Previous studies have shown high levels of both interleukin-6 (IL-6) and C-reactive protein confer an increased risk of long-term mortality in patients with non-ST-segment elevation acute coronary syndrome (NSTEACS).1–6 Moreover, early serial plasma level monitoring of these cytokines in acute coronary syndromes has also been found to be predictive of adverse outcomes,7–13 even in those patients with normal serum troponin T.14 However, most studies considered only early (24h to 72h) presenting values of these biomarkers, but the prognostic role of delayed serial measurements for long-term risk stratification in patients with NSTEACS remains uncertain.

Our objective was to evaluate the prognostic value of serial sampling of IL-6 in comparison to high-sensitivity C-reactive protein (hs-CRP), and to determine whether serial measurements provide additional information to isolated measurements in the long-term risk stratification of NSTEACS patients.

The study population consisted of 216 hospitalized patients with high-risk NSTEACS (Tables 1 and 2). IL-6day1 were positively correlated with baseline hs-CRP (hs-CRPday1) concentrations (r=0.59, P<.001). IL-6day1 weakly correlated with IL-6day30 (r=0.25; P=.002), age (r=0.24; P<.001); heart rate (r=0.16; P=.022), left ventricular ejection fraction (r=-0.16; P=.019); hemoglobin (r=−0.24; P<.001), leukocytes (r=0.22; P=.001), troponin T (r=0.30; P<.001), and GRACE (Global Registry for Acute Coronary Events) score (r=0.26; P<.001).

Over the study period, a total of 44 (20.3%) patients had the composite endpoint (16 deaths, 26 non-fatal myocardial infarctions and 15 acute decompensated heart failure hospitalizations). Of these, 23 (10.6%) patients had adverse events after the first 30 days (13 deaths, 21 non-fatal myocardial infarctions, and 11 acute decompensated heart failure). Median IL-6 levels were higher among patients with the composite endpoint than those without, both at presentation (15.38 [7.38-31.30] pg/mL vs 7.13 [4.29-16.35] pg/mL; P<.001) and at day 30 (8.59 [4.40-10.83] pg/mL vs 4.12 [2.67-5.80] pg/mL; P=.001). By contrast, only hs-CRPday1 levels (but not hs-CRP on day 30 [hs-CRPday30]) were higher among those with the composite endpoint (1.50 [0.36-3.99] mg/dL vs 0.53 [0.23-1.28] mg/dL; P=.009 for hs-CRPday1; and 0.39 [0.10-1.29] mg/dL vs 0.27 [0.13-0.44] mg/dL; P=.220 for hs-CRPday30).

Figure 1 shows a significant decrease in concentrations of both biomarkers from presentation to day 30 regardless of adverse events, although those with events had higher concentrations of IL-6 at both times. Given the higher levels at presentation, the absolute decreases of IL-6 and hs-CRP were actually higher in patients with events (–5.87[–0.19 to –19.01] pg/mL vs –2.38[0.58 to –10.81] pg/mL; P=.019 for IL-6 and –0.72[–0.26 to –1.60] mg/dL vs –0.19[0.04 to –0.90] mg/dL; P=.009 for hs-CRP; respectively), whereas the relative changes of both biomarkers were similar in both groups (–46%[–6 to –65] vs –41%[8 to –76]; P=.627 for IL-6 and –61%[–49 to –87] vs –43%[15 to –83]; P=.190 for hs-CRP).

Figure 1.

Change in A: interleukin-6. B: high-sensitivity C-reactive protein concentrations, from presentation to day 30 in subjects with and without adverse events. The median interleukin-6 and high-sensitivity C-reactive protein values and interquartile range (error bars) plotted against the day of sampling. hs-CRP, high-sensitivity C-reactive protein; IL-6, interleukin-6.

(0.11MB).

When receiver operating characteristic analyses were undertaken to evaluate the prognostic accuracy of both inflammatory biomarkers, IL-6day30 concentrations showed the best prognosis ability to discriminate between those patients with and without the composite endpoint (area under the curve [AUC]=0.72; 95% confidence interval [95%CI], 0.64-0.80), followed by IL-6day1 (AUC=0.67; 95%CI, 0.61-0.74), whilst hs-CRPday1 had overall lower prognostic accuracy (AUC=0.59; 95%CI, 0.52-0.66) (Table 3).

