Effect of a synbiotic on the response to seasonal influenza vaccination is strongly influenced by degree of immunosenescence

The study protocol was reviewed and approved by the University of Reading Research Ethics Committee (project number: 10/09) the National Health Service (NHS) Research Ethics Committee for Wales (10/MRE09/5). The trial was registered with clinicaltrials.gov (Identifier: NCT01066377) and conducted according to the guidelines laid down in the Declaration of Helsinki.

Prior to the influenza season of 2010–2011, young (18–35 y) and older (60–85 y) healthy adults were recruited from the population in and around Reading (UK) through newspaper and poster advertisements, email and radio. Inclusion criteria were: a signed consent form, age 18–35 y or 60–85 y, body mass index (BMI) 18.5–30 kg/m², good general health, as determined by medical questionnaires and laboratory data from screening blood and urine sample (fasting glucose, erythrocyte sedimentation rate, full blood count, liver function tests, renal profile, dipstick urinalysis), not pregnant, lactating or planning a pregnancy. Exclusion criteria included: allergy to the influenza vaccine, HIV infection, diabetes requiring any medication, asplenia and other acquired or congenital immunodeficiences, any autoimmune disease, including connective tissue diseases, malignancy, cirrhosis, connective tissue diseases, current use of immunomodulating medication (including oral and inhaled steroids), self–reported symptoms of acute or recent infection (including use of antibiotics within last 3 months), taking lactulose or any other treatment for constipation, alcoholism and drug misuse. Additional exclusion criteria for older volunteers included: laboratory data which were outside the normal range for this age group AND outside the ranges specified in the SENIEUR protocol [31], Barthel Index score of < 16/100, cumulative illness rating scale (CIRS) score of > 15 [32]. Additional exclusion criteria for the young subjects included laboratory data which were outside the normal range and influenza vaccination in the previous 12 months.

The primary outcome of the trial was the antibody response to vaccination, incorporating mean antibody titres, vaccine-specific Ig subclasses and seroprotection and seroconversion. Power calculations were based on mean antibody titres. Since the influenza vaccine is trivalent, it is unlikely that an intervention will alter the response to all three subunits in the same way. For example, in the study of Davidson et al. [33], there was no effect of probiotic on mean antibody titres in response to the H1N1 subunit, whereas the responses to both H3N2 and the B subunit were improved (72 vs 51 [SD 16.5] for H3N2 and 31 vs 25 [SD 7.1] for B subunit). Based on the smaller effect size for the B subunit, a sample size of 26 subjects per group within each cohort was determined to be sufficient for a two-tailed significance level of 5 % and a power of 80 %; this was adjusted to 30 subjects per group to allow for dropouts. Data on the co-primary endpoints, immunoglobulin subclasses, seroprotection and seroconversion, is very sparse, but a sample size of 26 subjects per group within each cohort was determined to be sufficient for a 376 mg/dL difference in circulating IgG levels in response to influenza vaccination, with an SD of 438 mg/dL, a two-tailed significance level of 5 % and a power of 80 % [23]. A total of 62 young subjects and 63 older subjects entered the study and 58 young and 54 older subjects completed the study (Fig. 1). Two subjects experienced adverse effects (gastrointestinal bloating) during the study, one on the placebo group and one in the B. longum + Gl-OS group; both withdrew from the study.

