Heart failure (HF) is defined as an abnormal cardiac function and/or structure that lead to the inability of the heart to deliver oxygen to the tissues at a rate that compensates for their metabolic requirements. A major cause of pediatric HF in developing countries is rheumatic fever (RF) and rheumatic heart disease (RHD) (1). RF and RHD present serious problems worldwide according to the Global Burden of Diseases in 2015(2). There have been crucial efforts by the World Health Organization (WHO) to prevent and control RHD (3). Improving the quality of housing and reducing overcrowding in combination with providing penicillin for early management of streptococcal infections are the most important steps needed to eradicate RHD (4). While there has been an evident decline in its prevalence in Egypt since the 1950s, acute rheumatic fever (ARF) and RHD remain national issues, and efforts by the Egyptian authorities to control ARF and RHD are ongoing (5).

Despite advancements in diagnostic work-up techniques, patients with RHD and HF may be diagnosed late or undiagnosed in developing countries due to the lack of echocardiography and experienced cardiologists in rural health centers. Therefore, sensitive and reliable serum markers as a simple laboratory test in conjunction with echocardiography are required for early detection of RHD and HF (6). The importance of studying cardiac biomarkers in the diagnosis and prognostic stratification of children with HF has been well emphasized; these biomarkers may act as future therapeutic targets (7). Similarly, noninvasive tools to monitor patients with HF before the occurrence of clinical deterioration are also required (8).

Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases that belong to the metzincin superfamily. MMPs have a key role in tissue remodeling in cardiovascular diseases (CVDs). MMPs are a promising target of therapy, as the administration of doxycycline ‘a tetracycline antibiotic that acts as a nonspecific MMP inhibitor’ in myocardial infarction patients improved ischemia/perfusion injury. Further studies on the implications of specific MMPs in different CVDs are required to clarify their role in disease progression and to aid the discovery of new therapeutic targets (9).

Gelatinases (MMP-2 and MMP-9) are the most frequently analyzed MMPs in HF. According to the majority of studies, active myocardium remodeling is associated with increased activity and concentrations of MMP-2 and MMP-9, which can be useful markers to identify patients at risk for HF development, and as indicators of patient outcome (10). In one study; circulating MMP-9, in contrast to MMP-2, was the only predictor of late-onset congestive HF (11).

We hypothesized that the serum MMP-9 concentration may be elevated in pediatric patients with RHD and HF, and its level can be correlated with the HF severity. Thus, the aim of the present study was to evaluate the sensitivity and accuracy of the MMP-9 level to predict HF development in children with RHD and to determine its effectiveness to indicate the degree of HF as detected by echocardiography.

Both patient groups were homogenous regarding demographic features (Table I). A total of 59 (60.0%) patients provided a past history (one to two years ago) of RF. The past history of RF, clinical manifestations, and distribution of valve lesions among patients are represented in Table SI.

The mean serum MMP-9 levels showed highly significant differences between the healthy controls (61.0±22.8 ng/ml) and the children with RHD without HF (356.5±74.7 ng/ml), between the healthy controls and children with RHD and HF (422.3±63.5 ng/ml), and between children with RHD without HF and children with RHD and HF (P<0.001 for each) (Table SII).

The univariate analysis showed that the serum MMP-9 level was associated with the risk of HF development [odds ratio (OR), 3.2; 95% CI, 1.3-7.3, P=0.008). Potential confounders from the univariate analysis (Table II) including hemoglobin (Hb)%, TLC, and ASO were introduced in a multivariate model, and the multivariate analysis revealed that MMP-9 was an independent indicator of HF in children with RHD (OR, 2.6; 95% CI, 1.0-6.7, P=0.04) (Table II).

The AUC of MMP-9 in patients was 0.85 (95% CI, 0.76-0.94) (Fig. 1). The MMP-9 level of 386.9 ng/ml was considered the best cutoff point with the maximal diagnostic accuracy to predict HF in children with RHD. At this level, MMP-9 predicted HF with 95% (95% CI, 83.08-99.39) sensitivity, 74.14% (95% CI, 60.96-84.74) specificity, a positive predictive value of 71.70% (95% CI, 61.96-79.75), a negative predictive value of 95.56% (95% CI, 84.67-98.82) and 82.65% (95% CI, 73.69-89.56) accuracy. Cross tabulation of the patients is shown in Table SIII.

There was no significant correlation in any of the parameters, including the duration of symptoms, CRP and ASO titer (Table SIV). MMP-9 appeared to significantly increase with the deterioration of EF and FS (P=0.01 and P=0.02, respectively) (Fig. 2; data not shown). In this cohort; 33 mild and 7 moderate cases were included. The mean ± SD MMP-9 was 413±62 and 464±53, respectively (P=0.055). By the application of the modified Ross classification; 36 patients were in Class II and 4 patients were in Class III. The mean ± SD MMP-9 was 416±63 and 478±29, respectively (P=0.059).

