CELLULAR IMMUNE CORRELATES OF PROTECTION AGAINST

Transcription

1 CELLULAR IMMUNE CORRELATES OF PROTECTION AGAINST SYMPTOMATIC PANDEMIC INFLUENZA Saranya Sridhar 1, Shaima Begom 1, Alison Bermingham 2, Katja Hoschler 2, Walt Adamson 3, William Carman 3, Thomas Bean 4, Wendy Barclay 5, Jonathan Deeks 6, Ajit Lalvani 1 1 Department of Respiratory Medicine, National Heart and Lung Institute, Imperial College London, Norfolk Place, LondonW2 1PG 2 Respiratory Virus Unit, Centre for Infections, Public Health England, 61 Colindale Avenue, London NW9 5HT 3 West of Scotland Specialist Virology Centre, Glasgow G12 0YN, Scotland, UK 4 HPA Microbiology Services, Porton Down, SP4 0JG, UK 5 Department of Virology, National Heart and Lung Institute, Imperial College London, Norfolk Place, London W2 1PG 6 Public Health, Epidemiology and Biostatistics, School of Health and Population Sciences, University of Birmingham, B15 2TT, UK 1

2 Supplementary Information SUPPLEMENTARY METHODS: Sample processing: PBMCs were isolated by Ficoll-Paque PLUS (Amersham Biosciences) density centrifugation and cryopreserved in heat-inactivated foetal calf serum supplemented with 10% DMSO (Sigma-Aldrich) at -180 C in liquid nitrogen as previously described 1. All assays were undertaken using cryopreserved PBMCs with >80% viability of cells after thawing. Serum was stored at -20 C. RT-PCR for ph1n1 09 virus: The influenza A(H1)v specific assay of the Health Protection Agency (HPA) contains primers and a dual-labelled TaqMan MGB probe (Applied Biosystems) targeting conserved sequences in the HA gene of A(H1N1)v viruses, and the positive control swine A(H1N1) virus A/Aragon/3218/2009, in a 1-step TaqMan PCR assay has been previously described in detail 2. Haemagglutination Inhibition assay Antibody responses to the virus strain A/England/195/2009, circulating in the UK during our study, were detected by use of Haemagglutination inhibition assay according to standard methods at the Centre for Infections, Health Protection Agency (London, UK) 2. Briefly, human sera were treated with receptor-destroying enzyme (RDE) II, (Denka Seiken, Japan) for 18 hours followed by heat inactivation for 1 hour at 56 C. Sera were screened in a limiting dilution range using the NIBRG122 virus and incubated with the haemagglutinin (HA) antigen suspension for 1 hour followed by addition of 0.5% RBC suspension (turkey 2

3 blood). The reaction is left for 1 hour at room temperature before reading. Each sample is titrated in duplicate and individual titres reported did not differ by more than a twofold serial dilution. The serum titre is equal to the highest reciprocal dilution, which induces a complete inhibition of haemagglutination. Suitable control serum samples were included in all analyses, with post-infection ferret serum samples raised against the ph1n1 virus strain as positive controls; human pooled serum samples from individuals with either high antibody titres to currently circulating influenza H1, H3, and B strains or from individuals with no antibody titres to these seasonal strains were used as negative controls. Microneutralisation assay Human sera were heat inactivated for 30 min at 56 C and twofold serial dilutions were produced using virus diluent (DMEM + 2µg ml -1 TPCK Trypsin) in a total volume of 100µl in immunoassay plates. The diluted sera were mixed with an equal volume of virus diluent containing 100xID50 influenza H1N1 A/California/7/09 virus and incubated for 2 h at 37 C in a 5% CO 2 humidified atmosphere, after which 100 µl of MDCK cells at cells ml -1 were added to each well. The plates were incubated for a further 18-20h at 37 C and 5% CO 2. Cell monolayers were washed with PBS and fixed in cold 80% (v/v) methanol for 10 min. The presence of viral protein was detected by ELISA with a monoclonal primary antibody against influenza A NP (MCA400 Mouse anti influenza a nucleoprotein, AbD Serotec) and a secondary HRP conjugate (Goat polyclonal secondary to Mouse IgG HRP, Abcam ab97023). After staining for 10 minute using o-phenylenediamine dihydrochloride, absorbance was read at 490nm in a plate reader. The mean of two replicates were read and samples were scored as positive for infection if the OD 490 was equal to or above a value 50% between the mean OD 490 from four virus negative control wells and the mean OD 490 from 3

