During the months of March and April 2009, the identification of a new serotype of the influenza A (H1N1) virus in Mexico and the United States of America1 marked the beginning of the first pandemic to be declared by the World Health Organization (WHO) in the 21st century. The WHO registered more than 625,000 cases worldwide, with 15,174 deaths.2 Although it presented a global mortality rate similar to that of the seasonal influenza virus, this new strain of influenza A (H1N1) virus has some particularities that make it a specific challenge for the intensivist. First, certain groups of patients, who are usually not at risk of complications or death from seasonal influenza, are at special risk with the influenza A (H1N1) virus, including pregnant or postpartum women, asthmatic or obese patients, and young subjects.3–5 Second, this virus has a special tropism for the lower airways. Accordingly, acute viral pneumonitis, with or without bacterial superinfection, and airflow limitation occur frequently.6,7 For the most part, patients with severe influenza A (H1N1) are admitted to hospital for respiratory failure that progresses rapidly, usually within one day, into severe respiratory failure necessitating admission to the intensive care unit (ICU) and rescue ventilatory strategies.3,7,8 Such patients present with moderate to severe acute respiratory distress syndrome (ARDS) associated with high morbidity and mortality rates (15-40%).3,9,10

These pulmonary and remote alterations present many similarities to “bacterial” sepsis, even though they may occur in the absence of bacterial infection. The H1N1 infection is associated with a strong pulmonary inflammatory response with increased production of interleukin (IL)-10, IL-4, and interferon-gamma (IFN-γ).11 This pro-inflammatory response is not restricted to the lungs. Recently, Bermejo-Martin et al.12 reported that systemic levels of several pro-inflammatory cytokines were increased in patients hospitalized with H1N1 infection, especially in those who were critically ill. Remote organ dysfunction may also be observed, including elevated liver enzymes, rhabdomyolysis, renal failure, and shock.9

In sepsis, severe microcirculatory alterations have been observed.13 These alterations may have important implications, as the microcirculation is the principal site of oxygen exchange between blood and underlying tissues. Additionally, anomalies of the microcirculatory network have been consistently demonstrated as being associated with organ dysfunction and mortality.14,15 However, microcirculatory alterations have not yet been reported in patients with viral infections. We hypothesized that microcirculatory abnormalities may be present in patients with severe influenza A (H1N1) infection, participating in local and remote tissue injury. During the H1N1 pandemic in the fall of 2009, we evaluated the microcirculation of critically ill patients admitted to hospital with severe respiratory failure associated with influenza-like syndrome in order to evaluate the degree of microcirculatory dysfunction in patients with severe influenza A (H1N1) infection.

Patients and methods

All patients with acute respiratory failure associated with influenza-like illness who were admitted to the 35-bed ICU of a teaching hospital from October to December 2009 were screened for inclusion in the study. Influenza-like illness was defined as two or more of the following symptoms: fever, chills, myalgia, arthralgia, headache, and sore throat and/or cough in the absence of another known cause during the epidemic period of H1N1 infection. Acute lung injury (ALI) was defined as PaO2/F i O2 below 300 in the absence of signs of raised left atrial pressure.16,17 These patients were subsequently screened for influenza A (H1N1) virus infection in the upper and/or lower respiratory secretions (viral culture, polymerase chain reaction, or immunofluorescence) and classified as having confirmed, probable, or suspected infections according to the WHO and the Canadian National Microbiology Laboratory definitions (Canadian Critical Care Trials Group, 9 A.D. 29 /id;Public Health Agency of Canada, 2009 30 /id). Exclusion criteria included age < 18 yr, psychomotor agitation, reliance on non-invasive ventilatory support, oral bleeding, and H1N1 infection diagnosed for > 96 hr. The study was approved by the Erasme Hospital Ethical Committee, and all patients or their spouse or children gave informed consent.

General measurements

On the day of microcirculatory assessment, severity of illness was assessed using the Sequential Organ Failure Assessment (SOFA) score,18 the Acute Physiology and Chronic Health Evaluation (APACHE II) score,19 and the Lung Injury Score (LIS).20 Length of ICU stay, length of hospital stay, ICU mortality, 28-day mortality, and hospital mortality were recorded.

We also recorded global hemodynamic parameters, demographic data, major comorbidities, variables (in terms of respiration, mechanisms, and oxygenation), blood serum markers of organ function, and the presence of bacterial superinfection. In addition, we noted organ support therapy, such as vasopressors, renal replacement therapy, mechanical ventilation, and related adjuvant therapies (nitric oxide inhalation, steroids, prone position, recruitment maneuvers, and extracorporeal membrane oxygenation) received during the ICU stay.

