Host adaptive immunity deficiency in severe pandemic influenza

Samples were stored at -80°C until cytokine, antibody and RNAm profiling. Attending to timing of seroconversion (production of antibodies against p2009A(H1N1)), day 9 from onset of symptoms was considered as the border between the innate and the adaptive immune response in the patients, establishing two moments in the evolution of the disease: an early phase (from onset of symptoms (day 0) to day 8) and a late phase (from day 9 and above). Patients were divided into two groups, depending on their respiratory status. The MV group needed invasive mechanical ventilation following admission; the NMV group was composed of those patients not needing mechanical ventilation at any moment during hospitalization. Cytokines, gene expression and viral load of MV patients were compared with those of NMV patients in both early and late phases separately. The number of samples analyzed in each phase is detailed in the Additional file 1.

Viral diagnosis was performed on RNA from pharyngeal swabs in the Microbiology Services of the participant hospitals by reverse transcription-polymerase chain reaction (RT-PCR)-based methods using reagents provided free of charge by the Centers for Disease Control (CDC, Atlanta, GA, USA) or purchased from Roche (Basel, Switzerland) (H1N1 detection set). These samples were also assessed by multiplex PCR (Luminex) with the xTAG RVP kit from Luminex-Abbott for coinfection with respiratory syncytial virus, influenza B virus, parainfluenza viruses 1-4, human metapneumovirus, enteroviruses, rhinovirus, adenovirus, bocavirus and coronaviruses NL63, HKU1, 229E, OC43, in accordance with the manufacturer's instructions. Viral load was quantified in both pharyngeal swabs and plasma in the Virology Lab of the WHO-associated center at Hospital Clinic in Barcelona, Spain, as detailed in Additional file 1. Oseltamivir resistance was directly detected in the initial positive pharyngeal swab by RT-PCR and sequencing of a 1296-bp fragment of the neuraminidase gene for the presence of the mutation H274Y by using an ABI 3130XL Genetic Analyzer.

HAI assays were performed on a 100-μl aliquot of the samples at University Health Network (UHN), Toronto, Ontario, Canada. The sera were treated with receptor-destroying enzyme (RDE) of V. cholerae by diluting one part serum with three parts enzyme and were incubated overnight in a 37°C water bath. The enzyme was inactivated by 30-min incubation at 56°C followed by the addition of six parts 0.85% physiological saline for a final dilution of 1/10. HI assays were performed in V-bottom 96-well microtiter plates (Corning Costar Co., Cambridge, MA, USA) with 0.5% turkey erythrocytes as previously described [14] using inactivated pandemic influenza A/California/07/2009(p2009A(H1N1)) antigens.

Microarrays were performed at University Health Network (UHN), Toronto, Ontario, Canada. More detailed explanation of microarray assays is provided in Additional file 1. Ingenuity Pathway Analysis 8.5 (IPA) (Ingenuity Systems, Redwood City, CA, USA) was used to select, annotate and visualize genes by function and pathway (gene ontology). IPA analysis identified those canonical pathways differentially expressed (P < 0.05) between comparison groups. Hierarchical clustering of those genes differentially expressed between groups by IPA analysis was performed using BRB-Array Tools v.3.8.1 stable release developed by Dr. Richard Simon and the BRB-array tools development team. Resulting microarray data sets have been uploaded at the GEO microarray data repository [GEO:GSE21802] [15]. We verified changes in microarray gene expression using quantitative real-time PCR (QRT-PCR) for representative genes from our analysis (Figure S1 in Additional file 2). Primers specific for human GAPDH mRNA were used to normalize samples.

The Mann-Whitney U test was employed for cytokine comparison purposes, since the Saphiro Wilk test evidenced absence of normal distribution of the data, and the Levene test demostrated absence of homogeneity of variance in both MV and NMV groups. Correlation studies between cytokine levels, gene expression levels, viral load and clinical parameters were done by calculating the Spearman correlation coefficients. All statistical tests were two-sided, and P < 0.05 was considered significant.