4 four no sera, virus positive control wells. The reciprocal serum dilution corresponding to the highest dilution to be scored negative for infection is the 50% neutralising antibody titre. Virus and peptides Live ph1n1 virus was obtained by growing a recombinant A/England/195/09 strain in MDCK cells. Peptide pools 9-mer peptides representing highly conserved reported class I restricted epitopes in influenza A virus from PB1, M1 and NP proteins were obtained from the NIH Biodefense and Emerging Infections Research Resources Repository, NIAID, NIH: Peptide Arrays, Peptides for Expected Conserved MHC Class I Epitopes of Influenza Virus A Proteins, NR-2667 (Supplementary table 1). For each protein, PB1, M1 and NP, a separate pool of peptides was made at a concentration of 5μg/ml for each peptide in the pool. Cell culture lysates from cells infected with CMV were also used (East Coast Biologicals Inc., USA). FACS staining: Staining of PBMCs was undertaken as previously described following 18 hours stimulation with media (negative control), phorbol myristate acetate (PMA)/Ionomycin (positive control), live ph1n1 (A/England/09/195) virus (MOI = 5) and CMV lysate (10 g/ml, control antigen) to maintain consistency with the Fluorescence-immunospot assay. Monesin A (Sigma-Aldrich) was added 2 hours after addition of stimulus. Staining with CD107a (clone H4A3, BD Biosciences) and CD107b (clone H4B4, BD Biosciences) was undertaken at the time of stimulation. Cells were blocked with 10 % human AB serum (Sigma) to prevent non-specific antibody binding prior to staining with amine-reactive viability dye (LIVE/DEAD Fixable Blue Dead Cell Stain Kit, Invitrogen) as a live/dead marker and surface staining was undertaken using a suitable combination of fluorochrome labelled anti-human CD3 (clone 4

5 UCHT1, BD Biosciences), CD4 (clone S3.5,Invitrogen), CD8 (clone RPA-T8, BD Biosciences), CD14 ( clone HCD14, Biolegend), CD19 (clonehib19, Biolegend), CD56 (clone HCD56, Biolegend), CCR7 (clone 3D12, BD Biosciences), CD45RA (clone MEM-56, Invitrogen) and CCR5 (clone 2D7, BD Biosciences) markers. Intracellular cytokine staining (ICS) was performed with BD Cytofix/Cytoperm Plus kit according to the manufacturer s instructions and cells were stained with IFN-γ (cloneb27, BD Biosciences), IL-2 (clone MQ1-17H12, BD Biosciences) and TNF-α (clone MAb11, BD Biosciences) antibodies. Fluorescence minus one controls stimulated with live ph1n1 virus were used for identifying positive populations of CCR7, CD107, CCR5 and CD45RA. In all samples, at least 1 million live events were collected and analysed. Antigen-specific cytokine responses were calculated only if the responses were >0.001% of the parent population. Flow cytometric analyses were performed using a Fortessa (BD Biosciences) and data were analyzed with Flow Jo (Tree Star) software and SPICE software (NIH). Fluorochrome Surface marker Clone PE-CF594 CD3 UCHT1 AF700 CD8 RPA-T8 QDot605 CD4 S3.5 Qdot655 CD45RA MEM56 PE-Cy7 CCR7 3D12 PerCP-Cy5.5 CD69 FN50 APC-Cy7 CCR5 2D7/CCR5 BV570 CD57 HCD57 V450 IFNg B27 PE IL-2 MQ1-17H12 FITC CD107A H4A3 FITC CD107B H4B4 APC TNFa MAb11 Biotin CD14 HCD14 Biotin CD19 HIB19 Biotin CD56 HCD56 PE-Cy5 Streptavidin N/A 5

6 Symptom score A summed symptom score was designed by totalling the weight for each of the canonical influenza symptoms used in clinical studies 3, 4 with a weight of 0 for none, 1 to mild symptoms and 4 for severe symptoms attributed to each symptom. Mild symptoms were of a severity that did not interfere with normal daily activities while severe symptoms were those that affected normal daily activity or required medical attention. Symptom Weight for no Weight for mild Weight for severe symptoms symptoms symptoms Fever Cough Sorethroat Headache Myalgia Statistical analysis A logistic model was used to model the relationship between number of cytokine secreting cells (SFCs/million PBMCs) and risk of illness with fever. The probability of illness was modelled as 1/1+ exp(α + βt) where t is the number of cytokine secreting cells. A curve of probability of illness against number of cytokine secreting cells represents a susceptibility curve 5. The curves were first estimated with only SFCs/million as the explanatory variable and then with age as a covariate. Goodness of fit of models compared using likelihood ratio chi-squared test showed no significant difference when age was used as a covariate. 6

