Background

Cardiovascular diseases (CVD) are still the leading cause of death and disability in type 2 diabetic patients despite the huge efforts to prevent and treat them precociously. In these patients, cardiovascular events typically occur 14.6 years earlier and with greater severity than in individuals without diabetes [1, 2]. The enhanced CVD risk and mortality in type 2 diabetes are due to greater atherosclerotic plaque burden, higher atheroma volume, and smaller coronary artery lumen diameter [3]. Multiple mechanisms underlie the atherogenic process in diabetes such as hyperglycemia, dyslipidemia, inflammation, reactive oxygen species, endothelial dysfunction, hypercoagulability, and vascular calcification [4,5,6,7].

Suppression of tumorigenicity 2 (ST2) is a transmembrane receptor expressed mostly on the surface of Th2 and mast cells, and belongs to the Toll-like/IL-1 receptor superfamily. In 2005, Schmitz uncovered the ligand for ST2, interleukin-33 (IL-33), a cytokine involved in Th2-dependent inflammatory processes [8]. The binding of IL-33 to its receptor determines phosphorylation of extracellular signal-regulated kinase (ERK) 1/2, p38 MAPK, JNKs and activation of NF-κB [9]. The soluble form of ST2 (sST2) is generated by alternative splicing and acts as a decoy receptor, and is implicated in the attenuation of the Th2 inflammatory response [10]. The IL33–ST2 pathway was implicated in several inflammatory diseases such as asthma, rheumatoid arthritis, sepsis, autoimmune, fibro-proliferative and cardiovascular diseases [11,12,13]. IL-33/ST2L signaling has cardioprotective effects; it preserves the myocardium against hypertrophy and cardiac fibrosis following pressure overload [14]. sST2 blocks the binding of IL-33 to his transmembrane receptor and the sequent cardioprotective cascade of events.

In the last decade, several evidences allowed to consider sST2 as a biomarker and predictor in CVD. Its role in acute and chronic heart failure is well demonstrated in numerous studies [15,16,17,18,19] enough to induce the FDA to authorize it for clinical use and the American College of Cardiology/American Heart Association to insert it in the update of HF guidelines in 2017 [20].

The aim of our study is to investigate the role of sST2 as a predictor of CV mortality and all-cause mortality in a cohort of patients affected by established atherosclerotic disease and its association with glucose tolerance.

Methods

Description of patients

The Tor Vergata Atherosclerosis Registry (TVAR) (ISRCTN42405215) was previously described [21]. Here, we show a secondary analysis based on follow-up performed in 2017. The data were available for 399 subjects for an average follow-up of 75 months, 6 patients did not answer to follow-up, and 85 patients had missing data on sST2 levels because of no serum available; however, these subjects had clinical characteristics similar to those included in the analysis.

At recruitment, all patients underwent thorough medical history, clinical examination and anthropometric measurements. Patients without history of type 2 diabetes (n = 325) performed an oral glucose tolerance test (OGTT) with 75 g of glucose. Patients were classified as normal glucose tolerant (NGT), impaired glucose tolerant (IGT) or type 2 diabetic patients followed ADA classification [22].

Informed consent and ethics committee approval

An informed written consent was obtained from all participants. The study was approved by the local ethics committee and the reported investigations were carried out in accordance with the principles of the Declaration of Helsinki as revised in 2000.

Laboratory measurements

Soluble ST2 was detected by double measurements in one batch by Quantikine ELISA—Human ST2/IL-1 R4 Immunoassay according to manufacturer’s instructions (Cat. Num. DST200—R&D Systems, Inc.); the average intra-assay coefficient of variation was 5.6% and interassay coefficient of variation was 17.3%. Other laboratory measurements were previously described [21]. GFR was estimated with the MDRD formula.