Subjects consumed Bifidobacterium longum bv. infantis CCUG 52,486 (B. longum, 10⁹ CFU in 1 g skim milk powder/day) combined with gluco-oligosaccharide (Gl-OS (BioEcolians, Solabia); 8 g/day) in a double-blind, placebo controlled randomised parallel study design for 8 weeks. The synbiotic approach was selected because in vitro data examining the growth and survival of this strain indicated that it was very vulnerable compared with other strains, but survived much better in the presence of an oligosaccharide substrate (data not shown). When comparing a number of possible substrates, the low water activity of Gl-OS, combined with its ability to support the growth of the probiotic strain, made it a clear choice for a powdered product. This prebiotic also has bifidogenic effects in batch culture models [34]. The placebo used was maltodextrin (9 g/day); both the placebo and the pre- and probiotic were sourced, packaged and blinded by BioAgro S.A. (Italy). The powders were consumed sprinkled into water or milk or with breakfast cereal. Microbiological safety of the product was independently verified by Leatherhead Food Research associates (UK) prior to commencement of the study and viability of the probiotic strain was confirmed on random samples on a weekly basis during the study. During the 3 weeks prior to the study and during the intervention itself, subjects were requested not to consume fermented products such as yogurts, kefir etc. or pre- or probiotic products. Dietary records were used to confirm avoidance of such foods. Subjects were randomized by a research nurse not involved in the analysis according to gender, age and BMI to receive the probiotic or placebo by covariate adaptive randomization [35]. All investigators were blinded to the treatments. After 4 weeks, subjects were administered with a single dose of the influenza vaccine (Influvac®sub-unit 2010/2011 season, Abbott Biologicals B.V., lot number 1070166) containing A/California/7/2009 (H1N1), A/Perth/16/2009 (H3N2) and the B/Brisbane/60/2008- like strain by intra-muscular injection in the deltoid. Vaccination was carried out by a research nurse in the presence of a qualified clinician (MG). Details of the study schedule and samples collected are detailed in Fig. 6. Compliance was assessed by counting returned sachets and by copy numbers of B. longum, assessed by qPCR. None of the subjects in the young cohort had previously received seasonal influenza vaccination or swine flu vaccination. Three subjects in the older cohort had received swine flu vaccination, and forty subjects had previously been vaccinated for seasonal influenza, of whom 37 had been vaccinated in the 2009/2010 period.

For serum, blood was collected into serum separator tubes and left at room temperature for 30mins to allow coagulation. Samples were centrifuged at 1300 × g for 10 min and aliquots of serum were collected and stored at −80 °C prior to analysis.

Serum samples (in duplicate) were sent to the Health Protection Agency (London, UK) for analysis of total antibody responses to each of the 3 viral subunits. This technique is based on the ability of specific anti-influenza antibodies to inhibit hemagglutination of red blood cells by influenza virus hemagglutinin protein. Serum antibody titers measured by hemagglutination inhibition (HI) assay was performed using an 8-step dilution protocol (1:8 to 1:1024). Serum samples were titrated in duplicate, and samples showing titres of 1024 were further titrated up to 1:4096. Titres which were < 8 (below the detection limit) were expressed as 4, to represent half of the detection threshold. Post-vaccination antibody titres were not normally distributed, and were therefore log transformed.

Vaccine-specific Ig subclasses were analysed using an in-house ELISA method. Briefly, 96-well maxisorb ELISA plates (Serotec) were coated with 500 ng/ml of the vaccine (Influvac sub-unit 2010/2011) in coating buffer (0.5 M Na₂CO₃). After overnight incubation at 4 °C, plates were washed three times with wash solution (50 nM TRIS, 0.14 M NaCl, 1 % BSA, 0.2 % Tween-20 in dH₂0). Blocking buffer (5 % BSA in PBS) was added and plates were incubated for 1 h at 37 °C. After three washes, plasma samples (neat for IgA, IgD, IgM, and diluted 1:100 in PBS for IgG1) were added, and plates were incubated for a further hour at RT. Plates were then washed to remove non-specific binding antibodies and mouse anti-human IgA, IgG, IgM or IgD was added. IgA, IgG1, and IgM were diluted with PBS 1:1000, whereas IgD was diluted 1:500. After a 1 h incubation, diluted horse radish peroxidase conjugated goat anti-mouse IgG (H/L) (AbD Serotec, Oxfordshire, UK) was added and the plate was incubated for a further 1 h at RT. Substrate (3.3’, 5.5’ tetramethylbenzidine) was added and the reaction was stopped with 0.16 M sulphuric acid. Plates were read at 450 nm in a micro-plate autoreader (Thermo Scientific) and the results were presented as a mean optical density units (OD) ± SEM.

To minimize inter-assay variation, three quality control samples were included in every assay, consisting of post-vaccination plasma samples from two young and one older subject.

Saliva samples were collected by passive drooling into a sterile screw-cap tube, centrifuged at 13,000 × g for 10 min at 4 °C to remove cells and debris and stored at −80 °C until analysis, which was completed within 6 months. The concentration of sIgA in saliva was analysed by sandwich ELISA using a commercially available kit (Immunodiagnostik AG, Bensheim, Germany) in accordance with the manufacturer’s instructions.

Concentrations of anti-CMV IgG antibodies were analysed by ELISA according to the manufacturer’s instructions (ab108724 Anti-Cytomegalovirus (CMV) IgG Human Elisa Kit, Abcam, UK) and read in a microplate reader (GENios) at 450 nm, with 620 nm as a reference wavelength.