After multivariate Cox regression analyses (Table 4), IL-6 levels, as quantitative variables, were independent predictors of adverse events at the 2 time points (IL-6day1 per pg/mL, HR=1.006; 95%CI, 1.002-1.010; P=.002; and IL-6day30 per pg/mL; HR=1.047; 95%CI, 1.021-1.075; P<.001). In addition, IL-6 levels above the median values both at presentation (>8.24 pg/mL; HR=1.290; 95%CI, 1.025-1.502; P=.027) and at day 30 (>4.45 pg/mL, HR=1.453; 95%CI, 1.059-1.994; P=.021) were significantly independent predictors of events (Fig. 2), as were its absolute changes (per Δ pg/mL; HR=1.050; 95%CI, 1.024-1.077; P<.001). However, both hs-CRP concentrations and the absolute changes in hs-CRP were not independently associated with adverse outcomes (Table 4).

Figure 2.

Cox regression risk dichotomized analysis for the composite endpoint (all cause death, nonfatal myocardial infarction, or acute decompensated heart failure) as a function of interleukin-6 levels above the median (>8.24 pg/mL for baseline interleukin-6 and >4.45 pg/mL for interleukin-6 levels on day 30). HR, hazard ratio; IL-6, interleukin-6.

(0.08MB).

The potential enhanced value from the combined use of serial measurements of IL-6 levels for prognosis risk stratification is shown in Figure 3. This figure shows that the lowest event rate was observed in those patients with IL-6 levels below the median values at presentation and at day 30 (n=43; 4.7%; score=0), whereas the event rate observed among patients who had IL-6 levels above the cut-off values at both time points assessed throughout the study protocol was considerably higher than in any other group (n=40; 35%; score=2; P<.001). In adjusted analysis, for each measure of elevated IL-6 levels in the combined score (from 0 to 2) a significantly higher risk of events was observed (per elevated measure, HR=1.35; 95%CI, 1.090-1.670; P=.005).

Figure 3.

Adverse events rate according to the presence of none, 1 or, both measurements of interleukin-6 above the median values at presentation and at day 30. IL-6day1, baseline interleukin-6; IL-6day30, interleukin levels on day 30.

(0.1MB).

Lastly, the addition of IL-6day30 concentration to a multivariable model, adjusted for other risk factors (including IL-6day1 and GRACE risk score), was associated with a significant improvement in prediction performance and accuracy. The net reclassification improvement from the addition of IL-6day30 concentrations was 0.39 (95%CI; 0.14-0.64, P=.002), whereas the relative integrated discrimination improvement was 0.49 (95%CI, 0.12-0.66; P=.001). The probability of correctly predicting events and no events when IL-6day30 was added to the clinical multivariate model was reflected particularly in the percentage of non-events correctly reclassified (48%), while the percentage of events reclassified was −9%. The C-index increased from 0.71 (95%CI, 0.63-0.78) to 0.80 (95%CI, 0.72-0.86); P=.042, with the addition of IL-6day30 to the clinical multivariate model.

In the present study, we examined the prognostic value of serial measurements of circulating levels of IL-6 and hs-CRP in a hospitalized population with NSTEACS. Consistent with the results of earlier studies,2,21,22 we found high IL-6day1 to be associated with poor long-term outcomes. More importantly, to the best of our knowledge, we demonstrated for the first time, that serial sampling of IL-6day30 levels of NSTEACS presentation offers complementary information to its baseline levels and other clinical variables (including GRACE risk score) for long-term risk stratification. These results are novelty, considering, the use of contemporary reclassification analyses for the assessment of the prognostic value of serial measurements of IL-6 levels in this clinical scenario. Accordingly, our results support the use of repeated IL-6 testing in NSTEACS patients once the patients have been stabilized. A good strength of our study is that blood was sampled at a time when patients were considered to be clinically stabilized (30 days after presentation); yet IL-6 values were able to identify those patients who were at a lower risk for long-term adverse clinical events with high negative predictive values (>90%). In this sense, IL-6 monitoring may help us to identify patients who might not benefit from more aggressive management and monitoring.

IL-6 is a ubiquitous cytokine with a crucial role in leukocyte and endothelial cell activation that promotes increased synthesis of a wide variety of acute phase reactant proteins, including C-reactive protein, and represents the main systemic mediator of the acute response to tissue damage. Furthermore, IL-6 is a key inflammatory cytokine in atherothrombotic disorders. IL-6 is expressed at the shoulder region of atherosclerotic plaques and acts locally to enhance recruitment of monocytes and T cells, with a consequent increase in plaque instability, and acts systemically to induce a pro-inflammatory and pro-thrombotic state.23–25 In this study, we found that IL-6day1 levels are closely related to long-term prognosis of patients with NSTEACS, which reaffirms the prognostic significance of the pro-inflammatory status during the initial phase of an acute coronary syndrome.