Microcirculatory assessment

The microcirculation was assessed using sidestream dark field (SDF) imaging (MicroScan, MicroVision Medical, Amsterdam, The Netherlands) and near-infrared spectroscopy (NIRS) (InSpectra 650, Hutchinson, MN, USA).

The SDF imaging technique has been presented elsewhere in detail.13 In brief, five sequence images of 20 sec each were recorded from the lateral sublingual area by a video camera (5 X objective), stored and analyzed offline. The density, the proportion of perfused vessels, and the microvascular flow index (MFI) of the vessels were quantified using the semi-quantitative method.21

The NIRS technique has been detailed elsewhere.22 Briefly, a probe was placed over the skin of the thenar eminence to register the baseline tissue oxygen saturation (StO2) and the total hemoglobin index (THI), i.e., an estimate of the local tissue hemoglobin concentration. A three-minute vascular occlusive test was then performed with an arterial cuff around the arm inflated up to 50 mmHg above the systolic arterial pressure. The rate of tissue oxygen consumption (linear regression of the downslope) and the tissue oxygen consumption at the end of the first minute of ischemia (NirVO2) were measured during the ischemic phase. The vasoreactivity response to hypoxia was measured as the linear regression of the incremental StO2 upslope that followed the arterial cuff release. Finally, the difference between the maximal StO2 and THI following the reperfusion and the baseline values of these variables were computed as ΔStO2 and ΔTHI.23

Statistical considerations

Continuous variables are depicted as median (interquartile range, 25-75%) or as mean ± standard deviation (SD), and categorical variables are depicted as number (%). Parametric (Pearson correlation) or non-parametric (Spearman test) linear regression was performed to evaluate the relationships between microcirculation (NIRS and SDF imaging) variables and indices of severity.

Results

Among the 30 patients with influenza-like syndrome who were admitted to the ICU during the study period, seven patients were included for microcirculatory evaluation. Twenty-three patients were excluded for use of non-invasive mechanical ventilation (n = 16) precluding investigation of the sublingual microcirculation, age < 18 yr (n = 2), time from diagnosis > 96 hr (n = 2), and missed cases (n = 3). Three of the included patients had confirmed H1N1 infection, one probable and three suspected. At the very least, one comorbidity was found in each of the patients: immunosuppression (n = 4), pregnancy (n = 2), diabetes (n = 2), heart disease (n = 2), chronic obstructive pulmonary disease (n = 1), neoplasia (n = 1), chronic renal failure (n = 1), and cirrhosis (n = 1). Three patients had bacterial superinfection. All patients were on antibiotics and six were receiving oseltamivir at the time of the microcirculatory assessment—in the remaining patient, therapy was initiated just after the microcirculatory assessment. The main clinical and laboratory characteristics of the patients are detailed in Table 1. Three patients were in shock, requiring norepinephrine administration with a mean dose of 0.57 ± 0.55 μg·kg−1·min−1. The main hemodynamic data are shown in Table 2.

Table 1 General medical characteristics of the patients who were studied on the day of the microcirculatory assessment
Full size table
Table 2 Ventilatory and oxyhemodynamic variables at the time of the microcirculatory assessment
Full size table

All patients received mechanical ventilatory assistance and fulfilled the ALI/ARDS criteria by way of low lung compliance, hypoxemia, and high LIS. Adjuvant therapy for ARDS was frequently applied, particularly steroids, which were reserved for the most severe cases (Table 2). At the time of microcirculatory assessment, the median LIS of all patients was 2.50 (2.25-3.25) points.

The microcirculatory assessment was performed 21 (7-45) hr after admission to the ICU. Both SDF imaging and NIRS revealed that the microcirculation was moderately to severely compromised (Table 3 and Figure). The proportion of perfused small vessels was 66% (60-86%; normal > 90%) and the MFI was 1.9 points (1.3-2.6 points; normal > 2.5 points). The NIRS descending slope was −0.19%·sec−1 (−0.15-0.22%·sec−1; normal range −0.22 ± 0.06%·sec−1), ascending slope 2.39%·sec−1 (1.75-2.67% ·sec−1; normal range 4.8 ± 1.6%·sec−1), and nirVO2 110.6 U (80.1-120.2 U; normal range 171 ± 57 U).

