Indication for a role of regulatory T cells for the advent of influenza A (H1N1)-related pneumonia

Abstract

Regulatory T cells (T_regs) have an anti-inflammatory role. A former study in a limited number of patients found that absolute counts of T_regs increase when infection by the new influenza H1N1 virus is complicated with pneumonia. These results generate the question if H1N1-related pneumonia is associated with a state of hypo-inflammation. A total of 135 patients were enrolled with blood sampling within less than 24 h from diagnosis; 23 with flu-like syndrome; 69 with uncomplicated H1N1-infection; seven with bacterial pneumonia; and 36 with H1N1-related pneumonia. T_regs and CD14/HLA-DR co-expression were estimated by flow cytometry; concentrations of tumour necrosis factor-alpha (TNF-α), of interleukin (IL)-6 and of soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) by an enzyme immunoassay; those of procalcitonin (PCT) by immuno-time-resolved amplified cryptate technology assay. Expression of human leucocyte antigen D-related (HLA-DR) on monocytes was similar between groups; absolute T_reg counts were greater among patients with H1N1-related pneumonia than flu-like syndrome or H1N1-uncomplicated infection. Serum TNF-α of patients with bacterial pneumonia was greater than those of other groups, but IL-10 was similar between groups. Serum PCT was greater among patients with H1N1-related pneumonia and sTREM-1 among those with H1N1-related pneumonia. Regression analysis revealed that the most important factors related with the advent of pneumonia were the existence of underlying illnesses (P = 0·006) and of T_regs equal to or above 16 mm³ (P = 0·013). It is concluded that the advent of H1N1-related pneumonia is related to an early increase of the absolute T_reg counts. This increase is probably not part of a hypo-inflammatory state of the host.

Introduction

Over the last year our world has been facing the pandemic of influenza H1N1 viral infection, with immense effects on daily lives and on the world economy. The new H1N1 virus has higher transmissibility than seasonal influenza. Clinical presentation differs considerably compared with pandemics caused by previous influenza viruses. In general, affected patients have a mild clinical course but most severe cases appear among young patients. Several factors have been defined to predispose to more severe illness, namely obesity, bronchial asthma, diabetes mellitus type 2 and pregnancy [1–3]. It is estimated that almost 200 000 deaths were caused during the course of this pandemic [4].

The pathogenesis of infection by the new H1N1 virus is still a matter of debate, and the mechanism by which some people develop serious pulmonary infection leading to acute lung injury (ALI) or adult respiratory distress syndrome (ARDS) and eventually die remains to be answered. A recent study by our group enrolled 31 patients with infection by the H1N1 virus; six of them had pneumonia [5]. It was shown that infection by the new H1N1 virus was accompanied by defective cytokine responses of blood monocytes, mainly of tumour necrosis factor (TNF)-α and interferon (IFN)-γ, after stimulation with heat-killed Streptococcus pneumoniae. It was also found that the absolute counts of regulatory T cells (T_regs) of patients with H1N1-related pneumonia were elevated significantly compared with patients infected by H1N1 but who did not progress to pneumonia.

The latter findings may lead to the hypothesis that in the event of H1N1 infection, particularly when pulmonary complications arrive, anti-inflammatory responses predominate. This hypothesis is based upon current knowledge that (i) T_regs blunt proinflammatory responses [6]; and (ii) blood monocytes of patients infected by H1N1 have defective cytokine responses to ex vivo stimulation [5]. The current study was undertaken to investigate further whether or not anti-inflammatory responses predominate during lung involvement of patients infected by the new H1N1 virus. Expression of human leucocyte antigen D-related (HLA-DR) on blood monocytes and absolute counts of T_regs were estimated early in a cohort of patients with H1N1 infection. This was correlated with early serum levels of the proinflammatory cytokine TNF-α and of the anti-inflammatory cytokine IL-10. Furthermore, serum levels of soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) were measured. sTREM-1 is the soluble shed counterpart of the proinflammatory TREM-1 receptor which is highly activated during inflammatory responses [7]. sTREM-1 appears to behave as an anti-inflammatory mediator; its measurement was considered a marker of the anti-inflammatory state of H1N1 infected patients.

