Sepsis is most frequently associated with hemostasis abnormalities 1 2 3 , especially disseminated intra-vascular coagulation (DIC). DIC is characterized by the association of systemic activation of coagulation, impairment of anticoagulant pathways and inhibition of fibrinolysis 4 5 . Manifestations of DIC, which is associated with an increased mortality during sepsis, include bleeding and microthrombi leading to organ failure 6 . Animal studies and clinical trials reported that impairment of several anticoagulant pathways, like Protein C 7 8 , tissue factor pathway inhibitor 9 or anti-thrombin 8 , could also occur during sepsis. Substitution therapy with these factors has not been shown to improve sepsis outcome 10 11 .

Beside coagulation disturbances, abnormalities of von Willebrand factor (VWF), related to endothelial damage, have been reported in sepsis patients during the last decade 12 13 . VWF is a multimeric glycoprotein crucial for platelet adhesion and aggregation at the high shear rates of blood flow present in the microcirculation. The largest VWF multimers are the most adhesive to platelets and thus, the most efficient for primary hemostasis. Ultra-large VWF multimers (ULVWF) are physiologically present in endothelial cells and platelets, but absent in plasma. A metalloprotease, ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type 1 repeats, member 13), specifically cleaves ULVWF multimers and thus, regulates its adhesive function. A severe ADAMTS13 deficiency (activity <5%) either inherited or most often acquired via specific autoantibodies, is the main actor in the pathophysiology of thrombotic thrombocytopenic purpura (TTP) 14 15 . TTP is a thrombotic microangiopathy (TMA) characterized by thrombopenia, mechanical hemolytic anemia, organ failure and VWF-rich microthrombi 16 17 18 . As a consequence of its implication on VWF multimer synthesis, sepsis has been reported as one of the factors that might trigger thrombotic microangiopathy 19 and TTP is one of the differential diagnoses of DIC. Some cases of ADAMTS13 deficiency have been reported recently in adult patients with sepsis-induced DIC and this ADAMTS13 deficiency was associated with acute renal failure 20 . Later clinical studies, also in adults and focusing on sepsis without DIC 21 22 23 24 25 or isolated DIC not specifically related to sepsis 26 , reported decreased ADAMTS13 activity. Interestingly, some of these studies 20 21 25 26 showed that partially decreased ADAMTS13 activity was associated with outcome. In pediatric patients with severe sepsis, ADAMTS13 deficiencies have also been reported 27 28 29 . In one of the latter studies focused on children with thrombocytopenia-associated multiple organ failure, mortality was associated with reduced ADAMTS13 activity and, interestingly, with VWF-rich microvascular thrombosis at autopsy 28 . Also, this study demonstrated that plasma exchange, a source of active ADAMTS13, was able to restore ADAMTS13 activity and organ function 28 .

Although these studies gave interesting insight into ADAMTS13 levels during sepsis, clinical relevancy, prognostic association and mechanisms of a potential deficiency remain uncertain. We thus designed the current prospective study with the primary objective to study the interaction between ADAMTS13 and DIC during septic shock and its impact on outcome. As a secondary objective, we investigated the mechanisms for ADAMTS13 functional variations in this clinical context. DIC, VWF and ADAMTS13 parameters were exhaustively analyzed and correlated with clinical data in 72 patients with septic shock prospectively enrolled from a single Medical ICU.

Sepsis-induced organ failure is associated with a microcirculation damage process involving mainly platelet/fibrin-thrombi supporting DIC. In some cases, the potential addition of platelet/VWF thrombi due to an acquired ADAMTS13 functional deficiency and leading to so-called TMA-like lesions is highly suspected in clinical arguments 41 42 43 but still remains debated in terms of prognosis and pathophysiology. Furthermore, in most studies published so far, the subtle combination of DIC and TMA-like lesions has been difficult to establish because of the great heterogeneity of the sepsis cohorts tested, involving mixed cases within the categories of systemic inflammation, sepsis, septic shock, and sepsis-related multiple organ failure 20 21 22 23 24 25 26 , some of them complicated by DIC and others not. In the current study, we focused on a cohort of 72 patients with septic shock prospectively enrolled from a single ICU in order to respond to two objectives: first, to analyze the interaction between DIC and ADAMTS13 and its impact on prognosis, and second, to elucidate the pathophysiological mechanisms for ADAMTS13 deficiency. The strength of our study is to propose an approach based on clinical practice supported by three groups to evaluate the prognostic value of ADAMTS13 in sepsis: we showed that septic shock outcome is associated with a functional deficiency of ADAMTS13 together with an increase of IL-6, independently of DIC.