Statistical analysis

Baseline parameters were reported as mean and standard deviation (SD) for continuous variables and frequencies, and percentages for categorical ones. Differences among groups were analyzed by one-way ANOVA with Hochberg post hoc analysis and χ2 test. Association between continuous variables was assessed by means of Spearman and partial correlation. Time-to-event data were analyzed by means of log rank test or univariate Cox models for categorical or continuous variables, respectively. Univariate and multivariate Cox models were also evaluated, to assess the effect of ST2 after adjusting for potential confounders. The final multivariable Cox model was chosen with forward stepwise regression based on Akaike Information Criterion (AIC), which guarantees a final parsimonious model with best prediction performance. All potential predictors were considered for final inclusion in the multivariable Cox model.

p values < 0.05 were considered as statistically significant. All analysis was performed with IBM software SPSS Version 23 (SPSS, Inc., Chicago, IL, USA) and R version 3.4.2 (Vienna, Austria).

Results

Patient clinical characteristics and impaired glucose metabolism

We divided the 399 patients on the basis of the status of glucose tolerance: 118 were NGT, 138 IGT, 69 had a new diagnosis of type 2 diabetes (neoT2D) and 74 were diabetic (oldT2D) at enrollment, as shown in Table 1. The four groups did not differ for age, sex or BMI. As expected oldT2D patients showed significantly higher levels of fasting plasma glucose and insulin (p < 0.000001), HbA1c (p < 0.000001) and were more insulin resistant (HOMA, p < 0.000001). The oldT2D group had a slight reduction of the glomerular filtration rate (p = 0.014) and lower levels of LDL cholesterol (p < 0.000001) and triglycerides (p = 0.05), probably due to a higher assumption of lipid-lowering therapies compared with the other groups, as shown in Table 1.

Table 1 Clinical and biochemical parameters of patients divided on the basis of the degree of glucose impairment
Full size table

Among all the several cytokines and inflammatory markers evaluated, we found that sST2 plasma levels were significantly increased in oldT2D compared with the three other groups (ANOVA p < 0.00001, Fig. 1a, post hoc Hochberg analysis NGT vs oldT2D p < 0.001, IGT vs oldT2D p < 0.00001, newT2D vs oldT2D p < 0.001), and the difference remained significant after adjustment for age and GFR (p < 0.00001).

Fig. 1
figure 1

Correlations of sST2 with glucose tolerance status and metabolic variables. a sST2 levels according to glucose tolerance status ANOVA p < 0.00001, post hoc Hochberg analysis NGT vs oldT2D p < 0.001, IGT vs oldT2D p < 0.00001, newT2D vs oldT2D p < 0.004 after adjustment for age and GFR (p < 0.00001); b correlation of sST2 with fasting plasma glucose levels (partial correlation R = 0.16, p = 0.002, after adjustment for age, gender and GFR); c correlation of sST2 with HbA1c levels (partial correlation R = 0.17, p = 0.002, after adjustment for age, gender and GFR); d correlation of sST2 with HOMA-IR (partial correlation R = 0.16, p = 0.004, after adjustment for age, gender and GFR)

Full size image

Soluble ST2 and impaired glucose metabolism

We observed that sST2 levels were positively correlated with age (R = 0.23, p < 0.0001) and negatively with GFR (R = − 0.20, p < 0.0001); no correlations were found with BMI, lipid profile, liver function and inflammatory markers (hs-CRP, fibrinogen, TNF-α, white blood cells).

The soluble receptor of IL33 was positively correlated with parameters of glucose metabolism such as fasting plasma glucose (R = 0.16, p = 0.002), HbA1c levels (R = 0.17, p = 0.002) and HOMA (R = 0.16, p = 0.004) after adjustment for age, gender and GFR, as shown in Fig. 1b–d.

We then stratified our cohort according to tertiles of sST2 levels and we found that patients belonging to the highest tertile (sST2 24.18±7.39 ng/ml) were significantly older (p = 0.0001) and showed lower levels of total (p < 0.0001) and LDL cholesterol (p < 0.0001), a worse kidney function measured by GFR (p = 0.03) and higher fasting plasma glucose (p = 0.05), HbA1c levels (p < 0.001) and HOMA index (p = 0.05) as shown in Table 2.