7 Supplementary Table 1. Individual haemagglutination inhibition and microneutralisation assay titres to ph1n1 virus at baseline for 51 individuals who subsequently developed infection. Study id HI titre to ph1n1 (Eng/195)* MN titre to ph1n1 (Cal/07)# Study id HI titre to ph1n1 (Eng/195)* MN titre to ph1n1 (Cal/07)# F004 5 <10 F179 5 <10 F007 5 <10 F181 5 <10 F009 5 <10 F F018 5 <10 F189 5 sample unavailable F019 5 <10 F190 5 <10 F021 5 <10 F196 5 <10 F025 8 sample unavailable F197 5 <10 F041 5 <10 F207 5 <10 F058 8 <10 F208 5 <10 F059 5 <10 F214 5 <10 F061 5 <10 F216 5 <10 F070 5 <10 F219 5 <10 F071 5 <10 F221 5 F073 5 <10 F229 5 <10 F075 5 <10 F238 5 <10 F087 8 <10 F278 5 <10 F089 5 <10 F282 5 F091 5 sample unavailable <10 sample unavailable F291 5 <10 F093 5 <10 F376 5 <10 F F377 5 sample unavailable F156 8 <10 F385 5 <10 F162 5 <10 F399 5 <10 F169 5 sample unavailable F401 5 sample unavailable F F413 5 <10 F172 8 <10 F421 5 <10 F * Haemagglutination Inhibition assay undertaken as described above. Titre of 8 or less was considered negative # Microneutralisation assay undertaken as described above. Titres of less than 20 considered negative. 7