In our cohort, the median of ADAMTS13 activity was 30%, which is in agreement with the literature 20 21 22 23 24 25 26 and most patients (57/72 = 79%) exhibited partial ADAMTS13 functional deficiency (values <50%). DIC was present in 19 patients (26%). The overlap between DIC and ADAMTS13 activity <30% was present only in 11 patients (15%), suggesting that these biological parameters were not linked, which was further confirmed by the absence of correlation between the ISTH score and ADAMTS13 activity and by the lack of a significant difference in ADAMTS13 between DIC-positive and DIC-negative patients and in the ISTH score between patients with ADAMTS13 activity >30% and <30%. Thus, the independence between DIC and ADAMTS13 activity prompted us to divide our cohort into three groups in order to compare patients with absence of both DIC and ADAMTS13 deficiency to other patients. We observed that the higher the number of hemostatic abnormalities, the higher the disease severity. We also found a strong significant difference between groups for all indicators of disease severity at ICU admission (LOD score, SAPS II score, systolic blood pressure, PaO/FiO2, lactates, IL-6 concentrations) suggesting that both isolated ADAMTS13 deficiency and DIC-associated ADAMTS13 deficiency were markers of severity in sepsis. Only one study dedicated to ADAMTS13 in 40 septic patients analyzed disease severity scores (LOD and SAP scores) but did not find any correlation with ADAMTS13 23 : this discrepancy with our study may be related to the presence of only eight patients with septic shock and no cases of DIC 23 .

Furthermore, in our study, there was also significant correlation between the number of hemostatic abnormalities and some outcome indicators that is, persistent shock at day 3. The need for dialysis, reflecting renal failure, was not significantly different between the three groups although the association of both DIC and ADAMTS13 deficiency <30% exhibited more renal failure than both of the other groups with a borderline significance (P = 0.05). These data are in agreement with other studies of the literature led in adult patients 20 21 24 25 26 : indeed, our group 3 is strongly comparable to the 109 patients with sepsis-induced DIC described by Ono et al. 20 who also found a higher rate of renal failure in patients whose ADAMTS13 activity was <20%. Also, Martin et al. 21 demonstrated that in 30 patients with severe sepsis, including 17 patients with septic shock and 4 patients with DIC, ADAMTS13 activity <30% was correlated with renal failure and as in our study, with hemodynamic shock. Other studies involving patients with severe sepsis/septic shock 24 25 26 also found a correlation between ADAMTS13 activity and organ dysfunction, in contrast to Kremer et al. 23 .

Finally, our study demonstrates that DIC and ADAMTS13 deficiency <30% are independent markers of mortality in septic shock, particularly as the mortality rate is strongly dependent on the position of the ADAMTS13 level towards the median value (significantly higher mortality rate when ADAMTS13 is <30%) and ADAMTS13 deficiency appears to be a potential prognosis marker. Although prognostic association between ADMATS13 deficiency and outcome remains after adjustment for confounders, the diagnostic performance of this test remains modest. In addition, confirmation of our findings in a validation cohort is needed. To corroborate this observation, both ICU and mortality rates were not significantly different in group-2 (mainly including isolated ADAMTS13 deficiencies) and group-3 patients. In other words, in our cohort, the addition of DIC to ADAMTS13 deficiency did not significantly change the mortality rate. Again, these data are in agreement with most studies of the literature 20 21 24 25 26 . Interestingly, in our cohort, four patients exhibited ADAMTS13 activity <5%. A previous study in 109 patients with sepsis-induced DIC also found that 17 patients had severe ADAMTS13 functional deficiency 20 , and also exhibited a TTP-like clinical presentation. The mechanism by which some septic patients reach a severe ADAMTS13 deficiency similar to TTP patients is very likely multifactorial and may also have an important impact on prognosis.