Table 3 Microcirculatory variables
Full size table
Figure
figure 1

Sidestream dark field and near-infrared spectroscopy values of individual patients. Gray areas represent normal ranges (+2 standard deviation [SD] and -2 SD around mean). Panel A) Proportion of perfused vessels (PPV); Panel B) Microvascular flow index (MFI); Panel C) Ascending slope; Panel D) Near infrared oxygen consumption (NIR VO2)

Full size image

The proportion of perfused small vessels and the MFI correlated negatively with the SOFA score (R² 0.67; P = 0.046 and R² 0.85; P = 0.009, respectively), while the NIRS ascending slope correlated negatively with the APACHE II score (R² 0.36; P = 0.05).

Discussion

Data from this series of seven patients demonstrates that patients with severe influenza A (H1N1) infection with ALI/ARDS have clinically significant microcirculatory abnormalities. In particular, both clinical (LIS) and microcirculatory insults were at least as severe in patients with only viral infection as in those with superimposed bacterial infection.

The microcirculation has a central role in the regulation of organ perfusion.24 In several studies and in different types of shock, the presence of microcirculatory dysfunction has been associated with poor outcomes, including a higher frequency of organ failure and mortality.14,15,25,26 In our small series of patients, the microcirculation was moderately to severely compromised, in line with previous studies of patients with severe sepsis and septic shock.14,22,23 For example, Creteur et al.22 reported NIRS ascending slope values of 2.3 ± 1.3%·sec−1 in patients with septic shock, a value similar to our findings. Skarda et al.23 found NirVO2values of 97 ± 36 U, again in the range seen in our patients. In addition, the proportion of perfused small vessels assessed by SDF imaging was 65% (53.1–68.9%) in patients with septic shock, similar to that reported by Sakr et al.14 As in sepsis, we observed a significant correlation between the microcirculatory abnormalities detected by the two methods (SDF imaging and NIRS) and the severity of disease as assessed by SOFA and APACHE II scores.

One of the most interesting findings was that bacterial superinfection was not mandatory for microvascular flow impairment to occur, although the small sample size limits the strength of this conclusion. Influenza A (H1N1) virus may induce microvascular alterations either directly or indirectly via the stimulation of host inflammatory factors. Administration of tumour necrosis factor (TNF), a pro-inflammatory cytokine, can lead to microcirculatory alterations in animals.27 This new influenza strain is more virulent than seasonal influenza strains and may, therefore, induce microcirculatory impairment by a direct viral effect. Using different mammalian models, Itoh et al.11 demonstrated that one of the first influenza A (H1N1) strains to be isolated was more lethal, persisted longer in the lower airways, caused more intense alveolitis, and induced higher concentrations of inflammatory cytokines in the lungs than a regular seasonal influenza strain. Compared with outpatients, the inflammatory response to severe influenza A (H1N1) is more pronounced in patients with severe disease who require hospitalization, and the response can be observed even in patients who do not have bacterial superinfection.12 The ability of the influenza and other acute respiratory viruses to provoke microvascular alterations—characterized by sludge of capillary flow associated with hypercoagulability and increased blood viscosity—has been suggested by others.28 Therefore, significant microcirculatory perfusion deficits could be an important pathophysiological mechanism in severe influenza A (H1N1) infections.

Our study has several limitations. First, although we studied most of the severe cases in our institution, the sample size was relatively limited, and we were unable to perform some exploratory analyses of subgroups. Second, no control groups of patients with less severe H1N1 infection or other causes of ARDS were included. Third, although the examinations were performed quite early, it was impossible to standardize their timing, and there were no follow-up tests to study disease progression.

What then are the implications of our data? First, our data suggest that severe viral infections, like bacterial infections, can induce microcirculatory alterations. The study was not designed to evaluate whether the observed microcirculatory alterations are specific to H1N1 infection or whether they simply reflect the fact that critically ill patients with severe sepsis have disturbed microcirculation. Second, the finding of microcirculatory abnormalities and their correlation with disease severity using two different techniques suggests that endothelial dysfunction may play a role in the pathophysiology of severe H1N1 infections. These findings are in line with other studies in sepsis that relate the severity of microcirculatory abnormalities to the severity of disease.14,22,29

In conclusion, these data suggest that microcirculatory abnormalities are present in patients with acute lung injury associated with severe influenza A (H1N1) infection independently of any bacterial co-infection.