8 Supplementary Table 2. Demographics of cohort included in the final analysis. 8

9 Supplementary Table 3. Sequences of predicted conserved MHC Class I epitopes from PB1, M1 and NP proteins, collectively denoted as CD8 conserved epitopes. Protein Peptide ID Peptide sequence HLA restriction PB1 PB1/1 4-NPTLLFLKV-12 HLA-A2 PB1 PB1/2 16-NAISTTFPY-24 HLA-A2 PB1 PB1/3 22-FPYTGDPPY-30 HLA-A24 PB1 PB1/4 30-YSHGTGTGY-38 PB1 PB1/5 77-NEPSGYAQT-85 HLA-A1 PB1 PB1/6 81-GYAQTDCVL-89 HLA-A11 PB1 PB1/7 83-AQTDCVLEA-91 HLA-A68 PB1 PB1/8 123-TQGRQTYDW-131 HLA-A26 PB1 PB1/9 127-QTYDWTLNR-135 HLA-B8 PB1 PB1/ ALTLNTMTK-229 HLA-B58 PB1 PB1/ TKDAERGKL-236 HLA-B15 PB1 PB1/ DAERGKLKR-238 HLA-B49 PB1 PB1/ RGKLKRRAI-241 HLA-A68 PB1 PB1/ KLKRRAIAT-243 HLA-B27 PB1 PB1/ RRAIATPGM-246 HLA-B27 PB1 PB1/ AIATPGMQI-248 HLA-B27 PB1 PB1/ TPGMQIRGF-251 HLA-A30 PB1 PB1/ LPVGGNEKK-279 HLA-A30 PB1 PB1/ GGNEKKAKL-282 HLA-A68 PB1 PB1/ GNEKKAKLA-283 HLA-B46 PB1 PB1/ EKKAKLANV-285 HLA-B35 PB1 PB1/ AKLANVVRK-288 HLA-A68 PB1 PB1/ KLANVVRKM-289 HLA-B58 PB1 PB1/ APIMFSNKM-348 HLA-B58 PB1 PB1/ IMFSNKMAR-350 HLA-B58 PB1 PB1/ MFSNKMARL-351 HLA-A3 PB1 PB1/ KMARLGKGY-355 HLA-B62 PB1 PB1/ ARLGKGYMF-357 HLA-B27 PB1 PB1/ LSPGMMMGM-411 HLA-A2 PB1 PB1/ SPGMMMGMF-412 HLA-A2 PB1 PB1/ MMMGMFNML-415 HLA-A2 PB1 PB1/ GMFNMLSTV-418 HLA-A2 PB1 PB1/ MFNMLSTVL-419 HLA-A2 PB1 PB1/ NMLSTVLGV-421 HLA-A2 PB1 PB1/ DGLQSSDDF-447 HLA-B27 PB1 PB1/ LQSSDDFAL-449 HLA-A68 PB1 PB1/ SSDDFALIV-451 HLA-B14 PB1 PB1/ INMSKKKSY-483 PB1 PB1/ TGTFEFTSF-495 PB1 PB1/ TFEFTSFFY-497 HLA-A1 PB1 PB1/ EFTSFFYRY-499 HLA-A2 PB1 PB1/ YRYGFVANF-505 HLA-A24 PB1 PB1/ FVANFSMEL-509 HLA-A2 PB1 PB1/ NFSMELPSF-512 HLA-A2 PB1 PB1/ LYNIRNLHI-606 HLA-A68 HLA-A1, A26, A30, A29 HLA-A3, A11, A31, A68 HLA-A03, A11, A31, A68 Protein Peptide ID Peptide sequence HLA restriction PB1 PB1/ DAVATTHSW-666 HLA-B40 PB1 PB1/ KRNRSILNT-677 HLA-A68 PB1 PB1/ EKFFPSSSY-705 HLA-A2 PB1 PB1/ KFFPSSSYR-706 HLA-B39 PB1 PB1/ FFPSSSYRR-707 HLA-B38 PB1 PB1/ EAMVSRARI-724 HLA-A29 M1 M1/1 29-EDVFAGKNT-37 HLA-A3 M1 M1/2 31-VFAGKNTDL-39 HLA-A2 M1 M1/3 37-TDLEALMEW-45 HLA-A30 M1 M1/4 49-RPILSPLTK-57 HLA-A3 M1 M1/5 51-ILSPLTKGI-59 HLA-A2 M1 M1/6 56-TKGILGFVF-64 HLA-A2 M1 M1/7 58-GILGFVFTL-66 HLA-A2 M1 M1/8 60-LGFVFTLTV-68 HLA-A2 M1 M1/9 66-LTVPSERGL-74 HLA-A2 M1 M1/10 68-VPSERGLQR-76 HLA-A2 M1 M1/11 71-ERGLQRRRF-79 HLA-A2 M1 M1/12 75-QRRRFVQNA-83 HLA-A2 M1 M1/13 76-RRRFVQNAL-84 HLA-A2 M1 M1/ GALASCMGL-130 HLA-B35 M1 M1/ ALASCMGLI-131 HLA-B35 M1 M1/ LASCMGLIY-132 HLA-B35 M1 M1/ SCMGLIYNR-134 HLA-A3, A11, A31, A33, A68, B27, B35 M1 M1/ NRMVLASTT-185 HLA-A3, A11 M1 M1/ MVLASTTAK-187 HLA-A3, A11 M1 M1/ VLASTTAKA-188 HLA-A3, A11 M1 M1/ LASTTAKAM-189 HLA-A3, A11 NP NP/1 88-DPKKTGGPI-96 HLA-A68 NP NP/2 89-PKKTGGPIY-97 HLA-A68 NP NP/3 147-TYQRTRALV-155 HLA-B15 NP NP/4 149-QRTRALVRT-157 HLA-A68 NP NP/5 158-GMDPRMCSL-166 HLA-2 NP NP/6 164-CSLMQGSTL-172 HLA-B15 NP NP/7 166-LMQGSTLPR-174 HLA-B15 NP NP/8 218-AYERMCNIL-226 HLA-B18 NP NP/9 225-ILKGKFQTA-233 HLA-B8 NP NP/ LKGKFQTAA-234 HLA-B15 NP NP/ SRNPGNAEI-253 HLA-B15 NP NP/ NAEIEDLIF-258 HLA-B40 NP NP/ LIFLARSAL-264 HLA-A2 NP NP/ IFLARSALI-265 HLA-A2 NP NP/ FLARSALIL-266 HLA-A3 NP NP/ YSLVGIDPF-304 HLA-B15 NP NP/ GVFELSDEK-470 HLA-B15 NP NP/ VFELSDEKA-471 HLA-B15 NP NP/ FELSDEKAT-472 HLA-B15 9