Peripheral blood mononuclear cells (PBMC: 1 × 10⁶) were stained with the following fluorochrome-conjugated monoclonal antibodies: (PerCP) labelled anti-CD3, (AmCyan) labelled anti-CD4, (APC-Cy7) labelled anti-CD8, (PE-Cy7) labelled anti-CD25, (Pacifice Blue) labelled anti-CD28, (APC) labelled anti-CD57, (FITC) labelled anti-CD26, (PE) labelled anti-CD127 in TruCOUNT tubes (BD Biosciences), washed and fixed before analysis by multiparamter flow cytometry (FACS Canto II, BD Biosciences) using BD FACSDiva™ software.

Telomere length of PBMC was measured by using the DAKO Telomere PNA Kit/FITC for Flow Cytometry (DAKO, UK), according to the manufacturer’s instructions. Relative telomere length (RTL) was determined by comparing isolated PBMC with a control cell line (1301; subline of the Epstein–Barr virus (EBV) genome negative T-cell leukaemia line CCRF-CEM; Sigma-Aldrich Company Ltd, UK). Quantification of RTL was carried out using an Accuri™ C6 flow cytometer (BD Biosciences, Oxford, UK).

Freshly voided faecal samples were collected in sterile plastic pots and stored at −20 °C.

DNA was extracted from faecal samples using a commercial kit (Fast DNA Spin kit, MP Biomedicals, Cambridge, UK) according to the manufacturers’ instructions. Quantity and purity of the DNA was assessed using NanoDrop (Thermo Fisher Scientific Inc., Loughborough, UK). Real-time quantitative PCR was performed in triplicate for enumeration of Bifidobacterium longum with the Rotor-Gene 6000 (Corbett, Mortlake, Australia) using 5 μL of extracted faecal DNA in 20 μL PCR master-mix according to Haarman and Knol (2005). The reaction was carried out after 1 cycle at 95 °C for 10 min, followed by 45 cycles at 95 °C × 10, 62 C° per 60 s, 72 °C per 40 s. Standard quantification was carried out using serially diluted DNA extract of Bifidobacterium longum bv. infantis CCUG 52,486 strain enumerated by plate count.

Freshly voided faecal samples were collected in sterile plastic pots, diluted 1 in 10 (wt:wt) in PBS and homogenised in a Stomacher 400 (Seward, Norfolk, United Kingdom) for 2 min. A 15 ml sample of faecal slurry was vortexed with 2 g of 3 mm diameter glass beads and then centrifuged to remove particulate matter (1500 × g, 2 min). Supernatant was fixed in paraformaldehyde [1:4 v/v in 4 % PFA in 0.1 mol PBS, pH 7.2] for 4 h at 4 °C, centrifuged (13,000 g 5 mins), washed twice with 0.1 M PBS, resuspended in 1:1 PBS:ethanol and stored at −20 °C. Oligonucleotide probes used were Cy-3 labeled and synthesized by Sigma-Aldrich (Poole, UK). Probes used were Bif164, Bac303, Chis150, Lab158, ATO291, Enter1432, Fprau655 and EUB338I/II/III specific for Bifidobacterium spp., Bacteroides/Prevotella group, Clostridium clusters I and II (including C. perfringens and C. histoliticum), Lactobacillus/Enterococcus subgroup, Atopobium, Enterobacterium group, Faecalibacterium cluster and total bacteria respectively. Samples were hybridized as described [36]. Data is expressed as log10 counts per g dry weight faeces.

Data were analysed using SPSS software (version 21). For the primary and continuous secondary endpoints, a Linear Mixed Model (LMM) was implemented. A first order autoregressive covariance structure was selected AR (1), with fixed factors of time (repeated measures), age and treatment and subject as a random effect. The main effects are reported first, followed by all two-way interactions of the variables. Following this initial analysis, the data were split by cohort (young/older) and the analysis was repeated in the same manner to determine time and treatment effects within each cohort. Vaccine-specific immunoglobulin subclass data were analysed as change following vaccination. The distribution of the data was checked using the Kolmogorov-Smirnov test. If data were not normally distributed, they were log transformed. Antibody titre data were transformed with binary log (log₂ n) and were not analysed as change from baseline. Seroprotection, seroconversion and self-reported illness data were analysed by the Fisher’s exact test. To account for multiple primary endpoints, two sided P values of 0.01 or less were considered statistically significant. All missing data were classed as missing at random and only available data were analysed.