Altered immune parameters in chronic alcoholic patients at the onset of infection and of septic shock

The history was recorded, and an alcoholism related questionnaire – the CAGE Questionnaire [21] – was administered to all patients. (The acronym CAGE stands for 'Cutting down, Annoyance by criticism, Guilty feeling, and Eye-openers'.) All chronic alcoholic patients met criteria for alcohol abuse or dependence (Diagnostic and Statistical Manual of Mental Disorders IV, from the American Psychiatric Association [22]), had a daily ethanol intake in excess of 60 g, and had a CAGE score of 3 or more. Patients with a daily ethanol intake below 25 g and a CAGE score of 1 or less were considered to be nonalcoholic. Conventional laboratory markers such as γ-glutamyl transferase and mean corpuscular volume were determined, in accordance with routine clinical practice. In addition, a marker of higher sensitivity and specificity, namely carbohydrate deficient transferrin, was measured.

A radial artery catheter and a central venous line were inserted for routine cardiovascular monitoring. A fibreoptic, pulmonary artery flotation catheter (Swan-Ganz Oximetry/TD-Catheter model 93A-741h-7.5F; Baxter Edwards Laboratories, Irvine, CA, USA) was inserted in all patients with sepsis. Fluids were administered to achieve optimal left ventricular filling pressure, reaching the plateau value for left ventricular stroke work. If the cardiac index was less than 2.5 l/min per m², then dobutamine was administered up to 20 μg/kg per min to maintain cardiac index between 3.0 and 3.5 l/min per m². Noradrenaline (norepinephrine) was administered to patients in whom mean arterial pressure was below 70 mmHg. All patients were mechanically ventilated and received continuous analgesic sedation.

Haemodynamic measurements included heart rate and cardiovascular pressures, as measured from the mid-axillary line. Also, cardiac output measurements (as measured using the thermodilution method with 10 ml iced physiological saline solution as injectate, employing a cardiac computer [SAT-2 Oximeter/Cardiac Output Computer; Baxter Edwards Laboratories]) were taken in triplicate, with results expressed as the mean value. Blood samples were drawn simultaneously, slowly and continuously over 30 s.

Arterial and mixed venous blood samples were analyzed for oxygen and carbon dioxide tensions (ABL 300; Radiometer Inc., Copenhagen, Denmark) and for their haemoglobin content and oxygen saturation (Hemoximeter OSM 3; Radiometer Inc.). Oxygen content, delivery and consumption were calculated according to the standard formulae.

Measurements were performed within 24 hours after the onset of infection (peritonitis/pneumonia), within 12 hours after the onset of septic shock, and 72–96 hours after the onset of septic shock. With each measurement, haemodynamic and oxygen transport related measures were recored and blood samples were drawn.

Data are presented as median (range). The Wilcoxon matched pairs signed rank sum test was used to compare intergroup variables. The adjusted significance (Bonferroni method) for twice measured haemodynamic and oxygen transport parameters was P/2 = 0.025; the adjusted significance for the thrice measured parameters was P/3 = 0.017.

The Friedman test was used to identify significant intragroup differences from infection to early or late septic shock. If the global test revealed a significant difference, then the Wilcoxon test was used to define at which time point a significant change occurred. P < 0.05 was considered statistically significant. Correlation coefficients were calculated according to the Spearman rank correlation. The receiver operating characteristic curve was used to provide a presentation of the relationship between sensitivity and sensitivity of mediators that were found to be significantly different between groups, and possibly to provide diagnostic cutoff levels. The area under the receiver operating curve represents the probability of discrimination between chronic alcoholic patients and nonalcoholic patients [23].

At the onset of infection IL-8 plasma levels were significantly lower in chronic alcoholic patients than in nonalcoholic patients (Fig. 1), whereas no differences between groups occurred with respect to plasma IL-1β and IL-6 levels (Figs 2 and 3). During early septic shock, IL-1β, IL-6 and IL-8 were significantly lower in chronic alcoholic patients (Figs 1, 2, 3). IL-1β significantly increased in nonalcoholic patients from the onset of infection to early septic shock (Fig. 2). Also, IL-1β and IL-6 significantly decreased in the nonalcoholic patients from early to late septic shock (Figs 2 and 3).

In the chronic alcoholic patients significant increases occurred for soluble TNF-RI from infection to early septic shock (Table 2). However, no significant intergroup differences for soluble TNF-RI (Table 2) were found. In addition, IL-10 plasma levels did not differ between groups during infection to septic shock.

For the significant intergroup differences, receiver operating curves were calculated for the progression from infection to septic shock. The corresponding area under the curve values were 0.60 for IL-1β, 0.62 for IL-6, and 0.66 for IL-8.

CRP was significantly increased in chronic alcoholic patients at the onset of infection as compared with non-alcoholic patients (Table 3). During early septic shock, leucocyte count was significantly increased in the chronic alcoholic patients (Table 3).

Chronic alcoholic patients had a significantly lower oxygenation index during early and late septic shock than did nonalcoholic patients (Table 4). The oxygenation index in chronic alcoholic patients with pneumonia was not statistically different from that in chronic alcoholic patients with peritonitis.

In the present study the period between establishing infection and development of septic shock did not differ between chronic alcoholic patients (5 [2–10] days) and nonalcoholic patients (4 [1–129] days; P < 0.171). Numbers of culture-positive versus culture-negative samples did not differ between groups (Table 1). The identities of the positive cultures isolated from blood culture or peritoneal swab/histology were as follows: Enterococcus faecium (n = 3), Streptococcus pyogenes (n = 2), Staphylococcus aureus (n = 1) and Enterococcus faecalis (n = 1). All patients with peritonitis were surgical patients. All infections were hospital acquired.