Patients and methods

Study design

The study was performed during the period September 2009 to January 2010. The study protocol was approved by the Ethics Committee of the ATTIKON University Hospital of Athens. All patients admitted to the emergency department for flu-like symptoms and who provided written informed consent were eligible. Only patients with core temperature greater than 38·5°C were screened further for enrolment. All the enrolled patients were different to those enrolled in a previous study by our group [5].

Inclusion criteria were: (i) written informed consent; (ii) age more than or equal to 18 years; (iii) symptoms compatible with infection by H1N1 as already defined [8]; (iv) core temperature above 38·5°C; and (v) start of symptoms within the last 24 h. Exclusion criteria were: (i) deny to consent; (ii) known infection by the human immunodeficiency virus (HIV); (iii) neutropenia defined as an absolute neutrophil count equal to or below 500 neutrophils per cubic millimetre (mm³) of blood; and (iv) oral intake of corticoids defined as more than 1 mg/kg of body weight of equivalent prednisone for more than 1 month.

Enrolled patients underwent a detailed work-out comprising case history, thorough physical examination, chest X-ray if considered necessary, white blood cell count, blood gas analysis, urine analysis for detection of leucocytes and nitrate and urine detection of antigens of S. pneumoniae and Legionella pneumophila. If the above work-out failed to disclose the presence of bacterial infection, a pharyngeal smear was collected with a swab. Infection by the H1N1 virus was diagnosed by real-time reverse transcriptase–polymerase chain reaction (rRT–PCR) detecting the presence of H1N1 RNA in the infected cells [9]. A patient was considered to present with pneumonia due to the new H1N1 virus when a local infiltrate was diagnosed by chest X-ray and H1N1 RNA was detected in the pharyngeal swab, as defined elsewhere [10,11]. Diagnosis of ALI and of ARDS was performed according to well-established definitions [12,13]. Patients were followed-up until complete resolution of symptoms.

In parallel with the collection of the pharyngeal smear, 10 ml of whole blood were collected after venipuncture of one forearm antecubital vein under sterile conditions; 5 ml were collected into ethylenediamine tetraacetic acid (EDTA)-coated tubes (Vacutainer, Becton Dickinson, Cockeysville, MD, USA); another 5 ml were collected into pyrogen-free tubes (Vacutainer). The latter tubes were centrifuged and serum was kept refrigerated at –70°C until assayed for cytokines.

Laboratory procedure

Red blood cells were lyzed in EDTA–whole blood with ammonium chloride 1·0 mM. White blood cells were washed three times with phosphate-buffered saline (PBS, pH 7·2) (Merck, Darmstadt, Germany) and subsequently incubated for 15 min in the dark with the monoclonal antibodies anti-CD14 at the flurochrome fluorescein isothiocyanate (FITC, emission 525 nm; Immunotech, Marseille, France); anti-HLA-DR at the fluorochrome phycoerythrin (PE, emission 575 nm; Immunotech); and anti-CD45 at the fluorochrome PC5 (emission 650 nm; Immunotech). Absolute counts of T_regs were estimated with the PE anti-human forkhead box P3 (FoxP3) staining set (eBioscience Inc., San Diego, CA, USA). Cells were analysed after running through the EPICS XL/MSL flow cytometer (Beckman Coulter Co., Miami, FL, USA) with gating for lymphocytes or monocytes based on their characteristic forward-/side-scatter (FS/SS). Flow cytometric results were also assessed in blood collected from 10 healthy volunteers.

Concentrations of TNF-α, IL-10 and sTREM-1 were estimated in serum in duplicate by an enzyme immunoassay (R&D Systems, Minneapolis, MN, USA). The lowest detections limits were: for TNF-α 80 pg/ml; for IL-10 25 pg/ml; and for sTREM-1 15·1 pg/ml.