All patients were positive for p2009A(H1N1) at admission to the Intensive Care Unit (ICU), with absence of any other respiratory virus in the pharyngeal swabs. None of the patients had received the vaccine against p2009A(H1N1). All patients received oseltamivir therapy by the day of admission to ICU. None of the viral samples examined showed the mutation H274Y conferring resistance to oseltamivir. Twelve patients showed respiratory work severe enough for need of invasive mechanical ventilation at ICU admission (these patients were classified as MV group), while none of the remaining seven were mechanically ventilated during their hospitalization (these patients were classified as NMV group). The most common symptoms at onset were fever > 38°C (cases in MV, cases in NMV) (11, 6), myalgias (8, 6), cough (11, 7) and dyspnea (12, 7). Nine patients of the most severe group and six of the less severe ones had concomitant preexistent conditions (Table 1). None of the patients except one patient of the NMV group who was receiving treatment with bortezomib were under immunosuppressory therapy by the day of admission. Five of 12 MV patients, for four of seven patients in the NMV group, were receiving steroids at the time of sample collection (Table 1). Seven MV patients failed to recover from disease and died with a mean illness duration time of 16.7 and 5.8 days (mean, SD). All NMV patients recovered from p2009A(H1N1) disease. The leading cause of death was primary respiratory failure with refractory hypoxemia in five patients and multiorganic failure in the remaining two patients (Table S1 in Additional file 3). While no differences were found in viral load between MV and NMV patients in the early stage of the disease, MV patients showed significantly higher viral loads than NMV in pharynx in the late stage of the disease (P < 0.05) (Figure 1). Three MV patients and one NMV showed detectable viremia at the day of ICU admission, with undetectable virus in plasma afterward. Viral load in pharynx showed a direct correlation with SOFA score (r = 0.4) and an inverse one with O₂ saturation (r = -0.3) during the course of the disease. Five MV patients suffered from a bacterial or fungal superinfection at some point during hospitalization (Table 1). Three patients suffering from bacterial superinfection died (Table S1, Additional file 3). Remarkably, none of the patients in the NMV group suffered from bacterial superinfection. All but three patients (two MV and one NMV) had produced antibodies (HAI titers > 1/40) by the last day of sample collection. Seroconversion took place 9.7 (4.6) days from the onset the symptoms in the MV group and 8.8 (1.6) days in the NMV group (mean, SD), with no significant difference between the two groups.

Comparison of gene expression profiles between MV and NMV patients in the early phase of the disease revealed the absence of differentially expressed genes between both groups. On the other hand, comparisons in the late phase of the disease revealed 4559 genes differentially expressed between MV and NMV patients (P < 0.05, FDR = 0.06). IPA analysis identified in this late phase a significant depression of a group of intracellular signaling pathways important for the development of the antiviral immune response in the most severe group of patients (MV) compared to NMV group (Figure 2). A number of genes involved in antigen presentation (Figure 3), B cell development (Figure S2 in Additional file 4), CD28 signaling in T helper cells (Figure S3 in Additional file 5), granzyme B signaling (Figure S4 in Additional file 6), T helper cell differentiation, protein ubiquitination, dendritic cell maturation, apoptosis and B-cell receptor (BCR) signaling were differentially regulated between these two groups and significantly lower in the MV group. On the contrary, the MV group showed significantly higher expression of the IL-6 and IL-10 signaling pathways (Figures S5 and S6 in Additional files 7 and 8). Tables S2-S9 (Additional files 9, 10, 11, 12, 13, 14, 15 and 16) show the genes differentially expressed between groups classified by individual pathway. IPA did not identify any signaling pathway differentially expressed between the seven MV patients with fatal outcome and the remaining five who survived the infection. Expression levels of HLA-DMA, HLA-DRB3, HLA-DRB4, HLA-DQA1, HLA-DRA1, HLA-DMB, HLA-DPA1, CD4, CD8A and CD8B showed a negative correlation with viral load in pharynx during the late phase of disease (P < 0.05, r ≤ (-0.4)). These genes, with the exception of HLA-DRB4, correlated positively with O₂ saturation in this phase (P < 0.05, r ≥ 0.5). They showed also a negative association with the SOFA score for severity, as also did CD74 and HLA-C (P < 0.05, r ≤ (-0.5)).

NMV and MV groups did not show major differences in the cytokine and chemokine profiles during the early phase of disease (Table S10 in Additional file 17). However, in the late phase, MV patients showed significantly higher levels of the chemokines IP-10 (CXCL10), IL8 (CXCL8) and MCP-1 (CCL2) (Figure 4 and Table S11 in Additional file 18). During the late phase, MV patients showed higher levels of two key Th1 cytokines (IL-12p70 and IFN-γ), IL6 (a Th17 related cytokine) and also two cellular growth factors (VEGF, GM-CSF) than the less severe patients (NMV). Also during the late phase, MV patients showed increased levels of two immunomodulatory cytokines (IL-10, IL-1ra) (Figure 4 and Table S11 in Additional file 18). Levels of IP-10, IL-6, and IL-8 showed a positive correlation with viral load in pharynx during the course of the disease (P < 0.05, r coefficient < 0.5). IL-6, IL-8, IL-10, IL-15, IL-12p70, GM-CSF and IFN-γ showed a positive correlation with the SOFA score for severity (P < 0.05, r coefficient < 0.5). IL-10 levels showed an inverse association with HLA-DRB3, HLADPB1 and CD74 expression levels; IL-1ra showed an inverse association with HLA-C expression levels (P < 0.05, r ≤ (-0.5)). IFN-α, IFN-λ (IL-28) and IL-23 were undetectable in the vast majority of the patients in both groups along the course of the disease.