3–5).

At the same time, parallel research programs have sought to understand the susceptibility to infection and the susceptibility to severe influenza illness in humans, including the new study from Sridhar and colleagues (pp. 325–332) in this issue of the Journal (6). What are the natural mechanisms within the human host that allow for protection or permit pathology, and how can those be exploited to improve population outcomes to influenza? It appears that the mechanisms that prevent infection may differ from those that reduce disease severity after infection. Seminal studies were done through the 1960s to the 1980s, establishing key principles of human influenza immunology. In 1976, Hoskins and coworkers found that influenza infection in a boarding school for boys allowed for longer-lasting immunity than did the killed virus vaccine (7). Another vaccine study found that vaccination by killed whole virus was able to elicit induction of memory cytotoxic T lymphocytes (CTLs), whereas the subunit vaccine was not (8). These studies provided evidence that natural infection provides more protection than vaccination against subsequent influenza infections. More recently, challenge studies have been employed to look at specific immunological effects of human influenza infection. Wilkinson and colleagues used H3N2 and H1N1 strains to experimentally infect healthy individuals to measure the response of CD4⁺ and CD8⁺ T cells, and found an increase in antigen-specific CD4⁺ T cells at Day 7 after infection, as well as associations between higher numbers of preexisting antigen-specific CD4⁺ T cells with lower viral shedding and milder illness outcome (9). Another challenge study using H1N1 (A/Brisbane/59/07) looked at the virus-specific B-cell response in healthy individuals, identifying influenza-specific antibody-secreting cells (ASCs) in peripheral blood mononuclear cells (PBMCs) in the first week after infection (10). Associations were found between the magnitude of the influenza-specific ASC response and viral load and duration of viral shedding.

While challenge studies provide a controlled experimental system for collecting samples and defining endpoints, by necessity they are performed on healthy adults and induce only mild disease. As a result, they are unable to answer questions relating to immunologic protection against natural infection and susceptibility to severe disease. Human studies of naturally acquired influenza infections have focused on how patients respond to the virus, and on the correlates of immunity between severe and mild cases. Oshansky and coworkers characterized the innate immune response in children and adults infected with influenza virus, including H3, pH1, and B strains (11). This study determined predictors of severe disease outcome, independent of viral load and age, including levels of nasal wash monocyte chemotactic protein-3, nasal wash IFNα2, and plasma IL-10. Immune characteristics of nasal washes and plasma were substantially different, highlighting the importance of cellular migration to site of infection and site-specific immune phenotypes. The results point to the innate response as an important contributor to disease severity. The emergence in 2009 of an H1N1 pandemic virus (H1N1pdm09) provided a rare chance to investigate host characteristics in response to a novel subtype. Denney and colleagues looked at the cellular composition in PBMCs of three severe cases in the United Kingdom (12). They determined that natural killer (NK) cells were significantly reduced in the patients with severe disease compared with normal control subjects. One patient who succumbed to the infection due to viremia had no detectable NK cells. Also, the CD4:CD8 T-cell ratio in patients with severe disease was much higher than in those with milder disease. A different study focused on the T-cell profiles in the blood and lung of patients infected with H1N1pdm09 with mild and severe outcomes (13). They concluded that the T-cell profile in the blood and lung are significantly different, with high levels of CD4⁺ T cells detected in the blood at early stage of infection and activated CD8⁺ T cells detected in high levels in the lungs. Acute disease was characterized by high levels of CD4⁺ T cells, measured from a PBMC sample collected 2 or 3 days after hospital admission, which were detected before clinical decline; therefore, the authors proposed high CD4⁺ T-cell levels in the blood as an early biomarker of severe influenza illness. He and colleagues found unique patterns of B-cell activation and priming in individuals after infection or vaccination with H1N1pdm09 (14). B-cell responses were significantly higher in subjects with H1N1pdm09 infection than in vaccinated subjects, and priming with natural H1N1pdm09 showed increased lymphocyte memory after subsequent immunization. A recently published study using households from the CoPanFlu-France cohort aimed to define the most essential factors determining susceptibility to H1N1pdm09 infection using structural equation modeling integrating a diverse set of factors, including host susceptibility (pre-epidemic HAI titer), contact network, and risk perception of H1N1pdm09 infection (15). The conclusions of the study provided evidence that host susceptibility and compliance with preventive behaviors were most associated with susceptibility to H1N1pdm09 infection, while vaccination was one of the factors most associated with protection.

Many of these natural infection studies have focused on the role of cellular responses after infection, as measuring cellular or humoral immunity before infection requires a prospective cohort design, which can be expensive and difficult to power effectively. Thus, the dynamics of humoral immunity from baseline through a natural infection and into memory have been difficult to capture. In this issue, Sridhar and colleagues report on a prospective cohort of 342 healthy individuals (16) established before the start of the second UK pandemic H1N1 wave (September 2009 to April 2010) (6). Participants were followed through two consecutive influenza seasons in which the pH1N1 was the predominant strain. Individuals lacking neutralizing antibodies to pH1N1 at enrollment who subsequently developed incident pH1N1 infection were included for analysis. The timing of this study provided a unique opportunity to collect biological samples before and after infection, allowing interrogation of humoral immunity in seronegative individuals, a situation that rarely occurs for influenza. Comparing antibody titers before and after natural pH1N1 infection showed a significant increase in antibody titer in these participants. Interestingly, the antibody titers remained constant and above the protective threshold (1:32) for at least 1.5 years. In addition to seronegative individuals who became infected, vaccinated individuals were also assessed for antibody titers to pH1N1. Compared with infected individuals, the antibody titers in vaccinated individuals were lower, but still remained above the protective threshold for up to 1.5 years. These findings that infection has a stronger effect on induction of antibodies than vaccination support data collected in the mouse. Finally, though the focus of the study was on humoral immunity, correlates of the antibody response with cellular immunity were also studied. Somewhat surprisingly, the primary correlate identified was negative, with CD4⁺IL-2⁺ T cells inversely related to the magnitude of antibody generation. The authors hypothesize that the induction of new CD4⁺ICOS⁺IL-21⁺ follicular helper (Tfh) cells are responsible for the induction of antibody response, rather than preexisting frequencies of CD4⁺ T cells—thus, the levels of CD4⁺IL-2⁺ T cells are a negative proxy for the development of a robust Tfh response that would promote humoral immunity. These findings establish that protective antibody levels are maintained for at least 1.5 years in human after natural pH1N1 infection, and suggest that CD4⁺IL-2⁺ T cells might play a negative regulatory role in determining antibody responses to influenza infection.

Given the long history of influenza research, it is remarkable that many of the key features of the immune response to natural infection are still poorly understood in humans. Key correlates discovered in human populations during the course of naturally acquired influenza infection are outlined in Figure 1. The natural infection study design used in this study has many features that should be considered in future protocols and by large cohort initiatives, including the broad collection of baseline samples for longitudinal antibody titer comparison and its rapid establishment after the emergence of a new pandemic, allowing study of a single infection to reduce the effects of multiple influenza exposures (6). Future studies that incorporated this prospective design with long-term follow-up and included infants, hospitalized patients, and the elderly would be of great value to the clinical influenza community for designing new interventions for vaccination, diagnosis, and treatment.

Figure 1.

Correlating immune responses to clinical features of influenza infection. Immunity to influenza can prevent infection after exposure (far left), mitigate the severity of disease and the extent of household transmission (middle), and contribute to the observed clinical pathology (middle to far right). Understanding the mechanisms underlying these correlates can lead to better diagnoses and interventions, including treatments and vaccines. Novel findings from this study (6) are indicated by bold text.