Association of Sepsis-Related Mortality with Early Increase of TIMP-1/MMP-9 Ratio

Objective Higher circulating levels of tissue inhibitor of matrix metalloproteinases (TIMP)-1 at the time of severe sepsis diagnosis have been reported in nonsurviving than in surviving patients. However, the following questions remain unanswered: 1) Does TIMP-1/MMP-9 ratio differ throughout the first week of intensive care between surviving and non-surviving patients? 2) Is there an association between TIMP-1/MMP-9 ratio and sepsis severity and mortality during such period? 3) Could TIMP-1/MMP-9 ratio during the first week be used as an early biomarker of sepsis outcome? 4) Is there an association between TIMP-1/MMP-9 ratio and coagulation state and circulating cytokine levels during the first week of intensive care in these patients? The present study sought to answer these questions. Methods Multicenter, observational and prospective study carried out in six Spanish Intensive Care Units (ICUs) of 295 patients with severe sepsis. Were measured circulating levels of TIMP-1, MMP-9, tumour necrosis factor (TNF)-alpha, interleukin (IL)-10 and plasminogen activator inhibitor (PAI)-1 at day 1, 4 and 8. End-point was 30-day mortality. Results We found higher TIMP-1/MMP-9 ratio during the first week in non-surviving (n = 98) than in surviving patients (n = 197) (p<0.01). Logistic regression analyses showed that TIMP-1/MMP-9 ratio at days 1, 4 and 8 was associated with mortality. Receiver operating characteristic (ROC) curves showed that TIMP-1/MMP-9 ratio at days 1, 4 and 8 could predict mortality. There was an association between TIMP-1/MMP-9 ratio and TNF-alpha, IL-10, PAI-1 and lactic acid levels, SOFA score and platelet count at days 1, 4 and 8. Conclusions The novel findings of our study were that non-surviving septic patients showed persistently higher TIMP-1/MMP-9 ratio than survivors ones during the first week, which was associated with severity, coagulation state, circulating cytokine levels and mortality; thus representing a new biomarker of sepsis outcome.


Introduction
Sepsis represents a systemic response of the immune system to infection leading to high mortality and costs [1,2]. Matrix metalloproteinases (MMPs) are a family of zinc-containing endoproteinases that are implicated in degradation and remodelling of the extracellular matrix (ECM) [3]; and also in proteolysis of intracellular protein [4][5][6], and are involved in innate immune defence and apoptosis. The regulation of MMP activity is complex and occurs at several levels [7], such as transcriptional (MMP gene expression in cells) which is modified by TNF-a, interleukin-1b and transforming growth factor (TGF)-b; post-transcriptional (stability of MMP transcripts in cells) which is influenced by glucocorticoids and TGF-b; translational (release of MMP from cells) which is regulated by plasmin and thrombin; and posttranslational (activation of MMPs after the release) which is affected by oxidative stress, nitrosative stress, phosphorylation [8], proteolysis and tissue inhibitors of matrix metalloproteinases (TIMPs). MMPs and TIMPs help modulate inflammatory [9] and prothrombotic responses [10,11].
Previous clinical studies including our own have shown higher circulating levels of MMP-9 and TIMP-1 in septic patients than in controls [12][13][14][15][16][17][18][19][20][21][22], and higher levels of TIMP-1 [15,19,20,23] at the time of severe sepsis diagnosis in non-surviving than in surviving patients. In addition, an association between circulating TIMP-1 and plasminogen activator inhibitor (PAI)-1 levels in septic patients at severe sepsis diagnosis has been reported [23]. However, the following questions remain unanswered: 1) Does TIMP-1/MMP-9 ratio differ throughout the first week of intensive care between surviving and non-surviving septic patients? 2) Is there an association between TIMP-1/MMP-9 ratio and sepsis severity during this period? 3) Is there an association between TIMP-1/MMP-9 ratio during the first week and sepsis mortality? 4) Could TIMP-1/MMP-9 ratio be used as an early biomarker of sepsis outcome? 5) Is there an association between TIMP-1/ MMP-9 ratio and coagulation state during the first week of intensive care in these patients? 6) Is there an association between TIMP-1/MMP-9 ratio and circulating cytokine levels during the first week of intensive care in septic patients? The present study sought to answer these questions. We focus on the determination of the TIMP-1/MMP-9 ratio because in previous studies we assessed serum TIMP-1 and MMP-9 levels [19,23], and because MMP-9 is inhibited by TIMP-1.