The survival rate for nonalcoholic patients with septic shock was 53% (10/19), whereas only 43% (9/19) of the chronic alcoholic patients survived. With regard to mortality rates, 47% (9/19) of the nonalcoholic patients and 57% (12/21) of the chronic alcoholic patients died (Table 1).

This is the first clinical study to demonstrate an altered plasma cytokine response in chronic alcoholic patients at the onset of infection and early septic shock.

Experimental studies have shown that ethanol modulates cytokine secretion and synthesis in vivo and in vitro [24]. In vitro, alcohol blunted the stimulation of cytokine production by lipopolysaccharide, and the alcohol induced decrease in cytokine synthesis was proportional to the level of alcohol consumption [6]. Also, levels of production of TNF-α and IL-8 in mast cells as well as in blood monocytes were downregulated by clinically relevant ethanol concentrations [25, 26].

In a clinical study [27], alveolar macrophages and their ability to produce cytokines locally were studied because of the high risk for pneumonia associated with chronic alcoholism. In vitro stimulation of alveolar macrophages from chronic alcoholic persons resulted in significant suppression of TNF-α as compared with those from healthy control individuals. The local pulmonary suppressive effects of acute or chronic alcoholism have been described [8, 10]. In chronic alcoholic patients, circulating IL-1β, IL-6 and TNF-α levels were significantly elevated as compared with nonusers of alcohol, and this elevation was correlated with liver disease [28]. In the present study, patients with active liver cirrhosis were excluded because of the different cytokine kinetics that occur in the presence of liver disease.

That levels of systemic proinflammatory cytokines were significantly lower in chronic alcoholic patients than in nonalcoholic patients in early sepsis is an important new finding. The proinflammatory response to invading micro-organisms might be fundamentally impaired in chronic alcoholism, and this may contribute to the clinically evident immune alteration in such patients. This is in contrast to a previous report that found that TNF-α is elevated in early infection in patients developing septic shock [29]. The latter may be the effect of a severe infection whereas the former may be the cause of progression of the disease. However, this remains speculative and requires further investigation.

In the present study no differences with respect to IL-10 levels were found between groups either during the initial development of infection or in early septic shock.

In previous studies IL-10 was produced in large amounts during septicaemia and septic shock [30]. Increased IL-10 production by human blood monocytes was observed after acute ethanol treatment [31]. Considering the ability of IL-10 to inhibit monocyte function, it is likely that elevated IL-10 levels contribute to the disturbed cellular immune response observed after acute alcohol treatment. However, we did not find a difference in response of chronic alcoholic patients with respect to IL-10. The reasons for this might be, first, that elevated IL-10 levels occur earlier in the development of infection and may not be seen later in the course of the disease [13]. Second, the reduced proinflammatory response might have contributed to lower IL-10 levels [14], and therefore no difference was found between groups. In addition, no significant intergroup differences were found with respect to soluble TNF-RI and TNF-RII. A significant rise in soluble TNF-RI was noted from the onset of infection to early septic shock in the chronic alcoholic patients. High values of soluble TNF-Rs were reported during severe sepsis [32], which was confirmed by previous data. One study evaluated the in vitro expression of TNF-Rs on macrophages in response to ethanol exposure [33]; ethanol significantly reduced the expression of TNF-Rs on interferon-γ stimulated pulmonary macrophages. In the present study we found that levels of TNF-Rs were not suppressed, and so hypothetically chronic alcoholic patients have a preferentially anti-inflammatory immune response.

In particular, plasma levels of IL-6 were significantly lower in chronic alcoholic patients in our study. Although IL-6 was initially found to be proinflammatory, recent findings suggest that IL-6 has anti-inflammatory effects [34]. Therefore, IL-6 might contribute to the resolution of acute and chronic inflammatory processes by direct suppression of IL-1β and TNF-α [35].

In the present study CRP was significantly increased in chronic alcoholic patients during infection. This is in accordance with a previous study conducted in patients with a daily alcohol intake of more than 80 g [36]. Chronic alcohol consumption is associated with higher CRP concentrations. CRP is mainly regulated by proinflammatory cytokines [37]. Proinflammatory cytokines were significantly decreased in early septic shock and did not differ in late septic shock. Additionally CRP might not be stimulated adequately in chronic alcoholic patients in early or late septic shock, which might be explained as an immune breakdown. There was a significant increase in leucocyte count in early septic shock. The leucocyte count has low predictive ability as a marker of infection [38]. A marked activation of the hypothalamic–pituitary–adrenal axis occurs during ethanol withdrawal, and this could result in an hypercortisolism mediated leucocytosis in early septic shock [39].

Chronic alcoholic patients had a significantly lower oxygenation index, which might be due to the higher number of smokers [40]. However, in the present study the percentage of smokers was not significantly different between the groups, and none of the patients were hypoxic because of a higher inspired oxygen fraction, indicating that these mechanisms were not relevant.

There were only a few significant differences in haemodynamic parameters. Chronic ethanol abuse can be associated with a variety of cardiovascular disorders, ranging from asymptomatic left ventricular dysfunction to hypertension, stroke, heart failure and sudden death from arrhythmias [2, 41]. The higher pulmonary capillary wedge pressure could be interpreted as a result of a higher end-diastolic pressure leading to heart failure if there were normal ventricular compliance [41].

Because the number of patients with dual addiction (i.e. to both alcohol and nicotine) is more frequently seen than isolated or no addictions, we cannot totally exclude that the effects of ethanol on cytokine interactions were also influenced by nicotine. However, in the present study no significant difference between groups was seen. Spies and coworkers [42] examined patients with comorbid chronic alcoholism and nicotine abuse, and found that only TNF-α plasma levels were significantly higher in chronic alcoholic smokers than in chronic alcoholic nonsmokers.