PCT was estimated in serum in duplicate by an immuno-time-resolved amplified cryptate technology assay (Kryptor PCT; BRAHMS GmbH, Henningdorf, Germany) with a functional assay sensitivity of 0·006 ng/ml.

Statistical analysis

Subjects were divided into four groups: flu-like syndrome; uncomplicated H1N1-infection; bacterial pneumonia; and H1N1-related pneumonia. Results for CD14/HLA-DR co-expression and for T_regs were expressed as means ± standard error (s.e.); those of serum cytokines by their median and 95% confidence intervals (CI). Comparisons between groups were compared by one-way analysis of variance (anova) with post-hoc Bonferroni analysis for the avoidance of random correlations. Comparisons for serum cytokines comprised outlier and extreme values. Logistic regression analysis was performed to estimate the risk factors for the advent of H1N1-related pneumonia. Advent of H1N1-related pneumonia was considered as the dependent variable; presence of underlying diseases, absolute T_regs counts equal to or higher than 16 mm³ and sTREM-1 higher than 180 pg/ml were considered as independent variables. Obesity, chronic obstructive pulmonary disease (COPD), bronchial asthma and chronic heart failure (CHF) were considered important underlying diseases as defined in previous publications [12,13]. Selection of 16 mm³ for T_regs was performed after designing a receiver operator curve with 90% sensitivity to differentiate patients with H1N1 infection and pneumonia from patients with H1N1 infection without pneumonia. Selection of sTREM-1 greater than 180 pg/ml was performed based on previous publications defining that cut-off as protective for the physical course of sepsis [14,15]. Odds ratios (ORs) and 95% CIs were estimated. Any value of P below 0·05 was considered significant.

Results

Demographic characteristics

A total of 135 patients were enrolled over the study period. Twenty-three patients were suffering from flu-like syndrome; 69 from uncomplicated H1N1-infection; seven from bacterial pneumonia; and 36 with H1N1-related pneumonia. From the latter group of patients three were presented with ALI and two with ARDS. Their demographic characteristics are shown in Table 1. Infection convalescence was noted in all enrolled patients. S. pneumoniae was identified as the causative pathogen for all seven patients with bacterial pneumonia. It was isolated from blood in two patients; it was isolated from sputum in one patient; another five patients had positive urine antigen detection.

Table 1.

Demographic characteristics of patients enrolled in the study.

	Flu-like syndrome	Uncomplicated H1N1-infection	Bacterial pneumonia	H1N1-related pneumonia
Number	23	69	7	36
Male/female	13/10	31/38	5/2	21/15
Age (mean ± s.d., years)	40·7 ± 16·9	31·4 ± 11·8	52·1 ± 19·1	39·5 ± 15·1
White blood cells (mean ± s.d./mm3)	8245·5 ± 4497·2	6448·9 ± 2947·3	10 457·1 ± 5 617·5	7521·6 ± 5216·0
Streptococcus pneumoniae as a pathogen (number, %)	0 (0)	0 (0)	7 (100)	0 (0)
Underlying conditions (number, %)
Obesity	0 (0)	6 (8·7)	0 (0)	12 (33·3)
Diabetes mellitus 2	1 (4·3)	1 (1·4)	1 (14·3)	2 (5·6)
Chronic heart failure	0 (0)	0 (0)	0 (0)	2 (5·6)
Chronic obstructive pulmonary disease	0 (0)	0 (0)	1 (14·3)	1 (2·8)

s.d.: standard deviation.

From the 10 healthy volunteers four were male and six were female. Their mean ± s.d. age was 34·5 ± 5·4 years.

Immunophenotyping

Co-expression of HLA-DR on monocytes did not differ between patients (Fig. 1). Absolute counts of CD14⁺/HLA-DR⁺ cells were greater among all four groups of enrolled patients compared with healthy volunteers (P of comparisons by anova 0·018).