Our results also emphasize the importance of monitoring “dynamic changes” in the inflammatory state of NSTEACS patients. Previous studies have shown that patients with unstable angina who had a decrease in IL-6 levels within 48 h to 72h of hospitalization, presented an uncomplicated in-hospital course, whereas patients with an increase in these levels had a higher rate of in-hospital14 or 30-day events.10,12 The delayed measurement of inflammatory biomarkers in assessing the long-term predictive value has 2 potential advantages. The first advantage is that we overcome the “problem” that is posed on prognosis by the easy access to revascularization, either during the hospital admission (urgently) or early after discharge (electively). This early invasive treatment in NSTEACS patients may change the natural history of coronary artery disease and thus attenuate the long-term prognostic value of an inflammatory marker measured during the acute phase. Indeed, Ozer et al. demonstrated that early coronary angioplasty modifies levels of inflammatory biomarkers, including the type of stent used.26 The second advantage is that, 4 weeks after the coronary event, the excess inflammatory response caused by the event should have decreased to the new steady state level or should be “normalized”, and inflammatory markers may then reflect the underlying chronic inflammatory process and thus, be better indicators of the severity of the disease. Consequently, a persistent elevation of inflammatory biomarkers beyond the acute phase suggests prolonged inflammation and could represent a subclinical hallmark of recurrent instability.27

In our study, IL-6 levels showed a significant decrease at 30 days regardless of the occurrence of events but measurement of IL-6day30 levels provides additional information to isolated measurements in the long-term risk stratification of these patients. It supports the hypothesis that a persistent inflammatory activation, as indicated by higher levels of IL-6, is an important predictor of long-term prognosis in patients with NSTEACS.

There are controversial results in the literature regarding the prognostic value of inflammatory markers. The SIESTA study (Systemic Inflammation Evaluation in patients with non-ST segment Elevation Acute Coronary Syndrome) evaluated the prognostic value of hs-CRP, IL-6, and other markers of inflammation in patients with low to moderate risk NSTEACS.28 In this study, inflammatory markers failed to provide independent prognostic information at 6 months and 12 months of follow-up. However, these apparently discrepant results may be attributed to differences in the population characteristics, including the higher risk profile and higher inflammatory burden of our patients (mainly reflected in higher levels of IL-6 in our population).29,30 A paper by Rallidis et al. also studied the prognostic value of cytokines monitoring after 6 weeks of presentation in patients with unstable angina, but IL-6 levels at 6 weeks were not independent predictors of long-term prognosis.13 One possible explanation could be the inclusion of the event revascularization because of new or worsening symptoms in the combined endpoint. Previous studies have failed to demonstrate an association between inflammatory markers and angina31 or inflammatory markers and late angiographic stent restenosis.32 Indeed, circulating biomarkers of inflammation may be more related to acute plaque rupture and thrombosis rather than to slowly progressive occlusive coronary aterosclerosis.33

On the other hand, our composite endpoint also included acute decompensated heart failure. Previous studies have shown that elevated IL-6 concentrations in patients with coronary artery disease preceded progression to clinically overt heart failure.34 Also IL-6 may by itself exacerbate the damage that results from minor myocardial necrosis/injury or may stimulate muscle atrophy and myocardial failure during acute coronary syndrome.35 This may be 1 of the reasons explaining why circulating IL-6 levels have been implicated on acute coronary syndrome and heart failure, and their increase is associated independently, with long-term mortality/acute decompensated heart failure in an acute coronary syndrome population.22

Our results may suggest that the prognostic value of serial IL-6 measurements may be superior to those of hs-CRP. The lack of predictive value of hs-CRP in our study contrasts with data in previous reports.3,36,37 This finding could be explained by the low statistical power of our study. However our results agree with other studies showing a limited role of hs-CRP for risk prediction. Thus, previous reports have showed that hs-CRP has added modest information to the overall prediction of risk resulting from the assessment of conventional cardiovascular risk factors.30,38 It probably suggests that hs-CRP predictive usefulness might not be universal. Further studies are needed to address this challenging issue in different populations.

Among NSTEACS patients, serial sampling of IL-6 levels at 30 days provides complementary information for the long-term risk stratification, and may help to identify patients who might not benefit from more aggressive management and monitoring.

This work was partially supported by a research grant FFIS/CM09/016 from Cajamurcia and a research grant Abbott/12399 from Murcia University.

None declared.