Endothelial cells have a key role in the control of vascular permeability and vessel tone, coagulation and fibrinolysis, and inflammatory response 1. There is an increasing body of evidence supporting the critical role of the vascular endothelium in the pathogenesis of multiple organ failure (MOF) in critically ill patients 2.

Endothelial dysfunction/or activation is associated with an imbalance in hemostatic functions. Endothelial cells are responsible for the release of tissue plasminogen activator (t-PA) and contribute to the release of plasminogen activator inhibitor-1 (PAI-1). Inhibition of the fibrinolytic system amplifies the pathogenic role of fibrin deposition during severe inflammation 3. Multiple factors, including lipoproteins, cytokines, and inflammatory proteins can modulate the endothelial cells to produce t-PA and PAI-1 4.

In infected patients, elevated concentrations of serum CRP are correlated with a risk of MOF and death 5, especially when these persist over time 5. However, the possible biological involvement of CRP in the development of MOF and death is unknown.

CRP can act directly on endothelial cells, inducing, for example, the expression of intercellular adhesion molecule (ICAM)-1 6 and the production of inflammatory cytokines such as interleukin- (IL)-6 7. It may also inhibit fibrinolysis by inducing PAI-1 release from human aortic endothelial cells 8. Moreover, the administration of recombinant CRP in volunteers may increase circulating PAI-1 levels 9.

In this study, we tested the hypothesis that CRP is associated with hypofibrinolysis as measured by ECLT in intensive care patients with and without sepsis.

The major cause of severe sepsis (9 patients) or septic shock (2 patients) was pneumonia (8 patients); angiocholitis was the cause in 1 patient, and in 2 patients the cause was not identified. In 8 patients (7 with pneumonia and 1 with angiocholitis), the infection was due to a Gram negative bacteria. Only 1 patient had documented bacteremia. Non-septic patients were admitted for postoperative surveillance (7 patients), intracerebral hemorrhage (3 patients), heart failure (3 patients), drug intoxication (3 patients), or aggravated chronic obstructive pulmonary disease (5 patients).

As expected, inflammatory parameters such as white blood cells and CRP levels, the SAPS II score and ECLT were higher in the septic than the non-septic population (Table 1). In an univariate analysis (in all patients), the ECLT was strongly correlated with serum CRP concentrations (R² = 0.45; p < 0.001) with no perceptible threshold (Fig 1). Surprisingly, there was no relationship between the SAPS score and ECLT (R² = 0.08; p = 0.15). In multivariate analysis, ECLT was best predicted by the CRP level (R² = 0.51; F = 25.6; p < 0.001) and not significantly by sepsis or the fibrinogen concentration. Interestingly, the ICU length of stay was significantly correlated with CRP (R² = 0.264, p = 0.003) and ECLT (R² = 0.259, p = 0.003) in all patients, and in the survivors (R² = 0.13, p = 0.05 and R² = 0.3, p = 0.003, respectively).

Many studies have demonstrated that sepsis is associated with endothelial cell dysfunction and promotes coagulation activation 14. Although levels of t-PA antigen increase in sepsis, fibrinolysis is inhibited by increased levels of PAI-1. ECLT is, therefore, an interesting test, because it represents the balance between t-PA and PAI-1 activities. Previously considered as an imprecise method, we have been able to improve the precision and reproducibility of the test with a new semi-automatic device 13.

Inhibition of the fibrinolytic system may contribute to MOF, as fibrin can activate endothelial cells leading to a disorganization of the monolayer and the release of inflammatory cytokines 4. Increased CRP levels are associated with worse outcomes and MOF in ICU patients 3. The role of CRP on fibrinolysis is unknown in vivo. Our data suggest that CRP could itself be involved in the processes leading to endothelium dysfunction. The observed relationship does not prove a direct biological link between increasing CRP and hypofibrinolysis; however, indirect arguments exist in support of the concept.

Several in vitro studies have reported the direct effects of CRP on endothelial cells 678. In vivo, Cleland et al. 15 reported a relationship between serum CRP levels and the forearm blood flow response to N^G-monomethyl-L-arginine (L-NMMA), reflecting endothelial dysfunction. Bisoendial et al. reported that the administration of CRP in volunteers impairs the fibrinolytic balance 9. In addition, CRP has a strong prognostic value in acute coronary syndromes 16. In a non-selected population with no inflammatory syndrome (CRP below 1.5 mg/dl, n = 160), we also observed that ECLT was significantly correlated with serum CRP levels 17.

CRP could also act indirectly on endothelial cells via the action of monocytes and the release of tumor necrosis factor-α (TNF-α). TNF-α is a strong inducer of PAI-1 production in-vitro and in-vivo. This mechanism seems to be important in sepsis, as high plasma levels of PAI-1 are associated with poor outcome 18. Moreover, an association between CRP and TNF-α has also been described 19. CRP can induce the monocyte release of cytokines such as IL-1β, IL-6, and also TNF-α through Fc receptors (γRI/CD64, FcγRIIa/CD32) 20.

CRP also has essential biological functions. No polymorphism of either the gene coding sequence or of the protein itself has been described in humans 21. Also, high levels of human CRP protect against lethal infection. Transgenic mice capable of produce human CRP are protected against lethal infection by Gram positive and negative bacteria, Szalai et al 2223.

This work is a pilot study. We have chosen to include patients in the first day of ICU admission to limit the possible rapid effect of inflammatory response on fibrinolysis, especially in septic patients. Indeed, this particular patient population has inflammatory reaction before signs of severe sepsis and thus before their admission to ICU. In fact, we have studied ECLT test at the onset of the organ dysfunction and the inflammatory reaction. Other studies with serial measurement of ECLT in patients who developed nosocomial ICU infections are needed to study the time course of these events. Moreover, we could not definitively exclude that all patients in the non-septic groups were non infected. For example, some non-septic patients with decompensated COPD may have minor infections, despite the negative microbiology cultures and the absence of antibiotic therapy. Viral infections were also possible in some patients.

Moreover, it would be of great interest to determine the interactions between CRP and ECLT with IL-6, TNF-α and endothelium dysfunction markers such as soluble thrombomodulin and soluble von Willebrand factor.

Despite accumulating evidence that the inflammatory and coagulation systems are activated in sepsis, little is known about the mechanisms that ultimately lead to organ dysfunction and death. Our data were obtained in the first 24 hours of ICU admission or of sepsis, thus, the relation between CRP and hypofibrinolysis appeared very quickly. Prospective studies including the time course of CRP and hypofibrinolysis would provide additional information about this relationship.

KZB: Laboratory analysis, writing of the manuscript and design of the study.

MP: patients recruitment and design of the study.

DB: coordination and design analysis of the results.

MG: design of the study.

PC: laboratory analysis.

JLV: design of the study and analysis of the results.

CR: design of the study and analysis of the results.

YB: patients recruitment.

MV: statistical analysis and coordination.