Matrix metalloproteinase MMP-8, TIMP-1 and MMP-8/TIMP-1 ratio in plasma in methicillin-sensitive Staphylococcus aureus bacteremia

Background Matrix metalloproteinase-8 (MMP-8) and tissue inhibitor of metalloproteinases-1 (TIMP-1) have been shown to predict prognosis in sepsis. However, MMP-8 and TIMP-1 in Staphylococcus aureus bacteremia (SAB) lacks evaluation and their role in the pathogenesis of SAB is unclear. Methods MMP-8 and TIMP-1 and MMP-8/TIMP-1 molar ratio were determined at days 3, 5 and 28 from positive blood cultures in patients with methicillin-sensitive SAB and the connection to disease severity and early mortality was determined. Results Altogether 395 SAB patients were included. Patients with severe sepsis or infection focus presented higher MMP-8 levels at day 3 and 5 (p<0.01). Higher day 3 and 5 MMP-8 levels were associated to mortality at day 14 and 28 (p<0.01) and day 90 (p<0.05). Day 3 MMP-8 cut-off value of 203 ng/ml predicted death within 14 days with an area under the curve (AUC) of 0.70 (95% CI 0.57–0.82) (p<0.01). Day 5 MMP-8 cut-off value of 239 ng/ml predicted death within 14 days with an AUC of 0.76 (95% CI 0.65–0.87) (p<0.001). The results for MMP-8/TIMP-1 resembled that of MMP-8. TIMP-1 had no prognostic impact. In Cox regression analysis day 3 or 5 MMP-8 or day 3 MMP-8/TIMP-1 had no prognostic impact whereas day 5 MMP-8/TIMP-1 predicted mortality within 14 days (HR, 4.71; CI, 95% 1.67–13.3; p<0.01). Conclusion MMP-8 and MMP-8/TIMP-1 ratio were high 3–5 days after MS-SAB diagnosis in patients with an infection focus, severe sepsis or mortality within 14 days suggesting that matrix metalloproteinase activation might play a role in severe SAB.

Staphylococcus aureus express surface structures of which cell-wall peptidoglycan is among the most important [8]. Staphylococcus aureus peptidoglycan induce matrix metalloproteinases (MMPs) levels in rat tissue [9] and induce elevated concentrations of neutrophil originated MMPs in human blood [10]. Furthermore, Staphylococcus aureus peptidoglycan displays inflammatory properties in human blood such as release of cytokines and the upregulation of tissue factor in human monocytes [11,12]. MMPs and their tissue inhibitors (TIMPs) have been studied as promising new diagnostic and prognostic biomarkers for critically ill and septic patients [13][14][15]. Structurally related but genetically distinct MMPs are zinc dependent proteolytic enzymes that control intra-and extracellular matrix proteins [16,17] and enhance migration of immune cells and pro-inflammatory and cytokine responses [18]. MMPs degrade almost all extracellular and basement membrane proteins as well as non-matrix bioactive molecules and mediators including α1-antitrypsin, bradykinin, insulin-receptor and complement component and thereby regulate immune responses [18,19]. The activities of MMPs are regulated by TIMPs which is the most potent endogenous inhibitor of MMP-8 and their ratio has been used in evaluation of effective pharmacological treatments [20]. In severe inflammation and critical illness the balance of MMPs and inhibitory TIMPs may become disturbed [21]. Among the most important MMP subgroups are the collagenases (MMP-1, -8, -13) and the gelatinases (MMP-2, -9) [22]. This data suggests that MMP activation could have a role in many pathogenic processes of complicated SAB and might be one potential therapeutic target.
The objective of the present study was to evaluate plasma levels of MMP-8, TIMP-1 and MMP-8/TIMP-1 ratio in methicillin-sensitive SAB (MS-SAB) and their possible role in disease pathogenesis by evaluating their predictive value on disease severity and mortality. All patients were provided with formal bedside ISC which guaranteed proper clinical management including non-delayed proper antibiotic therapy onset from the first day of SAB.

Settings and study population
This was a prospective multicenter study. Adult patients with blood cultures positive for S. aureus were included from seven central hospitals and five university hospitals in Finland from January 1999 to May 1999 and January 2000 to August 2002. A total of 430 SAB patients were included. Median time-interval between blood culture sampling and study inclusion was three days. All patients were provided with formal bedside ISC. The following were exclusion criteria: age < 18 years, pregnancy, breastfeeding, imprisonment, epilepsy, bacteremia 28 days prior to the study, poly-microbial bacteremia and meningitis [35]. No cases of methicillinresistant S. aureus (MRSA) were included. Data collection included gender, age, bacteremia acquisition, time to defervescence (axillary temperature below 37.5˚C), underlying diseases and length and administration route of any antibiotic therapy. Parameters required for Pitt bacteremia score calculation i.e. mental status, vital signs, any mechanical ventilation and recent cardiac arrest [36,37] as well as any ICU treatment were recorded. Infection focus documentation was based on clinical suspicion or verified by radiological, bacteriological, or pathological investigations. Plasma samples for MMP-8 and TIMP-1 concentrations were drawn at days 3, 5 and 28 from the positive blood culture sampling. Primary endpoint was mortality at 14 days. Secondary endpoints were mortality at 28 and 90 days as well as disease severity, prevalence of deep infection foci and time to defervescence.

Follow-up time period
Patients were followed prospectively for at least 90 days. Patients transferred to other hospitals were followed from patient records and direct contact to the hospital. Patients who were not hospitalized at 28 days had an additional laboratory appointment arranged to enable laboratory sampling at 28 days as well as an outpatient clinic follow-up visit at 90 days.

Definitions
Bacteremia with the first positive blood culture for S. aureus obtained 48 h after hospital admission was defined as healthcare-associated (HA) SAB. The same applied when the patient had undergone hemodialysis within the preceding two months or remained in a long-term care facility. McCabe's criteria were applied for classification of underlying diseases [38]. Severe sepsis was classified as sepsis in combination with hypotension, hypo-perfusion, or organ failure [39]. Endocarditis was defined according to the modified Duke criteria [40]. Deep infection foci included mediastinitis, pneumonia, endocarditis, purulent arthritis, osteomyelitis, deep-seated abscess, and any foreign-body infection. Deep infection foci were documented either based on clinical suspicion or verified by bacteriological, radiological or pathological findings [41].

Antibiotic therapy
Anti-staphylococcal penicillin (cloxacillin) was the first-line antibiotic therapy, whereas patients with contraindications for penicillin were provided with especially cefuroxime but also ceftriaxone, clindamycin or vancomycin. Fluoroquinolone (levofloxacin) and/or rifampicin were provided as additional therapy [35]. The dosages for intravenously administered antibiotics were as follows; cloxacillin 2 g q 4 h, cefuroxime 1.5 g q 6 h, ceftriaxone 2 g q 24 h clindamycin 600 mg q 6-8 h or vancomycin 1 g q 12 h. For oral treatment the dosages were as follows; levofloxacin 500 mg q 24 h for patients under 60 kg and 500 mg q 12 h for patients over 60 kg in weight and/or for rifampicin 450 mg q 24 h for patients under 50 kg and 600 mg q 24 h for patients over 50 kg in weight. Patients with diagnosed deep infection foci had intravenous antibiotic therapy for at least 28 days, whereas in the absence of any deep infection, 14 days was regarded as a proper length. For patients with renal or hepatic dysfunction the antibiotic doses were adjusted as recommended by the manufacturers.

MMP-8, TIMP-1 and C-reactive protein analysis
Plasma samples for MMP-8, TIMP-1 and CRP measurements were taken on days 3, 5 and 28 after positive blood cultures. Samples for MMP-8 and TIMP-1 analyzes were centrifuged at 2000 g for 10 min and the supernatants were stored at −80˚C until analysis. We measured MMP-8 plasma concentration by time-resolved immuno-fluorometric assay (IFMA) [31,42] by using monoclonal MMP-8 specific antibodies 8708 and 8706 (Medix Biochemica, Kauniainen, Finland) for catching and tracing. Europium chelate was applied for labelling the tracer antibody. TIMP-1 analysis was done by using commercially available enzyme-linked immunosorbent assay according to the manufacturer's instructions (Biotrak ELISA System; Amersham Biosciences, Buckinghamshire, UK). Concentrations are reported as ng/ml. Detection limits for MMP-8 and TIMP-1 are 0.08 ng/ml and 1.25 ng/ml, respectively. The levels of MMP-8 and TIMP-1 were denoted as ng/ml and for calculation of MMP-8/TIMP-1 molar ratio the levels were converted to mol/l [43,44]. The MMP-8 and TIMP-1 were properly stored (−80˚C storage) after collection and analyzed in 2015-2016. Laboratory quality testing ensured that MMP and TIMP remained stable at -80˚C and hence the standard of the laboratory samples were guaranteed despite longtime storage. Plasma C-reactive protein was subjected to automatic immune-turbidometric using analyzers 917 or Modular PP-analyzer (Hitachi Ltd, Tokyo, Japan) and Tina-quant CRP reagents (Roche Diagnostics, Tina-quant). The normal value of CRP concentration was < 10 mg/L for both methods. The anti-coagulant used for the blood sampling of MMP-8, TIMP-1 and CRP was heparinate or EDTA.

Statistical analysis
Data are presented either as absolute values and percentages or as mean with standard deviation. Categorical variables are compared with Pearson's X 2 -test whereas non-categorical variables are analyzed with Student's t-test. Odds ratios (OR) with 95% confidence intervals (CI) were calculated. The discriminative power of MMP-8, TIMP-1, MMP-8/TIMP-1 and CRP in predicting mortality was evaluated by receiver operating characteristic (ROC) curves. The area under the curve (AUC) was calculated. Youden index was defined as the sensitivity and specificity sum with the highest value or the ROC-curve point equally maximizing both sensitivity and specificity values to locate the cut-off point. The ROC-curve derived cut-off point results were applied for Cox regression model (proportional regression) predicting mortality. Generalized linear model (GLM) repeated measure analyze was applied for comparison of MMP-8, TIMP-1 and MMP-8/TIMP-1 throughout the 90 days follow-up period for patients who deceased versus those that survived. Univariate factors with p <0.1 were allowed for the Cox proportional regression model. All tests were two-tailed and p <0.05 was considered as significant. Analyzes were done using SPSS 12.0 (SPSS Inc., Chicago, IL, USA).

Patient characteristics
Altogether 430 SAB patients were included. All methicillin-resistant cases were excluded (N = 6). Due to missing plasma samples the results for 395 (day 3), 378 (day 5) and 315 (day 28) patients are presented. However, no patients were lost during the follow-up time-period for other reasons that death. The mean MMP-8 and TIMP-1 levels at day 3, 5 and 28 from positive blood culture were stratified according to patient characteristics (Table 1). Patient   Values are N (%) or mean (± standard deviation). https://doi.org/10.1371/journal.pone.0252046.t001 demographics and underlying conditions had no impact on MMP-8 or TIMP-1 levels on day 3 or 5 with the exception of chronic renal failure that was associated to lower MMP-8 (p<0.05) and higher TIMP-1 (p<0.01) values. Patient demographics and underlying conditions did not affect MMP-8 levels on day 28 but significantly higher TIMP-1 levels were associated to male sex (p<0.05), age > 60 years (p<0.01) and ultimately or rapidly fatal underlying disease (p<0.001) as well as coronary artery disease (p<0.01), chronic renal failure (p<0.001), diabetes mellitus (p<0.001) and alcoholism (p<0.05) ( Table 1).

Severity of illness and deep infection foci
A total of 7% of patients had severe sepsis and 15% needed ICU treatment at blood-culture collection time-point. High Pitt bacteremia score, severe sepsis and ICU treatment were associated to higher MMP-8 concentrations at day 3 and day 5 (p<0.01) and higher TIMP-1 at day 3 (p<0.05) whereas no connection to day TIMP-1 was observed. High Pitt bacteremia scores were associated to higher MMP-8 (p<0.05) on day 28 but TIMP-1 level was not affected by the Pitt bacteremia score (Table 1). Patients with any deep infection foci, as compared with patients without a deep infection foci, demonstrated higher MMP-8 and TIMP-1 levels in all sampling points whereas no differences in them were observed among patients who had endocarditis or not (Table 1).

Antibiotic therapy
All patients were treated with an antibiotic effective in vitro against the S. aureus blood isolate starting from the day of the positive blood culture. Majority of the patients (75%) received an anti-staphylococcal penicillin (cloxacillin) whereas 18% got cefuroxime and 3% ceftriaxone. Vancomycin was used in 2% and it was the only antibiotic in 0.5% of the patients. Rifampicin or a fluoroquinolone was given to 58% or 59% of patients, respectively.

Mortality and defervescence
The primary end-point was mortality at 14 day and it occurred in 8% (31 patients) whereas the secondary end-point, mortality at 28 days and at 90 days occurred in 13% (50) and 18% (72) of patients, respectively (Table 1 and Fig 1). The MMP-8 levels at day 3 and 5 were higher among patients who deceased within 14 days (p<0.01), 28 days (p<0.01) and 90 days (p<0.05) as compared to surviving patients. However, no difference was seen in TIMP-1 levels between surviving and non-surviving patients. Altogether 22 patients (6%) died after the first month and among these patients the MMP-8 levels at day 28 were higher as compared to surviving patients (p<0.05) whereas no difference was seen in the TIMP-1 levels ( Table 1). The mean time to defervescence was 4.1 ± 5.4 days (mean ±SD) and 46 patients (12%) had fever longer than 7 days. Patients with prolonged fever, compared to patients with defervescence within 7 days, had higher MMP-8 levels at day 3 (p<0.01), day 5 (p<0.05) and day 28 (p<0.05) and higher TIMP-1 at day 3 (p<0.05) ( Table 1). Furthermore, among background conditions chronic renal failure (N = 57) was the only factor associated significantly to lower MMP-8 levels (p<0.05) and higher TIMP-1 levels (p<0.01) on day 3. Hence, these patients were excluded from further statistical calculations and analyzed as a separate group. However, among patients with chronic renal failure, day 3, day 5 or day 28 MMP-8, TIMP-1 or MMP-8/ TIMP-1 had no significant predictive effect on mortality when analyzed by receiver operating characteristic.
When applying the generalized linear model repeated measure analyze for comparison of MMP-8, TIMP-1 and MMP-8/TIMP-1 throughout the 90 days follow-up period for patients who deceased versus those that survived, only the MMP-8 levels were significantly different (p-value<0.01) whereas the TIMP-1 or MMP-8/TIMP-1 levels did not differ significantly.

Cut-off values for MMP-8 and MMP-8/TIMP-1 in predicting mortality
To study the potential pathogenetic role of MMP activation in SAB we analyzed the predictive value of MMP-8 and MMP/TIMP-1 ratio at day 3 and 5 on mortality within 14, 28 or 90 days and compared it to C-reactive protein. MMP-8 and MMP-8/TIMP-1 ratio at day 5 were the strongest predictors of mortality within 14 days (Fig 2). The ROC analysis for day 5 MMP-8 for predicting death within 14 days was 0.76 (95% CI 0.65-0.87) (p<0.001) with cut-off value of 239 ng/ml with sensitivity of 74% and specificity of 70% whereas the ROC analysis for day 5 MMP-8/TIMP-1 ratio for predicting death within 14 days was 0.75 (95% CI 0.63-0.87) (p<0.001) with cut-off value 1.02 and sensitivity 79% and specificity 66% (Fig 2). Day 3 and day 5 MMP-8 levels predicted mortality within 28 days and day 5 MMP-8 level predicted mortality within 90 days, however, the AUCs and sensitivity and specificity were lower than for predicting 14 days mortality (S1 and S2 Figs). CRP was not predictive for mortality (Fig 2 and  S1 and S2 Figs).

Additional analyzes for TIMP-1
No association of TIMP-1 to mortality was observed in ROC-analyzes (Fig 2 and S1 and S2 Figs) with AUCs varying from 0.25-0.75 and 95% CI intervals varying between 0.031 and 0.97 and p-value non-significant for each measurement. To further evaluate the prognostic role of TIMP-1 we included only patients with ICU at blood culture collection or within 3 days, however, by this subgrouping the TIMP-1 had no connection to 14, 28 or 90 days mortality when analyzed by univariate analysis (p-value non-significant for each measurement).

Discussion
The main observation of the present study was that plasma MMP-8 and MMP-8/TIMP-1 molar ratio determined at an early stage (day 3 and 5 after positive blood-cultures) were significantly higher in SAB patients with a more severe disease i.e. severe sepsis, ICU treatment need and patients with a deep infection focus as compared to patients without these complications. Interestingly, higher MMP-8 and TIMP-1 levels were associated to presence of a deep infection focus but not to endocarditis. Higher MMP-8 levels on day 3 and especially on day 5 were not only associated to a more severe disease but were also predictive for mortality within 14 and 28 days. When all prognostic parameters were controlled for in Cox regression analysis high day 5 MMP-8/TIMP-1 ratio predicted mortality within 14 or 28 days but not at a later stage as analyzed by day 90 mortality. No association to mortality was seen for the commonly used inflammatory marker CRP. This data suggests that matrix metalloproteinase activation may be one link in the pathological processes in complicated SAB. Comparison of results from the present study to previous reports on the prognostic impact of MMPs and TIMPs in infections or critically ill patients is challenging due to variations in patient profiles, categorization according to illness, exclusion criteria, time-point of MMP or TIMP measurement and follow-up times. Furthermore, the present study is the only one that has included solely bacteremia patients due to one causative agent in contrast to sepsis or other critical illnesses in previous reports.
Second, we observed that chronic renal failure was associated to significantly lower day 3 MMP-8 levels and higher day 3 TIMP-1 levels. Moreover, among patients with chronic renal failure MMP-8, TIMP-1 and MMP-8/TIMP-1 levels had no prognostic impact in SAB. Thus, patients with chronic renal failure were excluded from the main analyzes. Previous reports on MMPs and/or TIMPs have analyzed patients as one group [14,25,28,30,33] or in specific subgroups including e.g. solely acute respiratory distress syndrome [15] or multi-organ failure patients [32]. Moreover, many reports on MMPs and TIMPs in sepsis or critical illness have excluded immunocompromised patients or patients with malignancies or liver disease [13,15,26,27,29]. However, we are not aware of reports on MMPs or TIMPs levels in renal failure or of reports excluding patients with renal failure.
Third, the present study included only bacteremia due to methicillin sensitive S. aureus and each patient received effective antimicrobial therapy from the day when the positive blood culture was drawn. Among previous reports on MMPs and TIMPs in septic or critically ill patients only three studies have reported that 13% -50% of septic patients have had positive blood cultures with two thirds having streptococcal and one third of miscellaneous bacterial etiology [27] or positive blood cultures for Bacteroides fragilis or Candida albicans [31] whereas one report did not mention the pathogens [26]. One study reported that 7% of patients had blood as a primary focus of infection [32]. Delayed effective antimicrobial therapy is a risk factors for poor prognosis [45]. MRSA is associated with poor prognosis and delay in effective therapy [9,46]. Vancomycin treatment increases the risk for persistent SAB compared to treatment with anti-staphylococcal penicillin [47]. Furthermore, all patients received formal ISC which is known to improve clinical management and prognosis of SAB patients [6,7]. All these pitfalls/sources of bias were controlled for in our study setup.
Fourth, the present study included only MS-SAB patients. Thus, the study cannot evaluate how MMP-8, TIMP-1 and MMP-8/TIMP-1 would perform in MRSA bacteremia. However, there were only 6 MRSA bacteremia patients and thus the excluded MRSA bacteremia patients represented only 1.3% (6/430) of the whole patient cohort. Hence, it is plausible to assume that this exclusion had not significantly affected the results.
Due to meticulous clinical and radiological investigations and ISC for each patient a total of 83% had a deep infection focus and 19% endocarditis diagnosed. We observed a strong connection of deep infection foci to higher MMP-8 and TIMP-1 levels throughout the 90 days observation period. However, this trend was not seen for endocarditis. Previous reports on MMPs and TIMPs in septic or critically ill patients have presented high occurrence of infection foci including 16-57% pneumonia, 4% endocarditis and 27-36% abdominal focus, however, the explicit connection of MMPs and TIMPs or MMP/TIMP molar ratio have not been evaluated in these previous reports [15,26,32].
In the present study, day 3, 5 and 28 MMP-8 levels were significantly higher in patients who deceased within 14, 28 or 90 days as compared to survivors. No corresponding trend at was seen for TIMP-1. Previous reports have observed higher MMP-8 on [25] and TIMP-1 [13,23,25,29] at sepsis onset or ICU admission in non-surviving sepsis patients compared to surviving ones. Furthermore, two reports observed that critically ill non-surviving patients at ICU admission or in need of mechanical ventilation, as compared to surviving ones, presented higher MMP-8 and TIMP-1 levels [14,15] but indifferent MMP-8/TIMP-1 molar ratios [15]. Hence, the observations in the present report of higher MMP-8 among non-surviving patients are in line with previous observations [25] whereas the lack of association of TIMP-1 to mortality is in contrast with many previous observations [13,23,25,29].
By ROC analyzes significantly higher day 3 MMP-8 levels were observed in patients who deceased early (within 14 or 28 days) and in day 5 MMP-8 levels of deceased patients regardless of time of death. Day 3 MMP-8/TIMP-1 molar ratio was associated to mortality at an early phase (within 14 days) whereas day 5 MMP-8/TIMP-1 predicted mortality at 14 and 28 days. Few previous reports have performed ROC analyzes to determine the capability of MMP-8, TIMP-1 and MMP-8/TIMP-1 to discriminate surviving and non-surviving septic or critically ill patients. To the best of our knowledge, three reports determined MMP-8 and TIMP-1 at hospital or ICU admission in septic or critically ill patients and presented TIMP-1 sensitivity 60-73% and specificity 45-76% with AUC 0.62-69 and cut-off values 331-531 ng/mL and MMP-8 sensitivity 49% and specificity 60% with AUC 0.55 and cut-off value 127 ng/mL in predicting 30-90 days mortality or overall sepsis mortality [13,15,29]. These results differ partly from the observations of the present study. First, we observed no connection of TIMP-1 to mortality. Second, the ROC analysis results for admission MMP-8 in predicting 90 days mortality in critically ill acute respiratory failure patients resembled the results in the present study where day 5 MMP-8 predicted death within 90 days with an AUC of 0.61, cut-off 149 ng/mL, sensitivity and specificity of 65% and 54% [15].
The observations in univariate and Cox proportional analyzes in the present study of age > 60 years, healthy-nonfatal underlying conditions, ICU treatment, endocarditis and adjunctive rifampicin therapy as parameters with strong prognostic impact have been reported earlier [2,3,4,48]. In univariate analysis, a strong connection of day 3-5 MMP-8 and MMP-8/ TIMP-1 to 14 and 28 days mortality was observed. However, only day 5 MMP-8 and MMP-8/ TIMP-1 -ratio were associated to 90 days mortality. When accounting for all prognostic parameters in Cox proportional regression analysis, day 5 MMP-8/TIMP-1 molar ratio was a strong predictor of 14-and 28-days but not 90-days mortality. However, day 3-5 MMP-8 did not predict mortality in Cox proportional regression analyzes and TIMP-1 was not included in this analysis as no connection of TIMP-1 to mortality was observed in the original univariate analysis. Previous reports have presented admission TIMP-1 as an independent negative prognostic marker for 30-90 days or overall mortality in critically ill, severe sepsis or acute respiratory failure patients [14,15,23,27,29]. These observations are in sharp contrast with the results of the present study but in line with one study that observed no significant predictive value of admission TIMP-1 in septic patients on 30 day mortality [13].
The results of the present study demonstrate a connection of MMP-8 and MMP-8/TIMP-1 and disease severity and mortality in SAB. This suggest that matrix metalloproteinase activation may be one link in the pathological processes linked to disease severity and mortality in SAB. Previous results in septic or critically ill patients suggest that MMP activation may play a role in pathologic processes in bacteremia and critical illness. However, the sensitivity and specificity of the MMP measurements in predicting mortality have been fairly low and suggest that they might not be useful as clinical markers. The possibility to affect MMP activation pharmacologically make these observations interesting and encourage further studies on the role of MMP activation in bacteremia and critical illness [20].
There are limitations and weaknesses of the present study that have to be taken into account when interpreting the results. The patient cohort of the present study was prospectively collected in 1999-2002, however, the MMP-8 and TIMP-1 laboratory analyzes were not performed until year 2015-2016. Clinical practice of SAB develops continuously, however, there are fundamental elements of SAB treatment that have remain unchanged e.g. the importance of non-delayed onset of proper antibiotic treatment and eradication of deep infection foci. The authors view that bedside formal infectious diseases specialist consultation in the present study has guaranteed recording of relevant clinical patient information and hence enabled high standard clinical management of SAB. Laboratory quality testing have ensured that MMP and TIMP remained stable at -80˚C and hence the standard of the laboratory samples are guaranteed despite longtime storage. Hence, the authors view that the patient data of the present study is not outdated and the laboratory samples have been correctly stored and analyzed and the results are reliable.
In conclusion, the present study is the first to demonstrate a connection of plasma MMP-8 and MMP-8/TIMP-1 to disease severity, presence of a deep infection focus and mortality in MS-SAB patients. However, comparison of the results to previous studies on MMPs and TIMPs in sepsis or critical illness is challenging due to differences in patient populations. Future studies need to evaluate the pathogenetic role of MMP-8 and MMP-8/TIMP-1 in bacteremia which might open possibilities for pharmacological therapies to control their activation in bacteremia.