Early Invasive Strategy for Unstable Angina: a New Meta-Analysis of Old Clinical Trials

Randomized controlled trials (RCTs) were conflicting to support whether unstable angina versus non-ST-elevation myocardial infarction (UA/NSTEMI) patients best undergo early invasive or a conservative revascularization strategy. RCTs with cardiac biomarkers, in MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials from 1975–2013 were reviewed considering all cause mortality, recurrent non-fatal myocardial infarction (MI) and their combination. Follow-up lasted from 6–24 months and the use of routine invasive strategy up to its end was associated with a significantly lower composite of all-cause mortality and recurrent non-fatal MI (Relative Risk [RR] 0.79; 95% confidence interval [CI], 0.70–0.90) in UA/NSTEMI. In NSTEMI, by the invasive strategy, there was no benefit (RR 1.19; 95% CI, 1.03–1.38). In the shorter time period, from randomization to discharge, a routine invasive strategy was associated with significantly higher odds of the combined end-point among UA/NSTEMI (RR 1.29; 95% CI, 1.05–1.58) and NSTEMI (RR 1.82; 95% CI, 1.34–2.48) patients. Therefore, in trials recruiting a large number of UA patients, by routine invasive strategy the largest benefit was seen, whereas in NSTEMI patients death and non-fatal MI were not lowered. Routine invasive treatment in UA patients is accordingly supported by the present study.


Baseline characteristics. Eight prospective randomized placebo-controlled clinical trials (RCTs) involv-
ing 10412 patients (range, 131 to 2457 patients per trial) were included in the analysis. The primary characteristics of the eight included trials are listed in Table 1. Patients were admitted to the hospital mainly because of UA/NSTEMI, and enrollment in the routine invasive intervention arm of the trial was completed within 98 hours of admission. Duration of the follow-up periods ranged from 6-24 months and few patients were lost to follow-up analysis. Some baseline patient clinical characteristics were different between the two randomized groups (NSTEMI versus UA/NSTEMI) of all the included studies (Table 1).
Overall clinical outcomes at the end of follow-up. Compared with the conservative strategy, the OR for death with an invasive strategy was 0.93 (95% CI, 0.79-1.09), the OR for recurrent nonfatal MI was 0.86 (95% CI, 0.76-0.97), and the OR for death or MI was 0.92 (95% CI, 0.84-1.02) (Fig. 2).
Outcomes for the two presentations. From randomization to the end of follow-up. Based on the pooled results, the use of routine invasive strategy in studies that randomized UA/NSTEMI patients was associated with a dramatic reduction for the composite ischemic events with 20-30% lower odds based on the definitions of death and MI used in these trials (Fig. 3a). In contrast, there was no benefit of the use of such strategy (RR, 1.19; 95% CI, 1.03-1.38) in NSTEMI. The use of a selective invasive strategy in studies that only randomized NSTEMI patients was associated with moderate reductions in the risk for death and MI and with a significant effect on overall composite ischemic events (Fig. 3b). The observed effects were consistent among most evaluated trials except MATE trial among UA/NSTEMI studies and except VINO trials among NSTEMI studies.
From randomization to discharge. A routine invasive strategy was associated with significantly higher odds of the combined endpoint in both UA/NSTEMI (RR, 1.29; 95% CI, 1.05-1.58) and NSTEMI (RR, 1.82; 95% CI, 1.34-2.48) (Fig. 4a,b). There was also a significant increase in the invasive arm of both the index death and MI (Fig. 4b).
Heterogeneity between trials. There was no evidence of heterogeneity among trials for the composite end points of death or MI (heterogeneity: Q:12.43, P:0.087, I 2 :43.7%), death (heterogeneity: Q:7.06, P:0.42, I 2 :1.4%) or MI (heterogeneity: Q:9.50, P:0.22, I 2 :28.0%) when the period of time from randomization to the end of follow-up was considered (Fig. 2). In addition, there was no evidence of heterogeneity among UA/NSTEMI trials. Some heterogeneity was found among NSTEMI trials during time from randomization to discharge (Fig. 4b).