Design and Subjects
A prospective, multicenter observational study was carried out in six Spanish intensive care units (ICUs Patients admitted to the ICU of the participating hospitals and meeting the criteria for severe sepsis were included in the present study. The diagnosis of severe sepsis was established according to the International Sepsis Definitions Conference [24]. Exclusion criteria were: age ,18 years, pregnancy, lactation, human immunodeficiency virus (HIV), white blood cell count ,103/ mm3, solid or hematological tumors or immunosuppressive, steroid or radiation therapy.

Variables recorded
The following variables were recorded for each patient: sex, age, diabetes mellitus, chronic obstructive pulmonary disease (COPD), ischemic heart disease, site of infection, microorganism responsible, bloodstream infection, antimicrobial treatment, pressure of arterial oxygen/fraction of inspired oxygen (PaO2/FIO2), creatinine, bilirubin, leukocytes, lactic acid, platelets, international normalized ratio (INR), activated partial thromboplastin time (aPTT), Sepsis-related Organ Failure Assessment [SOFA] score [25] and Acute Physiology and Chronic Health Evaluation (APACHE)-II score [26]. We assessed survival at 30 days as the endpoint.

Proteolytic, inflammatory and prothrombotic marker assays
Blood samples were collected at days 1, 4 and 8 of ICU admission. Serum separator tubes were used to determine MMP-9, TIMP-1, TNF-alpha and IL-10 concentrations, and venous citrated plasma tubes to determine PAI-1 concentration. Venous   Systems, Abingdon, United Kingdom) was used according to manufacturer's instructions with a serum dilution of 1:80 and 1:100 respectively. The interassay coefficients of variation (CV) were ,8% (n = 20) and the detection limits for the assays were 0.156 ng/ml and 0.08 ng/ml respectively. Assays for TNF-alpha, IL-10 and PAI-1 were performed at the Laboratory Department of the Hospital Universitario de Canarias (La Laguna, Santa Cruz de Tenerife, Spain). TNF-alpha and IL-10 were measured in serum by solid-phase chemiluminescent immunometric assays (Immulite, Siemens Healthcare Diagnostics Products, Llanberis, United Kingdom). The interassay CV were , 6.5% (n = 20) and ,9.9% (n = 40) respectively, and the detection limits for the assays were 1.7 pg/ml and 1 pg/ml respectively. PAI-1 antigen was assayed by specific ELISA (Imubind Plasma PAI-1 ElisaTM, American Diagnostica, Inc, Stanford, CT, USA). This assay detects latent (inactive) and active forms of PAI-1 and PAI-1 complexes. The interassay CV was ,5% (n = 20) and the detection limit for the assay was 1 ng/ml.

Statistical Methods
Continuous variables are reported as medians and interquartile ranges. Categorical variables are reported as frequencies and percentages. Comparisons of continuous variables between groups were carried out using Wilcoxon, Mann-Whitney or Friedman tests. Comparisons between groups for categorical variables were carried out with chi-square test. Logistic regression analyses were applied to determine the independent contribution of TIMP-1/ MMP-9 ratio at days 1, 4 and 8 on 30-day mortality, controlling for SOFA score and diabetes mellitus. To avoid the collinearity effect, we included only SOFA score and age as co-predictors. Odds ratio (OR) and 95% confidence intervals (CI) were calculated as measures of the clinical impact of the predictor variables. We plotted receiver operating characteristic (ROC) curves using survival at 30 days as the classification variable and TIMP-1/MMP-9 ratio at days 1, 4 and 8 as prognostic variables. The association between continuous variables was assessed using Spearman's rank correlation coefficient. A P value of less than 0.05 was considered statistically significant. Statistical analyses were  Table 3. Correlations between circulating TIMP-1/MMP-9 ratio levels and TNF-alpha, IL-10, PAI-1, lactic acid, SOFA score, platelet, INR and aPTT at days 1, 4 and 8.