10 A B Supplementary Figure 1. Pre-existing influenza-specific T cell responses at baseline. Baseline cytokine secreting T cells to live virus (A) and summed total of CD8 epitopes from M1, PB1 and NP (B) for each individual measured by fluorescence-immunospot. Frequencies of antigenspecific cells were calculated by subtracting the average number of Spot Forming Cells (SFCs) in negative no peptide control wells from SFCs in antigen-containing test wells. Pie charts represent the average of the relative proportions of the total cytokine-secreting cellular response contributed by the three distinct cytokine-secreting subsets. The frequency of ph1n1 live virus-specific T cells (median 152 SFCs/million PBMCs; IQR: ; range ) was higher (p<0.001) than frequencies of T cells to conserved CD8 epitopes from PB1, NP or M1 proteins (median 56 SFCs/million; IQR: ; range 0-268). The majority of the live virus-specific response on flow-cytometry was from CD3+ T cells (data not shown). 10

11 Supplementary Figure 2. Pre-existing influenza-specific total-cytokine-secreting T-cells is not different between individuals who develop infection versus age and gender-matched controls that remain uninfected. Comparison of the baseline total-cytokine-secreting T cells to live virus measured by fluorescenceimmunospot between individuals who subsequently develop infection (n=43) and age and gendermatched individuals who remain uninfected (n=34). The line represents the median response for each group. 11

12 Supplementary Figure 3. Risk of developing influenza episode with fever as a function of frequency of pre-existing heterosubtypic influenza-specific cells (odds ratio for total cells 0.14, 95% CI ; odds ratio for IFN- + IL , 95% CI ). A logistic model was estimated for the probability of developing a fever during illness episode for each ten-fold increase in the number of SFCs/million PBMCs enumerated by fluorescence-immunospot in response to live virus stimulation for each cytokine-secreting subset. Each 10-fold increase in the frequency of ph1n1 virus-specific T-cells was associated with a 7-fold decrease in risk of developing influenza illness with fever (odds ratio for total cytokine-secreting T-cells: 0.14, 95% CI: and for IFN-γ + IL-2 - T-cells: 0.16, 95% CI ). Frequencies of IFN- + IL-2 + and IFN - IL-2 + T-cells did not predict risk (odds ratio for IFN- + IL-2 + cells 0.53,95% CI ; odds ratio for IFN - IL-2 + cells 0.55, 95% CI ). Despite fewer infected individuals than estimated in our sample size calculation, this substantially greater-than-estimated effect size and a higher prevalence of crossreactive T-cells enabled us to detect a difference. 12

13 SFC per million PBMC Cellular immune correlate of protection against symptomatic pandemic influenza Summed response to conserved CD8 peptide pools IFN- IL-2 - IFN- - IL-2 + IFN- IL-2 + Undepleted PBMC CD8 depleted Supplementary Figure 4. Responses to CD8 conserved epitopes are from CD8 + T-cells PBMCs undepleted or depleted of CD8+ T-cells by labelling with CD8 microbeads and passing through a LD column (MACS Miltenyi) were stimulated overnight with CD8 conserved epitope pools (Supplementary table 3). The purity of the depletion in each case was > 98% by flow cytometry staining. Frequencies of different cytokine secreting T-cell subsets were enumerated using fluorescence-immunospot (n=3) and responses were abrogated when CD8 depletion was undertaken. Bars show mean and SEM. 13

14 Supplementary Figure 5. Correlation of pre-existing cross-reactive CD4 + IFN-γ + IL-2 - cells and symptom score. Phenotypic characterisation using multi-parameter flow cytometry of the different memory subsets of influenza virus-specific CD4 + IFN-γ + IL-2 - cells based on CCR7 and CD45RA surface expression following overnight stimulation of PBMCs with live ph1n1 virus in ph1n1-infected individuals (n=22; in 3 of 25 infected individuals, samples were of insufficient quantity to undertake flow-cytometry). (A) Proportion of CD4 + IFN-γ + IL-2 - secreting cells that were effector memory (CD45RA - CCR7 - ), late effector (CD45RA + CCR7 - ), central memory (CD45RA - CCR7 + ) or naive (CD45RA + CCR7 + ) phenotype. (B) Correlation between the proportion of pre-existing CD3 + CD4 + IFN-γ + IL-2 - cells of the late effector CD45RA + CCR7 - subset and total symptom score. r indicates the Spearman rank correlation coefficient. We found no association of symptom score and the proportion of CD4 + central-memory CD45RA - CCR7 + (r=0.0734; p=0.75) or the effector-memory CD45RA - CCR7 - (r= ; p=0.56) T- cells. 14