Discussion
We found that a routine invasive strategy is of most benefit in trials recruiting a large number of UA patients, whereas it cannot be proven to reduce death or non-fatal MI among NSTEMI patients. Potential clinical benefits from PCI do not seem to favorably affect the overall prognosis of the index MI at a follow-up exceeding 1-year.  (Table 1). Outcomes from randomization to the end of follow-up. Studies that only randomized biomarkerpositive patients (NSTEMI) were analyzed separately and showed a 1.6-fold increase in the relative risk of death and MI in the invasive arm at a mean follow-up of 13.7 months. On the opposite, the composite end point was significantly decreased by an invasive strategy in those studies that did not require positive cardiac biomarker status as an inclusion criterion (UA/NSTEMI). The observed effects were consistent among most evaluated trials except MATE among UA/NSTEMI and VINO among NSTEMI trials, respectively.
The exceptions. An exception to the potential hazards of an early invasive strategy in NSTEMI patients was the VINO trial. It should be noticed, however, that the VINO trial enrolled high-risk individuals with 53% of the patients having baseline Killip class II or III. Observational data have shown that Killip class II and III patients have significant mortality benefit from an invasive strategy while patients with Killip class I do not 33,34 . The high percentage of patients with heart failure in the VINO trial may explain, therefore, why a robust benefit for an early invasive strategy was found, despite the small cohort of NSTEMI. An exception to the benefits of an early invasive strategy among trials recruiting a large number of UA patients was the MATE trial. Yet, in this trial, approximately 30% of the patients enrolled presented with acute ST-elevation myocardial infarction (STEMI) and only 23% of the study population had segment depression on the initial ECG. Thus, differently from the other UA/NSTEMI studies in our analysis, MATE randomized patients with borderline selection criteria, in whom the overall prognosis was partially affected by a late revascularization of the index MI. According to the Occluded Artery Trial no discernible benefit at four year follow-up was found among patients with occlusion of the infarct-related artery following a strategy of routine PCI late after acute MI 35 . Outcomes from randomization to discharge. Our data are concordant with prior doctrine that early or inhospital coronary revascularization is fraught with hazard in patients with non-ST elevation acute coronary syndromes 5,36 . An updated meta-analysis of randomized trials showed that the routine invasive group had significantly more death or MI during the initial hospitalization compared with the selective conservative group 9 . In our study the primary outcome of death occurred in 1.7% of patients in the routine-intervention group and in 1% of those in the conservative strategy group. The difference was statistically significant among NSTEMI patients (RR, 1.96; 95% CI, 1.03-3.73) but not among UA/NSTEMI patients (RR, 1.42; 95% CI, 0.96-2.10). It is, therefore, reasonable to assume that overall there is an immediate hazard with any invasive procedure, which can be mitigated by the type of clinical presentation. The more severe is the disease, the greater is vulnerability. Patients with MI are known to be a more fragile group from the cardiovascular point of view than those with UA, and as so they have the greatest hazard 37,38 . Comparison with previous meta-analyses. Many old meta-analyses on this topic did not include the most recent trial, the ICTUS study, published in 2005 4,5 . More recent studies have shown that the invasive strategy had null effects on outcomes at 12 month follow-up [6][7][8][9] . The most recent meta-analysis 10 of the only trials with long-term outcomes (FRISC-II, ICTUS, RITA-3) showed a sustained reduction in the rate of cardiovascular death or MI by using a routine interventional strategy only at 5 year follow-up.
The findings from our analysis differ substantially from those of previous studies, as an early invasive approach was not of benefit at a mean follow-up of 13.7 months in those trials that included only NSTEMI patients, but benefited those including UA patients.