Fig. 1.

Expression of human leucocyte antigen D-related (HLA-DR) on monocytes and regulatory T cells (T_regs) in the enrolled cohort of patients: 23 patients with flu-like syndrome; 69 patients with uncomplicated H1N1-infection; seven patients with bacterial pneumonia and 36 patients with H1N1-related pneumonia. Comparisons between patients are shown; n.s.: non-significant.

The percentage of T_regs did not differ between patients. Absolute counts of T_regs were greater among all four groups of enrolled patients compared with healthy volunteers (P of comparisons by anova< 0·0001). Absolute counts of T_regs were greater among patients with H1N1-related pneumonia than either patients with flu-like syndrome or patients with H1N1-uncomplicated infection.

Absolute counts of T_regs did not differ between patients with bacterial pneumonia and those with H1N1-related pneumonia. Mean ± s.e. absolute counts of T_regs of 31 patients with H1N1-related pneumonia without ALI/ARDS were 19·1 ± 2·4/mm³; those of five patients with H1N1-related pneumonia and ALI/ARDS were 23·7 ± 4·6/mm³[P = not significant (n.s.)].

Comparison of absolute counts of T_regs between patients with uncomplicated H1N1 infection and patients with H1N1-related pneumonia in relation to their age quartiles are shown in Table 2. Absolute counts were greater among patients with H1N1-related pneumonia compared with patients with uncomplicated H1N1 infection of the first and second quartiles of age.

Table 2.

Absolute counts of regulatory T cells (T_regs) among patients with uncomplicated infection by the new H1N1 virus and among patients with H1N1-related pneumonia. Numbers in brackets respond to the number of patients at every age quartile.

	Tregs (mean ± s.e./mm3)
Age quartiles (years)	Uncomplicated H1N1 infection	H1N1-related pneumonia	P
< 24	6·9 ± 1·4 (23)	16·0 ± 6·0 (6)	0·027
25–38	17·1 ± 2·4 (17)	21·3 ± 4·7 (9)	0·028
39–48	12·3 ± 2·1 (19)	16·5 ± 5·7 (9)	n.s.
> 48	14·7 ± 7·7 (9)	19·8 ± 3·6 (12)	n.s.

n.s.: non-significant; s.e.: standard error.

Serum cytokines (Fig. 2)

Fig. 2.

Serum concentrations of tumour necrosis factor (TNF)-α, interleukin (IL)-10, procalcitonin (PCT) and soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) in the enrolled cohort of patients: 23 patients with flu-like syndrome; 69 patients with uncomplicated H1N1-infection; seven patients with bacterial pneumonia and 36 patients with H1N1-related pneumonia. Asterisks denote outliers and circles denote extremes. Dashed lines demonstrate lower limits of detection. Numbers in brackets respond to the respective number of patients with serum concentration below the limit of detection. Comparisons between patients are shown; n.s.: non-significant.

Concentrations of TNF-α in serum of patients with bacterial pneumonia were higher than those of other groups. Serum IL-10 did not differ between groups.

PCT has been considered a biomarker indicative of a systemic infection. Studies among patients with sepsis have indicated that serum PCT increase along with disease severity [16]. Concentrations of serum PCT of patients with H1N1-related pneumonia were greater than those with bacterial pneumonia.

sTREM-1 is the shed counterpart of the proinflammatory TREM-1 receptor. It is proposed to behave as an anti-inflammatory mediator [7]. Six patients with flu-like syndrome, 34 patients with uncomplicated H1N1 infection, three patients with bacterial pneumonia and 20 patients with H1N1-related pneumonia had sTREM-1 in serum below 180 pg/ml. Six patients with flu-like syndrome, 35 patients with uncomplicated H1N1 infection, four patients with bacterial pneumonia and 16 patients with H1N1-related pneumonia had sTREM-1 in serum above 180 pg/ml (P = n.s.).