In the present study, serum matrix metalloproteinase 9 (MMP-9) concentrations did not correlate with C reactive protein (CRP) levels in any group. Previous studies considering MMP-9 in rheumatic heart disease (RHD) that can support our results are rare. Previous research has shown that CRP has a positive feedback on serum MMP-9. For example, Cimmino et al examined the effect of CRP on MMP-9 gene expression in vitro in vascular smooth muscle and in vivo in patients with acute coronary syndrome (ACS) (20). Harfoosh et al detected a positive correlation between MMP-9 and CRP serum levels in patients with ACS (21). Although ACS encompasses myocardial inflammation and remodeling, the definite pathology could differ from that in RHD and HF. This can explain the contrast between our results and those of other studies.

In the present study, the mean serum MMP-9 level in the control group was lower than that in both the RHD without heart failure (HF) and RHD with HF groups, indicating that this serum marker can be a promising screening tool for subclinical RHD among school-aged children particularly in high risk endemic areas in conjunction with echocardiography. Although echocardiography has superiority in detecting subclinical RHD, Dougherty et al stated that portable echocardiography is expensive, heavy to transport, with limited battery capacity and functionality; in addition, it requires well-trained cardiologists or medical staff members (16).

Serum MMP-9 at a cutoff point 386.9 ng/ml was found to be a sensitive predictor of HF in children with RHD. Zhao et al (22) found that MMP-9 is overexpressed in the myocardium of patients with RHD congestive HF. Most of the studies concerning MMP-9 in HF, such as studies by Abou-Raya et al (23) Collier et al (24) and Morishita et al (25), suggest that active myocardial remodeling is associated with increased MMP-9 concentrations (10).

Lee et al reported that serum MMP-9 concentrations were significantly higher in patients with RHD >30 years of age, when compared to the controls, and in a patient group of <30 years there was no significant difference. These results are contradictory against ours, and can be explained by the small sample size in that group which included 10 patients and 10 age-matched controls (26).

Kantor and Rusconi reported that HF in childhood and young adolescence often presents at an advanced stage; thus, biomarkers that can support traditional diagnostic tools are valuable. These authors mentioned MMPs as possible markers but the existing data were not sufficient (27).

Similarly, de Couto et al adopted the term ‘pre-heart failure’ as the process of cardiac remodeling indicated by the drop of ejection fraction (EF) which precedes the appearance of symptoms by years, gains from the early interruption and reversal of cardiac remodeling is greater than just treating symptoms (28). Therefore, the follow-up of children with RHD by MMP-9 can predict the decline in left ventricular systolic dysfunction early before symptoms enabling the early intervention before overt irreversible failure occurs.

In the present study, serum levels of MMP-9 were negatively correlated with EF and fractional shortening (FS). These results were also demonstrated by Yan et al (29) who reported that elevated MMP-9 levels were connected with the deterioration in left ventricular (LF) function. According to Yoshihisa et al (30); reduced EF is an independent predictor of increased cardiac event rates. The authors stated that the initial assessment of LVEF is important for deciding treatment and predicting prognosis. Based on this; we assumed that the serum MMP-9 level may be associated with the prognostic status of patients. Radosinska et al (10) proposed that the MMP-9 level may not only be a useful marker to identify patients at risk for HF development but also, it can serve as an indicator of patient outcome. According to Kantor and Rusconi, the most challenging point in pediatric HF is the choice of treatments. This may be an additional value of biomarkers, which may guide decisions of how to manage and predict patient outcomes (27).

To the best of our knowledge; MMP-9 has not been previously studied in pediatric patients with RHD either in or outside Egypt, at the time of writing this manuscript. Therefore, this study adds to our general understading regarding the contribution of MMP-9 to the pathogenesis of RHD and HF in children and draws attention to a new marker and target of therapy for these patients.

On limitation of the present study was that it was conducted on symptomatic patients in a single center; the sample size was also relatively small. Thus, further multicenter studies with larger cohorts involving other types of MMPs are required to confirm these results. Future studies concerning MMP-9 in the screening of school-aged children for subclinical RHD in comparison with echocardiography, and screening of children with known RHD for asymptomatic cardiac dysfunction to allow early and effective intervention before the development of overt HF are highly recommended. Further investigations including the correlation of MMPs with other cardiac biomarkers such as natriuretic peptides (atrial or brain subtypes) or high sensitivity CRP, the severity of symptoms and valve lesions, prospective follow-up of children with HF to study the correlation between MMPs and the degree of deterioration in cardiac function to examine their role in monitoring these patients, and clinical trials to investigate the effects of doxycycline in children suffering from RHD and HF are also recommended.

In conclusion, the MMP-9 serum level can be considered as an independent sensitive indicator with which to identify children with RHD who are at risk of HF events. MMP-9 can be used to assess patients with HF; increased levels may indicate the need to intensify patient medical therapy. MMP-9 was higher in children with RHD either with or without HF than in healthy controls; therefore, MMP-9 is a promising diagnostic marker of subclinical RHD in school-aged children, although further studies are required.