Results
Comparison of demographic and clinical parameters between non-surviving (n = 98) and surviving septic patients (n = 197) is shown in Table 1. Non-survivors were older and showed higher diabetes mellitus, creatinine, lactic acid, INR, aPTT, SOFA and APACHE-II, and lower platelet count than survivors. No differences were observed regarding sex, COPD, ischemic heart disease, site of infection, microorganism responsible, bloodstream infection and antimicrobial treatment.
As shown in Table 3, TIMP-1/MMP-9 ratios at days 1, 4 and 8 correlated with circulating TNF-alpha, IL-10, PAI-1 and lactic acid levels, SOFA score, and platelet count, and at days 1 and 4 with INR and aPTT.

Discussion
A novel finding of our study was that non-surviving septic patients showed persistently higher circulating levels TIMP-1 levels, lower circulating MMP-9 levels and higher TIMP-1/MMP-9 ratio than survivors during the first week of ICU stay. In addition, we found for the first time that the TIMP-1/MMP-9 ratio during the first week was associated with mortality according to the results of logistic regression analyses. Two previous studies exploring MMP levels during the course of sepsis found no significant differences between surviving and non-surviving patients regarding TIMP-1 [21] and MMP-9 levels [21,22]. The higher sample size of our study (295 patients) compared with the smaller size (20 and 44 patients respectively) of previous studies [21,22] could explain the significant differences reported between surviving and non-surviving patients.
In a previous study by our team we found that TIMP-1 levels at the time of severe sepsis diagnosis could be used as a biomarker of sepsis outcome [19]. A novel finding of the current study was that TIMP-1/MMP-9 ratio during the first week could be used as a biomarker of sepsis outcome according to the results of ROC curve analysis. In addition, we previously reported that circulating TIMP-1 levels were associated with TNF-alpha, IL-10, PAI-1, SOFA score, lactic acid levels, platelet count, aPTT and INR [19]. Another novel finding of the present study was a significant association between TIMP-1/MMP-9 ratio and circulating cytokine levels, coagulation parameters and SOFA score during the first week.
Taken together, these results indicate that TIMP-1 and MMP-9 levels could play a pathophysiological role in septic patients. MMP-9 inhibits platelet aggregation [27,28] and an association between circulating TIMP-1 and PAI-1 levels has been found [29,30]. In addition, an association between PAI-1 levels and sepsis mortality has been reported [31][32][33]. Thus, it may be that the higher circulating levels TIMP-1 levels, lower circulating MMP-9 levels and higher TIMP-1/MMP-9 ratio found in nonsurviving patients during the first week could favour a prothrombotic state and contribute to capillary thrombosis, multiple organ dysfunction, and death in septic patients.
From a therapeutic perspective, the development of modulators of MMP/TIMP activity could be used as a new class of drugs for the treatment of severe sepsis [34,35].
The strengths of our study were the large sample size and the follow-up of circulating TIMP-1, MMP-9, TNF-alpha, IL-10, and PAI-1 levels throughout the first week after diagnosis of severe sepsis. However, our study has certain limitations. First, we did not analyze other MMPs and TIMPs; and it could be interesting to describe levels of other MMPs and TIMPs, and the ratios between them. Second, measurement of other cytokines and coagulation markers would be desirable in order to better evaluate the relationship between proteolysis, inflammation and coagulation in these patients. Third, we analyzed serum levels of MMP-9 and TIMP-1 but plasma determination is preferable because there could be a release of several MMPs or TIMPs from platelets in serum [36][37][38][39].
In conclusion, the novel findings of our study were that nonsurviving septic patients showed persistently higher TIMP-1/ MMP-9 ratio than survivors during the first week, which was associated with severity, coagulation state, circulating cytokine levels and mortality, thus representing a new biomarker of sepsis outcome.