The subjects were healthy adults aged 18 to 45 and serologically negative to the challenge virus (serum HAI titres ≤1:10). Detailed inclusion/exclusion criteria are described in clinicaltrials.gov under identifier NCT02014870 and NCT02071329. All subjects gave written informed consent to participate in the trial. The protocol and informed consent were approved by an independent ethics committee, Institutional Review Board, ZNA/OCMW, Antwerp, Belgium. The study was conducted in accordance with EU Directive 2001/20/EC and ICH GCP.

Data from two studies are included here. Clinical challenge volunteers were inoculated with influenza A/California/2009 (H1N1). The first study was a dose escalation study (NCT02014870) and details of this study have been reported elsewhere (Watson et al., 2015): three cohorts were employed: the low cohort were inoculated with 3.5 × 10⁴ TCID₅₀, the medium cohort received 3.5 × 10⁵ TCID₅₀ and the high cohort received 3.5 × 10⁶ TCID₅₀. Subjects entered the quarantine unit 24 h prior to virus challenge and remained for an additional 7 days. The second study is comprised of placebo subjects from a vaccine study (NCT02071329) who were infected with the 3.5 × 10⁶ TCID₅₀ of virus. Data from the first study is used to assess sera responses at day 29 after infection, which was not collected in study 2. For the evaluation of protection, data from the high infection cohort and study 2 are pooled for comparable infection dose.

Immediately before challenge, frozen virus was thawed rapidly in a 37°C water bath before dilution in pre-warmed (37°C) phosphate-buffered saline (PBS, Mediatech) to appropriate challenge dose. Subjects were placed in a semi-recumbent position, with the head tipped slightly back and were asked to close their palate. A total of 0.5 ml of diluted virus stock was administered intranasally to each subject (0.25 ml per nostril) using a pipette.

Symptoms were captured twice daily by self-reporting on 16 events associated with influenza illness (headache, nasal stuffiness, runny nose, sore throat, sneezing, hoarseness, earache, facial or eye pain, cough, wheezy chest, breathing difficulty, musculoskeletal ache, nausea/vomiting, feeling/hot/feverish/chills/rigor, fatigue and diarrhea) and classified as absent, mild, moderate or severe. Classifications were defined as follows: mild: the event causes a minor discomfort, does not interfere with daily activity of the subject or does not lead to establishment of a correcting treatment; moderate: the event perturbs the usual activity of the subject and is of a sufficient severity to make the subject uncomfortable; severe: the event prevents any usual routine activity of the patient and causes severe discomfort. In addition, a more objective targeted physical exam was performed twice daily by the study physician which assessed 6 clinical parameters (nasal discharge, otitis, pharyngitis, sinus tenderness, new wheezes, crackles or rhonchi on lung auscultation and percussion) using the same classification system. Vital signs and safety laboratory tests were performed daily. Temperature was measured twice daily. Nasal washes were performed daily by instilling 5 ml of pre-warmed (37°C) sterile phosphate-buffered saline into each nostril and the effluent was collected, diluted 1:1 in virus transport medium and stored at ≤-65°C. Daily nasal-discharge weights were determined by the collection of pre-weighed tissues in pre-weighed plastic bags assigned to each subject during each 24-h period. On the day of discharge from the quarantine unit subjects were required to have negative result from a negative rapid influenza test (Directigen^TM EZ Flu A+B, BD).

The haemagglutination-inhibition (HAI) assay at Day-1 and 29 (VisMederi srl, Siena, Italy) for all cohorts. Serum was pre-treated with receptor-destroying enzyme (RDE) II (Denka Seiken, Japan) and twofold serially diluted starting from a dilution of 1:10 to 1:2,560 in physiological saline. Virus (A/H1N1/California/04/2009, NIBSC) or (A/Victoria/361/2011) was added to each well at 4 haemagglutination units (HAU)/50 μl for 1 h at room temperature, followed by 0.35% turkey red blood cells for 1 h at room temperature. Reference sheep hyperimmune antisera were provided by NIBSC.

For real-time RT-PCR (rRT-PCR), the WHO protocol “CDC protocol of real-time RT-PCR for swine influenza A (H1N1)” was followed by VisMederi srl using but only the InfA primer designed for the universal detection influenza A viruses. A specimen was considered positive for influenza A if cycle threshold (Ct) values were within Ct values of 40. A virtual quantification tool was applied to the Ct values to allow the conversion of CDC rRT-PCR Ct values to virus RNA copy number, in the absence of a standard curve run in parallel to the samples (Piralla et al., 2013).

Antibodies specific to influenza were measured using a standardized ELISA (Russell et al., 2016). IgG responses were measured in sera and IgA responses in sera and nasal wash. To detect antigen specific responses, MaxiSorp 96-well plates (Nunc) were coated with 1 μg/ml H1N1 (A/California/09) or H3N2 (A/Texas/50/2012) surface antigens (GSK Vaccines, Sienna, Italy) and incubated overnight at 4°C. Plates were blocked with 1% BSA in PBS. Bound IgG was detected using biotinylated goat anti-human IgG (Sigma) and bound IgA was detected using a biotinylated anti-IgA (AbD Serotec) followed by Poly-HRP40 (Fitzgerald biotech). To quantify the concentration of antigen specific antibody, control wells were coated with a combination of anti-human lambda and kappa light chain specific antibodies (AbDSerotec, Oxford, United Kingdom) and a dilution series of control non-specific human IgG (Sigma) or IgA (Sigma) used as a standard in these wells. TMB with H₂SO₄ as stop solution was used to detect the response and optical densities read at 450 nm.