Factors involved with H1N1-related pneumonia

Analysis comprised the total of enrolled patients infected by the new H1N1 virus. It was found that the most important factors related independently to the advent of pneumonia were the existence of underlying illnesses (OR: 18·89, 95% CI: 2·35–151·46, P = 0·006) and of T_regs equal to or above 16 mm³ (OR: 1·11, 95% CI: 1·02–1·21, P = 0·013). Serum sTREM-1 greater than 180 pg/ml was not related to the advent of H1N1-related pneumonia (OR: 1·00, 95% CI: 0·99–1·00, P = 0·75).

Discussion

The emerging flu pandemic by the new H1N1 virus creates considerable dilemmas in all health-care authorities about the real threat for the human host. Death supervenes as a result of severe lung infection aggravated mainly by ALI or ARDS. The question remains unanswered as to what is the main driver leading patients infected by the new H1N1 virus to the development of pneumonia. A reply to that question is extremely urgent, as a high percentage of patients progressing to pneumonia are young [1–3]. Various epidemiological associations have indicated a positive role for various underlying diseases, mainly COPD, bronchial asthma, obesity and CHF [1,12,13].

Current knowledge from critically ill patients with severe sepsis indicates a major role for the derangement of normal immune responses of the host for the advent of ALI/ARDS. More precisely, as sepsis progresses proinflammatory T helper type 1 (Th1) responses are counterbalanced by anti-inflammatory Th2 responses [17]. The TNF-α/IL-10 ratio has been proposed as an indicator for the existing imbalance between pro- and anti-inflammatory responses of the septic host [18]. It seems that during the hypo-inflammatory state of sepsis the potency of blood monocytes for antigen presentation is severely diminished, as indicated by the decreased expression of HLA-DR on their cell surfaces [19]. At the same time, T_reg responses are activated, contributing further to this hypo-inflammatory state leading to multiple organ dysfunctions [6].

A recent study by our group of a limited number of patients infected by the H1N1 virus has shown increased counts of T_regs upon advent of H1N1-related pneumonia. Blood monocytes in these patients had blunted TNF-α responses after stimulation by S. pneumoniae[5]. These findings constitute considerable similarity to the hypo-inflammatory state of the septic host, creating the need to provide a reply as to whether predominance of anti-inflammatory responses is a major driver for H1N1-related pneumonia.

Results revealed that in the event of pneumonia patients infected by the new H1N1 virus are not under a fully defined hypo-inflammatory state. T_regs predominate at counts similar to those observed in bacterial pneumonia but the TNF-α/IL-10 ratio is not decreased. Moreover, expression of HLA-DR on monocytes remains intact.

Regression analysis showed clearly that elevation of T_regs cannot explain solely why some patients infected by the new H1N1 virus develop pneumonia and others do not. Although the increase of T_regs imposes considerably on the advent of pneumonia, the effect of underlying diseases is even greater. There are two main limitations of the present study that prevent distinguishing any cause and effect relationship between increase of T_regs and development of H1N1-related pneumonia: (i) T_regs in the bronchoalveolar fluid (BAL) of patients with pneumonia were not estimated. As a consequence, the existence of trafficking between T_regs in the lung and T_regs in blood cannot be defined; and (ii) functional analysis of circulating T_regs was not performed. This would define if T_regs were hyperfunctional, in order to confer anti-inflammatory responses, or hypo-functional, in order to confer proinflammatory responses. However, it should be underscored that absolute counts of T_regs were greater among patients with H1N1-related pneumonia compared with those with uncomplicated H1N1 infection when analysis focused upon patients of the first two quartiles of age distribution.