Unstable angina versus Non ST-Elevation Myocardial Infarction.
Observations on the natural history of UA have shown an alarming incidence of death (6%) and the need for coronary revascularization by either PCI or coronary artery bypass graft in 27% of patients at 1-year follow-up 39 . Thereafter, the long-term outcome of UA is more favorable with a mortality rate of 2-3% in the following 7 years from hospital discharge. These findings stand in stark contrast to the low (2.5%), 1-year mortality reported from the ICTUS trial, where only NSTEMI patients were enrolled. Yet in this study, mortality rates at 5-year follow-up were much greater than those reported for UA patients, respectively at 11.1 and 9.9% after routine or selective invasive procedures 40 . These diverging results cannot be explained by differences in clinical practice. The different findings in studies enrolling UA/NSTEMI participants may, therefore, be due to differences in the natural history of UA/NSTEMI. The strength of a prognostic factor may be greater during one period of time than another, which may underlie different pathogenetic substrates and obscure the perception of the best therapeutic strategy. In this case, the potential clinical benefits from PCI do not seem to favorably affect the overall short-term prognosis of NSTEMI, which is mainly given by the amount of ventricular injury due to the index event.
Benefits at short versus long-term follow up. Current literature indicates that an invasive strategy had favorable effect on mortality at 5 year follow-up 5 . It should be noted, however, that the risk of late adverse outcomes may be related to many factors 10 . Differences in patient selection and supportive care can lead to better outcomes late after PCI. Patients undergoing PCI are younger and more motivated. Thus, they are more likely to be adherent to medication prescriptions and are also more likely to participate in cardiac rehabilitation 41 . Earlier benefit of PCI, on the opposite, may support the plausibility of a true causal relationship. It takes a short time to demonstrate the potential of a new therapeutic strategy.
Guidelines and take home message. When we performed a PubMed search in April 2013, we found no recent trials on the subject of invasive versus conservative treatment in patients with unstable coronary syndromes. The AHA/ACC treatment guidelines for UA/NSTEMI recommend (class I) invasive treatment in patients with UA/NSTEMI who have an elevated risk for clinical events as defined by risk scores and additional risk factors (e.g LVEF less than 40%, prior CABG and PCI within 6 months) 1 . On the other hand, the Task Force on the Management of Myocardial Revascularization of the ESC recently stated in their document that timing of angiography and revascularization should be based on patients risk profile 42 . The same guidelines recommend an early invasive strategy within 24 hours in patients with at least one primary high-risk criterion (relevant rise in troponin, dynamic ECG changes, and GRACE score > 140), but large randomized trials for these indications are lacking. The results of our study support recommendation for routine invasive treatment of UA patients, even when thay have stabilized after an acute coronary syndrome.  Study limitations. We used the same trials included in the preceding meta-analyses that so far offer background to guidelines' recommendations. There are limitations in the published meta-analyses [4][5][6][7][8][9][10] . Indeed there was heterogeneity in studies with different inclusion criteria, variable times of interventions and differential use of glycoprotein inhibitors. The same limitations apply to our study. As well, the current as the other studies investigated the impact of troponin assays on the outcomes of myocardial infarction before the advent of hs-troponins and the more recent definition of myocardial infarction 43 . The use of hs-troponins and implementation of a lower diagnostic threshold disproportionately increases the incidence of MI. Yet UA still exists and our findings support the use of a routine invasive strategy in this clinical condition. Given that the trials used in this meta-analysis are more consistent with a ' delayed invasive' strategy, it is possible that the available data under-estimate the potential effectiveness of the invasive strategy [44][45][46] . Finally, the significance of peri-procedural myocardial infarction is a subject of considerable debate. A universal definition of myocardial infarction, including peri-procedural myocardial infarction, has been adopted only recently and defines peri-procedural myocardial infarction as elevated cTn values following procedure 43,47 . Unfortunately, the peri-procedural myocardial infarction was variably defined in the included studies and not for all studies (appendix table). However, end-points such as death are indisputable, and there was also a significant increase of death in the invasive arm.

Conclusions
Today, the evidence base for the most appropriate treatment of UA/NSTEMI is still limited, and there are differences between the European and the American guidelines. The current study supports class I recommendation of routine invasive treatment for patients diagnosed as UA. These results emphasize the need for further randomized trials to provide a better evidence base for the invasive versus conservative treatment of patients after UA/NSTEMI.

Study characteristics.
Eligible studies for inclusion were RCTs comparing a routine invasive versus a selective invasive strategy in patients presenting with UA/NSTEMI. Only studies reporting data on cardiac biomarkers were considered for inclusion. Trials were excluded if the majority of patients had STEMI or stable ischemic disease (Fig. 1). Outcome measures. The primary end-points were mortality, recurrent non-fatal MI and their combination identified in two periods of time: from randomization to discharge and from randomization to the end of follow-up. The follow-up was extracted from each original article and was calculated as the mean of all studies included. The mean follow-up was 16 months (range, 6-24).
Data extraction an statistical analysis. Data were extracted independently by two authors of the present study onto data extraction sheets. Disagreement was resolved first by consensus and then by consultation with another two co-authors.
We analyzed trials recruiting participants with positive cardiac biomarkers versus those that did not specify cardiac biomarker status as an inclusion criterion.
To standardize the reporting of our results we calculated relative risk (RR) and 95% confidence intervals (CI) from multiple standard 2 × 2 tables built based on the number of events among the participants for each group in every trial. We used fixed effect model meta-analysis to assess the effect of selective invasive treatment versus routine invasive treatment on the outcomes of interest. The Q statistic and I 2 index were used to assess statistical heterogeneity across trials. The Q statistic was considered significant if the p < 0.10, and heterogeneity was considered high if the I 2 index > 50%.
We also used a random effect model when the statistical test of heterogeneity showed significance. However, the results did not differ qualitatively from those of the primary analysis. Furthermore the results were double-checked performing analysis on the risk ratios' log scale, using the log of the ratio and its corresponding standard error for each study, calculated according to Woolf 's method. Results were displayed as Forest Plots using fixed-effects summary estimates.
To examine the influence of individual studies on the summary effect estimate, we performed an influence analysis, in which the meta-analysis estimates are computed omitting one study at a time. Finally, to analyze associations between treatment effect and study characteristics we used meta-regression for the possible confounders. Analyses were performed using STATA software version 11 and all tests were 2-sided.