The association between functional partial deficiency of ADAMTS13 and clinical prognosis factors in sepsis raises the question of the relevance of this metalloprotease as a passive marker or, in contrast, as an actor in the pathophysiology of organ failure. First, normal ADAMTS13:Ag levels found in our cohort suggested that ADAMTS13 functional deficiency was mainly related to a qualitative defect and not to a quantitative defect secondary to a synthesis/secretion deficiency or to a degradation process mediated by sepsis-related enzymes 20 44 45 46 47 48 . Second, direct catalytic inhibition of ADAMTS13 by IL-6, a cytokine highly secreted in sepsis, has been reported in vitro using ULVWF strings secreted from human umbilical vein endothelial cells under flowing conditions 49 . Clinical studies focused on septic patients are controversial, showing either no correlation between ADAMTS13 activity and IL-6 levels 23 or in contrast, a strong inverse correlation between these parameters 21 24 . Interestingly, in the current study, we also found a strong inverse correlation between ADAMTS13 activity and IL-6 levels. Although this inverse correlation does not allow extrapolation of any causal relationship, this result suggests a potential role of IL-6 in the functional deficiency of ADAMTS13 observed in septic shock. Additional studies are needed to confirm these findings and to more clearly assess the role of IL-6 on ADAMTS13 activity. Also, in our study, the link between IL-6 levels and prognosis already reported by other authors 50 may be mediated by ADAMTS13 deficiency. In contrast, no relevant specific autoantibodies to ADAMTS13 were detected, making unlikely the involvement of an autoimmune process in sepsis-associated ADAMTS13 functional deficiency. Third, in all our patients, we found increased VWF:Ag levels and the presence of UL-VWF multimers in plasma. An overwhelming release of ULVWF multimers from activated endothelial cells may exhaust ADAMTS13 activity by a consumption mechanism 44 46 49 50 51 52 53 54 . In septic patients, the literature is controversial either supporting this consumption process 20 21 24 25 or not 23 . Further studies are needed to confirm prognostic impact of ADAMTS13 deficiency and to validate a cutoff for ADAMTS13 activity that may be used to assess outcomes, independently of other determinants of death.

Our study has several limitations. First, our study does not include any control group. Indeed, without this control group we do not know whether the current findings are specific to septic shock and whether the observed relationship (or lack thereof) is similar to the relationship in critically ill patients who do not have septic shock. In addition, the compared group had, as expected, different severity at study inclusion. However, the prognostic impact persisted after adjustment for patient severity, suggesting the prognostic association between ADAMTS13 deficiency and outcome to be independent from initial severity differences.

Our study underlines that septic shock seems to be associated with partial functional deficiency of ADAMTS13, which mechanism is potentially related to IL-6-mediated inhibition but is independent of DIC. As a consequence, excessively released ULVWF multimers are more slowly cleared from the circulation and generate spontaneous formation of microvascular platelet thrombi responsible for multivisceral organ failure. Last but not least, our study emphasizes the role of ADAMTS13 as a prognosis factor in septic shock. These data are in agreement with the exciting concept of ADAMTS13 as a link between inflammation and thrombosis, which was recently developed thanks to animal models 55 56 57 58 . In addition, this study may bring new insights in the field of hemostasis inhibitor-based treatment in severe sepsis: indeed, although the efficiency of coagulation inhibitors is still controversial 59 , downregulation of VWF-mediated thrombosis by exogenous ADAMTS13 brought by plasma infusions, or in the future by plasma-purified or recombinant ADAMTS13, may be an interesting new approach targeted on primary hemostasis.

• Septic shock seems to be associated with partial functional deficiency of ADAMTS13, which mechanism might be related to IL-6-mediated inhibition.

• ADAMTS13 is an independent prognostic factor during septic shock.

• No correlation was observed in this study between ADAMTS13 deficiency and DIC.

• ADAMTS13 deficiency may decrease clearance of excessively released ULVWF multimers, leading to spontaneous microvascular platelet thrombi formation and multiple organ failure.

ADAMTS13: A disintegrin and metalloprotease with thrombospondin type 1 repeats member 13; ADAMTS13:Ag: ADAMTS13 antigen; DIC: Disseminated intravascular coagulation; ELISA: Enzyme-linked immunosorbent assay; Ig: Immunoglobulin; IL-6: Plasma interleukin (IL)-6; ISTH: International Society for Thrombosis and Haemostasis; LOD: Logistic Organ Dysfonction score; MV: Mechanical ventilation; OR: Odds ratio; PaO2/FiO2: arterial partial pressure of oxygen/inspired oxygen fraction; PI: Prothrombin index; RRT: Renal replacement therapy; SAPS II: Simplified Acute Physiology Score 2; TMA: Thrombotic microangiopathy; TTP: Thrombotic thrombocytopenic purpura; ULVWF: Ultralarge von Willebrand Factor multimers; VWF: von Willebrand Factor; VWF:Ag: von Willebrand Factor antigen.

The authors declare that they have no competing interests.

VP, EA, BS and AV conceived the study, participated in its design and coordination and helped to draft the manuscript. IC, EM, NA and MD participated in study design, patient recruitment and performed the statistical analysis. VP, PL, AMK, MW and AV carried out the biological testing and cytokine evaluation. All authors read and approved the final manuscript.