This is not the first study assessing serum levels of pro- and anti-inflammatory cytokines in the sera of patients with type A influenza infection. In an earlier study of 39 patients with either influenza A or B in Hong Kong during the first semester of 2006, i.e. well before the arrival of the new H1N1 virus, concentrations of proinflammatory cytokines were estimated in plasma of patients [20]. Cytokine serum levels were also estimated in a recently published cohort of 74 patients infected by the new H1N1 virus [21]; 23 developed ARDS; 14 were presented with oxygen desaturation; and 37 were presented with mild disease. Concentrations of TNF-α, IL-6 and IL-10 were higher among non-survivors than survivors. However, in both the above studies, serum concentrations of TNF-α and IL-10 were within the range of those reported in the present manuscript.

Apart from clinical studies in humans, one recent study in mice infected by a clinical isolate of H3N2 influenza A virus revealed that the key element for the earlier resolution of pneumonia was a potent proinflammatory response of the host. More precisely, mice were pretreated with an inactivated isolate of Haemophilus influenzae. Pretreatment resulted in prolongation of survival after infection by H3N2 and by increased concentrations of proinflammatory cytokines, namely TNF-α, IL-10 and IFN-γ, in the BAL of mice [22].

In the present study serum levels of sTREM-1 were also estimated. TREM-1 is a transmembrane pattern recognition receptor embedded on cell membranes of blood neutrophils and blood monocytes. This receptor is highly activated during serious bacterial infections. Once activated, production of TNF-α and of IL-8 is stimulated significantly [23]. The extracellular part of this receptor is shed into the systemic circulation, a molecule known as sTREM-1, and it is considered to behave as an anti-inflammatory mediator by binding to the TREM-1 agonist [7]. As a consequence sTREM-1 is an indirect estimation of the activated TREM-1. Serum levels of sTREM-1 greater than 180 pg/ml are indicative of favourable outcome among patients with severe sepsis [14,15]. Serum concentrations of sTREM-1 were detectable in most of the enrolled patients; however, they were not particularly pronounced among patients with H1N1-related pneumonia. Serum concentrations were mainly pronounced in patients with bacterial pneumonia.

Serum PCT was also estimated in the present study and it was found to be greater among patients with H1N1-related pneumonia than among patients with bacterial pneumonia. PCT is indicative of the severity of the systemic inflammatory response. Concentrations in serum among patients with sepsis increase along with disease severity [24–26]. Serum levels of PCT of patients with severe sepsis or septic shock reported in previous publications [27] are much greater than those described in the present study for H1N1-related pneumonia. This difference in PCT levels between H1N1-related pneumonia and severe sepsis probably implies that the systemic inflammatory reaction of severe lung infection by the H1N1 virus is not similar to that of sepsis. In any case, the increase of serum PCT and the lack of increase of the anti-inflammatory sTREM-1 among patients with H1N1-related pneumonia compared with patients with uncomplicated H1N1 infection probably implies that a proinflammatory response predominates in the former patients compared with the latter.

The results presented reveal clearly that the advent of H1N1-related pneumonia is related to an early increase of the absolute counts of T_regs. This increase is not part of a hypo-inflammatory state of the host, as other indicators of anti-inflammation such as expression of HLA-DR on CD14-monocytes and the TNF-α/IL-10 ratio do not participate (Fig. 3). These results may add considerably to our knowledge for the pathogenesis of H1N1-related pneumonia.

Fig. 3.

Summary of key findings for both spectra of infection by the new H1N1 virus i.e. uncomplicated infection and H1N1-related pneumonia. ALI: acute lung injury; ARDS: acute respiratory distress syndrome; COPD: chronic obstructive pulmonary disease; CHF: chronic heart failure; IL-10: interleukin-10; PCT: procalcitonin; sTREM-1: soluble triggering receptor expressed on myeloid cells-1; TNF-α: tumour necrosis factor-alpha; T_regs: regulatory T cells. Symbols: ↑: moderately increased; ↑↑↑: markedly increased; (−): not increased; (++): substantial additive effect.

Disclosures

None of the authors has any conflict of interest related to this study.