Table 2 Clinical and biochemical characteristics of patients divided in tertiles of plasma sST2
Full size table

Results from follow-up

399 patients from our cohort participate to annual follow-up for 74.98 ± 36.27 months. During this period, 99 deaths occurred (24.8%) of which 32 (8%) due to cardiovascular diseases. We did not find significantly differences in all-cause mortality and CVD mortality among the four categories of impaired glucose metabolism.

A Mantel–Cox analysis revealed that patients belonging to the highest tertile of sST2 displayed an increased rate of all-cause and cardiovascular mortality, as shown in Fig. 2 (all-cause mortality, log rank test, p = 0.045 and CVD mortality log rank test, p = 0.02).

Fig. 2
figure 2

All cause and cardiovascular mortality according to sST2 tertiles. Mantel–Cox analysis showed that the sST2 highest tertile was significantly associated with all cause (a, p < 0.04) and cardiovascular mortality (b, p < 0.02)

Full size image

Among the several parameters and biochemical variables available, sIL6R, TNFa, sVCAM, sICAM, ADMA, only age, sST2 and hs-CRP were found independently associated with all-cause and cardiovascular mortality with univariate Cox analysis, as shown in Table 3.

Table 3 Univariate Cox analysis for all-cause and cardiovascular mortality
Full size table

On the basis of these results, we performed a multivariate Cox analysis and we observed that sST2, included as continuous variable, increased the risk in cardiovascular mortality per SD by hazard ratio 1.050 [95% confidence interval (CI) 1.006–1.097, p = 0.025] after adjustment for age and hs-CRP (Table 4), while it did not significantly change the risk for all-cause mortality.

Table 4 Multivariate cox analysis for all-cause and cardiovascular mortality
Full size table

Discussion

In this study, we evaluated whether sST2 may predict all cause and CVD-mortality and its links with glucose tolerance in a cohort with established atherosclerotic vascular disease in secondary prevention. Our results revealed that increased sST2 is associated with insulin resistance and impaired glucose tolerance. The ST2-IL33 pathway is involved in several chronic inflammatory conditions such as atherosclerosis, obesity and metabolism impairment. In genetically obese diabetic (ob/ob) mice, the administration of recombinant IL-33 reduced adiposity, fasting glucose and improved glucose and insulin tolerance [23]. Furthermore, the selective depletion of fat-resident regulatory T cells via anti-ST2 antibody treatment increased adipose tissue insulin sensitivity in mice [24]. In 2012, Miller and colleagues found in a general population that only fasting plasma glucose, but not insulin resistance, was significantly correlated with ST2 after adjustment for age and sex [25]. Previously another study in 158 subjects revealed that patients with type 2 diabetes exhibited higher sST2 levels compared to healthy controls [26].

We also found that elevated levels of sST2 are associated with increased CVD mortality in our secondary prevention cohort. The role of this decoy receptor as a prognostic biomarker in heart failure is actually well recognized [16,17,18,19, 27]. The IL-33–ST2L signaling showed cardioprotective effects: in experimental models, it reduced myocardial fibrosis, prevented cardiomyocyte hypertrophy and improved myocardial function [28, 29]. The IL-33–ST2 pathway is also involved in the atherosclerotic process since IL-33 determined a reduction in aortic atherosclerotic plaque formation and, conversely, sST2 resulted in development of significantly larger atherosclerotic plaque and increased Th1 response in ApoE −/− mice [30]. In our study, all patients suffered from chronic atherosclerotic disease and those with higher levels of sST2 showed an increased CVD mortality at follow-up. To our knowledge, this is the first study demonstrating the possible role for sST2 as a predictor for a second fatal CV event in a high-risk population.

Our study has some limitations. First, we did not observe a specific effect of sST2 on cardiovascular mortality in subjects with impaired glucose metabolism. Given that sST2 levels correlated with insulin resistance, we can hypothesize that the neutral result on mortality in subjects of diabetes may be explained by the sample size that we analyzed. Second, our study observed a correlation between sST2 and GFR but we cannot ascertain whether the increase in the first is explained by the latter or vice versa sST2 may affect kidney function.

In conclusion, our study is suggestive for a role of sST2 in deterioration of cardiovascular function in subjects at high risk.