Correlation of data was performed using Pearson r-test. Comparison between 2 values was performed using t-tests, comparison of 3 variables was performed using ANOVA and Tukey’s multiple comparison test. All statistical analyses were performed using GraphPad prism 7.

The aim of the study was to define the antibody correlates of protection against influenza, focusing on IgA. To assess this we measured serum and local antibody responses in volunteers challenged intranasally with 3.5 × 10⁶ TCID₅₀ influenza A/California/2009 (H1N1). Clinical challenge volunteers from two studies were used for this study, the first study was a dose escalation and the second study was a vaccine and challenge study; the high dose challenge group from study one was combined with placebo patients from study two (all data is available in Supplementary Table S1). Patients were pre-screened for low H1N1 HAI titres (≤10) to increase the likelihood of infection because of the well-established correlation between HAI titre and protection against infection (Hobson et al., 1972).

There was a range of outcomes to viral challenge, with 14 out of 47 volunteers having no detectable virus by either PCR or culture. The number of days infected patients had culture positive samples recovered from the nose was normally distributed with the median being 4 days (Figure 1A). There was influenza RNA detectable in 33 volunteers (Figure 1B). Twenty volunteers showed no signs and symptoms of clinical illness and the median was for a mild response (Figure 1C). When plotted as the area under the curve of the RNA detected over the time course of the study, influenza RNA strongly correlated (r = 0.4428, p < 0.0001) with days virus was detectable (Figure 1D). Disease signs and symptoms also correlated with days culture positive (Figure 1E). Of the virus negative volunteers, 6 were also asymptomatic, interestingly there were virus positive – asymptomatic patients and virus negative – symptomatic patients. Since all the patients had low HAI titres, the variability in responses points to other host factors contributing to protection against infection.

HAI assays measure one facet of antibody immunity, we wished to determine whether H1N1 binding IgG was an independent correlate of protection. Serum was collected 1 day prior to infection and on day 29 after infection in study one only; responses in all three cohorts in this study were compared. There was a significant increase in the H1N1 specific IgG titre in the serum after infection; the median moved from 2,830 to 21,189 ng/ml (Figure 2A). The majority of volunteers increased in IgG titre after infection (26 out of 29) and there was no difference in the fold increase in antibody titre between individuals infected with different doses of influenza (Figure 2B). Whilst there was detectable IgG in the nose, there was very little H1N1 specific IgG (data not shown). The volunteers all had undetectable HAI prior to challenge, but this significantly increased after challenge (Figure 2C), and as seen with IgG challenge dose did not alter the change in HAI. Even though there was a spread of IgG ELISA values, HAI and ELISA IgG did not correlate prior to challenge, presumably as a consequence of pre-screening, but there was a significant correlation between HAI and IgG at day 29 after infection (Figure 2D, R²= 0.61, p < 0.0001).

In order to assess possible correlations between IgG and infection, data from the high dose group from the first cohort was combined with subjects from the second study. There was no significant correlation between H1N1 specific IgG and days culture positive (R² = 0.0014, p = 0.8; Figure 2E), viral RNA copies detected by PCR (data not shown) or cumulative signs and symptoms (R² = 0.023, p = 0.31; Figure 2F) infection. This suggests that individuals pre-screened for low HAI also had low serum IgG titres which were also sub-protective, and the two measures do not independently segregate as correlates of protection.

A number of studies have indicated that IgA is an independent correlate of protection to IgG (Rossen et al., 1970; Clements et al., 1986; Belshe et al., 2000; Ambrose et al., 2012). H1N1 specific IgA was measured in the serum of volunteers in study 1 on day-1 and day 29. H1N1 specific IgA in the serum significantly increased after infection (Figure 3A), the average fold change was 24. Interestingly, challenge dose had a significant effect on the increase in IgA titre, with the IgA increasing more in recipients of higher titre virus than those receiving a low dose (Figure 3B). Prior to infection, there was no correlation between serum IgA and IgG (Figure 3C). However, IgA and IgG did correlate on day 29 after infection (Figure 3D). There was no correlation between HAI and IgA (R² = 0.03, p = 0.384, data not shown).

The main aim of the study was to determine whether systemic or local IgA correlated with protection. H1 specific IgA was measured in nasal lavage; to compensate for dilution effects we normalized the values against total IgA recovered (Figure 4A). Prior to infection, H1N1 specific IgA in nasal wash and serum did not correlate (Figure 4B). Levels of serum and nasal IgA were compared with the same measures of infection tested for IgG. For both serum IgA (R² = 0.09293, p < 0.05; Figure 4C) and nasal IgA (R² = 0.08655, p < 0.05; Figure 4D) there was a weak but significant inverse correlation with the number of days detectable virus was shed, measured by culture. There was a similar, but not significant pattern when comparing IgA with either total RNA detection or symptoms. Therefore, IgA provides a separate correlate of protection to IgG against influenza.

To see if the effects we saw were virus serotype specific, we also measured responses against a recently circulating H3N2 virus (A/Texas/50/2012). After infection, there was a significant increase in H3N2 specific IgG in the sera (Figure 5A), but this was significantly lower than the H1 fold change (mean of 4 compared to 23 for H1, Figure 5B). This slight increase may reflect overlapping antigen similarity between H1 and H3, though there was no correlation between the individuals in terms of their increase in H1 and H3 titres. There was no increase in H3 HAI titre (Figure 5C) or H3 IgA in the sera (Figure 5D). There was no correlation between H3 IgA in the serum (R² = 0.0016, p = 0.79; Figure 5E) or the nasal lavage (R² = 0.0167, p = 0.37; Figure 5F) and protection against infection.