15 Supplementary Figure 6. Heterosubtypic T-cell responses associated with prevention of viral shedding. Responses to live ph1n1 virus stimulation (A) were measured by fluorescence-immunospot allowing enumeration of IFN-γ+IL-2- (purple), IFN-γ-IL-2+ (green) and IFN-γ+/IL-2+ dual (orange) cytokinesecreting cells. Box plots represent responses compared between individuals with viral shedding and those not shedding virus. p values estimated by Mann-Whitney non-parametric t test. The difference in total cytokine secreting cells did not reach statistical significance (p=0.09). In the box plots, the box represents the third centile (75%) and first centile (25%) with the horizontal line in the box represents the median (50%). The whiskers represent 1.5 times the interquartile range with outliers shown. The scatter plots represent the frequency of cellular response for each individual. 15

16 Supplementary Figure 7. Gating strategy for flow cytometric analysis of CD8 + IFN-γ + IL-2 - cells. Responses to live ph1n1 virus stimulation were measured by multi-parameter flow cytometry. PBMCs stimulated with live ph1n1 virus for 18 hours were stained for surface markers of memory, lung homing, degranulation and intracellular cytokines. PBMCs were gated on live, single cells, low forward and side scatter for lymphoctes, a negative dump gate for CD56, CD14 and CD19 (not shown), CD3 + (gate not shown), IFN-γ + IL-2 - and CD8 + CD4 -. CD8 + cells of the CD45RA + CCR7 - subset were analysed for expression of CCR5, CD107 and TNF-α using Flowjo software. The Fluorescence Minus One (FMO) controls are shown for CCR7, CD45RA,CD107 and CCR5 which were used to set the gates to identify positive populations. 16

17 Supplementary Figure 8. CD45RA and CCR7 expression is not changed by in vitro stimulation. Phenotypic characterisation using multi-parameter flow cytometry of the different memory subsets of influenza virus-specific CD4 + and CD8 + T-cells based on CCR7 and CD45RA surface expression following overnight stimulation of PBMCs with live ph1n1 virus (open symbols) or unstimulated controls (closed symbols) in 43 individuals. (A) Proportion of CD3 + CD4 + T-cells of effector memory (CD45RA - CCR7 - ), late effector (CD45RA + CCR7 - ), central memory (CD45RA - CCR7 + ) or naive (CD45RA + CCR7 + ) phenotype. (B) Proportion of CD3 + CD8 + T-cells of effector memory (CD45RA - CCR7 - ), late effector (CD45RA + CCR7 - ), central memory (CD45RA - CCR7 + ) or naive (CD45RA + CCR7 + ) phenotype. Symbols represent responses for each individual with the horizontal line representing the median response. Statistical analysis by one-way ANOVA with Dunn s post-test comparisons showed no statistically significant differences in proportions of the different memory subsets between unstimulated and live virus stimulated donor responses. 17

18 References: 1. Casey R, Blumenkrantz D, Millington K, et al. Enumeration of functional T-cell subsets by fluorescence-immunospot defines signatures of pathogen burden in tuberculosis. PLoS One 2010;5:e Miller E, Hoschler K, Hardelid P, Stanford E, Andrews N, Zambon M. Incidence of 2009 pandemic influenza A H1N1 infection in England: a cross-sectional serological study. Lancet 2010;375: Zambon M, Hays J, Webster A, Newman R, Keene O. Diagnosis of influenza in the community: relationship of clinical diagnosis to confirmed virological, serologic, or molecular detection of influenza. Arch Intern Med 2001;161: Monto AS, Gravenstein S, Elliott M, Colopy M, Schweinle J. Clinical signs and symptoms predicting influenza infection. Arch Intern Med 2000;160: Forrest BD, Pride MW, Dunning AJ, et al. Correlation of Cellular Immune Responses with Protection against Culture-Confirmed Influenza Virus in Young Children. Clin Vaccine Immunol 2008;15: