Data was collected between March 1999 and April 2002 as described earlier [17‐19]. Cases were ACS patients admitted in the heart intensive care unit at Lund University Hospital. The 343 patients included 108 unstable angina pectoris (UAP) and 235 acute myocardial infarction (AMI) patients. Forty-eight patients, who were invited to the study, chose not to participate and 21 patients died before the appointment. After a 6-month recovery period (median 350 days, IQR 434 days), 157 (45.8%) patients agreed to participate in the resampling.

The inclusion criteria for both cases and controls were age under 80 years, no cognitive intellectual disability, and no operations or chemotherapy within the previous 4 weeks. Twelve patients were excluded from the study because of aortic aneurysm, pulmonary embolism, myocarditis, pericarditis, unspecific precordial pain, or atrial fibrillation. Data on demographic factors were collected by a questionnaire. Patient history and use of medications were registered, and blood samples were drawn.

Cases were diagnosed with typical symptoms, laboratory measurements, and electrocardiogram (ECG). AMI and UAP were diagnosed according to prevailing criteria in 2000. AMI was diagnosed if the patient had two of the following criteria: chest pain related to exercise lasting over 20 min or changes in ECG, such as ST-elevations followed by T-wave inversion or new Q-waves, or an increase in Creatine kinase-MB (CK-MB) to more than twice the upper limit of the normal value (> 5 μg/l). UAP was diagnosed if two of following criteria were fulfilled: continuous chest pain, ST-segment depression in the ECG (<1 mm), or elevation of CK-MB (5 < CK-MB < 10 μg/l) or troponin T (0.05 < TnT < 0.10 μg/l).

All subjects of the study were followed up on average for 6 years (range 4.56–7.13) to the end of the study or to a major adverse cardiac event (MACE), i.e., cardiovascular death or hospitalization for an ACS event. During the follow-up, 150 patients suffered MACE, including 61 fatal and 89 non-fatal events. Among the patients, whose recovery samples were available, 63 patients suffered a MACE, including 14 fatal and 49 non-fatal events.

A subsample of patients who suffered a non-fatal (N = 23) or fatal (N = 7) MACE and whose both acute and recovery phase samples were available were randomly selected for gelatin zymography. Zymographies were run also on a set of controls (N = 28) that were matched for age and gender to the cases on a group level. The total number of samples in these analyses was 58. The patients in this subsample differed from the rest of the cases only as regards to systolic blood pressure (150 vs. 130 mmHg, p = 0.033).

Serum without activators was collected from the patients within 24 h after the diagnosis. The research nurse collected 70 ml of blood from controls. The blood samples were transferred on ice to the central laboratory for centrifugation. All samples were frozen (− 20 °C) until laboratory analyses. In the in-house laboratory quality control, no significant effect of storage time on biomarker levels has been observed [20].

TIMP-1-ELISA (R&D Systems, Minneapolis, MN, USA) and MMP-9-ELISA (GE Healthcare UK Limited, Amersham Place, UK) were performed according to manufacturer’s instructions on diluted samples (1:10 in TIMP-1 and 1:20 in MMP-9). The inter-assay CV% of TIMP-1 was 8.2% (N = 12) and for MMP-9 9.2% (N = 12). The detection limits were 0.08 and 0.05 ng/ml, respectively. TIMP-1 can form a complex with MMP-9 in a 1:1 stoichiometry with a high affinity [21]. For calculation of MMP-9/TIMP-1, stoichiometric molecular ratio molecular weights of 92,000 g/mol (MMP-9) and 28,000 g/mol (TIMP-1) were used. The recovery phase levels of MMP-9 and MMP-9/TIMP-1 were subtracted from those of the acute phase to calculate the Δ levels.

Gelatin zymography was performed using a previously described technique [22] with minor modifications to explore activatable MMP-9 in vitro. The incubation time (16 h) and sample volume (1 μl) were based on pilot zymography analyses with self-cast gels using fluorescent gelatin substrate. Gelatin zymography analyses of selected samples were performed with commercial gels (BIO-RAD 10% Ready Gel® Zymogram Gel, 10 well, 50 μl, #161-1167, CA, USA). Experiments were conducted with serum samples as such and with samples processed with APMA for in vitro activation of MMP-9. Serum was incubated at a final concentration of 1 mM APMA at + 37 °C for 1 h. Subsequently, all samples were incubated in Laemmli buffer at room temperature for 1 h. After incubation, the samples were applied on gels and electrophoresis was performed in + 4 °C for 2 h. After electrophoresis, the gels were incubated at room temperature in the first wash buffer (50 mM Tris-HCl, pH 7.5, containing 2.5% Tween 80 and 0.02% NaN₃) for 30 min, in the second buffer (50 mM Tris-HCl, pH 7.5, containing 2.5% Tween 80 and 0.02% NaN₃ and supplemented with 1 μM ZnCl₂, 5 mM CaCl₂) at room temperature for further 30 min, and in the last buffer (the same as the second buffer but without 2.5% Tween 80) at 37 °C for 16 h. The gels were stained with 0.1% Coomassie Brilliant Blue and destained with 20% methanol/10% acetic acid. Low-range prestained SDS-PAGE standards (BIO-RAD, #161-0305, CA, USA) were used in every gel. MMP-9 (Proteaimmun 100 ng/μl) was used as a positive control in every gel, and the molecular weights of the gelatinolytic zones were compared to the positive control. The gels were scanned with LI-COR ODYSSEY (Lincoln, NE, USA), and data was analyzed with Image Studio software. The data was presented as arbitrary scanning units of intensities.

The distribution of variables was tested before statistical analysis. Normally distributed continuous variables are presented as means and standard deviations (SD). The statistical significance of the differences between the groups was tested by Student’s t test or ANOVA of logarithmically transformed values. Categorical variables were tested by chi-square test. The ELISA measurements of MMP-9 and TIMP-1 as well as the gelatin zymography results displayed a skewed distribution and are presented as medians and interquartile ranges (IQR). Statistical significance was tested by using non-parametric Mann-Whitney U test or Wilcoxon signed-rank test. The diagnostic sensitivity and specificity of MMP-9 and MMP-9/TIMP-1 were calculated by receiver operating characteristics (ROC) from logarithmically transformed values. Two different multivariate logistic regression models were used to determine the association of MMP-9 and MMP-9/TIMP-1 molar ratio with ACS. The first model was stratified for age and sex; the second model was stratified for age, sex, and adjusted for C-reactive protein (CRP), cholesterol, and smoking. For the follow-up data, we used Cox regression model adjusted for age and sex. MACE and its subgroups, fatal and non-fatal MACE, were used as endpoints. Correlation analyses were performed by Spearman correlation. The analyses were performed using IBM SPSS Statistics 22.

Characteristics of the cases and controls are presented according to acute phase serum MMP-9 quartiles (Table 1). CRP and smoking status differed statistically significantly between the quartiles in both groups. No other significant differences were observed. More detailed characteristics of the subjects have been presented previously [18]. The mean (SD) age of the cases and controls was 63.3 (9.2) and 63.0 (9.2) years and they included 21.3% and 22.1% women, respectively. Of the cases, 235 patients (68.5%) were diagnosed with AMI and 108 (31.5%) with UAP.

The serum MMP-9 and MMP-9/TIMP-1 molar ratio were significantly higher in ACS patients than in controls (p < 0.001) (Table 2; Supplement 1). The same difference was seen in both AMI (p < 0.001) and UAP (p < 0.001). Both MMP-9 (p < 0.001) and MMP-9/TIMP-1 (p = 0.023) were higher in AMI compared to UAP. The association of serum levels with ACS was investigated with multivariate logistic regression models, which are presented in Table 3. Both MMP-9 concentration and MMP-9/TIMP-1 were strongly associated with ACS and its subgroups.

In ROC analyses, both MMP-9 and the MMP-9/TIMP-1 molar ratio distinguished ACS from the controls with an AUC (95%) of 0.742 (0.704–0.781, p < 0.001) and 0.702 (0.662–0.742, p < 0.001), respectively. Smoking decreased the diagnostic value of serum MMP-9 in ACS; in the ROC analysis, the sensitivity and specificity were higher in non-smokers than smokers with an AUC of 0.765 (0.722–0.808, p < 0.001) and 0.659 (0.560–0.758, p = 0.003), respectively. Serum MMP-9 had a significant correlation with CRP (r = 0.453, p < 0.001), CK-MB (0.278, p < 0.001), and troponin T (0.291, p < 0.001).

Recovery phase samples were collected from ACS patients not earlier than 6 months after the acute event. Characteristics of patients in recovery phase and according to the difference of MMP-9 concentration between acute and recovery phase, i.e., ΔMMP-9, are presented in the supplementary material (Supplements 2 and 3). In the recovery phase relative to the acute phase, serum MMP-9 concentrations and MMP-9/TIMP-1 decreased 49 and 34% (p < 0.001), respectively (Table 2; Supplement 1). The recovery phase MMP-9 concentrations did not differ significantly from those of controls, but the MMP-9/TIMP-1 ratio remained significantly higher than the levels observed in controls (Table 2; Supplement 1).

The cases were followed up for 6 years, and MACE was registered. Cases experiencing MACE during follow-up had significant differences in CRP between the acute phase MMP-9 quartiles (CRP increasing in a dose-dependent manner). No other significant differences were observed in the characteristics between the quartiles (data not shown).

We evaluated if serum MMP-9 and MMP-9/TIMP-1 molar ratio or their Δ values predict MACE. Elevated acute phase MMP-9 was a significant predictor of fatal events with a HR 2.88 (1.32–6.27, p = 0.025, Q4 vs. Q1), but not non-fatal ones. Elevated recovery phases MMP-9 and MMP-9/TIMP-1 were significant predictors of MACE with HRs 4.15 (1.87–9.23, p = 0.006) and 3.32 (1.48–7.42, p = 0.009) (Fig. 1), especially non-fatal events [6.05 (2.27–16.12), p = 0.004 and 6.89 (2.52–18.83), p = 0.001].

The highest ΔMMP-9 values presented protective predictive value of MACE with a HR 0.44 (0.20–0.97, p = 0.003) (Fig. 1) and especially non-fatal events [0.24 (0.09–0.59), p = 0.001]. Also, high ΔMMP-9/TIMP-1 was protective from non-fatal MACE with a HR 0.29 (0.11–0.73, p = 0.019).

Zymography was utilized to explore activatable MMP-9 in vitro. The serum MMP-9 concentrations obtained by ELISA correlated significantly with the total intensity units obtained by gelatin zymography (r = 0.397, p < 0.001) (Fig. 2a). In all samples, MMP-9 could be activated with APMA; the difference was significant for all comparisons (p ≤ 0.001 for all). Other gelatinolytic active MMP, MMP-2, was observed only in pro-form (72 kDa), but not in proteolytically activated form (64 kDa) (Fig. 2b). High molecular size (> 100 kDa) gelatinolytic species were also detected.

Medians of active and APMA-activatable MMP-9 were compared between cases and controls. When cases were analyzed as one group (endpoint being MACE), there were no statistically significant differences in the active and activatable MMP-9 between the cases and controls in acute phase. However, both molecular forms of MMP-9 decreased in the recovery phase (p = 0.179 for active and p = 0.003 for activatable) below the levels observed in controls. Additionally, values were calculated separately in cases with different endpoints both in the acute and recovery phase (Fig. 3). In cases with different endpoints, active and activatable MMP-9 decreased from the acute phase to the recovery phase, but the difference was significant only for activatable MMP-9 in cases suffering a non-fatal MACE (p = 0.018) (Fig. 3). Patients with fatal MACE during follow-up had significantly higher activatable MMP-9 zymography intensities (p = 0.028) and MMP-9 ELISA concentrations (p < 0.001) in acute phase than controls (Fig. 3). In the zymography subpopulation, MMP-9 concentrations obtained by ELISA differed statistically significantly between acute and recovery phase in cases with non-fatal (p < 0.001) and fatal MACE (p = 0.028) (Fig. 3).

Proteolytically activated MMP-2 lanes (64 kDa) were not observed in our zymography analysis, only MMP-9 was proteolytically activated (Fig. 2b). Medians of pro-MMP-2 (72 kDa) were significantly higher in cases with a fatal MACE when compared to controls both without (p = 0.013) and with (0.022) APMA-activation and when compared to non-fatal MACE without (p = 0.016) and with (p = 0.025) APMA-activation (Supplement 4). The corresponding values did not differ significantly between non-fatal MACE and controls.

MMP-9 can be utilized as an early stage biomarker, because its elevation reflects atherosclerotic plaque rupture and myocardial tissue destruction. Furthermore, MMP-9 has prognostic value, which is important in secondary prevention and planning of personalized treatment.